STATARY BEHAVIOR IN NOMADIC COLONIES O F ARMY ANTS: THE EFFECT O F OVERFEEDING

Nearctic colonies of the army ant Neivamyrmex nigrescens Cresson (subfamily Ecitoninae) exhibit behavioral cycles consisting of alternating nomadic and statary phases. During the statary phase, a colony remains at the same nesting site and forages irregularly for food. The nomadic phase, by contrast, is characterized by night-long raids and frequent emigrations to new bivouacs. According to Schneirla (1957, 1958), the nomadic phase is triggered by stimulation arising from newly-eclosed callows, and is subsequently maintained by comparable excitation from the developing larvae. Experimental support for brood-stimulation theory stems from studies showing: (1) an abrupt reduction in nomadism after removing a portion of a larval brood (Schneirla and Brown, 1950); and (2) the eclosion of a pupal brood (in the absence of newlyhatched larvae) is indeed sufficient to initiate a nomadic phase (Topoff et al., 1980a). Recent studies have suggested, however, that brood stimulation may in turn depend on the degree of brood satiation. Thus, in a preliminary field study involving food augmentation, Mirenda et al. (in press) was able to halt the occurrence of emigrations during a portion of the nomadic phase in colonies of N. nigrescens. This was followed by more prolonged laboratory studies (Topoff and Mirenda, 1980 a,b) showing that the frequency and direction of nomadic emigrations are indeed influenced by the amount and location of food. This paper reports findings from our continued studies of food augmentation for colonies of N. nigrescens. In previous studies, larval stimulation was reduced by artificially feeding colonies early in the nomadic phase, after callow eclosion. Because an additional

hatched larvae) is indeed sufficient to initiate a nomadic phase (Topoff et al., 1980a).Recent studies have suggested, however, that brood stimulation may in turn depend on the degree of brood satiation.Thus, in a preliminary field study involving food augmentation, Mirenda et al. (in press) was able to halt the occurrence of emigrations during a portion of the nomadic phase in colonies of N. nigrescens.This was followed by more prolonged laboratory studies (Topoff and Mirenda, 1980 a,b) showing that the frequency and direction of nomadic emigrations are indeed influenced by the amount and location of food.
This paper reports findings from our continued studies of food augmentation for colonies of N. nigrescens.In previous studies, larval stimulation was reduced by artificially feeding colonies early in the nomadic phase, after callow eclosion.Because an additional Psyche [Vol.88 goal of the present study was to reduce callow stimulation and thereby delay the onset of the nomadic phase, overfeeding commenced late in the statary period.For the first colony, emigrations were delayed approximately 6 days.In the second overfed colony, we were able to virtually eliminate the nomadic phase, together with all associated patterns of raiding and emigration behavior.

METHODS AND PROCEDURES
This study was conducted during July and August, 1980, in a desert-grassland habitat, 8 km east of Portal, Arizona.The site was chosen because the pavement-like substrate and patchy vegetation provided us with an excellent view of the ants' raiding and emigration activities.Surface soil temperatures averaged 50' C at 1500 hr (MST) and 17OC at 0200 hr throughout the summer (Mirenda et al. 1980).As a result of the severe daytime temperatures and aridity, colonies of N. nigrescens were usually active on the surface only between 1900-0500 hr.
Colonies were located by walking through the study area with gasoline lanterns or miner's cap lamps.Colony no. 1 was found on July 11, at the end of a nomadic phase, and observed nightly throughout its next statary period.During the subsequent nomadic phase, the colony was estimated to contain approximately 80,000 adults and 50,000 larvae.Colony no. 2 was collected on July 14, during its last nomadic emigration, and maintained in the laboratory (see Topoff et al. 1980b for details of the rearing procedure) until the pupal brood was fully pigmented.Prior to release in the field, the colony was culled to contain 4,000 adults and 4,000 pupae.By the next nomadic phase, approximately 4,500 larvae were also present in the colony.This small colony size was chosen for two reasons: (1) to increase our ability to appreciably overfeed the colony; and (2) a laboratory colony of comparable size had previously been released without food supplementation, as part of a study designed to show that laboratory rearing and population reduction do not alter qualitative aspects of nomadic behavior.This colony could therefore serve as a convenient control for our artificially-fed colony.
Food for both experimental colonies consisted of adult and brood individuals of the myrmicine ant Novomessor cockerelli, and workers of the termite genus Gnathamitermes.T o collect Novo-153 messor brood we made use of the panic-alarm behavior that t h i s prey species exhibits when raided by army ants.Accordingly, we released several hundred adult N. nigrescens into the nest e n t r a n c e of Novomessor, and aspirated the larvae and pupae that w e r e removed from the nest by their own adult workers.Whenever colonies were artificially fed, food was given at the start of r a i d i n g in the evening, while the column was within 2 m from their bivouac.If columns emerged from more than one exit hole, booty was p l a c e d a t the front of each ant column.
The raiding and emigration activities of colony no. 1 are s u m m a rized in Table 1.This colony was found on July 11, late i n its nomadic phase.It became statary on July 13, after settling i n t o a kangaroo rat mound (Dipodomys spectabilis).On the third s t a t a r y night, the colony conducted a 3-m long shift to the other side of the mound.A statary shift differs from a nomadic emigration in t h a t it is neither preceded nor followed by raiding.It consists instead of a single, unbranched column, and is presumably caused by a disturbance at the old site.For the next 13 statary days, the c o l o n y remained at the same bivouac, and staged either brief (1-3 hr) or n o predatory raids.On statary day 17, however, the colony conducted a longer shift to an adjacent mound.During the move, we observed that all of the pupae were deeply pigmented, and that a few callows were being transported by mature adults to the new site.As a result of detecting the onset of eclosion, we started artificial feeding of the colony on the next night (July 30), and continued to supply f o o d for a total of six consecutive nights (Table 1).
Each evening, a basal column appeared on the surface s h o r t l y after sunset (1800-1900 h?).As soon as the ants contacted the f o o d , the process of mass recruitment resulted in a sharp increase i n a n t traffic out of the nest.On the days of heaviest feeding, when m o r e than 30 g of booty were provided, the army ants required several hours to transport it back to the bivouac.The colony occasionally put out additional raiding columns later each night, but all c a p t u r e d booty was promptly brought back to the original bivouac, a n d no emigrations occurred.On the afternoon of August 5, the study area received 14 mm of rainfall between 1400-1550 hr.Phase L e n g t h : Because the colony was temporally anchored, J u l y 13 can be considered the first statary day, August 5 the first nomadic day.Thus, the statary interval becomes 23 days (Table 1).According to M i r e n d a and Topoff (1980), the range of statary-phase duration for N. nigrescens in the same study area is 15-19 days, with a modal length of 16 days.This suggests that the minimum delay i n nomadic onset f o r our colony was 4 days.If we use instead Mirenda and Topoffs m o d a l duration, the delay is calculated as 7 days.
Callow pigmentation: Newly eclosed callows of N. nigrescens a r e yellow and a c q u i r e adult-like pigmentation between 7-12 days.Several h u n d r e d callows were collected from the colony during its first emigration on August 5, and compared with preserved samples collected daily from nomadic colonies in previous years.Although this form of visual comparison can not always pinpoint the exact post-eclosion d a y , callows from the artificially-fed colony were substantially m o r e pigmented than those typically collected f r o m other colonies o n the first nomadic night.Our comparison between these callows a n d previously preserved specimens indicated a posteclosion age of between 5-8 days.
Larval size: Several hundred larvae were collected by aspiration from the first emigration.By visual inspection, we separated the 10 largest and 10 smallest larvae and measured them with the aid of a dissecting microscope fitted with an ocular micrometer.The mean length of the large group was 4.0 mm (range = 3.8-4.2mm), a s compared with a mean of 1.5 mm (range = 1.3-1.7 mm) for the small group.When these data are compared with Mirenda a n d Topoffs (1980) graph of larval growth versus nomadic day, they correspond t o a range of nomadic days between 4-6.
The nightly patterns of activity for colony no. 2 and for t h e control colony are summarized in Table 2.For this small colony, we  August 25, we excavated the colony and forced it to shift its s t a t a r y bivouac.This procedure verified that the colony's larvae h a d pupated.Thus, throughout a nomadic phase lasting 14 d a y s , t h e colony conducted only one completed emigration.On 4 n o m a d i c nights no raiding occurred.During the 10 nights in which r a i d i n g took place, the median time for raid onset was 2200 hr, a n d t h e median duration of each raid was 1.5 hr.
The control colony, which was also released from the l a b o r a t o r y at the end of a statary phase, exhibited more typical p a t t e r n s of nomadic behavior (Table 2).During a 15-day nomadic p h a s e , t h e colony emigrated on 11 nights.Some degree of raiding t o o k p l a c e on every nomadic night.The median time of raid onset f o r t h e control colony was 1850 hr, and the median duration of r a i d i n g was 9.7 hr.

Psyche [Vol. 88
Much of the discussion generated by Schneirla's brood-stimulation theory concerns the relative degree to which raiding and emigrations are influenced by interactions between brood and adults (internal processes), and by external environmental factors.Theoretical support for emphasizing brood-related processes stems not only from Schneirla's own research with army ants (Schneirla, 1957(Schneirla, , 1958(Schneirla, , 1971)), but from studies of other social insects as well.For example, honeybee workers can collect protein-rich pollen o r carbohydrate-rich nectar.Louveaux (1950) found that the amount of pollen collected by an incipient colony is small, but increases as the brood population increases.In another experiment (Louveaux, 1958), he removed the colony queen from a mature colony and found that pollen collection was unaffected until many of the larvae had pupated.Further evidence of larval stimulation of adult foraging came from Fukuda, 1960(in Free, 1967), who showed that foraging workers from a recently-divided colony collected very little pollen until the eggs laid by the new queen hatched into larvae.Finally, Free (1967) demonstrated that adult worker foraging was influenced more by direct access to the brood than by brood odor alone.Perhaps most significant was the additional finding that artificially feeding a colony with pollen resulted in a decrease in pollen collection and a corresponding increase in nectar collection.
Although Schneirla was primarily concerned with the role of callow and larval excitation, he did recognize the role of food as an ecological parameter.Thus, at an early stage of his field research with the neotropical genus Eciton, he reported (Schneirla, 1938) that colonies frequently emigrate along the heaviest raiding route of that day.Nevertheless, it was Rettenmeyer (1963) who first suggested that the location and amount of captured food might influence not only the path of colony movements, but the very tendency to emigrate in the first place.The idea that colony excitation could be related to brood satiation has received empirical support from Free's (1967) study of honeybees and from related research with the myrmicine ant genus Myrmica (Brian, 1957(Brian, , 1962;;Brian and Abbott, 1977;Brian and Hibble, 1963).It was therefore significant that by the time of Schneirla's last field study, concerning emigration behavior in the paleotropical army ant genus Aenictus, he conceded that short-term variations in colony excitation may 1 59 depend upon the "alimentary condition prevalent in the b r o o d " (Schneirla and Reyes, 1969), and that emigrations are likely to begin soon after food has run low.
In a recent series of field and laboratory studies of n o m a d i c behavior in nearctic colonies of N. nigrescens (Topoff and M i r e n d a , 1980 a,b;Mirenda et al., in press), we demonstrated: (1) that t h e location of booty clearly influences the direction of raiding a n d therefore of emigrations; and (2) that artificially-fed colonies e x h i b i t a lower frequency of emigrations.The present study differs f r o m these in that food augmentation began late in the statary p h a s e , before most of the callow population had eclosed.In addition t o delaying the onset of the nomadic phase by reducing excitation f r o m newly-eclosed callows (Topoff et al., 1980a), this was our f i r s t attempt to eliminate emigrations through a complete nomadic p h a s e in the field.
During the six days of food augmentation for colony no. 1, w e provided a total of 144 g of booty.Since the colony generated f e w additional raiding columns, the artificially-administered b o o t y represents over 90% of the colony's total food intake for that p e r i o d .According to Mirenda et al. (1980), colonies of N. nigrescens g a t h e r approximately 0.4 mg of booty/larva/ nomadic night.Thus, o n t h e average, we provided colony no. 1 each night with an amount o f food that would be collected by a colony containing about 60,000 larvae.Although our estimate of colony size contains an error o f Az 20%, we can be reasonably certain of having provided this c o l o n y with about 1.2 times the amount of food it would normally g a t h e r .Although the large size of this colony dictated that we could n o longer supplement its food to the same degree throughout t h e remainder of the nomadic phase, the evidence from phase length, callow pigmentation, and larval size supports the conclusion t h a t the onset of the nomadic phase was delayed for 4-8 days.
For colony no. 2, which was considerably smaller and m o r e precisely counted, intensive overfeeding was more feasible.O n t h e average, 8.8 g of booty were provided on food-supplemented nights.This is more than 5 times the amount of food that a colony of t h i s size would collect in the field.In view of this feeding regime, it is n o t surprising that the colony conducted only one completed emigration throughout its 14-day nomadic phase.We must emphasize, h o wever, that a reduction of the frequency of nomadic emigrations is b y itself not sufficient to infer a relationship between food supply and colony t o b e g g n raiding earlier than usual.Thus, although w e arrived at the s 5 t e by 1800 hr, a long (60 m) emigration was already in progress.G i v e n the large size of the colony, we decided t o terminate f o o d -augmentation.The colony remained nomadic f o r the next nine d a y s , during which time it emigrated on six nights.In order t o d e t e r m i n e whether we had been successful in delaying the onset of t h e nomadic phase, three independent types of evidence were analyzed: (1) phase length: (2) callow pigmentation; and (3) larval size.Collectively, our data indicate that the nomadic phase was indeed d e l a y e d for 4-8 days.
o monitor the time of onset and the duration o f e a c h night's raid, in addition to the emigration frequency.This c o l o n y was released from a laboratory nest at 1900 hr on August 7. B e c a u s e this was statary day 15, most of the pupae were fully pigmented.T h e colony promptly moved into a subterranean nest beneath a s m a l l hole in the desert floor.The first raiding column appeared s h o r t l y after 2200 hr, at which time 9.0 g of Novomessor brood and t e r m i t e s were placed near the raiding front.The army ants r e m o v e d t h e booty in less than 1 hr, after which all surface activity ceased.F o r the next seven nights, the colony was either not active on the s u r f a c e or, a t best, conducted brief raids (each of which was i m m e d i a t e l y followed by artificial feeding) but no emigrations.On August 1 5 w e arrived a t the study site after 2200 hr, and found the c o l o n y emigrating 25 m to the NW.Because previously-collected f o o d was being transported to the new nest, but no larvae had yet a p p e a r e d , we considered the emigration to be in an early stage.A c c o r d i n g l y , 10.2 grams of booty were placed near the emigration column, 1 m from the old bivouac.This resulted in recruitment of ants b o t h f r o m the short column leading to the old nest, and from the l o n g e r emigration column.All of the artificially-placed food was t a k e n back t o the old nest, and the emigration was aborted.O n August 19 (nomadic day 1 I), after 2 days of not h a v i n g b e e n fed, the colony conducted its only successful emigration.T h e m o v e took the colony 19 m to the N, beneath an Ephedra b u s h .O n

Table 1 .
The overcast sky, coupled with cool temperatures late in the afternoon, enabled the Activity schedule for Neivamyrmex nigrescens colony no. 1

Table 2 .
Activity schedule for Neivamymex nigrescens colony no. 2 and control colony 'Emigration started but reversed by feeding.See text for details.**Statary shift forced by excavation of bivouac