Pacific Lamprey Translocations to the Snake River Boost Abundance of All Life Stages

Translocation Efforts to Avoid Extirpation across a Species’ Range

Increasingly, global population abundance of some species has dwindled to the point that proactive conservation measures, such as translocations, are needed to restore target populations to a healthier level of abundance (Griffith et al. 1989; Seddon et al. 2014). Translocations are human interventions in which individuals from a source of abundance are collected and transported to a recipient site. This intensive intervention is often required to avoid extirpation of a population.

Declines of Pacific Lamprey Entosphenus tridentatus in the Snake River have prompted translocation efforts to recover this culturally and ecologically important species. Pacific Lamprey are anadromous fish that can spend at least 5 years in freshwater as filter-feeding larvae (synonymous with “ammocoete”) before transforming into juveniles by developing eyes, teeth, and osmoregulation ability and migrating to the ocean, where they feed parasitically (Clemens 2019). After some undefined period in the ocean, Pacific Lamprey return to freshwater in summertime but likely do not precisely home to natal streams like salmon (Spice et al. 2012), and they generally overwinter before spawning the following spring (Pletcher 1963).

Although Pacific Lamprey in the Snake River basin are not yet extirpated, the number of spawning-phase adults that volitionally migrate (i.e., “naturally” migrate) past eight dams to arrive this far upstream (confluence with the Columbia River is >500 km upstream from the Pacific Ocean) has decreased to the point that many Snake River basin streams are hypothesized to be no longer occupied (Cochnauer and Claire 2009). Consequently, translocation efforts initiated in 2007 by the Nez Perce Tribe to prevent extirpation are ongoing. Pre-spawning adult lamprey are trapped at three Columbia River dams (Bonneville, The Dalles, and John Day dams), hauled to overwintering facilities, and released into target streams the following spring. Streams targeted for translocations in the Snake River basin were historically occupied by Pacific Lamprey and contained suitable adult spawning and larval rearing habitat but were found to be unoccupied by larval Pacific Lamprey during repeated electrofishing surveys conducted several years prior to translocation (Cochnauer and Claire 2009). The primary objective of translocations is to boost larval abundance and restore a key role in the ecosystem (e.g., conversion of nutrients to stored biomass that becomes a food source for other animals; Close et al. 2002) while passage and habitat improvements for all life stages are completed basinwide (Close et al. 2009; Ward et al. 2012). Secondarily, translocations aim to provide additional benefits to boost abundance of all life stages via the following direct and indirect means: (1) directly through the successful production of offspring that survive to enter the marine environment and return to spawn as adults within the Columbia River basin and (2) indirectly through production of larval pheromonal attractants for adult spawners that come from many different regions (Yun et al. 2011). Ultimately, translocations are intended to help restore Pacific Lamprey ecosystem services and increase abundance to levels needed for sustainable tribal harvest in the interior Columbia River (CRITFC 2011).

Needs and Challenges for Monitoring Highly Dispersive Species

One of the challenges with monitoring dispersive species that have weak philopatry, like Pacific Lamprey (Spice et al. 2012), is the difficulty in attributing long-term gains in abundance across generations to a given conservation action. If lampreys lack homing ability, gains in larval productivity at a site receiving translocated adults cannot be assumed to provide consistent proportional gains in adult returns to that same site. Hence, indirect evidence, such as annual increases in juveniles or adult abundances at a translocation recipient site, cannot be used to quantify the success of translocation strategies.

Additional challenges to monitoring the success of Pacific Lamprey translocation releases include uncertainties pertaining to the protracted larval life stage, including the duration and rate of growth. In the first years following the inception of translocation releases, it was possible to demonstrate increases in larval abundance as the translocations initially targeted streams where Pacific Lamprey were extirpated (Ward et al. 2012). However, sizes of larvae are an imprecise proxy for age (Hess et al. 2021), so efficacy was difficult to quantify in subsequent years. Further, quantifying success for the secondary goal (increasing juvenile and adult abundance) using indirect evidence is logistically intractable with conventional means of tagging due to the diminutive size of larvae and the lack of natal homing. These characteristics of lampreys result in relatively unpredictable fates of fish after they exit their natal stream.

A prerequisite for monitoring the success of translocations is the ability to identify natal origins of large numbers of fish at any life stage. This can be accomplished by genetic methods (parentage and sibship analyses) in Pacific Lamprey and other dispersive species (Hess et al. 2021; Mateus et al. 2021) as genetic methods do not require traditionally intensive sampling and cost-effective methods have been developed to genotype large numbers of informative genetic markers (Hess et al. 2021). Parentage and sibship analyses represent a way to document direct success for both the primary and secondary objectives of the translocation program.

Study Objectives

We utilized genetic monitoring methods to provide a measure of translocation effectiveness by quantifying a direct increase to larval, juvenile, and adult abundances downstream from Snake River basin release sites via successful reproduction of translocated parents. Further, the genetic monitoring provided an accumulation of age, growth, and natal origin data collected over a 10-year period and across a vast geographic area. We exploited this unprecedented data to make a major contribution to the understanding of Pacific Lamprey and lampreys in general.

Sample Collection

The candidate parent baseline consisted of 6,899 translocated adults (henceforth, referred to as the “parent” data set, comprising the translocation fish denoted “TL” in Appendix Table A.1). Genetic sampling started with translocations to the Snake River in 2007 and continued through 2018. In some years, not all of the fish in translocated groups were sampled, and in at least 1 year (collection year 2009), the sampling was zero (Table A.1).

Larvae and juveniles from electrofishing surveys and screw traps throughout the Snake River basin were approximately half (N = 9,199) of the candidate “offspring” data set used to perform parentage assignments (N = 18,957). The offspring data set also included (1) larvae and juveniles collected farther downstream at John Day Dam on the Columbia River (river kilometer [rkm] 347; rkm 0 is the Columbia River mouth), where they were collected during salmonid smolt monitoring (N = 1,615); and (2) ocean-phase individuals from trawl survey collections along the U.S. West Coast, in Southeast Alaska, and in the Bering Sea (N = 1,368). We also collected adult Pacific Lamprey during their spawning migration in the Willamette River (rkm 42) at Willamette Falls (N = 4,436) and from fishways at Bonneville (rkm 235 on the Columbia River), The Dalles (rkm 306 on the Columbia River), and John Day (N = 2,339) dams (Figure 1; Table A.1).

Not all individuals in the offspring data set were expected to assign to translocation parents, so we needed to use sibship analysis to obtain additional information on natal origin that would not have been available based on results exclusively from parentage (i.e., parent assignments could not be made to translocation offspring whose parents were missing from the baseline or to offspring from volitional parent pairs). Therefore, we combined subsets of the offspring and parent data sets into a candidate “sibling” data set and tested for pairwise full-sibship assignments among all individuals in the sibling data set. For each pair of individuals that we identified as a full-sibling pair, we were able to use the location information of the sibling from the upstream-most collection site to estimate the natal origin of the other sibling that was collected further downstream. This sibling data set allowed us to comprehensively represent natal origins of all translocation production as well as volitional production from the Snake River basin that predated the translocation program. The location information for young larvae and juveniles that were included in this data set tended to provide natal origin information for other collections of older life stages of juveniles in the main-stem Columbia River, ocean-phase subadults, and prespawning adults during their freshwater migration.

Molecular Analysis

Tissue samples from Pacific Lamprey were dried on Whatman filter paper, and DNA was extracted from tissue using the same methods described by Hess et al. (2013). The fish were genotyped using a panel of single-nucleotide polymorphism (SNP) loci (Hess et al. 2021) and previously published protocols for genotyping-in-thousands by sequencing (Campbell et al. 2015). The panel includes highly variable (i.e., high minor allele frequency across the rangewide distribution of the species), independent loci (unlinked) that we used for parentage and sibship applications (includes 12 adaptive loci), 4 loci for species determination (Hess et al. 2015), and 28 extra loci for characterizing adaptive variation (defined by outlier tests; Hess et al. 2020). Previously, three loci involved in linkage disequilibrium (Etr_951, Etr_4455, and Etr_5465) had been inadvertently included in the set of 263 SNPs that were used for parentage (Hess et al. 2021), but those loci were excluded here (260 total SNPs). The three excluded loci were Etr_951, which was linked with Etr_3963 on chromosome 29, and Etr_4455 and Etr_5465, which were both linked with Etr_1806 on chromosome 4.

Parentage and Sibship Analysis

We used the genotypic data from 260 SNP loci (for quality control purposes, we excluded individuals missing 10% or more of their 260-SNP loci or duplicate genotypes that had fewer than five mismatching loci to prevent the rare case that an individual was sampled twice) and three methods to identify the progeny of translocations: parent pair assignments using SNPPIT version 1.0 (Anderson 2012); single-parent assignments calculated in SEQUOIA (Huisman 2017), with inputs assembled in SingleSEQUOIA (available at https://github.com/delomast/singleSequoia); and full-sibship assignments using COLONY version 2.0.6.5 (Jones and Wang 2010). The three software programs were chosen because they were well suited for each component of the analysis. Both SNPPIT (Hess et al. 2015) and SingleSEQUOIA (Hess et al. 2021) have been tested via simulation for accuracy of parent pair and single-parent assignments in Pacific Lamprey, and both programs run in less than 1 h for data sets of this size. COLONY takes significantly more time than the other parentage programs to analyze similar-sized data sets (i.e., days) but has been shown to provide accurate pedigrees of full sibship in Pacific Lamprey (Whitlock et al. 2017). SNPPIT was run with the parameter “–max-par-miss” set to 259 (total SNP N − 1), which effectively allows all parents and offspring in the data set to be compared regardless of missing data. All candidate parents were analyzed as a single population in SNPPIT, allowing all possible combinations of individuals to be evaluated as parent pairs (6,899 × 6,898 possible pairs). This design allowed the possibility that two candidate parents from different release sites or years could form a parent pair and avoided having to assume that adults spawned in the same year or at the same site in which they were released. We used a 0.5% per-allele error rate for the SNPPIT error parameter because it is larger than the observed per-locus error rate of 0.2% (Hess et al. 2021); additionally, we used a very stringent false discovery rate threshold of less than 0.01% (i.e., we expected the fraction of offspring assigned to incorrect parents to be less than 1 in 10,000). All candidate offspring with assignments that met this false discovery rate threshold were considered confident assignments and were excluded from the subsequent single-parent assignments made using SingleSEQUOIA. SingleSEQUOIA is an R package (R Development Core Team 2020) that interfaces with IDFGEN (R package available at https://github.com/mackerman44/idfgen) objects to build SEQUOIA (Huisman 2017) inputs for single-parent assignment. SEQUOIA infers parent–offspring relationships by comparing the likelihoods of the genotypes given the existence of a parent–offspring relationship or an alternative relationship (e.g., unrelated, aunt–niece). This calculation is made in part based on observed allele frequencies and Mendelian inheritance and by assuming independence between loci. The threshold for accepting single-parent assignments was set to a value of 0.5⋅log₁₀ (likelihood ratio) between a parent–offspring relationship versus unrelated, which was demonstrated to attain high accuracy by Hess et al. (2021).

To determine full-sibling pairs, we used COLONY to analyze genotypes of 24,265 candidate siblings (henceforth, referred to as the “sibling” data set). The sibling data set included 18,957 fish from the offspring data set and 5,496 fish from the parent data set, which were only the adults collected from recent years (2015–2018; Table A.1). The sibling data set also excluded 188 fish from the offspring data set to avoid overlap of multiple generations (see Supplemental Methods available in the online version of this article). Translocation adults collected during recent years were included in the sibling data set because they likely represented the same generation as the other larval and juvenile fish in the sibling data set (i.e., these adults arrived too recently to be plausible candidate parents of most age-classes represented among the larval and juvenile collections in the data set and they were old enough to be plausible siblings of the oldest collections of larvae and juveniles in the data set). We used the following parameter settings: polygamous mating for males and females without inbreeding, full likelihood, medium length of run, no allele updating, and no sibship priors. These parameter settings were chosen based on previous work that empirically demonstrated the polygamous system of mating and confirmed the accuracy of sibship reconstruction based on comparisons with known familial relationships (Hess et al. 2015). For each pair of full siblings, we assumed that the full-sibling family (i.e., all full siblings from a single pair of parents) originated nearest to the sibling collected at the upstream-most site and was present at least since the earliest collection year of either sibling. Therefore, all collections were ordered from most upstream to most downstream (and oldest to newest within sites), and natal origin was determined using the sibling from the upstream-most location.

Question 1: Did Translocation Increase Larval Abundance at Adult Release Sites in the Snake River Basin?

Using electrofishing and screw trap collections (predominantly larval collections), we calculated the proportion of parentage assignments across the “translocation” sites and putative “volitional” sites of the Snake River. We classified streams of the Snake River basin as sources of either “translocation” or putative “volitional” production of larvae and juveniles based on whether the streams had received translocated adults or not (translocation release sites 83–108;  Table A.1; Figure 1). These were only used as initial classifications for volitional production because any translocation parent assignments identified within those areas would be reclassified as translocation production. All larval and juvenile collections from translocation and volitional sites of the Snake River basin were used to estimate the presence of translocation offspring via calculation of the proportion of parentage assignments at each site. Generally, we expected that a direct boost to larval abundance would result in collections from each translocation site having a number of parentage assignments that was proportional to the tag rates of the parents released at that site during a given year. This expectation was based on the assumption that there were no larvae at any of the translocation release sites prior to the initiation of translocations. To test this assumption, we calculated the proportion of parent pair (trio) assignments out of the total parent assignments (trios and single-parent assignments combined) for each release group of parents (i.e., a group of adults released at a particular site and year). We expected that at high tag rates (e.g., >90%; equation 1), more trio assignments would be observed compared to all assignments to that release group. Assessment of this expectation was important because any release group that did not conform to expectations (e.g., had a low percentage of trio assignments in larval samples at a site despite a high adult tag rate) may indicate sites where a greater than expected number of unsampled, volitional spawners was present. Evidence of volitional spawners would not negate a finding that translocations boost larval abundance, but it would help to characterize sites where translocations were acting as a reintroduction as opposed to supplementation of existing volitional spawning.

Question 2: Did Translocation Increase Juvenile and Adult Abundance Downstream from the Snake River Adult Release Sites?

The following three locations were tested for the presence of juvenile translocation offspring: lower Snake River dams (primarily Lower Granite Dam, 606 rkm from the mouth of the Columbia River); John Day Dam; and ocean-phase samples from the U.S. West Coast, Southeast Alaska coast, and Bering Sea. For each location, we estimated the percentage of total Pacific Lamprey spawners that were released as translocated adults (i.e., “percent spawner abundance translocated”; Supplemental Methods). The estimates of percent spawner abundance translocated were used to test whether per capita juvenile production from translocated adults was equal to production from their volitional counterparts (i.e., the percent spawner abundance translocated should equal the percent juvenile production originating from translocations). Calculation of per capita juvenile production required estimation of the number of volitional and translocated Pacific Lamprey spawners at each location. We obtained 24-h dam counts across the years 2006–2017 (Fish Passage Center; www.fpc.org) to estimate the number of volitional Pacific Lamprey spawners at lower Snake River dams and John Day Dam, and we extrapolated an estimate of volitional spawners for the ocean-phase samples (see Supplemental Methods). Juvenile production abundance was estimated at lower Snake River dams and John Day Dam (see Supplemental Methods). For the ocean-phase samples, we used drainage areas to serve as an estimate for relative proportions of expected abundance. The three relevant drainage areas in our analysis were estimated to be 10.1 million km² (3.9 million mi²) for the entire Pacific Ocean, 585,337 km² (226,000 mi²) for portions of the Columbia River above John Day Dam, and 267,287 km² (103,200 mi²) above Lower Granite Dam. Assuming that abundances were proportional to drainage size, we expected that the ocean collections would comprise 6% of natal origins from above John Day Dam and 3% of natal origins from above Lower Granite Dam.

The following four locations were tested for the presence of adult translocation offspring: Willamette Falls, Bonneville Dam, The Dalles Dam, and John Day Dam (see Supplemental Methods for details on fish collection at each location). Adult production abundance at Willamette Falls was estimated by using a daily escapement estimator from Whitlock et al. (2019), and adult production abundance at Bonneville, The Dalles, and John Day dams was obtained from daytime dam counts (see Supplemental Methods).

We estimated the abundance of natal origin fish at three hierarchical levels: (1) translocation versus volitional, (2) stream site, and (3) PBT release group. Individuals in the data file were categorized by preferentially using information from parentage and secondarily using information from sibship assignments. In this way, individuals collected at sites outside of translocation streams could still be identified to a translocation release group by PBT or possibly identified to translocation streams with sibship. By default, if an individual had neither parentage nor sibship information, the individual’s “natal origin” was categorized using the location where it was collected.

Abundance estimates and 90% confidence intervals (based on 1,000 bootstraps; α = 0.10) were automated with the fishCompTools package in R (https://github.com/delomast/fishCompTools). Three input files were used for the estimate: individual sample data with natal origins, count data or abundance index, and PBT rates. The count data were stratified with a minimum sample size of 30 per stratum. A single stratum was used for the collection year if a more specific collection date was unknown. The “spibetr” (Salmonid Prior Information for Balancing Expansion by Tag Rates) function was used to balance PBT expansion by concordantly reducing the estimated abundance of unassigned fish that had the same natal origin category, thereby avoiding the “double-counting” of translocation fish (Delomas and Hess 2021; see Supplemental Methods).

Questions Related to Pacific Lamprey Biology

Basic biological information, including age, the ages at metamorphosis and migrations, growth rates, dispersal rates, and life span, is key to developing life cycle models and informing management decisions. These basic parameters have not been measured for Pacific Lamprey in the Columbia River basin, thus hindering development of management strategies tailored to specific regions. The large genetic sample sizes and time span needed to monitor the success of Pacific Lamprey translocations in the Snake River provided a means to supply this biological information. Parent pair assignments allowed us to determine whether parents deviated from the general expectation that they would produce offspring near sites where they were translocated and find mates among adults of the same release group. Specifically, biological data for candidate offspring (i.e., body length, collection date and location, and life stage classification) were matched with the release site and release date of the parents from the analyses and were used to address the following questions with direct relevance to adaptive management of the translocation program.

First, is basin size or the stream temperature of release sites correlated with larval and juvenile growth? Size at age was used as a proxy for growth, and this metric was calculated using the total body length (mm) of the larva or juvenile and its age. We calculated age using the parentage assignments of candidate offspring and candidate translocation parents. Age was estimated based on the number of days between the offspring’s date of capture and the most recent release date of one of the parents. Only 0–4-year-olds were utilized because growth is linear during that period (Hess et al. 2021). Watershed metrics were gathered from an ESRI online repository of spatial layers (Hill et al. 2016). Stream temperatures were based on mean August temperature (1993–2011) obtained from a regional database of modeled stream temperatures (https://www.fs.fed.us/rm/boise/AWAE/projects/NorWeST.html). We compared size at age using larvae between 0 and 4 years of age that were collected via electrofishing to allow comparisons across multiple streams, but we also compared rotary screw trap collections for a few streams in which large sample sizes could be obtained across multiple age-classes.

Second, are streams that produce fast-growing salmonids equally beneficial to Pacific Lamprey? We obtained the average size at ages 1 and 2 for steelhead Oncorhynchus mykiss (anadromous redband trout) smolts captured in rotary screw traps from the same streams surveyed for Pacific Lamprey in the Snake River, and we tested whether the size-at-age metrics of Pacific Lamprey larvae and steelhead smolts were correlated. Correlations were evaluated with a Mantel test using the ade4 package in R, and P-values were generated with 9,999 permutations.

Third, what are the ages of juvenile Pacific Lamprey when they are found migrating through the main-stem Snake River? We compared the average age and range of ages for all PBT release groups that were identified among larval and juvenile samples across all sites in our data set. We also examined the average age that specific cohorts were greatest in abundance at Lower Granite Dam; estimating the peak abundance of the offspring of each PBT release group as they passed a common point (Lower Granite Dam) provided a useful proxy for juvenile age at downstream migration (Table 1).

Fourth, how long is the ocean duration of the Pacific Lamprey parasitic phase? Juvenile Pacific Lamprey are expected to migrate to the ocean soon after passing major dams on the main stem of the Columbia River (Clemens et al. 2019). To estimate the ocean duration, we calculated the number of years between when the peak abundance (i.e., average year of abundance distribution) of full-sibling cohorts of juvenile Pacific Lamprey occurred at Lower Granite Dam (site 57 in Figure 1,  Table A.1) on the Snake River and when the peak abundance (average year of abundance distribution) of their adult siblings occurred at Willamette Falls (site 79), Bonneville Dam (site 81), The Dalles Dam, and John Day Dam (site 82). Overwintering behavior (i.e., protracted time in freshwater) of adults prior to their collection at Bonneville Dam is rare (<5%) among adults collected at that dam (Hess et al. 2014); hence, their collection year at Bonneville Dam was assumed to be the same as their freshwater entry year (i.e., adult migration year).

We used the results from our estimates of ocean duration to project the future number of returning adult translocation offspring at Bonneville Dam. For this projection, we used the cumulative return of adult abundance estimated at Bonneville Dam for the 2011 Lower Granite Dam cohort as a general model (i.e., each sibling juvenile cohort passing Lower Granite Dam will migrate as adults past Bonneville Dam in relative proportions of 4.1% in the fourth year after juvenile out-migration, followed by 37.3, 45.0, and 13.6% of the total cumulative adult abundance in subsequent years). We calculated the ratio of the total number of larvae and juveniles estimated at Lower Granite Dam to the number of Snake River-origin adults estimated at Bonneville Dam (i.e., juvenile-to-adult return [JAR]) for the Lower Granite Dam cohorts from 2011 to 2013. The average JAR for these years was then used to project what the total Snake River-origin adult Pacific Lamprey abundance would be at Bonneville Dam in future years (2019–2025) based on the abundance of larvae and juveniles at Lower Granite Dam from 2014 to 2018. Further, we assumed that the observed proportions of translocation offspring at Lower Granite Dam (larvae and juveniles) in 2014–2018 would be equivalent to the proportions of translocation offspring in the projected Snake River-origin adult Pacific Lamprey abundance at Bonneville Dam.

Finally, do adult Pacific Lamprey of the Snake River preferentially return to areas close to their natal origin? Similar to the previous question, we focused on the full-sibling cohorts of juvenile Pacific Lamprey at Lower Granite Dam (collection years 2010–2018) and compared our estimates of the abundance of adult full siblings from these cohorts at Willamette Falls and the main-stem Columbia River dams during the adult migration years (2014–2018).

Parentage Assignments

We observed 1,982 total parent assignments (N = 1,348 trios; N = 634 single parents) using our candidate parent baseline (N = 6,899 translocated adults) and offspring data set (N = 18,957). Most of these assignments (89%; N = 1,763) involved fish in the offspring data set that were specifically collected from “translocation areas” (N = 3,481; Figure 1). The total parent assignments confirmed direct reproductive success of 353 translocated adults in our parent baseline. In general, the total parent assignments represented 10 spawn years (SYs; 2007–2017), 13 streams, and 59 release groups (Table S1 available in the Supplement in the online version of this article). The total parent assignments were not uniformly distributed; a high proportion of assignments (83%) originated from SYs 2007–2013 and three streams (77% from Lolo Creek, Newsome Creek, and the South Fork Salmon River).

The total parent assignments included 189 unique parent pairs (both parents were sampled). For most of these parent pairs (73%), both parents had matching SY and release site information; however, the other 27% of parent pairs were mismatching in terms of SY, release site, or both. Eleven (6%) of the 189 unique parent pairs had mismatching release sites. This suggests that at least 11 translocated adults moved to a new location before spawning. Further, 33 of the translocated adults (9.3% of the 353 successful parents) observed in a parent pair had mismatching release years; that is, these adults spawned with adults released in subsequent years, which meant that one of the adults likely spawned 2 years after its migration year. This result represents evidence of delayed spawning in Pacific Lamprey, and this study is the first to quantify its frequency.

Most successful spawners were adults released in 2007–2014; for these release groups, the percent spawner success (i.e., percentage of adults with one or more offspring assignments out of the total number of sampled adults for each PBT release group) averaged 51.6% across years. After 2015, percent spawner success was 5% or less (Figure S1a available in the Supplement in the online version of this article). The success rates of each SY from 2007 to 2014 (hereafter, referred to as “maximum detection” SYs) were in a narrow range (50–60%) and did not correlate with the number of adults in each release (Figure S1b).

Sibship Assignments

For the sibling data set (N = 24,453), we identified 16,827 unique full-sibling families; there were 9,494 full-sibling families containing two or more siblings (average = 1.4; range = 1–110). There were 5,403 fish whose natal origins were identified to an “upstream” collection site (either located upstream of their collection site or originated from an earlier collection year) based on their sibling assignments. Among these sibship-determined origins, there were 24 different stream sites represented, 8 of which were considered translocation streams and included natal origins for 2,846 translocation progeny (53% of the sibship-determined origins).

Question 1: Did Translocation Increase Larval Abundance at Release Sites in the Snake River Basin?

We sampled 98 unique release groups across the release years 2007–2019. Based on the entire offspring data set, we could confirm that 59 release groups had direct evidence of reproductive success via parentage assignments (Table S2). Among the 34 release groups sampled from the maximum detection SYs (2007–2014), all but a single release group (Asotin Creek, SY 2014) had a least one confirmed offspring assignment (average = 49 assignments; range = 0–612), and 99.9% of these confirmed offspring were larvae or juveniles that had survived past their first summer (i.e., >1 year of age). Therefore, qualitatively, 97% of all release groups from the maximum detection SYs were verified as reproductively successful and boosted larval abundance across release streams of the Snake River.

In general, collections from translocation sites had high percentages of parent assignments compared to those from volitional sites (Figure 2). We found a linear trend when we plotted the percent of parentage assignments for each collection versus the average tag rates of the groups of adults released at each respective site (R² = 0.68; Figure S2).

Finally, to test the assumption that Pacific Lamprey were extirpated at all translocation sites prior to translocation (i.e., that all larvae found at translocation streams were the offspring of translocated parents), we calculated the proportion of observed parent pair (trio) assignments out of the total observed parent assignments (trios and single-parent assignments combined) for each release group of parents (i.e., a group of adults released at a particular site and year). For the 24 release groups with sufficient sample sizes (N > 5 parent assignments), we found that there was a logistic relationship of trio detection versus tag rate (Figure S3). Nineteen of the release groups had tag rates over 95% and were mostly observed with high trio detection (>75%). The other five release groups had tag rates less than 75% and low trio detection (<10%). The four exceptions were release groups that had over 95% tag rates but low trio detections (<50%): South Fork Salmon River SY 2013 and SY 2015; Asotin Creek SY 2013; and Newsome Creek SY 2013.

Question 2: Did Translocation Increase Juvenile and Adult Abundance Downstream from Snake River Adult Release Sites?

We estimated that translocation offspring made up 0% of the juvenile production at Lower Granite Dam until 2013, when translocation offspring were estimated to be above 2%; the percentage then increased during subsequent years to over 19% in 2018 (Figure S4). For the first 6 years (2006–2011) and the most recent 6 years (2012–2017) of our study period, the percent spawner abundance translocated above Lower Granite Dam was 42.8% and 44.6%, respectively (Figure 3a). Our estimate of the per capita juvenile production of translocated spawners in 2018 would therefore be less than half of volitional production. There was high variability in juvenile production across release groups (Table S3) and individual streams (Table S4).

We estimated that Snake River translocation offspring made up 3.5% and 3.0% of the juvenile production at John Day Dam in 2017 and 2018, respectively (Figure 4), while the volitional offspring made up 27.0% and 16.6% in those same years. The percent spawner abundance translocated above John Day Dam was 1.1% in 2006–2011 and 2.7% in 2012–2017 (Figure 3b). Therefore, per capita juvenile production was greater for translocated spawners relative to their volitional counterparts. As was the case for juveniles at Lower Granite Dam, juvenile production at John Day Dam varied by release group (Table S5) and Snake River stream (Table S6).

We estimated that Snake River translocation offspring made up 0.0% and 0.2% of the juvenile production in the ocean collections from 2017 and 2018, respectively (Figure S5), while volitional offspring made up 1.2% and 0.0% in those same years. During both 2017 and 2018, our ocean sample sizes were greater than 500 in each year, so they should have yielded 15–40 samples for these two natal origins if they were to meet the expected per capita production proportions of 6% and 3% of natal origins from above John Day Dam and Lower Granite Dam, respectively. However, the observed numbers of John Day Dam-origin fish in the ocean collections were nine (1.3% in 2017) and two (0.3% in 2018). The observed numbers of Lower Granite Dam-origin fish in the ocean collections were eight (1.2% in 2017) and one (0.2% in 2018). Snake River translocations were represented by one fish (0.2%) from Lolo Creek that was detected in the 2018 ocean collection (site 70 near the international border between the USA and Canada; Figure 1).

Finally, we estimated the abundance of Snake River-origin Pacific Lamprey among adults collected in the Columbia River (Table 2) and at Willamette Falls (Table 3). There were no parentage or sibling assignments from translocation sites that identified translocation offspring among the adult collections to date. However, we were able to project numbers of future returning adult translocation offspring (see the section on ocean duration below).

Questions Related to Pacific Lamprey Biology

How long is the ocean duration?

The earliest cohort of juveniles at Lower Granite Dam with a large sample size (N > 400) was sampled in 2011, and most of the abundance of this cohort (95.1%) was estimated to pass downstream at Lower Granite Dam in 2011 (Table S9). The adult siblings of this cohort were estimated to pass Bonneville Dam on average 5.6 years later (range = 4–7 years) according to the weighted average abundance, which peaked in 2017 at 3,014 adults (Figure S11; Table 2). The subsequent 2 years of juvenile cohorts from Lower Granite Dam (2012 and 2013) consistently fit an ocean duration of approximately 5 years; peak abundances of the juvenile cohorts that first appeared at Lower Granite Dam in 2012 and 2013 (Table S9) were estimated to pass Bonneville Dam on average 5.0 years later (peak abundance of 603 adults in 2017) and 4.7 years later (peak abundance of 158 adults in 2018). The only year from which we had length data for adults at Bonneville Dam was 2018, and 44 of these fish were identified as Snake River origin (average length = 701.4 mm; range = 610.0–773.0 mm). In contrast, the average length of all adults measured at Bonneville Dam in that year (N = 996) was 672.5 mm (range = 328.0–805.0 mm).

We combined data from the distribution of ages observed among the larvae and juvenile samples from this study and created a single figure of the hypothesized life span of a Snake River Pacific Lamprey (Figure 5). We added the observed distribution of ocean duration to the juvenile age distribution to estimate the prespawn adult age distribution—that is, the putative ages that would be represented among adults that appear at Bonneville Dam during each year. Finally, we added a distribution of years that were represented by the observed proportion of delayed spawners (adults that delayed spawning for 2 years; 9.3%) and regular spawners (adults that overwintered for a single year; 90.7%) to estimate the postspawn adult age distribution (Figure 5). The median age of juveniles was 4.8 years, and the median age of postspawn adults was 12.9 years. The upper 95th percentile of the age distribution for juveniles and postspawn adults was 10.4 and 17.7 years, respectively.

Benefits of Translocations

Translocations of adult Pacific Lamprey into the Snake River have continued for over a decade. We quantified the benefits of this restoration action and showed that the translocation program succeeded in its primary objective of restoring larval abundance to streams that were previously thought to no longer harbor Pacific Lamprey (97% of all release groups that were released between 2007 and 2014 were confirmed to have produced offspring). After the 2014 release year, there was a decrease in reproductive success, which we attributed to lower detection probabilities for more recent release years (2015–2018) due to our efforts to minimize lethality by not sampling small larval offspring.

We expected that the percentage of parentage assignments for each collection would be proportional to the average tag rates of the groups of adults released at each respective site. This was indeed the case as demonstrated by the high R² of the linear trend when the percentages of parent assignments were plotted versus tag rates. This result was a quantitative assessment of the release groups’ reproductive success. We also demonstrated with empirical evidence that most translocation sites had no lamprey production prior to translocations. This finding was based on the fact that for sites in which release groups were sampled at high rates (i.e., genetic tag rates > 95%), 79% had larvae with high rates of assignment to a pair of translocated parents. There were four outlier release groups, which may represent sites where Pacific Lamprey were not extirpated, as there appeared to be more unsampled parents than expected. However, in general our assumption that Pacific Lamprey were extirpated at the sites prior to translocation was supported by the fact that electrofishing surveys showed no presence of larvae prior to 2007 (Cochnauer and Claire 2009) and by results from most release groups in this test of trio detection rates.

Further, this study provides the first direct evidence that Pacific Lamprey translocation programs not only boosted the abundance of juveniles from the Snake River, but also affected juvenile abundance farther downstream in the main-stem Columbia River and even in the ocean. A single fish from a translocation stream (Lolo Creek) was identified in the northern-most ocean collections from 2018 near the international border between the USA and Canada.

During the same years, juvenile abundance at John Day Dam attributed to Snake River translocated adults exceeded the per capita production of their volitional counterparts. That is, translocated fish were apparently able to produce offspring at higher levels than the volitional migrants. This is likely due in part to the fact that volitionally migrating adults must still pass both large main-stem dams and many additional low-elevation obstacles to reach prime spawning habitats (Keefer et al. 2009; Moser et al. 2015). In contrast, translocated fish are overwintered in tanks supplied by water from the Clearwater River, which flows past the Nez Perce Tribal Hatchery facility, with little requirement for energy expenditure; the fish are then transported directly to optimum habitats for spawning. Moser et al. (2021) proposed that for anadromous species, passage barriers may be more important than habitat quality. Our genetic findings support this hypothesis. Additional support was provided by Hess et al. (2021), who showed that restoring passage to adequate habitat via dam removals for Pacific Lamprey in the Elwha River was sufficient to re-establish productivity.

One risk of translocations for conservation and restoration has been the potential disruption of population genetic structure when the individuals used for translocations are from a source that has mixed natal origins (Weeks et al. 2011). High levels of gene flow and the accompanying weak population structure exhibited by Pacific Lamprey within the Columbia River basin (Hess et al. 2013, 2014) serve to lessen this potential risk. The observed level of productivity demonstrated by this highly dispersive species in response to translocation suggests that translocation is an effective restoration strategy with benefits that outweigh the low potential genetic risks within the Columbia River basin. In fact, this may be an effective strategy to consider for other large, interior basins in the historic range that suffer chronically low Pacific Lamprey abundance due to passage barriers. For small, coastal basins, simply resolving passage impediments may suffice to allow natural recolonization that will restore Pacific Lamprey abundance (Reid and Goodman 2016; Hess et al. 2021).

Translocations Will Likely Boost Adult Returns to the Columbia River in Future Years

Although this study did not identify any offspring of translocated Pacific Lamprey in adult collections, our results indicated that detection of these adult offspring is imminent. Lack of detection thus far is due to the long life span of Pacific Lamprey (maximum age > 15 years). Using genetic analyses, we have shown conclusively that the freshwater phase of the larval life stage can exceed 10 years and that the ocean phase may extend past 5 years. Hence, there likely has not been sufficient time for adult offspring of translocations to return, despite the fact that over 10 years have passed since the first translocation in 2007.

Importantly, our projections indicated that average levels of Pacific Lamprey JAR (analogous to the smolt-to-adult returns used for salmon management) would result in adult translocation offspring adding to adult abundance in the interior Columbia River basin during future years. The magnitude of this boost was projected to be similar to the number of adults collected for translocations (e.g., 751 adults were collected by each treaty tribe in 2021). If our projections are realized, this boost would effectively replace the abundance used for translocations as soon as 2023, when we estimate that 722 adults would return to Bonneville Dam. Further, we predicted that the first translocation offspring (2007 release year) could be detected among adults that returned in 2021. This direct boost to adult abundance was unanticipated and is a significant extension of the benefits expected from translocation. Given the lack of philopatry in this species, the expected benefit to adult abundance from translocation was hypothesized to occur indirectly via greater attraction of volitional adults due to the increased levels of larval pheromones produced by translocation offspring (Ward et al. 2012; Moser et al. 2015).

New Biological Data Relevant to Adaptive Management

This project has generated biological data with direct relevance to adaptive management of the translocation programs for Pacific Lamprey: (1) identification of release group characteristics and habitat attributes that optimize larval growth and productivity, (2) estimation of the range and average age at metamorphosis and adult spawning migration, (3) ocean residence time, (4) estimation of the rate of return of adults to their natal drainage, and (5) delayed spawn timing. With our results, managers can target translocations to the most productive areas and forecast with better accuracy both out-migration timing and adult rates of return. Estimation of the range and average times of life stage transitions is helpful in setting expectations for how soon translocation offspring may potentially start to affect adult abundance.

Parentage and Sibship Analyses Can Estimate Movement of Highly Dispersive Species on a Time Scale Relevant for Management

Conventional metrics of gene flow provide limited information for characterizing subtle differences in dispersal rates. Specifically, the model that relates the genetic differentiation index F_ST to gene flow is on a time scale that is often coarser and more ancient than needed for management. This limitation is particularly problematic for species that exhibit intermediate or high gene flow (i.e., panmixia). Pacific Lamprey show intermediate levels of gene flow—somewhere between a state of panmixia and weak philopatry (Spice et al. 2012). For such species, F_ST decreases to near zero and the same magnitude of error in F_ST translates into a much larger error in effective migration (i.e., mN_e) estimates compared to estimates for species with restricted gene flow (Waples 1998).

As an alternative to F_ST metrics that use allele frequencies across collections, Mateus et al. (2021) proposed the use of individual-level genotypic data to perform either parentage analysis or full-sibship analysis for estimation of migration rates between potential sources and sinks of lamprey production. Hess et al. (2021) attempted to apply these parentage and sibship methods to Pacific Lamprey in coastal subbasins but were limited by insufficient recapture data. However, for the Snake River collections described here, these analysis methods were an effective tool for estimating dispersal rates in lampreys and required 4 years of juvenile collection (2011–2014) from one source and 4 years of adult collection (2015–2018) from two different sinks. This 8-year time frame is longer than the 5-year time frame suggested by Mateus et al. (2021) because the ocean duration for Pacific Lamprey was longer than expected. However, consistent with Mateus et al. (2021), sibship analysis required a shorter span of years to estimate dispersal rates in comparison with parentage analysis. Parentage would likely require a span of 15 or more years from the initial year of parent sampling until the offspring from those adults return to capture sites.

This quantification of subtle differences in Pacific Lamprey dispersal from the Snake River allowed us to identify vastly higher rates of return to Bonneville Dam relative to Willamette Falls. This detection ability provides guidance for how much we would expect translocation efforts in the interior Columbia River to replenish interior Columbia River streams versus augmenting harvest potential at Willamette Falls. We observed a difference in dispersal rates even though F_ST is low between sites within the Columbia River basin based on neutral genetic markers (average pairwise F_ST = 0.002; Hess et al. 2013). The fact that gene flow for Pacific Lamprey was relatively high for ancient time scales (as measured with F_ST) supports the idea that Pacific Lamprey in the Columbia River basin effectively constitute a single population (Hess et al. 2013). However, on recent time scales, Pacific Lamprey from the Snake River appear to exhibit a preference for returning to the main-stem Columbia River compared to the Willamette River—an important finding for contemporary management. Finally, on even finer spatial scales, we observed that once Pacific Lamprey overwintered and were translocated, our results supported the assumption that translocated fish had a high probability (94%) of spawning near the site of release.

Concluding Remarks

Translocation of Pacific Lamprey directly boosted the abundances of larvae and juveniles and likely will soon (2023) be supplementing counts of returning adults. Monitoring the success of this Snake River translocation program via genetic analysis has allowed for the quantification of translocation benefits while addressing several biological uncertainties. These biological data suggest that larger watersheds produce faster-growing larvae and that most juveniles may leave freshwater at about age 5–7. However, some larvae may linger for a decade of freshwater rearing, and after metamorphosis they could reside in the ocean for half a decade before returning to freshwater as adults, even then spending up to 2 years before spawning. Importantly, this species exhibits enough philopatry to ensure that the thousands of adults taken for the translocation program will be replaced by their returning offspring. Further, a majority of those Pacific Lamprey that return to this basin will likely bypass lower-river tributaries and occur at the first main-stem dam on their way to streams in the interior Columbia River basin.