Insights on the concept of indicator populations derived from parentage‐based tagging in a large‐scale coho salmon application in British Columbia, Canada

Abstract For Pacific salmon, the key fisheries management goal in British Columbia (BC) is to maintain and restore healthy and diverse Pacific salmon populations, making conservation of salmon biodiversity the highest priority for resource management decision‐making. Salmon status assessments are often conducted on coded‐wire‐tagged subsets of indicator populations based on assumptions of little differentiation within or among proximal populations. In the current study of southern BC coho salmon (Oncorhynchus kisutch) populations, parentage‐based tagging (PBT) analysis provided novel information on migration and life‐history patterns to test the assumptions of biological homogeneity over limited (generally < 100 km) geographic distances and, potentially, to inform management of fisheries and hatchery broodstocks. Heterogeneity for location and timing of fishery captures, family productivity, and exploitation rate was observed over small geographic scales, within regions that are, or might be expected to be, within the area encompassed by a single‐tagged indicator population. These results provide little support for the suggestion that information gained from tagged indicator populations is representative of marine distribution, productivity, and exploitation patterns of proximal populations.

(Oncorhynchus kisutch) in BC is based on life history, marine distribution, and exploitation information obtained from coded-wire-tagged (CWT) "indicator" populations, from which a portion of individuals carries a physical tag that can be recovered from a fish for identification purposes. Tagged coho salmon indicator assessment programs, and the geographic boundaries for populations that each indicator was intended to represent, were initiated over 40 years ago, predating the development of genomic technologies to assist in charac- an updated consideration of the assessment and management tools now available to characterize, monitor, and conserve biodiversity (Benestan et al., 2016;Hunter, Hoban, Brugord, Segelbacher, & Bernatchez, 2018). Strait-Southern Fjords, Homathko-Klinaklini Rivers) comprised 31% of the sample, but samples delivered to a central processing laboratory for potential CWT recovery from adipose-fin-clipped individuals for the same time period displayed no contribution of these two CUs to the sample (Beacham et al., 2019a).
Recent and ongoing improvements in genetic and genomic analytical methods indicate that a genetic-based approach may provide the most informative, viable, and cost-effective methodology for assessment and management of hatchery origin and wild Pacific salmon (Beacham et al., 2019a. Anderson and Garza (2006) noted that parentage-based tagging (PBT) provided a method of identifying both the age and population of origin for individual salmon. PBT uses molecular-based approaches to conduct largescale parentage assignments and has resulted in the unprecedented ability to identify genetically millions of hatchery-origin salmonids to their hatchery of release and age (Steele, Hess, Narum, & Campbell, 2019). Steele et al. (2013), Hess et al. (2016) Beacham et al. (2019a) provided evidence that PBT-GSI-based assessment and management of hatchery origin and wild coho salmon were a practical approach, as demonstrated by a large-scale application to fisheries management and assessment in BC. Population-and family-specific distributions among fisheries, origins, and productivity of hatchery broodstocks and associated stray rates among populations were also evaluated via PBT (Beacham et al., 2019b). As well as providing improved estimates of stock composition from mixedstock fisheries, PBT allows evaluation of family-specific productivity and exploitation, evaluation of productivity in specific components (differentiated by life-history type, spawn times, or rearing environment) of a hatchery broodstock, assessment of hatchery-wild interactions in salmonids (Araki, Berejikian, Ford, & Blouin, 2008, McClure et al., 2008, Jones & Wang, 2010, and the genetic basis for salmon migration and marine residency. Direct DNA sequencing is powering a revolutionary application of genetics to fisheries management and assessment, providing cost-effective genotyping at hundreds of single nucleotide polymorphism (SNP) loci (Campbell, Harmon, & Narum, 2015) or tens of microsatellites (Bradbury et al., 2018).
Harnessing the evolving power of genetic and genomic technologies will provide ongoing insight into the adaptive variation at the heart of biodiversity, and the impacts on wild populations resulting from fishery and hatchery broodstock activities as well as climate change and changing ocean regimes (Bernatchez et al., 2017).
In the current study, PBT methodology based on variability at 304 single nucleotide polymorphisms (SNPs) was applied to coho salmon sampled from fisheries and escapements (including both hatchery broodstock and nonbroodstock) in BC. Commercial and recreational coho salmon fisheries were sampled in 2017 and 2018, and hatchery broodstock and nonbroodstock river escapements Information on geographic variability among populations in fishery distribution and timing of catch, and family productivity and exploitation rates was generated. Our analysis of within and among population variability did not support the premise that indicator populations display abundance, distribution, and timing characteristics that are representative of other populations within management units based on geographic proximity.

| Fishery sample collection
Six fishery areas were defined for coho salmon sampled from fisher-  (Figure 1). Samples from commercial, F I G U R E 1 Map indicating geographic locations for fishery sampling and 26 populations for which parentage-based tagging was applied to identify individuals in both fishery and escapement sampling recreational, and First Nations fisheries within a fishery area were pooled, and samples from Barkley Sound and Alberni Inlet were pooled with WCVI samples. Further details on fishery sampling were outlined by Beacham et al. (2019a). There were 1,499 PBT identifications made in 2017 fishery samples (Beacham et al., 2019a) and 1,792 PBT identifications made in 2018 fishery samples, and these 3,291 PBT identifications comprised the fishery pool used to identify individuals to specific parents in either the 2014 for 2015 hatchery broodstocks.

| Escapement and broodstock sample collection
Twenty hatchery broodstocks were sampled in 2014, and 6,061 individuals were genotyped in these broodstocks .  Table 1.

| Genotyping
The detailed procedure for library preparation and genotyping was outlined by Beacham et al. (2017) and a summarized version provided by Beacham et al. (2019a). Summarized briefly, 768 individuals with up to 304 amplicons per individual were loaded on a P1 chip v3 with an Ion Chef, two chips were loaded consecutively with one run of the Ion Chef, both chips were then subsequently loaded on to an Ion Torrent Proton sequencer, and the genotype of each individual recorded with automated scoring of the genotype via Proton software Variant Caller® at one SNP site in each amplicon. Genotypes at all available SNPs for an individual were assembled to provide a multilocus individual genotype. This multilocus genotype was the basic input for PBT analysis.

| Identification of individuals
PBT was used to identify individuals in fishery and escapement samples by matching the genotype of the individual to the genotypes of prospective parents (COLONY, Jones & Wang, 2010;Wang, 2016).
The COLONY parentage assignment software was utilized as it can produce assignments when the genotype of one of the parents is missing, either due to a missing parental sample, or due to failure to produce a parental genotype from an existing sample. Given that PBT assignments for many potential populations were evaluated for each fishery and escapement sample, COLONY was run with all broodstock sampled during each year as input as a single unit for analysis of fishery and escapement samples, with no differentiation among populations. Although the COLONY assumption of a single population in the parent pool was violated, analysis of knownorigin samples indicated that very high levels of accuracy (100%) were achieved in population assignments when pooling of potential parents in contributing populations was conducted (Beacham et al., 2019a). Two-parent assignments were accepted only when both assigned parents originated from the same population. Twoparent and single-parent assignments were accepted only when the probability of correct assignment was ≥ 0.85 for the parent pair. An additional constraint on the single-parent assignment before it was accepted was that both the PBT assignment and GSI assignment corresponded to populations in the same CU. Polygamous mating was assumed for the COLONY analysis. Simple pairwise comparisons between offspring and potential parents were conducted. The PBT baseline for individuals sampled in the 2016 hatchery broodstock and nonbroodstock escapements (jacks) and 2017 fisheries and escapements and hatchery broodstocks included all broodstocks sampled in 2014 and 2015, as coho salmon in southern BC are predominately three years of age (Sandercock, 1991). Similarly, the PBT baseline for individuals sampled in 2018 fisheries and hatchery broodstocks included all broodstocks sampled in 2014, 2015, and 2016. Jacks selected in the 2017 escapement were identified via body size, as there is a fairly clear distinction between jack and adult body size within populations (Sandercock, 1991), and subsequently assigned to parents in the 2015 hatchery broodstocks. Individuals with more than 120 missing genotypes were eliminated from further analyses. An estimated genotyping error rate of 1% was used for COLONY assignments. Previously, Beacham et al. (2017) had reported that an average genotyping error rate of 1.07% (1,220 discrepancies in 114,105 comparisons) or an allele error rate of 0.53% (1,220 discrepancies in 228,210 comparisons) was observed over the 304 SNPs scored. The parent pair output file was the basic file used in subsequent analyses. Sound and Alberni Inlet sport, freshwater sport (Fresh), and unknown location (Unk). Total is the total number of fishery PBT identifications for the population.

| Geographical concordance
Four traits were compared among populations within regions.

| Fishery captures
The key assumption underlying an assessment method employ- In the Lower Fraser River CU, there were significant differences in marine fishery identifications among the six populations in the analysis ( 2 30 = 77.9, p < .001). Fishery identifications of the Chehalis River population were proportionately more prevalent in the JDF fishery, whereas those of the Stave River population were more prevalent in the JS and northern GS fisheries (Figure 2). The Inch Creek population is the only coded-wire-tagged population in the CU, and its distribution of fishery identifications was significantly different from those of the Chilliwack River ( 2 6 = 22.2, p < .001) and Kanaka Creek populations ( 2 6 = 18.3, p < .01), and approached significance for the Chehalis River ( 2 5 = 10.9, p < .06), Stave River ( 2 5 = 10.2, p < .06), and Norrish Creek ( 2 6 = 11.7, p < .07) populations. In the Howe Sound-Burrard Inlet CU, located on the southern BC mainland, there were differences in fishery recoveries of progeny from the four populations in the CU ( 2 18 = 82.8, p < .001), with only the Seymour River population coded-wire tagged in CU. The Tenderfoot Creek and Mamquam River populations contributed proportionately more to the CC and JS fisheries, whereas the Capilano River and Seymour River populations contributed proportionately more to the GS fishery ( Figure 2).
In the Boundary Bay CU, there was no significant difference in the distribution of fishery identifications between the two populations in the CU (p > .10) (Figure 2). In summary, there was little evidence to support the assumption that a single population chosen as an indicator in a CU is likely to display a distribution of captures among fishery areas that is representative of other populations in the CU.

| Timing of captures
The month of fishery capture was also evaluated for conform- In the Boundary Bay CU, there was no significant differentiation between populations in the CU with respect to monthly fishery progeny captures (p > .10). September was the dominant month of fishery captures for both populations (Figure 3). In Boundary Bay, there were restricted number of progeny observed for the two populations, and there was no significant difference in the distribution of total progeny per male between the two populations ( 2 7 = 11.7, p > .10) (Figure 4). Similarly, with a restricted total number of progeny identified per male for the two south Thompson River populations (Salmon River, Eagle River), there was no significant difference in the distribution of total progeny per male between the two populations ( 2 5 = 5.3, p > .10) (Figure 4)    that the marine migration pattern and associated ocean distribution of populations, and therefore fishery recoveries, are similar across populations within a CU. However, the data derived from the study were fishery-dependent, subject to any biases occurring due to location and timing of fishery sampling and possibly unrepresentative sampling. These results are in contrast to those reported previously by Weitkamp and Neely (2002) for coho salmon, who suggested that recovery patterns for tagged wild populations were consistent with those of hatchery populations from the same region and that marine distributions based on hatchery populations are reasonable proxies for marine distributions of wild populations from the same local area.

| Indicator populations
Part of the difference between the studies may relate to the smaller fishery recovery areas in southern BC employed in the current study in comparison with those of Weitkamp and Neely (2002), as well as a more extensive regional coverage for southern BC populations in the current study.
Estimation of age-specific exploitation rate is one of the more important outputs of the CWT assessment method, but we found significant differences in family-specific exploitation rates among populations within small geographic regions, putting into question the assumption that the exploitation rate of indicator populations will be representative of other populations in the management unit. Age-specific exploitation rates are of more management concern in Chinook salmon than in coho salmon. Satterthwaite Like the Puntledge River population, most of the Nitinat River production was released as fry, leading to lower observed productivity.
In the east coast of Vancouver Island-Georgia Strait CU, the  and epigenetic impact of hatchery fish on wild populations is becoming of concern with respect to wild population conservation. PBT provides the potential for realizing numerous goals in Canada's WSP (Fisheries and Oceans Canada (FOC) 2018) by enabling assessment on a CU and management unit basis using new assessment methods. Beacham et al. (2019a)  Quebec. Funding support was also previously provided by the Northern Endowment Fund of the Pacific Salmon Commission for baseline development.

CO N FLI C T O F I NTE R E S T
None declared.