Termites are social insects that undergo paurometabolous development and have morphologically distinct castes [1,2]. They are generally divided into two groups with the higher termites, members of the family Termitidae, representing a diversity of derived species while the remaining families, considered the ancestral lines, are termed the lower termites. The lower termites are known for their developmental plasticity [1]. The various castes and patterns of sex allocation displayed in the lower termites are especially intriguing, but the mechanism underlying aspects of their developmental biology are confused by inconsistencies reported in the literature [3,4,5,6,7,8,9,10,11]. The literature hints at an ontogeny complicated by flexibility. This confusion is exemplified by the fact that there have been at least nine versions of the life cycle proposed for the economically important pest genus Reticulitermes [2,6,12,13,14,15,16,17,18].

One acknowledged feature of subterranean termite life history is neoteny. Neotenics have long been reported and recognized as occurring in two phenotypes, ergatoids (without wing buds) and nymphoids (with wing buds), arising from worker and nymphal castes respectively [1,2,14,15,19,20]. In this study, we report three nymphoid neotenic phenotypes distinguished by the extent of cuticle pigmentation and present sex ratios of all castes from census of field and laboratory colonies. These observations and data provide valuable insight into the ontological options available to the subterranean termite Reticulitermes flavipes (Kollar). We also discuss published life cycles and caste determination/developmental models to illustrate the dynamic nature of Reticulitermes biology while suggesting the plasticity shown by this species may contribute to its status as an invasive species.

Neoteny in Reticulitermes has been reported as a common occurrence in the literature for the past 100 years [19,20]. Snyder [28] mentioned that neotenics developed in field populations of R. flavipes each spring, in quantities ranging from a few dozen to several hundred; an observation supported by the three swarm-time collections used in the present study. This is corroborated by additional field studies that report a higher proportion of neotenics than primary reproductives [29,30]. The literature is also rife with reports of Reticulitermes neotenics found in the presence of a primary reproductive from laboratory cultures [5,19,22,31,32,33]. It is evident that neoteny in R. flavipes is an independent process that occurs in the presence or absence of primary reproductives.

The factors that influence the production of neotenics in Reticulitermes have never been the topic of directed research although it was generally believed that environmental influences were the primary factor [1,34,35]. A number of recent studies with R. speratus have illuminated the potential of a genetic component [36,37,38]. The debate between proponents of environmental caste determination (ECD) and genetic caste determination (GCD) as outlined in Lo et al. [39] will likely continue as more data is accumulated using molecular techniques [40].

An examination of our neotenic data in combination with published reports on castes and caste development can point to avenues of further research. Neotenics have long been reported and recognized as occurring in two phenotypes, nymphoid and ergatoid, with two further, functional distinctions (supplementary, replacement) based on presence or absence of the primary reproductives [1,2,21]. However, our collections provided additional nymphoid phenotypes distinguished by the degree of cuticle pigmentation: RN, BHN, BN (Figure 1). We assume that BHN and BN are terminal stages that received mixed hormonal/abiotic signals as they exhibit the darkened cuticle characteristic of the adult stage but are clearly neotenic because wing development was incomplete and evidenced as 'buds' (did not form into functional wings) while the eyes were, likewise not fully developed (Figure 1). The reason these insects did not follow the 'typical' path to the adult stage but instead transitioned to a neotenic is unknown but the outcome is obvious from looking at the specimens.

There is ample evidence that neotenics are produced concurrently with alates as a normal development in the life history of certain R. flavipes populations. This evidence includes the field observations of Snyder [28]; our unpublished data from 80 field collection sites (inspection ports reported in Forschler and Townsend 1996 [21]) sampled four times a year for five years (2000 -2005) that always (N = 6) provided neotenics in collections that also contained alates; and the spring time collection of alates and neotenics from Log’s A, B and C in the present study. The concurrent production of alates and neotenics could be explained by dissemination of a colony-level ‘reproductive’ signal. In response to that signal nymphoid neotenics could occur as a result of timing, abiotic, and/or physiological conditions sending some nymphs to the ‘short-cut’ of neoteny rather than completing the path to adulthood. The Log A, B, and C collections also illustrate that some nymphs switch from the imaginal line to neoteny at a point where a remnant ‘adult’ signal remains - resulting in the gradation of pigmentation (partial sclerotization) evidenced in the BHN and BN.

The difficulty of maintaining healthy Reticulitermes cultures illustrates that something is missing in our attempts to mimic the field condition [22,41]. Production of alates and nymphs is rare in Reticulitermes cultures [21,32]. In the eight laboratory cultures we examined only three provided significant numbers of nymphs including two that had no primary reproductives. One five-yr-old culture had three primary reproductives one female, two males, and over 100 nymphs providing evidence that nymphs occur in colonies containing the royal pair in addition to highlighting the occurrence of polyandry also reported by Grube and Forschler [21]. Why most laboratory cultures of Reticulitermes do not produce a high proportion of nymphs may have something to do with the suggestion that a limited resource results in culling older workers that would normally go to the nymphal stage [21]. As is clear from the work of Lenz et al. [30] termites in the field preferentially exploit large food resources adding weight to the hypothesis that resource size and quality have an impact on data obtained from laboratory colony growth and development. Examination of Reticulitermes development under varied laboratory conditions must be conducted to elucidate the potential impact of extrinsic factors and without such data laboratory results must be interpreted with circumspection.

Genetic control of Caste Differentiation (GCD) has recently gained attention with parthenogenesis as the cornerstone of a queen-centric system [36,37,38,39]. That research is unanimous in using a simplified life cycle, assuming that neotenics differentiate (only) upon the death of the queen [39]. The GCD life cycle includes a sex-linked inheritance pattern that favors production of alates through inbreeding by worker-derived, ergatoid neotenics [37]. Hayashi et al. 2007 [37] used experimental laboratory pairings of various neotenic phenotypes then separated eggs to be reared by workers. They concluded that there is some hormonal control because “in the presence of reproductive individuals, the nymph genotype may be modified into a worker phenotype” [37]. Matsuura et al. [28] used natural populations and five microsatellite markers to suggest that neotenics were clones of the queen while workers and “alate nymphs” were almost all produced from sexual reproduction. This body of work is based on the model proposed by Thorne et al. [18] that assumes neotenics differentiate only as replacement reproductive and has been widely used as the basis for interpretation of microsatellite data [42]. A model that assumes only replacement neoteny, monogyny, an undefined generation time, and direct association of Simple Sequence Repeats (SSR’s) with strict Mendellian inheritance is clearly a good starting point but an oversimplification of the actual system. The fact that there is a genomic component to caste determination is indisputable. Genes are regulated and must be ‘turned on’ to produce mRNA that is translated into proteins, which result in particular phenotypes. The approach of using RNA Interference (RNAi) seems to hold promise for examining the processes underlying the interplay between hormone titers, gene expression, and caste phenotypes [43,44].

How does the sex ratio data from our three field populations and eight laboratory cultures/colonies match the literature? The terminology used to describe the various castes confounds comparisons between researchers. Buchli’s [15] 8^th pseudergate instar would be called workers using the terminology of Thorne [23]. We will assume that Buchli’s pseudergate is the same as the ‘worker phenotype with a nymph genotype’ mentioned in Hayashi et al. [37] or the pseudergate sensu lato of Korb and Hartfelder [2]. The confused physical features of partially-pigmented nymphoid neotenics (BN and BHN) would indicate that the hormonal cascade which results in tanning—a rite of passage to the terminal imago stage—is received and (partially) processed by a life stage that doesn’t carry the adult genotype. The Buchli [15] life cycle provides numerous developmental options and can account for the various nymphoid neotenics and caste sex ratios we obtained from the three Log collections in Table 1.

The Log A data demonstrates, assuming an equal colony-level sex ratio, a split by soldiers and neotenics from the same developmental path because of the opposite sex ratios in those castes (Table 1). In the Log A collection, the BN and BHN were male-biased, while soldiers were neutral tending to a female-bias. The others castes including workers, alates, RN, and nymphs were obvious neutral. Furthermore, it provided a neutral sex ratio for the combined neotenic and soldier data (Table 1). This information points to a developmental bifurcation with a male-biased neotenic or female-biased soldier as the choice. The life cycle proposed by Buchli [15] has this junction at the 6^th nymphal or psuedergate instar while the 7^th and 8^th instars do not have that option (Figure 2). This system also fits the hypothesis that the BHN and BNs received the signal to become reproductives too late to be alates because the RN, arising from later instars (7^th and 8^th instars) provided a neutral sex ratio.

Indicative of the plasticity of Reticulitermes development the Log B and C data display different pathways. Logs B and C provided RN ratios that were female biased which is consistent with descriptions in the literature [20,29,33,45,46,47,48,49,50]. Log B had no BN but of the 99 BHN all but five were male. Although this ratio was heavily male biased it does demonstrate that females can also be ‘caught’ in the hormonal confusion assumed to produce these pigmentation-differentiated phenotypes. The fact that the soldier sex ratio from Log B was obviously neutral is contrary to the sex ratios from Log A. The worker sex ratio was however male biased indicating that the bifurcation occurred at a different location in the developmental pathway. The Buchli [15] life cycle again provides that option at the 8^th pseudergate instar that is past the soldier pathway (Figure 2). Log C also provided a neutral sex ratio for soldiers (N = 32) that tended to female-bias along with female-biased RN (N = 18), BHN (N = 4) and nymphs (N = 18), while workers (N = 200) and BN (N = 5) were male-biased. These data contrast with the BHN from Log A and Log B that were male biased (Table 1). The three collections (Log A, B & C) provided indications of biased sex ratios in all neotenic phenotypes. Therefore it appears that adult maturation can follow different sex-linked pathways.

Reticulitermes appear to have two out of 21 chromosomes that are sex-linked and the presence of a sex-linked developmental pathway should be acknowledged in future research [37,51,52]. The occurrence of one pair of sex-determining chromosomes (XX in females, XY in males) is thought to be the basal sex-determination condition in termites, but fusions or translocations could complicate this simple line of reasoning [53,54]. The life cycle proposed by Vieau [6] indicated a definable morphological differentiation between the worker/soldier and imaginal lines in Reticulitermes that is reminiscent of Grasse [14] and Buchli [15]. That split was re-affirmed for R. speratus by Hayashi et al. [37] who examined 3^rd instar offspring after crossing various combinations of nymphoid and ergatoid neotenics and subsequently incubating the eggs in worker-only groups. They found that the sex ratios of nymphs and workers followed a single X-linked locus, dubbed worker (wk), with two alleles, A and B. Their model predicted that offspring in imago-competent colonies would be workers with a neutral sex ratio. Nymph production would only occur should the queen and/or king die and ergatoid neotenics develop [37,39].

Laboratory Colony V was the only group we examined that fit the expected one-locus-allele genetic model proposed by Hayashi et al. [37] by having a queen-right colony with a neutral worker sex ratio. Colony V at six years of age contained the primary reproductive pair and only workers and soldiers. Colony V provided a neutral sex ratio for both soldiers (N = 58; sex ratio = 1.23) and workers (N = 226; sex ratio = 1.48) although the worker ratio tended toward male-bias (Table 2). However, the other five queen-right colonies provided male-biased worker populations (Table 2). Nymph production in these same five colonies, however, was as predicted by Hayashi et al. [37] because nymphs represented ≤ 1% of the colony population (Table 2). The two ‘queenless’ colonies provided conflicting data regarding nymph production as predicted by the single X-linked genetic model. Colony I contained 61 neotenics including 11 BHN and a population with less than 2% nymphs in contrast to Colony II that had two female RN and a nymphal proportion of 19% (Table 2). Colony II fits the one-locus-allele model for proportional nymphal production but not with the male-biased worker population. Colony I is interesting because BHN were found in a laboratory culture and combined with the low nymph proportions suggests that this culture had some remnant of an ‘adult’ signal—to produce BHNs and low numbers of nymphs despite containing only neotenic reproductives. The variability recorded in the present study with R. flavipes may be due to intrinsic biological attributes of that invasive species [55,56,57] compared to other members of the genus.

Caste differentiation in Reticulitermes is much more plastic than predicted by recently published life cycles [6,18,58,59]. Genetic models predicated on neoteny only in the absence of the primary female draw conclusions that are oversimplifications of a complicated biological system. Although recent genetic-centric data sets are arguably valuable in understanding caste differentiation in Reticulitermes, the influence of extrinsic inputs must be part of a holistic approach to understanding the true beauty of this system.

The work of Grassé [14] and Buchli [15] outlined a developmental life history based on observation of morphologically distinct forms. The body of knowledge represented by the literature suggests that this 50-year-old research should be revisited and tested using the powerful tools available in the current century.

In conclusion, the information gained from this sex ratio study indicates a variety of developmental pathways in subterranean termites that are not consistent with any single model of caste differentiation. Numerous factors have been proposed to affect lower termite caste development including a sex-linked gene [37], juvenile hormone titers [8,9,60], hexamarins [43,44], pheromones [61] and quality/quantity of food resources [62]. The interaction between intrinsic and extrinsic factors must be considered in future studies of what is undoubtedly a complex biological system.