Introduction

Termites typically live within family units that consist of a reproductive pair and their sub-fertile offspring—the latter of which are specialized to help with brood care, foraging and nest defense (Nalepa and Jones 1991; Shellman-Reeve 1997; Korb et al. 2012). The specialization of worker and soldier castes into non-reproductive roles can evolve through indirect selection, provided that worker and soldier altruism is directed towards reproducing relatives (Hamilton 1972). Eusocial insects like termites should, therefore, have some capacity for kin or nestmate recognition (Breed 2014).

Nestmate recognition has often been reported in termites (e.g., Thorne 1982; Kaib and Brandl 1992) but its occurrence and expression seems to be conditional and potentially sensitive to genetic and ecological circumstance. Where it does occur, it remains uncertain how inherited (Adams 1991) versus environmentally acquired (Matsuura 2001; Florane et al. 2004) cues contribute to species or nestmate recognition. For subterranean species (Rhinotermitidae), chemically variable cues seem to play an important role (Clément et al. 1988)—at least for directing inter-colony aggression and delimiting territorial boundaries (Bagnères and Hanus 2015; Costa-Leonardo and Haifig 2014). Workers and soldiers can both defend territorial boundaries against intruders but soldiers in particular are well-adapted to this function, as evidenced by their conspicuous behavioral (e.g., positioning, blocking, attacking), mechanical (e.g., biting, crushing, slashing) and chemical (e.g., irritating, sticky, toxic secretions) defenses (Šobotník et al., 2010). Both workers and soldiers may also participate in more passive defenses like social avoidance or physical barriers to maintain kinship within colonies and reduce the potential costs of injury and death (Prestwich 1984; Shelton and Grace 1996; Matsuura and Nishida 2001; Li et al. 2010).

Controlled assays using Reticulitermes taxa have shown that antennation between non-nestmates can elicit aggressive, even lethal, contests (Clément and Bagnères 1998; Shelton and Grace 1996; Thorne and Haverty 1991; Polizzi and Forschler 1999; Huang et al. 2012). However, in other assays, such behaviors are rare (Fisher and Gold 2003) or absent (Grace 1996; Perdereau et al. 2011; Bulmer and Traniello 2002b). Some of this variation in behavior may be due to the experimental protocol used, with commonly used Petri-dish assays providing an easy but biologically unrealistic approach compared to study designs which allow termites to burrow in substrate (Cornelius and Osbrink 2003; Messenger and Su 2005; Chouvenc et al. 2011; Chouvenc and Su 2017). Intrinsic factors, such as the size or caste ratio of test populations (Polizzi and Forschler 1998) or differences in genetic background (Dronnet et al. 2005; Vargo and Husseneder 2009) may also contribute to variation in aggression and recognition behavior.

In this study, we use arena contests to test for evidence of aggression and nestmate recognition in populations of Reticulitermes flavipes in southern Ontario. Specifically, we predict that if aggressive behavior is present then it will increase as a function of geographic distance. Reticulitermes flavipes was introduced to Ontario on at least three separate occasions (Scaduto et al. 2012), and makes for an interesting case study in recognition and territorial defense. First, the different founding backgrounds provide a potential source of genetically variable recognition cues. Second, the Canadian populations show a mix of ‘open’ versus ‘closed’ societies (Matsuura and Nishida 2001; Clément 1986), and, therefore, vary in their propensity for inter-colony aggression and resource defensiveness.

The City of Toronto in particular is infested with termites that show low levels of genetic diversity (Scaduto et al. 2012) and have low inter-colony aggression (Grace 1996). Moreover, these urban colonies are diffuse, with multiple reproductive centers that are spread across whole city blocks (Myles 1996). The Toronto populations are, therefore, typical invasive ‘open societies’, as are other displaced populations of this species (e.g., Dronnet et al. 2005; Perdereau et al. 2015). In contrast, termites in the Pelee region, form discrete colonies that are genetically well-differentiated from each other (Raffoul et al. 2011). The Pelee termites live independently of any human habitation, are potentially native (Kirby 1965) and are effectively ‘closed societies’.

In this study, we use Petri-dish assays to test for short-term recognition in the form of antennation and aggression, and a shared-resource design assay combining field caught-colonies to monitor for evidence of nestmate sorting and non-nestmate conflict within a more natural context, over a longer timeframe.

Methods

Termite collections and housing

We sampled free-living termites between May and October 2015 from three field locations in Southwestern Ontario: Toronto (43.685°N, − 79.318°W), Point Pelee National Park (41.588°N, − 82.326°W) and Pelee Island (41.481°N, − 82.377°W). To trap termites, we used single face, two-ply corrugated cardboard roll traps (10 cm × 10 cm), each fitted with a plywood lid (15 cm × 15 cm), as previously described by Grace (1989) and Raffoul et al. (2011). We buried traps ~ 3 cm below the soil surface at each of the three locations. To ensure we obtained samples from at least two different colonies per location, we set multiple groups of traps (more than n = 10) at sites at least 1.5 km apart. For two such urban sites (Toronto ‘A’ and ‘B’; Table 1), we collected from residential properties in or around Danforth Village. For non-urban trap sites at Point Pelee and on Pelee Island, we collected termites under woody debris along or in close proximity to the Lake Erie shoreline. At Point Pelee National Park and Pelee Island colonies consisted of ~ 95% workers, 1% soldiers and 4% reproductives, while in Toronto collections consisted of ~ 98% workers, 1.5% soldiers and 0.5% reproductives. In total, we collected between 38,000 and 40,000 termites were collected throughout the field season. We housed all field-caught termites at 24 °C (± 1), 60% RH on a 16 L: 8 D photoperiod in 6 L plastic Tupperware containers filled with field-collected substrate and inert sand. We added 120 g of damp maple and oak wood shavings, along with pieces of softwood (cedar) plywood. Finally, we kept colonies hydrated by misting them 2–3 times per week, prior to using them within 1 month for behavioral assays.

Table 1 Distance (km) between Reticulitermes flavipes sites (at least two per location) at different locations in southwestern Ontario (Canada)
Full size table

Behavioral assays

To measure antennation and aggression between individuals from same versus different colonies, we established test populations of variable size (density) and caste type (Table 2). We replicated each of these conditions 10× with different sets of individuals, for a total of 210 assays. For Petri dish assays, we first transferred individual termites from their housing container onto clean plastic weigh boats, and paint marked them by colony of origin (as in Matsuura and Nishida 2001; Polizzi and Forschler 1998). We then briefly held (1-h) each group of marked termites in covered 3.5 cm Petri dishes (lined with moist filter paper) prior to transferring a single individual from one dish (the ‘intruder’) into the other (the ‘resident’). We then filmed and scored behavioral interactions under red light for 5 min (using a Sony HDR-CX700 camera in nightshot mode). From video footage, we scored level of aggression of each resident-intruder interaction along a four-point scale: (0) antennation where one termite contacts the other (either the head, body or anus) with antennae, but no aggression, (1) mandible contact between workers or mandible flaring by soldiers, (2) forward lunge body lunges forward and mandibles open by workers or soldiers, (3) biting by workers or soldiers.

Table 2 Details of Petri-dish and shared-resource design assays considering geographic origin of Reticulitermes flavipes colonies, the number of termites (density) in each assay and ratio of workers (w) to soldiers (s)
Full size table

To test for more subtle forms of recognition, we looked for evidence of nestmate sorting and resource competition between combined colonies over both 2- and 7-day periods. 2-day assays were carried out following the 7-day assays to confirm whether any mortality observed in pairings may have been due to aggressive interactions or factors unrelated to conflict such as starvation. In the 2-day assays, we used three-compartment chambers that allowed individuals from two compartments to intermingle as they tunneled and foraged through a common resource (the third compartment). We fed n = 15 workers on either plain or stained (0.1% w/w Nile Blue A) filter paper (as in Su et al. 1991), then placed them into left or right compartments of a modified Petri dish (Fig. 1a). In the seven-day assay, by contrast, we fed a larger number of workers (~ 500) per colony on either plain or stained filter paper prior to placing them in the left or right compartments of a custom-made termite complex (Fig. 1b). At the end of each experiment, we removed colonies and counted the number of living termites in all three compartments to determine survivorship. We also examined the proportion of workers recovered in their natal/home base or in another compartment. For both the 2- and 7-day shared-resource assays we compared nestmate and non-nestmate pairings as shown in Table 2.

Fig. 1
figure 1

Experimental set-up used to examine the distribution and survivorship of workers as they tunnel and forage through a shared resource. a Modified Petri dish (9.5 cm diameter) used to examine the distribution and survivorship of R. flavipes workers over a short (2-day) window. The upper left and right compartments house the termites and are filled with moist sand. The lower middle compartment has only a moistened Whatman #7 filter paper as a food resource. b Adapted from Grace (1996), this chamber is used to examine the distribution and survivorship of R. flavipes workers over a longer (7-day) window. The side compartments were 2.3 L Tupperware containers filled with moistened inert sand, joined by eight small glass tubes to the central chamber that contained moist sand, 60 g of water-soaked maple and oak shavings, two corrugated cardboard rolls and small wooden blocks. Two longer glass tubes also directly joined the compartments

Full size image

We analyzed all Petri dish behavioral assays using one-way ANOVAs, followed by Tukey’s HSD tests. We scored mean aggression score and antennation time(s) as response variables, and scored colony-of-origin (i.e., nestmate or non-nestmate) as a fixed factor. For the 7-day shared-resource assays, we quantified movement of termites using quasi-binomial logistic regression to account for over-dispersion in the model (dispersion factor = 22.5). This test was used to compare whether nestmate pairings [Toronto (TO): TO–TO, Pelee Island (PI): PI–PI] were more likely to move from their home compartment to other chambers compared to non-nestmate pairings (TO-PI). Finally, we arcsine-transformed the observed proportion of surviving termites before using a one-way ANOVA, and applying Tukey’s HSD test. We performed all statistical analyses using the R statistical programming platform (R Core Team 2015; version 3.2.2).

Results

One-on-One behavioral assays

Aggressive interactions were nearly absent in the one-on-one short-term Petri dish assays, regardless of geographic origin or caste (see supplementary material, Table S1). Specifically, all measures of aggression—that is, mandible contact (workers), mandible flaring (soldiers), lunging or biting (any caste)—were very rare and expressed in less than 3% of interactions. There was, therefore, no difference in mean aggression score between nestmate (TO–TO) and non-nestmate (TO–PI) pairings for worker–worker (ww: F3,36 = 1.80, P = 0.17), worker-soldier (ws: F3,36 = 1.80, P = 0.17) or soldier–soldier castes (ss: F3,36 = 1.35, P = 0.27). Further time spent antennating did not differ significantly regardless of their geographic origin in the different one-on-one assays (see supplementary material, Figure S1; TO resident–resident pairing: F3,36 = 2.82, P = 0.05; PI resident–resident pairing: F2,25 = 0.87, P = 0.43) or between residents and intruders (all p values were > 0.4; data not shown).

Changes in density: 5-on-1 and 5-on-5 behavioral assays

The frequency and intensity of aggressive behaviors remained low when the ratio of residents to intruders was varied, regardless of the castes involved (see supplementary material, Table S2). The 5-on-1 assays did, however, detect differences in antennation that involved soldiers (see supplementary material, Figure S2). Specifically, time spent antennating was longer for resident–intruder (i.e., non-nestmates) pairings that involved worker–soldier (F1,18 = 14.26, P < 0.001) and soldier–soldier (F1,18 = 5.26, P = 0.03) pairings. Pairings that did not involve soldiers, by contrast, showed no difference in antennation between resident and intruders (worker–worker pairings; F1,18 = 0.02, P = 0.88). The 5-on-5 assays showed similar patters (Supplementary material: Figure S2): antennation between resident and intruder was longer for soldier–soldier (F3,24 = 1.54, P = 0.23) and worker–soldier (F3,36 = 0.08, P = 0.05) pairings, but was not different for worker–worker pairings (F3,36 = 0.43, P = 0.74).

Shared-resource behavioral assays

In assays involving nestmates (i.e., TO−TO, PI−PI), the majority of workers were found in the middle (shared) resource area (Fig. 2). Furthermore, the left and right compartments always contained some non-nestmate migrants, as evidenced by the mix of stained and unstained workers. For assays involving non-nestmates, by contrast, the majority of workers were found in their home container, and there was little evidence of mixing (Fig. 2). When both stained and unstained individuals were found in the middle resource section during assays testing inter-colony differences, there was some evidence of colony separation—for example, in some replicates, one corrugated roll contained only stained workers while the other only had unstained workers.

Fig. 2
figure 2

The median number of stained (gray) and unstained (white) R. flavipes survivors in left or right compartments after 7 days when nestmates (TOcTO, PI−PI) or non-nestmates (TO−PI, TOA−TOB, TO−PP) shared the same resource. The shade of the background square indicates which side stained (gray) or unstained (white) individuals were placed at the start of the experiment and the pie charts indicate the proportion of survivors found after 7 days. The total number of survivors in middle compartment is indicated between the squares

Full size image

A logistic regression analysis suggests that termites are far more likely (92%) to migrate between chambers when they are nestmates (e.g., TO−TO, PI−PI; Odds ratio = 0.08, 95% CI0.02, 0.28, P < 0.001) then when pairings involve non-nestmates (e.g., TO−PI).

Survivorship

Worker survival varied within and between nest pairings for both 7-day and 2-day resource design assays. Survivorship was consistently higher in nestmate pairings (TO−TO, PI−PI), as well as non-nestmate pairings from within the same geographic region (TOA−TOB and PP−PI) than for geographically distant non-nestmate pairings (TO−PI) (F4,11 = 11.04, P < 0.001; Fig. 3). The majority of dead individuals were never found, possibly because R. flavipes workers are known to exhibit undertaking behavior (cannibalism, burial) within minutes of discovering a dead conspecific to reduce the probability of infection (Neoh et al. 2012; Sun et al. 2013; Sun and Zhou 2013). Those that were found were too decomposed to determine if aggressive interactions were the cause of death.

Fig. 3
figure 3

Survivorship (± SE) of colonies (stained and unstained combined) in 7-day shared-resource assays

Full size image

Survivorship was not associated with the use of Nile blue A cell stain, as there were no differences between stained and unstained workers in either the 7-day- (PI−PI and TO−TO: F5,18 = 10.27, P = 0.75; TO−PI: F5,18 = 10.27, P = 0.98) or the 2-day resource foraging assays (PI−PI: F5,80 = 1.53, P = 0.99; TO−TO: F5,80 = 0.80, P = 0.99; TO−PI: F5,80 = 0.84, P = 0.71).

Discussion

In this study, we used short- and long-term behavioral assays to test for evidence of aggression and nestmate recognition in Canadian populations of the Eastern subterranean termite. The results from all short-term Petri dish assays did not support the original hypothesis that inter-colony aggression would increase with increasing geographic distance. First, inter-colony aggression was almost absent, with no differences in aggression regardless of geographic origin, caste or density, and only a small number of soldiers exhibited mandible flares and lunging.

Second, while antennation time may be an indicator of nestmate discrimination (Clément and Bagnères 1998), there were no differences between nestmates and non-nestmates, with the exception of groups of five residents (workers or soldiers) antennating non-nestmate soldier intruders longer than nestmate intruders. The aggressive behaviors and longer antennation by soldiers may relate to the fact that they are the first line of defense, and engage in entrance guarding behaviors to prevent intruders from entering the nest.

These findings are consistent with previous observations suggesting that R. flavipes seems to lack inter-colony aggression and possibly kin or nestmate recognition (Grace 1996; Polizzi and Forschler 1998; 1999; Bulmer and Traniello 2002a, b; Fisher and Gold 2003; Perdereau et al. 2011). However, all these studies used Petri dish assays that never lasted more than 24 h. One possible explanation for the absence of aggression is that only a small proportion of workers are expected to be aggressive (Polizzi and Forschler 1998, 1999). However, as more than 200 workers were tested in the Petri dish assays, it seems unlikely that we failed to sample a single aggressive individual if one were present in the populations. A small number of soldiers repeatedly expressed more aggressive responses, so it is possible that there is polyethism in this caste, an aspect that merits further investigation.

Petri dish assays are convenient but do not incorporate ecological context. Our 7-day and 2-day shared-resource assays in which termites could tunnel and forage in soil revealed a significant increase in worker mortality in inter-colony pairings compared with intra-colony pairings. We suggest this mortality resulted from aggression, as other factors such as disease or starvation would be less likely to occur over a short (2-day) timeframe. In our resource assays, encountering non-nestmates while foraging for new resources may have elicited aggressive behavior. It is unknown whether aggressive interactions between non-nestmates occurred while feeding in the middle resource or when entering the other’s home environment. If initial aggressive interactions between non-nestmates occurred when foraging for resources, survivors may have remained in their respective containers and thus avoided individuals from the other colony. Unfortunately, our study does not distinguish between these two possibilities. Although mortality was still observed in nestmate pairings, we believe this is the result of disturbance created during initial set up of assays.

As the foraging sites of both intra- and interspecific colonies of Reticulitermes spp. frequently overlap in the field (Thorne et al. 1999; DeHeer and Vargo 2004; Perdereau et al. 2011), we would expect the evolution of behaviors that minimize the cost of aggressive encounters. For example, in a shared-resource-based assay R. flavipes and Coptotermes formosanus constructed separate foraging tunnels (Cornelius and Osbrink 2000). Similarly, in a planar arena study, agonistic interactions where observed when C. gestroi and C. formosanus tunnels intersected, followed by tunnel blockages at multiple locations to prevent further conflict (Li et al. 2010). In our study, the increased antennation of non-nestmates observed by soldiers in the Petri dish assays could be associated with entrance-guarding behavior that ensures only nestmates enter the territory.

Fisher et al. (2004) reported that when two unrelated colonies were mixed under laboratory conditions they shared food resources, and while workers engaged in trophyllaxis and brood care of unrelated nestmates, no interbreeding occurred. However, several genetic studies on R. flavipes field colonies, collected at the same foraging site, suggest that extensive intermixing and intercolony breeding is generally avoided in nature (Bulmer and Traniello 2002a; DeHeer and Vargo 2004, 2008; DeHeer et al. 2008), although familial structure is highly variable across populations (Bulmer et al. 2001), and colony fusion with mixed families (although rare) can occur, particularly in invasive populations (Perdereau et al. 2015). Territoriality is possibly due to cuticular hydrocarbon profiles, which are one way that termites can differentiate nestmates from non-nestmates (Bagnères and Hanus 2015). The near absence of inter-colony mixing observed in the 7-day assays suggests that R. flavipes are able to discriminate nestmates from non-nestmates. These results, therefore, question the idea that supercolony formation in Toronto or other invasive populations are the result of bottlenecks stripping founder population of hydrocarbon cues and a break-down in recognition behavior.

The Toronto colonies are approximately 90 years old (Urquhart 1953) and form expansive colonies that occupy whole city blocks (Myles 1996), and are spreading to other nearby regions (Kirby 1965; Grace 1990). By their single introduction, the Toronto colonies are, therefore, genetically similar and may have similar hydrocarbon profiles. The high level of intermixing observed in assays involving different colonies from Toronto (TOA−TOB) and the presence of open social colonies in Toronto may be the result of the “dear enemy effect” (Breed 2003) where related colonies avoid overt conflict, especially if resources are limited. In contrast, the Pelee populations are older and potentially native (Husby 1980), have discrete colonies (Raffoul et al. 2011) and are restricted to the sandy shore habitats of Lake Erie.

Unfortunately, R. flavipes lives in underground tunnels making it difficult to study behaviors under more realistic ecological conditions (Thorne et al., 1999; Deheer and Vargo 2004). Several previous studies have made use of a planar arena (Luscher 1949; Messenger and Su 2005; Chouvenc et al. 2011; Bernard et al. 2017), where individuals forage between stacked sheets of glass, with the benefit of allowing for the recording of movement and interactions of individuals in real time. We recommend the use of behavioral assays that incorporate natural substrate for improving the understanding of foraging behavior of R. flavipes and help clarify some of the apparent contradictions that exist in the current literature about the expression of aggression in this species. It will also help address questions such as: do foragers establish non-overlapping trails and block tunnels to avoid colony mixing? When and where do aggressive encounters normally occur, and which caste exhibits these aggressive behaviors? How might the addition of soldiers in different ratios effect worker movement? Does aggression and intermixing follow a graded response as a function of geographic distance, with higher levels of intermixing between neighboring colonies than with more distant colonies?