lnsectes Sociaux, Paris Masson, Paris, 1983 1983, Volume 30, n ~ 4, pp. 384-401

A BEHAVIORAL STUDY OF THE PRIMITIVE NOTHONYR2~ECIA MACROPS CLARK B, HOLLDOBLER (1) and R.W. TAYLOR (2) (I) Department ol Organismic and Evolutionary Biology, MCZ Laboratories, Harvard University, Cambridge, Mass. 02138, U.S.A. (2) Division af Entomology, C.S.LR.O., Canberra City, ACT 2601, Australia. Re~u Ie 18 octobre 1982. Accept6 le 10 avril 1983.

SUMMARY

1. Nothomyrmecia macrops workers forage individually; no recruitment communication or group retrieval of prey has been observed; prey items measure tess than 4 mm in length ; huntresses grasp live prey with the mandibles and forelegs and sting it to death. 2. Field and laboratory observations confirm that ~vorkers from neighboring nests can forage on a single tree 'without antagonism. Alien conspecifics might be attacked, however, if attempting to enter foreign nests. 3. Field and laboratory observations suggest that chemicaI trai!s do not play an impor- tant role:during homing by foragers, indeed they might not be produced. Mthough strictly nocturnal, Nothomyrmecia seems to navigate primarily by visual cues, possibly using the canopy silhouette overhead~ However, chemical markers might be important in localization and recognition of nest entrances. 4. No antagonistic behavior has been observed beZ'ween foragers from different colonies, yet in a laboratory test nestmates appear to shq',v a greater tendency to' cluster than non-nestmates. Wor&er interchange between neighboring laboratory nests ,was observed and experiments suggest that it is facilitated by loss of the queen in one colony. These results explain in part the findings of WAa~ and TAYLOR (1981) that Nothomyrmecia nestmates are not ~ways full sibs. 5. All records to date indicate that mature Nothomyrmecia colonies are monogynous, but can be founded by pleometrosis (TA~LOR, 1978). : In an experimentally assembled group oh I5 ~vor..kers ~vith 2 queens (neither of vghich was the mother of any of the worker~ i one queen exhibited dominance behavior to*,vards the other. The subordinate was finally expelled by the 'workers. 6, Nothomyrmecia employs chemical alarm communication when other ant attempt to enter its nests. The mandibular gland secretion is an effective close-range alarm . Dufour's gland secretions also elicit attraction, but the react more slowly. Secretions from the pygidial gland appear to function as an atarm-repelle,at substance. BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 385

ZUSAMMENFASSUNG

Fine verhaltensbiologische Untersuchun9 der primitiven Ameise Nothomyrmecia macrops Clark, 1. Nothomyrmecm macrops Arbeiterinnen furagieren einzeln ; Rekrutierungsvers~ndigung oder Eintragen yon Beute in Gruppen wurde nicht beobachtet. Die Beutestiicke sind gew.Shrdieh nicht Nnger als G.4 cm. Die Jggerinnen packen die lebenden Beuteinsekten rnit den Mandibeln und Vorderbeinen und jede Beute 'wird yon der Ameise gestochen. 2. Beobachtungen im Freiland und Laboratorium bestSitigen, dass Arbeiterinnen yon benachbarten Nestern am selben Baum Futter suchen kSnnen. Im Furagier-Areal ~urde kein antagonistiches Verhalten z"Mschen gleichartigen Fremden beobachtet. Gleichartige Arbeiterinnen k6•nen jedoch angegriffen werden, wenn sie in ein fremdes Nest eindringen. 3. Chemische Spuren spielen offensichtlich keine ~vesentliche Rotle bei der Heimfinde- Orientierung. Obgleich Nothomyrmecia bei Nacht furagiert, scheint sie sich haupts~ehlich visuell zu orientieren. Erste Versuehe deuten darauf hin, dass das "Baumdach-Muster" dabei eine ,wesentliche Rolle spielen kSnnte. Am Nesteingang 2<6nnten jedoch auch chemische MarJ~en entscheideade Erkennungs-und Orientierungshilfen darstellen. 4. Die Nesttreue von Furagier-Arneisen wird dutch die genaue Heimfinde-Orientierung und dutch eine Bevorzugung yon Nestgenossen gesichert. Arbeiterinnen vom selben Nest zeigen ira Versueh eine gr:Sssere Tendenz in dichten Gruppen zu sitzen als frerade Nothomyrmecia Arbeiterinnen. Dennoch haben wit ira Laboratorium Arbeiterinnen- Austausch zwischen Kolonien beobachtet. Die Ergebnisse mehrerer Versuche deuten darauf hin, dass der Verlust der K6nigin in einer Kolonie das Ab'wandern yon Arbeit- erinnen begfinstigt. Diese Ergebnisse erklfiren teiI'weise die genetischen Befunde yon WARY und TAYLOR (1981), wonach Nothomyrmecm Nestgenossen nicht iraraer voile Geschwister sind. 5. Alle bisherigen Ausgrabungen yon voli ent,wickelten Kolonien (10) deuten darauf hin, dass Nothomyrmecia in monogynen Kolonien lebt. Neue Kolonien :k6nnen jedoch pteometrotisch gegrfindet ~werden (TAYLOl~, t978). In einer experimentell zusarnmengesetzten Gruppe bestehend aus I5 Arbeiterinnen und zwei K/Sniginnen (keine davon ~war die Mutter der Arbeiterinnen), zeigte eine K,fnigin Dominanzverhalten gegenfiber der anderen. Die untergeordnete KSnigin wurde schliesslich yon den Arbeiterinnen ausgestossem 6. Nothomyrmecia setzt ein chemisches Alarm-Komraunfkationssystem ein, ~wenn Arneisen anderer Arten versuchen in ihr Nest einzudringen. Unsere Versuche zeigen, dass das Mandibeldrfisen-Sekret ein besonders effektives Alarrapherornon darstellt. Auch das Sekret dec Dufourschen Drfise 1,Sst Attraktion aus, jedoch ist die Reaktion tier Ameisen lang- saraer. Das Sekret der Pygidialdrfise scheint als A1arm.Repetlent-Substanz gegen andere Ameisenarten eingesetzt zu werden.

INTRODUCTION

The Australian ant Nothomyrmecia macrops (CLARK) is arguably the most primitive li~ng formicid, and therefore a species of special interest to systematic and beha~oral biologists (WILSON, 1971). Its original description was based upon two damaged worker specimens originating from southern Western Australia and coI!ected in the Summer of 1931-32 (CLARK, 1934). Despite much effort Nothomyrmecia was lost to prospective colIectors, and 386 B. HOLLDOBLER AND R.W. TAYLOR to science, for a further 46 years, until rediscovered by TAYLOR ill 1977. This resulted in the collection and study of captive colonies, initiating a major program of field and laboratory research conducted by TAYLOR and others. Some results of this ~r have been published (TAYLOR, 1978 ; H/fLLDOBLER and ENGEL, 1978 ; KUGLER, 1980 ; WARD and TAYLOR, 1981 ; WHEELER, WI~ELER and TAYLOR, 1980), and other studies are in progress in Australia, Europe and North America. We report here the results of further behavioral studies on communication and colony distinctiveness in Nothomyrmecia~ The single presently known population of N. macrops resides in Euca- lyptus oleosa-dominated mallee woodland near Poochera (32 ~ 43' S ; 134 ~ 50' E), southwest of Ceduna, South Australia, about 1 000 km east of the original collection site. It is more extensive than previously supposed but all its known loci lie veithin a 1 km radius.

MATERIAL AND METHODS

Our research ,was conducted during a brief field visit to the Poochera area in late September and early October 3980, supplemented by Tater experiments in Canberra, using three colonies excavated then (and designated A, B and C), along ~vith a number of captured foraging ~workers and three foraging queens. Each laboratory colony was housed in a stoppered test-tube nest (8 cm • O 2.5 era) fitted 'with a narro~v glass entry tube (O 0.6 era). The interior of each nest was kept moist *with ,water trapped at the bottom of the :tube by a Cotton-~r plug. Early observations showed that dim ambient light did not visibly disturb the ants, so the nests ~were kept uncovered. Each colony had access to a foraging arena, ~vhere small containers of honey~water ~vere provided, and prey ~vas available. This included several species of field-collected flies, and laboratory reared adults. Other nestS, described belo-,v, 'were used for some experimental procedures.

RESULTS

Hunting behavior and homing We noted in the field several phenomena previously reported by TAYLOR (1978). Namely that Nothomyrmecia workers forage nocturnally on Euca- lyptus oteosa trees, leaving the nests only after dusk and returning by d~wn, with all above-ground activity ceasing by approximately 30 minutes after sunrise; that the ants forage individually, with no evidence of recruitment or group retrieval of prey ; that prey objects included various small , generally less than 4 mm long; that foragers observed on tree treks late at night did not return positively in a manner allcrvcing them to be easily tracked, even when carrying prey or fed to satiation with honey (*) and that

(*) We observed three workers carrying insect prey down a tree trunk (~ig. 2) at 22:30; 23:05;23:30 hours. These ~vere lost in the leaf litter at the base of the tree; so it is not ~mcmm if they returned directly to their nest. BEHAVIOR OF THE PRIMITIVE ANT N. MACKOPS 387

nests could be most easily located at first light, when the last returning foragers move very positively down the tree trunks and are readily tracked from there across the ground to their nest entrances. We also confirmed by direct observation the findings of WARD and TAYLOR (1981), that workers from neighboring nests can forage on the same tree (fig, 1).

.,o NesfE ; !m / Fig. 1. -- Map of ~wo neighboring nests of N. macrops. Foragers:from both nests had simultaneously visited the 11 Workers same tree, as indicated by their homing routes to the nests at da~n ; 17 ~vorkers ~vere counted returning to nest I and tl workers to nest II.

Abb. 1, -- Lage zweier N. macrops Nester. Arbeiter yon beiden Neste~ fura- Ne't ~~sTree gierten am selben Baum. 17 Arbeiter kehrten zu Nest I zurfick und 11 Arbeiter zu Nest II.

In addition we observed in the field a phenomenon of great interest noticed previously by TAYLOR and WARD. Foragers seem very definitely to be more active at times when temperatures are low (5-10 ~ C), and are more difficult to locate on warmer nights. We suggest that low temperatures hamper the escape of potential prey encountered in the tree tops by Notho. foragers, increasing their hunting success. We atso noted that other diurnal ant species continue to forage actively on warm nights until much later than on cold ones. NotHomyrrnecia might thus encounter fewer aggressive competitors when abroad on cold nights, and this could contribute to the adaptiveness of its foraging behavior. We tracked foraging workers in low canopy, ~vhere they searched small apical twigs in succession, thus systematically examining leaf Clusters for prey, We 'were unable to witness prey-capture in the field. It was studied in the laboratory. Nothomyrmecia foragers were observed usually to move quite slowly. However, when a huntress came within 1 to 2 cm of a potential prey item she suddenly opened her mandibles and rushed forwards, pursuing the prey with very fast, short zigzag movements. We observed Drosophila adults being captured by workers. In a typical encounter the ant simul- taneously sprung shut her mandibles and grasped the fly ~r her forelegs 388 B. HOLLDOBLER AND R.W. TAYLOR

(fig. 2). The prey was then stung (fig. 2) and almost instantly immobilized. The whole procedure usually lasted less than 2 seconds. Booty was normally carried back to the nest and through the entrance tube to the inner chamber. Often, however, it was dropped while the ants continued hunting. This occurred several times just outside a nest entrance.

Fig. 2. -- (a) A N. macrops foraging ,wor- 'ker photographed in the field 'while retrie~ing a freshly captm'ed ,wasp. (b) A Nothomyrmecia 'worker with a freshly caught Drosophila just cap- tured as prey, photographed in the foraging arena of a laboratory nest. Note how the ant has grasped the fly with the forelegs. (e) A Nothomyrmecia huntress hol- ding a freshly caught Drosophila in her mandibles, photographed just after catching a second fly ,with her forelegs, and in the process of stinging it.

Abb. 2. -- (a) Eine N. macrops Axbeiterin tr~igt eine frisch erbeutete 3Nespe ein (im Freiland photographiert). (b) Eine Nothomyrmecia ,tggerin hat soeben in der Furagier-Arena des Labornestcs eine Drosophila Fliege gefangen. Die Fliege .wird mit den Mandibetn und Vorderbeinen fest- geha/ten. (c) Eine Jiigerin, die eine fdsch erlegte Drosophila Fliege in den ! :!:i::/Yi~? Mandibeln h~ilt, f~ingt eine zweite Fliege mit den Vorderbeinen und macht sic mit einem, Stich be~ve- gungslos.

In:the field the return of foragers across the ground, from trees to nests, was not markedty affected when the leaf litter and soil being traversed was disturbed by our manipulations. In the laboratory homing ants were offered artificial trails drawn with secretions from the hindgut, Dufour's gland, poison gland and pygidial gland (fig. 3), No trail-following behavior was elicited. We conclude from these observations tl~at chemical trails do not play a pre- dominant role in homing orientation, and that Nothomyrmecia probably does not lay such trails. It is possible, however, that nest entrances are chemically marked (see below). In several field experiments we observed ants experiencing difficulty in BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 389

DG

Fig. 3. -- Schematic illustration of the major exocrine glands .which could be sources for in Nothomyrmevia Honkers. Detailed descriptions of the morpho- logy of the glands ~vill be published elsewhere; MG : mandibular gland ; MPG : meta- pleura[ gland; PoG: poison gland ; DG: Dugout's gland; PG: pygidial gland; HG : hindgut, Abb, 3, -- Schematisehe Darstellung der wichtigsten exakrinen Drtisen, die als Pheromon- driisen in Frage Nommen, Eine genaue Beschreibung der Morphologie der Dr~isen 'wird an anderer Stelle ver6ffentlich. MG : Ma~dibeldrtise ; MPG : Metapleuraidriise ; PoG : Giftdriise ; "'DG : Dufoursche Driise ; PG : Pygidialdr~se ; HG : Enddarm, locating nest entrances ~,vhen within 5 to 20 cm of them, ~vhen the surrounding leaf litter and soil had been artificially disturbed. In these circumstances they moved in searching loops ~vithin a radius of about 10 to 20 cm until finally locating the entrance hole. We noticed moreover that homing foragers became disoriented when their overhead view was obscured. This suggests that visuat orientation, and possibly canopy navigation (HSLLDOBL~R, 1980), might be the major mechanism used by homing Nothornyrmecia workers. To a human observer the canopy pattern is strikingly clear and sharply contrasted against the sky, even on very dark nights. Returning Nothomyrmecia foragers seem normally to travel less than 4 meters across the ground from the trees, the radiating canopy pattern of which could well provide reference for precise navigation on the ground. Simple geotaxis could enable effective orientation by the ants while foraging on tree trunks or canopy. Recent sensory physiological investigations in Switzerland indicate that the of Nothomyrmecia are excellently adapted for the perception of silhouettes in low light intensity (R. WEANER, pers. comm.). The directional accuracy of Nothomyrmecia workers returning to their nests is impressively precise; so much so that we believe the chance of an individual encountering and entering an alien nest must be very lgw, and that this factor must contribute positively to colony fidelity.

Colony fidelity and nest entrance perception

The precision of colony fidelity 'was investigated in the following laboratory experiments using two colonies of Nothomyrmecia given access to opposite ends of the same large plexiglass foraging arena (47.5 cm • 23.7 cm). Each 390 B. HOLLDOBLER AND R.W. TAYLOR was housed in a standard test-tube nest fitted with a long entrance tube (16 cm • O 0.6 cm), and each contained a queen, 43 workers and advanced larval brood. The gaster of every worker was marked with a dot of Testors | PLA paint -- yello~v for colony A, red for colony B. The arena floor was covered with a sheet of paper and divided at first by a glass wall, which limited foragers from each colony to half the arena space. This partition was removed after 4 days when the experiments commenced. Foraging activity was intermittently observed on 4 nights, during which each colony had between 3 and 17 foragers abroad at any given time. The foraging areas of both colonies soon overlapped, to occupy essentially the whole arena. When alien workers met they usually parted very quickly; encounters 'with nestmates were similarly brief. During an 8-day period of intermittent inspection no workers were found in the nest of an alien colony : foragers remained strongly bound to their own nests. The orientation cues used by foragers were investigated using the same experimental arrangement. Nest A 'was initially positioned at the end of the arena nearest a laboratory window, and nest B towards an opposite blank wall. A black cardboard surround (12 cm square), pierced by the entrance tube, was placed in front of nest B. The experiments were conducted in the afternoon ; since we had observed previously that ants artificially displaced from their nests to the foraging arena during the day would return to the nests, allowing easy observation of homing behavior. Ten ants were displaced in turn to the centre of the foraging arena and observed for a maximum period of 15 minutes, during which the frequencies of 2 orientation events were recorded: 1) the crossing of a semicircle (r = 6 cm) drawn around the nest entrance, to define a "home area ", and 2) movement fully into the nest entrance. Multiple entry of the home area or nest entrance by some ants occurred ; each such event was recorded separately, The results of this experiment can be summarized as follows: 70 % of colony A workers entered nest A within 15 minutes, and none entered nest B ; 50 % of colony B workers entered nest B, while one briefly entered nest A, only to emerge seconds tater. The arena was then turned through 180~ and the cardboard surround transferred to nest A, reversing the visual environment of each colony. The paper cover on the arena floor was retained, so that any possible chemical markers deposited on it by the ants remained in context relative to the nest entrances. In this situation 30 % of colony A workers entered colony A, and 40 % entered colony B; 20 % of colony B workers entered colony B, while 30 % entered colony A. All individuals which entered a foreign nest re-emerged within the 15 minutes observation period. BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 391

The ants clearly continued to navigate using the previously learned visual cues, and thus frequently returned to the alien nest opposite their own (though now in its original position). Although there was a high frequency of entry into alien home areas, well in excess of actual nest entry, ants in the vicinity of foreign nests clearly hesitated to enter them, even though they had visually navigated to the vicinity. These results suggest that nest entrances are marked around with colony- specific scent signals by their resident ants. This possibility is further supported by a pilot experiment in which we covered an area of 10 • 12 cm around a nest entrance with fresh paper, thus blanketing any possible scent marks. Homing ants then took considerably longer than usual to find the nest entrance. The hindgut provides a likely source of chemical markers of the type implied. We observed that paper near the entrance of a nest containing ants fed several days previously with the stain Azurubin-S was marked xvith tiny red spots. Dissection of 2 workers revealed that the contents of their rectal bladders were stained similarly. Although antagonistic confrontations between alien workers were not observed in the foraging arena they were seen repeatedly between resident ants and aliens which had entered the wrong nest. The intruder was grasped with the mandibles of the resident and both ants often flexed their gasters as if attempting to sting the opponent (fig. 4 a). Such encounters lasted up to several minutes, and usually resulted in eviction of the intruder. On occasions no immediate fighting was witnessed, but injuries (such as loss of limb extremeties) to known alien workers were sometimes observed several days afterwards. These presumably resulted from delayed fighting. On other occasions no discrimination by resident ants against intruders was evidenced. Our 3 experimental colonies differed markedly in this regard. Colonies A and B accepted foreign workers more readily than did colony C (which 'was the largest, including 84 workers at collection). Colony C workers invariably attacked aliens, whether they came from the pool of workers originally collected as foragers, or from one of the other colonies.

Colony distL~etiveness Observations reported above suggest that aggressiveness towards conspecific aliens is stronger near nest entrances than it is between alien foragers away from nests. TAYLOR (I978), and WARD and TAYLOR (1981) have reported that alien N. macrops workers collected from scattered trees show little aggressiveness when mixed. The following experiments ~r designed to compare levels of attraction and aggression observed between nestmates with those seen between aliens, Four series, each containing 10 Nothomyrmecia individuals were established in separate gypsum-floored Petri-dish nests (0 = 9 cm). They 392 B. HOLLDOBLER AND R.W. TAYLOR

Fig. 4. -- (a) ~wo N. macrops 'wor- kers fighting after one of them ~zas experimentally in- troduced into a foreign labo- ratory nest. (b) Dominance behavior bet- .ween two N. macrops queens, in a laboratory nest. The do- minant individual s t a n d s above the crouching subor- dinate.

Abb. 4. -- (a) Zwei N. macrops Arbeiterinnen kimpfen, nach- dem eine yon ihnen in einem Versuch in das fremde Nest gesetzt wurde. (b) Dominanz-Verhalten, z~i- schen z~vei N. macrops K6ni- ginnen. Die dominante K6ni- gin steht fiber der sich duck- enden subordinierten KSnigin.

included respectively: I and II, foraging workers collected randomly on separate well-spaced trees ;III, workers collected foraging on a single tree, and IV, 'workers taken from an excavated colony. The nests were inspected 4 times daily for 5 days, and those ants huddled together in groups ~ere counted and scored. The results are summarized in figure 5. Ants of series III and IV clearly had a greater tendency to sit close together (huddle) than did the randomly collected foragers of series I and II. We assume that most workers collected foraging on a single tree (III) were nestmates and therefore behaved very similar to those ants taken directly from the nest (IV). On the other hand, foragers collected on different trees (I, II) most likely belonged to different colonies and therefore tended to avoid each other at least during the first two days of the experiment ; no aggressive interactions were observed. These results suggest 1) that nestmate recognition occurs in BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 393

Ooy t 0Qy 2 O~y 3 Oay 4 Oay 5

I0 - A 9 A X A >O< ..:::c i c:~ 7- A A X

Jl, XX ,I, O

X ~' A XX A 0 A O

X & X 0 X Q 0 XX A

X 0 X | 0

~: 2 .co X 0 0 X O X t O0 O0 O0 lll~e O0 0 0

I ~ ]2~2 III ~

Fig; 5. -- Tendency of N, macrops ~or.kers to sit close together (huddle): Each group contained 10 workers and 4 times a day we counted how many ~vo~kers were sitting closely together. I, II : foragers collected from different trees ; ItI : foragers collected from one tree ; IV : workers taken from one colony. Abb. 5. -- Tendenz yon N. macrops Arbeiterinnen, in Gruppen zusammenzusitzem Jede Versuchsgruppe enthielt 10 Arbeiterinnen und 4 real am Tag veurde gez~hlt, ~vieviele Arbeiterinnen dicht zusammensassen. I, tI : Arbeiterinnen ~vurden yon verschi~xlenen Baumen gesammelt; III: Arbeiterinnen ~r yon einem Baum gesammelt; IV: Alle Arbeiterinnen stammen yore selben Nest.

Nothomyrmecia, and 2) that colony integrity is not achieved solely by the site fidelity of foragers during home-range orientation. Some laboratory observations suggest that worker interchange between nests in the field is possible, and might occur. In the two-colony arena experiments discussed above no interchange accurred between colonies A and B at first. After about 3 weeks, however, we found the queen and 3 workers dead in nest B. Five days later 2 colony B workers were found in nest A, and on following days workers from A were seen in B. For the next 2 'weeks both nests were occupied by workers from both colonies, but the queen and approximately 2/3 of the combined worker force at~ays occupied nest A. Quantitative studies were not conducted, but no differential treatment of adopted versus resident workers was evidenced. These observations clearly indicate the need for further laboratory and field studies of 'worker interchange between neighboring nests and the effect upon 394 B. HOLLDOBLER AND R.W, TAYLOR

it generated by loss of queens from established colonies. The following experiment was conducted to this end. After initial collection 12 workers taken from a single tree were housed in a test-tube nest (D), while t4 workers and a queen collected from various trees were established in a identical nest (E). The two groups were kept separate until several 'weeks later, when the experiment was initiated by placing both in the same foraging arena (36 x 51 cm). By day 2 five group D workers had moved into nest E, and after 4 days only 1 ant remained in nest D. Some interchange between the nests continued -- 1 to 4 ants usually being present in nest D when it 'was randomly checked -- although by far the largest 'worker group was always in nest E, with the queen. It can be reasonably assumed that the original nest D ants, collected from a single tree, were more nearly homogeneous than those in nest E, which were collected froth several trees. The evident attraction of nest E was presumably related to the presence there of a queen, albeit one associated with a relatively heterogenous group of 'workers. Although the resulting artificially created colony initially lacked brood (the queen later laid eggs), and the workers did not exhibit the same level of beha- vioral activity as those in fully constituted natural colonies, they did show increasing mutual attraction and "allegiance" to the queen. We believe that these results support a hypothesis that the presence of a queen is of major importance to the maintenance of colony cohesion in Nothomyrmecia.

Number of queens in colonies The available evidence suggests that mature Nothomyrmecia colonies are monogynous., A total of ten such colonies had been excavated, either before or during our studies. Al~ contained a single queen, except one from which no queen was recovered (TAYLOR, 1978, and unpublished data). On the other hand TAYLOR (1978) found evidence that two or more young queens can associate for colony foundation, since the only founding colony ever excavated contained two queens, and these later cooperated in brood rearing, WARD and TAYLOR (1981) suggested that one of the explanations for the high levels of observed in Nothomvrmecia colonies which they studied could be that some workers might become inseminated and act as supplementary reproductive females. Several of our observations are relevant to these three propositions. We dissected 6 queens, the 3 from our excavated colonies, and 3 collected foraging on trees. They each had between 8 and 10 ovarioles, and all were inseminated. WARDand TAYLOR (1981) reported 9 to 10 ovarioles in 5 queens, We also dissected 67 'workers, 45 collected with colonies, the remainder taken as foragers. None were inseminated. All had a spermatheca and well developed ovaries, with between 4 and 7 ovarioles. A majority of 49 had BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 395

6 ovarioles. Corpora lutea were noted in approximately 30 % of the dissections. Thus, while 'workers seem to possess reproductive potential this seems not to be realized in nature, perhaps because they do not become inseminated. The feeding of (presumably unfertilized) worker-laid eggs to larvae and adults (including mother queens and alate reproductives of both sexes) has been seen in laboratory nests by TAYLOR (1978). This does not preclude the possibility of male production from worker-laid haploid eggs, as pointed out by TAYLOR. The severe shortage of live individuals allowed us only one experiment to test whether two queens can co-exist in a mature N. macrops colony ; the results are however worthy of report. Fifteen 'workers from colony C were separated from their parent nest for 25 days in a petri-dish nest (0 = 9 cm) connected to a foraging arena (21 • 30 cm). ~wi~ queens, which had been kept together for over 4 weeks since their collection as foragers on separate trees, were then introduced into the nest. These were individuaUy recognizable by wing-stub differences. Although colony C, when intact, had shown consistent hostility towards aliens (see fig 4 a), no initial aggression was observed. After about 2 weeks, however, the 2 queens became involved in a peculiar behavior, which had not been observed when they 'were cultured without workers. One stood above the other, even stepping onto it, while pointing downwards with her head. The loxver individual adopted an unusual crouching posture (fig. 4 b). This behavior was later observed frequently, with the same queen (designated ~) always standing above her partner (designated ~). This behavior closely ressembled the dominance interactions described for Leptothorax ants by COLE (1981), and for the wasps Potistes and Mischo- cytharus by PARDI (1948), WEST (1967) and JEANNE (1972). No fighting was observed between these queens, and workers were not initially involved with them. Several days later, however, a worker was seen pulling the ~ queen by a leg, and two days later she was seen outside the nest, but had returned the following day. By now workers were observed repeatedly pulling and dragging her, and she was rescued from the colony for use in other experiments. Both queens were shown by later dissection to have 8 ovarioles, and to have been inseminated. Similar dragging of queens by workers had been observed previously by TAYLOR in a nest containing several queens with workers, all collected separately as foragers. These results suggest that a dominance order is established among multiple queens in Nothomyrmecia colonies and that the workers may finally expel subordinate queens. This could be the process by which colonies founded by several queens become monogynous.

Colony defense and alarm communication Foragers returning to our test-tube nests usually had to pass one or two guard ants standing in the entrance tube. Conspecific aliens were either 396 B. HOLLDOBLER AND R.W. TAYLOR attacked there, allowed to enter the nest chamber and later driven out, tolerated and finally adopted. We cannot explain these differences, but note that colonies A and B were more tolerant than colony C. Our attempts to introduce single individuals of other ant species into nests involved australis Erichson, a species of Cerapachys, and one of Camponotus. These always elicited attack by Nothomyrmecia guards, followed almost instantly by an aggressive reaction involving from 2 to 5 additional workers. These moved into the entrance tube, and sometimes rushed from its exit with mandible widely opened (fig: 6). This group response implies that Nothomysmecia employs some form of alarm communication.

Fig. 6. -- (a) An alarmed Nothomyrmecia guard in the entrance of a laboratory nest. (b) An Ambtyopone australis in the laboratory nest being attacked by the N. macrops guard (Note: These species are not known to occur sympatrically in the field). Abb. 6. -- (a) Nothomyrmecia W~chterin im Nesteingang, (b) Eine Amblyopone australis Arbeiterin, die dem Nothomyrmecia Nesteingang zu nahe .karn, ,wird yon der N. macrops 'W~fichterin angegriffen. BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 397

All castes of N. macrops stridulate ~hen prevented from moving freely (TAYLOR, 1978 ; MARKL and TAYLOR, unpublished), but our observations suggest that plays no part in the communication of general alarm.

Chemical communication, utilizing releaser pheromones, is widely involved in alarm communication among ants. We therefore tested the responses of Nothomyrmecia workers to secretions of the major exocrine glands. Tests were conducted in a 9 cm diameter petri-dish cell containing a Nothomyrmecia colony, and fitted 'with 5 small, equally spaced holes around its side ,wall, through vchich an applicator stick could be inserted. A grid marked on the floor of ~he ceil enabled measurement of the distances between responding ants and the chemically loaded appli- cator. For each test one gland of a kind was dissected from a freshly ~tled ant, crushed on the tip of a hard~vood applicator stick, and immediately inserted into the test ceil, thus exposing to the ants, in relatively still air, any secretions present. Records ~vere made of the number of ants 'which (1) promptly waved their anterm~e towards the" applicator, (2) opened their mandibles; or (3)were attracted towards the applicator during the first 30 seconds. Either before or after the test with glandular secretions the ants' reactions to,yard an applicator loaded with ~ater served as Control Approximately 3 minutes elapsed betaveen the presentation of control or test applicator respectively. The experiments 'were conducted ,with two colonies of equal size (43 ,workers). The glands were dissected out of workers ~hich did not belong to the test colonies (previous pilot experiments did not indicate colony specificity in the alarm response evoked by the glandular secretions). After each experiment at least 30 minutes elapsed before the colony ~.vas tested again. The results are summarized in table L

Table L -- Mean number of ants (-+- standard deviation) responding to glandular secretions (for additional explanation see text).

Tabelle I. -- Mittelwerte der Anzahl von Ameisen (-+- Standardabweichung) die auf verschiedene Driisensekrete reagierten (fi2r ,weitere ErklSrungen siehe Text).

Antennee Mandibles Attracted waved opened

Control n = 11 1.5 +_ t.4 0;2 -+- 0.4 0.4 ~ 0.7 1 mandibular gland n = 11 6.6 • 3.0 3.4 __+ 2.3 4.8 ~ 1.8

Control n = 12 0.4 ~ 0.7 0.I --+-_ 0.3 0.i _ 0.3 1 metapleural gland n = 12 1.3 ~ 1.4 0.3 -+- 0.7 0A ~ 0.7

Control n = 12 0.7 ~ 1.2 0.1 • 0.3 0.2 ~ 0.4 Pygidial gland n = 12 3.5 ~ 2.7 1.2 __+ 1.2 2.8 ~ 1.5

Control n = 10 0 0 0 Hindgut n = 10 0.8 ~__ 1.0 0.2 ~ 0.4 0.1 -+- 0.3

Control n = 9 0.1 ___ 0.3 0 0 Poison gland n = 9 1.0 __+ 1.4 0.3 -+- 1.0 0.4 ~ 0.7

Control n = 8 0.8 • 1.2 0 0.3 __+ 0.7 Dufour's gland n = 8 4.9 ~ 2.0 1.9 -4- 1-6 3.4 ~ 2.1 398 B. HOLLDOBLER AND R.W. TAYLOR

Mandibular gland extracts elicited the strongest responses: most ants reacted within 5 seconds; the signal had apparently faded after 30 seconds; ants more than 5 cm away from the source of glandular extract waved their antennae, and some of them moved slowly towards the applicator. Most ants within 2 to 3 cm Of the mandibular gland extract rushed in random zig zag Ioops towards the applicator, with wide-opened mandibles. This behavioral response closely resembles that described by WILSON and BOSSERT (t963) as the typical close-range alarm reaction in ants. Workers were also attracted to Dufour's gland extract. The reactions were slower, but the signal lasted longer than that from mandibular gland extract. Ants were attracted from as far away as 6 cm, and the open-mandible aggressive response was less pronounced. Pygidial gland secretions also released some attraction and aggressive responses. These results suggest that secretions from the mandibular gland, in particular, function as an effective close-range alarm pheromone. It is possible that, in a more escalated defense reaction, such as when an intruder is stung by a guard, Dufour's gland secretions might attract more distant workers, enhancing colony defense. We also tested glandular secretions for a possible repellent effect on a nocturnal Camponotus species from the Poochera site, and two other species of that collected near Canberra. Positive results were obtained only with pygidial gland material, a whitish liquid easily discharged from the gland reservoir by slightly stretching the in tersegmentaI membrane between abdominal tergites VI and VII (tt6LLDOBLER and ENGEL, 1978). We suggest that the pygidial gland in Nothomyrmecia produces an alarm-defense secretion, perhaps employed by foragers upon encounter 'with other ant species. We have witnessed in the field attacks on Nothomyrmecia foragers by Camponotus and workers. Nothomyrmecia workers actively avoid contact with workers of Camponotus consobrinus (ERIcHS0N) introduced experimentally into observation nests. On these occasions, however, Camponotus which had attempted to grasp Nothomyrmecia workers in their jaws showed a clear-cut repellent response, characteristics of those we have observed in other ants,

DISCUSSION

On the basis of altozyme variation data and the estimation of genetic relatedness of Nothomyrmecia workers from the same nest, WARD and TAYLOR (1981) concluded that nestmates are not all full sibs (mean coefficient of relatedness 0.371 ---+ 0.038). As they pointed out, several factors could be responsible for the low level of genetic relatedness among nesmates : multiple inseminations of the queen, polygyny (including reproduction by mated workers), and worker interchange between neighboring nests. Since it was BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 399 also found that workers foraging on the same Eucalyptus tree have a mean coefficient of relatedness of 0.172 ~ 0.068, it was concluded that foraging areas between neighboring colonies widely overlap, which could facilitate worker interchange. The results presented here suggest that polygyny (including reproduction by mated workers) is very unlikely in mature Nothomyrmecia colonies. However, the incorporation of workers from foreign conspecific colonies appears to be possible, especially when the queen has been lost from one of the colonies. Intraspecific aggression seems to vary greatly with the situation and context. In the foraging area or on "neutral" ground foreign conspecifics show avoidance behavior but no fighting. Inside the nest. however, conspecific foreigners can be attacked and driven out. In particular, the largest of our three experimental colonies behaved aggressively toward conspecific drifters. Perhaps this colony was either a very homogeneous group (consisting primarily of full sibs), and therefore more discriminatory against outsiders, or its worker force had reached an optimal size and was therefore closed to further worker adoption. The laboratory experiments suggest that workers which have lost their queen show an increased tendency to drift to neighboring nests. Furthermore, our results indicate that worker groups that have lost their queen will more readily accept a foreign queen and will develop an allegiance to her. We also found evidence that queens housed in one nest-tube will establish a dominance order and that the workers (which in our experiment were not the offspring of either queen) wilt subsequently behave aggressively towards the subordinate queen, leading eventually to her expulsion. From this we might speculate that queens of Nothornyrmecia establish " personal" nest territories with their own worker entourages, and that the very scattered and relatively small Nothomyrmecia populations might somehow represent a unicolonial system of oligogynous organization (H6LLDOBLER, 1962 ; BUSCHINGER, 1974; Hi3LLOOBLER and WILSON, 1977). The brachyptery of virgin queens of Nothomyrrnecia (TAYLOR, 1978) could well relate to the above considerations. The queens' wings do not appear to be very suitable for flying and therefore virgin queens probably do not conduct nuptial flights and thus engage in long-range dispersal. We hypo- thesize that virgin queens, ready for mating, remain on the ground or climb trees and from there call tomes by releasing sex pheromones. The males have very well developed wings and are presumably good flyers. It is easily conceivable that some of them might even fly in from more distant populations, although we do not yet know how far apart neighboring patches of Nothomyrmecia populations are. In any case, freshly mated females do not appear able to diperse far from the area 'where they were raised. This means that the population has a considerable degree of viscosity which in turn should favor worker inter- 400 B. HOLLDOBLER AND R,W. TAYLOR change. Workers losing their own mother queen night still enhance their inclusive fitness by helping a reproductive sister, aunt or other related queen. The finding by WARD and TAYLOR (1981) that colonies of Nothornyrmecia are not simple family units, consisting entirely of full sibs, are partly explained by our results on worker interchange. In addition the possibility remains that multiple insemination of the queens could atso account for this phenomenon. Our results suggest a strong correlation between the presence of an active queen in the colony and the fidelity of workers to that colony. Yet, the colonies do not necessarily include 'workers which are full siblings. This indicates that the queen's attractiveness has a unifying effect even on a genetically diverse worker group. It remains to be determined ~vhether there are nevertheless subtle differences in recognition and treatment among differently related 'workers and ~vhether the queen's chemical and behavioral dominance might mask these differences (H~SLLDOBLER and MICHENER, 1980).

ACKNOWLEOGMENTS. -- We thank RJ. BARrEtt, F. H~LLDOBLER, E.S. LOCKIE and H. MARKL for assistance in the field, and for many discussions throughout this study. N. FRANKS and E.O. WILSON read and commenfed on the manuscript. This ~ork 'was supported in part by grants to Bert Ir~iSLLI3OSLER from the National Geographic Society, National Science Foundation {U.S.A,) and lohn Simon Guggenheim Foundation. and by facilities provided by the Divisionof Entomology, Commor~wealth Scientific and Industrial Organi- zation, Canberra. Australia.

References

BUSCmNC~R A,, 1974, -- Monogynie und Polygynie in Insektensoziet~iten. In G.H. Schmidt (ed.), Sozialpolymorphismus bei Insekten, Wissenscha#liche, Verlagsgesellschaft MBH, Stuttgart.

CLARK J., 1934. -- Notes on Australian ants, with description of new species and a new genus. Mere. Natl. Mus. Victoria. Melbourne, 8, 5-20. COLE BJ., 1981. -- Dominance hierarchies in Leptothorax ants. Science, 212. 83-84.

H~LLI3OBLeR B., 1962. -- Zur Frage tier Oligogynie bei Camponotus ligniperda und Campo- notus hercuteanus L. (Hym. Formicidm). Z, ange~v. Entomotogie, 49, 33%352. H6LLDOBLER B., 1980. -- Canopy orientation: a new kind of orientation in ants. Science, 210, 86-88. H6LLt~O~LmZ B., ENCEL H., 1978, -- Tergal and sternal glands in ants. Psyche, 85, 285.330. H~ILLt~OBtZR B., MlCra~NaR C,D., 1980. -- Mechanisms of identification and discrimination in social , In Markl H, (ed.), Evolution of Social Behavior: Hypo- theses and Empirical Tests, pp. 35-58. Dahlem Konferenzen 1980. Weinheim : Verlag Chemic GmbH. H~LLI~OBLER B., a/qILSON E.O., 1977. -- The number of queens: an important trait in ant ex~oIution. Naturwissenscha#en, 64, 8-t5. JEANNI~ R.L., 1972. -- Social bioIogy of the neotropical wasp Mischocytharus drewsenL Bull. Mus. Comp. Zoology, Harvard, 144, 63-150. KUGLER C.. 1980. -- The sting apparatus in the primitive ants Nothomyrmecia and Myrmecia. 1. Aust. Ent. Soc., 19, 263-267. PARDI L.. 1948. -- Dominance order in Polistes ~vasps. Physiol. Zoology, 21, 1-13. TAYLOR R:W., t978. -- Nothomyrmecta macrops : a living-fossil ant rediscovered. Science, 201, 979-985. WARD P.S,, TAYLOR R.~N., 1981, ~ Allozyme variation, colony structure and genetic relatedness in the primitive ant Nothomyrrnecia macrops Cla~k. I. Aust, Ent, Soc., 20, 177-183, BEHAVIOR OF THE PRIMITIVE ANT N. MACROPS 40I

W~T M J, 1967. -- Fonndress associations in poHstine ~vasps : dominance hierarchies in the evolution of social behavior. Science, 157, 1584-1585. WHirR E.C., ~REEI.~R J., TAYLOR R.~N., 1980. -- The larval and egg stages of the primitive ant Nothomyrmecia macrops Clarrk. I. Aust. Ent. Soc., 19, 131-137. WILSON E.O., 1971. -- The insect societies. The Belknap Press of Harvard Univ. Press, Cambridge, Mass., 1971. WImsoN E.O., BOSSERT W.H., 1963. -- Chemical communication among . Recent Progress in Hormone Research, 19, 673-716.