- D. Pergolesietal.,Nat. Mater. 9 , 846–852 (2010).
- Y. Wuet al.,ACS Appl. Energy Mater. 1 , 580–588 (2018).
- Y. Xinget al.,ACS Energy Lett. 4 , 2601–2607 (2019).
- J. Garcia-Barriocanalet al.,Science 321 , 676– 680
(2008). - C. Zhaoet al.,Energy Environ. Sci. 13 , 53–85 (2020).
ACKNOWLEDGMENTS
We thank B. Y. Wang’s group from Hubei University for verifying
fuel cell results. We acknowledge Z. Y. Huang from Nanyang
Technological University and J. Su from the Center for Nanoscale
Characterization and Devices, WNLO, of the Huazhong University
of Science and Technology (HUST) for high-resolution TEM
characterization and discussion. We also thank J. Tang from HUST
for helpful discussion.Funding:We acknowledge the support
from the National Natural Science Foundation of China (grants
51774259 and 51772080).Author contributions:Y.W., H.B.S.,
and B.Z. conceived of the idea, designed the experiments, and
analyzed the data. L.L. and Q.S. carried out most of the
characterizations and device optimizations. M.H. performed the
theoretical simulations and analyzed the results. Y.W., M.H.,
C.C., L.L., Q.S., J.S.K., H.B.S., and B.Z. participated in device
optimization and data analysis. J.W. and J.F.L. carried out Raman
characterizations and analyzed the results. L.R.Z. carried out
XANES characterizations and analyzed the results. M.A. verified
fuel cell results. Y.W., H.B.S., and B.Z. wrote the paper. All
authors commented on the manuscript.Competing interests:
The authors declare no competing interests.Data and materialsavailability:All data are available in the manuscript or the
supplementary materials.SUPPLEMENTARY MATERIALS
science.sciencemag.org/content/369/6500/184/suppl/DC1
Materials and Methods
Supplementary Text
Figs. S1 to S12
Tables S1 to S3
References ( 27 – 37 )
19 October 2019; resubmitted 6 March 2020
Accepted 26 May 2020
10.1126/science.aaz9139COGNITIVE MAPS
Cognitive map–basednavigation in wild bats revealed
by a new high-throughput tracking system
Sivan Toledo^1 †, David Shohami^2 †, Ingo Schiffner^2 ‡, Emmanuel Lourie^2 , Yotam Orchan^2 ,
Yoav Bartan^2 , Ran Nathan^2 *
Seven decades of research on the“cognitive map,”the allocentric representation of space, have yielded
key neurobiological insights, yet field evidence from free-ranging wild animals is still lacking. Using
a system capable of tracking dozens of animals simultaneously at high accuracy and resolution,
we assembled a large dataset of 172 foraging Egyptian fruit bats comprising >18 million localizations
collected over 3449 bat-nights across 4 years. Detailed track analysis, combined with translocation
experiments and exhaustive mapping of fruit trees, revealed that wild bats seldom exhibit random
search but instead repeatedly forage in goal-directed, long, and straight flights that include frequent
shortcuts. Alternative, non–map-based strategies were ruled out by simulations, time-lag embedding,
and other trajectory analyses. Our results are consistent with expectations from cognitive map–like
navigation and support previous neurobiological evidence from captive bats.
G
oal-directed navigation, a fundamen-
tal component of animal movement in-
volving a variety of sensory systems,
stimuli, computational mechanisms,
and memory encoding and retrieval,
is essential for accomplishing many spatial
tasks, ranging from nearly global migration
of birds or whales to local foraging of ants or
bees. The most sophisticated form, map-based
navigation, does not require systematically
following known routes or landmarks (pilot-
ing) or directly sensing some goal-emanating
cue (beaconing), which are simpler forms of
goal-directed navigation ( 1 ). Instead, map-based
navigation requires a frame of reference for the
animalÕs current position in relation to a known
yet undetected goal ( 1 ).
For migratory and other animals moving
through unfamiliar areas across large scales,
such a frame of reference is thought to be pro-
vided by predictable environmental gradients
such as EarthÕs magnetic field [Òmap-and-
compassÓnavigation ( 2 , 3 )]. Most animals,
however, move much more frequently within
their smaller-scale familiar home range. At
these scales, the frame of reference for map-
based navigation may be provided by an inter-
nal allocentric representation of space that
supports self-localization and vector calcu-
lation, along with long-term memory of the
spatiotemporal features of the environmentÑ
aÒcognitive mapÓ(sensu Tolman, OÕKeefe, and
Nadel) ( 1 , 4 – 7 ).
The neurobiological basis of spatial repre-
sentation and navigation has been exten-
sively studied in the laboratory ( 5 , 8 ), mostly
on the traditional rodent model species and
primates ( 7 ). The Egyptian fruit bat (Rousettus
aegyptiacus, Pteropodidae) has recently be-
come a model for many such studies, and lab-
based evidence is mounting for its spatial
cognitive abilities and their underlying neuro-
biology ( 7 , 9 , 10 ). Yet, field studies supporting
these findings in this, or other, bat species have
been limited to displacement experiments re-
porting successful homing of bats translocated
well outside their familiar area, suggesting a
map-and-compass navigation ( 11 , 12 ). The ca-
pacity of bats to navigate to multiple targetson a smaller, familiar-area scale relevant for a
cognitive map (which may still be considera-
bly large, spanning ~1000 km^2 for Egyptian
fruit bats) has not yet been examined. Our pre-
vious GPS-tracking showed that bats of this
species regularly fly far and straight to specific
fruit trees within their home range, but the
evidence was limited to partial tracking data
collected over only a few nights ( 11 ).
In this study, we tagged and followed 172 free-
ranging Egyptian fruit bats in the Hula Valley,
Israel, for 3449 full nights (summed across all
bats) using ATLASÑan innovative reverse-GPS
system that localizes extremely light-weight,
low-cost tags ( 13 , 14 )(Fig.1,inset).EachATLAS
tag transmits a distinct radio signal detected
by a base-station network distributed in the
study area. Tag localization is computed using
nanosecond-scale differences in signal time-of-
arrival to each station, enabling nearly real-time
tracking and alleviating the need to retrieve the
tagortohavesomepower-consumingremote-
downloadcapability.Wewerethusabletotrack
wild bats at 0.125 to 1 Hz for up to 219 con-
secutive nights with 4-g tags, <4% of the batÕs
body weight, yielding ~18.2 × 10^6 localizations.
On the basis of the neurobiological evidence
obtained from laboratory animals, we hypoth-
esized that Egyptian fruit bats use spatial
memory to guide their routine foraging navi-
gation in a manner compatible with posses-
sion of a cognitive map. We thus tested a set
of predictions (elaborated below) comparing
cognitive map to simpler alternative navigation
mechanisms (random search, path integration,
piloting and beaconing), applying multiple data
analysis approaches to our dataset.
Bats were mist-netted on fruit trees or cave
entrances and tagged with ATLAS in 38 cap-
ture sessions spanning all seasons between
2015 and 2019. Bats were tagged by gluing
the tag to their back (138 individuals) or by
a custom-made collar (34 individuals) (Fig. 1,
inset), and were tracked for an average of 13
(maximum 38) and 47 (maximum 219) nights,
respectively. We mapped every individual fruit
tree (14,314) and an estimated 18,111 orchard
fruit trees that can potentially be eaten by
Egyptian fruit bats in a 19,000-ha area com-
prising the core area of bat foraging, where188 10 JULY 2020•VOL 369 ISSUE 6500 sciencemag.org SCIENCE
(^1) Blavatnik School of Computer Science, Tel-Aviv University,
Israel.^2 Movement Ecology Lab, Department of Ecology,
Evolution and Behavior, Alexander Silberman Institute of Life
Sciences, Faculty of Science, The Hebrew University of
Jerusalem, Israel.
*Corresponding author. Email: [email protected] (S.T.); david.
[email protected] (D.S.); [email protected] (R.N.)
†These authors contributed equally to this work.
‡Present address: School of Natural Sciences, Bangor University,
Deniol Road, Bangor LL57 2UW, UK.
RESEARCH | REPORTS