Behavioural Processes 102 (2014) 51–61

Contents lists available at ScienceDirect

Behavioural Processes
journal homepage: www.elsevier.com/locate/behavproc

Beginnings of a synthetic approach to desert ant navigation

Ken Cheng a,∗ , Patrick Schultheiss a , Sebastian Schwarz b , Antoine Wystrach c ,
Rüdiger Wehner d
a
Department of Biological Sciences, Macquarie University, Australia
b
Department of Psychology, Neuroscience & Behaviour, McMaster University, Canada
c
School of Life Sciences, University of Sussex, United Kingdom
d
Brain Research Institute, University of Zürich, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: In a synthetic approach to studying navigational abilities in desert ants, we review recent work compar-
Received 1 July 2013 ing ants living in different visual ecologies. Those living in a visually rich habitat strewn with tussocks,
Received in revised form bushes, and trees are compared to those living in visually barren salt pans, as exemplified by the Central
27 September 2013
Australian Melophorus bagoti and the North African Cataglyphis fortis, respectively. In bare habitats the
Accepted 5 October 2013
navigator must rely primarily on path integration, keeping track of the distance and direction in which
it has travelled, while in visually rich habitats the navigator can rely more on guidance by the visual
Keywords:
panorama. Consistent with these expectations, C. fortis performs better than M. bagoti on various meas-
Desert ant
Navigation
ures of precision at path integration. In contrast, M. bagoti learned a visually based associative task better
Path integration than C. fortis, the latter generally failing at the task. Both these ants, however, exhibit a similar pattern
Visual panorama of systematic search as a ‘back up’ strategy when other navigational strategies fail. A newly investigated
Systematic search salt-pan species of Melophorus (as yet unnamed) resembles C. fortis more, and its congener M. bagoti less,
Synthetic approach in its path integration. The synthetic approach would benefit from comparing more species chosen to
address evolutionary questions.
This article is part of a Special Issue entitled: CO3 2013.
© 2013 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2. Thermophilic scavengers and their navigational toolkits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3. Precision in path integration in C. fortis and M. bagoti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4. C. fortis stays with path integration longer than does M. bagoti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5. M. bagoti is better than C. fortis in a task of visual associative learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6. Systematic searching: similarities and differences between species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
7. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

1. Introduction a red ant with long legs dashed quickly across the hot bitumen.
Eureka! Thus began the study of navigation, foraging, and learn-
On a January day in 2001, two of the authors (KC, RW) and ing in the thermophilic red honey ant Melophorus bagoti (Cheng
Sibylle Wehner pulled in at the parking lot at Simpson’s Gap, a gor- et al., 2009). More than a decade later, the study of M. bagoti
geous desert place in the West MacDonnell ranges ∼18 km west now complements research on the far more studied desert ants
of Alice Springs, Australia. As the group stepped out of the car, of North Africa, genus Cataglyphis (see Wehner, 2013, for a per-
sonal historical account), as well as desert ants in southern Africa,
genus Ocymyrmex (Wehner, 2003). We have learned something
∗ Corresponding author at: Department of Biological Sciences, Macquarie Univer- about its foraging ecology (Muser et al., 2005; Schultheiss and
sity, Sydney, NSW 2109, Australia. Tel.: +61 2 98508613; fax: +61 2 98509231. Nooten, 2013), its use of the surrounding panorama (Graham and
E-mail address: ken.cheng@mq.edu.au (K. Cheng). Cheng, 2009; Wystrach et al., 2011a, 2012), its route following

0376-6357/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.beproc.2013.10.001
52 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
(Kohler and Wehner, 2005; Narendra, 2007b; Sommer et al., 2008;
Wystrach et al., 2011b), and compared its performance with other
desert ants (Bühlmann et al., 2011; Schwarz and Cheng, 2010). The
study of Melophorus ants has now expanded to a species (as yet
unnamed) that lives on the salt pans of South Australia (Schultheiss
et al., 2012). Comparing ants in different habitats that fill a similar
ecological niche of a thermophilic scavenger has yielded interest-
ing similarities and differences that form the topic of the present
account.
Research on navigation in desert ants on different continents
reflects to a large extent Kamil’s (1987) synthetic approach to the
study of animal intelligence, especially in the comparisons between
species. The synthetic approach focuses on intelligent behaviour
that an animal needs to use in its natural life, much more so than
on laboratory tasks that experimenters can impose in standard-
ized conditions. Lab work is not eschewed, but often inspired and
informed by field work in the natural habitats of the animals. A
wider range of intelligent behaviour was called for in Kamil’s (1987)
paper, including numerical, timing, and spatial abilities, as well as
placing the animal’s behaviour in an appropriate ecological and
evolutionary context.
In Kamil’s own work on food-storing corvids for example,
the interest started with field observations suggesting excellent
spatial memory in the food-storing Clark’s nutcracker (Nucifraga
columbiana; Tomback, 1980), which stores and retrieves tens of
thousands of food caches over a winter (Vander Wall and Balda,
1977). Lab work inspired by such findings confirmed the role of
spatial memory (Kamil and Balda, 1985). Further comparative work
with a number of corvids showed that the Clark’s nutcracker, the
most prolific food-storer among North American corvids, performs
best on lab tasks of spatial memory, but did not stand out in a lab
task requiring memory for colour (Olson et al., 1995). Birds that are
more prolific food storers also remember caches stored in the lab for
longer (Bednekoff et al., 1997). More recent work has expanded to
compare corvids on other kinds of tasks, such as transitive inference
(Bond et al., 2010).
The work on desert ant navigation focuses on an important task
in their natural life, that of finding their way to food sources soli-
tarily – for desert ants are solitary foragers – and then finding their
way back home again after suitable prey has been found. Most of
the research takes place in the ants’ natural habitats, in the form of
field experiments, but lab work complements the enterprise when
it is needed, for example in examining the eyes (Schwarz et al.,
2011b) or the brains of ants (Stieb et al., 2010). The subjects of
study travel the scale of distance in the range of their normal trav-
els on foraging excursions, unlike, for example, the much smaller
and restricted arenas typically foisted on lab rodents in the study
of rodent navigation. Functional and mechanistic questions have
both been entertained in the course of the research. And with the
most recent comparisons across genera, questions of evolution are
beginning to be addressed about the navigational toolkit of these
animals. In our account, we will show that all desert ants possess a
basic navigational toolkit. Natural selection from living in different
habitats, however, has drawn different emphases and specializa-
tions. We discuss briefly the bases for such specializations. First,
however, we need to present the key players in the story.
2. Thermophilic scavengers and their navigational toolkits
The desert ants of the genera Cataglyphis (in North Africa),
Ocymyrmex (in southern Africa), and Melophorus (in Australia) all
fill the niche of thermophilic scavengers (Wehner, 1987; Wehner
and Wehner, 2011). These ants specialize in heat tolerance, with
one of our study species, M. bagoti, being the most heat tolerant
ant on its continent (Christian and Morton, 1992). This allows the
ants to forage in the heat of the day in summer, when it is too hot
for other foragers. M. bagoti, for example, shares a geographic niche
with other ants (of the genera Iridomyrmex, Camponotus, and oth-
ers), but when the ground temperature exceeds about 50 ◦ C, only
they are out foraging (Schultheiss and Nooten, 2013). They forage
solitarily, running on long legs (Sommer and Wehner, 2012) and
scavenging dead animal bits, largely arthropods (Wehner et al.,
1983). In ants that live in plant-rich habitats, plant materials are
also collected (Muser et al., 2005; Schultheiss and Nooten, 2013;
Schultheiss et al., 2012). One species, Cataglyphis floricola in south-
western Spain, even uses the latter in the form of petals as its main
food source (Cerdá et al., 1992).
Their navigational toolkit includes a number of what Wiener
et al. (2011) called spatial primitives, such as the encoding of com-
pass direction, based on both celestial and terrestrial cues, and
distance travelled or odometric measure. It is possible that the more
complex level of spatial constructs is also encoded. The major sys-
tems in the toolkit are generally recognized as path integration, use
of the visual panorama, and systematic searching (Wystrach et al.,
2013b), although a fourth system of backtracking has just been pro-
posed for M. bagoti (Wystrach et al., 2013b). Less well investigated
are the use of cues of other modalities, including vibrational, mag-
netic (Buehlmann et al., 2012) and olfactory ones (Buehlmann et al.,
2012; Steck et al., 2009; Wolf and Wehner, 2000), and a differ-
ent compass mediated by the ocelli (Schwarz et al., 2011a, 2011c)
rather than the dorsal rim area of the compound eyes (Wehner,
1994).
In path integration, the navigator keeps track of the straight-
line distance and direction that it has travelled from a starting
point (typically home), and uses the calculated vector for homing
(Wehner and Srinivasan, 2003). This system depends on registering
the direction in which the ant is travelling, the distance travelled
en route (odometry), and importantly, integrating the two kinds of
information, so that the traveller encodes which direction a partic-
ular step has moved in. The directional component is based largely,
but not solely, on a celestial compass derived from the pattern of
polarized light in the sky. Light is scattered in a systematic fashion
as it enters the Earth’s atmosphere (Wehner, 1994). A visual system
with appropriate receptors and processing systems can perceive a
pattern in the polarized light that indicates the direction of the
sun. The position of the sun itself is also used as a cue (Wehner and
Müller, 2006), as are spectral patterns (Wehner, 1994). The latter
refer to the distribution of different wavelengths of light reflected
by terrestrial objects as a function of the position of the sun. The
pattern of polarized light, spanning the entire sky, is not surpris-
ingly weighted more than the position of the sun (Wehner and
Müller, 2006). Odometry is dependent largely on a stride integra-
tor that functions like a step counter (Wittlinger et al., 2006, 2007),
although the pattern of optic flow beneath the ant is also used to
some small extent (Ronacher and Wehner, 1995). These studies
were all on the North African C. fortis.
In using the visual panorama, some visual cues in the surround-
ing panorama are used to determine a direction of travel. These
cues might encompass large segments or the entire visual surround
(Graham and Cheng, 2009; Wystrach et al., 2011a, 2012). Thus, the
skyline is used by M. bagoti as a terrestrial compass source (Graham
and Cheng, 2009). The skyline is a record of the heights of ter-
restrial objects in the visual panorama. Graham and Cheng (2009)
showed that desert ants would follow the dictates of an artificially
constructed skyline made of black cloth. The contour of the cloth
enclosure matched roughly the skyline heights measured at the
feeder to which the ants were trained to visit, with the elevations
matched at 15◦ intervals. Even large landmarks, obvious and promi-
nent to human observers, might not be used by the ants when the
rest of the visual panorama is unfamiliar (Wystrach et al., 2011a).
This and other results led to proposals that the entire panoramic
 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61 53

terrestrial scene guides ants as some kind of terrestrial visual com-
pass (Graham et al., 2010; Wystrach et al., 2012, 2013b). Different
models of how this is done have been proposed (e.g., Baddeley et al.,
2012; Möller, 2012; see Möller, 2012 for a brief overview), and this
topic forms an agenda for present research.
M. bagoti, at least, backtracks under appropriate conditions
(Wystrach et al., 2013b). The ants reverse the last direction in which
they were travelling, when (1) their vector based on path integra-
tion is near zero, (2) the current scene looks unfamiliar, and (3)
when they have just seen a familiar scene, that surrounding their
nest. According to Wystrach et al. (2013b), backtracking functions
to move the ant in the direction that is most likely to lead to some
familiar visual scene after sudden displacement, for instance by
wind. Thus, if a forager nearing her nest is blown off by wind, and
ends up somewhere unfamiliar, it is probable that she has gone
beyond the nest along the route that she had travelled on. Had she
been blown back along her route of travel, the scene should look
familiar.
Even with these strategies operating normally, the homing ant
might miss the nest because of inevitable errors of biological sys-
tems. We have often observed returning ants searching in the
vicinity of their nests for the entrance on a normal excursion. In
such cases, the ant resorts to a back up strategy of systematic
searching. In all systematic searches, the ant loops around the start-
ing point of the search in ever larger loops, often returning to the
starting point of search (Wehner and Srinivasan, 1981). But as we
explain later, searching in a familiar area, such as in the vicinity
of a nest or an oft-visited food site, appears to be a different pro-
cess from searching in unfamiliar territory (Schultheiss and Cheng,
2011, 2013; Schultheiss et al., 2013).
While we focus on comparing desert ants, other ants – and for
that matter other insects – share the major tools of the toolkit. Thus,
rainforest ants Gigantiops destructor integrate a path (Beugnon
et al., 2005) and follow signature routes (Macquart et al., 2006),
wood ants, Formica japonica, rely mainly on distant panoramic
cues, which can fully outcompete information provided by the path
integrator (Fukushi, 2001; Fukushi and Wehner, 2004), and night-
active bull ants Myrmecia pyriformis use a celestial compass and the
visual panorama for directional cues (Reid et al., 2011). Even trail-
laying ants learn to use the panorama (review: Collett and Zeil,
1998). And the navigation of honeybees, as well as other kinds of
learning, has been much studied (reviews: Avarguès-Weber et al.,
2011; Perry et al., 2013).
Different species of desert ants live in different habitats, and in
the comparisons we are about to present, we address how these dif-
ferences in habitats might have shaped their use of the navigational
toolkit. Some species live in bare salt pans (Fig. 1A and C), in which
natural terrestrial visual cues are largely absent. These include
the much studied North African ant C. fortis (Fig. 1A, taken from
Wehner, 2012), and a ‘new’ (taxonomically speaking) Melophorus
ant that has yet to be named (which at present we will call Melopho-
rus sp.; see Schultheiss et al., 2012 and Fig. 1C). Other ants, such

Fig. 1. Different habitats where desert ants live, with photos of the ants. (A) Chott el Djerid, a typical North African salt pan habitat for C. fortis. Reprinted from Wehner
(2012,). (B) Near Alice Springs Airport, Alice Springs, Australia, a typical habitat for M. bagoti. C. Eucolo Creek at Island Lagoon, South Australia, a typical habitat for the salt-pan
dwelling Melophorus sp.
Photo credits: R. Wehner (A), Eliza J. T. Middleton (B, habitat), S. Schwarz (B, ant, and C, habitat and ant).
54 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
Fig. 2. A field near Alice Springs Airport that has been largely cleared of vegeta-
tion. Bühlmann et al. (2011) compared a nest of Melophorus bagoti found there with
Cataglyphis fortis ants.
Photo by Eliza J. T. Middleton.
as M. bagoti, live in visually cluttered habitats (Fig. 1B). In bare
habitats, path integration is at a premium because useful terres-
trial visual cues are largely absent. We might expect ants in such
habitats to specialize in path integration. In a cluttered habitat on
the other hand, terrestrial visual cues might be used to form stereo-
typical routes (e.g., Kohler and Wehner, 2005; Mangan and Webb,
2012; Wystrach et al., 2011b). We may expect ants in such habi-
tats to rely more on terrestrial visual cues for navigation, and to be
better able to learn visually based tasks. In what follows, most of
the comparisons revolve around C. fortis (a North African salt-pan
ant) and M. bagoti (an Australian ant in a cluttered habitat), because
these species have been studied the most.
3. Precision in path integration in C. fortis and M. bagoti
Most of the comparisons of path integration have been between
C. fortis and M. bagoti. The data to be described come largely from
experimental set ups in which a feeder at a constant location provi-
sions food for the foragers. After foragers have made regular visits
to the feeder for a while, picking up some food to take home on each
visit, a forager is tested after she grabs a bit of food in her mandibles
at the feeder. Such a forager with booty is motivated to head home.
Because C. fortis lives in a bare habitat, our predictions were
that it would be more precise in path integration than M. bagoti.
Because M. bagoti lives in a visually cluttered landscape and can
rely on terrestrial visual cues for guidance for most of its travel,
finding a suitable nest for comparing with C. fortis was a challenge.
An artificially cleared field near the Alice Springs airport (Fig. 2)
was the most ‘level playing field’ we could find. This cleared plot
with quite a uniform ring of trees around it was used for launching
hot-air balloons, and on it one nest of M. bagoti could be located.
We could thus test the precision of path integration of this nest as
they travel over terrain resembling what faces the typical C. fortis
forager (Bühlmann et al., 2011).
C. fortis and balloon-field M. bagoti ants, each in their habitat,
were trained to visit a feeder at a constant location, and then dis-
placed to a distant location with some food in their mandibles. The
(circular) scatter in their initial heading direction provided a mea-
sure of directional precision. Without nearby guiding visual cues,
and little by way of distant panorama, the ants had to rely mostly
on their sky compass (Wehner, 1994; Wehner and Müller, 2006)
to strike an initial heading. On this measure, however, no signif-
icant differences between species was evident. This was the case
0.4
6 m outbound distance
SD of group / M of group
12 m outbound distance
0.3
0.2
0.1
0
C. fortis M. bagoti
Fig. 3. Precision in odometry of Cataglyphis fortis and Melophorus bagoti ants. Ants
travelled in a narrow channel (10 cm × 10 cm) for 6 m or 12 m to a feeder. They then
returned in a different channel with a bit of food, and the first point at which an
ant turned back and initiated back-and-forth searching was noted. The dependent
measure is the standard deviation of the group divided by the mean of the group. A
number of similar conditions were pooled.
Data on C. fortis came from Cheng et al. (2006). Data on M. bagoti from Narendra
et al. (2007a).
whether artificial landmarks (black cylinders) were provided along
the nest-feeder route or not.
These results do not support our predictions. In retrospect, M.
bagoti might well need precision in using their sky compass for
more than path integration. Wystrach et al. (2012), in modelling
their data, suggested that panoramic visual patterns would benefit
from being aligned with respect to the sky compass in some situ-
ations. The sky compass, serving as a key component of the path
integration system, might also function in learning walks of ants.
Before they begin foraging, desert ants make short walks in the
area of their nest, striking out in different directions and returning
to the nest in loops (C. bicolor: Wehner et al., 2004; Ocymyrmex
robustior: Müller and Wehner, 2010). O. robustior ‘learners’ turn to
face their invisible nest at regular intervals during learning walks,
and path integration is said to “provide a scaffold” for learning the
view towards the nest (Müller and Wehner, 2010). The sky com-
pass might also help ants recapitulating a familiar route to orient
in the correct direction along a route. Thus, the sky compass might
serve both path integration and panorama-based navigation in M.
bagoti, and precision in the sky compass might be just as important
for M. bagoti as for C. fortis.
Another aspect of path integration is odometry, estimating the
distance travelled on the outbound trip. On this front, we were
not limited to the balloon-field M. bagoti nest because narrow
channels could be constructed that largely excluded any views
outside of the channel. Ants of each species travelled a fixed dis-
tance to a feeder in channels of the same construction. Ants with
food were then displaced to a long channel, and their initial run
in the homeward direction before they turned back and started a
back-and-forth search pattern constituted an odometric estimate
(C. fortis: Cheng et al., 2006; Sommer and Wehner, 2004; M. bagoti:
Narendra et al., 2007a). On average, both species travelled roughly
the outbound distance in channels before starting to search, but
the scatter across animals was significantly larger in M. bagoti, thus
showing less odometric precision (Fig. 3). The species difference
was confirmed statistically by a (conservative) test of differences
in variance between multiple groups (O’Brien’s test), and was found
at both 6 m and 12 m outbound distance. The results confirm that
C. fortis is better at this aspect of path integration.
Behind this species difference in odometric performance is one
commonality: both species perform as well as they do from trial
1, and no improvement with practice is found. Both the studies on
 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
C. fortis (Cheng et al., 2006) and M. bagoti (Narendra et al., 2007a)
trained different groups of ants with different numbers of trials
(visits to the feeder) before testing, and no significant differences
were found between these groups in either species. Repeated train-
ing does not improve path integration on the natural open field
either (Merkle and Wehner, 2009b). This makes some functional
sense, as the ant needs to be good enough to return to the nest on
its first foraging trip or it will perish.
Perhaps most telling are comparisons of performance in path
integration on the open field, comparing the balloon-field M. bagoti
nest with C. fortis foragers. In one condition, Bühlmann et al. (2011)
let ants at a feeder run home with food on the training field
itself, with no displacement of the animals. The situation matched
the ants’ training runs between nest and feeder completely. The
authors observed that M. bagoti ants were far worse at finding their
nest. This is in good part because the scatter in their odometric
estimates (SD) was 4.7 times larger than the scatter in C. fortis. Stri-
kingly, most C. fortis ants (18/20) found their way directly to their
nest without exhibiting looping search patterns at all, whereas only
2/18 M. bagoti ants managed to find their nest without having to
search in loops for it. These differences likewise confirm that C.
fortis is better at path integration than M. bagoti.
Intriguing and telling in the comparisons of odometric precision
are recent data from Melophorus sp. (work in prep.), the Melophorus
ant that lives on barren salt pans (Fig. 1C). Measured at 4 m out-
bound distance over open field (i.e., not in channels), the variation
in odometric estimates of Melophorus sp., adjusted by the outbound
distance, resembled that found in C. fortis more than that found in
M. bagoti. Melophorus sp. lives in a habitat with few if any usable ter-
restrial visual cues for navigation and must rely primarily on path
integration. A species signature is evident, but the line is drawn
between ecological habitats and not between genera.
4. C. fortis stays with path integration longer than does M.
bagoti
In a bare habitat, path integration and systematic search are
at a premium because the navigator lacks usable visual cues in the
panorama. One might expect such a navigator to stick with the vec-
tor estimated from path integration for its entirety before starting
to search systematically. When a route is rich in visual cues on the
other hand, and the navigator has been displaced off its familiar
route, the option of searching for familiar route cues might com-
pete with the strategy of path integration. Such a navigator might
be expected to turn to systematic searching before the calculated
vector has been run off. These expectations have been confirmed
in comparing cluttered-habitat ants (M. bagoti) with salt-pan ants
(both C. fortis and the salt-pan Melophorus sp.).
Considering first comparisons between C. fortis and M. bagoti,
Bühlmann et al. (2011) trained ants to visit a feeder regularly, and
then took trained ants one at a time from the feeder to a distant test
site. They measured the length of the initial homing run on the open
field on the test, and found that under a variety of conditions, M.
bagoti from the nest on the balloon-field ran off a shorter distance
than did C. fortis before initiating systematic search, some 10–20%
shorter. This was the case when landmarks were provided en route
in both training and test situations, when no additional landmarks
were provided, or when landmarks en route were provided in train-
ing but absent on test. In addition, when a nest-defining landmark is
presented at an earlier stage of the ant’s homebound trip, i.e., before
the fictive position of the nest is reached, M. bagoti gets attracted
by this landmark much more readily than does C. fortis (Bregy et al.,
2008; Bühlmann et al., 2011). Interestingly, initial data on the salt-
pan Melophorus sp. show that they behave more like C. fortis, which
inhabit the same kind of bare habitat, than like their congener M.
55
bagoti (Fig. 4, from Schultheiss et al., 2012). When trained and tested
on the open field, Melophorus sp. ran the full outbound distance
before initiating search. A species signature is again evident, with
the line again drawn between ecological habitats and not genera.
In addition to a species signature is what we might call a signa-
ture of environmental mismatch. In C. fortis, if artificial landmarks
along the route were removed on a test, they run slightly but sig-
nificantly less than the full outbound distance (Bühlmann et al.,
2011). In M. bagoti, the proportion of the outbound distance that
they run before initiating search varies across conditions of testing
(Bühlmann et al., 2011; Cheng et al., 2012; Narendra, 2007a, 2007b).
As already mentioned, if trained and tested in narrow channels that
blocked the visual surround, M. bagoti run the full outbound dis-
tance back (Narendra, 2007a; Narendra et al., 2007a, 2008). When
displaced from one open-field location to another open-field loca-
tion, they run some ∼75–90% of the outbound distance on average
(Bühlmann et al., 2011; Cheng et al., 2012). And when displaced
from one of their typical cluttered habitats to an unfamiliar loca-
tion, they run ∼45–50% of the outbound distance on average (Cheng
et al., 2012; Narendra, 2007b).
Species differences are manifested not only in the initiation
point of systematic search, but also before that, while the ants
are running off their vector or a part thereof. We have observed
that when M. bagoti ants are running home on unfamiliar terri-
tory, or even if they are a small distance away from their usual
route, their paths meander, compared with their largely straight
runs on a familiar route. In recent studies, the straightness of the
path that they run has been quantified (Cheng et al., 2012; Schwarz
and Wystrach, 2011; Wystrach et al., 2011b). A ‘dose response’
curve has been found in that the more the mismatch between
the scene at the release point and the scene on a familiar route,
the more the ants meander, and the less straight their paths. In
Cheng et al. (2012), ants from 3 nests in typical cluttered habitats
were displaced from a feeder to the open balloon-launching field.
Their initial paths were less straight than those of the nest living
on the balloon-launching field. Comparing M. bagoti with C. fortis
ants under similar conditions (that is, using the M. bagoti nest on
the balloon-launching field), Bühlmann et al. (2011) found that the
paths of C. fortis were straighter. This was the case whether experi-
mental landmarks were provided along the route or not in training
and testing. We interpret the meandering as a sign of conflict with
the strategy of systematic searching, with systematic search even-
tually winning out when the ant turns and makes looping searches.
In this regard too, both a species signature and a signature of envi-
ronmental mismatch are evident.
In sum, under similar conditions, C. fortis runs a longer propor-
tion of their calculated vector than M. bagoti does before engaging
in systematic search. And the runs of C. fortis are straighter. Both
differences indicate a higher propensity to rely on path integration
in C. fortis. The one test performed so far on the salt-pan Melopho-
rus sp. shows that they, like C. fortis, run off the full vector before
starting systematic search.
5. M. bagoti is better than C. fortis in a task of visual
associative learning
An ant living in a visually cluttered habitat is expected to rely
more on terrestrial visual cues. It might be expected to do better
on a visually based learning task. One study compared M. bagoti
and C. fortis explicitly in the same visual learning task (Schwarz
and Cheng, 2010). The ants were trained and tested in channels.
After coming out to a feeder, they were trained to choose one side
of a bifurcation along the way home, in a version of a Y-maze. One
side was coloured black, while the other side was white, and the
correct choice was the black side. The task was mastered within
56 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61

Fig. 4. Comparison of the average homing distance between different desert ant species after displacement from a familiar feeder to an unfamiliar test field; the position
of the fictive nest is marked by a small cross. Melophorus sp. and Cataglyphis fortis inhabit similar visual habitats and exhibit similar homing paths. On average, they run off
their entire vector before they start to search for the fictive nest. M. bagoti ants differ in their homing behaviour depending on whether they are trained in visually open or
cluttered surroundings. The ants were tested on an open field. Only when living and being trained in open surroundings do homing paths of M. bagoti resemble those of
the other two species, although the ants still run off a shorter proportion of their path-integration vector. In contrast, when taken from their typical cluttered habitats M.
bagoti ants search on average much earlier for their fictive nest. Paths of C. fortis (outbound distance of 10 m) are adapted with permission from the Journal of Experimental
Biology (Bühlmann et al., 2011), paths of M. bagoti (outbound distance of 12 m) are from Cheng et al. (2012), and paths of Melophorus sp. (outbound distance of 3 m) are from
Schultheiss et al. (2012). For purposes of clarity, most groups show paths that were selected randomly from a larger number of paths (Melophorus sp.: n = 10 of 36, C. fortis:
n = 23 of 23, ‘open’ M. bagoti: n = 8 of 20, ‘cluttered’ M. bagoti: n = 12 of 25).
Reprinted with permission from Schultheiss et al. (2012) (http://www.publish.csiro.au/nid/90/paper/ZO12096.htm, Fig. 9).

30 trials by a slim majority of M. bagoti ants, all of which contin-
ued successfully on subsequent tests. In sharp contrast, only 1 of 60
C. fortis foragers succeeded within 30 trials of training and main-
tained the success on the subsequent tests. The study included a
rare C. fortis nest located in a visually cluttered habitat, as well
as a nest provided with artificial landmarks, neither of which had
any successful ants. This one data set supported our comparative
predictions.
On the other hand, all desert ants can learn to use a set of exper-
imental landmarks to help locate their nest (C. fortis: Ziegler and
Wehner, 1997; C. bicolor: Wehner et al., 1996; Wehner and Räber,
1979; M. bagoti: Narendra et al., 2007b). These studies used 2 to 4
black cylinders, in the middle of which was located the nest. Ants
were displaced to a distant test field with the landmarks set up
for testing. The results from these studies, however, are not com-
parable. Ziegler and Wehner (1997) reported training C. fortis ants
for at least 2 days, and thus the exact amount of training was not
specified. Narendra et al. (2007b) trained M. bagoti ants for up to
15 trials, spread unsystematically over days depending on how fre-
quently the ants returned to the feeder. Learning was best after 15
training trials spread over more than 2 days. Test performance was
also not comparable: M. bagoti ants, having been displaced to a dis-
tant test site, faced a different panoramic surround from that found
at their nest. The significance of panoramic visual information was
not fully appreciated at that time.
Although species differences in learning to use a set of experi-
mentally provided visual cues (landmarks) cannot be ascertained
at this point, underlying similarities between all desert ant species
in this tool can be pointed out once the goal location with respect
to these cues has been learned. Thus, in both C. fortis (Ziegler and
Wehner, 1997) and M. bagoti (Narendra et al., 2007b), these visually
based memories last a long time, as long a delay as experimenters
cared to impose, probably meaning that the memories last the life-
time of the ant. Doubling the distance of each landmark from the
fictive target (the place where the nest would be had the ants not
been displaced on a test) leads to a breakdown in searching in C.
bicolor (Wehner et al., 1996; Wehner and Räber, 1979) and in M.
bagoti (Narendra et al., 2007b). If the landmarks double their dis-
tance and also double their size, however, both these species then
search at the centre of the array. Some matching of the projected
size of these objects, beyond matching their mean azimuths, seems
to matter to these ants. Important theoretical progress has been
made on this front (Baddeley et al., 2012; Möller, 2012; Wystrach
et al., 2013a), but much theoretical and empirical research remain
to be done to specify what panoramic characteristics are used in
matching.
Another similarity across species is the propensity to follow a
stereotypical route when visual cues along the route are available.
Melophorus bagoti’s major mode of travel is following a route that
is similar from one trip to the next, as demonstrated in a number
of studies now (Kohler and Wehner, 2005; Sommer et al., 2008;
Wehner et al., 2006; Wystrach et al., 2011b). Ants from a C. for-
tis nest found in an unusually cluttered habitat also recapitulate
stereotypical routes (Wehner et al., 1996). When displaced near
their nest back to a part of their route, these ants readily run off
their route once more. Ants of other Cataglyphis species living in
 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61 57

cluttered habitats also recapitulate stereotypical routes (C. velox:
Mangan and Webb, 2012).
6. Systematic searching: similarities and differences
between species
Sometimes a forager may fail to find a specific location it has
visited before, for example the nest entrance or a food location. In
this case it engages a ‘backup’ mechanism, the systematic search
behaviour. As mentioned, the searching ant moves around in ever
expanding loops, while re-visiting the vicinity where it expects to
find the target repeatedly. This basic pattern of expanding loops
is found in every species of desert ant studied so far (C. fortis, C.
bicolor, M. bagoti, Melophorus sp., Fig. 5). The overall spread of these
search paths (the area that is covered) is flexible and depends on
the context of the search and the experience of the forager, and can
be used as a proxy for search accuracy. In ants searching for their
nest, those having run off a longer vector before being displaced
for a test engage in a larger spread in the subsequent nest search,
probably an adaptation to an increase in the navigational error of
path integration (C. fortis: Merkle et al., 2006; M. bagoti: Schultheiss
and Cheng, 2011). Further work on M. bagoti has revealed a similar
pattern in relation to navigation in a visually familiar environment.
When these ants search for their nest in the vicinity of their actual
nest, their precision is greater (their search spread is smaller) in
a visually rich environment with experimentally added landmarks
around their nest than in a normal ‘unenriched’ visual environment
(Schultheiss et al., 2013).
In outbound M. bagoti foragers searching for a known food
source, the size of the search can also depend on the type of food
that the ant is looking for. Searches for protein food are larger than
searches for carbohydrate food, which matches the different distri-
butions of naturally occurring food items of these kinds (Schultheiss
and Cheng, 2013). The size of food searches in C. fortis is additionally
influenced by productivity and familiarity of food sites (Bolek et al.,
2012). It remains to be seen whether the other species of desert ants
respond to variations in food type, quantity and familiarity in a sim-
ilar fashion. In other forcimine ants (Formica pallidefulva, formerly
F. schaufussi), however, similar differences in patterns of search-
ing for carbohydrates vs. protein have been found (Fourcassié and
Traniello, 1994; Traniello et al., 1992).
Another line of inquiry has focussed on the movement strategies
of searching desert ants, and what they can tell us about the mech-
anistic basis of searching behaviour. For this, an animal’s trajectory
is discretised into a series of straight-line movements (segments),
connected by instances of re-orientation, or turns. In M. bagoti, the
distribution of search segment lengths displays an exponential dis-
tribution of movement lengths with lots of short segments and
very few long segments when searching in a familiar visual context,
either for a single food source (Schultheiss and Cheng, 2013) or for
the nest entrance (Schultheiss et al., 2013). This pattern changes to
a ‘two-process’ strategy, a mixture of two exponentials, when the
ants are displaced into a distant test-field and search for their nest
in an unfamiliar visual environment (Schultheiss and Cheng, 2011).
The picture is different in the salt-pan dwelling Melophorus sp.,
which displays a single-exponential strategy (similar to a Brown-
ian walk) when searching for the nest in a distant test-field (work
in prep.). However, the visual appearance of the distant test-field
in this scenario (the open, featureless salt-pan) will look almost
identical to the familiar nest environment, even though the ant has
never visited this place before. Based purely on visual input, the
searching Melophorus sp. foragers may not be able to discriminate
between the two. In the Namibian desert ant Ocymyrmex robustior
even a single nearby landmark can concentrate search movements
and reduce search time dramatically (Wehner and Müller, 2010).
So far, these data seem to indicate that visual scene familiarity may

Fig. 5. Expanding loops in the search pattern of desert ants: (A) Cataglyphis bicolor, (B) Melophorus bagoti. Each panel shows one example path from a single searching ant.
The distance of a searching ant from the origin is plotted against path length, showing that the path has a looping structure (the ant repeatedly returns close to the origin),
and that loop size gradually increases. Panel A is adapted from Wehner and Srinivasan (1981) (Fig. 8) with kind permission from Springer Science+Business Media, panel B
shows data from Schultheiss and Cheng (2011), but not presented in that work.
58 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
play an important role in shaping the search strategies of desert
ants, and the notion of familiarity plays a key explanatory role in a
recent account (Wystrach et al., 2013b), to which we will return in
considering differences across species. Yet, the distribution of seg-
ment lengths in the search strategies of Cataglyphis ants remains to
be studied.
Intriguingly, the exponential search strategies of desert ants are
distinct from the strategies of another well-studied hymenopteran,
honeybees Apis mellifera. In similar contexts, i.e., for nest or food
searches in unfamiliar environments, the search paths of bees have
scale-free power-law properties, consistent with the execution of
so-called Lévy walks (Reynolds et al., 2007a, 2007b). In Lévy walks,
the distribution of segment lengths (l) obeys a power function,
distributed as l− , with the exponent ␮ around 2 to be optimal.
Compared to exponential strategies, these have a much “heavier”
tail, with a higher frequency of very long movement lengths. These
Lévy searches are optimal for finding single targets in unfamiliar
environments (Reynolds, 2008), so why are they seen in bees but
not in ants? One possibility is that a property of bees’ odometric
system is conducive to producing Lévy search patterns. In bees,
odometric errors are proportional to the distance being estimated
(Cheng et al., 1999). Theoretically, Lévy walks could be derived
from this property (Reynolds et al., 2013). In desert ants, odo-
metric errors are not proportional (Sommer and Wehner, 2004),
and Lévy walks cannot be derived. It is therefore possible that the
double-exponential strategy of M. bagoti is a way of approximating
a scale-free Lévy walk in unfamiliar environments, which would be
the optimal strategy (Reynolds et al., 2014).
7. Discussion
While the study of desert ant navigation has had a history going
back more than a century, at least to Santschi’s (1911, 1913) work
(for a biographical account of this almost forgotten scientist and his
work, see Wehner, 1990), the systematic comparison of desert ants
living in different habitats has had a far shorter history. What we
have gathered together here supports predictions that ants living
in different habitats would specialize in different aspects of navi-
gation. All desert ants, however, seem to share a basic navigational
toolkit of spatial primitives (Wiener et al., 2011) consisting at least
of path integration, visually guided navigation based on panoramic
terrestrial cues, and systematic search. It is how they weight these
different tools that seems to differ according to ecological habitat.
Based on some recent models of insect navigation, we can spec-
ulate on some of what the adaptive specializations consist in. Far
more comparisons are necessary, however, for a synthetic approach
to desert ant navigation, or to expand further to ants in general
or even to hymenopterans or insects as a group. In fact, many
aspects of navigation are shared by all terrestrial navigators, includ-
ing the sun, visual panorama, and the laws of geometry governing
the relations between paths of travel and points on the surface of
the Earth. This suggests the likelihood of convergent mechanisms,
and even deep homologies (for example, with vertebrate animals)
are not out of the question. We elaborate on these points in this
discussion.
Based on the navigational cues available in different habitats,
we predicted that ants living in visually bare habitats would spe-
cialize more in path integration, while ants living in visually rich
habitats would specialize more in navigation based on terrestrial
visual cues. A number of differences reviewed here confirm these
expectations. Cataglyphis fortis, living in a bare salt-pan habitat, per-
forms better in various ways in path integration than does M. bagoti,
living in a visually cluttered habitat. Compared with M. bagoti, C.
fortis is more precise in odometry, estimating the distance trav-
elled on a trip (Cheng et al., 2006; Narendra et al., 2007a), and
better at finding its nest by using path integration, without having
to resort to searching (Bühlmann et al., 2011). Cataglyphis fortis also
runs off a longer portion of the estimated vector before resorting
to searching, in comparison with M. bagoti under various condi-
tions (Bühlmann et al., 2011). Importantly, M. bagoti’s congener
Melophorus sp., which lives in a bare salt-pan habitat like C. fortis,
looks more like C. fortis than like M. bagoti in its path integration
(Schultheiss et al., 2012; work in prep.). In the use of visual cues,
explicit comparisons are limited to date, and far more needs to be
done. In one task of visual associative learning (choosing a black
side in a Y-maze), M. bagoti performed much better than C. fortis
(Schwarz and Cheng, 2010).
These differences between desert ants living in different habi-
tats should be seen against the backdrop of a common navigational
toolkit likely extending well beyond desert ants to other ants, other
hymenopterans, and probably most insects and many arthropods. If
ecological habitats have driven the evolution of navigational strate-
gies, it is in specialization in relying more on one tool or the other,
and probably not in the fabrication of a brand new ‘tool’. Thus,
path integration, navigation based on the visual panorama, and
systematic searching are found in all these desert ants. While M.
bagoti may not be as precise as C. fortis or Melophorus sp., it has
been clearly demonstrated to integrate a path travelled (Narendra,
2007a, 2007b; Wehner et al., 2006). And while C. fortis may not
perform as well as M. bagoti in a visual associative task, many stud-
ies have shown that it uses experimentally provided landmarks
to locate a target (e.g., Ziegler and Wehner, 1997). And C. fortis, C.
bicolor, C. albicans, M. bagoti, and Melophorus sp. all resort to sys-
tematic search strategies when other navigational strategies fail
(Merkle et al., 2006; Merkle and Wehner, 2009a; Schultheiss and
Cheng, 2011, 2013; Schultheiss et al., 2013; Wehner and Srinivasan,
1981; work in prep.).
Recent models of insect navigation (Collett, 2012; Cruse and
Wehner, 2011; Wystrach et al., 2013b) suggest how the adaptive
specialization of relying more or less on a particular navigational
system might be realized neurobiologically, and this in turn sug-
gests how natural selection or ontogenetic experience might find
a locus for their effects. The basic idea in these models is that the
various systems or modules are not switched on one at a time, as if
a switch can only connect with one system at a time. Rather, they
are all in operation simultaneously, converging on a central sum-
mator that sums the inputs in a weighted basis. The operation of
one system or module means that it has a high weight while the
other modules have near zero weights. Different conditions might
dictate which system is assigned more weight. Thus, Wystrach et al.
(2013b) posit explicitly that systematic search has a continuous low
weight. That makes this module a fall-back option, running when
current conditions make the weights for all the other modules very
low. The path integration system is weighted by the magnitude of
the scalar component (vector length), while the influence of visual
matching of terrestrial cues is weighted by the familiarity of the
current view.
A signature of the weighted average scheme, whatever its form
or underlying specifications, is that sometimes, more than one sys-
tem can be witnessed in operation. Empirically, this means that the
animal follows a dictate that is an average of the dictates of different
systems. Reid et al. (2011) set the pattern of polarized light and ter-
restrial panorama in conflict by using a polarizing filter with which
they can sysetmatically manipulate the polarized light. They found
that night active bull ants Myrmecia pyriformis averaged the dictates
of polarized light and terrestrial panorama in such conflicts. Collett
(2012) displaced C. fortis ants on their familiar outbound journey
to a feeder, another way of setting the dictates of path integration
in conflict with those of the terrestrial visual panorama. He found
that the ants struck a course intermediate between the dictates of
those two systems. We have found similar intermediate courses in
 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
homebound M. bagoti ants in work in preparation. And we have
also interpreted the meandering during path integration or view
based matching as a partial tendency for systematic searching (see
Section 4).
A scheme of weighted averaging points to a locus, the weights,
on which both natural selection and ontogenetic experience may
act, at least in shaping the propensity to rely more on one system
rather than another. For example to break off path integration ear-
lier or later and switch to searching, the threshold for the point
at which searching dominates over following path integration may
be changed by adjusting weights appropriately. Species differences
might consist in assigning more weight, or potentiating the inputs
of one system to the summator. Likewise, experience in navi-
gating under different circumstances might adjust those weights.
Whether such a scheme might also contribute to differences in
performance at odometry or visual associative learning, for exam-
ple by affecting attentional or motivational factors, remains to be
determined. It would be interesting to investigate whether such a
weighting scheme might apply to other taxa, such as the propensity
for some food-storing birds to rely on spatial cues for remember-
ing a location (Brodbeck, 1994; Brodbeck and Shettleworth, 1995).
Schemes of weighted averaging have certainly been proposed for
vertebrate animals in navigation (Cheng et al., 2006; Collett et al.,
1986). It would also be exciting to unravel the neurobiological and
genomic bases of such a plausible computational scheme.
Systematic search is considered a ‘fall back’ strategy when other
navigational strategies fail. As such, some common characteristics
across species may be expected. For example, one ideal strat-
egy that has been discussed (Reynolds et al., 2013; Wehner and
Srinivasan, 1981) is an expanding spiral pattern. If the spiral is
perfectly executed, and perception of the target when one comes
near it is fail-safe, this is the most efficient search strategy pos-
sible. Of course, the antecedent conditions following the “if” are
not realized in any error-prone biological system, and no animal
has shown a spiral systematic search pattern (but see the “hid-
den spiral” in Müller and Wehner, 1994). The expanding looping
searches cater to error-prone searching found in animals, and can
be found in other taxa than ants (e.g., isopods: Hoffmann, 1983a,
1983b, 1985a, 1985b). On the other hand, the target of search may
be different in different species and search conditions, perhaps gen-
erating differences in systematic search strategies. When looking
for a nest or a food site in a visually bare habitat, the searcher needs
to locate the exact spot of the target. But when looking for a target
in an unfamiliar but visually cluttered habitat, finding some famil-
iar scenery is usually good enough for finding the target (Wystrach
et al., 2012). Once in familiar territory, reliable route strategies can
usually deliver. Further comparisons of systematic searching can
be fruitful in a synthetic approach to desert ant navigation.
Indeed, the comparisons that we have presented need to be
expanded. Mostly, we have single comparisons between a pair of
species, in a sense amounting to n = 1. In the near future, paral-
lel comparisons between Cataglyphis species on the one hand, and
between Melophorus species on the other hand, would expand the
comparative base, and also provide some ‘phylogenetic control’, in
that pairs (or more) of congeners may be compared. Recently, not
only has work begun on a salt-pan Melophorus species, but also on
a Cataglyphis species, C. velox, living in a visually cluttered Spanish
habitat resembling M. bagoti’s mix of tussocks, bushes, and trees
(Mangan and Webb, 2012; see their Fig. 1). Like M. bagoti (Kohler
and Wehner, 2005; Sommer et al., 2008), C. velox relies a lot on
stereotypical routes (Mangan and Webb, 2012). Cataglyphis velox
spontaneously forms multiple routes. Whether it is more like M.
bagoti in other regards than like C. fortis remains to be investigated.
Cataglyphis bicolor, in former years much studied before its habitat
was encroached upon by economic development (Wehner, 2013),
is also often found in visually cluttered habitats.
59
Looking further afield, a far larger crop of species needs to be
examined to unravel some evolutionary questions about navigation
in desert ants. Large comparative studies requiring the cooperation
of many teams, as trumpeted in a recent manifesto by MacLean
and numerous co-authors (2012), are likely necessary to trace the
origins of basic navigational abilities such as path integration or
systematic search strategies. These abilities, along with other basic
forms of learning, may be widespread among arthropods, or per-
haps inherited from a common ancestor. The scientific world is
biased against reporting negative data, but in tracing the origins
of learning and cognitive abilities, it is absolutely necessary to doc-
ument both negative and rudimentary cases of the ability (Perry
et al., 2013).
Any terrestrial animal navigator (or even a mechanical robot)
faces a set of frequently encountered problems. Given a particular
outbound path, the direct homebound path is fully determined by
the laws of geometry as applied to a limited region of the Earth’s
surface. The visual panorama changes in predictable ways as one
travels: objects that one approaches take up more of the view while
objects that one moves away from take up less of the view. The
sun moves in the same way across the day for all. And so on. This
suggests the likelihood of convergent mechanisms or at least simi-
larities across the navigational toolkits of diverse taxa, for example
between insects and vertebrate animals (Cheng, 2012; Wiener et al.,
2011; Wystrach and Graham, 2012). Recently, a deep homology
between parts of arthropod and vertebrate brains (central com-
plex in arthropods and basal ganglia in vertebrates) has even been
suggested (Strausfeld and Hirth, 2013). Deep homologies in naviga-
tional mechanisms are not out of the question. A fully comparative
approach (Shettleworth, 2010) should be an integral part of the
synthetic approach to animal navigation.
Acknowledgements
We would like to thank Sibylle Wehner for all her support in
launching the research on both Melophorus bagoti and the salt-
pan Melophorus sp., as well as her help on Cataglyphis research
over the years. Funding for the desert ant research reported here
has come from the Australian Research Council (DP0770300 and
DP110100608 to KC and RW), the Swiss National Science Founda-
tion (3675/1-1 to RW), and Macquarie University in supporting PS,
SS, and AW.
References
Avarguès-Weber, A., Deisig, N., Giurfa, M., 2011. Visual cognition in social insects.
Ann. Rev. Entomol. 56, 423–443.
Baddeley, B., Graham, P., Husbands, P., Philippides, A., 2012. A model of ant route
navigation driven by scene familiarity. PLoS Comp. Biol. 8, e1002336.
Bednekoff, P.A., Balda, R.P., Kamil, A.C., Hile, A.G., 1997. Long-term spatial memory
in four seed-caching corvid species. Anim. Behav. 53, 335–341.
Beugnon, G., Lachaud, J.-P., Chagné, P., 2005. Use of long-term stored vector infor-
mation in the neotropical ant Gigantiops destructor. J. Insect Behav. 18, 415–432.
Bolek, S., Wittlinger, M., Wolf, H., 2012. Establishing food site vectors in desert ants.
J. Exp. Biol. 215, 653–656.
Bond, A.B., Wei, C.A., Kamil, A.C., 2010. Cognitive representation in transitive infer-
ence: a comparison of four corvid species. Behav. Proc. 85, 283–292.
Bregy, P., Sommer, S., Wehner, R., 2008. Nest-mark orientation versus vector navi-
gation in desert ants. J. Exp. Biol. 211, 1868–1873.
Brodbeck, D.R., 1994. Memory for spatial and local cues: a comparison of a storing
and a nonstoring species. Anim. Learn. Behav. 22, 119–133.
Brodbeck, D.R., Shettleworth, S.J., 1995. Matching location and color of a compound
stimulus: comparison of a food-storing and a nonstoring bird species. J. Exp.
Psychol.: Anim. Behav. Process. 21, 64–77.
Buehlmann, C., Hansson, B.S., Knaden, M., 2012. Desert ants learn vibration and
magnetic landmarks. PLoS One 7, e33117.
Bühlmann, C., Cheng, K., Wehner, R., 2011. Vector-based and landmark-guided
navigation in desert ants inhabiting landmark-free and landmark-rich environ-
ments. J. Exp. Biol. 214, 2845–2853.
Cerdá, X., Retana, J., Carpintero, S., Cros, S., 1992. Petals as the main resource col-
lected by the ant Cataglyphis floricola (Hymenoptera Formicidae). Sociobiology
20, 315–319.
60 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
Cheng, K., 2012. How to navigate without maps: the power of taxon-like navigation
in ants. Comp. Cogn. Behav. Rev. 7, 1–22.
Cheng, K., Middleton, E.J.T., Wehner, R., 2012. Vector-based and landmark-guided
navigation in desert ants of the same species inhabiting landmark-free and
landmark-rich environments. J. Exp. Biol. 215, 3169–3174.
Cheng, K., Narendra, A., Sommer, S., Wehner, R., 2009. Traveling in clutter: navigation
in the Central Australian desert ant Melophorus bagoti. Behav. Proc. 80, 261–268.
Cheng, K., Narendra, A., Wehner, R., 2006. Behavioral ecology of odometric memories
in desert ants: acquisition, retention, and integration. Behav. Ecol. 17, 227–235.
Cheng, K., Srinivasan, M., Zhang, S., 1999. Error is proportional to distance measured
by honeybees: Weber’s law in the odometer. Anim. Cogn. 2, 11–16.
Christian, K.A., Morton, S.R., 1992. Extreme thermophilia in a Central Australian ant,
Melophorus bagoti. Physiol. Zool. 65, 885–905.
Collett, M., 2012. How navigational guidance systems are combined in a desert ant.
Curr. Biol. 22, 927–932.
Collett, T.S., Cartwright, B.A., Smith, B.A., 1986. Landmark learning and visuo-spatial
memories in gerbils. J. Comp. Physiol. A 158, 835–851.
Collett, T.S., Zeil, J., 1998. Places and landmarks: an arthropod perspective. In: Healy,
S. (Ed.), Spatial Representations in Animals. Oxford University Press, Oxford/New
York.
Cruse, H., Wehner, R., 2011. No need for a cognitive map: decentralized memory for
insect navigation. PLoS Comput. Biol. 7, e1002009.
Fourcassié, V., Traniello, J.F.A., 1994. Food searching behaviour in the ant Formica
schaufussi (Hymenoptera Formicidae): response of naive foragers to protein and
carbohydrate food. Anim. Behav. 48, 69–79.
Fukushi, T., 2001. Homing in wood ants Formica japonica: use of the skyline
panorama. J. Exp. Biol. 206, 535–541.
Fukushi, T., Wehner, R., 2004. Navigation in wood ants Formica japonica: context
dependent use of landmarks. J. Exp. Biol. 207, 3431–3439.
Graham, P., Cheng, K., 2009. Ants use the panoramic skyline as a visual cue during
navigation. Curr. Biol. 19, R935–R937.
Graham, P., Philippides, A., Baddeley, B., 2010. Animal cognition: multi-modal inter-
actions in ant learning. Curr. Biol. 20, R639–R640.
Hoffmann, G., 1983a. The random elements in the systematic search behavior of the
desert isopod Hemilepistus reaumuri. Behav. Ecol. Sociobiol. 13, 81–92.
Hoffmann, G., 1983b. The search behavior of the desert isopod Hemilepistus reaumuri
as compared with a systematic search. Behav. Ecol. Sociobiol. 13, 93–106.
Hoffmann, G., 1985a. The influence of landmarks on the systematic search behavior
of the desert isopod Hemilepistus reaumuri I. Role of the landmark made by the
animal. Behav. Ecol. Sociobiol. 17, 325–334.
Hoffmann, G., 1985b. The influence of landmarks on the systematic search behavior
of the desert isopod Hemilepistus reaumuri II. Problems with similar landmarks
and their solution. Behav. Ecol. Sociobiol. 17, 335–348.
Kamil, A.C., 1987. A synthetic approach to the study of animal intelligence. Nebraska
Symp. Motiv. 35, 257–308.
Kamil, A.C., Balda, R.P., 1985. Cache recovery and spatial memory in Clark’s nutcrack-
ers (Nucifraga columbiana). J. Exp. Psychol. Anim. Behav. Proc. 11, 95–111.
Kohler, M., Wehner, R., 2005. Idiosyncratic route memories in desert ants Melophorus
bagoti: how do they interact with path integration vectors? Neurobiol. Learn.
Mem. 83, 1–12.
MacLean, E.L., Matthews, L.J., Hare, B.A., Nunn, C.L., Anderson, R.C., Aureli, F., Brannon,
E.M., Call, J., Drea, C.M., Emery, N.J., Haun, D.B.M., Herrmann, E., Jacobs, L.F., Platt,
M.L., Rosati, A.G., Sandel, A.A., Schroepfer, K.K., Seed, A.M., Tan, J.Z., van Schaik,
C.P., Wobber, V., 2012. How does cognition evolve? Phylogenetic comparative
psychology. Anim. Cogn. 15, 223–238.
Macquart, D., Garnier, L., Combe, M., Beugnon, G., 2006. Ant navigation en route to
the goal: signature routes facilitate way-finding of Gigantiops destructor. J. Comp.
Physiol. A 192, 221–234.
Mangan, M., Webb, B., 2012. Spontaneous formation of multiple routes in individual
desert ants (Cataglyphis velox). Behav. Ecol. 23, 944–954.
Merkle, T., Knaden, M., Wehner, R., 2006. Uncertainty about nest position influences
systematic search strategies in desert ants. J. Exp. Biol. 209, 3545–3549.
Merkle, T., Wehner, R., 2009a. How flexible is the systematic search behaviour of
desert ants? Anim. Behav. 77, 1051–1056.
Merkle, T., Wehner, R., 2009b. Repeated training does not improve the path integra-
tor in desert ants. Behav. Ecol. Sociobiol. 63, 391–402.
Möller, R., 2012. A model of ant navigation based on visual prediction. J. Theor. Biol.
305, 118–130.
Müller, M., Wehner, R., 1994. The hidden spiral: systematic search and path inte-
gration in desert ants, Cataglyphis fortis. J. Comp. Physiol. A. 175, 525–530.
Müller, M., Wehner, R., 2010. Path integration provides a scaffold for landmark
learning in desert ants. Curr. Biol. 20, 1368–1371.
Muser, B., Sommer, S., Wolf, H., Wehner, R., 2005. Foraging ecology of the ther-
mophilic Australian desert ant, Melophorus bagoti. Aust. J. Zool. 53, 301–311.
Narendra, A., 2007a. Homing strategies of the Australian desert ant Melophorus
bagoti I. Proportional path integration takes the ant half-way home. J. Exp. Biol.
210, 1798–1803.
Narendra, A., 2007b. Homing strategies of the Australian desert ant Melophorus
bagoti II. Interaction of the path integrator with visual cue information. J. Exp.
Biol. 210, 1804–1812.
Narendra, A., Cheng, K., Sulikowski, D., Wehner, R., 2008. Search strate-
gies of ants in landmark-rich habitats. J. Comp. Physiol. A 194,
929–938.
Narendra, A., Cheng, K., Wehner, R., 2007a. Acquiring, retaining and integrating
memories of the outbound distance in the Australian desert ant Melophorus
bagoti. J. Exp. Biol. 210, 570–577.
Narendra, A., Si, A., Sulikowski, D., Cheng, K., 2007b. Learning, retention and coding
of nest-associated visual cues by the Australian desert ant, Melophorus bagoti.
Behav. Ecol. Sociobiol. 61, 1543–1553.
Olson, D.J., Kamil, A.C., Balda, R.P., Nims, P.J., 1995. Performance of four seed caching
corvid species in operant tests of nonspatial and spatial memory. J. Comp.
Psychol. 109, 173–181.
Perry, C.J., Barron, A.B., Cheng, K., 2013. Invertebrate learning and cognition: relat-
ing phenomena to neural substrate. Wiley Interdisciplinary Rev. Cogn. Sci. 4,
561–582, http://dx.doi.org/10.1002/wcs.1248.
Reid, S.F., Narendra, A., Hemmi, J.M., Zeil, J., 2011. Polarised skylight and the land-
mark panorama provide night-active bull ants with compass information during
route following. J. Exp. Biol. 214, 363–370.
Reynolds, A.M., 2008. Optimal random Levy-loop searching: new insights into the
searching behaviours of central-place foragers. Europhysics Lett. 82, 20001.
Reynolds, A.M., Schultheiss, P., Cheng, K., 2013. Are Lévy flight patterns derived
from the Weber–Fechner law in distance estimation? Behav. Ecol. Sociobiol.
67, 1219–1226.
Reynolds, A.M., Schultheiss, P., Cheng, K., 2014. Does the Australian desert ant
Melophorus bagoti approximate a Lévy search by an intrinsic bi-modal walk?
J. Theor. Biol. 340, 17–22.
Reynolds, A.M., Smith, A.D., Menzel, R., Greggers, U., Reynolds, D.R., Riley, J.R., 2007a.
Displaced honey bees perform optimal scale-free search flights. Ecology 88,
1955–1961.
Reynolds, A.M., Smith, A.D., Reynolds, D.R., Carreck, N.L., Osborne, J.L., 2007b. Hon-
eybees perform optimal scale-free searching flights when attempting to locate
a food source. J. Exp. Biol. 210, 3763–3770.
Ronacher, B., Wehner, R., 1995. Desert ants Cataglyphis fortis use self-induced optic
flow to measure distances travelled. J. Comp. Physiol. A 177, 21–27.
Santschi, F., 1911. Sur le mécanisme de l’orientation chez les fourmis. Revue Suisse
Zool. 19, 303–338.
Santschi, F., 1913. Comments’ orientent les fourmis. Revue Suisse Zool. 21, 347–426.
Schultheiss, P., Cheng, K., 2011. Finding the nest: inbound searching behaviour in
the Australian desert ant, Melophorus bagoti. Anim. Behav. 81, 1031–1038.
Schultheiss, P., Cheng, K., 2013. Finding food: outbound searching behavior in the
Australian desert ant Melophorus bagoti. Behav. Ecol. 24, 128–135.
Schultheiss, P., Nooten, S.S., 2013. Foraging patterns and strategies in an Australian
desert ant. Austral Ecol., http://dx.doi.org/10.1111/aec.12037.
Schultheiss, P., Schwarz, S., Cheng, K., Wehner, R., 2012. Foraging ecology of an
Australian salt-pan desert ant (genus Melophorus). Aust. J. Zool. 60, 311–319.
Schultheiss, P., Wystrach, A., Legge, E.L.G., Cheng, K., 2013. Information content
of visual scenes influences systematic search of desert ants. J. Exp. Biol. 216,
742–749.
Schwarz, S., Albert, L., Wystrach, A., Cheng, K., 2011a. Ocelli contribute to the encod-
ing of celestial compass information in the Australian desert ant Melophorus
bagoti. J. Exp. Biol. 214, 901–906.
Schwarz, S., Cheng, K., 2010. Visual associative learning in two
desert ant species. Behav. Ecol. Sociobiol. 64, 2033–2041,
http://dx.doi.org/10.1007/s00265-010-1016-y.
Schwarz, S., Narendra, A., Zeil, J., 2011b. The properties of the visual system in the
Australian desert ant Melophorus bagoti. Arthropod Struct. Dev. 40, 128–134.
Schwarz, S., Wystrach, A., 2011. Visual input and path stabilization in walking ants.
Comm. Integr. Biol. 4, 758–760.
Schwarz, S., Wystrach, A., Cheng, K., 2011c. A new navigational mechanism mediated
by ant ocelli. Biol. Lett. 7, 856–858.
Shettleworth, S.J., 2010. Cognition, Evolution, and Behavior, second ed. Oxford Uni-
versity Press, New York.
Sommer, S., von Beeren, C., Wehner, R., 2008. Multiroute memories in desert ants.
Proc. Natl. Acad. Sci. U S A 105, 317–322.
Sommer, S., Wehner, R., 2004. The ant’s estimation of distance travelled:
experiments with desert ants, Cataglyphis fortis. J. Comp. Physiol. A 190,
1–6.
Sommer, S., Wehner, R., 2012. Leg allometry in ants: extreme long-leggedness in
thermophilic species. Arthropod Struct. Dev. 41, 71–77.
Steck, K., Hansson, B.S., Knaden, M., 2009. Smells like home: desert ants
Cataglyphis fortis, use olfactory landmarks to pinpoint the nest. Front. Zool. 6,
5, http://dx.doi.org/10.1186/1742-9994-6-5.
Stieb, S.M., Muenz, T.S., Wehner, R., Rössler, W., 2010. Visual experience and
age affect synaptic organization in the mushroom bodies of the desert ant
Cataglyphis fortis. Dev. Neurobiol. 70, 408–423.
Strausfeld, N.J., Hirth, F., 2013. Deep homology of arthropod central complex and
vertebrate basal ganglia. Science 340, 157–161.
Tomback, D.F., 1980. How nutcrackers find their seed stores. Condor 82, 10–19.
Traniello, J.F.A., Kozol, A.J., Fournier, M.A., 1992. Resource-related spatial patterns of
search in the ant Formica schaufussi: a field study. Psyche 99, 87–94.
Vander Wall, S.B., Balda, R.P., 1977. Coadaptations of the Clark’s Nutcracker and the
piñon pine for efficient seed harvest and dispersal. Ecol. Monogr. 47, 79–111.
Wehner, R., 1987. Spatial organization of the foraging behavior in individually
searching desert ants Cataglyphis (Sahara desert) and Ocymyrmex (Namib
desert). In: Pasteels, J.M., Deneubourg, J.M. (Eds.), From Individual to Collective
Behavior in Insects. Birkhäuser, Basel.
Wehner, R., 1990. On the brink of introducing sensory ecology: Felix Santschi
(1872–1940) – Tabib-eb-Neml. Behav. Ecol. Sociobiol. 27, 295–306.
Wehner, R., 1994. The polarization-vision project: championing organismic biology.
Fortschritte Zool. 39, 103–143.
Wehner, R., 2003. Desert ant navigation: how miniature brains solve complex tasks.
J. Comp. Physiol. A 189, 579–588.
 K. Cheng et al. / Behavioural Processes 102 (2014) 51–61
Wehner, R., 2012. Wüstennavigatoren en miniature: Rennpferde der Insektenwelt.
Biol. unserer Zeit 42, 364–373.
Wehner, R., 2013. Life as a cataglyphologist – and beyond. Ann. Rev. Entomol. 58,
1–18.
Wehner, R., Boyer, M., Loertscher, F., Sommer, S., Menzi, U., 2006. Ant navigation:
one-way routes rather than maps. Curr. Biol. 16, 75–79.
Wehner, R., Harkness, R.D., Schmid-Hempel, P., 1983. Foraging strategies in indi-
vidually searching ants Cataglyphis bicolor (Hymenoptera: Formicidae). Gustav
Fischer Verlag, Stuttgart.
Wehner, R., Meier, C., Zollikofer, C., 2004. The ontogeny of foraging behaviour in
desert ants, Cataglyphis bicolor. Ecol. Entomol. 29, 240–250.
Wehner, R., Michel, B., Antonsen, P., 1996. Visual navigation in insects: coupling of
egocentric and geocentric information. J. Exp. Biol. 199, 129–140.
Wehner, R., Müller, M., 2006. The significance of direct sunlight and polarized sky-
light in the ant’s celestial system of navigation. Proc. Natl. Acad. Sci. U S A 103,
12575–12579.
Wehner, R., Müller, M., 2010. Piloting in desert ants: pinpointing the goal by discrete
landmarks. J. Exp. Biol. 213, 4174–4179.
Wehner, R., Räber, F., 1979. Visual spatial memory in desert ants, genus Cataglyphis
(Formicidae, Hymenoptera). Experientia 35, 1569–1571.
Wehner, R., Srinivasan, M.V., 1981. Searching behaviour of desert ants, genus
Cataglyphis (Formicidae, Hymenoptera). J. Comp. Physiol. A 142, 315–338.
Wehner, R., Srinivasan, M.V., 2003. Path integration in insects. In: Jeffery, K.J. (Ed.),
The Neurobiology of Spatial Behaviour. Oxford University Press, Oxford.
Wehner, R., Wehner, S., 2011. Parallel evolution of thermophilia: daily and sea-
sonal foraging patterns of heat-adapted desert ants: Cataglyphis and Ocymyrmex
species. Physiol. Entomol. 36, 271–281.
61
Wiener, J., Shettleworth, S., Bingman, V.P., Cheng, K., Healy, S., Jacobs, L.F., Jeffery,
K.J., Mallot, H.A., Menzel, R., Newcombe, N.S., 2011. Animal navigation: a syn-
thesis. In: Menzel, R., Fischer, J. (Eds.), Animal Thinking: Contemporary Issues in
Comparative Cognition. MIT Press, Cambridge, MA/London.
Wittlinger, M., Wehner, R., Wolf, H., 2006. The ant odometer: stepping on stilts and
stumps. Science 312, 1965–1967.
Wittlinger, M., Wehner, R., Wolf, H., 2007. The desert ant odometer: a stride integra-
tor that accounts for stride length and walking speed. J. Exp. Biol. 210, 198–207.
Wolf, H., Wehner, R., 2000. Pinpointing food sources: olfactory and anemotactic
orientation in desert ants, Cataglyphis fortis. J. Exp. Biol. 203, 857–868.
Wystrach, A., Beugnon, G., Cheng, K., 2011a. Landmarks or panoramas: what do
navigating ants attend to for guidance? Front. Zool. 8, 21.
Wystrach, A., Beugnon, G., Cheng, K., 2012. Ants might use different view-matching
strategies on and off the route. J. Exp. Biol. 215, 44–55.
Wystrach, A., Graham, P., 2012. What can we learn from studies of insect navigation?
Anim. Behav. 84, 13–20.
Wystrach, A., Mangan, M., Philippides, A., Graham, P., 2013a. Snapshots in ants? New
interpretations of paradigmatic experiments. J. Exp. Biol. 216, 1766–1770.
Wystrach, A., Schwarz, S., Baniel, A., Cheng, K., 2013b. Backtracking behaviour in lost
ants: an additional strategy in their navigational toolkit. Proc. R. Soc. B-Biol. Sci.
280, 20131677.
Wystrach, A., Schwarz, S., Schultheiss, P., Beugnon, G., Cheng, K., 2011b. Views, land-
marks, and routes: how do desert ants negotiate an obstacle course? J. Comp.
Physiol. A 197, 167–179.
Ziegler, P., Wehner, R., 1997. Time-courses of memory decay in vector-based and
landmark-based systems of navigation in desert ants, Cataglyphis fortis. J. Comp.
Physiol. A 181, 13–20.