Termites are eusocial by possessing two subfertile or sterile castes, the worker and the soldier. The consensus is that eusociality in termites is the result of a suite of factors (Thorne, 1997), though the relative importance accorded to each in the literature has shifted over time (Howard and Thorne, 2011). For the purpose of a quick review, I have not looked into mechanisms that appear to have been set-aside in the literature, such as asymmetric relatedness and cycles of inbreeding and outbreeding (cited in Howard and Thorne, 2011).


Soldiers are monophyletic for all extant termite taxa (citations in Howard and Thorne, 2011), and so it is generally taken as a given that soldiers evolved before workers (e.g Korb et al., 2012; Roisin and Korb, 2010; Boomsma, 2013; Tian and Zhou, 2014, though this depends upon how one defines “workers”). It is unclear whether weapons evolved before or after the evolution of juvenile helper forms (Howard and Thorne, 2011), however Nalepa (2010) makes the convincing argument that alloparental care must have evolved first because soldiers are dependent upon nestmates for feeding.

Comparing across taxa there are a number of powerful predictors of the existence of a soldier caste, including: nesting within an abundant food resource soldiers and high relatedness (Boomsma, 2013; Tian and Zhou, 2014). Two main hypotheses have been put forward for the evolution of soldiers in termites specifically (references in Saiki et al., 2014): (1) intra-colony competition between replacement reproductives (Myles 1986), perhaps in the context of the death of primary reproductives during inter-colony competition and colony fusion (Thorne et al., 2003), and (2) inclusive fitness and colony defence against inter- and conspecific enemies (Roisin and Korb, 2010).

The intra-colony competition hypotheses (Myles cited in Roisin, 1999) has fallen out of favour. For example, in Zootermopsis, where soldier-like reproductives are observed, neotenics without soldier traits are about as common, raising the question as to why the latter persisted (review of other criticisms in Roisin, 1999). The intra-colony competition hypotheses implies that the sterile soldier caste evolved from a reproductive soldier-like morph. It is possible that soldier-like characteristics in reproductives evolved to increase the individual’s chance of defeating and usurping the position of primary reproductive during intercolony warfare (Thorne et al., 2003). This is supported by the observation that soldier-like reproductives commonly become the replacement reproductives after colony fusion (Johns et al., 2009). However a study into hormonal regulation during the development of soldiers versus soldier-like reproductives suggests that soldiers evolved independently rather than via an intermediate soldier-like phase (Saiki et al., 2014).

The role of relatedness in soldier-caste evolution is supported by the pattern of its occurrence in relation to the genetic system across taxa (Higashi et al., 2000). Nalepa (2010) makes the compelling argument that that alloparental care was a necessary antecedent to soldiers, and relatedness was the key driver of alloparental care. Changes in morphology resulting from the evolution of alloparental care may have also favoured the evolution of a soldier caste. In Cryptocercus, the primary reproductives are responsible for defence as well as brood care. If evolution of alloparental care involved heterochronic changes that led to paedomorphosis, then this would favour the allocation of defence from primary reproductives to a soldier caste (Nalepa and Bandi, 2000).

The particulars of Cryptocercus defence may also provide some hints about the drivers involved in soldier-caste evolution. Cryptocercus displays alarm behaviour (“complex oscillatory movement”, Howse 1995) that has an analogue in Zootermopsis (Seelinger and Seelinger, 1983). The behaviour is produced by nymphs in response to both conspecific intruders and predators. Adult defensiveness is correlated with presence of juveniles, however adults rarely attacked conspecific intruders smaller than their own offspring (Park and Choe, 2003). Instead, resident nymphs themselves frequently attacked smaller intruders, chasing and then trying to bite them around the anal area (Park and Choe, 2003), which is also the preferred area for live cannibalism of sick nestmates (Nalepa, 1994), probably due to its urate storage.

Other roles for soldiers include accompanying alate dispersal (Roux and Korb, 2004), as foraging scouts (citations in Tian and Zhou, 2014), exploration, recruitment, and egg care (citations in Roisin and Korb, 2010).


Two drivers feature most prominently in the current literature for the evolution of delayed dispersal: inclusive fitness via alloparental care, and hopeful reproductives. While the two drivers are not necessarily mutually exclusive, a tension exists between them in the termite case because the proposed mechanism for the hopeful reproductives hypothesis – inter-colony warfare – will erode the high relatedness that promotes alloparental care via inclusive fitness.

Helpers and inclusive fitness</h3>

The narrative for inclusive fitness is best described in the works of Nalepa on Cryptocercus. In short, the proto-termite ancestor is hypothesised to have been similar to Cryptocercus, a subsocial one-piece nester. Strong nitrogen limitation was a key selective pressure, imposing a heavy cost upon the reproductive pair, preventing the resumption of egg-laying. Alloparental care therefore had a high potential payoff: to free the primary reproductives to resume egg-laying within the already-obtained nesting/food resource. Given strong selection for high parental care, alloparental care may have been initiated as correlated behavioural character in juveniles (Nalepa, 2010), and coprophagy provided the substrate upon which proctodeal trophallaxis could develop (Nalepa and Bandi, 2000). This triggered a cascade of events, including two potential positive feedbacks. First, the hormonal changes that modified the moulting sequence had the side-effect of disrupting symbiont cyst formation between moults, such that newly moulted termites became dependent upon proctodeal trophallaxis from nestmates, thus reinforcing social cohesion (Nalepa, 1994). Second, an increase in the number of juveniles in the colony plus paedomorphosis favours the evolution of a defensive caste, which reinforced the small-bodied and fragile morphotype (Nalepa, 2015).

The inclusive fitness mechanism is enhanced by high relatedness, which could have been achieved due to monogamy (Boomsma, 2013), inbreeding (Bratz to read) and/or chromosomal translocations (cited in Tian and Zhou, 2014). The mechanism is also strengthened when the assured fitness returns of alloparental care (Queller 1989 cited in Howard and Thorne, 2011) compare favourably against the low probability of successful dispersal and nest founding. Predation risks during dispersal are high in termites (Korb and Schmidinger 2004 cited in Hoffmann and Korb, 2011), on the order of less than 1 percent of dispersers succeed in founding a new nest (citations in Roisin and Korb, 2010).

Helpers and hopeful reproductives

The hopeful reproductive theory is that offspring will delay dispersal for the chance to inherit the ter- ritory. The idea has two main components: (1) the ecological constraint, that the offspring’s probability of succeeding after dispersal is low; and (2) inheritance, that the long-term benefit of staying is the chance to inherit the valuable territory. Therefore alloparental care evolved in two stages: first delayed dispersal, then second cooperative breeding when helpers received sufficient direct and/or indirect benefits to contribute (Korb et al., 2012).

Termite primary reproductives can be long-lived, however the accelerated inheritance hypothesis notes that opportunities for nest inheritance arise during to inter-colony warfare and colony fusions (Thorne et al., 2003). Colony fusion is found in three termite families thought to resemble proto-termites in various ways: Mastotermitidae, Archotermopsidae, and Kalotermitidae (Howard et al., 2013). Specific examples can be found for Zootermopsis (Johns et al., 2009; Howard et al., 2013) and Cryptotermes (Korb and Foster, 2010). Whether colony fusions are peaceful or aggressive has been found to depend upon colony size, which may reflect some mechanistic constraint (Korb and Roux, 2012) or may reflect a relationship between colony size and payoffs.

The conflict between the two processes

The main conflicts between the inclusive fitness and hopeful reproductive hypotheses involve the issue of relatedness. In contrast to inclusive fitness, theoretical models show that the hopeful reproductives mechanism can work in situations of low relatedness. There are many examples of unrelated cooperative breeding birds, and differences in relatedness do not appear to explain variance in helping behaviour (citations in Kokko et al., 2001). Furthermore, colony fusions can lead to nonrelatives inheriting the nest (citations in Johns et al., 2009). High genetic mixing has been observed in colonies of Kalotermes (Luchetti et al., 2013) and Zootermopsis (cited in Korb, 2007), and Nasutitermes colonies can have multiple reproductive kings and queens (Atkinson and Adams, 1997).

If delayed dispersal is also about alloparental care and inclusive fitness, then we should expect increased alate production with decrease in relatedness, however experimental results have been equivocal (Korb and Schneider, 2007): reducing relatedness within colonies did not have an effect, but individuals in inbred colonies were less likely to disperse. Korb (2007) observed that most nestmate interactions in Cryptotermes appeared to be reciprocal and only soldiers were net recipients, however this is disputed (Nalepa, 2015, and citations therein). Nevertheless, the importance of nitrogen to termites and Cryptocercus point to the potential for selection for alloparental care in the proto-termite ancestor, therefore it is hard to discount the possibility that these species’ lack of alloparental care is secondarily derived (Korb et al., 2012).

Recent phylogenetic evidence shows that “true workers” (in contrast with pseudergates) evolved once only, rather than on three separate occasions, as previously thought (Thompson et al., 2000). It is unclear how to relate this to alloparental care. The difference between these castes is not functional but developmental (the bifurcated development pathway vs totipotency): “true workers” are not necessarily sterile, and pseudergates can take part in the same worker activities as larvae and nymphs (Roisin and Korb, 2010). If the bifurcated development path is the ancestral rather than derived condition, then the flexibility of the pseudergate caste may have evolved as a bet-hedging strategy for delayed dispersal in response to increased environmental variability, regardless of what process drove dispersal delay.

This issue also has implications for the evolution of the soldier caste. The existence of a solider caste in termites appears to fit the cross-taxa pattern of soldier castes evolving in highly-related fortress defenders (Boomsma, 2013). Yet colony fusions erode relatedness, so what is their effect on soldier evolution? It is noted as well that high relatedness may be a consequence rather than a cause of cooperation if cooperation keeps a group intact (Nonacs, 2011).

Environmental correlates of worker behaviour may shed some light on the processes that drive it. In general, alloparental care is low in wood dwellers compared to foragers, where the former is considered the ancestral lifestyle (Korb et al., 2012). A general trend of shift to earlier caste determination in more derived species with larger colonies has also been reported (Howard and Thorne, 2011). Sterile workers are absent in one-piece nesters, and it is generally thought that the transition from one-piece to foragers increased monogamy which led to sterile helpers (obligatory rather than facultative helping) (citations in Higashi et al., 1991; Boomsma, 2013). Grooming in general also depends upon pathogen load (Korb et al., 2012), which is higher in one-piece nesters and in dampwood species. An alternative narrative is that, by the time the transition to foraging had been made, helpers had already evolved, and the evolutionary trajectory had past a point of no return such that reversion to solitary or subsocial life was no longer possible (citations in Johns et al., 2009).

Modelling opportunities

Given that all termites left extant are eusocial, the best that can be done to uncover the evolutionary path is to develop hypotheses that are both consistent with the biology of termites and with the principles of evolution (Thorne, 1997). Theoretical modelling is well suited for this task, but work to date has come under strong criticism for downplaying termites’ taxon-specific properties (Nalepa, 2015). From what modelling literature I have read so far, there appears to be no effort to reconcile the controversies in the literature that is recent, and the most recent models cited are not termite-specific.

The literature appears to have two somewhat incompatible narratives for the evolution of termite eusociality: inclusive fitness and accelerated inheritance. Howard and Thorne (2011) states that “it may be less important to discuss which process is key or driving than to attempt to measure the relative contributions of each to social evolution in different situations”. An older model by Higashi et al. (1991) did just this; they explored the difference in conditions under which false versus true workers would evolve. However the model was interested in how one-piece nesting versus foraging led to the evolution of reproductive versus sterile workers, where the primary difference between nesting styles was nest stability in terms of log collapse. This model could be updated and adapted to make nest stability relate instead to intercolony warfare and fusion, and how that effects inclusive fitness versus accelerated inheritance as drivers for delayed dispersal. The goal would be to distinguish the preconditions for each mechanism, which may elucidate the most likely sequence of events.

Some of the work of Kokko may also provide a basis upon which to build a model. The models are largely inspired by birds, however changes to the details of the model might be made to account for the biological particulars of a hypothetical proto-termite ancestor. For example, the model of Kokko and Johnstone (1999) accounts for the individual fitness of a subordinate, dominant, and lone disperser, in terms of reproductive share and as a hopeful reproductive. The model is useful because it provides a concrete example of how re- latedness can be accounted for in individual fitness, and how to account for the future transition probabilities between states.

Kokko and Ekman (2002) models safe havens theory, which is the idea that the natal territory provides a base from which the individual can survey neighbouring territories for recent vacancies, or wait for the chance to inherit the nest. The model excludes inclusive fitness considerations. It is primarily concerned with comparing a safe haven to a floater strategy, the latter of which allowing the individual to survey a larger number of territories but with a lower survival rate. Alates are generally considered the only dispersing termite morph and do not survey neighbouring territories, yet this model is commonly cited by proponents of the accelerated inheritance hypotheses (e.g. Korb and Schmidinger, 2004; Roisin and Korb, 2010). However if the ancestral condition includes wingless dispersal (as in Cryptocercus and as hypothesised in Bourguignon et al. (2016)), then neighbourhood surveying might be a possibility, providing a a stronger justification for invoking safe haven theory. We could also simulate the hypothesised transition from warm tropical lowlands to more challenging environments with patchy resource distribution by shifting the surveying component and investigating its effect on the proclivity to delay dispersal. It would be interesting to see if switching between the inclusive fitness versus accelerated inheritance mechanisms is possible in both directions. Kokko et al. (2001) shows how the two-rank models above can be extended to a larger number of individ- uals, and also addresses the issue of group augmentation, where individuals survive and reproduce at higher probabilities in larger groups. The model would require some modification to fit the survival probability description to the context of inter-colony warfare. Also, the difficulty Kokko et al. (2001) encountered of describing inheritance probabilities for large numbers of subordinates may be simplified in our case, if we assume that termites do not form a dominance hierarchy (as modelled in Kokko et al., 2001) but rather that inheritance precedence is a combination of honest signal (via proctodeal trophallaxis frequency) and fair lottery (as was suggested in Hoffmann and Korb, 2011).

Another possibility would be to look at delayed dispersal as a bet-hedging strategy (Yoshimura and Clark, 1991, for a simple example of bet-hedging). The nest-inheritance theories above trade off the high probability of obtaining low fitness from remaining as a helper for another season against the low probability of high fitness gains by founding a new nest and raising one’s own offspring. The trade-off may find some dispersal delay time that optimises expected fitness, such that the expected gain from delaying dispersal any longer is equal to that of dispersing now. However an added dimension of interest is in the variance of the two strategies. In temporally variable environments, fitness is taken from the geometric, not the arithmetic, mean growth rate (Gillespie, 1974). If the variance in the payoff from dispersing is higher than from staying, then the true stochastically optimal evolutionary strategy may favour a longer delay in dispersal than would be expected from the expected value of the fitness alone. The dynamics become more interesting when we take into account that the payoff from delay includes nest inheritance, the probability of which is dependent upon the number of siblings present, which in turn depends upon their own delay-time strategies.

The question of why colonies should fuse (as opposed to, for example, individuals from losing colonies being cannibalised for their nitrogen) appears quite open from a theoretical/modelling standpoint. The work of Korb and Roux (2012) suggests some possible benefits of fusion, including increased survival of fused colonies, and higher individual survival and colony growth when fusions are peaceful.

Finally it is noted that the timing of colony fusion is under direct control of the workers as they are responsible for tunnel digging (Korb and Roux, 2012). It may be in the interests of a hopeful reproductive to initiate an inter-colony war that will kill the primary reproductives. This may inspire some game theoretic model.


Atkinson, L. and Adams, E. S. (1997). The origins and relatedness of multiple reproductives in colonies of the termite nasutitermes corniger, Proceedings of the Royal Society of London B: Biological Sciences 264(1385): 1131–1136.

Boomsma, J. J. (2013). Beyond promiscuity: mate-choice commitments in social breeding, Philosophical Transactions of the Royal Society of London B: Biological Sciences 368(1613): 20120050.

Bourguignon, T., Chisholm, R. and Evans, T. (2016). The termite worker phenotype evolved as a dispersal strategy for fertile wingless individuals before eusociality., The American Naturalist 187(3): 372.

Gillespie, J. H. (1974). Natural selection for within-generation variance in offspring number, Genetics 76(3): 601–606.

Higashi, M., Yamamura, N. and Abe, T. (2000). Theories on the sociality of termites, Termites: Evolution, sociality, symbioses, ecology, Springer, pp. 169–187.

Higashi, M., Yamamura, N., Abe, T. and Burns, T. P. (1991). Why don’t all termite species have a sterile worker caste?, Proceedings of the Royal Society of London B: Biological Sciences 246(1315): 25–29.

Hoffmann, K. and Korb, J. (2011). Is there conflict over direct reproduction in lower termite colonies?, Animal behaviour 81(1): 265–274.

Howard, K. J., Johns, P. M., Breisch, N. L. and Thorne, B. L. (2013). Frequent colony fusions provide opportunities for helpers to become reproductives in the termite zootermopsis nevadensis, Behavioral Ecology and Sociobiology 67(10): 1575–1585.

Howard, K. J. and Thorne, B. L. (2011). Eusocial evolution in termites and hymenoptera, Biology of Termites: a Modern Synthesis, Springer, pp. 97–132.

Johns, P. M., Howard, K. J., Breisch, N. L., Rivera, A. and Thorne, B. L. (2009). Nonrelatives inherit colony resources in a primitive termite, Proceedings of the National Academy of Sciences 106(41): 17452–17456.

Kokko, H. and Ekman, J. (2002). Delayed dispersal as a route to breeding: territorial inheritance, safe havens, and ecological constraints, The American Naturalist 160(4): 468–484.

Kokko, H. and Johnstone, R. A. (1999). Social queuing in animal societies: a dynamic model of reproductive skew, Proceedings of the Royal Society of London B: Biological Sciences 266(1419): 571–578.

Kokko, H., Johnstone, R. A. and Clutton-Brock, T. (2001). The evolution of cooperative breeding through group augmentation, Proceedings of the Royal Society of London B: Biological Sciences 268(1463): 187– 196.

Korb, J. (2007). Workers of a drywood termite do not work, Frontiers in Zoology 4(1): 1.

Korb, J., Buschmann, M., Schafberg, S., Liebig, J. and Bagnères, A.-G. (2012). Brood care and social evolution in termites, Proceedings of the Royal Society of London B: Biological Sciences 279(1738): 2662– 2671.

Korb, J. and Foster, K. R. (2010). Ecological competition favours cooperation in termite societies, Ecology Letters 13(6): 754–760.

Korb, J. and Roux, E. (2012). Why join a neighbour: fitness consequences of colony fusions in termites, Journal of Evolutionary Biology 25(11): 2161–2170.

Korb, J. and Schmidinger, S. (2004). Help or disperse? cooperation in termites influenced by food conditions, Behavioral Ecology and Sociobiology 56(1): 89–95.

Korb, J. and Schneider, K. (2007). Does kin structure explain the occurrence of workers in a lower termite?, Evolutionary Ecology 21(6): 817–828.

Luchetti, A., Dedeine, F., Velonà, A. and Mantovani, B. (2013). Extreme genetic mixing within colonies of the wood-dwelling termite kalotermes flavicollis (isoptera, kalotermitidae), Molecular Ecology 22(12): 3391–3402.

Nalepa, C. A. (1994). Nourishment and the origin of termite eusociality, booktitle pp. 57–104.

Nalepa, C. A. (2010). Altricial development in wood-feeding cockroaches: the key antecedent of termite eusociality, Biology of termites: A modern synthesis, Springer, pp. 69–95.

Nalepa, C. A. (2015). Origin of termite eusociality: trophallaxis integrates the social, nutritional, and microbial environments, Ecological Entomology 40(4): 323–335.

Nalepa, C. A. and Bandi, C. (2000). Characterizing the ancestors: paedomorphosis and termite evolution, Termites: Evolution, Sociality, Symbioses, Ecology, Springer, pp. 53–75.

Nonacs, P. (2011). Monogamy and high relatedness do not preferentially favor the evolution of cooperation, BMC Evolutionary Biology 11(1): 1.

Park, Y. and Choe, J. (2003). Territorial behavior of the korean wood-feeding cockroach, cryptocercus kyebangensis, Journal of Ethology 21(2): 79–85.

Roisin, Y. (1999). Philopatric reproduction, a prime mover in the evolution of termite sociality?, Insectes Sociaux 46(4): 297–305.

Roisin, Y. and Korb, J. (2010). Social organisation and the status of workers in termites, Biology of Termites: A Modern Synthesis, Springer, pp. 133–164.

Roux, E. and Korb, J. (2004). Evolution of eusociality and the soldier caste in termites: a validation of the intrinsic benefit hypothesis, Journal of Evolutionary Biology 17(4): 869–875.

Saiki, R., Yaguchi, H., Hashimoto, Y., Kawamura, S. and Maekawa, K. (2014). Reproductive soldier-like individuals induced by juvenile hormone analog treatment in zootermopsis nevadensis (isoptera, archotermopsidae), Zoological Science 31(9): 573–581.

Seelinger, G. and Seelinger, U. (1983). On the social organisation, alarm and fighting in the primitive cockroach cryptocercus punctulatus scudder, Zeitschrift für Tierpsychologie 61(4): 315–333.

Thompson, G., Kitade, O., Lo, N. and Crozier, R. (2000). Phylogenetic evidence for a single, ancestral origin of a trueworker caste in termites, Journal of Evolutionary Biology 13(6): 869–881.

Thorne, B. L. (1997). Evolution of eusociality in termites, Annual Review of Ecology and Systematics pp. 27–54.

Thorne, B. L., Breisch, N. L. and Muscedere, M. L. (2003). Evolution of eusociality and the soldier caste in termites: influence of intraspecific competition and accelerated inheritance, Proceedings of the National Academy of Sciences 100(22): 12808–12813.

Tian, L. and Zhou, X. (2014). The soldiers in societies: defense, regulation, and evolution, International Journal of Biological Sciences 10(3): 296.

Yoshimura, J. and Clark, C. W. (1991). Individual adaptations in stochastic environments, Evolutionary Ecology 5(2): 173–192.