4. Discussion
Theory predicts that both life-history and dispersal traits will evolve during range expansion (Peischl et al., 2015; Peischl & Excoffier, 2015; e.g. Phillips et al., 2010b; Shine et al., 2011), with different evolutionary processes driving distinct patterns of evolution. It is unclear how well theoretical predictions hold up in complex natural systems including biocontrol agents, and which evolutionary processes appear to drive trait changes across range expansions (Szűcs et al., 2019). We studied the recent range expansion of the biocontrol agent D. carinulata in the western US and infer the dominant evolutionary processes acting during the range expansion by comparing populations at the core and edge of the range.
We found selection at low densities to be the dominant process over expansion load driving evolution of reproductive life-history traits at the edge of the range. Females from the edge laid on average one more egg than those from the core on the first day of reproduction. If this difference persists throughout the multi-week lifespan of adults, it could sum to a substantial difference in fitness between core and edge individuals. Although we were not able to measure lifespan or lifetime fecundity in this study, previous studies of D. carinulata indicate that early fecundity is a good predictor of lifetime fecundity (Bitume et al., 2017). We used mass as an additional gauge of fecundity, since insect body size is often related to egg production and how many eggs females can carry (Berger et al., 2012). We found this association between mass and fecundity to hold in D. carinulata and that individuals from the edge were larger than those from the core. Age at first reproduction is an often-overlooked trait that can be just as important for fitness as fecundity itself, since reproducing earlier can increase the total time available for an individual to reproduce and early offspring often have an advantage over later offspring because there are fewer other offspring to compete with (Stearns, 1976). Developing faster and reproducing earlier can also allow more generations per year, which might allow acceleration of the range expansion from edge populations of D. carinulata, which are also less constrained by cold temperatures in the winter (Jamison et al., 2018). Early fecundity, mass of females, and age at first reproduction all indicate that selection has increased reproductive capacity at the edge of the range expansion. While the effect sizes are small for each individual trait, when viewed together, this provides strong evidence for a shift in reproductive life-history traits over about 15 years.
We did not find evidence of increased genetic load in edge populations relative to core populations. Genetic diversity has been retained along one D. carinulata expansion front (Stahlke et al., 2021), which may have contributed to rapid evolution of diapause induction timing (D. W. Bean et al., 2012). D. carinulata tends to aggregate (D. W. Bean, Wang, et al., 2007; Cossé et al., 2005), which may allow it to maintain high enough population sizes on the range edge to reduce the consequences of serial founder events and gene surfing inherent in range expansion. Populations of Diorhabda might also maintain high genetic variation since they were collected from multiple source populations and population sizes were deliberately large to avoid reducing variation that could increase establishment in the field (Stahlke et al., 2021; Szűcs et al., 2017).
Dispersal is an inherently and notoriously variable trait (Bowler & Benton, 2005) and we found this to be true for D. carinulata, even when measuring dispersal in a controlled lab environment. Accounting for the density and mating context of dispersal decisions in our experiments allowed us to observe evolution between core and edge more clearly and form hypotheses about the mechanisms behind the patterns we see. The occurrence of flight was affected by mating status along the range expansion such that unmated beetles from the edge flew more than those from the core. This implies that the response to mating has evolved between core and edge. This could be due to low mate availability in edge environments, so males on the edge are more likely to need to disperse before finding a mate. For the number of flights, dispersal became negatively density-dependent at the edge, such that the number of flights increased in low density environments compared to core, while staying about the same in high density environments. This implies that the response to density has evolved during range expansion. This could be due to selection for increased dispersal at range edges at low density, as predicted by theory (De Bona et al., 2019; Fronhofer, Nitsche, et al., 2017; Travis et al., 2009).
Edge beetles flew further than core beetles across all density and mating treatments, and density and mating status interacted as predicted by condition-dependent dispersal theory. Unlike with occurrence of flight and number of flights, the relationship of distance with mating status and density did not change over the range, but we do find a weak signature of spatial sorting of dispersal ability. In this species, spatial sorting might primarily act on occurrence or frequency of flights rather than flight distance or speed if most dispersal flights driving the range expansion are comprised of multiple frequent flights to catch air currents, instead of long-distance flights. Future studies will be needed to explore the nature of dispersal in this species and how spatial sorting acts on different dispersal elements in natural systems.
The effect of spatial sorting in the range expansion of D. carinulata system could be small because of maladaptation to novel environments on the edge of the range expansion that slow down range expansion and reduce assortative mating between dispersive individuals at the edge (Andrade-Restrepo et al., 2019; Hillaert et al., 2015). Early in its range expansion, D. carinulata was maladapted to photoperiod cues (D. W. Bean et al., 2012; D. W. Bean, Dudley, et al., 2007) and possibly higher summer temperatures in southern latitudes (Herrera et al., 2005). Adaptation to photoperiod has limited the rate of southern range expansion in this beetle and thus may reduce the effect of spatial sorting of dispersal. Despite this, our results suggest that spatial processes during range expansion may be important to natural range expansions even over heterogenous environments.
In many species, dispersal evolves along with suites of traits, called dispersal syndromes (Ronce & Clobert, 2012), and in some cases, many life-history traits may correlate well with dispersal (Stevens et al., 2013). Trade-offs between dispersal and reproductive ability are widely hypothesized to be present due to allocation of finite resources (Bonte & Dahirel, 2017; Stearns, 1989) though support for such trade-offs during range expansion is mixed (e.g. Hughes et al., 2003; Jan et al., 2019; Kelehear & Shine, 2020; Tabassum & Leishman, 2018; Therry et al., 2015). In the D. carinulata range expansions, we do not see evidence of a trade-off between dispersal and life-history traits, though we were unable to measure all traits within the same individuals. We measured dispersal in male D. carinulata, but there are many reasons for dispersal to differ between the sexes (reviewed in Li & Kokko, 2019). There may also be trade-offs between other traits, such as lifespan or immune system development or function (reviewed in Chuang & Peterson, 2016).
Long-term success of the Tamarix-Diorhabda biocontrol program requires D. carinulata to continue its spread to cover the range of the target weed and to adapt to new environments. Our results contribute further evidence of sufficient genetic variation for adaptive evolution to occur. We show an increase in both reproductive output and dispersal ability at the edge and low genetic load, which may enable an accelerating expansion front and will likely contribute to the establishment and persistence of D. carinulata populations at the edge (Phillips et al., 2010a). Evolution of these traits and others previously studied (e.g. D. W. Bean et al., 2012; Stahlke et al., 2021) suggests that there is sufficient genetic variation for populations to continue to adapt to novel environments during the expansion. As the first test of evolutionary theory of range expansions in a modern biocontrol agent, we show that these theoretical predictions can be applied to range expansions across heterogeneous environments, especially when the ecological context of individuals is included. We may expect to find selection at low densities to be the dominant evolutionary process over expansion load and for spatial sorting to act on other biocontrol agents that share many characteristics with D. carinulata (e.g. Bartelt et al., 2008; Muller-Scharer et al., 2014). Our results suggest that evolutionary processes impacting range expansions of natural populations can act simultaneously with adaptation to environmental gradients.