vol. 173, no. 6 the american naturalist june 2009 Notes and Comments Should Environmental Caste Determination Be Assumed for Termites? Nathan Lo, 1,* Yoshinobu Hayashi, 2 and Osamu Kitade 3 1. Australian Museum, 6 College Street, Sydney, New South Wales 2010, Australia; 2. National Institute of Agrobiological Science, Tsukuba, Ibaraki 305-8634, Japan; 3. College of Science, Ibaraki University, Mito, Ibaraki 310-8512, Japan Submitted September 22, 2008; Accepted January 12, 2009; Electronically published April 8, 2009 abstract: A defining feature of social insects is the differentiation of colony members into either royal or worker castes. It is widely believed that caste is determined by environmental factors, with ge- notype playing little or no role. Social insects are thus considered key examples of polyphenism. In termites, various data support en- vironmental caste determination (ECD) in two of the seven families; however, the evidence is less clear-cut for the other five. An alternative to ECD is genetically influenced caste determination (GCD), in which genotype has a significant influence over caste. We have recently discovered the first case of GCD of royals and workers in a termite. Here we examine previous colony composition studies of various termite species, and we find a number of results that are suggestive of GCD. On the basis of these results, we suggest that ECD should not necessarily be assumed for most termite families, and we outline methods for testing between ECD and GCD. Keywords: sociobiology, social insect, kin selection, sterility, worker. Introduction Social insects such as ants and termites dominate many ecosystems (Wilson 1971). Key to this success is the spe- cialization of colony members into royal and worker castes. Royals are specialist reproducers, while workers are typically sterile or subfertile. The existence of the worker caste is explained by kin selection theory (Hamilton 1964; Maynard Smith 1964). Genes associated with sterility can be favored by selection, so long as they are also present in a royal relative whose reproduction is enhanced by the sterile carrier. Caste is expected to be determined primarily by environmental factors, such as nutrients, pheromones, or temperature (environmental caste determination [ECD]), with genotype playing little or no role (Charles- worth 1980; Crozier and Pamilo 1996; Queller and Strass- mann 1998). Thus, genes and developmental programs * Corresponding author; e-mail: nathan@usyd.edu.au. Am. Nat. 2009. Vol. 173, pp. 848–853.  2009 by The University of Chicago. 0003-0147/2009/17306-50754$15.00. All rights reserved. DOI: 10.1086/598498 causing sterility are “switched on” in workers but remain unexpressed in royals. Early empirical studies of hymenopteran social insects (all ants and many bees and wasps) supported the pre- diction that ECD should be widespread (Wheeler 1986). Studies on termites, which focused on representatives of two of the seven families (Kalotermitidae and Termopsi- dae), also found evidence for ECD (reviewed in Miller 1969). For the other families, the evidence supporting ECD tended to be weaker; data showing that larvae have no predisposition toward a particular caste were not obtained. Nonetheless, it became universally accepted that ECD of royals and workers occurs in all termites (Noirot 1969; Thorne 1997; Roisin 2000; Korb and Katrantzis 2004; Korb and Hartfelder 2008). An alternative to ECD is genetically influenced caste determination (GCD; reviewed in Crozier 1979). In GCD, the genotype of an individual has a significant influence on what caste it will become. Environmental stimuli may still modulate genetic predispositions toward a particular caste (weak GCD), or they may be relatively unimportant (strong GCD; Anderson et al. 2008). Following a few iso- lated reports of GCD in stingless bees (Kerr 1950) and slave-making ants (Winter and Buschinger 1986) during the twentieth century, several examples of GCD have been discovered in ants this decade (reviewed in Keller 2007; Anderson et al. 2008). We have recently demonstrated the first case of GCD of royals versus workers in the termite Reticulitermes spe- ratus (Hayashi et al. 2007; see also Crozier and Schlu ¨ns 2008). This followed an earlier demonstration of genotype having a weak influence among two sterile castes (workers vs. soldiers) in Mastotermes darwiniensis (Goodisman and Crozier 2003). These findings, combined with the fact that experimental verification of the caste determination system of most major termite lineages is lacking, suggest that it may no longer be prudent to assume ECD. Here we ex- amine previous studies of termite caste composition (Myles 1999) and find a number of results that are sug-