Discussion

Knowledge of evolutionary and demographic processes is crucial for our understanding of how native species will respond to biological invasion and the mechanisms that facilitate or inhibit their co-existence with invasive species. We investigated the interaction between adaptation and demography to gain insight into the persistence of a native gastropod (Amnicola limosus ) following approximately 12 years of exposure to an invasive predator, the round goby, in the Upper St. Lawrence River. Our genomic results indicate that A. limosus has locally adapted to the invasion in the span of \(\leq\)12 generations. We also find evidence for adaptation to differences in water calcium over the longer geological history of the ecosystem. Despite evidence of local adaptation in invaded populations, they are experiencing demographic decline, whereas refuge populations show relative stability. While these results could imply that the low-calcium refuge populations have the potential to provide migrants and generate demographic and genetic rescue of invaded populations, this hypothesis was not supported. We detected restricted gene flow and strong population structure between the physiological refuge and invaded populations. Moreover, we found evidence that individuals in uninvaded refuges appear to be maladapted for life history traits, showing low fitness overall. Therefore, despite the current persistence of native A. limosus gastropods in the Upper St. Lawrence River system following the invasion by round gobies, this native gastropod could become vulnerable due to reductions in effective population sizes and limited potential for genetic rescue of impacted populations by the populations in physiological refugia.

Genomic signatures of local adaptation to round goby invasion and low water calcium.

Our genomic data (population structure, Fig. 5 and environmental association analyses, Fig. 2) provide general evidence for local adaptation to the two distinct environment types in A. limosus , i.e., low calcium/goby absent and high calcium/goby present. The exceptions were for two populations (RAF-LCGP and PDC-HCGA) that experienced inverse conditions for selection than the other sampled populations, both clustering with the invaded populations. For RAF-LCGP, the results support strong selection from goby predation even under lower calcium conditions, which are less optimal environmental conditions for round goby feeding and performance (Iacarella & Ricciardi, 2015). For PDC-HCGA, the results suggest strong migration from adjacent invaded sites, which was confirmed by our demographic analyses (see below). PDC-HCGA itself remained uninvaded despite higher water calcium concentrations, likely because the site was located within a wetland, which provides less optimal conditions for round-goby establishment due to the substrate properties (Astorg et al., 2021).
Our EA analyses with Baypass and poolFreqDiff potentially allowed us to disentangle the signals of the two putative selective pressures (i.e., the effect of selection from goby predation at invaded sites and low calcium levels at uninvaded sites), even though invasion status and calcium concentration were strongly correlated. We found SNPs uniquely associated with invasion status and calcium concentration, which can be interpreted as signatures of local adaptation to predation by the round goby fish, and to the more limiting calcium concentrations at uninvaded sites. However, a more comprehensive understanding of the relative roles of these distinct selection mechanisms on genomic variation will require functional validation and ideally a different sampling design that includes additional sites with less correlation between these two environmental factors. Due to the unavailability of an annotated reference genome for A. limosus or a closely related species, we were unable to investigate putative physiological functions of the SNPs showing significant differentiation between environment types. Adaptation to calcium likely involves different functions from adaptation to predation. Differences in calcium concentrations between the water masses from the two rivers are related to the geological characteristics of the river watersheds and therefore represent environmental differences over the long evolutionary history of this species in the St. Lawrence River. On the other hand, predation from the invasive round goby on mollusks is a recent and novel stressor in the St. Lawrence River. Putative physiological functions that would be worth investigating in future studies include transmembrane calcium transport and biomineralization pathways that might be involved in adaptation to low calcium concentration (Clark et al., 2020), as well as shell development regulatory genes that could play a role in the evolution of smaller-sized shells at maturity, which has been observed in populations subject to goby predation (Kipp et al., 2012; Johnson et al., 2019).

(Mal)adaptive responses to round goby invasion and water calcium levels

Our results from the reciprocal transplant experiment give insight into potential adaptive and maladaptive responses in life history traits between SL (HCGP) and OR (LCGA) populations to divergent calcium concentrations and round goby predation regimes (Fig. 3). Indeed, we found important differences in life-history traits (fecundity and survival as fitness components) between the populations from the two habitat types. Both fecundity and survival were higher in the HCGP populations than in the uninvaded LCGA populations, regardless of water treatment. The potential for local adaptation to the calcium gradient is suggested by a slight home advantage in fecundity for populations from both habitats in their origin water versus transplant water (water treatment LCGA vs HCGP), although the interaction between origin and treatment water was not significant. While our laboratory reciprocal transplant experiment could indicate that A. limosus responded to round goby invasion through shifts in life-history traits, populations from the uninvaded habitats are also possibly maladapted, as shown by low fitness across treatment water and goby cue treatments. This suggests the LCGA populations might be generally maladapted (Brady et al., 2019), which could occur through a trade-off of adaptation to low calcium water, as intracellular transport of calcium is energetically costly (Clark et al., 2020). In addition, individuals from the Ottawa River might allocate more resources toward calcium transport and be less able to invest in life-history traits such as reproduction.
Survival rates between populations varied widely, especially for the HCGP populations. This could reflect potential local (mal)adaptation to other biotic or abiotic parameters that we did not consider in the present study (e.g., temperature, substrate, nutrient availability, and food quality). As it was conducted within a single generation, we acknowledge that our reciprocal transplant experiment did not allow us to differentiate plastic vs genetic vs maternal effects on the measured traits. However, our genomic results support the idea that the life-history differences observed between the two population types could be at least partially explained by adaptive genetic differences between the environments.
  1. Demographic and genetic effects of the invasion.

Given prior findings showing a decline in gastropod population abundance following invasion by round gobies, we hypothesized that invaded populations might suffer from population bottlenecks and reduced genetic diversity. However, we did not find a negative effect of the invasion on genetic diversity (Fig. 4), with high levels of nucleotide diversity found in all populations. The nucleotide diversity reported here is relatively high compared to other species observed across phyla (Leffler et al., 2012), although lower than what has been observed in other gastropod species (Redak et al., 2021; Oswald et al., 2022). The slightly negative Tajima’s D found in all populations indicates that there was an excess of rare alleles compared to the neutral model, which could reflect a recent population expansion or positive selection. However, our demographic analysis revealed the occurrence of genetic bottlenecks or reductions in Ne in invaded populations, and even in one refuge population (Table 1, Table S2). This reduction in Ne was only followed by recovery in one of the cases, hinting at a possible case of genetic rescue (i.e., due to an increase in genetic diversity), although the results must be interpreted with caution as there was considerable uncertainty around the parameter estimates. The source of migrants that could potentially be generating a rescue is also presently unknown and is unlikely to be due to the migration from the low-calcium physiological refugia populations as the gene flow in this direction was non-significant. Further validation of the potential for genetic rescue would require obtaining census data and collecting new genomic samples targeted at identifying the potential source populations and quantifying the level of hybridization in recipient-invaded populations (Fitzpatrick et al., 2020).
Surprisingly, the effective population size reductions did not trigger major declines in genetic diversity levels (Fig. 4). This could be due to low but significant gene flow (0.1 < 4Nem < 11; Hämälä et al., 2018) within habitats, as suggested by the gene flow estimates detected between PDC-HCGA and GOY-HCGP populations and the lower FST within clusters (Table S2, Fig. 5). Similar rates as observed here have been shown to be sufficient to maintain genetic diversity despite low effective population size (Gompert et al., 2021). Invaded populations thus do not appear to be currently in need of genetic rescue from refuge populations to recover genetic diversity, or perhaps genetic rescue has already occurred or is ongoing and is the reason for the high genetic diversity observed at invaded sites.

Interaction between local adaptation and demographic/genetic rescue

A core goal of this study was to determine if local adaptation could interact with demographic and genetic rescue. We found relatively low or non-significant levels of gene flow between the populations from the two habitat types (from LCGA refugia to HCGP populations and inversely), but high gene flow within habitat types (Table S2). Pairwise-FST values were within the range of what has been observed in other egg-laying marine gastropod species . Thus, our analyses of gene flow indicate that refuge populations in low calcium habitats do not provide migrants to invaded populations, and this is further supported by the significant pattern of isolation by environment (Wang & Bradburd, 2014). This pattern could be generated by selection against migrants (Nosil et al., 2008; Orsini et al., 2013; Tigano & Friesen, 2016) related to calcium limitation and round goby predation. Individuals from the LCGA populations had lower fitness overall compared to HCGP populations and thus might have low reproductive output and survival in invaded habitats. Given that goby predation has been shown to cause selection for smaller shell sizes at maturity (Kipp et al., 2012), LCGA individuals might be also more vulnerable to predation than HCGP individuals if they are more conspicuous due to larger shell sizes. Similarly, hybrids might be selected against if intermediate phenotypes have lower fitness in their local environment (Thompson et al., 2022). In addition, because individuals from uninvaded populations have lower reproductive output and survival, they provide a more limited demographic subsidy to invaded populations. This will depend on the magnitude of the relative fitness difference between source and recipient populations (i.e., how detrimental migrant alleles will be in the recipient populations; Bolnick & Nosil, 2007). The recent adaptation to goby predation could have thus reinforced the existing effect of isolation by environment from the adaptation to low calcium, thereby limiting the potential of low calcium physiological refugia to provide migrants necessary for both the demographic and genetic rescue of invaded populations.
In contrast to the low gene flow found between LCGA and HCGP populations, gene flow was relatively high between the wetland refuge PDC-HCGA and the invaded population GOY-HCGP, although it did not result in Ne recovery from the bottleneck in the latter (Table S2). Wetland habitats provide refuge from goby predation by reducing their abundance at a local scale and are known to enhance fish and macroinvertebrate diversity (Astorg et al., 2021; Morissette et al., 2023). Their role as a refuge has also been previously recognized in other invaded systems (Reid et al., 2013). Given the prevalence of wetlands in the Upper St. Laurence River (Morissette et al., 2023), wetland refugia with high calcium concentration thus have the potential to provide migrants to invaded sites, particularly due to the lower adaptive divergence between these populations. This suggests that migration of individuals from larger wetland refuge populations might be providing not only a demographic subsidy (demographic rescue; Hufbauer et al., 2015) but could also be replenishing the genetic diversity if indeed it was lost due to population declines in invaded populations (genetic rescue; Whiteley et al., 2015). The role of wetlands as a refuge and their potential implication in the demographic and genetic rescue of invaded gastropod populations therefore warrants further investigation.
Based on our demographic modeling and population structure results (Table S2, Fig. 5), populations from high calcium habitats are more likely to be exchanging migrants, which appears to be sufficient for preserving genetic diversity. However, this beneficial effect of gene flow might have limitations as shown by the absence of effective population size recovery in two out of three populations for which we detected a bottleneck (Table 1). This is particularly important because strong selection such as that detected in the invaded populations can also lead to reduced population sizes and drift, with negative effects on population fitness (Falk et al., 2012). The net outcome of this conflict between adaptation to the invasive predator and low calcium concentrations, and genetic rescue, will thus depend on the severity of population decline in recipient populations (Hufbauer et al., 2015), the extent of adaptive differentiation between populations in each habitat type, and the rate of immigration from high calcium wetland populations.

Potential limitations of genetic rescue during population management.

This study documents the impact of an aquatic invasive predator on evolutionary and demographic processes in a native prey species. Evaluating the potential evolutionary impacts of invasive species on native species is important because they can lead to surprising, unforeseen negative effects, such as the disruption of existing local adaptation . Although genetic rescue has been proposed as a valuable tool for the conservation of small, isolated populations (Whiteley et al., 2015; Ralls et al., 2018), it has also been recognized that genetic rescue carries the risk of outbreeding depression if there is an adaptive genetic divergence between source and recipient populations (Frankham et al., 2011). The present study highlights a case from natural, unmanaged populations where the potential for genetic rescue from physiological refugia is potentially limited by adaptive divergence, and only appears to be possible in the presence of migration from refuge populations of similar habitat types. This implies that the presence of physiological refugia will not necessarily translate into the demographic or genetic rescue of imperiled populations if strong genetic differentiation exists between refuge and recipient populations, for example stemming from isolation by environment. It thus reiterates the importance of considering the local (mal) adaptation of donor and recipient populations during managed introductions that aim to produce genetic rescue (Hoffmann et al., 2021).