Costs and benefits of pleometrosis, as understood from social Hymenoptera, have never been tested in the independently evolved termites. To understand the extent to which such co-founding may be advantageous for colony survival and growth, we tracked the survival and reproduction of 5000 laboratory-established incipient colonies of the facultatively polygamous neotropical termite Nasutitermes corniger. Significantly more pleometrotic groups than monogamous queen–king pairs failed within the first 90 days of establishment, and 99 per cent of pleometrotic groups lost at least one founding member. Oviposition commenced earlier in larger groups, but colony growth was slower and production of workers and soldiers was delayed compared with pairs. Thus, pleometrosis does not increase colony fitness and is in fact highly disadvantageous.
Monogamy is the usual breeding strategy for eusocial insects, including termites , bees, wasps and ants . While monogamy is the ancestral condition in the social Hymenoptera , deviations into polygyny (in the Hymenoptera, associations between multiple queens) and polyandry (one queen mating with multiple males) are successful [4–8] but occur only in a small number of derived taxa . No such systematic studies exist for the termites, however, this group is potentially instructive because males (kings) co-found colonies with queens, resulting in potential for relatively higher promiscuity at different times in the colony life cycle.
Excluding cases in which offspring replace or supplement reproduction by their colony-founding parents, primary reproductives are generally monogamous, although multiple primary reproductives have been reported in nests of various higher termites . Higher termites have an irreversibly sterile worker caste, suggesting they may have passed through an evolutionary ‘monogamy window’ to advanced eusociality , similar to that which occurred in ants . While few studies have assessed relatedness of queens and kings, multiple unrelated reproductives have been documented in two species. In the case of the African fungus-growing Macrotermes species, more than a single unrelated queen–king pair can be found in about 12 per cent of colonies . Pleometrosis, colony foundation by multiple reproductives, has been directly observed in Macrotermes [12,13] and may be adaptive under marginal ecological conditions and/or competition for unevenly distributed resources [11,14]. In the neotropical Nasutitermes corniger, 20–25% of mature colonies are headed by multiple unrelated primary reproductives [15,16]. Although never documented in this species, pleometrosis was also thought to explain the origin of these multiple unrelated queens and kings [17–19]. More recently, experimental fusion of mature colonies  suggests that mergers later in colony development could contribute to the observed pattern. No previous research has systematically examined the costs and benefits of pleometrosis during colony foundation.
In this study, we exploit the fact that N. corniger is highly abundant and produces large quantities of alates that can be experimentally manipulated to establish colonies. We monitored 5000 pleometrotic or monogamous incipient colonies to assess whether multiple founders increase survival or growth during colony foundation. We first examined individual and colony survival as a function of group size and co-founder relatedness, then evaluated colony productivity and offspring caste ratio. We provide multiple lines of evidence that co-founding does not increase group fitness and is almost always disadvantageous.
2. Material and methods
(a) Experimental colony establishment and monitoring
Nasutitermes corniger alates (winged reproductives) were collected from mature (parent) nests in and around Gamboa and Galeta, Panama, just prior to dispersal flights in April–June 2006–2009. They were sorted by sex, then maintained in moist filter paper-lined Petri dishes until use in experimental incipient colonies.
Experimental colonies were formed by placing alates in moist filter paper-lined 30 mm Petri dishes with a fragment of dead white birch (laboratory) or mango (field) wood. Replicate incipient colonies were established for each relatedness (nest-mates, non-nest-mates) × founding group size (pairs, trios and quintets) treatment combination (see the electronic supplementary material, table S1). Although the mating strategy of the parent nests was unknown, nest-mates are assumed to be more closely related than non-nest-mates. In the non-nest-mates treatment, founders were either all from different colonies or all members of the same sex were from the same colony. No significant differences were found between the two non-nest-mate types for any of the measurements collected, so the data were pooled for analysis. Trios and quintets were established with both male and female-biased (2 : 1 and 3 : 2) sex ratios. Female-biased groups were initially chosen to reflect the queen : king ratio of dissected mature nests from this area of Panama ([17,20] and T. R. Hartke 2007–2009, personal observation), however, the majority of colonies were established with a male-biased sex ratio to reflect the natural sex ratio of alates at the time of dispersal and of newly established colonies in the field (; n = 100).
Experimentally established incipient colonies were stacked in boxes and either maintained in an environmental chamber set to 28°C and 80 per cent humidity (2006, 2007) or under ambient conditions in Panama (2008, 2009). Survival of founders and colony growth were monitored daily for 90 days. In 2006 and 2007, observations extended to 220 days post-establishment. The monitoring period was short compared with the natural lifespan of successful colonies, but our field observations and previous studies suggest that this is (i) the most critical period for success in the colony life cycle and (ii) within the median lifespan of incipient colonies established in a season [22–26]. Tracking colonies for longer periods of time would have required destructive sampling, imposing significant repair costs and risks to our young colonies and reducing the reliability of the data over time. While reproductive maturity of colonies would be the ultimate measure of their success, the vast majority of incipient colonies in nature will not survive their first season, so this time span is an appropriate measure of the consequences of colony founding decisions. A total of 5000 colonies were established for this experiment, tracking 16 418 individual alates.
(b) Statistical analysis
Less than 1 per cent of incipient colonies established as trios or quintets survived to 90 days with all founders alive; for analysis, such reduced colonies were retained in their original founding group size category. Although the number of reproductives was reduced, the historical fact of having been established by three or five individuals made those colonies qualitatively more similar to each other than to colonies originally established with fewer individuals. Removing reduced colonies from the analysis entirely was not desirable, because we wanted to explicitly test whether larger founding groups were reducing to an optimal size and whether mortality was affected by founding group size. Colonies were considered ‘dead’ when less than two original founders remained and the reproductive potential of the colony lost.
Individual and colony survival, and time courses to reproductive milestones, were analysed with Cox Proportional Regression using the Efron approximation for ties . All possible covariates (founding group size, relatedness, sex) and their interactions, as well as clustering variables (year, establishment date, parental colony), were used in the initial model, which was simplified in a step-wise manner. The proportional hazards assumption, that there is no difference in the hazard rate of death over time, was checked on the final model. The influence of group size and relatedness on both individual and colony survival and hatching time of eggs, and of relatedness on soldier production, changed over time (failed test of proportional hazards assumption). Interactions with time were added as necessary, and step-wise simplification again attempted.
Significant χ2-tests on the proportion of founders surviving to census points were followed by the Marascuilo procedure to determine homogeneous subgroups. Kruskall–Wallis tests (KW) were performed to determine differences in offspring production at census points and the worker : soldier ratio at 90 days. Significant tests were followed by non-parametric multiple comparisons . The Kolmogorov–Smirnov test was used to compare distributions and Fisher's F-test was used for comparison of variances.
To explore the factors affecting colony growth, we used a generalized linear mixed model (GLMM) fit by the Laplace approximation [29, lmer], with colony of origin of the female and male founders as random factors and accommodating the repeated measure as a random effect between time and replicate. Fixed factors in the model were time post-establishment, relatedness of founders, founding group size, year, sex bias of the original founders and numbers of queens and kings alive at the census point. A founder relatedness by founding group size interaction term, significant in the survival and reproduction analyses, was initially included, but it was not significant in the GLMM and was removed. All statistical analysis was performed in R [30, v. 2.12.2].
Mortality was acute in the first 15 days, declined through 90 days post-establishment, then levelled off (see the electronic supplementary material, table S1) and was within the range of failure (15–100%) found in studies of other species [22–26]. Both founding group size and founder relatedness were significant predictors of mortality; smaller founding groups survived better than larger groups, and nest-mates survived better than non-nest-mates for all founding group sizes (figure 1). When tested at the level of the individual, relatedness of the co-founders was the greatest predictor of mortality (reference group nest-mate pairs; hazard ratio 523.78, p < 0.001), although group size (hazard ratio 5.88, p < 0.001) and the interaction between the two factors (hazard ratio 0.35, p < 0.001) were also significant. When examined at the colony level, the effects of group size (reference group nest-mate pairs; hazard ratio 2.23, p < 0.001) and relatedness (hazard ratio 3.85, p< 0.001) were similar in magnitude and did not interact significantly.
(b) Reproductive milestones
For an incipient colony to be viable, founders must successfully raise offspring from eggs into dependent larvae which eventually mature into sterile workers and soldiers. Workers, in particular, are important for colony growth and development. Founding group size was a significant predictive factor in time to first appearance of all offspring classes (figure 2). Relatedness was a significant factor for all but the worker caste. No interaction was found between number of founders and their relatedness for any offspring class, nor was sex bias of the founders a significant factor in the time course of reproduction.
Founding group size significantly predicted time to oviposition (reference group trios; likelihood ratio 1.07 and 1.14; p = 0.135 and 0.003, for pairs and quintets, respectively), with quintets ovipositing soonest. Time to egg hatch, indicative of effective parental care and founder health, was strongly predicted by group size (reference group quintets: likelihood ratio 1.33 and 1.61 for trios and pairs, respectively; p < 0.001 for both). Thus, although quintets oviposited first, they produced larvae later than pairs and trios.
Only founding group size significantly predicted the development of workers in a colony (reference group quintets: likelihood ratio 1.77 and 2.04 for trios and pairs, respectively; p < 0.001 for both). Smaller founding groups were more likely to produce soldiers than larger founding groups (reference group quintets: likelihood ratio 1.92 and 2.71 for trios and pairs, respectively; p < 0.001 for both). Of the 450 colonies producing at least 1 mature offspring in 2007–2009, most produced a worker first (205, 45.6%; proportion test, χ2 = 18.03, p < 0.001). While 31.6 per cent of colonies (142) produced a soldier first, the remaining 22.9 per cent (103) matured both on the same day. Caste of the earliest maturing offspring did not differ between treatments. The proportion of colonies with a worker first were not significantly different by relatedness (nest-mates , non-nest-mates ; χ2 = 2.65, p = 0.103) or founding group size (pairs , trios , quintets ; χ2 = 4.45, p = 0.108).
(c) Colony growth
The number of surviving queens and the relatedness of colony founders were significant predictors of the total number of offspring at the census points (test of fixed effects in GLMM; z = 5.00, p < 0.001 and z = 2.70, p = 0.006, respectively). Colony size at 15 days was negatively correlated with colony size at the later time points, although colony size was positively correlated between the later time points. Median numbers of offspring were significantly different between treatments at 15 days (figure 3; KW p < 0.001) and 30 days post-establishment (KW p < 0.001), the point when more than 50 per cent of surviving colonies had laid eggs and some had hatched larvae. Median numbers of offspring did not differ between treatments at 60 or 90 days. The number of offspring per queen differed between treatments at all time points (KW p < 0.001), however, because all offspring become sterile helpers at this stage of colony development, this is not an indication of relative fitness of individual queens but may suggest that not all queens contributed equally to the offspring pool in larger groups. The distributions and variances of the data differed between founding group sizes (Kolmogorov–Smirnov tests and Fisher's F-tests; see also electronic supplementary material, figure S1). Caste ratios of offspring at 90 days post-establishment were worker biased (mean 2.8 workers/soldier, median 2.4) and not significantly different between group size (KW χ2 = 1.93, p = 0.381) or relatedness (KW χ2 = 0.75, p = 0.385) treatments.
Pleometrotic colony foundation in N. corniger appears to be disadvantageous compared with colony foundation by monogamous pairs, as shown by a higher likelihood of death early in colony development (figure 1), delayed offspring production (figure 2) and slower colony growth than pairs (figure 3). We find no support for the hypothesized benefit of group colony foundation to reduce incipient colony mortality [18,31,32]. Not only were individual queens and kings more likely to die in pleometrotic groups, but the entire colony was more likely to fail. Behavioural observations and examination of deceased founders did not find evidence for any overt conflicts between founders in this experiment (T. R. Hartke, J. LaRosa, R. B. Rosengaus 2006–2007, unpublished data).
Neither is the hypothesized benefit of increased growth rate under pleometrosis [12,13,18,19] supported by our data. Our pleometrotically established incipient colonies grew more slowly than pairs and had no more offspring than pairs at 60 or 90 days, when the first workers and soldiers developed. This is in direct contrast to the case of facultatively polygynous ants [7,33,34]. In our study, incipient colonies founded by groups were less likely to achieve important reproductive milestones, particularly development of workers and soldiers, than pairs and took longer to do so.
Similar to previous studies [22,24,35], nest-mates in our study survived better than non-nest-mates, however, this is probably an artificial situation with little biological relevance as alates participating in natural mass dispersal are unlikely to pair with nest-mates [36–38]. In N. corniger, workers and soldiers actively exclude or kill alates remaining in a queen- and king-right parent colony after the normal dispersal flights have occurred (, and T. R. Hartke 2007–2009, personal observations), and thus remaining in the natal nest as supplementary or replacement reproductives is unlikely. Delayed oviposition by nest-mates could indicate a preference for non-nest-mate partners, inhibition of reproduction due to continued presence of the odour of their natal nest, or both, and may be an adaptive trait fostering dispersal. Delayed reproduction and thereby less rapid depletion of bodily reserves could also contribute to the increased survival of nest-mates compared with non-nest-mates.
Although laboratory results may not perfectly reflect what occurs in the field, our results reflect both constant climactic conditions (2006 and 2007) and limited exposure to ambient conditions of this species' natural habitat (2008 and 2009), and within the large scale of this experiment any advantage to pleometrotic colony foundation should have been readily apparent. The absence of benefits, and indeed the survival and reproductive disadvantages of group colony foundation, indicate that pleometrosis is not a viable strategy in this species, and that mature polygamous colonies of N. corniger in this population are not typically established via pleometrosis. Data from naturally established colonies and semi-natural microcosm and mesocosm experiments  also support this view. We suggest that polygamous colonies containing unrelated reproductives may instead result from coalescence of young monogamous colonies during development.
Unlike basal termites, which kill or repel additional reproductives during colony foundation [40,41], preliminary experiments suggest that founding N. corniger pairs readily accept additional reproductives during the first 60 days post-establishment, and that young colonies at least nine months post-establishment (approx. 200 workers and soldiers) and probably longer, peacefully meet and merge under laboratory conditions. If colonies merge rather than engage in costly battles for territory, they double their workforce and potential rate of increase, gaining an advantage in future inter- and intra-specific competition. While workers and soldiers from mature N. corniger colonies generally attack non-nest-mates ([39,42] and T. R. Hartke 2007–2009, personal observation), workers and soldiers in incipient colonies seem to be highly tolerant of non-nest-mates, including young queens and kings. Experimental mergers of natural colonies  suggest that some colonies remain tolerant and retain the potential for mergers during maturity. In this way, colonies could accrue the benefits of larger colony size at a younger age, without the risks of pleometrosis. Since multiple kings are not necessary to provide sperm for multiple queens, supernumerary kings are probably eventually eliminated [13,17,39,43].
Colony fusion is relatively common in the basal termites [40,44,45], in which all colony members retain some direct fitness potential which is exercised when colonies fuse or are otherwise stressed [46,47]. In these cases, multiple secondary (neotenic) reproductives may coexist, but not multiple primary reproductives. The intermediate genera Reticulitermes and Coptotermes, with both sterile workers and totipotent helpers, are well known for reproduction by ‘extended families’  but colonies seldom fuse . Within the higher termites, colony fusion has thus far been documented only in N. corniger . Further experimental and natural history work is needed to understand when and how this phenomenon occurs.
We would like to thank the Smithsonian Tropical Research Institute and ANAM for permit and logistical assistance in Panama, and Casey Hamilton and Digna Mattias for assistance in the field and laboratory. We thank Koos Boomsma for helpful comments on an earlier version of the manuscript and four anonymous reviewers whose comments and suggestions improved this manuscript. This research was supported by an NSF CAREER grant (DEB 0447316 and DEB-0930162) and NSF REU (REU-DEB-0822492) to R.B.R.
- Received October 30, 2012.
- Accepted January 15, 2013.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.