Slightly modified from Holldobler and Wilson 1990. The references cited can be found in The Ants.
Alarm is the most difficult of behavioral responses merely to define, because investigators have used it as a portmanteau category into which all responses to danger are placed. In the broad sense worker ants are said to be in a state of alarm when they move away from a potentially dangerous stimulus, either calmly or in panic, or charge toward it aggressively, or simply mill about in a state of heightened alertness. Nevertheless, for each particular species in turn it is usually easy to devise a bioassay based on the precise reactions of the colonies, without having to take into account all the variations of behavior displayed by other species.
Most alarm signals are multicomponent. They typically consist of two or more pheromones, which often serve simultaneously to alert, attract, and evoke aggression. Acoustical signals are sometimes added to the pheromones, including especially stridulatory chirps that enhance attraction. The ants also occasionally touch nestmates with their antennae and forelegs during ritualized motor displays.
Alarm behavior is made additionally difficult to characterize because it so often blends into other major behavioral responses. Most ant species engage in some form of alarm-defense, in which the same chemicals are used to repel enemies and to alert nestmates. Examples of alarm-defense substances that have been identified in various ant species are citronellal, dendrolasin, dimethyl sulfate, and undecane. Some secretions are probably purely defensive, such as the formic acid employed by at least some formicine species. It is also possible that other compounds are purely communicative, especially those produced in minute quantities, but this specialized function remains to be documented in many cases of which we are aware. Probably the great majority of compounds used in alarm communication also serve in defense. Put in the sequence more likely to have occurred in evolution, many chemicals employed in defense have also been ritualized to some extent into alarm signals.
A second intergradient category that has been documented is alarm-recruitment. Alarm signals both alarm and attract in some species, while in others the alarm pheromones are combined with odor trails that lead nestmates toward or away from the source of the danger. Many species employ a single alarm-recruitment procedure to alert nestmates to both enemies and prey, and in fact the distinction between the two may be wholly blurred with reference to communication.
Whether joined with recruitment or not, alarm behavior can be conveniently classified into one or the other of two broad categories (Wilson and Regnier, 1971). In aggressive alarm some of the colony members (often the majors or "soldiers") are drawn toward the threatening stimulus and seek to attack it. In panic alarm the colony as a whole flees from the stimulus or dashes around in erratic patterns. If disturbed strongly enough by the waves of alarm, they may even evacuate the nest.
A wide range of stimuli evokes alarm communication. Oddly, substratal vibrations and air currents disturb the entire colony and may even induce evacuation if severe enough, but they seldom trigger the release of alarm pheromones. Alarm communication is initiated far more predictably when an enemy penetrates the close environs of the nest. Certain stimuli work better than others in this context. In the phenomenon called "enemy specification," dangerous species are more effective than other, less threatening ones in evoking a response. For example, minor workers of Pheidole dentata recruit majors to only a single fire ant forager approaching the nest (Wilson, 1976b). Those of Pheidole desertorum and Pheidole hyatti initiate panic alarm leading to nest evacuation when they become aware of the approach of army ants of the genus Neivamyrmex (Droual and Topoff, 1981). Such hair-trigger responses are not limited to living enemies. A single drop of water in the nest entrance of Pheidole cephalica is often enough to cause alarm-recruitment and the full retreat of the colony away from that part of the nest (Wilson, 1986c).
Alarm is easy to define but technically one of the most difficult forms of communication to study with precision. At lower stimulus intensities the responses may be subtle, entailing nothing more than increased alertness or an increase in running velocity combined with a reduced sinuosity in the direction taken (Cammaerts et al., 1982). It is further possible to get an alarm response from any volatile compound extracted from ants, if the concentrations used are high enough. A positive reaction cannot be used as evidence that the substance is an alarm pheromone if the experimental concentrations are higher than those that could be generated by the ants themselves. An essential part of the analysis is to create active spaces of the size and geometry resembling those expected under natural conditions. Investigators usually accomplish this end crudely by crushing single glands containing the pheromone.
Let us now examine a particular case of a relatively uncomplicated aggressive alarm system. When a worker of the subterranean formicine ant Lasius claviger is severely threatened, say attacked by a member of a rival colony or an insect predator, it reacts strongly by simultaneously discharging the contents of the reservoirs of its mandibular and Dufour's glands. After a brief delay, other workers resting a short distance away display the following response: the antennae are raised, extended, and swept in an exploratory fashion through the air; the mandibles are opened; and the ant begins to walk, then run, in the general direction of the disturbance (Regnier and Wilson, 1968). Workers sitting a few millimeters away begin to react within seconds, while those a few centimeters distant take a minute or longer. In other words, the signal appears to obey the laws of gas diffusion. Experiments have implicated some of the terpenes, hydrocarbons, and ketones as the alarm pheromones; these are shown in Figure 7-30. Undecane and the mandibular gland substances (all terpenes) evoke the alarm response at concentrations of 109-1012 molecules per cubic centimeter, reflecting a moderate amount of sensitivity as far as pheromones go. These same substances are individually present in amounts ranging from as low as 44 ng to as high as 4.3 µg per ant, and altogether they total about 8 µg. Released as a vapor during experiments, similar quantities of the synthetic pheromones produce the same responses. Apparently the L. claviger workers rely entirely on these pheromones for alarm communication. Their system seems designed to bring workers to the aid of a distressed nestmate over distances of up to 10 cm. Unless the signal is then reinforced by additional emissions, it dies out within a few minutes. The Q/K ratios are on the order of 103-104. If the entire contents of the Dufour's gland, containing about 2.5 µg of undecane, are discharged as a puff from the poison gland, the diffusion model of Bossert and Wilson (1963) predicts that the pheromone signal will reach a maximum of about 20 cm in still air. If on the other hand only 0.1 percent was to be discharged, this active space can still reach a maximum of 2 cm. Hence the match with the observations of natural behavior is reasonably close.
The alerted Lasius claviger workers approach their target in a truculent manner. This aggressive defensive strategy is in keeping with the structure of their colonies, which are large in size and often densely concentrated in the narrow subterranean galleries. It would not pay the colonies to try to disperse when their nests are invaded, and, consequently, the workers have apparently evolved so as to meet danger head-on.
A different strategy, based on panic and escape, is employed in the chemical alarm-defense system of the related ant Lasius alienus (Regnier and Wilson, 1969). Colonies of this species are smaller and normally nest under rocks or in pieces of rotting wood on the ground; such nest sites give the ants ready egress when the colonies are seriously disturbed. L. alienus produce the same volatile substances as Lasius claviger, with the exception of citronellal and citral. Their principal volatile component is undecane. When they smell the latter pheromone, the Lasius alienus workers scatter and run frantically in an erratic pattern. They are more sensitive to undecane than the Lasius claviger workers, being activated by only 107-108 molecules per cubic centimeter. Thus in contrast to L. claviger, L. alienus utilizes an "early warning" system and subsequent evacuation in coping with serious intrusion.
Alarm systems are nearly universal in ants, having been discovered in every species in which a test for the phenomenon was performed. This is true even of the species considered to be among the most primitive, Stigmatomma pallipes (Traniello, 1982), Myrmecia gulosa (Robertson, 1971), and Nothomyrmecia macrops (Hölldobler and Taylor, 1983). By far the most common mode of alarm communication is chemical. Many species, such as the members of Lasius, employ alarm pheromones without the accompaniment of acoustical or tactile displays. In general, it appears that where acoustical and tactile signals exist they have been added onto chemical alarm and recruitment as modulators. However, this does not mean that such signals are specialized or derived with reference to the evolution of the ants as a whole. Stridulatory organs are present in Nothomyrmecia and many poneromorphs. Among members of the latter group, Amblyopone australis and S. pallipes utilize a tactile signal during a vibratory display in which the ants vigorously jerk the head and thorax up and down while coming into contact with nestmates (Hölldobler, 1977; Traniello, 1982). In S. pallipes at least, mandibular gland pheromones also serve as weak attractants. They might serve in alarm, recruitment, or both--not enough experiments have been made to make a distinction.
A remarkable alarm behavior in the Australian bulldog ant, Myrmecia, was recently discovered by Hubert Markl (unpublished observations). On the basis of models of the two-dimensional random alarm process (Frehland et al., 1985), Markl and his collaborators give the following account: "Around an undisturbed nest one usually finds a small number of workers randomly distributed over a few square meters, sitting still or moving about slowly. If an intruder disturbs one of these 'sentinels,' it may trigger it into a frantic, erratic run for a number of seconds. If it thereby comes within sight of a second nestmate, this will lunge forward as if in attack. Upon direct contact, however, combat is avoided and now the second ant starts the same alarm-run while the first one may continue its own. Depending on density and distribution of ants over the guarded area and on number of guards initially stimulated, the alarm can spread two-dimensionally over an extended area. Soon, one or the other of the alarmed workers will return to the nest and there recruit additional forces to join the excited crowd. Intruders into the nest territory will thus be reliably detected and attacked by ants (who can sting ferociously) notwithstanding the stochasticity of the environment. After intruders have been driven out, the alarm subsides by not receiving more stimulating input and/or by having spread out over too large an area. Thus, with a small number of widely distributed guards, the colony can efficiently control a fairly large territory which cannot completely be overrun by any single individual, without engaging more work force than necessary at any given time."
Most of the alarm pheromones identified to the present time are listed in Table 7-4. Their great structural diversity is a reflection of the antiquity and phylogenetic complexity of the ants themselves. Additional variety comes from the fact that most alarm pheromones also serve as defensive substances and have probably obtained their communicative function through ritualization many times independently. In some instances certain exocrine glands initially involved in alarm communication are secondarily hypertrophied and function as the major defensive devices of the ants (Buschinger and Maschwitz, 1984). Defensive chemicals, basic to the biology of social insects, are in turn tied into varying strategies that involve entire syndromes of anatomical structure and behavior. These syndromes are known to evolve across species in correlation with colony size, nest site, and other aspects of natural history. In short, the diversity of alarm pheromones is an expected consequence of phylogeny and the close linkage that exists in ants between defense and alarm. The one common feature the substances seem to share is molecular weight. The great majority of the compounds are in the C6 to C10 range. This is in accord with the prediction (Wilson and Bossert, 1963) that alarm pheromones should evolve in the lower molecular size range because of the need for substantial volatility and a lower Q/K ratio, permitting the rapid expansion and fade-out of the active space. Further, there is little need for privacy in communication, so that molecular complexity and size have not been at a premium during evolution.
- Hölldobler, B. and Wilson, E. O. 1990. The Ants. Cambridge, Mass. Harvard University Press. Text used with permission of the authors.