Ethology 102 99-108 (1996)
0 1 996 Blackwell Wissenschafts-Verlag, Berlin
Dipartimento di Biologia e Fisiologia generali, Universita di Pama and Dipartimento di
Biologia animale e Genetica, Universitd di Firen7e
Homing Behaviour in Polyergus mfescens Latr.
DoNAro ANTONIO GRASSO,
GRASS), D. A,, UGOLINI
Formicidae). Ethology 102, 99-108.
A. & I.E Moi,~, F. 1996: Homing behaviour in Po/yetps mf.scens Latr. (Hymenoptera,
Homing mechanisms of the European slave-making ant Po/yequs mjscens Latr. are investigated by field
experiments. The analysis of the behaviour and paths of both homing scouts and raiders after passive
displacement showed that: I. Scouts probably home by using a path integrauon system based on celestial
cues; and ii. Displaced raiders do not seem to adopt such a vectorial orientation mechanism. Moreover, we
found that passively displaced scouts exhibit a systematic search strategy for the nest after a rectilinear path.
By contrast, raiders perform a similar search pattern just after release. Similanties between Cutu&his
I-’o/yegys homing behaviour are discussed.
Correspondence: F. I.F Moi,~, Dipartimento di Biologia e Fisiologja generali, Universita degli Studi,
Vide delle Scienze, 1-43100 Parma, Italy.
To reach goals, such as food or a nest site, ants can basically employ two systems:
1. Trail-following; and 2. Orientation based on visual cues. As far as the latter is
concerned, an important role is played by mechanisms involving celestial cues and path
integration (see reviews by HOLLDOBLER
& WILSON 1990; WEHNER
Slave-making (dulotic) ants are social parasites that raid other ant colonies in order
to paage the brood by which they obtain slaves for domestic tasks. Hence, the worker
force of a dulotic colony is composed of slave foragers, and slave-making scouts and
raiders whose role is to locate and attack target nests, respectively (TOPOFF
LE MOLI et al. 1994). In the afternoon, Po&tgus rcrfescens workers leave their colony
searching for target nests. The outbound route is generally tortuous, but is followed by a
return trip along a different, ths time straight path. After a successful trip, the scout
recruits hundreds of nestmates and leads them towards the target nest. Paths engaged
by raiders are quite rectilinear. Moreover, returning home, raiding swarms follow exactly
et al. 1984;
U. S. Copynght Clearance Center Statement. 01 79-161 3/96/1022-0099$11.50/0
& I.F. MOLI, Homing in Pobepf njmcens
the same path used during the outbound trip since they probably use a chemical trail to
reach home (LE MOLI et al. 1994).
At present, there is no evidence that Po&egys scouts use trd-following
mechanisms during their journeys (see LE MOLI et al. 1994). Hence, we hypothesize
that P. mjscenf homing ability is based on an orientation mechanism similar to that
adopted by Cata&his
ants, although they live in completely different habitats
(grasslands and desert areas, respectively; see WEHNER 1992).
In this paper, we report the results of a field investigation that provides the first
data on the homing mechanisms adopted by P. mjscens scouts. Moreover, we present
further findmgs on the homing behaviour of raiders.
The experimental area (near Parma, northern Italy) was a patched network of fallow fields and small
woods. Within the fields, five dulotic colonies of Po/yeergrrs mjmm were found. All of these were populated by
workers of Fomira runiwhria Latr. as slaves.
The paths engaged by scouts were marked with pins that were inserted in the ground every time the
ants changed drection (by at least 15") after having covered a distance of at least 10-15 cm. In order to
analyse the paths, the angles and distances between consecutive pins were measured.
The homing behaviour of the scouts after passive displacements was investigated In 13 dfferent
indviduals. The experiments were carried out under the sun and blue sky. However, it was occasionally
pssible to conduct some experiments with a completely overcast sky. We followed each scout on their
outward path and induced them to stop searching by using a freshly collected host-colony fragment
(composed of nest material, brood, adult workers). This was placed in front of the walking ant. When scouts
started the inbound trip they were captured (using an opaque box) and displaced to a pre-established release
site. Releases were made in unfamiliar sites located far away from the mother nest (up to 160 m) and other
P mjmens colonies in order to avoid the influence of chemical trails on scouts' orientation. Moreover,
where possible, release sites were chosen so as to be mametrically opposite to the dxection capture point-
For each dsplaced scout's path, we determined the trend of the rectilinearity index (0 < RI 5 1, defined
as the ratio between the minimum distance from the release point and the actual length of the path).
Following a method similar to that used by WEHNER et al. (1983), we divided the inbound path into two
parts: 1. Return trip (usually straight and long); and 2. Tormous searching path for the nest. We considered as
the origin of search the point were RI decreased abruptly (Fig. 1). Angles and dstances between capture
point and nest, as well as between release point and origin of search were recorded.
The analysis of the ant search phase was made by plotting the distance from the origin of search against
the path length covered until that point for each segment of the path (see also WEHNER & SRINIVASAN 1981).
To compare with scouts' behaviour, some raider ants (n = 20) carrying pupae were individually
captured from inbound columns, displaced and released at different distances from the mother colony (from
3.5 up to 20 m). Their return behaviour and search paths were recorded and analysed by the same methods
used for dsplaced scouts.
Return Routes of Displaced Scouts
Soon after release, scouts moved along a straight route for a whde and then
engaged in a searching path (Fig. 2).
We found a highly sipficant correlation between home drections at the moment
of capture (capture point-nest) and directions selected by the scouts after the
displacement and release (release point-origm of search) (Fig. 3a). There is no statistical
difference between the regression line and the theoretical line for a perfect coincidence
between drections. This means that, independently of the position of the release sites,
dsplaced ants selected the &ration that, before capture, would have led them back to
the mother colony.
We also found a good correlation between the &stances the ants should have
covered from the capture point to the nest, and the &stances actually covered by
the scouts in the rectilinear phase after release (Fig. 3b). However, the regression h e
differs from the theoretical line of perfect coincidence. In fact, dlsplaced scouts often
stopped their straight route and began searchtng before covering the total homing
distance. These errors were more evident for ants that performed long pre-
dsplacement trips. Indeed, we found a good correlation between the outbound path-
lengths and such errors (differences between theoretical and covered homing &stances;
Searching Paths of Displaced Scouts
The analysis of the search for the nest shows that scouts engaged in ever enlarging
loops that led them further and further away; however, the ants repeatedly returned to
near the origin of the search (Fig. 5).
Scouts occasionally displaced during overcast conditions showed a completely
different pattern of movement. They moved slowly, often stopping, performing a
tortuous path that lacked a rectilinear phase.
D. A. GRASSO.
A. UGOIJNI & F. I.E Moi.~
scouting task; b. Homing path of the same scout after having been captured at point C and passively
displaced to a test site and released (R). Spatial co-ordinates are the same in both cases
Homing scouts: a. Example of homing path of a scout returning to the nest from an undisturbed
Homing of Displaced Raiders
Raider ants (n = 2) collected during the return trip, passively displaced and released
within 3.5 m of their nest, were able to home without searching. In contrast to this,
homing raiders displaced 11-20 m from the mother colony, on release immediately
started with a search strategy very similar to the search phase performed by displaced
scouts (Fig. 6). Unlike in the case of scouts, a return along a straight path was completely
absent in raiders. Only six out of 20 displaced ants, after having searched at the release
site for 11-32 min, eventually succeeded in finding one of the chemical trails placed by
earlier swarming raiders, and reached the nest.
Trips performed by P. mjscem scouts are characterized by hfferent and non-
overlapping outbound and inbound paths (LE MOLI et al. 1994). This means that such
3 6 0 ' -
c7 2000 ;
n = 13
y = 1.03~
r = 0.98 p<O.OOl
t = 0.57
p = ns
180" 200" 220' 240" 260" 280" 300" 320" 340' 360'
ANGLE CAPTURE POINT - NEST
4: 2 15
n = 13
y = 0.64X +2.23
. _ _
L = 0.d
I I I
25 3D 35 40 45 50
DISTANCE CAPTURE POINT - NEST (m)
F&. 3 :
after displacement (release point+)rigin of search); b. Correlation between homing distances (measured
from capture point to the nest) and &stances covered during the rectilinear phase of the post-dJsplacement
path. n = number of ants; r = correlation coefficient; t = Student's t-test value for companson between
observed and expected regression lines; equation of the regression line is also given
Homing scouts: a. Correlation between home directions (capture point-nest) and directions taken
ants do not usually home using a trail-following mechanism. Our experiments of passive
displacement under the sun and blue sky showed that, after release, ants moved along
linear trajectories that terminated in a systematic search. The correlation between home
directions and those taken by scouts soon after release shows that they moved in the
D. A. GRASSO,
A. UGOLINI & F. LE MOLI
n = 13
y = 0.32X -5.32
r = 0.71 p<0.01
t = 7.12 p<O.O1
I I 0
PATH LENGTH OF THE OUTBOUND TRIP (in)
F&. 4: Homing scouts. Correlation between the path length of the outbound trips and the errors in the
evaluation of the homing distances. Error = differences between theoretical (capture point-nest) and
covered (release pint-search origin) homing distances; n = number of ants; r = correlation coefficient; t =
Student's t-test value; equation of the regression line is also given. For further explanations, see Fig. 3
directions that before the capture would have led them home. For some of the scouts,
there was also a very good correspondence between the homing &stance and the
distance covered during the linear route of the post-displacement trip. Thus, the results
show that scouts exhibit a vectorial orientation mechanism that enable them to reach
the vicinity of their colony. A vectorial orientation mechanism based on a path
integration system has been thoroughly stu&ed in the desert ants of the genus
1986; MULLER & WEHNER 1988; SCHMIDT
WEHNER 1992). Cata&bis
foragers continuously record their location relative to home
and compute a mean vector pointing from their actual position towards the nest based
on the past trajectories (MULLER
& WEHNER 1988, 1994; WEHNER 1992). Th~s system
allows the ants to home by a straight route rather than retracing the outbound path.
Although P. mjscens lives in grasslands -
with desert areas -
our results show a strilung similarity in the homing behaviour of
Pabepa scouts and Cata&bis foragers. Thus, P. mjscens scouts probably home by path
integration. Scouts displaced under overcast sky were disoriented. This observation
suggests that celestial cues are involved in the scouts' homing ability. Indeed, scouting
activity seldom occurs during overcast days (LE MOLI et al. 1994). There is experimental
evidence that the scouts of the related American species Pabergas bevzcqs basically resort
to visual cues during their trips (TOPOFF
et al. 1984, 1985).
~ E H N E R & WEHNER et al. 1992;
a very heterogeneous habitat compared
30 - a
€ 2 0 -
Z 15 -
I,, , ,
I,, , ,
PATH LENGTH (in)
100 110 120 130 140 150 160
( I ,
70 80 90
PATH LENGTH (m)
Fig. 5; Homing scouts. Two examples (a, b) of search strategy adopted by displaced scouts once they
reached the fictive position of the nest. The graphs were obtained by plotting (for each segment of the
path) the &stance of the ant’s position from the origin of search (ordinate) against the path length actually
covered by the ant to that point (abscissa)
During the outbound trip, Cata&bzs
do not allow them to reach exactly the position of the nest (WEHNER & SRINIVASAN
1981; MULLER & WEHNER 1988; WEHNER 1992). This could also be the case with the
dsplaced P. mjscens scouts that covered a shorter rednear path as compared with the
real homing distance. In fact, these dscrepancies were generally greater with increasing
length of the outbound path. Moreover, it is possible that P. mjscens memorizes
as found in CatagCyphis (WEHNER & RABER 1979; WEHNER et al. 1983;
COLLETT et al. 1992) -
and/or resorts to chemical trails present near the nest: these
ants accumulate computational errors that
D. A. GMSSO, A. UGOIJNI
& F. I.E MOLI
PATH LENGTH (m)
transported passively to a test site. The path consists exclusvely of searching behaviour, the geometry of
which (b) is very similar to that obaen~d in scouts (see Fig. 5). The straight return path that scouts display
in the initial phase of homing (see Fig. 2) is totally absrnt in raiders. This example is one of many showing
that, in the homing task, raiders rely on olfactory cues more than on other onenting factors
Homing raiders: a. Example of homing path of a raider captured during the return trip and
additional orienting factors, which could affect the final part of the homing trip, are
obviously not available in the unknown release site.
Displaced scouts searching for their nest also show a systematic search strategy
similar to that exhibited by Cutu&hzs.
This searching behaviour consists of ever
expandmg loops starting and ending near the origin of search, the point where home
was most likely to be ~ E . H N E R & SRINIVASAN
& WEHNER 1994). This kind of search strategy, common to many other arthropods (see,
for example, HOFFMANN 1983), enables the ant to reach the nest despite possible errors
that have accumulated during the path-integration process.
1981; WEHNER & WEHNER 1986; MULLER
Homing in Pobergus rufescens
Raiders passively dsplaced during the inbound trip, although strongly motivated
to return to the nest, did not take the correct homeward direction. Soon after release,
raider ants dsplaced far away from the nest performed a search strategy very similar to
that typically adopted by displaced scouts at the end of the return trip, without engaging
in a rectilinear path as scouts did. This finding suggests that homing raiders do not use
path integration during raiding journeys. They were able to home only when released
near the nest or after having engaged in a more-or-less prolonged exploration of the
surrounding area. This is in accordance with our previous findings (~t: Mor.r et al. 1994)
that Pobetgm raiders home relying m d y on chemical trds. The adaptive value of such
a strategy was directly observed in some cases when displaced raiders, after having
performed several loops, encountered the chemical trail deposited by a raiding phalanx,
which allowed them to return home. During the season, we verified that scouts
@reviously marked) repeatedly took part in raids. Hence, such individuals could
function both as scouts and raiders on different days. This suggests that these ants have
the capacity of vector orientation, but when raiding do not use it. Thus, we hypothesize
that the same indwidual can use dfferent cues in different tasks. This resembles several
behavioural aspects in the honeybee (see LEHRER 1993).
The authors wish to give their sincere thanks to Prof. Dr. R. !&HNb.R
Universitat Zurich, Swtzerland) for his critical reading of the manuscript and helpful suggesuons. Thanks
are due to Mr J. HOI:~;MAN
(School of Biological Sciences, University of Wales, Swansea, UK) for his
invaluable aid in collecting field data during the course at the University of Pama as a student of the
Erasmus Project. This research was supported by qants from the Ministero dell’Universita e d e b Ricerca
Scientifica e Tecnologica assiqed to F. I.E MOIL
(Zoologisches lnstitut der
COI.I.EI.I., T. S., DIIJXA~~,
ants. I. Comp. Physiol. A 170, 435442.
E., GKXR, A. & W~HNFR,
R. 1992: Visual landmarks and route following in descrt
HoI:f.MANN, G. 1983: The search behavior of the desert isopod Hemikpisfus naumuti as compared with a
systematic search. Behav. Ecol. Sociobiol. 13, 93-106.
LEHRFJ, M. 1993: Spatial vision in the honeybee: the use of different cues in dfferent tasks. Vision Res. 34,
I . E Mor.~, F., GRASSO,
D.A., Mi)RI, A. & L~OI,INI,
A. 1994: Eco-ethological factors affecting the scouting and
raiding behaviour of the slave-malung ant, f’o/yqu.r
MCUER, M. & WWKER, R. 1988: Path integration in desen ants, Cbfug&hzs,iJ-fortls. Proc. Natl. Acad. Sci. USA
_ _ & --
1994: The hidden spiral: systemauc search and path intep-ation in desert ants, (btu&bhisfirtis.
J. Comp. Physiol. A 175, 525-530.
I., COI.I.FTT, T. S., DII.I.IFX,
F.-S. & K’FHNF.R,
on their way home. J. Comp. Physiol. A 171, 285-288.
TOPOFF, H., 1.h Mi)N, B., Gi)i)txoE, I.. & GOI.DSTF.IN, M. 1984: Social and orientation behavior of /-)olvergus
hui+ during slave-making raids. Behav. Ecol. Sociobiol. 15, 273-279.
_ _ , PAGANI,
M. & MACK, L. 1985: Onentation behavior of the slave-making ant f’o/yeTus
bmtkep in an oak-woodland habitat. I. New York Entomol. Soc. 93, 1041-1046.
WEHNFX, R. 1992 Anhropods. In: Animal Homing (FAPI,
_ _ , HARKNESS,
R.D. & Si:HMID-HF.MPEI., P. 1983: Foraging Strategies in Individually Searching Ants
Iicdor (Hymenoptera: Formicidae). Fischer-Verlag, Stuttgart, New York.
B. & WIISON, E.O. 1990: The Ants. Springer-Verlag, Berlin, Heidelberg.
yfesrens Latr. (Hymenoptera, Formicidae). Ethology
R. 1992: How desert ants cope with enforced detours
F., ed.). Chapman & Hall, London, New York. pp.
- _ & SRINIVASAN,
- - & WEHNER, S. 1986 Path integration in desert ants. Approaching a long-standing puzzle in mect
navigation. Monitore zml. ital. (N.S.) 20, 309-331.
UGOLINI & LE MOLI, Homing in Polyergvs m-scens
& RALER, F. 1979: Visual spatial memory i n desert ants, Cata&bis bicoior(Hymenoptera: Formicidae).
Experientia 35, 1569-1 571.
M.V. 1981: Searching behaviour of desert ants, genus Cda&hzs
Hymenoptera). J. Comp. Physiol. 142, 315-338.
Received Aprii 10, 1995
Accepted August 25, 1995 ('W Wckbr)