A puzzling question in conservation biology is how to maintain overall fitness of individuals  bred in captive environment upon release into the wild, especially for rehabilitating declining or threatened species [1,2]. For salmonid species, a heritable change in fitness related traits and gene expression has been reported to occur in a single generation of captivity in hatchery environment [3–5]. Such rapid changes are congruent with models of  inadvertent domestication selection which may lead to maladaptation in the natural environment [4]. Arguably, the underlying mechanism by which captivity may induce  fitness difference between wild and captive congeners is still poorly understood. Short- term selection on complex phenotypic traits is expected to induce subtle changes in allele  frequency over multiple loci [7–9]. Yet, most studies investigating the molecular basis for  rapid change in fitness related traits occurring in hatchery have concentrated their effort on 34 finding evidence for selection at the genome level by identifying loci with large effect.  Numerous wild stocks of Pacific anadromous salmon and trout (genus Oncorhynchus  and Salmo) have experienced fluctuating abundance over the past century, with a series of  sharp declines [6–8]. With the objectives of preserving ecosystem integrity, enhancing. declining populations and sustaining fisheries, conservation hatcheries have been  flourishing. This is particularly true along the North American Pacific coast where billions of  salmonids, all species included, are released each year. Despite substantial improvement of  production management, the beneficial ecological role of hatcheries in enhancing and  restoring wild stocks is still debated, mainly because of the reduced fitness and  maladaptation of hatchery-fish when released in the wild [3,5,9]. Although previous studies  showed that domestication selection was involved in such fitness impairment, they also  observed that different environmental conditions (e.g. reduced fish density) significantly  modulated the physiological acclimation to hatchery environment [4].  

Environmental stimuli are especially relevant during early embryonic development,  which also correspond to a sensitive methylation reprogramming window in vertebrates  [10,11]. It is therefore plausible that differences in rearing environment during early  development may result in epigenetic modifications that could in turn impact on fitness.  However, the only epigenetic study to date pertaining to captive rearing in salmonids and  performed using methylation-sensitive amplified fragments (MSAP) failed to identify  significant changes in methylation profile associated with hatchery rearing [12]  

Here, we used a higher resolution approach to compare the genome-wide pattern of  methylation in hatchery-reared juvenile (smolt) Coho Salmon with that of their wild  counterparts in two geographically distant rivers in British Columbia, Canada. Using a  reduced representation bisulfite sequencing (RRBS) approach covering an average per  individual of about 70 million cytosines in CpG context, we identified 100 methylated  regions (DMRs) that differed in parallel between hatchery and natural origin salmon in both  rivers. The total variance of epigenetic variation among individuals explained by river or  origin and rearing environment in a RDA model was 16% (adj.R2=0.16), and both variables  equally explained about 8% of the variance after controlling for each other. The gene  ontology analysis revealed that regions with different methylation levels between hatchery  and natural origin salmon showed enrichment for ion homeostasis, synaptic and  neuromuscular regulation, immune and stress response, and control of locomotion  functions. We further identified 15,044 SNPs that allowed detection of significant  differences between either rivers or sexes. However, no effect of rearing environment was  observed, confirming that hatchery and natural origin fish of a given river belong to the  same panmictic population, as expected based on the hatchery programs applied in these  rivers (see Supplementary experimental procedures). Moreover, neither a standard  genome-scan approach nor a polygenic statistical framework allowed detection of selective  effects within a single generation between hatchery and natural origin salmon. Therefore,  this is the first study to demonstrate that parallel epigenetic modifications induced by  hatchery rearing during early development may represent a potential explanatory  mechanism for rapid change in fitness-related traits previously reported in salmonids.