Proc. Nati. Acad. Sci. USA Vol. 88, pp. 2031-2034, March 1991 Population Biology Nest-mate recognition based on heritable odors in the Microcerotermes arboreus (aggression/Isoptera/kin recognition/social insects) ELDRIDGE S. ADAMS* Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Panama Communicated by Mary Jane West-Eberhard, December 7, 1990 (receivedfor review November 10, 1989)

ABSTRACT Workers ofthe Neotropical termite Microcer- heritable cues. I also tested whether can distinguish otermes arboreus distinguish nest mates from other conspecifics degrees of relatedness among unfamiliar kin and whether by odor. A controlled breeding experiment demonstrated a inheritance through the mother has a stronger effect on genetic component to variation in colony odors. Workers were colony odor than inheritance through the father. less aggressive toward unfamiliar relatives than toward non- relatives and distinguished degree of relatedness among unfa- miliar workers. Unfamiliar relatives were attacked more often MATERIALS AND METHODS than nest mates, despite similar levels of genetic relatedness These studies were conducted from January 1988 through thus, nest-mate recognition is not based solely upon heritable January 1989 on Barro Colorado Island at the Smithsonian characteristics of individual workers. No difference was de- Tropical Research Institute in the Republic of Panama. At tected between the effects ofcues inherited through the mother this site, M. arboreus is an abundant termite that builds nests and cues inherited through the father. in trees or on the ground surface. This species has also been identified as Microcerotermes exiguus, but here I follow the Many eusocial insects distinguish nest mates from non-nest- catalog of Araujo (13). All termites were collected from mate conspecifics (1, 2). The recent sociobiological literature arboreal nests <2 m from the ground. has focused on the mechanisms and ecology of these recog- In tests for nest-mate recognition, I recorded the reactions nition abilities, in part because of their implications for the of workers within a group of "resident" termites to "intrud- evolution of the highly altruistic social systems that insect ers" presented in a standardized manner. This terminology colonies represent. Evolution of social behavior by kin reflects the natural context of nest-mate discrimination, selection requires that animals differentially direct altruistic which is territory defense (8-11). One to 12 hr before each behavior toward genetic relatives (3). If insects discriminate test, 30 workers from the resident colony were placed in a among conspecifics on the basis of heritable cues, this plastic dish measuring 5 cm in diameter, which was lined with discriminative ability may provide the basis for preferentially moist filter paper. Workers were always taken from nest helping close kin. carton collected in the field within 2 wk before laboratory Previous studies on colony recognition have been con- assays. Responses to intruders were then measured during ducted almost entirely with the social -, short presentations to individual workers chosen haphaz- bees, and (2). The Isoptera, or termites, have inde- ardly within the group of 30. pendently evolved similar eusocial colony structures and are To present an intruder, the termite was grasped by the neck an important group for comparative studies of social evolu- with a pair offorceps and killed by severing the head from the tion (4). Although the Hymenoptera have a haplodiploid body. This technique prevented movement and biting by the mechanism of sex determination, the Isoptera are fully dip- intruder, which could affect responses by the resident ter- loid (5). Theoretical analyses of heritable nest-mate recogni- mites. The head was then presented directly in front of a tion cues show that accurate nest-mate recognition is more resident worker so that the palps of the resident came in easily achieved in haplodiploid than in diploid organisms (6); contact with the face of the intruder. This relative position however, many diploid organisms recognize familiar or un- was maintained for either 5 or 20 sec, depending upon the familiar relatives in other contexts (7). Full sisters of haplo- experiment. All reactions were recorded. Ten levels of ag- diploid species share a greater proportion of alleles identical gressiveness were recognized and assigned numerical scores by descent than do full siblings ofdiploid species (5); thus, the on a 10-point scale (Table 1). When multiple presentations genetic contrast between full sisters and either half-sisters or were made to the same group of residents, the presentations nonrelatives is greater (6). As a result, the potential for were rotated among the 30 available workers. After each distinguishing relatives from nonrelatives or sisters from trial, the forceps and plastic dishes were washed with water half-sisters on the basis ofheritable characters is much lower and ethanol, and the paper linings were replaced. for termites than for social Hymenoptera. Although many To test whether workers distinguish nest mates from other species of termites are known to aggressively reject non- termites, workers were collected from fragments of nests nest-mate conspecifics (8-11), whether their colony recog- distributed along 5 km of trails. The reactions of workers nition cues have heritable components is not known. from 10 randomly selected colonies to three kinds of termite The genetic basis of recognition cues can be examined by intruders were measured by using the procedure described experimentally breeding insects to control the pedigree co- above: (i) nestmates, (ii) non-nest-mate conspecifics from a efficient of relatedness, G (12), between interacting pairs. I randomly selected nest, and (iii) workers of a sympatric adopted this approach to determine whether nest-mate rec- arboreal termite, Nasutitermes corniger. Reactions to clean ognition in the termite Microcerotermes arboreus is based on forceps were also measured to control for any effects of the movement or odor of the forceps themselves. Thirty presen- The publication costs ofthis article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" *Present address: Department of Biology, University of Rochester, in accordance with 18 U.S.C. §1734 solely to indicate this fact. Rochester, NY 14627. 2031 Downloaded by guest on September 30, 2021 2032 Population Biology: Adams Proc. Natl. Acad. Sci. USA 88 (1991)

Table 1. Behavioral reactions by resident termites to conspecific colony was presented sequentially to five resident workers intruders during 20-sec laboratory assays of aggression for 20 sec. The numeric scores shown in Table 1 were used Score Behavior to calculate a mean score of aggression for the five encoun- 1 Neutral response: contact with the intruder without ters. All tests were blind; that is, the ancestry of the intro- examination or attack duced termite was unknown to the human observer at the 2 Brief examination: orientation toward the intruder for time of trial. The probability of attack, calculated as the <10 sec proportion of scores of 5 or greater (Table 1), and the mean 3 Prolonged examination: orientation toward the scores of aggression were both analyzed. Aggression scores intruder for at least 10 sec and/or palpation of the were examined with a two-level analysis of variance intruder's head and mouth parts (ANOVA) with workers nested in colonies and colonies 4 Avoidance: rapid reorientation and increase in walking nested within levels of relatedness. To equalize variances, I rate used the square roots ofaggression scores and the arcsines of 5 Defecation: deposition of a droplet of anal fluid on or the square roots of attack proportions. near the intruder In February 1989, the field colony that had served as a 6 Nip: biting of the intruder for <2 sec, terminated source of resident workers and of alate reproductives was without visible damage dissected to examine its reproductives. 7 Attack within 20 sec: strong bites that pierce the intruder's exoskeleton RESULTS 8 Attack within 15 sec 9 Attack within 10 sec Although workers of M. arboreus showed little or no aggres- 10 Attack within 5 sec sion toward nest mates or clean forceps, they often attacked and killed alien conspecifics and workers of the sympatric termite species, N. corniger (Fig. 1; Kruskal-Wallis test, H tations of 5-sec duration were made with each of the four = 34.5, df = 3, P < 0.001). M. arboreus workers were more stimuli to each termite group. likely to attack N. corniger workers than non-nest-mate To vary experimentally the genetic relatedness between conspecifics (73.6% of presentations versus 29.0%6). Agonis- resident and intruder termites, colonies were bred from kings tic behaviors ranged from brief examinations to immediate and queens of known origin. Alate queens and kings matured attacks (Table 1). within many large nests by March or April and mated after The probability of attack varied significantly with genetic brief nuptial flights at the start of the wet season (May and relatedness between resident and intruder termites, even June 1988). A single field colony, hereafter called the resident though they had never previously encountered one another colony, was selected as a source of resident workers to be (Fig. 2, F2.24 = 10.47, P < 0.001). Among relatives, workers used in subsequent tests of nest-mate discrimination. Kings distinguished degrees of relatedness: the proportion of close and queens were collected from this colony and from two relatives attacked was significantly smaller than the propor- other colonies separated by at least 150 m and showing tion ofdistant relatives attacked [a posteriori comparison by mutual aggression in preliminary tests. The reproductives the GT2 method (14); P < 0.05]. Furthermore, the mean were netted as they emerged during nuptial flights or were scores of aggression toward introduced workers varied sig- removed directly by cutting the nest open with a machete nificantly with relatedness (two-level nested analysis of vari- (May 6 to June 12, 1988). The alates were allowed to fly ance, F2,23 = 7.31, P < 0.005). Mean scores of aggression briefly in a large plastic bin, measuring 90 cm tall and 45 cm (Table 1) for the three levels of relatedness were (weighted X in diameter, after which they dropped their wings and began ± SE): N = 3.02 + 0.26 for close relatives; x = 4.97 + 0.49 to form tandem-running pairs. These pairs were separated for distant relatives; and X = 5.91 ± 0.78 for nonrelatives. before mating, and the termites were randomly assigned to Variation in relatedness accounted for 27.5% of variation in one of three breeding treatments: (i) the king and the queen scores of aggression. came from the resident colony, (ii) the king or the queen came Considering those trials in which the resident and intruder from the resident colony, whereas the other reproductive termites were related through only one parent, there was no came from a separate colony, or (iii) neither reproductive significant difference in aggressiveness toward termites re- came from the resident colony. lated through the queen (five colonies) and those related The reproductives were placed in plastic containers 10 cm through the king (six colonies; F1,242 = 1.02, not significant). in diameter and 7 cm in height with blocks of decaying wood measuring -8 x 3 x 4 cm, all cut from the same log on the forest floor. The containers were stored in a small open-air shed in a shaded region on the forest floor and watered as necessary to keep the wood moist. Colonies were maintained until January 1989. Colonies with at least five fully pigmented workers were then used in recognition tests. This procedure ~0 yielded 6 colonies (59 workers) related to the resident colony through both parents, 11 colonies (93 workers) related s0 0.5 through one parent, and 10 colonies (117 workers) unrelated 0. F 0 to the resident colony. Hereafter, these three kinds of work- CL ers will be called "close relatives," "distant relatives," and 0. ['1 "nonrelatives," respectively, of termites from the resident colony. To test for the ability to recognize unfamiliar kin, I assayed the responses of termites from the resident colony, freshly collected from the field, to intruders from laboratory-reared colonies. These termites had never previously encountered one another. A separate group of 30 workers from the Forceps Nestmates Conspecifics Nasutitermes resident colony was used to assay responses to each labora- FIG. 1. Proportions of termites or controls attacked in 5-sec tory-reared colony. Each worker from the laboratory-reared encounters (x ± SE). Downloaded by guest on September 30, 2021 Population Biology: Adams Proc. Natl. Acad. Sci. USA 88 (1991) 2033 components of colony odor (17). However, the resident termites used to test for nest-mate recognition were taken from the field, where they were exposed to natural environ- mental inputs and variation. Because these workers discrim- inated among conspecifics on the basis ofinherited attributes, la it is clear that in natural conditions, workers develop tem- - a plates of colony odor that encode information about genetic a relatedness. Environmental differences may create further cs 0.5 variation in these templates and in recognition cues. 0 The degree of relatedness between resident termites and go the "close" and "distant" relatives presented in this study 0 A-., cannot be determined precisely. Colonies ofM. arboreus are founded by a single queen and king (unpublished data), but in the resident colony used in this breeding experiment, the original pair were replaced by nymphoid reproductives. The high number of replacements and the low degree of physo- F ] gastry of the queens suggests that this replacement occurred during the course of the experiment. Most probably, the Close Distant Non relatives relatives relatives secondary queens and kings were nymphs that remained within the nest in 1988 instead of completing development of FIG. 2. Probability of attack (x ± SE) toward termites during the flight apparatus and dispersing during the nuptial flights. 20-sec encounters. Close relatives were related to the attacker If so, the replacement reproductives were not parents of the through both parents, distant relatives were related through one alates used to breed laboratory colonies (collected in May parent, and nonrelatives were unrelated through either parent (see 1988), but they may have produced some or all of resident text). workers used to test for nest-mate recognition (collected in When the resident nest was dissected in February 1989, no January 1989). primary reproductives were found. Instead, 117 female and Analysis of these possible patterns of ancestry yields 29 male replacement reproductives were found within the coefficients of relatedness of0.5 between residents and close central areas ofthe nest. Both sexes were fully pigmented and relatives, of0.25 between residents and distant relatives, and had short wing buds, resembling those offirst instar nymphs. of 0 between residents and nonrelatives. These estimates The queens were slightly physogastric, suggesting a recent hold whether the resident workers were offspring of primary origin. or secondary reproductives. The calculations assume that replacement reproductives were offspring ofthe primary pair and that there had been no previous . Other DISCUSSION pedigrees are possible, so these values must be regarded as Workers of M. arboreus recognized unfamiliar relatives and approximations. In any case, the average degree of related- distinguished degrees of similarity among unfamiliar kin. ness to close relatives was approximately twice the degree of These results demonstrate that there is a strong inherited relatedness to distant relatives. component to the cues used in nest-mate recognition. Inher- Genetic effects upon colony recognition labels have also itance through the mother and the father were equally im- been demonstrated in the social Hymenoptera. Tests for portant. Because workers are blind and tactile stimuli were inherited components of individual or colony odors have controlled in laboratory assays, recognition appears to be been made by controlled breeding designs, by cross-fostering based primarily on variation in odor. of eggs or immature forms, or by attempting to remove As in many other breeding experiments, the possibility of sources of environmental variation (1, 2, 18). The strongest nongenetic inheritance cannot be ruled out. Odors may be evidence for genetic effects on colony odor comes from carried from the source nest by alate reproductives and studies ofthe sweat , Lasioglossum zephyrum (1, 19), the transferred directly to their offspring. However, this alter- honeybee, Apis mellifera (2), and the acacia , Pseudo- native requires that compounds carried by the alate repro- myrmex ferruginea (20). Breeding experiments with these ductives persist for 7 mo in concentrations sufficient to species have shown that acceptance of unfamiliar workers supersede other sources of variation, despite dilution among correlates with genetic relatedness across a range of values. the offspring. Symbiotic bacteria used in digestion of cellu- In the social wasps, there have been no comparable breeding lose are also transferred from parent to offspring (4), but there experiments, but heritability of colony labels may be indi- is no evidence that these affect surface pheromones. cated by preferences for relatives separated before adulthood Nest-mate recognition in M. arboreus is not based solely (18). upon heritable characteristics of individual workers. The Genetic mechanisms of kin discrimination are more easily termites attacked unfamiliar close relatives more often than realized in haplodiploid species, such as the social Hy- nest mates, despite nearly equivalent levels of genetic relat- menoptera, than in diploid species, such as the termites. In edness. Two non-exclusive possibilities could produce this models of nest-mate recognition in which each allele pro- effect. (i) There may be environmental contributions to the duces an identifiable odor, error rates for a given genetic recognition labels, as there are in many social Hymenoptera system, defined by the number of loci and by the allele (1, 2). (ii) Heritable colony recognition labels may derive frequencies at each locus, are several times greater in diploid primarily from the queen and king (15) or be transferred from species (6). The sensitivity of M. arboreus workers to heri- worker to worker to produce a- shared mix of odors (16). table odors thus suggests that many loci or alleles affect the Under this hypothesis, unfamiliar close relatives can be colony recognition cues. distinguished from nest mates due to their exposure to hypotheses of social evolution require that different, but related, reproductives or workers. animals be able to direct altruistic behavior preferentially In this study, environmental variation in odor was reduced toward relatives (3). These results show that worker assess- by raising all colonies in similar laboratory conditions. This ment of heritable odors could be a mechanism for this ability design may exaggerate the apparent importance of heritable in termites. By aggressively excluding nonrelatives, workers Downloaded by guest on September 30, 2021 2034 Population Biology: Adams Proc. NatI. Acad. Sci. USA 88 (1991)

maintain high levels of genetic relatedness within the colony 7. Fletcher, D. J. C. & Michener, C. D., eds. (1987) Kin Recog- and ensure that their labor primarily benefits close kin. nition in Animals (Wiley, Chichester, U.K.). 8. Clement, J.-L. (1978) C. R. Acad. Sci. Ser. D 286, 351-354. I thank the Smithsonian Tropical Research Institute and the 9. Thorne, B. L. (1982) Psyche 89, 133-150. Republic of Panama for the opportunity to work on Barro Colorado 10. Levings, S. C. & Adams, E. S. (1984) J. Anim. Ecol. 53, Island. M. J. West-Eberhard and N. Franks offered useful advice. C. 705-714. Haverkate helped care for the young termite colonies. R. Condit, G. 11. Adams, E. S. & Levings, S. C. (1987) J. Anim. Ecol. 56, Roderick, B. Mitchell, and U. Smith made helpful comments on an 1069-1081. earlier draft of the manuscript. This study was funded by a Post- 12. Pamilo, P. & Crozier, R. H. (1982) Theor. Popul. Biol. 21, doctoral Fellowship from the Smithsonian Institution. 171-193. 13. Araujo, R. L. (1977) Catalogo dos Isoptera do Novo Mundo 1. Michener, C. D. & Smith, B. H. (1987) in Kin Recognition in (Academia Brasileira de Ciencias, Rio de Janeiro). Animals, eds. Fletcher, D. J. C. & Michener, C. D. (Wiley, 14. Sokal, R. R. & Rohlf, F. J. (1981) Biometry (Freeman, San Chichester, U.K.), pp. 209-242. Francisco), 2nd Ed. 2. Breed, M. D. & Bennet, B. (1987) in Kin Recognition in 15. Carlin, N. F. & Holldobler, B. (1986) Behav. Ecol. Sociobiol. Animals, eds. Fletcher, D. J. C. & Michener, C. D. (Wiley, 19, 123-134. Chichester, U.K.), pp. 243-285. 16. Stuart; R. L. (1988) Proc. Natl. Acad. Sci. USA 85, 4572-4575. 3. Hamilton, W. D. (1964) J. Theor. Biol. 7, 1-52. 17. Carlin, N. F. (1989) Neth. J. Zool. 39, 86-100. 4. Wilson, E. 0. (1971) The Insect Societies (Harvard Univ. 18. Gamboa, G. J., Reeve, H. K. & Pfennig, D. W. (1985) Annu. Press, Cambridge, MA). Rev. Entomol. 31, 431-454. 5. Crozier, R. H. (1979) in Social Insects, ed. Hermann, H. R. 19. Greenberg, L. (1979) Science 206, 1095-1097. (Academic, New York), Vol. 1, pp. 223-286. 20. Mintzer, A. & Vinson, S. B. (1985) Behav. Ecol. Sociobiol. 17, 6. Getz, W. M. (1981) J. Theor. Biol. 92, 209-226. 75-78. Downloaded by guest on September 30, 2021