Discussion
Here we provide evidence that polygyne fire ant colonies in Texas are not unicolonial and may engage in low levels of nepotism. Polygyne colonies did not exchange resources or workers, indicating that they maintain strict colony boundaries (Fig 3). Polygyne workers also preferentially fed larval siblings and may have cannibalized non-siblings during times of stress (Fig 6), suggesting that within-colony conflict exists. Our behavioral results also corresponded with higher levels of within-mound relatedness between workers in the field than those previously reported in North America (Fig 5; but see Ross, 1993). Our results suggest that polygyne fire ant colonies are actually multicolonial and may engage in high levels of intraspecific competition in at least parts of their introduced range (Weeks, Wilson, & Vinson, 2004).
Counter to our expectations, we detected distinct colony boundaries between almost all mounds in the field regardless of social form (i.e., polygyne mounds were no more likely to share than monogyne mounds; Fig 3) and within-mound relatedness between workers (Appendix S5). The mounds that did share with each other were likely part of the same polydomous colony based on genetic results and spatial distance between mounds (Fig 4). Although our results seem to run counter to previous assumptions about polygyne fire ants (e.g., Greenberg et al., 1992; Holway et al., 2002; Morel et al., 1990; Plowes et al., 2007; Porter et al., 1992; Vander Meer et al., 1990), several other studies have found evidence of boundaries at least on some level between polygyne colonies (Goodisman et al., 2007; Krushelnycky et al., 2010; Weeks et al., 2004). For example, polygyne mounds in Georgia, USA, showed distinct genotypic frequencies and worker weight profiles, suggesting that workers and queens are not moving freely between mounds (Goodisman et al., 2007). Moreover, although polygyne workers do not aggressively attack non-nestmates like their monogyne counterparts (Vander Meer et al., 1990), workers will antennate and occasionally bite non-nestmates in the laboratory, indicating well-developed nestmate recognition (Obin, Morel, & Vander Meer, 1993). There is also evidence of exploitative competition between polygyne mounds in the field (Weeks et al., 2004).
Our results confirm that polygyne mounds located near each other are not completely interconnected (Goodisman et al., 2007; Krushelnycky et al., 2010; Weeks et al., 2004); in fact, polygyne mounds in our study were just as isolated as monogyne mounds. Moreover, boundaries appear to be present at relatively small spatial scales, as many mounds of both social forms did not exchange resources despite being within 5m, and sometimes even less than 1m, from each other (Appendix S6). Weeks et al. (2004) found that most labeled polygyne fire ant workers remained within 4m of their colony. In our study, we found mounds with distinct boundaries separated by less than 1m, suggesting mounds very close to each other may belong to different colonies. These results imply that fire ants are able to distinguish nestmates from non-nestmates, even when environmental odor cues may be similar from living in close proximity. Heritable and environmental odor cues are thought to be additive in fire ants, but monogyne and polygyne fire ant workers have been shown to distinguish nestmate from non-nestmate despite similar environmental odor cues (Obin et al., 1993; Obin, 1986).
Our laboratory experiment provides further evidence that polygyne fire ants are not as cooperative as previously assumed, as workers preferentially fed sibling over non-sibling brood (Fig 6a). Workers may have even preferentially cannibalized non-sibling brood, because there was a significantly greater reduction in the number of non-sibling brood remaining at the end of the experiment compared with sibling brood (Fig 6b). All colonies were kept in standardized laboratory conditions and fed standardized diets to minimize acquired, environmental identification cues (Obin et al., 1993), so worker recognition of sibling over non-sibling larvae is likely based on heritable as opposed to environmental odor cues. One reason for the greater disappearance of non-sibling brood could be due to differential brood viability between families, which can cause “sham nepotism” (Holzer, Kümmerli, Keller, & Chapuisat, 2006). We believe this scenario is unlikely, however, given the short time frame and reciprocal nature of our experiment. Furthermore, queens allowed to found in isolation do not express differential viability of offspring (Vargo & Ross, 1989). Larvae were given to the colonies by placing them outside of the nest dishes and allowing the workers to bring them into the nest, so it is also possible that workers collected greater numbers of sibling brood than non-sibling brood. We found no desiccated larvae, however, in or around the experimental colonies. Instead, we hypothesize that polygyne fire ant workers preferentially cannibalize less related brood in times of stress. High levels of cannibalism are known from this species (Sorensen, Busch, & Vinson, 1983; Tschinkel, 1993) and often occur when resources are in short supply (e.g., a lack of proteinaceous food). Any hereditary predisposition towards nepotism should be under positive selection in a eusocial system, as preferential care of related offspring should increase inclusive fitness (Keller, 1997; Wilson, 2008; Wilson & Hölldobler, 2005). However, extremes in nepotism should be selected against in polygyne systems (Keller, 1997), as they tend to create an environment of competition within the colony and diminish colony fitness. While strong nepotism should be selected against, slightly nepotistic ants may have an advantage over their less discriminating counterparts as nepotism toward sexual larvae, regardless of the strength of the interaction, should increase inclusive fitness (Nonacs, 1988). This would suggest that eusociality should favor low levels of nepotism, such as those found in this study, which straddles the gap between selfless and selfish endeavors.
The lack of sharing between polygyne colonies in the field and evidence of within-colony nepotism in the laboratory corresponds with high relatedness coefficients observed in the field. Although relatedness between workers was lower within polygyne mounds than within monogyne mounds, relatedness coefficients in polygyne mounds were much higher (mean and standard errors: 0.269 ± 0.037) than those previously observed in other introduced populations in the USA (Fig 5; DeHeer & Ross, 1997; Goodisman et al., 2007; Ross, 1993; Ross & Fletcher, 1985; Ross et al., 1996). Past studies have reported values that were not significantly different from zero (i.e., many unrelated queens producing workers within the same mound; but see Ross, 1993), but our results suggest that workers within polygyne mounds in Texas may even be half-sisters (expected r for half-sisters = 0.25). One explanation is that the polygyne mounds that we surveyed contained fewer queens than those sampled in past studies. Ross (1993) demonstrated that relatedness between workers within polygyne mounds in Georgia was negatively correlated with queen number. Geographic variation in colony genetic structure may also explain the higher within-mound relatedness and pairwise F ST values in polygyne mounds compared with those in other states (DeHeer & Ross, 1997; Goodisman et al., 2007; Ross & Fletcher, 1985; Ross et al., 1996; but see Ross, 1993). Much of the population genetics data of introduced polygyne fire ants in the USA has focused on one or a few geographic regions (DeHeer & Ross, 1997; Ross, 1993; Ross & Fletcher, 1985; Ross & Keller, 1995; Ross et al., 1996). Although fire ant populations in Texas have been shown to vary genetically from other parts of the country (Shoemaker, Deheer, Krieger, & Ross, 2006), only a few studies have examined colony genetic structure in Texas (Chen, Lu, Skow, & Vinson, 2003; Ross, Vargo, Keller, & Trager, 1993; Ross et al., 1996), and none that we know of have reported within-colony relatedness between workers (Fig 5).
It is also possible that colony genetic structure has changed over time. For example, relatedness was almost twice as high in older compared with younger populations (i.e., over 100 years old vs. 17 years old) in the polygyne ant Formica fusca (Hannonen, Helanterä, & Sundström, 2004). Past studies of polygyne fire ant queens in Texas reported a near zero relatedness between co-occurring queens (Chen et al., 2003; Ross et al., 1996), which should result in similarly low relatedness between workers, but it is possible that within-colony relatedness has increased over the past 20 years. The ecological impact of polygyne fire ants weakened significantly over a 10-year period in parts of Texas (Morrison, 2002), which may have corresponded with a change in genetic structure. Interestingly, our relatedness coefficients between workers were much more similar to those reported in native polygyne fire ant populations (Fig 5; Ross et al., 1996). In these native populations, polygyne colonies are multicolonial; nestmate queens are highly related (Ross et al., 1996), workers recognize nestmate from non-nestmate (Chirino, Gilbert, & Folgarait, 2012), and colony densities are 4-7 times lower than those observed throughout North America (Porter, Williams, Patterson, & Fowler, 1997). Although we did not measure relatedness between nestmate queens, our behavioral results in the field and in the laboratory support the conclusion that polygyne fire ants in Texas likely function similarly to native conspecifics, in that colonies are likely multicolonial and engage in high levels of intraspecific competition. Our within-mound relatedness coefficients between polygyne workers were also similar to those reported in Australia (Fig 5; Henshaw, Kunzmann, Vanderwoude, Sanetra, & Crozier, 2005), so it would be interesting to determine if Australian polygyne fire ant colonies behave similarly to those in Texas and in the native range.
This does not explain, however, why some studies have detected colony boundaries despite very low relatedness between polygyne fire ant mounds (e.g., Goodisman et al., 2007). Our results suggest that relatedness alone does not predict sharing between mounds, as several neighboring mounds had low pairwise F ST and relatedness values but did not share with each other (Appendix S6). In other ant species, kinship does not always correlate with cooperation between nests (Procter et al., 2016). For example, nests of the polygyne antFormica lugubris did not share workers or resources with each other despite high genetic relatedness (Procter et al., 2016). Similarly, Argentine ants (L. humile ) did not freely exchange workers between all nests within a single supercolony, even though there were no detectable genetic differences between nests (Heller et al., 2008). Likewise, gene flow was limited and some workers were unexpectedly aggressive towards each other within the same supercolony in the unicolonial ant Formica pressilabris , suggesting that supercolonies do not always function as a single unit (Hakala, Ittonen, Seppä, & Helanterä, 2020). This highlights the importance of quantifying colony boundaries using several different methods (Ellis et al., 2017), as genetic relationships do not always imply a free exchange of workers or resources. Low relatedness among nestmate workers may also result from extreme polygyny, where workers originate from numerous unrelated queens (Keller 1995). In this case, each polygyne colony can contain as much genetic diversity as the background population, with nestmate workers being as related to each other than to any random worker within this population, leading to a zero relatedness within the colony (Queller and Goodnight 1989). Future research should examine the exchange of resources between polygyne fire ant mounds in other parts of their invaded range where within-mound relatedness has been reported to be much lower to ultimately determine the relationship between sharing and genetic relatedness.
Our study tests fundamental assumptions about the role of inter- and intracolonial cooperation in polygyne fire ant abundance. Polygyne fire ants are highly abundant throughout their invaded range (Porter et al., 1991), but our results suggest that their high abundance is not due to higher levels of cooperation between neighboring colonies, at least in parts of Texas. Moreover, workers appear to preferentially direct care towards more related brood, indicating some level of nepotism within multiple-queen colonies. Our results add to a growing body of evidence that unicoloniality does not always explain invasive ant abundance (e.g., Eyer et al., 2020; Garnas, Drummond, & Groden, 2014; Zhang et al., 2019). Other factors such as dietary flexibility, competition with other species, and abiotic features may be more important to invasions than intraspecific cooperation in some species. Future research should further examine the relationship between unicoloniality and ant invasions by directly quantifying sharing between nests in the field. Identifying the factors that promote invasive ant success is critical to understanding and managing their ecological impacts.