Differential infestation of juvenile Pacific salmon by parasitic sea lice in British Columbia, Canada

: Fraser River Pacific salmon have declined in recent decades possibly from parasitism by sea lice ( Caligus clemensi and Lepeophtheirus salmonis ). We describe the abundance of both louse species infesting co-migrating juvenile pink ( Oncorhynchus gorbuscha) , chum ( O. keta ), and sockeye ( O. nerka) salmon over five years in the Discovery Islands and Johnstone Strait, British Columbia. The generalist louse, C. clemensi, was 5-, 7-, and 39-times more abundant than the salmonid specialist, L. salmonis , on pink , chum, and sockeye salmon, respectively. Caligus 31 clemensi abundance was higher on pinks (0.45, 95% CI: 0.38-0.55) and sockeye (0.39 95% CI: 32 0.33-0.47) than chum. L. salmonis abundance was highest on pinks (0.09, 95% CI = 0.06-0.15). Caligus clemensi had higher abundances in Johnstone Strait than the Discovery Islands. These results suggest differences in host specialization and transmission dynamics between louse species. Because both lice infest farmed salmon, but only C. clemensi infests Pacific herring ( Clupea pallasii ), conservation science and management regarding lice and Fraser River salmon should further consider C. clemensi and transmission from farmed salmon and wild herring.

In this paper we compare L. salmonis and C. clemensi abundance from co-migrating 87 groups of juvenile pink, chum, and sockeye salmon in the Discovery Islands and Johnstone 88 Strait, BC, over five years of field surveys. We investigate possible sources of variation between 89 louse species in their specialization among Pacific salmon species by focusing on the relative 90 abundances of the two louse species on the three salmon species in our study. We also 91 characterize the dynamics of sea lice on wild salmon relative to other areas with salmon farming  Islands from the Strait of Georgia and the exit points from Johnstone Strait to Queen Charlotte 110 Strait (Fig. 1). We deployed the purse seine nets from open, 6-8 m twin-outboard research 111 vessels to capture heterospecific schools comprised of juvenile pink, chum, sockeye, chinook, 112 and coho salmon, along with Pacific Herring. Visual-survey transects of surface activity were 113 used to identify areas with juvenile salmon, with the purse-seine net only being deployed if 114 juvenile salmon were observed.

115
Once we deployed the seine, it was used to corral the fish beside the boat in a submerged 116 section of the bunt end of the net so that the captured fish remained in the water and had space to

139
To facilitate comparisons of louse abundance among the salmon species, we filtered our 140 data for collections in which we retained at least five individuals each of pink, chum, and 141 sockeye salmon. This was done to guarantee that we only included schools of fish with all three 142 species co-migrating together, and to ensure that no bias was introduced into our analysis by 143 under-representing a given species within and among collections. All fish in a collection were 144 retained for analysis, so no bias was introduced by filtering data within collections, which is the 145 level at which the comparisons were made. While a higher cut-off would have reduced species 146 under-representation even further, a five-fish cut-off struck the best balance because increasing it 147 any more would have drastically reduced the number of collections available to analyze (e.g., 148 using a ten-fish cut-off would have resulted in an overall sample size of 1,217 instead of 149 2,262).We specifically targeted these species with our field methodology, and therefore they 150 were by far the most commonly captured fishes in our collections; coho were captured often but 151 generally in low numbers and chinook were caught infrequently (Fig. S1). Our final dataset was   156 To investigate potential differences in sea-louse parasitism between sampling areas and 157 among pink, chum, and sockeye salmon, we fit a suite of generalized linear mixed-effects 158 models (GLMMs) with louse abundance per fish as the response variable. The models employed 159 a negative binomial (type II) error distribution with a logarithmic link function to account for 160 overdispersion in the parasite counts. The models involved six fixed effects: salmon species, 161 sampling year, sampling area (Discovery Islands or Johnstone Strait), and the two-way 162 interactions between these three predictors. In accordance with the hierarchical nature of our 163 data, every model included both the sampling area and year as fixed effects. We therefore fit ten 164 models for each louse species. All our models included a random effect on the intercept for  The models that received the most support from the data differed between the two louse 199 species (Tables 1 & 2, Table S1). The highest ranking model for C. clemensi included fixed non-zero weight candidate models, rather than using a delta-AIC threshold to denote which 212 models were considered.

213
Our model-averaged predictions for L. salmonis and C. clemensi were consistent with 214 observed abundances and showed obvious differences among salmon species, years, and 215 sampling areas ( Fig. 3 and 4). Caligus clemensi were more than five times as abundant as L. 216 salmonis, on average, and our mean predictions for C. clemensi were higher than L. salmonis for 217 every combination of salmon species, year, and sampling area. Generally, pink salmon had the 218 highest L. salmonis abundance of any salmon species (Fig. 4 and 5). Chum salmon harboured the fewest C. clemensi in both areas.

228
Most of the sockeye salmon in our study were from the Fraser River. Of the 673 sockeye 229 salmon that were genetically identified to stock, 89% were from the Fraser River, just over half 230 the fish originating from Chilko (26%), Lower Adams (12%), and Lower Shuswap (12%) stocks.

231
In total, 38 separate stocks were represented in our subsample of sockeye from 2015-2017 (Table   232 S2). Broughton Archipelago, where in nine of the ten years data were collected, juvenile chum 248 salmon showed higher louse abundances than juvenile pink salmon (Patanasatienkul et al. 2013). 249 Sockeye salmon also experienced the largest difference in parasite abundance between the two 250 louse species (Figs. 3 and 4). This result corroborates previous, more anecdotal reports that C.  could also influence survival rates of host fish. However, it is unclear from our current data how 300 environmental drivers interact with other relevant factor to shape infestation patterns as a whole.

301
If salinity and temperature were the only factors influencing infestation, we would expect 302 coherence of infestation patterns among salmon and louse species. The lack of this coherence 303 suggests a more complex relationship between the various drivers of infestation. Further work is 304 needed to gain a more complete understanding of this multi-host-parasite system not only as it 305 currently stands, but how further environmental change will alter its dynamics in the future.   Canada, but many populations are seeing declines, and are the focus of considerable 351 conservation concern. Fraser River sockeye, which represented almost 90% of the genetically- generalist parasites like C. clemensi -whose abundance on sockeye salmon was on average 39-356 fold higher than L. salmonis in our study -are of particular concern because their additional