All termites secrete trail pheromones from their sternal gland, whereas ants use a variety of glands for this purpose. This and the diversity of chemical compounds that serve as trail pheromones among ants, and the uniformity of chemicals among termite trails, suggest a different evolutionary historical dynamics for the development of chemical mass recruitment in both taxa. Termites in addition show pheromonal parsimony. This suggest a single evolutionary origin of pheromone trails in Isoptera, whereas chemical mass recruitment among Formicidae seems to have evolved many times and in different ways. Despite these very different evolutionary histories, both taxa evolved chemical recruitment systems involving attractants and orientation signals, and at least two divergent decision making system for recruitment. This evolutionary analogy suggests that chemical mass recruitment is constraint by fundamental physical dynamic laws. Artificial intelligence including “mass intelligence” and “ant intelligence”, emulates mass recruitment in interacting virtual agents in search of optimal solutions. This approach, however, has copied only the “Democratic” recruitment dynamics with a single compound pheromone. Ant and termite evolution shows more sophisticated recruitment dynamics which, if understood properly, will improve our understanding of nature and applications of artificial “swarm intelligence”.

1. Introduction

One of the great advantages of society is the use of large numbers of individuals to perform tasks that a lonely individual is unable to perform [2, 3].One of the most studied group task in social insects is recruitment for food retrieval,after an individual discovers a food source that is much larger than what it can handle on its own. Some of the communication signals modulating this recruitment are based on auditory or visual signals, but the most important communication signal used in recruitment, in the great majority of ant and termite species, is chemical. In recruitment to food, these signals are at least of two different kinds as first detailed for ants [4] : one used to orient workers to the food source, that is trail pheromones; another to attract workers to the trail and thus to the food source, that is attractants for food recruitment. Some species use chemicals for only one of these signals and signal the other function by means of tactile or acoustic signals. An illustrative intermediate recruitment system is called “Tandem Running” [5], where the scout physically carries a nestmate to the food source. In tandem calling [6], the recruiting workers lead nestmates to the newly discovered food source by physically guiding them to the source, sometimes using chemical trails to help orientate to the food. Other species lay chemical trails that fulfill both functions, requiring different chemicals for attracting and orienting ants [7, 8]. These intermediate stages in the evolution of chemical mass recruitment, starting from individual foraging, allow us to suggest phylogenies for recruitment systems illuminating the possible evolutionary history of chemical mass recruitment. Such comparisons suggest that the evolution of chemical recruitment seems to have happened several times, at least in ants [9].

Termites seem also to use both type of chemicals, attractants and orientation signals, in their foraging trails [10, 11], although the details of the chemical communication system used by termites are less well known than in ants.

As both, ants and termites, are terrestrial and arboreal, and that both use chemical mass recruitment, we can compare the different chemical recruitment systems known among ants and among termite in order to extract some general rules.

2. Methods

We explored the existing literature collected by Pherobase [1] for publications on trail pheromones for ants and Isoptera and by Bordereau and Pasteels [12] for additional data on Isoptera. Pherobase, among many other things, reports for Isoptera and for Formicidae all publications mentioning chemicals that had been related to trail pheromones by the author of the publication. Pherobase provides, if available, the exact molecular structure of the chemical and the link to the reference where the trail pheromone was published, grouped by taxonomic or by chemical criteria. Thus, for more details of all references indicated in the Tables, the reader should consult Pherobase for ants and Isoptera, and Bordereau and Pasteels for Isoptera.

3. Results

3.1. Ants

Reports of the chemical nature of the communication signal used for recruitment in ants revealed an interesting pattern of chemical compounds. The summary of available data for ants is presented in Table 1. This table shows that, in many cases, the various compounds produced by a single species are very similar as they constitute small variants of a common chemical skeleton, as is the case for Monomorium pharaonis. We suggest that this might be due to the fact that in the biochemical process leading to the synthesis of one or a few active compounds, other chemicals are produced in the process. Indirect evidence for this suggestion comes from other insects where it was shown that synthesizing pure chemicals in pheromone secreting glands is very difficult, if not impossible [13]. In other cases, an adaptive purposeful chemical diversity seems to be present, as chemicals from completely different biochemical pathways are produced as a substrate for the chemical recruitment signal. This is the case for the Atta and Acromyrmex species and Daceton armigerum. In these cases, as shown in Table 1, some compounds have high carbon numbers and low volatility, and others have high volatility, appropriate for the fulfillment of different communication functions such as orientation and attraction.

The chemical survey presented in Table 1 reveals that the pattern of chemical compounds related to trail pheromones in ants correlates with what we know about the decision making behavior used during chemical mass recruitment to food [14]. We know that ants use either one of two decision making systems regulating chemical mass recruitment. The “Democratic” mass recruitment was described in detail for Solenopsis invicta [15] and the “Autocratic” system first described for Atta cephalotes [16]. The main difference is that in the Democratic system, all workers eventually perform all tasks as in Solenopsis; in the Autocratic system, workers specialize either in scouting or in food retrieval [17], as in Atta. The Democratic recruitment system is adapted for fast recruitment towards ephemeral food sources. Here all workers participating in the recruitment process have the same responsibility and add a fixed amount of recruitment pheromone to the trail. The more trail pheromone, the stronger the signal, the more workers are recruited. This leads to an increase in the workforce allowing engaging the maximum worker strength in the shortest possible time, so as to collect a scarce recourse (a recently discovered dead cockroach for example) before a competitor does.

The Autocratic recruitment system is adapted for the simultaneous exploitation of a diversity of durable food sources. Here workers specialize in chemical communication or in food retrieval. Communication specialists then visit different food sources and signal the palatability, quality, or quantity of a food source with varying levels of chemical concentrations. Thus, a very good food source will trigger trail laying with plenty of an attractive chemical, whereas food sources of low quality will be signaled with low amounts of this chemical laid on the trail. This system allows for the fine tuning of sophisticated recruitment activity such as described for several Atta species, where one group of workers recruit nestmates to the tree canopy where they cut large leaves at their base, so that they fall whole to the ground. There, another group of workers is recruited to each of the leaves that accumulate on the ground, where the workers cut the leave in smaller pieces and transport these pieces to intermediate sites, from where another group of workers transport the leaf fragments to the nest [18].

In both cases, the trail needs to be marked with a chemical that will orient workers towards the food source. If the food source is ephemeral in its existence, an efficient chemical mark does not need to last long. As soon as the food has been collected, the chemical evaporates and the trail disappears. For the simultaneous exploitation of several food sources, however, several longer lasting chemical signals could be very useful, as the source could be revisited fast after spots of inactivity due to rain, heat, cold, or other daily rhythmic patterns. Yet a long lasting chemical signal is not appropriate if it has to work also as an attractant, as any changes in the required workforce will take a long time to achieve if the long lasting chemical need, to evaporate first. Therefore in this later case, highly volatile chemicals, together with some of low volatility, are required to modulate recruitment. Species using chemicals to only attract or orient ants need only one—or a few—chemical compounds to perform this function, whereas species using chemical trails for both, attraction and orientation of nestmates, have to produce a range of chemicals for these two purposes.

As Table 1 shows, most ant species seem to use a few compounds as trail pheromone. Only 14 out of 57 species (25%) seem to use more than 3 chemicals, and only 10% of the species listed use six or more compounds. The use of a few compounds corresponds well to Tandem Calling or even to a Democratic recruitment system. In contrast, species such as the leaf cutting and fungus growing ants Atta, Trachymyrmex, and Acromyrmex secrete over six different chemicals on their trails. Other species using the Democratic system, such as Solenopsis, seem to produce much simpler trail pheromones from the standpoint of chemical diversity of compounds. The trail pheromone composition of Solenopsis invicta recalls the case of Monomorium pharaonis discussed above. Although over 5 different chemicals can be recognized in Table 1, all these chemicals have the same chemical skeleton. Thus, the Autocratic chemical recruitment system could be associated to a more advanced chemical signaling. The case of the hunting and recruiting foragers of Daceton armigerum [19] that use a multitude of recruitment strategies is interesting. Table 1 shows that its trail pheromone has many chemical compounds, hinting to a sophisticated diverse chemical communication system.

Many ant species in the subfamily Myrmicinae with large colonies and a sophisticated social structure, use carboxylates and pyrazines to lay their pheromone trail. These are semivolatile compounds. The Myrmicinae, Atta, and Acromyrmex, for example, need to constantly recruit many workers to supply big colonies with a great quantity of leaves which they use as a substrate to grow their fungus. In contrast, ants with less developed societies living in smaller colonies, such as species of the subfamily Ponerinae, use alcohols and acetate, which are more volatile and thus might serve as chemical attractants to trigger foraging to collect ephemeral food sources. Ponerinae individuals feed opportunistically on dispersed food items. This requires quick recruitment of workers, and, as a consequence, the compounds of the pheromone trail are more volatile and less permanent in time, compared to the carboxylates of the leaf cutter ants. In some species of Ponerinae, chemical trails also regulate nest moving [20].

The Formicinae ants are mostly predators but differ from Ponerinae by their greater social complexity, larger colonies, and more diverse worker castes or polymorphism. The trail pheromones of Formicinae species use a mix of compounds that are more complex than that of Ponerinae, probably due to a more elaborate recruitment system. Table 1 reflects this showing among Formicinae, compounds with elevated molecular weights, such as mullein, in addition to compound of low molecular weight and probably low volatility. Formicinae trail pheromone chemistry seems to be closer to the Myrmicinae than the Ponerinae. This suggests trails with both short-term attractant and long-term orientation function. In the case of Dolichoderinae species, the information is scarcer. In the Argentine ant, Linepithema humile, a tramp species with supercolonies of hundreds of thousands of workers, the trail pheromone has short-chain volatile aldehydes, suggesting a foraging strategy with fast short term bouts of recruitment. The continuous reinforcement of a trail made with short lasting volatiles can last long if it is reinforced by hundreds of workers.

3.2. Termites

Termite species also show diverse ecological life types. We know species that live and feed in the same piece of wood, and species that have their nest separated from their food source [21]. But even the “one-piece” life type species possess trail pheromones which they use to recruit workers for defense or nest moving. Termites of “one-piece” life type do not require orientation systems a priori. Secretions of their sternal gland are considered to function in the recruitment of nestmates to source disturbance within the nest. These termites might also use trail following pheromones to colonize new food sources to where they move their nest [22, 23]. Most termites forage on relatively durable food sources containing cellulose. In addition, most termite species forage on several food sources simultaneously, suggesting a recruitment system closer to the above described Autocratic chemical recruitment system, which seem to be the case in the only termite species where this has been explored so far [24]. Table 2 presents what we know about the chemicals used in trail pheromones by termites. The available data shows that pheromone trails among each termite species are constructed with one or a few compounds among a total of 8 chemicals. For the families where chemical trail pheromones have been reported, the Rhinotermitidae, Termitidae, and Kalotermitidae seem to use mainly neocembrene and a dodecatrienol; Nasutitermes corniger uses in addition to these two compounds trinervitatriene; whereas Mastotermitidae and Termopsidae use a trimethylundecadienol for trail following. That is, all trail pheromones in Isoptera are synthesized from a much conserved metabolic route eventually leading to a compound with the same carbon skeleton as that of a dodecatrienol, where two subfamilies have diverged somewhat from the rest in that they synthesize trimethylundecadienol instead of dodecatrienol as the rest of termite species does.

In the case of termites, volatile chemicals for modulating the recruitment of workers are most likely to be used outside the nest or outside the covered galleries. This seems natural if we take into account that most termites forage in galleries which orient workers to their food sources. Very volatile chemicals are of little use in closed environments where they cannot disperse. Long-lasting, low-volatility chemicals may be useful for trail orientation outside the nest and might form the substrate around which galleries are built [25]. Thus, other signals seem to be more appropriate here in modulating communication. Many termite species add feces, saliva, and other secretions to the trail. This explains foraging trails that are reused after several years. A different situation may occur among termites foraging on grasses or leaves in open habitats and foraging at the end of their galleries where they can display very sophisticated foraging and recruitment strategies [26]. When recruitment behavior was explored in an open setting in a Nasutitermitinae [24], the decision making systems used to modulate the recruitment dynamics conformed to the Autocratic kind described for Atta.

4. Discussion

This paper is based only on published reports, and many more compounds used as trail pheromones are surely to be discovered in the future. For example, it is very likely that Atta texana uses a larger pool of compounds as trail pheromones as that reported in Table 1, as it is unlikely to differ very much from other Atta species in this regard. Thus, results in the Tables are biased towards species that have drawn more attention from researchers. Another cautionary remark regards the assessment of volatility based on chemical structure alone. In general, compounds of the same kind of lower molecular weight are more volatile than the ones of higher molecular weight or longer carbon chains. Biologically relevant volatility, however, depends not only on the compound but also on the substrate on which the chemical is secreted, on its concentrations on the substrate, and on the humidity and temperature of the surrounding air. Thus, simple direct correlations between molecular weight, assumed volatility, and behavioral function of a compound should be avoided.

The work behind the literature used for this study, evidently, was not performed with our objectives in mind, but it is unlikely that methodological limitations explain the lack of more chemical compound associated with trail pheromones among termites than among ants. Despite many possible limitations of this study, the large extend of the research effort explored and the large number of species covered guarantee a minimum of robustness that makes drawing conclusion from these data reasonable.

Despite these and other limitations of this paper, we might suggest two basic trends: (1) evolutionary history of the evolution of ant and termite trails is very different, and (2) the dynamics of interacting individuals achieving a recruitment process mediated by chemicals follow basic rules.

4.1. Different Evolutionary Histories between Ants and Termites

The diversity of chemical structures among ant trail pheromones and the uniformity of chemical compounds among termite trails suggest a different evolutionary history for the development of chemical mass recruitment in both taxa. In termites, often trail pheromone compounds are synthesized also by other exocrine glands and are used as sex pheromones. This pheromonal parsimony seems to be characteristic of termites [12] and is not common among ants.

Chemical mass recruitment among ants seems to have evolved at least 8 times [9], whereas chemical mass recruitment among termites seems to be a more conservative phenomenon where all species seem to share a common ancestor that had already developed chemical recruitment. This explains also the large difference between ants and termites in the glands responsible for the secretion of the trail pheromones. Many different glands are used by different species among ants [27], whereas only the sternal gland is used by termites [12]. Another factor explaining this difference is the ecological diversity of ant species, each exploring different food source. Termites in contrast exploit more uniform ecological niches in their search for cellulose.

4.2. Basic Rules Govern the Recruitment Dynamics

The main conclusion from this study is that despite the fact that the evolutionary history of the chemical mass recruitment of ants and termites is different, a similar recruitment dynamics has evolved in both groups. This evolutionary analogy suggests that chemical mass recruitment is constraint by basic physical-dynamic laws. This would explain the convergence to chemical mass recruitment in the two evolutionary processes studied. A third convergence towards similar solution for the modulation of mass recruitment dynamics is nowadays repeated in the development of artificial intelligence, where the “mass intelligence” of ants copied in the interaction of simple virtual computer agents is in search of optimal solutions. Artificial intelligence, however, has copied only the simple recruitment dynamics named here as the Democratic system with a single compound pheromone. More sophisticated modeling could bear fruits to artificial intelligence that might echo the fruits chemical mass recruitment that has brought to social insect species evolving them.