Vol. 136: 157–162, 2019 DISEASES OF AQUATIC ORGANISMS Published online October 17 https://doi.org/10.3354/dao03399 Dis Aquat Org

NOTE

A geographic hot spot of infection in the southern Salish Sea, USA

P. K. Hershberger1,*, A. H. MacKenzie1, J. L. Gregg1, A. Lindquist2, T. Sandell2, M. L. Groner3,4, D. Lowry2

1US Geological Survey, Western Fisheries Research Center, Marrowstone Marine Field Station, Nordland, WA 98358, USA 2Washington Department of and Wildlife, Fish Management Division, Marine Fish Science Unit, Olympia, WA 98501, USA 3US Geological Survey, Western Fisheries Research Center, Seattle, WA 98115, USA 4Prince William Sound Science Center, Cordova, AK 99574, USA

ABSTRACT: The prevalence of Ichthyophonus infection in Pacific herring Clupea pallasii was spatially heterogeneous in the southern Salish Sea, Washington State, USA. Over the course of 13 mo, 2232 Pacific herring were sampled from 38 midwater trawls throughout the region. Fork length was positively correlated with Ichthyophonus infection at all sites. After controlling for the positive relationship between host size and Ichthyophonus infection, the probability of infection was approximately 6-fold higher in North Hood Canal than in Puget Sound and the northern Straits (12 vs. 2% predicted probability for a 100 mm fish and 30 vs. 7% predicted probability for a 180 mm fish). Temporal changes in Ichthyophonus infection probability were explained by sea- sonal differences in fish length, owing to Pacific herring life history and movement patterns. Rea- sons for the spatial heterogeneity remain uncertain but may be associated with density-dependent factors inherent to the boom−bust cycles that commonly occur in clupeid populations.

KEY WORDS: Ichthyophonus · Pacific herring · Salish Sea · Puget Sound · Hood Canal

Resale or republication not permitted without written consent of the publisher

1. INTRODUCTION sent a species complex (Gregg et al. 2016). To avoid further perpetuation of these ambiguities, the organ- Ichthyophonus hoferi occurs in freshwater, marine, ism should be referred to generically, as Ichthyo - and anadromous throughout the Northern phonus, until differences are clearly delineated by Hemisphere, where it is periodically associated with further phylogenetic studies. disease epizootics and population-level impacts Many basic characteristics of Ichthyophonus life (McVicar 2011). A lack of distinguishable morpho- history remain poorly described, owing largely to logical characteristics contributed to prior taxonomic challenges inherent in tracking a parasite in highly un certainties; however, recent molecular phylo - vagile marine hosts. One of the most pertinent in - genetic studies convincingly place the organism in formation gaps involves unknown transmission the eukaryotic clade Opisthokonta, within a class of routes that operate on clupeids and other largely organisms re ferred to as the or planktivorous marine fishes. Although the possibil- Ichthyosporea (syn.) (Mendoza et al. 2002). Taxo- ity of a planktivorous host has been hypothesized nomic inaccuracies continue to persist within the repeatedly (reviewed in McVicar 2011), the para- genus, and previous reports of I. hoferi likely repre- site has not yet been conclusively identified from

*Corresponding author: [email protected] © Inter-Research 2019 · www.int-res.com 158 Dis Aquat Org 136: 157–162, 2019

any intermediate or paratenic hosts. To date, ef - wards implementing this targeted field approach, the forts to understand possible Ichthyophonus trans- objective of this study was to identify whether local- mission routes in planktivorous fishes have mostly ized hot spots of elevated Ichthyophonus infection involved epidemiological associations between en- prevalence occur within Pacific herring Clupea pal- vironmental parameters and in fection prevalence lasii aggregations in the southern Salish Sea, Wash- (e.g. Kramer-Schadt et al. 2010, Hershberger et al. ington State, USA. 2016); however, recent investigations have applied reductionistic approaches with controlled labora- tory experiments (Hershberger et al. 2015). Unfor- 2. MATERIALS AND METHODS tunately, neither approach has been particularly informative for elucidating transmission dynamics Pacific herring were collected bimonthly (February in planktivorous marine fishes. 2016−February 2017) using a midwater trawl at each Focusing parasite surveillance efforts in localized of 12 randomized and indexed sites throughout the areas where Ichthyophonus infection pressures are southern Salish Sea (Table 1). Each bi monthly sam- highest may be a more effective approach towards pling block included the samples from a semi-contin- identifying its transmission routes. As a first step to- uous 3 wk research cruise. Target sample size from

Table 1. Pacific herring sample date, location, and biological data

Sampling Sampling site Geographic n Mean (SD) Mean (SD) Ichthyophonus date bin fork length (mm) mass (g) prevalence (%)

Feb 10, 2016 Dabob Bay Hood Canal 60 133 (9.5) 19.4 (6.16) 20 Feb 10, 2016 Dabob Bay Hood Canal 60 126 (8.3) 17.3 (3.24) 27 Feb 16, 2016 Dallas Bank Straits 59 114 (8.5) 12.6 (3.26) 1.7 Feb 17, 2016 South Lopez Island Straits 30 112 (7.6) 11.4 (3.15) 0 Feb 23, 2016 Strait of Georgia Straits 60 121 (12.1) 14.7 (4.82) 5 Apr 4, 2016 Dabob Bay Hood Canal 60 169 (11.9) 42.4 (9.53) 13 Apr 4, 2016 Squamish Harbor Hood Canal 60 131 (13.6) 19.9 (7.30) 15 Apr 4, 2016 Squamish Harbor Hood Canal 50 157 (16.5) 34.5 (9.07) 18 Apr 5, 2016 Saratoga Passage Puget Sound 60 132 (7.28) 18.4 (3.68) 1.7 Apr 6, 2016 Saratoga Passage Puget Sound 58 156 (16.8) 32.7 (11.4) 1.7 Apr 5, 2016 Oak Bay Puget Sound 62 143 (11.2) 24.5 (5.76) 11 Apr 13, 2016 Port Angeles Straits 60 126 (11.5) 18.0 (6.88) 0 Apr 18, 2016 Yukon Harbor Puget Sound 60 136 (10.8) 24.3 (6.30) 6.7 Apr 19, 2016 Colvos Passage Puget Sound 60 142 (8.98) 26.5 (4.91) 3.3 Apr 20, 2016 Drayton Pass Puget Sound 60 151 (10.0) 35.1 (6.76) 0 Jun 1, 2016 Drayton Pass Puget Sound 60 155 (10.0) 39.2 (9.16) 13 Jun 8, 2016 Saratoga Passage Puget Sound 60 172 (18.3) 50.0 (15.4) 6.7 Jun 13, 2016 Strait of Georgia Straits 60 126 (3.91) 19.9 (2.63) 12 Aug 24, 2016 Squamish Harbor Hood Canal 60 89.8 (6.61) 7.42 (1.90) 5.0 Aug 24, 2016 Dabob Bay Hood Canal 60 177 (7.29) 56.8 (6.13) 40 Aug 25, 2016 Saratoga Passage Puget Sound 60 150 (4.82) 39.8 (4.33) 5.0 Aug 26, 2016 S. Lopez Isl. Straits 60 90.7 (4.55) 7.44 (1.30) 1.7 Aug 27, 2016 President Channel Straits 60 89.2 (6.68) 7.03 (1.46) 0 Oct 4, 2016 President Channel Straits 60 102 (3.56) 9.78 (1.14) 1.7 Oct 5, 2016 Strait of Georgia Straits 60 173 (15.0) 57.7 (14.5) 5.0 Oct 11, 2016 Yukon Harbor Puget Sound 60 173 (6.90) 61.1 (8.98) 1.7 Oct 11, 2016 Colvos Passage Puget Sound 60 117 (5.51) 16.1 (2.44) 0 Oct 12, 2016 Saratoga Passage Puget Sound 59 130 (24.3) 25.7 (17.7) 3.4 Oct 12, 2016 Saratoga Passage Puget Sound 60 178 (11.2) 63.1 (11.9) 10 Oct 18, 2016 Oak Bay Puget Sound 59 106 (7.98) 10.4 (2.08) 15 Oct 19, 2016 Dabob Bay Hood Canal 58 179 (11.8) 61.9 (9.67) 55 Dec 7, 2016 Saratoga Passage Puget Sound 60 177 (10.6) 66.5 (12.9) 13 Dec 13, 2016 Dallas Bank Straits 60 99.6 (7.72) 8.19 (1.68) 1.7 Feb 7, 2017 Colvos Passage Puget Sound 60 130 (11.4) 21.1 (9.07) 1.7 Feb 8, 2017 Colvos Passage Puget Sound 60 126 (6.62) 17.9 (2.66) 3.3 Feb 21, 2017 Saratoga Passage Puget Sound 60 156 (21.9) 41.3 (16.1) 6.7 Feb 22, 2017 President Channel Straits 60 103 (3.97) 8.88 (1.15) 5.0 Feb 22, 2017 Strait of Georgia Straits 57 170 (11.4) 47.8 (10.8) 3.5 Hershberger et al.: Ichthyophonus in Salish Sea herring 159

each trawl was 60 Pacific herring, and Ichthyophonus with 5% fetal bovine serum, 100 IU ml−1 penicillin, infection prevalence was assessed from sites where 100 µg ml−1 streptomycin, and 100 µg m−1 genta - n ≥ 30 (Table 1). For statistical comparisons, sampling mycin. Heart explant cultures were incubated at sites were binned into 3 geographic sub-basins: 15°C and examined microscopically (40× magnifica- Hood Canal, the Straits (Strait of Juan de Fuca, San tion) after 7 and 14 d of incubation. Cultures were Juan Islands, and southern Strait of Georgia), and scored positive when Ichthyophonus life stages were Puget Sound (Fig. 1, Table 1). Overall, 2232 herring detected in the broth medium. were sampled during 38 sampling events. Ichthyophonus infection status was analyzed as a Pacific herring carcasses were transported on ice to function of fish length, sampling day, and sub-basin the laboratory, where they were processed within using mixed effects logistic regression models. Indi- 24 h of capture. Fork length (mm) and mass (g) were vidual fish samples were nested within their respec- measured, and Ichthyophonus infection status (posi- tive midwater trawl collections, which were treated tive or negative) was determined by explant culture as separate random effects. Length was included as a of heart tissue. A piece of heart tissue (≤0.5 cm3) was covariate in all models because Ichthyophonus pre - aseptically removed and placed in a 5 ml culture tube valence typically increases with host size (Marty et containing 3 ml of Tris-buffered (pH 7.8) Eagle’s al. 1998, Hershberger et al. 2002, 2016). Sampling Minimum Essential Medium (MEM), supplemented month was not expected to have a linear effect on infection, so Fourier terms (sine and cosine curves with a 366 d periodicity) were fit to the data and incorporated as a covariate in the analyses (Bhas - karan et al. 2013). Multiple models were run, testing all possible inde- pendent and interactive effects of the terms for fish length, day (Fourier terms), and sub-basin. Model selec- tion was informed using Akaike’s in - formation criteria corrected for small

sample sizes (AICc). Analyses were run in R v. 3.5.3 using the ‘lmerTest’ package (Kuznetsova et al. 2017) and the ‘AICcmodavg’ package (Mazerolle 2019). Confidence intervals around the predictors in the best-fitting model were bootstrapped using the ‘mer- Tools’ package.

3. RESULTS

The best predictors of Ichthyo - phonus infection prevalence in Pacific herring from the southern Salish Sea included fish length and sub-basin as additive terms. Fourier terms for sam- pling day were not included in the best-fitting model, which had the low-

est AICc value (Tables 2 & 3), although they did occur in the next best-fitting

models ( AICc from best-fitting model = 3). Prevalence of Ichthyo phonus in - Fig. 1. Southern Salish Sea, Washington State, USA, depicting the 3 sub- fection increased with fork length basins and individual Pacific herring sampling locations across all 3 sub-basins and was signif- 160 Dis Aquat Org 136: 157–162, 2019

Table 2. Top models to predict the probability that a Pacific icantly higher in Hood Canal relative to Puget Sound herring has an Ichthyophonus infection. The best-fitting and the Straits (p < 0.001; Table 3). The probability of model has the lowest AICc. Length: herring fork length infection was approximately 6× higher in North Hood (mm); region: geographic region; time (Fourier): harmonic terms fit to the model to account for seasonal variation in Canal than in Puget Sound and the northern Straits infection. All models included a random effect of sampling for any given size class (e.g. 12 vs. 2% predicted event probability for a 100 mm fish and 30 vs. 7% predicted probability for a 180 mm fish; Fig. 2).

Model AICc

Length + Region 1143.5 4. DISCUSSION Length + Region + Time (Fourier) 1146.5 Length × Region 1147.3 Length + Region × Time (Fourier) 1148.2 High infection prevalence in Hood Canal, particu- Length × Region + Time (Fourier) 1150.4 larly among larger fish, indicates that an Ichthyo- phonus hot spot can occur within a relatively narrow geographical area, even in a highly vagile species like Table 3. Best-fitting mixed effect logistic regression model Pacific herring. A geographic hot spot was previously to predict the probability that a Pacific herring has an Ich - documented in Atlantic herring Clupea harengus be- thyophonus infection. The baseline model is for Hood Canal. A random effect for sampling event was included in this tween the Norwegian and the Barents Seas (Kramer- model Schadt et al. 2010). It was hypothesized that the At- lantic hot spot resulted from the accumulation of Estimate SE z p heavily infected fish that were unable to continue their migration to the preferred feeding grounds, Intercept −3.47 0.71 −4.8 >0.00001 thereby producing a density-dependent infectious Length 0.015 0.004 3.3 >0.001 belt of high parasite transmission. It is unlikely that an Region: Puget Sound −1.71 0.34 −5.1 >0.00001 analogous process accounted for the Hood Canal hot Region: Straits −1.98 0.41 −4.8 >0.00001 spot because, unlike in the Atlantic, there was no indi- cation from gross disease signs that 0.8 the Pacific herring demonstrated heavy 0.6 infections. As with the Norwegian ex- ample, inter-annual stability of the Hood 0.4 Canal hot spot remains unknown. 0.2 Straits However, some support for the tempo- 0.0 ral stability of the Hood Canal hot spot is indicated by previous Ichthyo phonus 0.8 surveys, where the total infection pre- 0.6 valence — not normalized for herring 0.4 size or age — was generally higher in adult herring from Hood Canal (e.g. 0.2 Hood Canal

infection prevalence (circles) 69% in April 2005, 73% in June 2005, 0.0 and 57% in May 2013) than from other 0.8 locations throughout the Salish Sea 0.6 and eastern North Pacific (Hersh- berger et al. 2016). Further evidence Predicted probability of infection (lines) 0.4 for the temporal stability of the Ich thy- Ichthyophonus 0.2 o phonus hot spot was provided by the Puget Sound 0.0 best-fitting model, which provided no indication that the probability of infec- 100140 180 tion changed throughout the sampling Herring fork length (mm) year within any of the 3 sub-basins or Fig. 2. Icthyophonus infection prevalence as a function of Pacific herring throughout the entire southern Salish length in each geographic region of the southern Salish Sea. Points: mean prevalence from a unique sampling event; lines and shading: mean and 95% Sea region. predicted interval which were bootstrapped from the best-fitting model Determining the cause(s) of the Ich - (shown in Table 3) thyophonus hot spot in Hood Canal Hershberger et al.: Ichthyophonus in Salish Sea herring 161

was beyond the scope of this study; however, insights sistently demonstrate high infection pressures. We into possible density-dependent factors were pro- recommend that future field investigations into Ich - vided by several epidemiological observations. Peri- thyophonus transmission mechanisms include Hood ods of high Ichthyophonus infection prevalence often Canal and similar geographic hot spots and consider coincide with periods of high clupeid biomass; for host size and sampling time. example, elevated Ichthyophonus infection preva- lence corresponded with high biomass of American Acknowledgements. This study was performed in collabora- shad Alosa sapidissima on the Columbia River (Her- tion with a trawl surveillance effort performed by the Wash- shberger et al. 2010) and Atlantic herring C. haren- ington Department of Fish and Wildlife (WDFW) and funded gus from the western (e.g. Cox 1914, Fish 1934, Sin- by an appropriation from the state legislature. Support for dermann 1956) and eastern North Atlantic (e.g. the laboratory component was provided by the Exxon Valdez Oil Spill Trustee Council (Project #19 120111-E) and Rahimian & Thulin 1996, Óskarsson et al. 2018). An US Geological Survey (USGS)−Fisheries Program, Ecosys- analogous high biomass period currently exists in the tems Mission Area. We gratefully acknowledge the techni- north Hood Canal sub-basin, where the biomass of cal contributions of P. Biondo, M. Burger, C. Fanshier, and adult, pre-spawn Pacific herring increased from 2810 M. Burger (WDFW), M. Wilmot (USGS−Marrowstone Mar- ine Field Station), and the crew of the F/V ‘Chasina’. The Mt in 2014 to 5816 Mt in 2018. Consequently, >50% use of trade, firm, or corporation names in this publication is of the Pacific herring metapopulation in the southern for the information and convenience of the reader. Such use Salish Sea currently spawns in north Hood Canal. does not constitute an official endorsement or approval by Large boom−bust cycles such as this are common in the US Government of any product or service to the exclu- Pacific herring and other marine forage fishes (Bakun sion of others that may be suitable. 2006), and it is possible that environmental condi- tions conducive to high clupeid biomass also facili- LITERATURE CITED tate high Ichthyophonus infection prevalence. For example, the north Hood Canal area is located at the Bakun A (2006) Wasp-waist populations and marine ecosys- tem dynamics: navigating the ‘predator pit’ topogra- intersection of 3 major basins in the Salish Sea, in- phies. Prog Oceanogr 68:271−288 cluding Hood Canal, Puget Sound, and the Strait of Bhaskaran K, Gasparrini A, Hajat S, Smeeth L, Armstrong B Juan de Fuca; water movement patterns in this tran- (2013) Time series regression studies in environmental sition zone create a highly productive feeding area epidemiology. Int J Epidemiol 42: 1187−1195 Bishop MA, Eiler JH (2018) Migration patterns of post- for Pacific herring. If we speculate that Pacific her- spawning Pacific herring in a subarctic sound. Deep Sea ring become infected after consuming infected food Res II 147: 108−115 (possibly intermediate hosts; reviewed in McVicar Cox P (1914) Investigation of a disease of the herring (Clu- 2011), then the high infection pressures in north pea harengus) in the Gulf of St. Lawrence, 1914. Contrib Hood Canal may simply reflect the rich feeding area. Can Biol Fish 1914-1915: 81−85 Fish FF (1934) A fungus disease in fishes of the Gulf of Alternatively, if we speculate that that the parasite is Maine. Parasitology 26: 1−16 transferred directly, from fish-to-fish via a water- Gregg JL, Powers RL, Purcell MK, Friedman CS, Hersh- borne stage (Kocan 2019), then infection pressures berger PK (2016) Ichthyophonus parasite phylogeny would be highest during periods of high host density, based on ITS rDNA structure prediction and alignment identifies six clades, with a single dominant marine type. such as the current Pacific herring boom conditions Dis Aquat Org 120:125−141 in Hood Canal. Hershberger PK, Stick K, Bui B, Carroll C and others (2002) The Hood Canal hot spot may be useful for advanc- Incidence of Ichthyophonus hoferi in Puget Sound fishes ing our understanding of basic Ichthyophonus life and its increase with age of adult Pacific herring. J Aquat history and transmission. For example, although the Anim Health 14: 50−56 Hershberger PK, van der Leeuw BK, Gregg JL, Grady CA possibilities of a paratenic or intermediate host and and others (2010) Amplification and transport of an direct fish-to-fish or waterborne transmission for endemic fish disease by an invasive species. Biol Inva- Ich thyophonus have been proposed for decades, con- sions 12:3665−3675 vincing evidence has not yet been provided to Hershberger PK, Hart LM, MacKenzie AH, Yanney ML, Conway C, Elliott D (2015) Infecting Pacific herring with support either hypothesis. These information gaps Ichthyophonus sp. in the laboratory. J Aquat Anim re main largely unaddressed because of observa- Health 27: 217−221 tional challenges inherent in studying a marine para- Hershberger PK, Gregg JL, Hart LM, Moffitt S and others site in a highly vagile host that can quickly move (2016) The parasite Ichthyophonus sp. in Pacific herring from the coastal NE Pacific. J Fish Dis 39: 395−410 large distances (Bishop & Eiler 2018). Identification Kocan RM (2019) Transmission models for the fish pathogen of Ichthyophonus hot spots narrows the geographic Ichthyophonus: synthesis of field observations and scope of epidemiological efforts into areas that con- empirical studies. Can J Fish Aquat Sci 76: 636−642 162 Dis Aquat Org 136: 157–162, 2019

Kramer-Schadt S, Holst JC, Skagen D (2010) Analysis of (eds) Fish diseases and disorders, Vol 3: viral, bacterial, variables associated with the Ichthyophonus hoferi epi- and fungal infections. CABI Publishing, New York, NY, zootics in Norwegian spring spawning herring, 1992– p 721−747 2008. Can J Fish Aquat Sci 67: 1862−1873 Mendoza L, Taylor JW, Ajello L (2002) The class Mesomy - Kuznetsova A, Brockhoff PB, Christensen RHB (2017) lmer - cetozoea: a heterogenous group of microorganisms at Test package: tests in linear mixed effects models. J Stat the -fungal boundary. Annu Rev Microbiol 56: Softw 82: 1−26 315−344 Marty GD, Freiberg EF, Meyers TR, Wilcock J, Farver TB, Óskarsson GJ, Pálsson J, Gudmundsdottir A (2018) An Hinton DE (1998) Viral hemorrhagic septicemia virus, ichthyophoniasis spizootic in Atlantic herring in marine Ichthyophonus hoferi, and other causes of morbidity in waters around Iceland. Can J Fish Aquat Sci 75: 1106−1116 Pacific herring Clupea pallasi in Prince William Sound, Rahimian H, Thulin J (1996) Epizootiology of Ichthyophonus Alaska, USA. Dis Aquat Org 32:15−40 hoferi in herring populations off the Swedish west coast. Mazerolle MJ (2019) AICcmodavg: model selection and Dis Aquat Org 27:187−195 multimodel inference based on (Q)AIC(c). R package Sindermann CJ (1956) Diseases of fishes of the western version 2.2-1. https://cran.r-project.org/web/packages/ North Atlantic. IV. Fungus disease and resultant mortali- AICcmodavg/index.html ties of herring in the Gulf of St. Lawrence in 1955. Maine McVicar AH (2011) Ichthyophonus. In: Woo PTK, Bruno DW Dept Sea Shore Fish Res Bull 25:1−23

Editorial responsibility: Dieter Steinhagen, Submitted: May 31, 2019; Accepted: July 19, 2019 Hannover, Germany Proofs received from author(s): October 2, 2019