Introduction

Individuals of many species benefit from the presence of conspecifics. The inverse density dependence at low density, which is broadly referred to as the Allee effect (1931), has gained considerable attention in population and community ecology (Courchamp et al. 1999), conservation biology (Lande 1988; Groom 1998; Stephens and Sutherland 1999), and invasion biology (Liebhold and Bascompte 2003; Taylor and Hastings 2005; Tobin et al. 2011; Wang et al. 2011; Luque 2013). There are a lot of mechanisms that can cause the inverse density dependence in plants and animals including increased risk of predation (Andrewartha and Birch 1954; Clark 1974), inbreeding depression, loss of heterozygosity (Frankel and Soulé 1981; Charlesworth and Charlesworth 1987; Lamont et al. 1993), and the lack of cooperation (Courchamp et al. 1999; Stephens and Sutherland 1999; Stephens et al. 1999). The most notable cause of the Allee effect is the difficulty among sexually reproducing species in finding mates in low-density populations (Andrewartha and Birch 1954; Dennis 1989; McCarthy 1997; Kuussaari et al. 1998; Courchamp et al. 1999; Boukal and Berec 2002).

Allee effects play an important role in the evolution of mating systems and reproductive phenology (Stephens et al. 1999; Calabrese and Fagan 2004), and theoretical models predict that monogamy is more susceptible to the Allee effect, where males are constrained from having more than one mate (Legendre et al. 1999; Bessa-Gomes et al. 2003). In termites, monogamous pairs of primary reproductives (primary king and queen) found colonies and form lifetime partnerships (Nutting 1969; Nalepa and Jones 1991; Shellman-Reeve 1997). The phenomenon of a general synchronized flight from colonies has been reported for multiple termite species (Weesner 1960). Such synchronized flights reduce predation due to predator satiation (Nutting 1979), increase the likelihood of mate finding (Thorne 1983), and increase breeding between members of different colonies (Wynne-Edwards 1962; Luykx 1986). In the life cycle of termites, alates are at greatest risk of predation during the period from swarming to colony foundation, as they are away from nest-associated soldier defenses (Sheppe 1970; Deligne et al. 1981). After landing, males and females are also exposed to high predation risks from ants, resulting in limited mate-searching periods on the ground. Therefore, termite alates are highly susceptible to the Allee effect. However, little is known about this phenomenon in respect to termite mating and colony foundation.

Reticulitermes speratus Kolbe is the most common and widespread termite species in Japan. In Western Japan, R. speratus alates swarm during the day in May, and although the complete dispersal term lasts 1 week in each specific area, a massive synchronized flight from multiple colonies occurs on sunny days following rain (Matsuura 2002, 2006). An alate breaks off its wings upon alighting, and the resulting dealate travels randomly until it encounters a partner with whom it forms a tandem running pair. The pair then relocates to a suitable nest site, with the male following the female (Matsuura and Nishida 2001). Both heterosexual and homosexual tandems are formed (Matsuura et al. 2002b, 2004); female–female pairs are able to found new colonies by parthenogenesis (Matsuura and Nishida 2001; Matsuura et al. 2002a) and male–male pairs promote survivorship and provide the chance to replace a male in an incipient colony after colony fusion (Mizumoto et al. 2016).

In this study, we investigated the effects of flight timing and alate density on pairing and colony foundation success in R. speratus by a release–recapture experiment in a semi-natural field. The effects of alate density on nest-site preference were also assessed by providing different types of rotten wood at different depths within each experimental section.

Materials and methods

Termite collection

The nest wood of four mature R. speratus colonies (A, B, C, and D) were collected from late April to early May 2014 from pine forests at Hieidaira, Takaragaike, Okamotoguchi and Daigo, in Kyoto, Japan. The colonies were maintained primarily at 20 °C, and the temperature was subsequently raised to 25 °C to control alate maturation and flight timing. Following the emergence of alates from the wood, they were separated by sex and maintained in Petri dishes containing moist filter paper. Wings were removed at the basal structure to mimic dealation (wing shedding) behavior (Matsuura and Nishida 2002). Alates were isolated from the wood just before experiment and used in the experiment within 8 h.

Colony foundation experiment under semi-natural conditions

In an open-air greenhouse (3.0 × 7.2 m, plastic roof and fine mesh walls), four 1.0 × 1.0-m experimental sections (I, II, III, and IV) were established consisting of three layers of humus and soil, which was covered by litter (Fig. 1). Section bottoms and surrounding walls were covered by polyvinyl sheets, and the seams were sealed with waterproof tape. Adhesives were also applied to the outer wall to prevent ants and other predatory arthropods from entering the field. Thirty-six wood pieces consisting of brown rotten pine Pinus densiflora (100 × 100 × 30 mm), and thick (60-mm diameter × 150 mm), medium-sized (30-mm diameter × 150 mm), and thin branches (15-mm diameter × 150 mm) of white rotten oak Quercus serrate were arranged within each of the three layers. The wood species were arranged in a 6 × 6 lattice in each of the three layers in patterns based on random numbers generated by R software (Fig. 1).

Fig. 1
figure 1

Diagram and photograph of the experimental setup. a A diagram showing the setup from above (left), accompanied by a photo image (right). Density sections (I, II, III, and IV) are arranged in order; 10, 20, 40, and 80 pairs of alates, respectively, were released at each time. b Vertical arrangement of the experimental field, consisting of three layers of humus and soil covered by litter. The upper and middle layers consisted of humus, and the bottom layer of soil. Thirty-six wood samples were arranged in each layer. The seams of a polyvinyl sheet and the plastic board were sealed using waterproof tape to prevent termites from escaping the field. Similarly, to prevent ants and other predatory arthropods from entering the field, adhesive sealed the outer wall. c The arrangement of the four types of wood within each layer. Pieces of brown rotten pine and three size classes of white rotten oak were arranged in a 6 × 6 lattice in each of the three layers following random numbers generated by the R program

Full size image

To investigate the relationship between release timing and pairing probability, dealates were released from four different colonies at 3-day intervals. Sib and non-sib dealates showed no difference in pair formation or in their time to excavation (Matsuura and Nishida 2001). On the day of the first release, 10, 20, 40, and 80 pairs of dealates from colony A were released into sections I, II, III, and IV, respectively. Male and female dealates were scattered throughout the entire surface of each section. Three days later, 10, 20, 40, and 80 pairs of dealates from colony B were released in the same manner. The dealates from colonies C and D were also released at 3-day intervals in the same manner. In total, 40 males and 40 females were released into section I (10-pair section), 80 males and 80 females into section II (20-pair section), 160 males and 160 females into section III (40-pair section), and 320 males and 320 females into section IV (80-pair section). One month following the last release, all males and females were collected from each wood pieces (Table S1). Males and females were extracted from the same nest so that all members of an incipient colony were preserved together in a 2.0-ml tube and then used for colony identification by microsatellite analysis.

Microsatellite genotyping for colony identification

DNA was extracted from ten alates from each original colony and from all males and females isolated from the wood pieces. Multiple microsatellite loci were analyzed, including Rf21-1, Rf24-2, Rf6-1 (Vargo 2000), and Rs15 (Dronnet et al. 2004), and compared among samples from the original (Table S2) and incipient colonies (Table S3). One primer from each pair was fluorescent-labeled for detection purposes. Termite DNA was extracted from alate legs using a modified Chelex extraction protocol (Walsh et al. 1991). Polymerase chain reaction (PCR) amplifications were performed in 15.25-μl reactions containing 0.3 μl of 25 mM MgCl2, 0.3 μl of 10 mM dNTPs, 1.5 μl of 10X PCR Buffer, 0.2 μl of 5 U/μl Taq DNA Polymerase (Qiagen, Valencia, CA, USA), 5 pmol of each of the multiplex primers, and 1 μl of the DNA template. Amplification consisted of initial denaturation at 95 °C for 5 min, followed by 35 cycles consisting of denaturation at 98 °C for 12 s, annealing at 60 °C for 30 s, and an extension at 72 °C for 1 min. Following amplification, 1 μl of product was combined with 0.5 μl of GS-600 (LIZ) size standard and 10 μl of HI-DI formamide. Samples were denatured at 95 °C and snap-cooled on ice for 5 min. Sample detection was performed using an Applied BioSystems 3500 Genetic Analyzer. Raw data were analyzed using GeneMapper 5.0 software (Applied Biosystems, Inc., Foster City, CA, USA).

Data analysis

To investigate whether alate density had an effect on pairing success (binary variable: success or failure), a generalized linear model (GLM) with a binomial distribution and logit link function was performed using STATISTICA 10 (StatSoft Inc., Tulsa, OK, USA).

In all models, colony of origin, sex and alate density were treated as fixed factors and the effects of the fixed factors were tested using the likelihood ratio test. Similarly, alate density effect on survival (binary variable: dead or alive) was tested using a GLM with binomial distribution and logit link function. To examine the effects of release timing on pairing probability, the proportion of the combinations of monogamous pairs formed by the alates released on the same day (0-day gap) was assessed and compared to those with 3-, 6-, and 9-day gaps. The proportions were then compared to the null hypothesis, i.e., that the proportion of pairs formed would be the same as that expected from random pairings. A three-way ANOVA was used to investigate nest-site preference, and the effects of alate density, layer, and wood type on the number of males and females extracted from each piece of wood were examined. For these analyses, the density section was included as a categorical variable, and the distribution of individuals among wood types was compared using χ2 tests.

Results

Allee effect on pairing success and survivorship

Pairing efficiency was low at low alate densities. In the 10-pairs/m2/day section, two single males and two single females were recovered from the wood, and no successful pairs were observed (Fig. 2). Alate density was positively correlated with pairing success (general linear model (GLM): estimate  ±   SE = 0.016  ±  0.0030, likelihood ratio χ 2 = 18.93, P < 0.0001; Fig. 3a). The colony of origin also had a significant effect on pairing success (likelihood ratio χ 2 = 20.83, P < 0.0001). There was no significant difference in pairing success between the sexes (likelihood ratio χ 2 = 0.021, P = 0.88). The survival rates of alates were also significantly affected by alate density (estimate  ±  SE = 0.016  ±  0.0038, likelihood ratio χ 2 = 29.63, P < 0.0001; Fig. 3b) and colony (likelihood ratio χ 2 = 33.50, P < 0.001), whereas sex had no significant effect (likelihood ratio χ 2 = 0.91, P = 0.34).

Fig. 2
figure 2

Composition of surviving units extracted from the wood. The number of units observed in each density section is shown in parenthesis. F, single female; M, single male; MF, male–female pair; MM, male–male pair; FF, female–female pair; MMF, a female with two males; MFF, a male with two females; MMM, three males

Full size image
Fig. 3
figure 3

Comparison of pairing success rates (a) and survival rates (b) among the sections receiving 10, 20, 40, and 80 pairs/m2/day. The means of four original colonies released at 3-day intervals were shown. Bars denote standard errors. The alate density had significant effect on pairing success (likelihood ratio χ 2 = 18.93, P < 0.0001) and on survival rates (likelihood ratio χ 2 = 29.63, P < 0.0001)

Full size image

Flight timing and pairing

The Rf21-1, Rf24-2, Rf6-1 (Vargo 2000), and Rs15 (Dronnet et al. 2004) microsatellites were allele-rich, containing four to six alleles each (Table S2). Given that the first three loci exhibited sufficient diversity for source colony identification, Rf21-1, Rf24-2, and Rf6-1 were used for profiling (Table S3).

The timing of alate release significantly affected pairing probability, with pair combinations being significantly different from expectations of random pairing regardless of release timing (χ 2 = 127.74, df = 3, P < 0.0001). Among 75 monogamous pairs recovered from wood, 59 (78.7%) contained males and females released on the same day (0-day gap), 13 (17.3%) showed the 3-day gap, 3 (4.0%) showed the 6-day gap, and no pairs showed the 9-day gap (Fig. 4). The density effects on pairing success and alate survival rate remained highly significant even when the data of 3- and 6-day-gap pairs were excluded from the analyses (pairing success: likelihood ratio χ 2 = 11.05, P < 0.001; survival rate: likelihood ratio χ 2 = 21.94, P < 0.0001).

Fig. 4
figure 4

Release timing and pairing probability. The proportion of monogamous pairs formed by the alates released on the same day (0-day gap), and those with a 3-, 6-, and 9-day gaps were compared with the expected proportion from random pairings. ***P < 0.0001

Full size image

Nest-site preference

Males and females were found in the upper and middle nest layers, but none was found in the bottom layer (Fig. 5). The number of individuals extracted from each piece of wood was significantly influenced by alate density, layer, and wood type (density: F 3,384 = 10.29, P < 0.0001; layer: F 2,384 = 10.53, P < 0.0001; wood type: F 3,384 = 4.06, P < 0.01; Fig. 5). An association between density and layer was also observed (F 6,384 = 6.51, P < 0.0001 by three-way ANOVA), indicating that layer preference differed according to density. The observed distribution among wood types differed from a random distribution (χ 2 = 103.02, P < 0.0001), and also differed according to wood size (χ 2 = 97.70, P < 0.0001). In general, termites preferred rotten brown pine for colony foundation to rotten white oak.

Fig. 5
figure 5

Nest-site preference. Number of individuals extracted from each type of wood in each section

Full size image

Discussion

A positive density-dependent relationship was observed between the number of alates released in each section and the rate of pairing, indicating the Allee effect. Surprisingly, however, no pairs were observed in the 10-pair/m2/day density section, where 10 males and 10 females were released at four different times (i.e., 40 males and 40 females in total) in a 1.0 × 1.0-m area. These data suggest that termites have low pairing efficiencies at low densities. Since this experiment was performed without predators such as ants and spiders, the Allee effect observed may be conservative.

The Allee effect could select for specific termite behaviors during swarming and colony foundation, as it is known to influence dispersal in other animals (Serrano et al. 2005; Fowler 2009; Travis and Dytham 2002). The intercolonial synchrony of swarming would increase alate density, and thus should limit the Allee effect. In addition, synchronized flight would also increase the chance for mating with an alate from a different colony of origin and therefore increase the chance for outbreeding. This study revealed that most pairs were formed by males and females released on the same day (Fig. 4). This result suggests that a colony which disperses alates at a different timing from the synchronous flight of the other colonies would have a reduced fitness. Therefore, the Allee effect may be a selective force behind the maintenance of synchronous flight. The Allee effect may also reduce selection for mate choice because alates have little or no opportunity to choose a mate at low density (Møller and Legendre 2001), although it is known that males prefer larger females in laboratory experiments (Matsuura and Nishida 2001).

The finding that termites are susceptible to the Allee effect following dispersal enhances our understanding of termite invasion. Although alate flight is considered limited, alates aided by the wind can travel long distances, such as between islands, and be introduced to new regions (Nutting 1969). In such situations, however, it is unlikely that alates will successfully locate partners. Our empirical data support earlier studies, suggesting termite invasions are mediated by human transportation of infested timber and that the capacity to generate new colonies from colony fragments facilitates the spread of subterranean termites (Dronnet et al. 2005; Vargo and Husseneder 2009). Indeed, the introduced populations of R. flavipes in Europe exhibit a particular colony breeding structure that is characterized by hundreds of inbreeding neotenic reproductives (Perdereau et al. 2015). In introduced populations or peripheral areas, neotenic-headed colonies are more common than primary-headed colonies of Reticulitermes (Vieau 1996) and Coptotermes (Lenz and Barrett 1982). Strong Allee effects in conjunction with low population-level genetic diversity must diminish the advantage of alate production during invasion. Therefore, as suggested by theoretical models (Travis and Dytham 2002) and empirical studies (Dronnet et al. 2005; Vargo and Husseneder 2009; Lenz and Barrett 1982; Husseneder et al. 2012), termites should exhibit reduced dispersal rates and lower rates of spread during invasion and early range expansion, while human transportation would mediate range expansion. Once established, swarm densities increase rapidly with the additional mature colonies founded in a given location, which would contribute to weaken the Allee effect.

In addition to monogamous pairs, other types of surviving units were observed, including single males, single females, male–male pairs, female–female pairs, a female with two males, a male with two females, and three males (Fig. 2). Although female-only units can reproduce by parthenogenesis (Matsuura and Nishida 2001), single males and male-only units cannot reproduce. However, fusion of the units may provide an opportunity to reproduce, as multiple units were observed in individual wood pieces (Table S1). For example, single males and single females may pair following wood excavation, or fusion of male–male and female–female pairs may also lead to monogamous relationships. Among the 75 monogamous pairs, three pairs showed the 6-day gap of release timing (Fig. 4). Thus, it is unlikely that those males and females stayed in a specific test site for 6 days. These data suggest that pair formation is rare following excavation, but not impossible.

In this study, major alate predators, including ants, spiders, and lizards, were eliminated. Thus, the Allee effect observed may be conservative. Additional studies are needed to evaluate the influence of predators on the Alee effect. Additionally, the sex ratio of alates released into each section was balanced throughout the study. Controversy remains over the prediction that monogamy is more susceptible to the Allee effect than other mating systems are. Although some theoretical models predict that monogamy is more susceptible to the Allee effect (Legendre et al. 1999; Bessa-Gomes et al. 2003), one recent study claimed that susceptibility to the Allee effect is not a general property of a given mating system but depends largely on the sex ratio (Engen et al. 2003; Bessa-Gomes et al. 2004; Lee et al. 2011). The alate sex ratio varies among Reticulitermes spp., such that species with an asexual queen succession (AQS) system (Matsuura et al. 2009; Vargo et al. 2012; Luchetti et al. 2013) have female-biased sex ratios, whereas non-AQS species have equal sex ratios (Kobayashi et al. 2013). Therefore, susceptibility to the Alee effect differs between AQS and non-AQS species in Reticulitermes termites. It must also be noted that sex ratios involves other unidentified factors in other termite taxa. In conclusion, termites provide an ideal system to further understand the relationship between the Allee effect and sex ratio.