Evaluation of Three Atlantic Salmon Strains for Resistance to Copepodid Sea Lice Attachment

Abstract

Sea lice have been a persistent pest of the salmon farming industry for more than 50 years. In this study, we aimed to identify if different strains of Atlantic salmon with discrete long-term lice exposure histories had variable resistance to copepodid attachment and/or different attachment-specific transcriptome patterns. We additionally sought to characterize lice distributions on fins, head, and skin and identify if attachment location influenced transcriptomic profiles of lice. Lice counts were correlated with body size and highest on St. John River (SJR; open ocean-run) relative to Grand Lakes Stream (GLS; 200-year restricted ocean-run) or Sebago Lake (CAS; ~11,000 years landlocked) Atlantic salmon. However, lice density was similar between strains. Skin and fins had expectedly different transcriptomic profiles; however, notable differences were not observed between salmon strains. Variance in lice transcriptomes was minimally affected by attachment location even though lice strongly preferred fins relative to head or body. Attached lice did have different transcriptomic profiles on GLS relative to CAS or SJR. This study cumulatively identified a minimal host evolutionary component for sea lice attachment resistance, although lice behavior post-attachment appeared somewhat affected by strain. Non-uniform settlement distributions and tank-specific variability in lice attachment were observed across populations.

1. Introduction

Sea lice, particularly Lepeophtheirus salmonis, are a significant economic pest to Atlantic salmon aquaculture [1,2]. Sea lice mitigation costs have been estimated at around 9% of farm gate value or just under one billion EUR a year for the global industry [1,2]. These costs are due mainly to loss of growth, inefficient harvesting, and direct and indirect management efforts [1,3,4,5,6].

Initial sea lice management beginning in the late 1970s focused on the use of antiparasitic drugs, and these have remained a major tool in the management of sea lice through to the present day [4,7,8,9,10,11]. However, as early as the late 1980s, drug resistance has repeatedly developed [12,13] and drug regulatory approval processes have lengthened, forcing farmers to look for non-drug-based alternatives for managing sea lice in many circumstances [14].

Over the last 15 years, numerous alternative methods for managing sea lice have been developed. These include increased use of cleaner fish such as lumpfish, the development of lasers, various barrier technologies such as snorkel cages, acoustic deterrence devices, functional feeds, thermal and salinity-based de-lousing, and selective breeding [15,16,17,18,19,20,21,22,23,24,25,26,27]. In response, lice have developed ways to avoid some of these treatment methods. For example, some families of lice have adapted to inhabit deeper waters than is typical, allowing them to avoid various partial barrier technologies such as plankton skirts or snorkel cages [28]. Likewise, there is evidence that lice may have the capacity to evolve tolerance to both thermal and freshwater treatment [29,30,31]. Thus, to mitigate effective lice adaptation, lice management is now trending towards utilizing multiple strategies within an integrated pest management system that incorporates multiple strategies for preventing lice attachment either by parallel or series application [32].

Selective breeding has long been seen as a potential drug-alternative method to mitigate sea lice in aquaculture in an integrated pest management approach where specific emphasis has been directed at inhibiting attachment and/or enabling host-directed removal early in the louse’s life cycle. Glover, et al. [33] first evaluated five strains of Norwegian salmon—three wild strains and two farmed strains—and showed significant differences in lice attachment between strains. Soon after, the first heritability estimates were reported based on natural infections [34], and a potential for resistance heritability has now been corroborated by multiple groups for various populations under selection, which demonstrate a heritability component for settlement resistance between 0.10 and 0.35 [35]. Published estimates from our program at the USDA ARS National Cold Water Marine Aquaculture Center (NCWMAC) fall within this general range (0.20–0.30) [25,36]. Selection for resistance to sea lice settlement has further evolved from pedigreed family-based programs to genomic selection-based programs as genomic tools and genotyping costs have come down. It has also been identified that resistance to sea lice settlement is highly polygenetic across host genomes with no one genetic marker accounting for more than 3% of the genetic variance [25,37,38]. Genomic selection, which accounts for both intra- and inter-family genetic variance, is thus better suited to selection on polygenetic traits than pedigreed family-based selection, which only accounts for inter-family variation.

The mechanism(s) for resistance in Atlantic salmon has yet to be clearly demonstrated. Immune-mediated mechanisms are thought to be the primary means of resistance in Pacific salmon, but the importance of behavioral mechanisms to avoid initial infestation has also been highlighted recently [39,40]. Previous work has demonstrated that the majority, roughly 60–80%, of copepodid settlement occurs on the fins of Atlantic salmon, with lower settlement rates on the body and head, hypothesized to be a result of variability in hydrographic and physical forces across fish surfaces [41,42,43]. The immune response of Atlantic salmon lice attachment, including at the copepodid and chalimus stages, has also been well studied [39,44]. Previous work characterizing the immune response of salmon to infestation by L. salmonis has focused on factors such as which primary immune pathways are being upregulated or downregulated, gene response at various time points after infestation, and response to secretory/excretory products. Most of these studies utilize skin samples taken from the body of the fish just behind the dorsal fin. Holm, et al. [45] examined the response of newly settled copepodids on skin samples with scales (at the traditional area just posterior to the dorsal fin) and a scaleless skin area from the head and identified that immune responses were different between the two regions. The immune responses of the fin have not been looked at previously. Given that there are differences between skin with scales and skin without scales, the potential for differences in fins may also exist. This may explain the preference for settlement on the fins over other body regions.

To identify if natural selection has differentially influenced genetic resistance of salmon strains to sea lice, this study assessed the relative susceptibility to sea lice attachment of Atlantic salmon with varying evolutionary histories for sea lice exposure. This included a historically landlocked stock (~11,000 years with no sea-lice exposure), a recently sea-run restricted stock (~200 years with intermittent and restricted sea-run capacity), and a continual ocean-run stock of Atlantic salmon. Additionally, we sought to identify if transcriptomic responses at the site of lice attachment were different for each salmon strain being infected, and if lice transcriptomic responses following attachment could give an insight into why lice settlement is non-uniform across fish surfaces (i.e., why lice appear to settle more on fins than on the head or body).

2. Materials and Methods

2.1. Experimental Fish and Husbandry

Three strains of North American Atlantic salmon were used in this study. One was a commercially developed strain originating from Saint John River stocks (SJR strain), which has maintained sea-run exposure to sea lice throughout its known history [36]. The second strain was obtained from the Grand Lakes Stream Fish Hatchery (GLS strain) operated by the State of Maine, USA, which has mainly been a landlocked stock for approximately the last 200 years, since dams in the water system were built and precluded or severely limited sea-run potential. The third strain was obtained from the State of Maine, USA, Casco Hatchery (CAS strain), whose populations are from Sebego Lake stock believed to be landlocked since the last glacial recession, approximately 11,000 years ago. Fish from all strains were obtained as eyed eggs at the same time and grown to approximately 20 months of age at the USDA-ARS National Cold Water Marine Aquaculture Center (NCWMAC) in Franklin, ME, USA. It should be noted that growth rates were highly variable between strains, resulting in substantial size discrepancies by 20 months (e.g., SJR fish have been growth selected for over a decade and were approximately 2× the mean body weight relative to GLS or CAS; Table S1). All fish were individually tagged with passive integrated transponders and smolted in 15ppt brackish well water at approximately 13 months of age. Thirty fish from each strain were then stocked into four replicate 1 m³ challenge tanks for a total of 90 fish per tank. The tanks were on a single recirculating aquaculture system with natural sea water (32 ppt) at 18.2 °C ± 1.6 °C. Fish were acclimated to the system for 10 days prior to sea lice challenge.

2.2. Sea Lice Collection and Rearing

Sea lice egg strings were collected from gravid females at commercial salmon farms during net-pen harvest events and transported back to the NCWMAC, where they were hatched and reared to the copepodid stage. Lice were reared in a recirculating system kept in natural sea water (32 ppt) filtered to 1 µm. Unhatched egg strings were placed in PVC culture chambers with 67 µm mesh on the bottom and side holes to allow water flow. Water was introduced from the top, creating a downwelling flow. Unhatched egg strings were separated from hatched copepodids and transferred to new culture chambers daily to create known 1-day cohorts of lice. All copepodids used in the challenges were 2–4 days old.

2.3. Sea Lice Exposure Challenge

Fish were challenged with lice in a static four-hour bath according to the standard protocol at the NCWMAC [36]. In brief, an estimated 100 copepodids per fish were acclimated to the temperature of the challenge system for a minimum of 30 min. Water flow to the challenge tanks was temporarily shut off, supplemental oxygen supplied, and the water levels reduced to approximately 1/3 of the normal volume (~300 L). Temperature-acclimated copepodids were added in a static four-hour bath, after which normal water flow and water volume was restored. Lice were allowed to develop for 7–10 days post infestation (dpi) to the second chalimus stage, after which fish were euthanized with an overdose of MS-222 (Syndel, Ferndale, WA, USA) and placed in a freshwater staining bath containing 8.3 mg/L of Neutral Red (Fisher Scientific, Fairlawn, NJ, USA) for 20 min to improve lice visualization for counting. Fish were removed from the stain, individually bagged, and lice counted under a dissecting microscope at 1–5× magnification. For all four replicate challenge tanks, the total number of lice per fish was recorded. For the fourth challenge tank, total lice counts were additionally subdivided by location to differentiate the head (including the operculum), the body, and the fins. The boundary between the head and body of the fish was defined as the area where scales began. The fins included the pectoral, pelvic, anal, adipose, and caudal fins, again with the appearance of scales as the boundary between the body and fin. Lice attached to the gills were not counted as they are considered an artifact of the challenge method [46].

2.4. Lice Enumeration

Lice count (LC) per fish at 7–10 dpi was recorded and further standardized to body weight (BW) by two different methods. The first method used to determine lice density (LD₁) was performed as defined by Gjerde, et al. [47], where LD₁ = LC/BW^2/3. The second method for determining lice density (LD₂) was calculated using the formula LD₂ = LC/SA, where surface area (SA) was derived using the equation provided by Frederick, et al. [48]: SA = 13.9 BW^0.61. To determine the portion of the SA that was on the head, fins, and body, measurements were taken from 13 fish, which included fish from all three strains and with weights ranging from 104 to 453 g. The SA for the fins and head was determined by cutting the fins off the fish, placing them on copy paper, then wrapping the head in paper towel and tracing both fin and head outlines on to the paper with a pencil. Outlines were then scanned, and ImageJ (version 1.15p) was used to measure the SA of each part [48,49]. The measured area of the fins was doubled to account for both sides. The body SA was calculated by using the SA model [48] to calculate the entire SA then subtracting the measured fin and head SA. A linear regression was fit using least squares to estimate the SA for each body segment, head, fins, and body (GraphPad Prism 9.5.1).

2.5. Transcriptomic Profiling

Tissue samples were opportunistically collected from four fish per strain from one replicate infection tank (Tank 4). Samples were taken with an 8 mm biopsy punch and included a fin and a body sample (body sample included scales, skin, and a modest amount of underlying muscle), each with a louse attached at the center of the punch. Samples were immediately frozen in liquid nitrogen until RNA extraction with Tri Reagent (Sigma Aldrich, St. Louis, MO, USA) following manufacturer’s instructions. Eluted RNA was further purified via DNAse-1 treatment and RNeasy micro kit cleanup (Qiagen, New York, NY, USA) following manufacturer protocols. Cleaned RNA was then sent to the Oklahoma Medical Research Foundation NGS Core facility in Oklahoma City, USA, where stranded mRNA libraries were prepared using the IDT xGen RNA Library Prep kit (Integrated DNA Technologies, Coralville, IA, USA) and sequenced on Illumina NovaSeq to achieve 2× by 150 bp reads targeting 20 million reads per sequencing library.

Raw demultiplexed FASTQ files were transferred to the USDA-ARS SCINet high-performance computing cluster, where adapter trimming was completed using TrimGalore (ver. 0.6.6) with default parameters. Trimmed reads were aligned to the combined Atlantic salmon (Refseq GCF_905237065.1) and Salmon louse (Refseq GCF_016086655.3) reference genomes using HISAT2 (ver. 2.2.1). Alignments were then separated by species (Atlantic salmon or salmon louse) and assembled and quantified using StringTie (ver. 2.2.0) following a standard differential expression analysis workflow [50]. One salmon (skin and fin) and the corresponding two louse libraries were excluded due to inadequate read coverage. Resulting transcript variants in remaining libraries were then consolidated into unigenes using the R package tximport (ver. 1.30.0).

2.6. Statistical Analysis

LC, LD₁, and LD₂ data were assessed for normality according to Shaprio–Wilk and Kolmogorov–Smirnov tests, and box-cox transformations were applied to each dataset to achieve normal distributions (Table S1). A two-way ANOVA was used to compare LC, LD₁, and LD₂ data between fish strains and challenge tanks (all data) followed by Tukey’s multiple comparison tests to determine differences between fish strains within each challenge replicate. A one-way ANOVA was used to examine the preference for lice settling on the three body surface regions: body, head, and fin. Tukey’s multiple comparison tests were used to assess differences between the three regions. These analyses were conducted in Graphpad Prism (10.4.2).

Principle component analyses of transcriptome data were completed separately for Atlantic salmon and sea lice, and differential expression of unigenes of sea lice relative to the host Atlantic salmon strain was assessed using the R package DEseq2 (ver. 1.42.0).

3. Results

3.1. Challenge Produced High Variation in Sea Lice Attachment Among Replicate Tanks

Infestation rates were highly variable between the four replicate challenge tanks used in this study. Mean counts ranged from 12.7 (Tank 3) to 32.4 (Tank 4) lice per fish (Figure 1). Lice density was similarly variable between tanks, with mean LD₁ ranging from 0.365 to 0.948 lice per ^2/3√g weight and LD₂ ranging from 0.035 to 0.092 lice per cm². Two-way ANOVA of normalized LC, LD₁, and LD₂ data further identified that challenge replicate (i.e., tank effect) accounted for a large portion of total variance (between 44 and 50%), while Atlantic salmon strain accounted for a minor portion of total variance (between 1 and 7%), with the interaction also having a minor contribution (accounting for 1–2%) (Table S2).

3.2. Copepodid Attachment Was Correlated with the Size (Weight) of Host Fish

A consistent correlation between lice attachment and fish weight was observed in all four replicate challenge tanks in this study, with positive Spearman r values between 0.28 and 0.54, p < 0.01 (Figure 2). The largest fish in all tanks were SJR salmon, with mean weights of 182–198% (p < 0.001) of the age-matched GLS or CAS salmon used in this trial. It was therefore not surprising that salmon lice counts were highest on SJR salmon relative to either CAS or GLS in two of the four replicate challenge tanks in this study (p < 0.001; Table S3).

3.3. Strain History of Sea Lice Exposure Contributed Minimally to Differences in Copepodid Attachment Density

Although salmon lice counts were highest on SJR salmon relative to either CAS or GLS in two of the four replicate challenge tanks (Table S3), strain-specific preference for SJR was confounded in this instance by the correlation of LC with fish weight, where SJR salmon were consistently the largest fish in each challenge tank. In comparing lice density—which standardizes for size differences—there was little observable difference in attachment density relative to the three host strains used in this study. In one of the four replicate challenges, CAS fish had lice densities slightly higher than GLS or SJR, but in the other three replicate challenge tanks, attachment density of lice on each salmon strain was indistinguishable (Figure 3).

3.4. Copepodid Settlement Occurred More on Fins Relative to Head or Body Independent of Atlantic Salmon Strain

In the single challenge tank where lice counts were segregated by location to the head, fins, and body (Tank 4), lice attached most on the fins (Figure 4). Specifically, an average of 65% of lice were attached to fins relative to 25% on the body and 10% on the head. High attachment counts on the fins also corresponded to the highest relative lice density (mean 0.24 lice per cm² fin), which was 3–6 times higher (p < 0.001) than on the body or head, respectively. Although lice density was second highest on the head (mean 0.07 lice per cm²) relative to the body (mean 0.04 lice per cm²), the body represented a substantially larger portion of the total surface area of the fish, and overall lice counts were higher on the body (mean 25%) relative to the head (mean 10%).

3.5. Atlantic Salmon Strain Was Not a Major Source of Skin/Fin Transcriptome Variation Following Sea Lice Attachment in This Study

Seventeen of the eighteen libraries (eight fin and nine skin with attached sea lice) submitted for sequencing passed quality control filtering, yielding an average of 24 ± 7 million total reads per library (Table S4). Total aligned reads after filtering resulted in an average of 18 ± 11 million reads (mean 74% total reads), for which most were associated with the Atlantic salmon genome (16 ± 6 million reads; mean 89% of filtered reads). For the Atlantic salmon-associated genes, global transcriptional expression patterns at sea lice attachment sites were reasonably discrete for skin and fin samples, contributing strongly to the main source of inter-sample variation in this study as seen via principal component analysis (Figure 5A). We therefore considered host evolutionary strain effects in a location-specific (skin or fin) manner. This identified no differentially expressed genes between the Atlantic salmon strains in either skin or fin samples at a false-discovery rate adjusted p-value < 0.05.

3.6. Sea Lice Transcriptomes Were Suggested to Vary Based on Host Atlantic Salmon Strain

Of the 17 libraries containing both salmon and sea lice transcripts, only an average of 11% of total filtered transcripts mapped to the sea louse genome in each library, with 5 libraries having less than 1% of total filtered transcripts mapped to the sea louse genome (Table S4). These latter 5 libraries were excluded from further sea lice considerations due to lack of coverage, leaving 12 individual louse libraries: 6 from fin, 6 from body, with 4 from each of the 3 strains of Atlantic salmon (Figure 5B). Interestingly, variability in louse transcriptome profiles did not appear notably affected by attachment location (i.e., fin vs. body). However, variation in transcriptome profile was observed for lice that were attached to GLS salmon compared to either SJR or CAS salmon, which were similar. Differential expression for lice attached to GLS vs. SJR/CAS identified 33 differentially regulated genes, with FDR adjusted p-values < 0.05—6 upregulated genes and 27 downregulated genes. The majority of the DEGs identified (24) were of unknown functions (Table S5).

4. Discussion

This study did not identify notable variation in lice susceptibility between the three different Atlantic salmon strains evaluated. Only one of the four replicated challenges had significantly different settlement rates between the strains, with the landlocked strain being potentially more susceptible to lice in this isolated instance. This is in contrast to a similar study conducted by Bui, et al. [51] that showed domesticated strains were more susceptible to lice infestation than sea-run or landlocked strains. However, Bui, et al. acknowledge that lice susceptibility can vary greatly across strains [33]. We speculate that this variability may also be confounded with general challenge variability, which in our present study, accounted for a substantial portion of total lice settlement variation—far higher than what was observed between strains. Our results could be interpreted to suggest that the potential for genetic improvement for sea lice resistance is limited if one assumes that salmon–louse interaction has been strongly driving natural selection since the last glacial recession (11,000 years ago). As this seemingly contradicts numerous genetic studies showing the potential for genetic selection [25,33,34,35,36], several other factors warrant further consideration. First, resistance to sea lice may not be a major evolutionary pressure in wild Atlantic salmon, at least in the Western Atlantic where these three salmon strains can be found. If sea-run Atlantic salmon survival were not influenced historically by sea lice presence, then there would be no evolutionary pressure to develop a more robust resistance to sea lice than their landlocked counterparts. Second, Atlantic salmon may place a greater reliance on alternative strategies than physical/chemical resistance to copepodid attachment such as behaviors aimed to avoid infestation with sea lice [52,53]. Third, the cost of resistance mechanisms existing at the time of the evolutionary split may not be sufficient to drive their loss in landlocked strains. It is worth noting that the same resistance mechanisms may be beneficial in landlocked strains for responding to other parasites. Fourth, the farming environment (net-pen) is clearly different from the natural environment, and so genetic effects on resistance to sea lice may be distinct between the farm/challenge and natural environments. Recent work has also questioned the role of group protection and whether current common sea lice challenge practices for assessing lice susceptibility accurately reflect genetic resistance at a group level of protection on farms [54]. Our present study confirms the need for additional research to assess the effectiveness of current genetic selection programs on production farms.

Our study found that lice preferentially settle on the fins of salmon, with the body and head next, in that preference order irrespective of the fish strain. This agrees with previous work that has also noted fin as a preferred settlement location [41,43]. Our results are within the ranges reported by Samsing, Solstorm, Oppedal, Solstorm, and Dempster [43] and slightly above the ranges reported for Atlantic salmon by Dawson, Pike, Houlihan, and McVicar [41]; however, these authors also reported that roughly 50% of the chalimus stages settled on the gills, which likely artificially suppressed (challenge artifact) the proportions on the body and fins. When looking at the louse transcriptome when attached to the body vs. the fins, there was no notable response difference between these two locations. This suggests that lice are not preferentially seeking out the fins to incur a life history benefit sufficient to change their transcriptional behavior and increase their chances of survival. This is consistent with the current hypothesis that lice settlement site is driven primarily by hydrodynamic forces [41,42,43]. It also indirectly suggests that Atlantic salmon rely on non-immunological anti-louse defense measures, as suggested by Bui et al. [51,52,53] and Odegard et al. 2024 [54], given that host transcriptomes were very different in tissues at these two settlement locations yet had no notable impact on sea lice post-attachment transcriptional behavior. Indeed, this is where the host strain of Atlantic salmon appeared to create more of an effect, suggesting the louse may be somewhat passively opportunistic in “deciding” which host it attaches to and must behaviorally adapt to its environment once attached.

In conclusion, several experimental factors were important in the current study. The greatest amount of variation observed was the average tank-to-tank infestation level. The level of challenge-to-challenge variation is in line with other sea lice infestation challenges conducted at our facility (unpublished data) and underscores the importance of conducting multiple replicated sea lice challenges. Modeling lice infestation and maintaining tanks/challenge as a model factor showed a potential modest strain effect, with CAS strain fish being slightly more susceptible to infestation in one of four challenges compared to the other two strains. Additionally, a significant correlation was noted between fish size and LC in all four challenges. While the fish in the study were age matched, they were not size matched, especially considering that SJR fish have been under intensive selection pressure for fast growth [36]. Thus, normalizing fish weight either through LD₁ or LD₂ allows for an improved interpretation of lice settlement data compared to LC. This is true for comparing between studies as well, and reporting individual LC and fish BW, or data normalized to LD, would facilitate stronger inter-study comparisons of results.