THE ROLE OF SCOUTING IN SLAVE RAIDS BY POL YERGUS BREVZCEPS ^HYMENOPTERA: FORMICIDAE)*

The formicine ant genus Polyergus contains four species, all of which are obligatory social parasites of the related genus Formica. Slave ants are obtained during group raids, in which a swarm of Polyergus workers penetrates a nest of Formica, disperses the adult workers and queen, and carries off the pupal brood (Topoff et al. 1984, 1985). Although many of these pupae are subsequently consumed in the slave-maker's nest (Kwait and Topoff 1984), a significant portion of the Formica brood is reared through pupal development. Workers eclosing from this pupal population subsequently perform their typical functions (i-e., foraging, feeding, nest defense) as permanent members of a mixed-species nest. Ever since the pioneering studies on Polyergus rufescens by Huber (1810) and Emery (1908), on P. lucidus by Talbot (1967) and Harman (1968), and on P. breviceps by Wheeler (1916), it has been well known that slave-making raids are usually initiated by a small group of workers called scouts. These individuals locate target colonies of Formica, return to their colony of origin, recruit nestmates, and lead the raiders back to the Formica nest. Despite the generalization that Polyergus slave raids are typically preceded by scouting, virtually no field studies exist showing the actual paths travelled by scouts, or their overall importance in initiating slave raids. In their study of laboratory colonies of P. lucidus, Kwait and Topoff (1984) found that most raids were indeed directed towards

reaching the fence were scouts (and not the circlers, which routinely emerge and mill around the nest entrance prior to raiding).To enable scouts to move in and out of the enclosure, the fence contained four gaps (15 cm wide) at 90-degree intervals, starting at the north end.Beginning at 1400 hr (MST), one person walked continuously around the circular fence.Scouts reaching the fence usually ran along the inside wall, and left the enclosure immediately upon encountering the nearest gap.Sometimes, a scout ran back and forth along the same small section of the fence's inner wall.On these occasions, a garden trowel was placed in the scout's path.When the scout moved onto the trowel, it was lifted and placed on the ground on the outer side of the fence.The path of each scout was marked by placing the stick end of cotton swabs (painted yellow and numbered sequentially) into the ground behind the scout, at approximately l-m intervals.Because of the grass and leaf litter, this     1987]   Topoff Bodini, Sherman, & GoodloemPolyergus 263 procedure required two persons, the first to constantly monitor the scout's path, and the second to place the numbered markers.Using a compass and a rolling measuring wheel, a map was made depicting the scout's path.

Results
During the three-week period of this study, 18 scouts were followed on 12 different days.Of these 18 scouts, five were tracked only on their outbound trip to a target colony of Formica gnava.For three other scouts, we succeeded in following their return trip as well.On five occasions, our tracking was abruptly halted when the scout was seized and killed by a spider.The remaining five scouts disappeared beneath the leaf litter before reaching a target colony.
The outbound paths of all scouts consisted of two distinct phases.
Because of the essential similarity in movement among the scouts, the route illustrated in Figure can serve as a typical example of scouting behavior.The target Formica nest on this day was located approximately 40 rn to the west of the Polyergus nest.Phase one of scouting began when individuals left the swarm of ants circling around the nest, and moved in a relatively constant compass direc- tion.For the scout on the afternoon of July 23, this straight run extended about 30 rn to the west (the range for phase-one scouting on all days was 25-45 m).During this phase the scouts ran continu- ously, without stopping to search for Formica colonies.The second phase of scouting involved a qualitatively different pattern of movement, with scouts changing direction after running small dis- tances over all compass directions.More importantly, it was only during this second phase that scouts periodically stopped and searched beneath rocks and patches of leaf litter.On July 23, this searching phase extended the scout's path for an additional 10 m to the west, with the north and south displacement combined covering about 14 m.Thus the searching phase of this particular scout covered an area of approximately 140 m2.
A representative map of the complete sequence of successful scouting, followed by a slave raid, is illustrated in Figure 2. To clarify the distinctness of the paths, the erratic phase-two move- ments of the scout were replaced by a line extending from the end of the phase-one run, directly to the target nest.The most significant finding here is occurrence of three non-overlapping routes for the outbound run, return trip, and slave raid respectively.After locating Psyche [Vol. 94Fig. 1.Outbound path of Polyergus scout, consisting of a linear phase, followed by a circuitous (searching) phase.Scales of distance are shown at the lower left corner.N north; P location of Polyergus nest; F location of Formica nest.
a Formica nest, the scout's return route paralleled its outbound course, but was displaced about m to the east.The slave raid back to the Formica colony was displaced an additional 2 m to the east.Because the scout was not marked, we can not be certain that the slave raid was led by the same individual.Nevertheless, our previous studies with marked scouts (Topoff et al. 1984) indicate that success- ful scouts of P. breviceps typically run (at least intermittently) at the head of raid swarms.

Materials and methods
To determine the importance of scouts for initiating slave raids, another aluminum-flashing fence (15 cm high, 7 m radius, and con- taining four exits spaced at 90-degree intervals) was constructed around a second colony of P. breviceps.Every afternoon, from 1400 hr to 1800 hr, the perimeter of the enclosure was continuously moni- tored for the presence of scouts.For each scout, we recorded the time of arrival at the fence and the compass direction in which it was travelling.Each scout was then removed and placed in a holding container until the end of the day, at which time all the captured SCOUTS were placed at the nest entrance.This procedure for eliminat- ing scouting was repeated for 19 consecutive days during the period July 2-July 20.Starting on July 21, and continuing until August 4, the fence was removed to permit unlimited scouting.Without the barrier, it was not possible to determine accurately the number of scouts departing each day.During this phase of the study, we simply verified that scouting occurred, and noted the time of raid onset.

Results
A comparison of the frequency of slave raids by P. breviceps with and without scouts is summarized in Table 1.The Table does not include data from days when scouting was prevented by rain or heavy overcast.Slave raids occurred on only 2 of 11 days (18%) during the period of scout removal.The two raids that occurred on 7/12 and 7/20 were stopped at the fence (by sealing the exits) to ensure that no Formica pupae would be brought back to the nest.When the fence was removed to permit scouting, the frequency of slave raids jumped to 69%, as 9 raids were conducted on 13 different days (X 2 7.4, df l, P 0.01).All but one of these raids resulted in the capture of Formica brood, and the only unsuccessful one (July 23) was aborted by an abrupt thunderstorm.

DISCUSSION
Polyergus breviceps is similar to other obligatory parasites in that workers do not search for food.Nevertheless, the location of target nests by scouts can be thought of as indirect foraging, because much of the raided Formica brood is fed to the Polyergus workers and queen by their resident slaves.It is therefore not surprising that the searching pattern of Polyergus scouts illustrated in Figure is almost identical to that described for the ant Cataglyphis bicolor, which forages alone for dead arthropods (Harkness and Maroudas  1985).In this desert-dwelling species, foragers also move away from the nest in a linear path, followed by random searching throughout a particular sector.Cataglyphis can measure angular directions from the pattern of polarized light (Wehner and Menzel 1969), so it could forage in a straight line until it found food, and then return to 1987] its home nest on the same bearing.But Harkness and Maroudas (1985) show that the sideways search pattern decreases the time for the return trip, and the same argument could easily hold for Polyer- gus.In addition, the linear phase of the scout's route ensures min- imum overlap among the sectors searched by all of the scouts on any given day.
Polyergus is also similar to Cataglyphis (and other formicine ants) in its ability to orient to polarized light, and we recently dem- onstrated that scouts use visual (and not chemical) orientation dur- ing their outbound and return run, and when leading the raid swarm back to the target nest (Topoff et al. 1984).Further support for this hypothesis stems from our map of the complete route taken by a scout on July 13 (Fig. 2), in which a different path was taken for each of the scout's three runs.A model based upon chemical cues would predict a single path for the outbound run, return trip, and slave raid respectively.However, because a chemical trail is depos- ited during the slave raid, we have not ruled out the possibility that scouts might follow trails deposited on previous days.
The initiation of slave-making raids by scouts has so far been reported for the myrmicine genera Strongylognathus, Harpagoxenus, Leptothorax, Epimyrma, and Chalepoxenus, and in the formi- cine genera Formica, Polyergus, and Rossomyrmex (see review by  Buschinger et al. 1980).In both subfamilies, details of scouting show several striking similarities.For example, scouts of the myrmicine ant Harpagoxenus sublaevis tend to be experienced workers which are at least one year old (Buschinger 1968; Buschinger and Winter  1977).This is almost identical to the findings of Kwait and Topoff  (1984) for the formicine Polyergus lucidus, in which ants functioned as scouts only during their second season after eclosing.
A second convergence between the two subfamilies concerns the role of scouting as a prerequisite for raiding.In a laboratory study of the myrmicine species Harpagoxenus americanus and Lepto- thorax duloticus, Alloway (1979) recorded a total of 23 slave raids.For both genera, scouting preceded all of the slave raids, and scout- ing never took place when raids did not occur.In a similar study with P. lucidus, Kwait and Topoff (1984) removed scouts from laboratory nests on each of 7 days, during which time no slave raids occurred.Furthermore, observations showed that 25 out of 27 raids were directed towards Formica schaufussi nests that were scouted on the same day.The remaining two raids were directed at target Psyche [Vol. 94  nests that were scouted on the previous raid day.Our present field study with P. breviceps confirms the importance of scouting in the natural habitat.Because all scouts collected were returned to their nest in late afternoon, the colony's "interpretation" might have been that these scouts were simply unsuccessful in locating a colony of Formica.If this were the case, an adaptive colony response would be to send out more scouts on successive days.Note (in Table 1) that the number of scouts from July 7-July 11 ranged between and 8, but that 15-25 scouts were collected between July 12 and July 20.It is plausible to hypothesize that the magnitude of scouting can be varied according to the colony's requirements, but a more rigorous, statistical analysis must await a larger sample size.
The data from July 12 and 20 also show that raids can be organ- ized even when scouting has not taken place for at least one week, but we do not know how recruitment takes place when this occurs.When groups of Polyergus are led by same-day scouts, the raid typically mimics the biphasic nature of scouting.Thus, the raid starts with a relatively linear movement away from the nest, during which the swarm advances without stopping.This is followed by the second phase in which the ants periodically stop advancing, fan out in all directions, and search under rocks and leaf litter (Topoff et al.  1984).It will be interesting in future field studies to observe how slave raids proceed when not organized by same-day scouts.Under such circumstances, the linear phase of raiding might be eliminated, so that the Polyergus swarm is forced to advance in a more deliber- ate manner, with more numerous searching stops even close to the nest.Finally, formicine ants such as Formica rufa exhibit site alle- giance, a process that is based upon individual memory of spatially- organized visual cues (Rosengren and Fortelius 1986).Using individually-marked Polyergus, we now plan to determine whether each scout has a particular compass direction and sector in which it regularly searches for Formica colonies.

SUMMARY
Slave raids by Polyergus breviceps are initiated by one or more scouts which locate target colonies of Formica gnava, and recruit nestmates to participate in group raids.Field studies in southeastern Arizona showed that scouting comprises two distinct phases.The first is a linear movement away from the nest, during which no 1987] Topoff BodinL Sherman, & Goodloe--Polyergus 269 searching occurs.This is followed by sideways movements, a phase characterized by intensive searching under rocks and leaf litter throughout a limited sector.Sometimes, the scout takes a different route on the outbound run, return trip, and slave raid respectively.The distinctness among these three paths is consistent with the hypothesis that scouts rely primarily on visual orientation.The importance of scouts was determined by their systematic removal, resulting in the absence of raiding on 9 out of 11 days.When scout- ing was subsequently permitted, raids occurred on 9 out of 12 days.

Fig. 2 .
Fig. 2. Comparison of routes taken by Polyergus scout during outbound trip, return run, and subsequent slave-raid.Scales of distance are shown at the lower left corner.N north; P location of Polyergus nest; F location of Formica nest.

Table 1 .
Comparison of slave-raid frequency with and without removal of scouts