Wind energy production in forests conflicts with tree-roosting bats

Abstract

Many countries are investing heavily in wind power generation,1 triggering a high demand for suitable land. As a result, wind energy facilities are increasingly being installed in forests,2,3 despite the fact that forests are crucial for the protection of terrestrial biodiversity.4 This green-green dilemma is particularly evident for bats, as most species at risk of colliding with wind turbines roost in trees.2 With some of these species reported to be declining,5,6,7,8 we see an urgent need to understand how bats respond to wind turbines in forested areas, especially in Europe where all bat species are legally protected. We used miniaturized global positioning system (GPS) units to study how European common noctule bats (Nyctalus noctula), a species that is highly vulnerable at turbines,9 respond to wind turbines in forests. Data from 60 tagged common noctules yielded a total of 8,129 positions, of which 2.3% were recorded at distances <100 m from the nearest turbine. Bats were particularly active at turbines <500 m near roosts, which may require such turbines to be shut down more frequently at times of high bat activity to reduce collision risk. Beyond roosts, bats avoided turbines over several kilometers, supporting earlier findings on habitat loss for forest-associated bats.10 This habitat loss should be compensated by developing parts of the forest as refugia for bats. Our study highlights that it can be particularly challenging to generate wind energy in forested areas in an ecologically sustainable manner with minimal impact on forests and the wildlife that inhabit them.

Full text

Report
Wind energy production in forests conflicts with
tree-roosting bats
Graphical abstract
Highlights
dIn forests, bats were more often present at wind turbines
(WTs) near roosts
dBeyond roosts, bats avoided WTs, but individual responses
varied
dPlacing WTs near roosts may increase the casualty risk and
the need for curtailments
dHabitat loss due to WT operation must be compensated
Authors
Christine Reusch, Ana Ailin Paul,
Marcus Fritze,
Stephanie Kramer-Schadt,
Christian C. Voigt
Correspondence
voigt@izw-berlin.de
In brief
Using high-resolution GPS tracking from
common noctules in Germany, Reusch
et al. reveal that bats were most active at
wind turbines (WTs) in forests when WTs
were placed close to daytime roosts, yet
that bats avoided WTs over several km
distances beyond roosts, indicating that
WTs at forest sites may lead to high
numbers of casualties and habitat loss.
Reusch et al., 2023, Current Biology 33, 1–7
February 27, 2023 ª2022 The Author(s). Published by Elsevier Inc.
https://doi.org/10.1016/j.cub.2022.12.050 ll

Report
Wind energy production in forests
conflicts with tree-roosting bats
Christine Reusch,
1
Ana Ailin Paul,
1,2
Marcus Fritze,
1,5
Stephanie Kramer-Schadt,
1,3,4,7
and Christian C. Voigt
1,2,4,6,8,
*
1
Leibniz Institute for Zoo and Wildlife Research, Alfred-Kowalke-Str. 17, 10315 Berlin, Germany
2
Institute of Biology, Freie Universit€
at Berlin, Ko
¨nigin-Luise-Str. 1-3, 14195 Berlin, Germany
3
Institute of Ecology, Technische Universit€
at Berlin, Rothenburgstr. 12, 12165 Berlin, Germany
4
These authors contributed equally
5
Present address: Zoological Institute and Museum, University of Greifswald, Loitzer Str. 26, 17489 Greifswald, Germany
6
Twitter: @voigtbatlab
7
Twitter: @EcoDynIZW
8
Lead contact
*Correspondence: voigt@izw-berlin.de
https://doi.org/10.1016/j.cub.2022.12.050
SUMMARY
Many countries are investing heavily in wind power generation,
1
triggering a high demand for suitable land. As
a result, wind energy facilities are increasingly being installed in forests,
2,3
despite the fact that forests are
crucial for the protection of terrestrial biodiversity.
4
This green-green dilemma is particularly evident for
bats, as most species at risk of colliding with wind turbines roost in trees.
2
With some of these species re-
ported to be declining,
5–8
we see an urgent need to understand how bats respond to wind turbines in forested
areas, especially in Europe where all bat species are legally protected. We used miniaturized global posi-
tioning system (GPS) units to study how European common noctule bats (Nyctalus noctula), a species that
is highly vulnerable at turbines,
9
respond to wind turbines in forests. Data from 60 tagged common noctules
yielded a total of 8,129 positions, of which 2.3% were recorded at distances <100 m from the nearest turbine.
Bats were particularly active at turbines <500 m near roosts, which may require such turbines to be shut down
more frequently at times of high bat activity to reduce collision risk. Beyond roosts, bats avoided turbines
over several kilometers, supporting earlier findings on habitat loss for forest-associated bats.
10
This habitat
loss should be compensated by developing parts of the forest as refugia for bats. Our study highlights that it
can be particularly challenging to generate wind energy in forested areas in an ecologically sustainable
manner with minimal impact on forests and the wildlife that inhabit them.
RESULTS AND DISCUSSION
We used high-resolution biologging to investigate the response
of common noctule bats (Nyctalus noctula) to wind turbines at
a predominantly forested site in late spring and late summer,
which is the time when most casualties occur at turbines in Cen-
tral Europe.
9
We hypothesized that the movement activity of bats
would depend on the distance to turbines and roosts. We pre-
dicted that common noctules would be most likely active at
wind turbines next to daytime roosts and that beyond the vicinity
of daytime roosts common noctules avoid wind turbines.
10,11
Specifically, bats were expected to avoid central wind turbines
more than peripheral wind turbines and particularly those with
large rotors.
10
We retrieved global positioning system (GPS) tags from 60 of
80 tagged common noctule bats, yielding 8,129 spatial positions
(Figure 1). After emerging from roosts at around 12 min after sun-
set (median), bats traveled over distances of 16 km (median) per
night, thereby covering areas of 11 ± 34 km
2
(median ± one SD;
kernel density area, KDE 95) (Table S1). The assignment of
movement modes to spatial positions by hidden Markov models
revealed that about 40% belonged to commuting (COM), 51% to
area-restricted movement (ARM) indicating insect hunting, and
9% to undefined movements. ARM was observed mostly over
forests and COM over farmland and meadows (Figure S1). The
estimates on flight height confirm that common noctule bats
were flying in the range of the rotor-swept area of turbines (me-
dian 60 m; range 0–614 m; after neglecting negative values; Fig-
ure 2). Overall, 2.3% of observed spatial positions were recorded
at distances <100 m from the nearest wind turbine (for compar-
ison, 1.4% of the created random positions). A resource selec-
tion analysis (Table S3) revealed a higher presence probability
of bats at wind turbines sited in forests when these turbines
were placed next to daytime roosts of bats (Figure 3A1). This ef-
fect was absent for wind turbines placed outside forests,
although roosts were also present there, for example, at avenue
trees. However, 79% of documented roosts were in the forest.
Our study highlights that it is key for the permitting process to
search intensively for bat roosts in forests designated for wind
turbine deployments and to inform stakeholders about suitable
and unsuitable sites for wind turbines. We recommend that
wind turbines are not installed near daytime roosts because of
Current Biology 33, 1–7, February 27, 2023 ª2022 The Author(s). Published by Elsevier Inc. 1
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
ll
OPEN ACCESS
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050

the following two reasons: (1) a high activity of bats at wind tur-
bines close to roosts may lead to high numbers of casualties.
Many aerial insectivores such as common noctule bats use tree
hollows for maternity colonies (late spring and early summer)
and for mating (late summer and early fall), which may lead to
an increased abundance of individuals in the vicinity of roosts
during these periods. After weaning, juveniles may use the area
around maternity roosts for their first flights, starting around
Figure 1. Flight paths of 60 common noctule bats
(A) Study area (black box) in Europe drawn with ‘‘rworldmap.’’
12
(B) Flight paths, tree roosts (black triangles), wind turbines (WTs; circles with a black center; n = 80),
13
and forest (green), as well as water (blue) (Corine: GeoBasis-
DE/BKG 2018).
(C) WT symbol size correlates to the rotor diameter. Colors indicate avoidance (blue), prefere nce (red), or indifferent (purple) behavior of bats toward WTs given as
the relationship between random versus observed locations within a 100 m radius around wind turbines. White: no locations.
See also Figure S1 and Table S1.
ll
OPEN ACCESS
2Current Biology 33, 1–7, February 27, 2023
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

July in the case of temperate zone bats of the Northern hemi-
sphere. Apparently, juvenile bats are particularly vulnerable at
wind turbines.
14
In late summer, bats may swarm around mating
roosts, which may also cause an increased abundance of bats
around such roosts. Frequent bat casualties at wind turbines
next to daytime roosts may deplete local populations and cause
local or regional extinction of species.
5,6
It should be noted that
the availability of natural roosts may change during the operation
period of wind turbines owing to the natural decay of trees and the
activity of woodpeckers. Therefore, even when avoiding bat
roosts at the time of turbine erection, it cannot be ruled out that
roosts establish during the later operation period of turbines,
particularly in case of cavities created by woodpeckers, which
seems to be preferred by common noctules.
15
(2) If permitting au-
thorities grant companies to set up wind turbines in forests, it
seems necessary to curtail the operation of wind turbines during
times of high bat activity.
2
In Germany, acoustic surveys are con-
ducted to identify ambient conditions when bats are most active
in the rotor-swept area. These data can then be used to formulate
criteria for stopping temporarily the operation of wind turbines
when bats are active.
2
Such curtailment schemes may reduce
the number of casualties by more than 80% at low revenue losses
for the company; however, this has only been shown for turbines
sited in open landscapes.
16
We argue that it is likely that such cur-
tailments are stricter for wind turbines operating in than outside
forests. This will lower the potential for wind power generation
at forested sites, leading to lowered contributions to the reduction
of greenhouse gas emissions and lowered monetary revenues for
companies. However, how do bats respond to wind turbines in
forests at some distance to bat roosts?
To account for the roost effect at forested sites, we excluded
all spatial positions of bats at distances <500 m from the nearest
roost based on the median of the spatial position in relation to the
wind turbines rounded up to the nearest hundredth digit
(removing 25,949 spatial positions [53.2%] of the total dataset
and 18,203 [55.2%] for the male subset). Removing this effect,
we observed the avoidance of bats toward wind turbines inside
and outside forests (Figures 3A2,S2A2, S2A4, and S2A6). The
rotor diameter of wind turbines had no effect on the avoidance
response of common noctule bats (Figures 3B4 and 3B6). How-
ever, we observed that bats tend to be less active at central than
at peripheral wind turbines at forest sites, whereas bats selected
central wind turbines more than peripheral wind turbines outside
forests (Figures 3C2, 3C4, and 3C6). The avoidance behavior of
bats at the population level was most apparent for males in late
summer (Figures S2F2, S2F4, and S2F6). Avoidance responses
of common noctule bats over several km distances have already
been observed in a study conducted at a coastal migration
corridor, which suggested that bats lose habitats in landscapes
with high densities of wind turbines.
11
Acoustic surveys at farm-
land sites also confirm that the acoustic activity of bats de-
creases over 1 km distance toward wind turbines.
17,18
Recently,
it was shown that forest specialist bats, which belong to the guild
of narrow-space foragers,
19
halved their acoustic activity along a
transect from 450 m distance to 80 m distance from wind tur-
bines at forested site.
10
Accordingly, avoidance behavior toward
wind turbines may be a general phenomenon observed over
larger spatial scales in various bat species across different func-
tional guilds. This avoidance behavior may be driven by turbine
noises, which may startle bats or interfere with their acoustic
orientation—possibly also with the network foraging style of
open-space foraging bats like common noctules that depend
on eavesdropping on conspecifics.
20,21
However, do all individ-
uals respond in a similar way to wind turbines?
To elucidate individual responses of common noctule bats to
wind turbines,we analyzed the data usingresource selection func-
tions (distance to closest wind turbine; integrated as individual
random slopes in models). This analysis confirmed that on the
population level (fixed effects) common noctule bats avoided
wind turbines overseveral km distances; however, responses var-
ied largely betweenindividuals (Figure 4;Table S3). Across all sea-
sons and sexes, only 14% of the individuals differed from the
observed general avoidance response toward wind turbines,
and this difference was not explained by habitat. In addition, we
conducted a seasonal comparison of individual responsestoward
wind turbines to elucidate whether the response differs between
early and late summer. This analysis was only performed with
data from male bats since we were not allowed to work with preg-
nant or lactating females in early summer. Based on the data of
males, we observed indifferent responses of males toward wind
turbines in early summer, but avoidance of wind turbines in late
summer (Figures 4B and 4C; Table S3). We were not able to iden-
tify specific factors that explained the inter-individual variation of
the response behavior of bats toward wind turbines (Figure S3).
22
Figure 2. Estimated flight height above
ground (m) in relation to main land cover
and rotor blade density
Background color intensity depicts the rotor blade
density in the study area at the respective heights.
The x axis shows the different main land cover
categories (50 m radius around the GPS location):
farmland, meadow, shrubs and herbaceous vege-
tation (shrubs), forest, sealed surface (sealed),
water, and wetland. ‘‘Diverse’’ describes GPS lo-
cations covering several land cover categories with
no category exceeding 50% within the 50 m buffer.
Negative flight height estimations were disregarded
(Dis. neg val, number of disregarded values, and
percentage per main land cover in parenthesis).
See also Table S1.
ll
OPEN ACCESS
Current Biology 33, 1–7, February 27, 2023 3
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

(legend on next page)
ll
OPEN ACCESS
4Current Biology 33, 1–7, February 27, 2023
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

It is possible that unaccounted intrinsic or extrinsic factors might
cause bats to responddifferently. For example,bats could engage
in exploratorybehavior and mistake wind turbines for largetrees.
23
The observation of a less pronounced avoidance behavior toward
wind turbines in late compared with early summer is consistent
with this idea. Juvenile bats mightexplore turbine structures;how-
ever, we could not test age as a factor since fully grown juvenile
and adult bats looked similar. Bats may also respond to wind tur-
bines depending on whether or not wind turbines operate under a
curtailment scheme. Unfortunately, we had no access to the oper-
ation schemes of local turbines, which prevents us from con-
ducting a more detailed analysis.
Conclusion
Forests are preferred habitats for many bat species worldwide.
Building wind turbines in forests goes along with fragmentation
when maintenance roads are built and with habitat loss when
clearings are created for wind turbines.
2
In Europe, monitoring
schemes are usually required to assess the potential of wind tur-
bines to impact forest-associated bat species. Accordingly,
habitat loss caused by clearings has to be compensated by
setting aside other forest patches for bats. Based on our GPS
study, we now reveal that common noctule bats are most likely
active at turbines within a 500 m distance to daytime roosts.
Therefore, we recommend maintaining a minimum distance of
500 m between bat roosts and wind turbines. Further, strict cur-
tailments should be put into practice to avoid high numbers of
casualties at wind turbines in forested areas. In addition, we
observed that—beyond 500 m distance to roosts—common
noctule bats avoid wind turbines. This avoidance behavior to-
ward wind turbines is consistent with recent observations in
the same species and also in bats of other functional
guilds.
10,11,17,18
Avoidance of bats toward wind turbines might
have remained unnoticed until recently because previous sur-
veys have focused primarily on the rotor-swept area of turbines
using either ultrasonic detectors or thermal imagery.
2,23
Howev-
er, the detection ranges of these techniques for monitoring bats
are limited to a few tens of meters.
24
In summary, wind turbines
at forested sites impact bats in several negative ways. If wind tur-
bines have to be built in forests, i.e., in the absence of alternative
sites or alternative sources of renewable energy, we call for
engaging in detailed pre-construction surveys that involve
searching for potential roosts in the vicinity of prospected sites.
The observation of bats avoiding wind turbines over several hun-
dred meters
10
or even several km suggests that wind turbine
operation in forests leads to habitat loss for bats over a larger
spatial scale than currently considered during the permitting pro-
cess. We therefore request that this habitat loss be considered in
the planning of wind energy facilities. If wind turbines are planned
to be located in forests, we suggest that forest areas larger than
the cumulative area of turbine clearings and maintenance roads
be provided to compensate for the habitat degradation
associated with wind turbine operation. The selected forest
areas should be at a sufficient distance from wind turbines to
avoid any disturbances caused by wind turbines over long dis-
tances. Further, the set-aside forests should be of similar struc-
ture if not even of higher quality.
STAR+METHODS
Detailed methods are provided in the online version of this paper
and include the following:
dKEY RESOURCES TABLE
dRESOURCE AVAILABILITY
BLead contact
BMaterials availability
BData and code availability
dEXPERIMENTAL MODEL AND SUBJECT DETAILS
BExperimental bats and site
dMETHOD DETAILS
BGPS Attachment and Data Collection
BData preparation and environmental predictor vari-
ables
dQUANTIFICATION AND STATISTICAL ANALYSIS
SUPPLEMENTAL INFORMATION
Supplemental information can be found online at https://doi.org/10.1016/j.
cub.2022.12.050.
ACKNOWLEDGMENTS
We thank Lukas Schwarzmeier, Manuel Roeleke, and Gabriel Pelz for helping
at various stages of the project. Big thanks go to C
edric Scherer and Moritz
Wenzler for preparing the environmental layers. Additional thanks go to C
edric
Scherer for advice on improving our figures. This project was funded by the
Deutsche Bundesstiftung Umwelt DBU (34411/01-43/0).
AUTHOR CONTRIBUTIONS
C.C.V. and S.K.-S. conceptualized the study. C.R., A.A.P., and M.F. carried
out the fieldwork. C.R. conducted the analyses with the support of A.A.P.
and S.K.-S. C.C.V. and C.R. drafted the original manuscript. S.K.-S. reviewed
and edited the manuscript. All authors commented on the manuscript.
DECLARATION OF INTERESTS
The authors declare no competing interests.
INCLUSION AND DIVERSITY
We support inclusive, diverse, and equitable conduct of research.
Received: August 10, 2022
Revised: October 18, 2022
Accepted: December 20, 2022
Published: January 20, 2023
Figure 3. Presence probability (95% CI) of common noctule bats in relation to WT characteristics
(A1–A6) Distance to roost (km), (B1–B6) rotor diameter (m), and (C1–C6) centrality (number of WTs within 1 km radius). The response was estimated for (A) bats
close to WTs in different main land covers around turbines (red box), as well as (B) turbines in open areas (yellow) and in forests (green) regarding distance to WTs
(near [0.1 km]; far [2 km]). The remaining fixed effects were set to the median. The red dashed line describes the 16.7% threshold (observed:random posi-
tions = 1:5) separating habitat avoidance (below) from preference (above).
See also Figures S2–S4 and Tables S2 and S3.
ll
OPEN ACCESS
Current Biology 33, 1–7, February 27, 2023 5
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

REFERENCES
1. GWEC (Global Wind Energy Council) (2021). Global wind report 2021.
https://gwec.net/wp-content/uploads/2021/03/GWEC-Global-Wind-Report-
2021.pdf.
2. Arnett, E.B., Baerwald, E.F., Mathews, F., Rodrigues, L., Rodrı
´guez-Dura
´n,
A., Rydell,J., Villegas-Patraca,R., and Voigt, C.C. (2016).Impacts of winden-
ergy development on bats:a globalperspective. In Batsin the Anthropocene:
Conservation of Bats in a Changing World (Springer), pp. 295–323.
3. Bunzel, K., Bovet, J., Thr€
an, D., and Eichhorn, M. (2019). Hidden outlaws in
the forest? A legal and spatial analysis of onshore wind energy in Germany.
Energy Res. Soc. Sci. 55, 14–25.
4. FAO (2015). Global Forest Resources Assessment 2015—how are the
world’s forests changing? (Food and Agriculture Organization of the
United Nations).
5. Frick, W.F., Baerwald, E.F., Pollock, J.F., Barclay, R.M.R., Szymanski,
J.A., Weller, T.J., Russell, A.L., Loeb, S.C., Medellin, R.A., and McGuire,
L.P. (2017). Fatalities at wind turbines may threaten population viability
of a migratory bat. Biol. Conserv. 209, 172–177.
6. Friedenberg, N.A., and Frick, W.F. (2021). Assessing fatality minimization
for hoary bats amid continued wind energy development. Biol. Conserv.
262, 109309.
7. BfN (Bundesamt fu
¨r Naturschutz) (2018). National implementation report
to EUROBATS. https://www.eurobats.org/sites/default/files/documents/
pdf/Meeting_of_Parties/Inf.MoP8_.21_NIR_Germany.pdf.
8. Printz, L., Tschapka, M., and Vogeler, A. (2021). The common noctule bat
(Nyctalus noctula): population trends from artificial roosts and the effect of
biotic and abiotic parameters on the probability of occupation. J. Urban
Ecol. 7, juab033.
9. Rydell, J., Bach, L., Dubourg-Savage, M.J., Green, M., Rodrigues, L., and
Hedenstro
¨m, A. (2010). Bat mortality at wind turbines in northwestern
Europe. Acta Chiropterol. 12, 261–274.
10. Ellerbrok,J.S., Delius, A.,Peter, F., Farwig,N., and Voigt, C.C.(2022). Activity
of forest specialist bats decreases towards wind turbines at forest sites.
J. Appl. Ecol. 59, 2497–2506. https://doi.org/10.1111/1365-2664.14249.
11. Reusch, C., Lozar, M., Kramer-Schadt, S., and Voigt, C.C. (2022). Coastal
onshore wind turbines lead to habitat loss for bats in Northern Germany.
J. Environ. Manag. 310, 114715. https://doi.org/10.1016/j.jenvman.2022.
114715.
12. South, A. (2011). rworldmap: a new R package for mapping global data. R
J. 3, 35–43.
13. LfU (Landesamt fu
¨r Umwelt Brandenburg) (2020). Windkraftanlagen des
Landes Brandenburg. https://mlul.brandenburg.de/lua/gis/wka.zip.
14. Kruszynski, C., Bailey, L.D., Bach, L., Bach, P., Fritze, M., Lindecke, O.,
Teige, T., and Voigt, C.C. (2022). High vulnerability of juvenile Nathusius’
pipistrelle bats (Pipistrellus nathusii) at wind turbines. Ecol. Appl. 32,
e2513.
15. Boonman, M. (2000). Roost selection by noctules (Nyctalus noctula) and
Daubenton’s bats (Myotis daubentonii). J. Zool. 251, 385–389.
16. Arnett, E.B., Huso, M.M., Schirmacher, M.R., and Hayes, J.P. (2011).
Altering turbine speed reduces bat mortality at wind-energy facilities.
Front. Ecol. Environ. 9, 209–214.
17. Barr
e, K., Le Viol, I., Bas, Y., Julliard, R., and Kerbiriou, C. (2018).
Estimating habitat loss due to wind turbine avoidance by bats: implica-
tions for European siting guidance. Biol. Conserv. 226, 205–214.
18. Millon, L., Colin, C., Brescia, F., and Kerbiriou, C. (2018). Wind turbines
impact bat activity, leading to high losses of habitat use in a biodiversity
hotspot. Ecol. Eng. 112, 51–54.
19. Denzinger, A., and Schnitzler, H.U. (2013). Bat guilds, a concept to classify
the highly diverse foraging and echolocation behaviors of microchirop-
teran bats. Front. Physiol. 4, 164.
20. Roeleke, M., Blohm, T., Hoffmeister, U., Marggraf, L., Schl€
agel, U.E.,
Teige, T., and Voigt, C.C. (2020). Landscape structure influences the use
of social information in an insectivorous bat. Oikos 129, 912–923.
21. Roeleke, M., Schl€
agel, U.E., Gallagher, C., Pufelski, J., Blohm, T., Nathan,
R., Toledo, S., Jeltsch, F., and Voigt, C.C. (2022). Insectivorous bats form
mobile sensory networks to optimize prey localization: the case of the
common noctule bat. Proc. Natl. Acad. Sci. USA 119. e2203663119.
22. Roeleke, M., Blohm, T., Kramer-Schadt, S., Yovel, Y., and Voigt, C.C.
(2016). Habitat use of bats in relation to wind turbines revealed by GPS
tracking. Sci. Rep. 6, 28961.
23. Cryan, P.M., Gorresen, P.M., Hein, C.D., Schirmacher, M.R., Diehl, R.H.,
Huso, M.M., Hayman, D.T.S., Fricker, P.D., Bonaccorso, F.J., Johnson,
D.H., et al. (2014). Behavior of bats at wind turbines. Proc. Natl. Acad.
Sci. USA 111, 15126–15131.
24. Voigt, C.C., Russo, D., Runkel, V., and Goerlitz, H.R. (2021). Limitations of
acoustic monitoring at wind turbines to evaluate fatality risk of bats.
Mamm. Rev. 51, 559–570. https://doi.org/10.1111/mam.12248.
Figure 4. Individual responses of common noctule bats toward wind
turbines (WTs) excluding positions <500 m near roosts
Presence probability in relation to wind turbines for both sexes and seasons
(A), and males only in early (B) and late summer (C). Solid lines represent in-
dividuals (dark red, negative slope; blue, positive slope), and dashed lines
represent population levels (gray polygon, prediction interval). The red dashed
line separates avoidance (below) from preference (above). Stacked boxplots
next to the figures show the percentage of negative and positive slopes and
the total sample size on top.
See also Figures S2–S4 and Tables S2 and S3.
ll
OPEN ACCESS
6Current Biology 33, 1–7, February 27, 2023
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

25. R Core Team (2019). R: a language and environment for statistical
computing (R Foundation for Statistical Computing). https://www.
R-project.org/.
26. RStudio Team (2016). RStudio: integrated development for R (RStudio).
http://www.rstudio.com/.
27. GeoBasis-DE/BKG (Federal Agency for Cartography and Geodesy).
(2018). Digitales Landbedeckungsmodell fu
¨r Deutschland (LBM-
DE2018). https://gdz.bkg.bund.de.
28. GeoBasis-DE/BKG (Federal Agency for Cartography and Geodesy)
(2015). Digitales Gel€
andemodell Gitterweite 200m fu
¨r Deutschland (DGM
200). https://gdz.bkg.bund.de.
29. Cvikel, N., Egert Berg, K.E., Levin, E., Hurme, E., Borissov, I., Boonman,
A., Amichai, E., and Yovel, Y. (2015). Bats aggregate to improve prey
search but might be impaired when their density becomes too high.
Curr. Biol. 25, 206–211.
30. Egert-Berg, K., Hurme, E.R., Greif, S., Goldstein, A., Harten, L., Herrera M,
L.G., Flores-Martı
´nez, J.J., Vald
es, A.T., Johnston, D.S., Eitan, O., et al.
(2018). Resource ephemerality drives social foraging in bats. Curr. Biol.
28, 3667–3673.e5.
31. Voigt, C.C., Scholl, J.M., Bauer, J., Teige, T., Yovel, Y., Kramer-Sc hadt, S.,
and Gras, P. (2020). Movement responses of common noctule bats to the
illuminated urban landscape. Landsc. Ecol. 35, 189–201. https://doi.org/
10.1007/s10980-019-00942-4.
32. P
eron, G., Calabrese, J.M., Duriez, O., Fleming, C.H., Garcı
´a-Jim
enez, R.,
Johnston, A., Lambertucci, S.A., and Shepard, E.L. (2020). The chall enges
of estimating the distribution of flight heights from telemetry or altimetry
data. Anim. Biotelem. 8, 1–13.
33. Michelot, T., Langrock, R., and Patterson, T.A. (2016). moveHMM: an R
package for the statistical modelling of animal movement data using hid-
den Markov models. Methods Ecol. Evol. 7, 1308–1315.
34. Brooks, M., Kristensen, K., Benthem, K., Magnusson, A., Berg, C.,
Nielsen, A., Skaug, H., M€
achler, M., and Bolker, B. (2017). glmmTMB bal-
ances speed and flexibility among packages for zero-inflated generalized
linear mixed modeling. R J. 9, 378–400.
35. Burnham, K.P., Anderson, D.R., and Huyvaert, K.P. (2011). AIC model se-
lection and multimodel inference in behavioral ecology: some back-
ground, observations, and comparisons. Behav. Ecol. Sociobiol.
65, 23–35.
36. Culina, A., Linton, D.M., and Macdonald, D.W. (2017). Age, sex, and
climate factors show different effects on survival of three different bat spe-
cies in a woodland bat community. Glob. Ecol. Conserv. 12, 263–271.
https://doi.org/10.1016/j.gecco.2017.11.009.
37. Mukaka, M.M. (2012). Statistics corner: a guide to appropriate use of cor-
relation coefficient in medical research. Malawi Med. J. 24, 69–71.
ll
OPEN ACCESS
Current Biology 33, 1–7, February 27, 2023 7
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

STAR+METHODS
KEY RESOURCES TABLE
RESOURCE AVAILABILITY
Lead contact
Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Christian C. Voigt (voigt@
izw-berlin.de).
Materials availability
This study did not generate new unique reagents.
Data and code availability
The datasets used generated in this study are available at https://doi.org/10.5281/zenodo.7535030. R code used in analysis is avail-
able from the lead contact.
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Experimental bats and site
Our work was conducted during late spring (May and June) and late summer (August and September) in 2019 and 2020 in Branden-
burg (Germany) under the animal welfare license #2347-5-2019 and conservation license #4543/131+3#27171/2019. Our study site
was dominated by forests (43 % of local land cover) and farmland (28 %), where wind turbines were placed both either in pine sil-
vicultures (30 % of wind turbines) or farmland (50 %). The study area included three wind parks with a total of 80 wind turbines
(on average 27 per wind park).
REAGENT or RESOURCE SOURCE IDENTIFIER
Deposited data
Annotated GPS locations This study https://doi.org/10.5281/
zenodo.7535030
Experimental models: Organisms/strains
Common noctule bat Nyctalus noctula Northeastern Germany; n
atural tree roosts
N/A
Software and algorithms
R 3.6.2
25
The R Project for
Statistical computing
25
https://cran.r-project.org/
mirrors.html
R Studio 1.1.383 R Studio
26
https://www.rstudio.com/
products/rstudio/download/
QGIS 3.10.3 QGIS Geographic Information
System, QGIS Association
https://www.qgis.org
Inkscape Inkscape Project, 2020 https://inkscape.org
Other
GPS nanoFix GEO-MINI Pathtrack, Otley, United Kingdom https://www.pathtrack.co.uk/
Corine land cover map ("Digitales L
andbedeckungsmodell fu
¨r Deutschland",
2018 (LBM-DE2018))
GeoBasis-DE/ Federal Agency f
or Cartography and Geodesy (BKG)
27
https://gdz.bkg.bund.de
Digital terrain model ("Digitales Gel€
andemodell
Gitterweite 200m fu
¨r Deutschland‘‘,
2015 (DGM 200))
GeoBasis-DE/ Federal Agency
for Cartography and Geodesy (BKG)
28
https://gdz.bkg.bund.de
Wind turbine data ("Windkraftanlagen d
es Landes
Brandenburg"; last update 10.01.2020)
Landesamt fu
¨r Umwelt Brandenburg
13
https://mlul.brandenburg.de/
lua/gis/wka.zip
ll
OPEN ACCESS
e1 Current Biology 33, 1–7.e1–e3, February 27, 2023
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

METHOD DETAILS
GPS Attachment and Data Collection
During late spring, we focused on males to avoid disturbance of maternity roosts. At daytime, we monitored bat boxes in the vicinity
of wind parks for the presence of common noctule bats. We selected bats for our study from these boxes. In case we observed bats in
natural roosts, we captured emerging bats with mistnets (8-10m length, 10 mm mesh size, Solida, Steinbach, Germany; 6-9m length,
16mm mesh size, Ecotone, Gdynia, Poland) set up at various heights (2-18 m) in front of the roost. We noted the sex and the age
(based on epiphyseal closure) in each captured bat. In addition, we measured the forearm length (0.1mm, digital caliper, Ecotone,
Gdynia, Poland) and body mass (precision 0.1 g, electronic pocket scale, Ecotone, Gdynia, Poland). The GPS-VHF unit was pro-
tected from the elements by a light-weight rubber bag, which was glued to the back of bats with skin glue (Torbot bonding cement,
Torbot, Cranston, USA). By attaching the unit to the fur (and not the skin), we ensured that the GPS-VHF unit would fall off the bat
within about 5 days. For retrieving the tagged bats or the GPS-VHF unit, we located the VHF signal of the radio-transmitter by homing
in using a receiver and an antenna (ICOM IC-R30, ICOM, Japan; Australis 26K Receiver, Titley Scientific, Australia). The complete unit
made up on average 7.9 % of the bat’s body mass, which we considered acceptable given the shortness of our experiment, and since
past studies could not detect negative effects on bats equipped with these or heavier tags.
14,20–22,29–31
In total, we equipped 80 com-
mon noctules with GPS logger units (nanoFix GEO-MINI. Pathtrack, Otley, United Kingdom), of which we retrieved 60. Each tag was
also equipped with a radiotransmitter (Telemetrie-Dessau, Dessau, Germany) for retrieving either the tagged bat or the separated
GPS unit. Units weighed about 2.3 g, which made up 7.9 ± 0.9 % (mean ± one standard deviation) of the bats’ body masses
(29.1 ± 3.1 g; range 24.1-39.7 g). GPS units were programmed to start recordings at the day when tags were employed if the individual
was caught during the day or in the subsequent night when bats were captured while emerging from the roost. This ensured that
caught individuals habituated to the attached GPS logger. For recording nights, GPS units sampled spatial positions every minute,
starting at 20:00 or 21:00 hours (CET; depending on sunset) and lasting until 02:00 hours (CET), the presumed latest return time of
common noctules.
22,31
Data preparation and environmental predictor variables
For all data processing and analysis we used the software R
25
and R Studio.
26
We described flight paths by using basic parameters
such as total distance traveled, duration (Table S1;Figure S1), step length (distance between subsequent spatial positions of a track),
turning angles (angle between three subsequent spatial positions of a track) and speed (step length divided by time elapsed between
the corresponding spatial positions). We determined the flight height above the ground (m) (see results and discussion, as well as,
Figure 2) by subtracting the height of the Earth’s surface
28
from the height above the geoid estimated by our GPS devices. Negative
values were disregarded (Figure 2). Due to the high level of imprecision of altitude measured by GPS devices,
32
we did not use the
flight height above ground estimates in further analyses. We used Hidden Markov models to assign one of two movement modes,
namely Area Restricted Movement behavior (ARM) or COMmuting behavior (COM), to each GPS location based on step length and
turning angles. We set mean step lengths to starting values of 40 m (± 40 m standard deviation SD; state one/ ARM) and 200 m (±
200 m SD; state two/ COM). Furthermore, turning angle means were set to p(state one/ ARM) and 0(state two/ COM). We fitted the
Hidden Markov model with a gamma distribution for step length and von Mises distribution for turning angles (fitHMM function, R
package ‘moveHMM’).
33
A threshold of 0.75 was set to correctly assign one of the states to a GPS location. We characterized
GPS locations below this certainty level as undefined. The two resulting states were later used for separating foraging from search
behavior. Specifically, short step lengths and larger turning angles were used to identify foraging behavior (indicated by ARM).
Straight trajectories defined by small turning angles and rapid flights defined by large step lengths were used to identify search
behavior (indicated by COM). Based on a 20 m raster of the Corine land cover map (LBM-DE2018; see key resources table), we as-
signed the main underlying land cover type within a buffer of 50 m to each GPS location. The land cover types were summarized into
nine categories: sealed surface, city green, farmland, meadows, shrubs and herbaceous vegetation, forest, open natural areas,
wetland and water. After estimating the proportion of each land cover category within the 50 m buffer, we selected the category
with the highest proportion inside the buffer. If the main category represented >50 % of the 50 m buffer, the GPS location was as-
signed to this land cover type, otherwise it was characterized as ‘‘diverse’’. Furthermore, we calculated the distances of the GPS
locations to the closest WT.
11,27
QUANTIFICATION AND STATISTICAL ANALYSIS
The statistical analysis was also done with the software R
25
and R Studio.
26
We tested whether avoidance of or attraction to wind
turbines, expressed as the probability of bat presence, depended on features (rotor diameter) and site characteristics of wind tur-
bines (WT; distance to potential bat roosts and density of wind turbines, i.e. central or peripheral location of WT in relation to others
in a radius of 1 km around wind turbines, hereafter ‘centrality’). We set the rotor diameter, distance to roosts and WT centrality in
statistical interactions with the surrounding main land cover of the WT (forest and non-forest areas) and distance of bat spatial po-
sitions to wind turbines (Table S3). Thus, the full model formula was: Presence (rotor diameter + distance to potential bat roosts +
centrality) * forest and non-forest + (rotor diameter + distance to potential bat roosts + centrality) * distance bat-WT + (distance bat-
WT | bat ID). A subset of data from only males was used to test for seasonal differences by including three-way-interactions with field
season in the model selection. This leads to the following full model formula for the male data subset: Presence ((rotor diameter +
ll
OPEN ACCESS
Current Biology 33, 1–7.e1–e3, February 27, 2023 e2
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report

distance to potential bat roosts + centrality) * forest and non-forest) * season + ((rotor diameter + distance to potential bat roosts +
centrality) * distance bat-WT) * season + (distance bat-WT | bat ID). The analyses were based on resource selection functions (RSF), in
which we analyzed preference for certain WT characteristics against randomized spatial positions within the 100% minimum convex
polygon of individual flight paths (RSF-MCP100
flightpath
) in a use-vs.-availability design (see Figure S4), based on the protocol of Re-
usch et al.
11
We placed five random positions per observed GPS location randomly anywhere within the 100% MCP of the according
individual flight path and fitted generalized linear mixed effect models with Template Model Builder (glmmTMB function from R pack-
age ‘glmmTMB’)
34
with binomial error distribution (see Tables S2 and S3). The resulting ratio of observed to random positions of 1:5
leads to a 16.7 % threshold (calculation: (1/(1+5))*100 = (1/6)*100 = 16.7 %) separating habitat avoidance (below) from preference
(above) (e.g. see Figures 3 and S3). We included individuals as random effects in the RSF to test for differences in the response
of individuals toward WT (see Figures 4 and S4;Table S3). Specifically, individuals were included as random intercept and the dis-
tance to closest wind turbine was integrated as individual random slopes in the models (see Tables S2 and S3). Model selection was
based on the Akaike information criterion corrected for small sample sizes (AICc) and we selected simpler models whenever
dAICci < 2, for dAICci = AICci – AICcmin (see Table S2).
35,36
We, additionally, performed model averaging for the candidate models
with a dAICc < 2 to confirm our results (see Table S3). Numeric variables included as fixed effects in the model selection were tested
for multicollinearity and only one of the compared variables was included in the model if |Kendall’s tau| > 0.7.
37
This was the case for
rotor diameter and hub height of WT in the study area (Kendall’s tau = 0.97, N
observations
=48774). We chose rotor diameter for the
further analyses based on it indicating the actual risk area for bats. To assess model quality we determined the area under the curve
(AUC) (see Table S3). For better visualization 0.1 km (close) and 2 km (far) were chosen as representative values for distance to WT in
figures based on observations in the field.
ll
OPEN ACCESS
e3 Current Biology 33, 1–7.e1–e3, February 27, 2023
Please cite this article in press as: Reusch et al., Wind energy production in forests conflicts with tree-roosting bats, Current Biology (2022), https://
doi.org/10.1016/j.cub.2022.12.050
Report