4 | DISCUSSION
Our study provides insights into the invasion pathway of the termiteReticulitermes flavipes , highlighting numerous and recent human-mediated jump dispersals in both the native and introduced range of this species. We first revealed a strong clustering among individuals within the native range of this species in the eastern USA. Yet, these individual differences were not due to geography, as highly different individuals were found in the same locality and highly similar ones in localities separated by several thousand kilometers. This finding indicates extensive movement of colonies throughout the native range, likely through human transportation. We also highlight a complex invasion history with multiple events out of the native range and bridgehead introductions from the introduced population in France. The apparent genetic shuffling within the native range limits our ability to assign an exact source population(s) for the different introduced ranges. However, similar to the effect of multiple introductions into the invasive range, admixture in the native range prior to invasion can potentially favor invasion success by increasing the genetic diversity later conveyed to the introduced ranges.
Our findings revealed the occurrence of multiple introductions from different native localities serving as sources for the invasive ranges of France, Chile and Canada. Additionally, Canada and Chile received secondary invasions from the introduced population in France, which acted as a bridgehead. Some previous results indicated there may have been several introductions into France (Perdereau et al. 2019). Reticulitermes flavipes was first reported in Austria in 1837; however, it probably occurred earlier in Western Europe, when it was identified as R. santonensis (Kollar 1837). Despite being unable to definitively link its source population(s) to the New Orleans region as previously suggested (Perdereau et al. 2013, 2015), our data, based on a larger sample size and more informative markers, do not rule out this possibility, suggesting that this invasive population originated from somewhere in the southeastern USA, from Louisiana to South Carolina. However, it is possible that the French population originated from colonies originally coming from the New Orleans region that had been transported elsewhere within the native range, such as South Carolina. Such long-distance jump dispersal within the native range necessarily hampers clear identification of the source population(s). Likewise, although our results suggest that the Canadian and Chilean introduced populations originated from admixture between the introduced population of France and native localities in the northern range of R. flavipes, these results suffer from low confidence, potentially due to genetic mixing between native localities. Overall, these findings indicate that jump dispersal may not be restricted to a single region within the native range of this species. Instead, such dispersal appears common with R. flavipes , suggesting that the evolutionary mechanisms promoting this phenomenon are globally distributed across the species distribution range.
The genetic patterns observed in R. flavipes may be explained by numerous and recent jump dispersal events across the native range, likely mediated via human trade and transportation. This finding exemplifies species spread by stratified dispersal , whereby individuals disperse at different spatial scales, from local to long-distance movement (Shigesada et al. 1995). Local scale dispersal relies on the biological dispersal ability of the species, ranging from limited (i.e., budding) to moderate dispersal (i.e., nuptial flight). In contrast, long-distance dispersal is often human-mediated and therefore considered stochastic and difficult to determine. Notably, our study revealed both 1) genetically distinct individuals inhabiting the same locality and 2) genetically similar individuals separated by several thousand kilometers. The geographic distance separating highly similar individuals far exceeds the biological dispersal ability of this species, suggesting these individuals were artificially transported to a different locality. Additionally, the finding of genetically distinct individuals within the same or adjacent localities indicates a low level of mixing between those individuals. This may stem from reduced local dispersal, by which transported individuals interbreed and do not disperse far from their landing point. A high proportion of new reproductives of R. flavipes do in fact interbreed with their nestmates during mating flights (25%); however, the proportion of inbred successful founders is significantly reduced among established colonies (DeHeer and Vargo 2006). Therefore, this inbreeding depression may select against the interbreeding of artificially transported colonies. Also, R. flavipes usually disperse through nuptial flights, which should enhance gene flow over large scales (Vargo 2003). Consequently, a scenario by which transported individuals interbreed and do not disperse far from their landing point may not alone explain the pattern observed in this study. The finding of highly genetically different individuals within the same locality suggests that some of the long-distance jump dispersal events are probably too recent to allow transported individuals to admix with local colonies to homogenize the gene pool within populations.
The global spread of invasive species is strongly influenced by long-distance jump dispersal, even once established within an introduced range (Suarez et al. 2001). These long-distance jump dispersal events are more effective, and often required, for rapidly reaching widespread distributions. The worldwide distribution of the Argentine ant has been shown to primarily stem from human-mediated jump dispersal, rather than from its classical spread through colony budding, as the latter would have to be three orders of magnitude higher to explain its actual distribution (Suarez et al. 2001). This finding is also exemplified in the global distribution of the red imported fire antSolenopsis invicta , which utilized long-range jump dispersal to first invade the southeastern US, and subsequently Asia and Australia from its bridgehead in the USA (Ascunce et al. 2011). In general, eusocial organisms like ants (Bertelsmeier et al. 2018) and termites (Buczkowski and Bertelsmeier 2017, Blumenfeld and Vargo 2020) appear adept at utilizing human-mediated jump dispersal to broaden their global distributions. These serial long-distance movements are also observed among regions within invasive ranges, across a wide variety of taxa, such as the aforementioned S. invicta throughout the southern USA (Lofgren 1986) and China (Ascunce et al. 2011), the western mosquitofish Gambusia affinis in New Zealand (Purcell and Stockwell 2015), and plants in China (Horvitz et al. 2017). Although most studies demonstrate the importance of human-mediated dispersal in shaping invasion dynamics following establishment, it often remains unclear whether long-distance jump dispersal pre-exists in the native range of invasive species, and its relative importance in the pattern of genetic diversity observed at the global scale of these species.
Native ranges of many invasive species often remain geographically structured (Voisin et al. 2005, Beck et al. 2008, Leinonen et al. 2008, Verhoeven et al. 2011). For example, native populations of S. invicta are strongly geographically differentiated (Ross et al. 2007). Though rare long-distance dispersals have been reported (Ahrens et al. 2005), these events occurred far in the past and have been attributed to strong winds during nuptial flights or the rafting of entire colonies during flooding events (Hölldobler and Wilson 1990), rather than from human-mediated transport (Ahrens et al. 2005). Native populations of another termite invaderCoptotermes formosanus in China are highly structured, with distinct native populations representing different genetic clusters (Blumenfeld et al. 2020). This structuring suggests reduced gene flow across populations, and therefore a limited number of human-mediated dispersal events within the native range of this species. Our results stand in sharp contrast with the strong population structure commonly uncovered within the native ranges of invasive species, as frequent jump dispersal appears to have occurred in the native range of R. flavipes . Understanding the factors driving the differences between these two invasive termite species may shed light on key evolutionary mechanisms underlying their invasion success. Furthermore, while most studies focus on unraveling invasion pathways out of a native range, our results stress the need to consider evolutionary processes and human-mediated dispersal that may already be present within the native range of an invasive species, as these can affect the level and distribution of genetic diversity in both the native and invasive ranges.
Extensive human-mediated jump dispersal has been reported in the native range of a few species. For example, in the invasive tree Acacia pycnantha , extensive transport and replanting throughout its native Australian range prior to its introduction to South Africa resulted in highly admixed genotypes already present in the native range. This feature has consequently prevented an accurate identification of the native source population(s), as highly admixed genotypes and comparable genetic diversity were present in both ranges of the species (Le Roux et al. 2013). A similar pattern has been found in the North American rangeland weed, Centaurea diffusa , where an extremely low level of population structure in the native range hindered the assignment of its introduced population to its likely native source location (Marrs et al. 2008). However, the genetic patterns observed in these studies are slightly different than the one observed in R. flavipes , as the inability to pinpoint the origins of invasive populations stems from the near-panmixia found across the native range. Therefore, the patterns in these other species most likely stem from an ancient and continuous genetic shuffling throughout the native range. In contrast, the lack of geographic structure despite highly genetically different individuals indicates recent and stochastic long-distance dispersal in R. flavipes .
The invasion success of termites is tightly linked with their ability to eat wood, nest in wood and cultivated plants and readily generate secondary reproductives, as all 28 species of invasive termites share these three traits (Evans et al. 2013). These traits are common in lower termites like R. flavipes andC. formosanus , and ensure that any piece of wood serving as a nest or foraging site has the potential to become a viable propagule (Evans et al. 2010, Evans et al. 2013). Although these traits may enhance the frequency of human-mediated dispersal in R. flavipes , their occurrence in both species cannot explain the difference in long-distance dispersal in the native ranges of these two species. In R. flavipes , repeated human-mediated dispersal could reflect a higher degree of propagule pressure from different USA regions, representing multiple hubs of intense human activity and timber production. Forests and timber production are unequally distributed across the eastern USA (Brown et al. 1999, Howard and Liang 2019), and may therefore require significant wood transportation throughout this part of the country from high to low timber-producing regions. Similarly, the frequency of human-mediated dispersal may reflect the connectivity between native regions. In the introduced population of R. flavipes in France, the distribution of genetic diversity is associated with the railway network, highlighting its possible role in displacing termites over long distances (Andrieu et al. 2017, Suppo et al. 2018, Perdereau et al. 2019). In the USA, about 14,000km of track were active by 1850, mainly in the eastern USA (141,000km in 1880 and over 400,000km in 1916) (United States Census Bureau 1890, Chandler 1965). In contrast, the first 10km railway was built in China in 1881, but less than 13,000km were in use by 1948 for the whole country. This difference in connectivity may explain the numerous long-distance dispersal events in the native range of R. flavipes and their absence in the Chinese native range ofC. formosanus . Interestingly, the railroad network in the USA has also been suggested to represent a major dispersal mode for C. formosanus in its invasive range in this country (Austin et al. 2008). Overall, many invasive social insect species originate from South America or East Asia (Tsutsui et al. 2000, Heller 2004, Ross et al. 2007, Eyer et al. 2018a, Eyer et al. 2018b, Eyer et al. 2020). The population structure observed in most native populations may simply reflect the reduced connectivity between native regions in these areas, potentially resulting from a lack of internal trade among regions or difficulty in reaching isolated geographic areas. Our findings in R. flavipesmay shed light on frequent long-distance dispersal already present within native ranges of invasive species, especially those originating from regions with a long history of dense transport networks.
While the invasion scenario of numerous introductions from distinct source populations and their admixture in the invasive range may explain the levels of admixture observed in the introduced populations of France and Chile, we cannot rule out the possibility that admixed introduced populations re-invaded the native range of R. flavipes.Similarly, it is possible that populations were already admixed before propagules were transported worldwide. Native populations of many invasive species often remain geographically isolated and locally adapted. It has been suggested that a temporary loss of local adaptation in recent invaders fundamentally alters the fitness consequences of admixture (Verhoeven et al. 2011). Long-distance dispersal in the native range enhances gene flow between distant populations that are otherwise isolated. Similar to post-introduction increases of genetic diversity through multiple introduction events (Kolbe et al. 2004, Stenoien et al. 2005, García et al. 2017), admixture between native populations prior to an introduction event may enhance the amount of genetic diversity brought to the invasive range. Admixture may improve invasion success through recombination between distinct genotypes, potentially creating novel combinations of traits, and/or increasing the level of genetic diversity upon which natural selection can act. Pre- or post-introduction admixture may also reduce the inbreeding load by reducing the expression of recessive deleterious alleles or lead to heterosis effects, potentially improving the establishment and early success of invasive species (Ellstrand and Schierenbeck 2000, Drake 2006, Keller and Taylor 2008, Hahn and Rieseberg 2016). Overall, increased genetic diversity via admixture may favor subsequent introductions given the novel selection pressures invasive species face in their new environments (Verhoeven et al. 2011).