Intraspecific genetic admixture (henceforth ‘admixture’) occurs when separated populations starts interbreeding [1]. It occurs naturally in many terrestrial and aquatic species in the wild as a consequence of dispersal [2], and can also occur as a result of anthropogenic activities, for example management actions to support populations (e.g. supplementations and translocations) [3, 4], removal of dispersal barriers [5], and escapes of farmed individuals [6].
Admixture affects the genetic composition of the involved populations and tends to increase the genetic variation in the receiving population. Increased genetic variation is generally considered to positively influence aspects of population performance. For example, it has been shown that genetically and phenotypically more diverse populations are better able to cope with environmental change, and to colonize novel environments [2, 7–10]. Admixture can also have positive fitness effects by allowing creation of novel gene combinations and dampening inbreeding depression by masking detrimental effects of deleterious recessive alleles (heterosis) [1, 3]. However, the influx of new genetic material can also result in negative fitness effects. If the involved populations are highly differentiated, genomic incompatibilities might exist (e.g. as a result of chromosomal rearrangements in the parental populations) [11, 12], and admixture may thus lead to outbreeding depression. In addition, admixture between populations that have adapted to different environmental conditions might dilute favorable alleles [1], give rise to offspring with intermediate phenotypes that are not optimal in either of the parental environments [4, 13], and has potential to impair fitness in subsequent generations by breaking up co-adapted gene complexes or by underdominance (i.e. heterozygote disadvantage) [1, 4, 13–15]. The net outcome (with regards to both magnitude and direction) of the response to admixture will be determined by an interplay between these mechanisms, and will thus depend on the genetic composition of the parental populations, local adaptations and environmental conditions [13].
The motivation for management and conservation efforts resulting in admixture is to increase productivity, viability, and adaptability of populations [3, 11, 14]. To avoid undesirable outcomes of such efforts, it would be valuable to have a reliable proxy that could be used to predict the response to admixture. Genetic similarity has been put forward as such a candidate proxy [11, 14]. However, empirical studies show conflicting results [3, 11, 16–19], and further studies are therefore required to increase the understanding about potential associations between genetic similarity and the response to admixture, and to understand the reasons for the observed inconsistencies. Previous studies have commonly estimated genetic similarity based on a modest number (4–32) of microsatellite markers [20–23]. Microsatellites are mainly neutral markers that only occasionally reflect functional evolution, e.g. by residing within coding or regulatory regions or by being linked to functional loci [24, 25]. Microsatellites have high mutation rates, and loci with relatively high allelic variation are commonly selected during the marker development [26, 27]. As a consequence, genetic diversity estimates based on small numbers of microsatellites generally do not reflect the genome-wide diversity [26, 28]. It is therefore possible that the inconsistencies in associations between genetic similarity and admixture effects in previous studies [3, 11, 14, 16, 20, 29] can be partly attributed to the use of non-representative estimates of genetic diversity, and estimates of genome-wide diversity might be better able to predict the outcome of admixture. In addition, estimates of parental genetic similarity might be more informative than population-based similarity, as the former accounts for the inter-individual variation [23].
The development of next generation sequencing techniques, such as Restriction-site associated DNA sequencing (RADseq) has enabled to genotype many markers at a low cost, even for non-model species [30]. RADseq commonly yields thousands of single nucleotide polymorphisms (SNPs), which provides better estimates of genome-wide diversity. It has also been shown that even relatively low numbers of SNPs (≥ 50) generally has the same, if not more, statistical power compared to 20 microsatellites in relatedness studies [31], and that RADseq SNPs generally tend to outperform microsatellites in population genetic studies [32], but see [33]. In addition, it is functional – not neutral - genetic variation that is key for the evolvability of populations [34] with potential to influence the outcome of admixture. It can therefore be hypothesized that estimates of genetic similarity based on adaptive genetic variation (or divergence) offer a better predictor of the response to admixture [35]. RADseq SNPs thus offer a viable alternative as they can provide information on both neutral and adaptive genetic variation and differentiation.
The overall aim of the present study was to evaluate the potential of neutral (microsatellites), genome-wide neutral (RADseq SNPs), and adaptive (outlier SNPs under selection) parental genetic similarity to predict admixture effects on offspring performance (hatching success) resulting from experimental matings of individuals representing genetically separated and locally adapted natural populations. To this end, we used previously published genetic and phenotypic data for three populations (Harfjärden, Lerviksbäcken, and Oknebäck; henceforth Harfjärden, Lervik and Okne) of anadromous Baltic Sea pike (Esox lucius) [32, 36]. Pike is an ecologically and socio-economically important species that has suffered decreases during the last decades, and that has been the target of extensive management efforts [37–40]. More specifically, we investigated whether: i) variation in parental similarity estimates among pairs of experimentally mated males and females were consistent or differed among the markers; ii) parental similarity was associated with offspring hatching success; and iii) the associations between parental similarity and offspring performance were consistent or depended on whether parental similarity was estimated based on neutral (microsatellites), genome-wide neutral (RADseq SNPs) or adaptive (RADseq SNPs under selection) genetic markers.