Infectious agents and their physiological correlates in early marine Chinook salmon (Oncorhynchus tshawytscha)

Physiological parameters of early marine Chinook salmon, measured via gene expression and plasma chemistry, were correlated with the load of infectious agents confirmed in the host through in situ hybridization. Agents demonstrating the most associations with deviation from host homeostasis included Piscine orthoreovirus, Candidatus Branchiomonas cysticola, Parvicapsula pseudobranchicola and Parvicapsula minibicornis.


Introduction
Infectious agents, including bacteria, parasites and viruses, are found in aquatic ecosystems worldwide, and impact both wild and cultured organisms.While diseased fish and associated mortalities are easily detected in an aquaculture setting, the impact of infectious diseases in wild fish populations is difficult to observe.Diseased organisms are more vulnerable to predation, making it difficult to sample individuals at a late stage of disease (Bakke and Harris, 1998;Miller et al., 2014;Furey et al., 2021).For this reason, and because fish can be easily accessed and a multitude of factors controlled in a closed system, much of our knowledge of the histopathological and physiological impacts of infectious agents comes from laboratory or aquaculture-based experiments.This is certainly the case for Pacific salmon, a group of species in which dozens of infectious agents have been documented.
Pacific salmon (Oncorhynchus spp.) are iconic species on the west coast of North America due to their ecological, cultural, economic and recreational importance.The productivity of Pacific salmon in the North Pacific Ocean is characterized by high interannual variability but steady declines in some species and populations have been evident for decades, especially towards the southern distribution of the genus (Riddell et al., 2013;Welch et al., 2021).In British Columbia, marine survival of Chinook salmon (Oncorhynchus tshawytscha) has declined over the past 20 years with smolt-to-adult returns for many populations now often falling below 1% (Welch et al., 2021).The Committee on the Status of Endangered Wildlife in Canada recently designated 12 of 26 southern British Columbia Chinook populations as endangered (COSEWIC, 2019).Although the reasons behind the decline in Chinook abundance are still in question, predation (Nelson et al., 2019), shifting marine conditions (Welch et al., 2021), global climate change (Crozier et al., 2021) and infectious disease (Miller et al., 2014) are suspected to contribute to the decline.
The early marine life history of Pacific salmon is a critical period where salmon are thought to have very low survival (Beamish and Mahnken, 2001;Cunningham et al., 2018;Claiborne et al., 2020).Infectious agents are considered to have sporadic impacts because of their ability to reproduce rapidly and to influence the host as population regulators and selective agents (Bakke and Harris, 1998).A recent study found that some infectious agents detected in fish during early marine residence were negatively associated with population-level survival and body condition for Chinook salmon (Bass et al., 2022).However, this study was correlative and included no assessment of disease state of individual fish.While providing evidence of causal relationships between pathogens and survival in wild populations is nearly impossible, physiological associations with agents could provide another layer of correlative evidence.These physiological associations could be established by utilizing well-established assessments, including blood metabolites and ions and histopathology, combined with modern molecular techniques.
The advancement of molecular technology has created more possibilities for studying infectious agents in wild populations.High-throughput quantitative polymerase chain reaction (HT-qPCR) enables rapid and highly sensitive screening of infectious agents across many samples (Miller et al., 2014(Miller et al., , 2016)).On the same HT-qPCR platform, host gene expression can be simultaneously profiled through the inclusion of assays targeting aspects of host health, such as immune status, stress level and osmoregulation (Jeffries et al., 2014;Teffer et al., 2017;Houde et al., 2019;Deeg et al., 2022).Such curated panels of genes can also be used to identify patterns indicating the host's response to viral disease (Miller et al., 2017).Molecular technology, combined with traditional methods of studying fish physiology, such as blood chemistry and histopathology, is likely to broaden our understanding of infectious agents carried by wild fish and the relationship between infectious agents and early marine mortality.
In this study, we assessed associations between pathogen infection intensity and the physiology of wild juvenile Chinook salmon during their first year of marine residence.We compared infectious agent presence and loads to fish physiological conditions, including host transcription profiles, blood plasma metabolic and osmoregulatory variables (e.g.glucose, lactate, chloride, sodium and osmolality), and evidence of cellular damage through histopathology.We applied in situ analysis to co-localize specific infectious agents within the regions where cellular damage occurred to strengthen the observed relationships between infection and disease.We examined juvenile Chinook salmon from a broad geographic area, encompassing the waters surrounding Vancouver Island and the Salish Sea.This research is unique as we investigated associations between pathogens and three layers of physiological information: molecular (host gene expression), metabolic and ionic (blood plasma chemistry) and cellular (histopathology) of juvenile Chinook salmon in their early marine residence.

Methods overview
Juvenile Chinook salmon in their first marine year were collected by trawl along the British Columbia coast.Blood and multiple organs were collected from each fish.From blood samples, metabolic and ionic variables were assessed.A subsample of the organs was used for histopathological analysis including both hematoxylin and eosin staining and in situ hybridization.Six organs were pooled and screened for 46 pathogens by molecular analysis.Gill and liver tissue were separately prepared to measure the expression of host genes from several functional groups.For each pathogen, loads were compared to blood parameters and gene expression in linear mixed models that accounted for additional variation  imposed by sampling season, location and life history.Model results were visualized in a heatmap so that pathogens with multiple associations across blood variables or gene groups, or those with particularly strong associations, could be identified.

Fish, blood and tissue collection
Samples were obtained from Fisheries and Oceans Canada (DFO) research sampling programs along the southern coast of BC (Figure 1).All samples were obtained from the winter, summer and fall of 2012, 2013 and 2014.The sampling methods are described in Tucker et al. (2018).Briefly, fish were captured by midwater trawl (DFO marine sampling vessels) for 15 to 30 min at 5 knots and brought onboard.Juvenile Chinook salmon were haphazardly selected, and length (mm) and mass (g) were measured.Fish were dissected within 30 minutes of capture to optimize tissue quality for gene expression profiling.To ensure that only juvenile Chinook salmon were collected, seasonal size limits were applied as follows: May to August < 300 mm, October to November < 350 mm, February to March < 400 mm.Fin clip samples were collected from individuals using sterile scissors and preserved in 95% ethanol for genetic stock population identification (GSI) in the lab (Beacham et al., 2006).Whether an individual represented the subyearling or yearling life history at ocean entry was determined by the dominant life history in the natal population as determined by GSI.
Blood was collected from each individual using a sterile 26 gauge needle on a 1.0 mL syringe within 30 minutes of capture.Needles and syringes were flushed with heparin solution prior to blood extraction from the caudal peduncle.The collected blood samples were centrifuged at 6900 G for 5 minutes to isolate plasma for the measurement of physiological parameters in the laboratory, and then immediately stored at −80 • C. Tissue biopsies were taken from the brain, kidney, liver and heart.Two separate sets of sterile dissection tools were used for exterior (gill) and interior tissues.Samples used for infectious agent detection and host gene expression profiling were collected first.Whole or half brain and tissue pieces of each of the other organs were individually preserved in vials of RNAlater (Qiagen, MD, USA), kept for 24 hours at 4 • C and then frozen at −80 • C or on dry ice.Additionally, tissue samples from gills, skeletal muscle, heart, liver, spleen, kidney, pancreas and brain (half) were also collected into histology cassettes and preserved in 10% buffered formalin.To maintain constant size of tissue samples, in the case of small specimens, the whole organ may be included (e.g.spleen); in the case of larger specimens, each tissue was then subsampled to a size not thicker than 4 mm in at least one of the three dimensions to allow formalin to penetrate and fix the tissues quickly.Nanaimo, BC, Canada).High-throughput quantitative PCR (HT-qPCR) using assays with TaqMan probes was run on the Fluidigm BioMark™ HT microfluidics platform (Fluidigm, CA, USA) to test for the presence and quantity of infectious agents and the expression of host genes.This technology has been adopted for salmon research (Miller et al., 2016) and used in several studies of Pacific salmon featuring both juvenile and adult fish (Jeffries et al., 2014;Teffer et al., 2017;Deeg et al., 2022).The platform performs independent PCR reactions for each of 96 samples against each of 96 assays for a total of 9216 reactions.The specificity, sensitivity and repeatability of the platform have been validated for use in salmon infectious agent detection and quantification (Miller et al., 2016).

Infectious
BioMark™ dynamic arrays were run separately for infectious agents and host genes to maximize the number of agents and host genes surveyed.For infectious agent detection, 47 assays to 46 infectious agents (two assays to infectious salmon anemia virus) and one reference gene were selected to run in duplicate on each dynamic array.For infectious agent monitoring, each dynamic array contained combined DNA and cDNA from pooled samples from the brain, gill, kidney, liver and heart, positive and negative processing controls and six standard serial dilutions of combined artificial positive constructs (clones of DNA sequences corresponding to all infectious agent assays).Combining cDNA and DNA allows for detection of both RNA and DNA viruses, as well as detection of pathogens in both active and dormant states.For host gene expression, only a subset of gill samples (n = 218) and a subset of liver samples (n = 263) were available for analyses.cDNA from these single tissue samples was run separately on different dynamic arrays.At least one positive and one negative processing control and six standard serial dilutions (made by pooling host cDNA using 1 μL from every sample) were also allocated on every chip.Eighty-nine host gene assays were selected to run as singletons based on their known contributions to immune response, general stress response, osmolality, thermal and hypoxia stress, along with three reference genes (Table 1; Miller et al., 2011Miller et al., , 2016;;Akbarzadeh et al., 2018).The selection included a set of genes identified previously as a "mortality-related signature" that has been predictive of wild adult salmon migration and spawning failure (Miller et al., 2011), as well as speed of migration into freshwater (Drenner et al., 2018).A panel of viral disease development (VDD) host genes was also included, which when co-expressed, can distinguish fish in a viral disease state from a carrier or no virus state (Miller et al., 2017).The VDD biomarker panel was developed using challenge studies and farm outbreak samples from multiple viral agents and salmon species.Not all assayed genes were included in subsequent analyses, including those which had poor assay efficiency and those that were part of a thermal stress panel which we decided was irrelevant for correlations with pathogen load (Table 1; sequences provided in Supplementary Material, Table S2).
Laboratory procedures for nucleic acid preparation and qPCR protocol are described in (Miller et al., 2014) and (Miller et al., 2016), and the same process has been applied in several recent studies (Jeffries et al., 2014;Teffer et al., 2017;Tucker et al., 2018;Deeg et al., 2022).In short, tissues from each sample were individually homogenized.For infectious agent detection, the aqueous phase of multiple tissues from the same fish was pooled to extract RNA and the organic layer pooled to extract DNA.For host gene profiling, aqueous samples for gill and liver tissue homogenates were used separately for RNA extraction.Extracted DNA and RNA were assessed for purity and normalized.cDNA was synthesized from normalized RNA, and for infectious agent monitoring, equal aliquots of cDNA and DNA were combined.cDNA alone was used for host transcriptome analyses.Because the individual wells on the microfluidics dynamic arrays for the BioMark platform are small volume (7 nL), samples must undergo an initial enrichment step involving amplification with all target assay primers combined according to the Fluidigm protocol (Miller et al., 2016).Prior to individual assay qPCR on the dynamic arrays, excess or unincorporated nucleotides and primers were removed, and samples were diluted 5-fold.Cycle threshold (Ct) was determined in the BioMark Real-Time PCR software.Amplification curves of all reactions between each assay and each sample were visually examined in case of any abnormal curve shape (in which case they could be manually scored as N/A or negative).Assay efficiencies were calculated based on a fitted curve from serial dilutions.Assays with efficiencies less than 80% or greater than 120% or coefficients of determination (R 2 ) of the fitted curve less than 0.98 were removed from subsequent analyses.Host gene expression of gill and liver samples were normalized with the 2 − Ct method (Livak and Schmittgen, 2001) in which relative expression of each gene was calculated using the reference genes and the pooled sample made with all available sametissue samples of the entire study included.

DNA natal population identification
Fin clip samples preserved in ethanol were used to assess the population of origin for individual fish following methods outlined in Beacham et al. (2006).Stock IDs were grouped into six geographical natal groups: West Coast of Vancouver Island, East Coast of Vancouver Island, Fraser River system (upper and lower Fraser River and Thompson River), Mainland BC (including streams in Northern, Central and Southern mainland BC that were not included in the other five region groups), the Columbia River system (including Columbia River and Snake River) and Washington (including tributaries to the Puget Sound and Strait of Juan de Fuca).Hereafter, they will be referred to as WCVI, ECVI, Fraser, Mainland, Columbia and Washington, respectively.In the analysis, individuals from an unknown natal population or with a probability of assigned natal population less than 0.50 were excluded.

Blood analyses
Blood samples were collected for a subset of juvenile Chinook salmon  Functional group represents inclusion into specific biomarker panels in our analysis; note that while many genes can be involved in multiple physiological pathways, they were only included in a single functional group for our analysis.Mortality related signature (MRS) is a curated panel discovered in association with migratory losses in salmon (Miller et al., 2011) and contains genes from multiple physiological pathways.Viral disease development (VDD) is a curated panel that when co-expressed indicates a viral disease state across multiple RNA viruses infecting salmon (Miller et al., 2017).Sequences for all assays provided in Supplementary Material, Table S2.

Symbol
Host Chloride concentration and osmolality were measured as the average of the duplicates using a Model 4 425 000 Haake Buchler digital chloridometer and the Advanced Instruments 3320 freezing point osmometer, respectively.If the disagreement between the duplicates was greater than 3 mmol/L for chloride or 3 mOsm/kg for osmolality, measurements were repeated and the average was taken from the two closest measurements.At a later date, transferred plasma samples were thawed again and were diluted at 1:100 dilution for sodium analysis by a BWB XP flame photometer.The photometer was calibrated against a four-point standard curve that was created using sodium standard solutions at every start-up or after a change was observed during checks performed every ten samples.If the difference between the two results was greater than 6 mmol/L, the measurements were repeated and the averages were calculated using the two closest results.
Because the capture process causes rapid changes to most of these plasma variables (Pickering et al., 1982) and blood was collected anywhere from 15 to 60 minutes following the initiation of a trawl tow, the measures we collected should be considered representative of fish experiencing a capture stressor.

Histology
Based on the work of Tucker et al. (2018), findings herein and other observations during the course of the research conducted under the Strategic Salmon Health Initiative, Ceratonova shasta, Parvicapsula minibicornis, Paranucleospora theridion, PRV, 'Candidatus Branchiomonas cysticola' and Ichthyophonus hoferi were chosen for histological analysis.Forty-four histology samples that tested PCRpositive (generally with loads exceeding 1000 copies per μg tissue) for at least one of these six infectious agents were used (Supplementary Material, Table S3).Samples were dehydrated through an ascending gradient of ethanol solutions.Samples were then embedded in paraffin wax, and consecutive serial sections were cut at 3.5 μm thickness.One section per sample was stained with standard hematoxylin and eosin (H&E) for histological examination.

Statistical analyses
All statistical analyses were performed in R statistical software, version 3.4.2(R Core Team, 2017).Infectious agent load was defined as the amount of infectious agent nucleic acids in a given sample.The infectious agent Ct values were first averaged between replicates.In the case where an infectious agent was not positive for both replicates, no detection was assigned.The infectious agent Ct values were then converted to copy numbers using the standard curve method (Larionov et al., 2005).For this study, we defined limit of detection (LOD) as a cycle threshold (Ct) number below which true positive results were expected 75% of the time for a given assay, based on reanalysis (unpublished) of Miller et al. (2016) data that included multiple independent serial dilutions.Due to the high sensitivity of the BioMark platform, this is a conservative estimate.Data exceeding the LOD for a given pathogen were replaced with zero.
For each individual, we calculated the total number of infectious agent taxa detected (total pathogen taxa).Because this metric does not take the load of infectious agents per individual into account, Relative Infectious Burden (RIB) (Bass et al., 2019) was also calculated.This metric is a composite score that divides the load of each agent detected in an individual by the highest load of that agent in the population and then sums the values for the individual across agents.For each infectious agent, prevalence was calculated as the percentage of positive detections within the entire study population.
To test for correlations between infectious agents (plus RIB and total pathogen taxa) and blood plasma variables, we employed linear mixed models individually for each pathogen and blood plasma variable combination (100 models).For a given model, the blood plasma variable was the response, pathogen load was the explanatory variable of interest and smolt age (subyearling vs yearling), season and capture region were included to account for variation in the plasma variable unrelated to pathogen load.Season and capture region were random intercepts.Capture regions were West Coast Vancouver Island, Northeast Vancouver Island and Southeast Vancouver Island (Figure 1).The plasma variable and pathogen load were both scaled by one standard deviation so that associations could be visually compared across pathogens and associations could be discussed in terms of standard deviation changes in both predictor and response.Pathogen load was transformed by adding 1 and log-transforming prior to scaling.This analysis and similar models for gene expression were only conducted for infectious agents with 1% prevalence or greater (18 pathogens), excluding Nanophyetus salmincola, which happened to have an insufficient number of positive fish for which samples were run for plasma and gene expression.Note that all models were also applied to RIB and total pathogen taxa.
A similar approach was used to test for correlations between infectious agent load and fish physiology at the molecular level in both gill and liver tissue.The response variable for gene expression was the first principal component (PC1) from separate principal component analyses for each of six functional gene groups (Table 1, Figure 2), resulting in 120 models per tissue (18 pathogens and 2 coinfection variables modeled against 6 functional gene groups).Most of these functional groups were derived from previously developed, curated biomarker panels (mortality-related signature; Miller et al., 2011); osmoregulation (Houde et al. 2019a;Akbarzadeh et al., 2018) and viral disease development (Miller et al., 2017).For most curated panels, the presence of a given physiological state is predicted by the co-upregulation of most or all of the genes on the panel.The remaining three groups were innate immune response, adaptive immune response and metabolism.Note that many genes can have roles in more than one physiological pathway/group, but in our analysis, they were only represented once.Separate PCAs were conducted for each tissue and the PC1 axis was flipped where necessary so that PC1 had similar gene loadings between gill and liver tissue for each functional group (Figure 2).The primary difference in model structure from the plasma analysis was that dynamic array ID number (Biomark PCR chip) was included as a random intercept to account for variance due to analytical (laboratory) artifacts.Because samples were non-randomly distributed across dynamic arrays, capture region was confounded by dynamic array ID and thus could not be included as a random intercept.For confirmation in the interpretation of associations between pathogens and PCs, the same model structure was applied to each gene in gill and liver individually (1560 models for gill, 1520 for liver; heatmaps provided in Supplementary Material, Fig. S5, S6).
Coefficients for pathogen loads were presented for plasma variables and gene expression in a heatmap using the pheatmap() function from the R package pheatmap (Kolde and Kolde, 2015).We allowed the pheatmap() function to cluster both pathogens and response variables (plasma variables and gene expression) using the default "complete linkage" clustering method.To reduce Type I errors, we adjusted all the p values using the false discovery rate (FDR) approach across all 323 model results composing the heatmap (Benjamini and Hochberg, 1995).For inference we prioritized pathogens with correlations that were significant following FDR P value adjustment and showed consistent patterns across response variables.However, pathogens with  multiple P values < 0.05 prior to FDR adjustment are also described.

Results
Twenty-seven of 46 assayed infectious agent taxa were detected at least once within the limit of detection among 319 juvenile Chinook salmon (Figure 3).Nineteen infectious agents had an overall prevalence greater than 1%, including two viruses, four bacteria and 13 parasites (Figure 3).'Ca.Branchiomonas cysticola' was the most prevalent and was found in 71.8% of the total sample, whereas viral hemorrhagic septicemia virus (VHSV) was the least prevalent, only detected in one individual (Figure 3).Across the entire study population, total pathogen taxa ranged from 0 to 7 pathogens per individual with a median of 3 and a mean of 2.97 (± SD, 1.37).

Model results
Eighteen pathogens, total pathogen number and RIB were included as the explanatory variable of interest in mixed models with plasma variables, PC1s for each gill gene group and PC1s for each liver gene group as the response variables.On the y-axis of the heatmap (Figure 4), physiological responses sometimes clustered in logical ways (e.g.plasma osmolality and chloride, immune gene groups) and for gene expression, often within tissues (i.e.liver and gill gene groups largely grouped separately, Figure 4).Viral disease development was the only gene group to cluster across tissues.Plasma glucose and gill osmoregulation genes clustered closely with many similarities across pathogens (Figure 4).This clustering arose partly  from a high number of negative associations for each of these physiological parameters.Osmoregulation in gill tissue, for example, was negatively associated with pathogen load for five pathogens and total pathogen taxa prior to FDR adjustment (Figure 4).In both tissues, mortality-related signature genes clustered among the immune gene groups (Figure 4).
For gene groupings, associations with pathogen load tended to be exclusive to gene expression in one tissue or another but not both, with the exception being the VDD genes and Piscine orthoreovirus (PRV; Figures 4, 5).Note that correlations between Neoparamoeba perurans and VDD genes appear very similar to those for PRV, but are reflective of co-infections with PRV (described below).'Candidatus Branchiomonas cysticola' and Parvicapsula pseudobranchicola demonstrated positive associations between gene expression and pathogen load across multiple gene groups in gill tissue but neutral associations in liver tissue (Figures 4, Supplementary Material, Fig. S5).Parvicapsula minibicornis had negative associations between gene expression and pathogen load in gill tissue but no strong patterns in liver tissue.Loma salmonae showed positive associations between liver gene expression and pathogen load and neutral expression in gill tissue (Figures 4, Supplementary Material, Fig. S5).Finally, 'Candidatus Syngnamydia salmonis' showed negative associations in liver tissue and neutral expression in gills (Figures 4, Supplementary Material, Fig. S5).

Piscine orthoreovirus
The first and second strongest associations between pathogen load and a response variable in the study were between PRV load and the viral disease development PC1 (VDD-PC1) in liver (β = 0.60, T 247.4 = 12.2, p < 0.001, n = 263) and gill (β = 0.44, T 281.4 = 9.4, p < 0.001, n = 292) tissue (Figures 4, 5).Viral disease development gene expression was well correlated between gill and liver tissue for PRVinfected fish (Figure 5C).Prior to FDR adjustment, PRV was also negatively associated with plasma lactate and gill metabolism and positively associated with adaptive immunity genes in liver tissue (Figures 4, Supplementary Material, Fig. S6).
While it appears that N. perurans load was associated with VDD-PC1 (Figure 4), this pattern was an artifact of the data.All of the six fish with positive detections of N. perurans were captured in Quatsino Sound on March 10 and 11, 2013 and were from the Marble River population.Four of these six fish tested positive for PRV (there were a total of nine PRV-positive fish in the study), with three of those with PRV copy numbers exceeding 1000.It is highly unlikely that N. perurans, an extracellular parasite, would trigger activation of VDD genes, which show no activation in the presence of extracellular bacteria or a parasite (Miller et al., 2017).

'Candidatus Branchiomonas cysticola'
'Ca.B. cysticola' was positively associated with PC1 for metabolism, adaptive immunity and innate immunity in gill tissue (Figure 4).Plasma lactate was also positively associated with 'Ca.B. cysticola' load.Prior to FDR adjustment, this bacterium was positively associated with mortality-related signature expression in gill tissue.

Parvicapsula pseudobranchicola
This marine-transmitted myxozoan had positive associations with gill gene expression in all the functional groups assayed excluding viral disease development.Following FDR adjustment, osmoregulation, metabolism, innate immunity and mortality-related signature were all significantly associated (Figure 4).Of the plasma variables, lactate, glucose and sodium were positively associated with P. pseudobranchicola prior to FDR adjustment.Parvicapsula pseudobranchicola and 'Ca.B. cysticola', both known gill pathogens, presented similar patterns in gill gene expression, which led to their clustering in Figure 4.

Parvicapsula minibicornis
Another Parvicapsula species, P. minibicornis, had a pattern of associations with plasma variables and gill gene expression that was complementary to the pattern of P. pseudobranchicola (Figure 4).There was a negative association across gill gene expression groups and metabolism, osmoregulation, innate immunity and mortality-related signature following FDR adjustment.Plasma glucose and sodium were negatively associated with P. minibicornis load and plasma osmolality was positively associated, all prior to FDR adjustment.

Loma salmonae
This marine microsporidian was positively associated with gene groups in liver tissue.Mortality-related signature genes were significantly positively associated following FDR adjustment and immune cell signaling and adaptive and innate immunity genes were positively associated prior to adjustment.ride and osmolality and negatively associated with plasma glucose prior to FDR adjustment (Figure 4).

Histology
Lesions were observed on host tissues consistent with damage and disease development associated with five of the six agents of interest (C.shasta, P. minibicornis, PRV, 'Ca.B. cysticola' and I. hoferi, but not P. theridion).The majority of lesions associated with these agents were on spleen and kidney, tissues we did not examine for gene expression (Supplementary Material, Table S3).Associations between lesions and the suspected causal agents were further supported by localization of the target agents near and/or in the lesions after applying ISH on the set of consecutive sections to those used for H&E staining (Supplementary Material, Table S3).No lesions consistent with disease development stemming from infection by Paranucleospora theridion were observed among the four individuals examined.Noteworthy findings of lesions consistent with each agent are described below with additional images provided in Supplementary Material, Figs.S1-S4.

Piscine orthoreovirus
Six of the eight fish examined contained at least one lesion consistent with PRV-related jaundice/anemia, with one containing mild, focal endocarditis in the spongy layer of the myocardium (Figure 6A, Supplementary Material, Table S3), one with renal tubular vacuolar degeneration leading to necrosis of kidney tubules (Figure 6C); two with hemosiderin (excess of hemoglobin byproduct indicative of red blood cell [RBC] lysis) and mild congestion and hyperplasia of the white pulp in the spleen (Figure 6D) and six with mild congestion and hyperplasia of the hemopoietic tissue in the kidney.One fish also presented with inflammation in the white skeletal muscle.Overall, inflammation in the spleen and hyperplasia of the hemopoietic tissue in the kidney was noted, along with some cases of mild necrosis in liver, spleen ellipsoids and renal tubules.
Using ISH, PRV was found in the host heart, posterior kidney, spleen, intestine and liver.In the heart, PRV was present in the cardiomyocytes involved in the focal inflammatory lesions observed in the spongy layer of the ventricle  (Figure 6B).PRV was widely distributed in the red blood cells and macrophages of the spleen, where blood congestion and hemosiderin deposits were also observed (Supplementary Material, Fig. S1B).Red blood cells and macrophages in the posterior kidney were also heavily infected by the virus, and PRV was observed in association with a few necrotic renal tubules (Supplementary Material, Fig. S1A).In the intestine and liver, PRV was also found in the enterocytes and hepatocytes, respectively (Supplementary Material,Fig S1C and S1D).

'Candidatus Branchiomonas cysticola'
Two of seven samples tested showed the presence of epitheliocystis in the gills' secondary lamellae (Figure 7A), but only one of them also showed mild inflammation in the gills.By ISH, 'C.B. cysticola' was found inside the epitheliocystis observed in the gills, with some clusters of the bacteria where the cysts were developing (Figure 7B).

Parvicapsula minibicornis
Two of 14 fish examined had mild to moderate glomerulonephritis with mild necrosis of the renal tubules (Figure 8A), hypertrophy/hyperplasia of Bowman's capsule (Figure 8C) and generalized interstitial hyperplasia in the kidney (Figure 8 A-C).Parvicapsula minibicornis was found by ISH in both glomeruli and the lumen of renal tubules (Figure 8D).

Ichthyophonus hoferi
Two of five samples showed fungal cysts in the endocardium, sometimes associated with nodules of granulomatous inflammation (Supplementary Material, Fig. S2A).I. hoferi was clearly identified in the center of the granulomatous lesions in every organ infected, although the inflammatory reaction was not always associated with the fungal cysts (Supplementary Material, Fig. S2B).

Ceratonova shasta
One of six fish examined for C. shasta had moderate chronic enteritis (Supplementary Material, Fig. S3A), and one showed mild inflammation and proliferation of the epithelial cells of the gills from the developing stage of C. shasta (Supplementary Material, Fig. S4B).Ceratonova shasta was detected in the host lamina propria of the intestine (Supplementary Material, Fig. S3B), while some immature forms were detected in the host gill tissue (Supplementary Material, Fig. S4C and  S4D).

Discussion
Identifying physiological correlates of infection intensity in wild salmon is a means of identifying pathogens with the potential to impact early marine survival.Ours is the first study to combine transcriptional, metabolic and ionic and   histopathological data to examine infectious agents carried by wild salmon.For three infectious agents recently discovered to be relevant to salmon health, including 'Ca.B. cysticola' (Gjessing et al., 2021), P. pseudobranchicola (Karlsbakk et al., 2002;Nylund et al., 2005) and PRV (Di Cicco et al., 2018;Polinski et al., 2021), this is the first study to report associations with physiological condition in wild fish.We provided the first evidence of the potential impacts of PRV on both host gene expression and tissue pathology in wild juvenile Chinook salmon, findings that were highly consistent with observations in cultured fish of the same species (Di Cicco et al., 2018).We also provide the first evidence that two agents known to cause disease in freshwater, C. shasta and P. minibicornis, are also associated with pathology in fish caught in the marine environment, some sampled after several months at sea.Typically, studies investigating correlations between physiology and infectious agents are conducted under controlled conditions in laboratories.However, many of the agents we investigated have not been cultured previously, and to conduct such a controlled experiment across so many agents at once would require vast resources.The value of a study such as ours is the ability to test for physiological associations with levels of infection ecologically relevant to wild salmon across many agents at once.Results from such a study can prioritize pathogens that may warrant further investigation using controlled experiments capable of causal inference.Some negatives of our approach are that acute infections are difficult to detect since highly impacted fish are rarely encountered (Bakke and Harris, 1998), and there are many non-pathogen sources of variation that confound potential associations.Furthermore, while co-infections are likely playing an important role in physiological responses (Kotob et al., 2017), we are limited in our ability to explore them.Nonetheless, by looking across dozens of genes, five plasma variables and histopathology, we were able to consider the weight of evidence when determining which pathogens may warrant further study.
Of all the plasma variables and gene expression groups that we considered, none had more associations across pathogens than the gill osmoregulation genes.PC1 for gill osmoregulation was negatively associated with five pathogens and total pathogen taxa and strongly positively associated with P. pseudobranchicola load.A closer investigation of the genes involved (Supplementary Material, Fig. S5) revealed that genes coding components of sodium potassium adenosine triphosphate (Na,K-ATPase) in gill were those frequently negatively associated with pathogen load (although not for P. minibicornis).Such a relationship has been observed previously where Amoebic Gill Disease was associated with a decrease in Na,K-ATPase, likely due to the reduction in chloride cells caused by the host response to infection (Nowak et al., 2013;Chang et al., 2019).It is possible that pathogens with negative associations with expression of Na,K-ATPase isoforms in our study cause similar damage, however we conducted histopathology on only a subset of these (C.shasta did show some evidence of gill lesions and also a negative association with gill osmoregulation).
Plasma glucose, the only plasma variable not to show considerable deviations from literature baseline values (Supplementary Material, Table S1), shared a similar pattern of correlations with gill osmoregulation.Parvicapsula minibicornis and C. shasta infections were negatively associated with plasma glucose levels in our study.In addition, both RIB and total pathogen taxa were negatively associated with glucose.Starvation can lead to depressed plasma glucose levels (Furné et al., 2012), and a commonly observed clinical sign of disease for many pathogens is a reduction or cessation of feeding (Conte, 2004), sometimes accompanied by anorexia (Nowlan et al., 2020).For example, Mesa et al (Mesa et al., 2000) found that Chinook salmon experimentally infected with Renibacterium salmoninarum exhibited reduced plasma glucose levels, as well as lower mass, relative to controls.Parvicapsula pseudobranchicola was the only pathogen demonstrating a positive association with plasma glucose.In a recent study (Bass et al., 2022), P. pseudobranchicola had a strong, positive association with mass at length for Chinook salmon, suggesting that highly infected fish were above average weight for their length.One interpretation of the gill osmoregulation and plasma glucose results in combination is that, for some pathogens, highly infected fish are experiencing impacts to ionoregulatory mechanisms in the gill and as a result (or concurrently) are less capable of foraging.We note that this pattern occurred for the total pathogen taxa metric, one of our variables that was representative of coinfection.This variable has been previously associated with decreased populationlevel survival and elevated predation (Miller et al., 2014;Furey et al., 2021;Bass et al., 2022).
For almost every pathogen with significant physiological associations, correlations between pathogen load and gene expression differed dramatically between gill and liver tissue, leading to separate clusters for liver and gill gene groups (Figure 4).The notable exception to this rule was the strong upregulation of VDD genes in both gill and liver tissue associated with PRV load, but this is unsurprising given how organisms systemically respond to viruses (Miller et al., 2017).We suspect that this dichotomy in gene expression between tissues is a result of tissue tropism, for instance, with known gill parasites 'Ca.B. cysticola' and P. pseudobranchicola (primarily a parasite of the pseudobranch but also infecting and damaging gills (Nylund et al., 2018)) showing strong upregulation of gill immune gene expression.The diverse mechanisms through which pathogens interact with host physiology likely resulted in a complex array of responses in our study, limiting our ability to generalize host response across pathogens (Holzer et al., 2021).Furthermore, by limiting our molecular tests to gill and liver tissue, we could be missing important associations between pathogens and physiological responses, for example in the case of Myxobolus arcticus which targets brain tissue or C. shasta which targets the intestine (although histology did reveal that C. shasta was the only pathogen tested that showed pathology in the gastrointestinal tract).
We observed few strong associations between blood plasma variables and pathogen loads, which could be due to the fact that many of the plasma variables we measured are highly sensitive to the acute stress response which is triggered by capture and handling (Donaldson et al., 2014).The time from the initiation of the capture experience to blood sampling ranged from 15 to 60 min, and throughout this range, some increase in several of the plasma variables will occur (e.g.lactate).As a result, abnormal concentrations of any of these plasma variables caused by infectious agent pathology would likely be overwhelmed by large deviations from homeostasis caused by the acute stress response and thus unlikely to be observed.The exception to this pattern may be plasma glucose, the elevation of which is a secondary response to acute stress that ensues around 2 hours after a stressor is applied (Pickering et al., 1982;Dick et al., 2018).Researchers conducting similar studies in the future should weigh the utility of plasma variables given their reactivity to capture stress and carefully select parameters featuring slower response times to acute stress.
Wild fish are rarely collected in a symptomatic disease state and thus histology is not routinely used for observational studies of wild fish such as ours.Because fish in an advanced disease state are unlikely to survive in the presence of predators, wild fish usually present with only mild to moderate lesions, making the observation of tissue modifications due to pathological processes very challenging to detect.Regardless, histopathological observations of a disease agent and associated tissue damage represent the ultimate evidence of the physiological effect of infectious agents in the host.By applying ISH in this study, which allowed us to localize specific agents in tissues, we were able to confirm the association between specific agents and corresponding lesions observed in the fish tested.

Piscine orthoreovirus
Despite only nine fish with detections, PRV demonstrated the strongest associations with gene expression in the study, consistent across both gill and liver tissues.PRV was recently identified as one of the pathogens most likely to be associated with decreases in survival and condition for Chinook and Coho salmon in British Columbia (Bass et al., 2022).Our study demonstrates that PRV occurs in wild Chinook salmon in their first marine year in British Columbia, and that these fish have gene expression patterns indicating a response to viral infection in multiple tissues, as well as cellular changes consistent with PRV-related jaundice/anemia.The relationship between PRV and the activation of the VDD genes was first observed in samples from a Chinook salmon farm outbreak of jaundice/anemia (aka jaundice syndrome) (Miller et al., 2017) and confirmed in a subsequent independent assessment of audit samples (Di Cicco et al., 2018).Importantly, Di Cicco et al. (2018) showed that the VDD response did not appear to activate in the host until the virus was observed in the extracellular space outside of its primary infective tissue, red blood cells (RBCs), likely due to a massive hemolysis.Unlike in Atlantic salmon, Chinook salmon infected with PRV could present anemia (Miller et al., 2017;Di Cicco et al., 2018) and show a threshold response between PRV load and VDD activation potentially suggesting a greater vulnerability in Chinook than Atlantic salmon to the PRV-1a variant in BC (Di Cicco et al., 2018).The innate antiviral response induced by salmonids in the presence of PRV was recently described by Dahle et al. (2022), who listed immune components that were upregulated in the presence of PRV infection, including CD4, CD8, IgMs, MHCI, MHCII, IRF, STAT1, IFNa and Mx, many of which we included as assays in this study (Supplementary Material, Fig. S5 and S6).For the majority of these genes we observed significant positive associations with PRV load, particularly in liver tissue (Supplementary Material, Fig. S6).Dahle et al. (2022) suggest that a strong innate immune response may be associated with pathological outcomes although there is currently no evidence that these immune mechanisms offer protection against PRVassociated pathological changes.
In our study, loads of PRV were overall lower compared with the farm audit Chinook salmon presented in Di Cicco et al. (2018).This finding was expected as the wild fish were sampled live, indicating an overall early phase of the infection, while the farm audit fish, in which high PRV load is usually observed, were all moribund or dead fish.However, in the present case, the VDD signal was still strongly associated with PRV load in liver and gill.Di Cicco et al. (2018) showed that milder lesions associated with earlier stages of the development of jaundice/anemia disease are present in fish that did not yet contain the clinical signs of jaundice (external yellowing) or anemia (pale gills), but only in fish classified as VDD+.Similar lesions occurring in the presence of PRV, affecting primarily the heart (mild endo/myocarditis), the spleen (congestions and accumulation of hemosiderin) and the kidney (hemopoietic tissue hyperplasia and renal tubule necrosis), were observed in the juvenile chinook included in the present study.We therefore suspect that these fish were in an early stage of development of jaundice/anemia.Whether wild Chinook salmon with a late-stage disease would survive long enough to be sampled, and if they are physiologically compromised at early stages of disease development, are certainly questions worth pursuing in the future.
Importantly, all of the Chinook with high loads of PRV showing signs of disease in this study were sampled in the cool fall/winter period, the same temporal time that jaundice/anemia occurs in Chinook salmon on farms (Di Cicco et al., 2018) and Heart and Skeletal Muscle Inflammation occurs in Atlantic salmon (Di Cicco et al., 2017).While two PRV challenge studies have been carried out in Chinook salmon from the Pacific Northwest, both studies were not carried out in temperatures typically experienced overwinter and focused on recapitulation of clinical signs of disease (mortality, jaundice) rather than recapitulation of the pathological lesions that lead to clinical manifestations (Garver et al., 2016;Purcell et al., 2020).Given that most PRV challenge studies worldwide have successfully demonstrated the pathology but failed to induce disease powerful enough to induce clinical signs, including mortality, the failure of these studies to demonstrate a cause-and-effect relationship with "disease" was unsurprising.Future studies need to feature low-temperature trials and focus on pathological outcomes, as described in Di Cicco et al. (2018).

'Candidatus Branchiomonas cysticola'
The bacterium 'Candidatus Branchiomonas cysticola' is commonly found in gill epitheliocysts in farmed Atlantic salmon (Toenshoff et al., 2012;Mitchell et al., 2013) and was recently identified as a major contributor to inflammatory gill disease in that species (Gjessing et al., 2021).However, some have suggested that, as this bacterium is a member of the fish gill microbiota in healthy fish, it may not be pathogenic (Gunnarsson et al., 2017).Mitchell et al. (2013) posited that, although 'Ca.B. cysticola' loads in Atlantic salmon were positively associated with the severity of gill inflammation, the negative impacts may be load-dependent or more likely to cause disease when additional agents are presentthus providing two explanations for why many infections of this high-prevalence agent may not be impactful to the host.
Consistent with the theory that this bacterium may impact gills, we observed several positive associations between immune gene groups and 'Ca.B. cysticola' load in gill tissue but no associations in liver tissue.Specifically, 'Ca.B. cysticola' load was positively associated with the expression of innate and adaptive immunity as well as immune cell signaling, with a strong activation of inflammatory genes including CD8, MMP25, MMP13, IL-1b, IL-8 and SAA (Supplementary Material, Fig. S5).We also observed a positive association between 'Ca.B. cysticola' load and plasma lactate, potentially a result of earlier onset of anaerobic metabolism during capture due to decreased gas exchange by inflamed gill tissue.Finally, in situ hybridization localized 'Ca.B. cysticola' to epitheliocysts in two Chinook salmon.These results indicate that, similarly to Atlantic salmon, 'Ca.B. cysticola' is found in gill epitheliocysts of Chinook salmon and likely triggers a pro-inflammatory immune response in the gill tissue of this Pacific species which may result in impacted gill function.
Although in previous research 'Ca.B. cysticola' has been highly prevalent in out-migrating salmonid smolts (Healy et al., 2018;Stevenson et al., 2020) and adult Chinook salmon returning to spawn (Bass et al., 2017(Bass et al., , 2019)), no studies of this agent in wild fish have revealed associations with survival.By acquiring fish after they left the freshwater environment but before they matured, our study focused on a different life stage from these previous studies.Studying the same life stage, Tucker et al. (2018) found that declines in prevalence were coupled with load truncation, a combination of characteristics consistent with pathogen-mediated mortality.Bass et al. (2022) found that in the fall and winter, Chinook body mass was negatively associated with 'Ca.B. cysticola' load, indicating that as infection intensity increased, fish became more underweight (although the opposite was true in the spring and summer).The findings from the two aforementioned studies coupled with the physiological findings from our current study suggest that although 'Ca.B. cysticola' appears extremely common in Pacific salmon, further research is needed to determine whether under stressful environmental conditions, this bacterium may act cumulatively or synergistically to reductions in the health and survival of wild salmon.

Parvicapsula pseudobranchicola
Correlations between the load of this highly prevalent marine myxozoan and gene expression (and to a lesser extent, blood parameters) were very similar to those we described above for the bacterium 'Ca.B. cysticola'.In addition to associations in gill tissue genes suggestive of immune response and inflammation, there was a significant positive association between P. pseudobranchicola load and osmoregulatory gene expression (unique in the study).Plasma glucose, lactate and sodium also increased with P. pseudobranchicola load but not significantly after FDR correction.
Much of what is known regarding the biology of this pathogen comes from a single longitudinal study of Atlantic salmon held in aquaculture pens in Norway (Nylund et al., 2018).In that study, P. pseudobranchicola was 100% prevalent in gills 49 days after introduction of juvenile salmon to seawater and PCR-measured loads in gill were highly correlated with loads in pseudobranchs.Because previous work had not led us to anticipate the many correlations between P. pseudobranchicola load and gene expression, we did not conduct histopathology targeting this agent.However, the presence of many significant correlations between P. pseudobranchicola load and gene expression suggest that gill tissue, in comparison to liver, is the more likely location where related impacts occur.While Nylund et al. (2018) found that the primary tissue infected by P. pseudobranchicola in Atlantic salmon was the pseudobranch, the only other tissue where the parasite was localized using ISH was the gill.We did not sample pseudobranchs for any purpose in this study, but these organs are adjacent to the gills.Nylund et al. (2018) observed an increase in P. pseudobranchicola spores and normalized expression (via PCR) in Atlantic salmon pseudobranchs until around day 147 following seawater transfer, after which both of these measures decreased dramatically, likely due to the parasite rupturing from pseudobranch cells and dispersing into the surrounding environment.Such concentrated development and subsequent tissue damage is likely to trigger the upregulation of immune-related genes.Thus, if the trajectory of the pathology of the parasite is similar in Chinook salmon, the positive association between gill genes associated with immune response and inflammation that we observed might be expected.Using the same PCR platform used in our study, Deeg et al. (2022) observed positive associations between P. pseudobranchicola load and the expression of immuneand inflammation-related genes in the gill tissue of pink and sockeye salmon collected in the Gulf of Alaska.Given that P. pseudobranchicola was one of the agents with the most associations in our study, that it appeared important for Deeg et al. (2022), and that it had high consistency in associations with survival between Chinook and Coho salmon in a recent  (Bass et al., 2022), we suggest that this marine parasite should receive further detailed study in Pacific salmon.

Parvicapsula minibicornis
Parvicapsula minibicornis is a myxozoan parasite that targets the glomeruli of the kidney, thus compromising host osmoequilibrium and sometimes resulting in mortality for adult salmon (Bradford et al., 2010b).Accordingly, in adult sockeye salmon conducting freshwater migrations, plasma osmolality was negatively correlated with P. minibicornis abundance in kidney tubules (Bradford et al., 2010a).In our study, evidence of damage caused by P. minibicornis was observed in histopathology, including lesions in the kidney characteristic of P. minibicornis pathology and localization of the pathogen in these tissues via ISH.In gill tissue, another component of the osmoregulatory system, genes associated with osmoregulation were down-regulated.Perhaps as an outcome of these impacts on multiple aspects of the osmoregulatory system, we found a positive association (prior to FDR correction) between plasma osmolality and P. minibicornis load.This is the opposite result found by Bradford et al. (2010a), likely because in this case fish were collected in saltwater.To date, all studies linking P. minibicornis to disease have focussed on returning adult salmon in freshwater (e.g.Bradford et al., 2010).However, the co-localization of lesions in the host kidney with P. minibicornis and the stress-related signals we discovered in the host genes indicate the high likelihood that this parasite is capable of causing disease in juvenile salmon in the ocean.
Myxozoans are a broad group of multicellular parasites with a diversity of life history strategies and methods for evading host immune systems (Holzer et al., 2021).We observed a widespread downregulation of host gill genes involved with intracellular immunity and inflammation associated with increasing loads of P. minibicornis.Similar to our study, experimental myxozoan infections in gilthead sea bream and rainbow trout have resulted in broad downregulation of immune-related genes (Davey et al., 2011;Barrett and Bartholomew, 2021).In contrast, P. pseudobranchicola appeared to show patterns of gene expression complementary to those of P. minibicornis in nearly all cases.This opposite pattern of molecular response could be driven by pathogen tissue tropism.Of 14 fish receiving histopathological sampling targeting P. minibicornis in this study, one fish showed pathology in the gills while five showed pathology in kidney (Supplementary Material, Table S3).These results are consistent with those of Bradford et al. (2010b), who showed that adult sockeye in freshwater experienced higher P. minibicronis loads in kidney before gills.Thus, immune function in gills could be less of a priority for a resource limited host battling P. minibicornis infection in other tissues relative to a host contending with P. pseudobranchicola in gills and the adjacent pseudobranchs.Alternatively, it is possible that although these two myxozoans are classified in the same genus, they may have completely different approaches to evading the immune response of the host organism (Holzer et al., 2021).

Loma salmonae
This marine microsporidian was the only pathogen that demonstrated a stronger gene expression response in liver tissue compared to gill tissue.This result is unexpected as gill is the target tissue for the formation of cyst-like xenomas that eventually rupture and induce chronic inflammation and stress in the host.Additionally, liver tissue has not been demonstrated as a target tissue through molecular methods or histology (Speare et al., 2012).Nevertheless, we saw positive associations between L. salmonae load and several MRS genes in liver tissue.Components of adaptive immunity and immune cell signaling were also positively associated with L. salmonae load.Although long considered an organ for metabolism and energy storage, teleost livers do show some immune response to pathogens (Causey et al., 2018).

Ichthyophonus hoferi
I. hoferi is a mesomycetozoan parasite of over 100 species of fish across marine, brackish and freshwater habitats (McVicar, 2011).It was prevalent among returning Chinook salmon in the Yukon River and was suspected of causing pre-spawn mortality (Kocan et al., 2004).In our study, we observed cysts typical of the parasite, localized primarily in the heart, associated with a mild to moderate granulomatous inflammatory reaction in areas where the cysts were damaged.We observed an immune stimulation of the hemopoietic tissue of the kidney, but I. hoferi cysts and other immature stages of the parasite were not visible in the histology sections, and only very rare granulomata were observed through ISH in the kidney, with no other microscopic cysts or "free", immature stages of the agent present in the tissue.A weak but consistently positive upregulation across multiple gene groups in gill tissue was observed.

Ceratonova shasta
Ceratonova shasta is a myxozoan parasite of fish intestines, and it is commonly found in Chinook salmon in several large river systems from BC to California (Fujiwara et al., 2011).The similar life history shared by P. minibicornis and C. shasta (Bartholomew et al., 2006) and the fact that 82% of C. shasta positive fish were also positive for P. minibicornis probably partially explained the similar patterns between these two agents and gene expression (Figure 4).We found minimal associations between C. shasta and physiological disturbance at the molecular or protein level, but histology revealed moderate inflammatory lesions in the gastrointestinal system and in the tips of gill lamellae where developing C. shasta spores were forming.Osmoregulatory genes in gill appeared slightly down-regulated, perhaps due to gill lesions caused by the presence of C. shasta.they leave freshwater may experience physiological impacts in the ocean-a potential example of a pathogen-mediated carryover effect.Recent studies have indicated that C. shasta infections are associated with reduced weight of Chinook salmon during freshwater migrations (Mauduit et al., 2022) and soon after ocean entry (Bass et al., 2022).

Conclusions
Through the combination of transcriptional, metabolic and ionic and histopathological data, we identified several infectious agents with multiple lines of evidence suggesting physiological impact to wild Chinook salmon with increasing pathogen loads.Our results support the use of molecular methods to monitor the impact of infectious agents on wild populations, which can be applied alongside regular molecular monitoring of infectious agents among Pacific salmon.Once a more definitive relationship between infection, disease and survival is established, incorporation of infectious agents and host transcriptome may enhance the accuracy of models estimating survival.
We found that wild Chinook salmon with PRV demonstrated a molecular response to viral disease development and pathology consistent with jaundice/anemia in farmed Chinook salmon (Di Cicco et al., 2018) as well as similar diseases in Pacific salmon caused by other strains of PRV (reviewed in Di Cicco et al., 2018).These findings, coupled with the higher PRV infection rates for populations exposed to salmon farms (Mordecai et al. 2021) and evidence indicating population-level impacts (Bass et al., 2022), highlight the threat PRV may pose to wild Chinook salmon populations and the need for precaution in our interpretation of risks related to PRV.To conclusively determine whether PRV (or other pathogens of concern) impacts wild or hatchery salmon populations, well-designed paired-release studies featuring experimental manipulations that alter host resistance to infection (i.e.vaccines, which have not yet been developed to scale (Dahle et al., 2022)) are necessary (e.g.Vollset et al. (2018)).

Figure 1 :
Figure 1: Capture locations (points) of Chinook salmon sampled around Vancouver Island (center) by mid-water trawl from 2012 to 2014.Red rectangle in inset map indicates the extent of the main map.Point color and shape represents the capture regions assigned for statistical analysis (red circles = West Coast Vancouver Island, green triangles = Southeast VI, blue squares = Northeast VI).For each capture region, pie charts depict the population origin for the sampled fish (colors in legend correspond to pie charts only; ECVI = East Coast Vancouver Island, WCVI = West Coast Vancouver Island).

Figure 4 :
Figure 4: Heatmap of coefficients for each pathogen in models for blood plasma variables and the first principal components of functional gene groups in gill and liver tissue.Cell values represent the change in units of standard deviation for the plasma or gene variable associated with a standard deviation increase in log pathogen load (from mixed tissue).Two asterisks in a cell indicate an FDR-adjusted p value < 0.05 and a single asterisk indicates a p value < 0.05 prior to FDR adjustment.

Table 1 :
Assays for host genes and infectious agents used in HT-qPCR analyses on juvenile Chinook salmon (Oncorhynchus tshawytscha).
Teffer et al. (2017)odium osmolality, using methods identical toTeffer et al. (2017).The analyses were conducted at the DFO West Vancouver Laboratory, BC.Immediately after thawing the 1500 μL centrifuge tube, the plasma layer inside the tube was carefully transferred into a new 500 μL centrifuge tube by a single-use pipet.Plasma glucose and lactate concentrations were measured using a YSI 2300 Stat Plus lactate/glucose analyzer (Yellow Springs Instruments, OH, USA).