Proc. Nati. Acad. Sci. USA Vol. 78, No. 3, pp. 1976-1979, March 1981 Population Biology

Enemy deterrence in the recruitment strategy of a : Soldier-organized foraging in costalis (communication/ predation/chemical ecology/defensive behavior/caste evolution) JAMES F. A. TRANIELLO* Department of Biology, Museum of Comparative Zoology Laboratories, Harvard University, Cambridge, Massachusetts 02138 Communicated by Edward 0. Wilson, December 15, 1980

ABSTRACT The nasute soldiers of the neotropical termite eight 45 x 60 cm glass foraging plates supported on oil traps. Nasutitermes costalis function as scouts by exploring new terrain Decayed birch was used as food, and large test tubes fitted with for food in advance of the worker caste and regulate foraging ac- tight cotton plugs provided moisture. The colony of N. costalis tivity by laying trails composed of sternal gland pheromone. Ad- ditional soldiers are at first recruited in large numbers, and sub- on which most observations and experiments were performed sequently workers appear as the pheromone concentration has thrived in the laboratory for 7 yr, continually expanding its increases. The role of the nasutes in the organization of foraging nest. is extremely unusual for the soldier caste in social and ap- In the laboratory, groups of soldiers of N. costalis periodically pears tobe a component of a foraging/defense system that controls emerge in "pseudopod" fashion from the nest or the termini of the recruitment of foragers and effectively deters attacks by , existing foraging trails and slowly extend and expand. When the most fierce and important predators of . offered a foraging arena with food, "pseudopods" of 100-200 soldiers slowly advance and explore the new area. After the food In both the higher and lower termites, the soldier caste is often is located, the investigative web of soldier trails coalesces into equipped with elaborate and bizarre armament for defensive one or two main routes and is immediately followed by the re- purposes. In some the mandibles are enlarged and cruitment of additional soldiers and subsequently workers. highly sclerotized, whereas in others they are vestigial and the Typically 700-1000 soldiers arrive at the food and patrol the head capsule has become an ampule-shaped structure for dis- surrounding area during the next 30-45 min. The recruitment pensing defensive secretions (1-3). The extreme degree of de- of workers parallels the soldier pattern, but is delayed by ap- fensive specialization is reflected not only in morphological proximately 30 min and reaches a maximum 1.5 hr later (Fig. modifications but also in the impoverished behavioral reper- 1). This pattern was invariably observed in more than 50 ob- toire of the termite soldier and suggests a limited, exclusively servation periods. protective role in the organization of termite societies. How- The recruitment of soldiers and the initial stages of worker ever, in addition to their defensive functions, soldiers of Na- recruitment appear to be regulated entirely by the soldier caste. sutitermes costalis are responsible for discovering new food The following observations and experiments support this sources and communicating their location to workers. In this hypothesis. species the soldier caste is responsible for two social processes (i) Soldiers consistently explore new areas for food without that appear to be closely associated: the organization of foraging any worker accompaniment. When clean glass plates are con- and colony defense. nected to occupied areas by a wire-gauze bridge, soldiers in- The Nasutitermes is one of the most widespread and vestigate the new area and recruit additional soldiers in large ecologically significant genera of the higher termites (4). Non- numbers. After 50-60 soldiers explore the surface, a few work- reproductive castes are divided into (i) small and large workers rise dra- which tend brood, construct the nest, and forage (5, 6) and (ii) ers appear. While the number of soldiers continues to soldiers, or nasutes, which respond to defensive contingencies. matically, less than 5% of the total number of termites are work- When a soldier contacts an enemy, the mandibular muscles ers. In the absence of food, workers soon return to the occupied contract and project from the snout a sticky, irritating secretion area while soldiers continue exploratory behavior. Therefore, produced in the frontal gland (3, 7). Colonies of N. costalis are scouting parties are composed almost entirely of soldiers that arboreal, and foraging trails guide termites from the carton nest mass recruit to new areas. to decaying wood on the ground. Soon after foraging is in prog- (ii) After soldiers locate wood, workers are recruitedfollowing ress, workers cover the food source in a sheath of fecal material a 20- to 30-min delay. However, the delay in worker recruit- and debris; the foraging trail is protected by an arcade (semi- ment is not an intrinsic temporal characteristic of the recruit- circular in cross section) of similar composition (8). Although ment system. If soldiers are prevented from returning to the this prominent aspect of the biology of Nasutitermes has long nest after contacting wood, workers do not appear, although been reported in the literature (9-12), details of the mecha- other soldiers are continually recruited. Worker recruitment nisms involved in the organization of foraging have been can be delayed indefinitely in this way and can be induced at undescribed. any time by allowing soldiers that have contacted the food METHODS AND RESULTS source to return to the nest. Colonies of N. costalis and Nasutitermes corniger were reared (iii) Both stages of foraging behavior-soldier and worker in the laboratory in large plastic containers and given access to recruitment-can be experimentally induced with extracts of the sternal gland, the source of trail pheromone in termites (13). The publication costs ofthis article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertise- * Present address: Boston University, Department of Biology, 2 Cum- ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. mington Street, Boston, MA 02215. 1976 Downloaded by guest on September 27, 2021 Population Biology: Traniello Proc. Natl. Acad. Sci. USA 78 (1981) 1977

1000. protected and subject to predation, as during the early stages of foraging, when thousands of termites file in the open be- tween the nest and food source. It is possible that the early re- cruitment of a large soldier force provides an initial line of de- fense that deters ant predation by eliminating scouts that potentially could organize swift, massive raids. Predatory strat- egies of ants can be divided into three principal types: (i) fa- cultative predation by solitary foraging ants such as the ponerine species Pachycondyla and Odontomachus, (ii) organized attacks by species such as Solenopsis that are able to assemble a large worker force in a brief time period through mass recruitment, and (iii) predation by obligate termitophagous species [Pachy- condyla (= Termitopone) laevigata]. The effectiveness of nasute defensive tactics in coping with these types of predation was tested in the following manner. 100 In the first series of experiments, groups of 25 soldiers were placed in the foraging arenas ofvarious species ofponerine ants, and the results of 1:1 encounters between the ants and soldiers E were recorded. The results are presented in Table 1. Among the solitary foraging species, even the most efficient species Rhytidoponera metallica and Odontomachus bauri were successful in disabling and retrieving soldiers in only 18% 00 and 26% of encounters, respectively. The ability of 0. bauri workers to prey upon the nasutes might be attributed to their "strike and recoil" predatory behavior (21), which permits them to avoid the discharge of a soldier following contact. The spe- cialized termite predator P. laevigata never organized a raid after encountering nasutes, and most interactions resulted in avoidance or grooming. A particularly pugnacious ant, Solenopsis invicta, was used to simulate predation by a mass-recruiting species. In this series 1 2 3 4 5 6 7 of experiments, termite colonies were given access to a new Time, hr foraging plate (24 x 60 cm divided into 2 x 2 cm quadrats) on which was located a piece of birch. As soon as the nasutes ex- FIG. 1. Recruitment responses of soldiers and workers during for- plored the new area and began recruitment, a colony of S. in- aging organization. At time = 0, a fresh foraging plate with wood was connected to an occupied plate. Curves were fitted to the data by eye. victa was connected to the same plate with a 125-cm-long dowel O. Workers; *, soldiers. bridge. The plate was photographed at various time intervals to record the defensive behavior of the termites. The results are presented in Table 2. Solutions made from a single soldier sternal gland have a re- cruitment effect only on other soldiers. However, worker re- Table 1. Results of encounters between workers of various cruitment can be initiated by increasing the concentration of trails prepared from the sternal glands of soldiers. The results species of ponerine ants and termite soldiers of this procedure suggest that workers are recruited because Results of encounters, no.* of the cumulative effect of soldier-generated trails. The ex- Total perimental manipulation of pheromone concentration permits encounters, C the artificial induction of the typical pattern of foraging no. A B I II D organization. Species The soldiers, which are dependent on the workers for nu- Ectatomma trition, make the "decision" to begin foraging. As worker traffic tuberculatum 80 38 7 20 15 0 increases, the trail takes on its characteristic form and soldiers E. ruidum 114 46 26 16 26 0 flank the sides of the trail and the of the others Odontomachus bauri 81 23 25 6 6 21 periphery wood; O. chelifer 31 14 5 8 4 0 disperse throughout the area. Surveilling soldiers slowly retreat Pachycondyla to either the wood or the trail, and after 24 hr. arcade construc- obscuricornis 72 48 12 7 5 0 tion is generally complete. P. apicalis 64 45 15 3 1 0 Soldier precedence in foraging might be regarded as a strat- P. laevigata 147 99 48 0 0 0 egy to decrease predation by ants, many of which are termite Paltothyreus predators (14-16). Although considerable progress has been tarsatust 71 11 42 11 2 5 made in the study of the stimuli releasing defensive behavior Rhytidoponera (17-20), little is understood of the socioecological role of the metallicat 200 52 74 13 25 36 termite soldier. In field observations of termites were attacked Data was pooled from three to four 15-min observation periods. Nasutitermes, * Results of encounters: A, contact avoided; B, ant is sprayed, retreats, by workers of Crematogaster, Camponotus, Wasmannia, Az- and grooms; C, ant attacks, is sprayed, releases soldier [which is teca, Pachycondyla, and Odontomachus soon after breaking mobile (I), or disabled (II)], and grooms; D, ant attack is successful open 5-cm-long sections of their arcades. This manipulation and soldier is retrieved. simulates a condition in which the termites are not structurally t Species that do not occur in the range of N. costalis. Downloaded by guest on September 27, 2021 1978 Population Biology: Traniello Proc. NatL Acad. Sci. USA 78 (1981)

Table 2. Changes in the distribution pattern of soldiers after a colony of S. invicta was connected to a nasute-patrolled arena divided into 264 2 x 2 cm quadrats Area Soldiers/ % distribution of soldiers Time, Soldiers, occupied, quadrat, by no./quadrat min no. cm2 mean no. c3 4-9 10-17 ::-18 +15 207 312 3.03 75.6 23.1 1.3 0 +30 624 468 6.97 45.3 37.6 14.5 2.6 +60 809 476 8.54 36.1 32.8 26.1 5.0 +105 777 380 8.04 29.5 31.8 29.5 9.5 +120 435 260 7.31 38.5 33.8 18.5 9.2 +150 225 88 10.0 31.8 27.3 13.6 27.2 Soldier distribution is expressed as the number of quadrats with a given number of soldiers relative to the total number of quadrats occupied. Time is relative to the introduction of the first Solenopsis scout.

Two stages of defense can be recognized. First, the initial The time to incapacitation (net motion, <1 cm/min) is rapid contact between the scout ants and the termites induces the (mean = 92.7 + 52.2 sec; mean number of soldiers discharging mass recruitment of soldiers from the trail to the areas of con- = 17.9 6.5; N = 15). As the soldier flanks become progres- frontation: the size of the soldier force quadruples within 60 sivelv restricted to the periphery of the wood, soldier density min. Second, as the ants advance into the nasute-patrolled area increases and infiltration of the nasute line becomes impossible. and the number of contacts increases, the spatial distribution In more than 10 experimental series, neither ponerine hunt- of soldiers becomes strongly clumped. After contacting an ant, resses nor Solenopsis scouts could penetrate nasute flanks in a soldier discharges frontal gland material and recoils 3-4 cm spite of the close proximity of colonies and large numbers of while dragging the tip of the nasus on the substrate. Within workers. During these encounters, ant casualities are extremely seconds soldiers 3-4 cm away move toward the area of dis- high. In one typical case, 69% of 416 Solenopsis scouts entering charge; some adopt guard positions while others depart in trail- the foraging area of N. costalis were dead or disabled within 3 laying posture. Soon the soldier density in the area increases hr; onlv 11 nasutes were killed and retrieved by the ants. In dramatically. In the absence of confrontation with ants, such addition, tropical species of Solenopsis, Pheidole spp., Was- aggregations of soldiers are not observed. These changes in mannia, and Paratrechina were tested; the results were similar. soldier distribution appear to be regulated by the secretions of Small mvrmicine and formicine workers that are sprayed are the frontal and sternal glands. When extracts of these sub- rapidly debilitated and suffer high mortality in comparison to stances or crushed whole glands were offered on filter-paper larger ponerine ants, which are strongly affected by the noxious squares to patrolling soldiers, the greatest approach rate was properties of the spray but seldom die. mediated by the sternal gland secretion, whereas both frontal The scouting behavior of soldiers and their involvement in and sternal gland pheromones had a pronounced effect on ar- the regulation of the early stages of foraging are components resting motion (Table 3). of the defensive strategy of N. costalis. In comparative field and laboratory studies with N. corniger and N. ephratae, one dif- DISCUSSION ference in foraging behavior was noted: in most cases members The constant tightening of the soldier defensive perimeter sug- of the large worker caste accompanied the soldiers. Although gests a continual "retreat and regroup" strategy. Apparently the the results of Stuart (22) suggest that the mechanism of foraging secretion of the sternal gland serves both a stimulating recruit- organization in N. corniger is similar to that in N. costalis, fur- ment function and also acts as an arrestant, depending upon ther studies are required. In another nasutitermitine, Hospi- context. Scout ants are efficiently eliminated by the concerted talitermes, soldiers form the front and rear of foraging columns efforts of soldiers. Ants sprayed once or twice immediately re- (23). It appears that termite species that must leave the cover treat and groom, while heavily sprayed ants are immobilized. of the nest or the soil to forage have similar foraging systems.

Table 3. Response of soldiers to secretions of the frontal and sternal glands Soldiers remaining Duration of patrolling,t Approach rate* or patrollingt sec N =6 N = 6 N = 40 Control 15.8 ± 10.4 1 (0-4) 18 (2-135) Frontal gland secretion 15.2 ± 11.2 7 (3-9) 45 (5-306) Sternal gland secretion 32.0 ± 6.9 8 (6-10) 58 (6-244) Single whole glands were crushed on filter paper squares (4 cm2) and offered to patrolling soldiers. Untreated filter paper squares served as controls. N, Number of replicates. The difference from controls in approach rate to sternal gland secretion is significant (0.02 > P > 0.01; Student's t test), and a sig- nificantly higher number of soldiers remained on squares treated with frontal gland secretion (P <0.005; Mann-Whitney U test) or sternal gland secretion (P <0.001). Also, soldiers patrolled in the area of treated squares for significantly longer durations (P <0.001). No synergistic effect was observed when sternal and frontal gland secretions were offered together. * Mean numbers of soldiers per min ± SD. t Average number of soldiers remaining on the square or patrolling after 5 min with range in parenthesis. t Median duration per soldier of patrolling behavior with range in parenthesis. Downloaded by guest on September 27, 2021 Population Biology: Traniello Proc. NatL Acad. Sci. USA 78 (1981) 1979

Caste-specific responses during foraging appear to be based 2.. Noirot, C. (1969) in Biology of Termites, eds. Krishna, K. & Wees- on quantitative differences in the sizes of worker and soldier ner, F. M. (Academic, New York), Vol. 1, pp. 311-350. sternal glands or qualitative differences in their constituents. 3. Prestwich, G. D. (1979) J. Chem. EcoL 5, 459-480. 4. Araujo, R. L. (1969) in Biology of Termites, eds. Krishna, K. & Trails prepared from extracts of soldier sternal glands were Weesner, F. M. (Academic, New York), Vol. 2, pp. 527-576. found to be only one fourth as attractive as similar trails pre- 5. Pasteels, J. M. (1965) Biol Gabonica 1, 191-205. pared from the sternal gland of the large worker caste (LW3), 6. McMahan, E. (1970) Ins. Soc. 17, 113-120. thus supporting the latter hypothesis. A similar concentration 7. Vrko\, J., Ubik, K., Dolej§, L. & Hrd9, I. (1973) Acta Entomol effect has been described in Trinervitermes trinervoides (24), Bohemoslov. 70, 74-80. although this species or other species in this genus do not appear 8. Grass6, P. P. & Noirot, C. (1951) Annk PsychoL 50, 273-280. 9. Beaumont, J. & Dudley, P. H. (1889) Trans. N.Y. Acad. Sci. 8, to have soldier-organized foraging (25). 35-114. Ant predation is likely to be the major selective pressure that 10. Andrews, E. A. (1911)J. Anim. Behav. 1, 193-228. has led to not only the evolution of a well-armed soldier caste 11. Stuart, A. M. (1960) Dissertation, (Harvard Univ., Cambridge, but also to the development of the integrated systems of for- MA). aging and defense exhibited by N. costalis. The importance of 12. Stuart, A. M. (1969) in Biology of Termites, eds. Krishna, K. & predation is also manifest in the proportion of soldiers (20-30%) Weesner, F. M. (Academic, New York), Vol. 1, pp. 193-232. 13. Stuart, A. M. (1963) Physiol Zool 36, 69-84. in Nasutitermes (17, 26). This reflects a significant investment 14. Wheeler, W. M. (1936) Proc. Am. Acad. Arts Sci. 71, 159-243. in a caste that must be nutritionally supported but has no en- 15. Sheppe, W. (1970) Ins. Soc. 17, 205-218. ergetic input into colony growth. Prestwich et al. (27) have re- 16. Wilson, E. 0. (1971) The Societies (Belknap Press of Har- cently demonstrated that Nasutitermes soldiers have unusually vard Univ. Press, Cambridge, MA). high rates and may be self-sufficient as a caste. 17. Ernst, E. (1959) Rev. Suisse Zool 66, 289-295. This nutritional independence may permit the maintenance of 18. Stuart, A. M. (1967) Science 156, 1123-1125. 19. Eisner, T., Kriston, I. & Aneshansley, D. J. (1976) Behav. Ecol. the large defensive forces common in Nasutitermes. Termite SociobioL 1, 83-126. species have a wide variety of food habits (28), foraging systems, 20. Nutting, W. L., Blum, M. S. & Fales, H. M. (1975) Psyche 81, and highly variable caste ratios (26), and comparative studies 167-177. that focus on these parameters as a function of the predatory 21. Brown, W. L. (1976) Stud. Entomol 19, 67-171. environment should provide an understanding of the ecological 22. Stuart, A. M. (1975) in Olfaction and Taste (Academic, New York), determinants of social organization. Vol. 5, pp. 343-348. 23. jander, R. & Daumer, K. (1974) Ins. Soc. 21, 45-69. I thank B. Holdobler, E. 0. Wilson, R. E. Silberglied, G. Alpert, 24. Tschinkel, W. R. & Close, P. G. (1973)J. Ins. Physiol. 19, 707-721. G. Prestwich, and S. Levings for reading the manuscript. A. Aiello,. R. 25. Oloo, G. W. & Leuthold, R. H. (1979) EntomoL Exp. Appl, 26, E. Silberglied, and B. Thorne provided valuable assistance at the 267-278. Smithsonian Tropical Research Institute in , and D. A. Traniello 26. Haverty, M. (1977) Sociobiology 2, 199-216. and K. Horton typed various forms of the manuscript. This work was 27. Prestwich, G. D., Bentley, B. L. & Carpenter E. J. (1980) Oec- supported by National Science Foundation Grant BNS 77-03884 (B. ologia 46, 397-401. Holidobler, sponsor) and by the Richmond Fund of Harvard University. 28. Wood, T. G. (1978) in Production Ecology ofAnts and Termites, ed. Brian, M. V. (Cambridge Univ. Press, Cambridge, England), 1. Weesner, F. M. (1969) in Biology of Termites, eds. Krishna, K. pp. 55-80. & Weesner, F. M. (Academic, New York), Vol. 1, pp. 19-47. Downloaded by guest on September 27, 2021