魚 病 研 究 Fish Pathology,38(3),105-112,2003.9 2003 The Japanese Society of Fish Pathology

Effect of Water Temperature and Flow Rate on the Transmission of Microsporidial GillDisease Caused by salmonae in Rainbow Trout mykiss

Joy A. Becker*, David J. Speare and Ian R. Dohoo

Atlantic Veterinary College, University of Prince Edward Island Charlottetown, Prince Edward Island Canada, C1A 4P5

(Received May 21,2003)

ABSTRACT―Two studies were designed to quantify the effect of water temperature and flow rate

on the transmission potential of the important salmonid gill pathogen, Loma salmonae. Using sur

vival analysis, increased water temperature and low flow rates were determined as risk factors for

the transmission of microsporidial gill disease caused by L. salmonae in rainbow trout

Oncorhynchus mykiss. Fish were experimentally infected with L. salmonae via a cohabitation

exposure model and monitored for the development of branchial xenomas. On any given day, fish

held at 11℃ and 15℃ had a hazard ratio equal to 0.80 and 0.68, respectively, for the development

of branchial xenomas compared with fish held at 19℃. From the flow rate trial, fish housed in a

low flow tank(0.83 L/min)had an increased chance of developing branchial xenomas when com

pared to fish in tanks at normal(1.67 L/min)and high(2.5 L/min)flow rates with hazard ratios reported as 0.69.

Key words: survival analysis, Loma salmonae, temperature, flow rate, disease management, hazard function, Oncorhynchus mykiss, rainbow trout

Diseases are an integralpart of the existence of all (Georgiadis et al.,2001). Regarding Pacific salmon animals including cultured fish populations with the culture,microsporidiosis caused by Loma salmonae was disease response dependent on the interactions of identifiedas one of the most problematic infectious dis variables defined for the host, the pathogen and the eases. environment(Hedrick,1998). This interactionis often Infectionscaused by the protozoan parasite, L. diagramed as three interlockingcircles with disease salmonae are gaining in importance in salmon aquacul occurring at the intersectionof the circles. A recent ture because of theirhigh prevalence and severe mortal examination of the health management problems for the ity(Magor,1987;Kent et al.,1995;Ramsay et al., principalfish species(Atlantic salmon Salmo salar, 2001). Allspecies of Oncorhynchus are susceptibleto Pacific salmon Oncorhynchus spp., rainbow trout L.salmonae infection,although both cultured and wild Oncorhynchus mykiss and channel catfish Ictalurus hinook salmon(Oncorhynchus tshawytscha)are thec punctatus)produced by North American aquaculture most vulnerable(Kent et al.,1995;Shaw et al.,2000). revealed infectious diseases as a major obstacle to The infectionoccurs in the gillsand in other vascularized future growth(Georgiadis et al.,2001). The diseases tissues with a finaldevelopment stage of a spore-filled that were identifiedas the most problematic in North withinthe endothelialand pillarcells of the gill American salmon aquaculture share common character (Speare et al.,1998a). Experimental infectionmodels isticsin that they are almost all infectiousagents that have been developed for L. salmonae using a high dose respond poorly to drug therapy and often have transmis oral exposure(per os)and a low dose cohabitation sion patterns that are unknown or poorly understood model(Kent et al.,1995;Shaw et al.,1998;Speare et al.,1998b;Ramsay et al.,2001). The differences *Corresponding author between these two models indicated the transmission E-mail:[email protected] potentialof L. salmonae appeared to extend beyond the 106 J.A. Becker, D. J. Speare and I. R. Dohoo

visible signs of disease(Ramsay et al.,2001). This in mykiss)(RBT)(15-25 g)were obtained from a certified dicated that the cohabitation model is a more reasonable disease-free(notifiable pathogens)commercial hatchery demonstration of the actual disease events occurring in on Prince Edward Island with no history of L. the sea cage. The cohabitation model allowed for salmonae. Approximately 75 RBT were randomly dis chronic low dose exposure to spores as compared to a tributed to each circular fibreglass tank with a flow single point exposure to a large number of spores. through system of 70 L of habitable volume. The tanks

Although microsporidian research has become of were supplied from a well water source with constant

increasing importance in veterinary and human medi aeration and the oxygen levels were monitored using a cine, there is still little knowledge regarding mechanisms Campbell Scientific Data Logger(model CR-7, Campbell of pathogenesis(Becker et al.,2002). Scientific Data Logging Inc., Logan, Utah). The RBT

The environment is perhaps the least defined ele were acclimatized to these conditions for 8 days prior to

ment of the host, pathogen and environment paradigm initiating the experiments. The flow rate was main

mentioned above(Hedrick,1998). In an aquaculture tained at 2.0 L/min because water turnover rate may situation, specific components of the environment(e.g. affect the transmission of L.salmonae(Speare et al., dissolved gases, pH, temperature, flows and turbidity) 1998c). All procedures were conducted in accordance are evident and are monitored either continuously or with the guidelines of the Canadian Council on Animal after specific times(e.g. after disease outbreaks) Care.

(Hedrick,1998). Temperature has been identified as The water temperature treatments were 11℃(cold), having a defining role in the life cycle of L. salmonae 15℃(moderate)and 19℃(warm)(±0.3), which were

(Beaman et al.,1999). The permissible temperature completed in duplicate with the second 15℃ tank being range for this parasite to proceed to sporogony and the control tank(not exposed to L. salmonae). This xenoma formation is between 9℃ and 20℃(Beaman et was chosen as the control tank because 15℃ is the opti

al.,1999). Additionally, a thermal unit model was mum temperature for L. salmonae proliferation(Beaman developed for predicting disease onset based on tem et al.,1999;Speare et al.,1998a). A two header tank

perature(Beaman et al.,1999;Speare et al.,1999). system was used to ensure precision and consistency in However, these studies were completed using a high water temperatures. One tank contained ambient well dose oral point exposure infection model. Recently, the water at approximately 11℃, the other contained heated oral infection model has been shown to produce signifi well water at 27℃. Heated water passed through an cantly higher disease prevalence and xenoma intensity aeration/degassing column to prevent gas supersatur than fish exposed via a cohabitation infection model ation. The two tanks of water mixed just before enter

(Ramsay et al.,2001). Differences between exposure ing the experimental tanks to provide the desired methods have been demonstrated in terms of the ability temperatures. Water temperatures were monitored of an infected fish to transmit L. salmonae to a naive fish daily using a Fluke armoured thermocouple(Fluke

(Ramsay et al.,2001). Corp., Everett, WA, USA)and were recorded every 10 Flow rate is another environmental factor that was min using a Campbell Scientific Data Logger.

identified as a possible disease transmission modifier in an aquaculture environment(Hedrick,1998;Reno, Trial II 1998). Flow rates are quite easily explored in the labo The purpose of thistrial was to evaluate the effectof ratory setting and currently there are no reports investi flow rate on L. salmonae disseminationin a rainbow trout gating the transmission potential of L. salmonae under population. Juvenile RBT(30-35 g)were obtained flow rate manipulation. Under the cohabitation infection from a certifieddisease-free(notifiable pathogens)

model, presumably the faster the infective spores are commercial hatchery on Prince Edward Island with no cleared from the water column, the less likely a suscep history of L salmonae. Fish(approximately 40 per tible host will ingest a spore and develop disease. tank)were housed in a flow through system of 50 L(hab

The objective of this study was to use the cohabita itablevolume)circular fibreglasstanks with a diameter tion infection model with a chronic low dose exposure to and water depth of 50 cm and 32 cm, respectively. The

infective spores to determine the transmission potential tanks were supplied from the same water source as of L. salmonae to naive fish under different temperature described above with the same monitoring systems for and flow rate regimes. temperature and water quality. Fish were acclimatized to these conditions for 5 days priorto initiating the experiments. The flow ratetreatments were set at 0.83 Materials and Methods L/min(low),1.67 L/min(normal)or 2.5 L/min(high). TrialI Each flow rate was carriedout with 2 replicateswith one The purpose of thistrial was to evaluate the effectof controltank not exposed to L. salmonae with a flow rate water temperature on L. salmonae transmission in a adjusted for 1.67 L/min(normal rate). rainbow trout population. Juvenile rainbow trout(O. Survival analysis of Loma salmonae transmission 107

Cohabitation exposure model collectedfrom each study with the survivaltime defined

Separately for each trial, a large pool of rainbow as the number of days PE untilfirst visible branchial trout was per os exposed(high dose)to L. salmonae xenoma and the failuredescribed as the first visible spores via infectious gill macerated material as previ branchial xenoma. All of the survivalanalyses were ously described and subsequently held at 15℃(Kent et completed using the software package STATAT al.,1995;Speare et al.,1998b). These RBT were rou (StataTM Corporation,College Station,Texas)using-st M tinely non-lethally evaluated for the intensity of visible procedures as outlinedby Cleves et al.(2002). xenomas(Table 1). Once a sufficient number of highly The firststep in the data analysiswas obtainingsur infectious RBT scoring a 3 or higher on the intensity vival probabilitiesfor the differenttreatment groups, index were observed(25 and 30 RBT for trial I and II, which is in the form of the survivorfunction. The survi respectively), these fish were fin-clipped for identification vor functionS(t), is the probabilitythat an individualwill and randomly allocated to the treatment tanks so that survive longer than some specifiedtime, t(Kleinbaum, five highly infectious RBT were added to each 1996). Survivalfunctions are non-increasingand start tank. This was considered the initiation of each trial at one and drop to zero ifall subjects experience the out and subsequently referred to as day 0 post exposure come of interest(Kleinbaum,1996). For both the tem

(PE). The infectious fish were allowed to cohabitate perature and flow rate studies,the Kaplan-Meier(K-M) with the naive RBT in the treatment tanks until day 21 estimator(as described in Cleves et al.,2002)was used PE. to calculate the survivor curve for each tank. The Wilcoxon's Test was used to compare the survivor Table 1. Xenoma intensityindex for measurement of the num functions. This test was specificallyused to identify ber of visiblebranchial xenomas during a Loma salmonae infec possible differenceswithin replicatesof a given treat tionin rainbow trout ment group. For each trial,fish that did not experience the disease event were considered censored observa tionsfor the survivalanalysis. The next step was to fita proportional hazards model to each of the data sets. The hazard function is the probabilityof a fish failingat a specifictime(i.e showing a visiblebranchial xenomas)given that. itdid not show xenomas before that time(Kleinbaum,1996). In other words, the hazard functionrepresents the instanta neous failurerate over time. A parametric proportional Screening methods and data collection hazards model based on a Weibull distribution was Beginning at day 21 PE and continuing on a chosen over the commonly used semi-parametric Cox biweekly basis untilday 91 PE for trialI and day 80 PE proportionalhazard regression model. Semi-paramet fortrial II,the first left gill arch forall fish in the treatment ric methods proceed by making comparisons between and controltanks were non-lethallyexamined under a individualsat the times when failureshappen to occur stereoscope to determine ifthe RBT showed branchial whereas parametric methods use probabilities that xenomas. In the case that L. salmonae was detected, depict what occurs over the whole time interval(Cleves the adipose fin was clipped and the fishwas declared et al.,2002). Consequently, parametric models are sta disease positive for the remainder of the trial. tisticallymore efficient,provided that the distribution Furthermore, xenoma intensitywas measured by an in assumptions are fullymet. In thisstudy, a Weibull haz dex value based on the number of visiblexenomas on ard functionwas assumed and this was evaluated by the firstleft gill arch(Table 1). Fish were anaesthetized graphing the empirical hazard function for each using benzocaine at a concentrationof 60 mg/L in water group. A Weibull function is either monotonically forall screening procedures. increasing or decreasing with the shape of the curve depending on the shape and scale parameters(Cleves Data analysis et al.,2002). Overall,survival analysis is a collectionof statistical The finalstep in the survivalanalysis process was to procedures forthe analysisof data in which the outcome assess the validityof the proportionalhazards model by variable of interest is time until an event occurs ensuring that no assumptions were violatedand com (Kleinbaum,1996). An event is any designated experi pletingresidual diagnostic procedures. A proportional ence of interest,for example death, disease event, re hazard regression model assumes that the hazard ratio covery or sero-conversion. Generally when using sur is constant over time. This assumption can be evalu vivalanalysis, the time variableis referredto as survival ated by lookingfor parallellines on a plotof the log of the time and the event is referredto as failure(Kleinbaum, cumulative hazard against log time or by testing for 1996). Survivalanalysis was used to interpretthe data a zero slope in a linear regression of the scaled 108 J.A. Becker, D. J. Speare and I. R. Dohoo

Schoenfeld residualsover time(Kleinbaum,1996; Cleves etal.,2002). The assumptionof a Weibulldistri bution was investigated as described above. Additionally,the possibilityof tank effectwas investi gated by buildinga shared frailtymodel. A frailtymodel is a generalizationof a survivalregression model and accounts for heterogeneity and random effects (Gutierrez,2002). Finally,Cox-Snell residualswere calculatedand graphed to identifyany influentialobser vationsand to investigateoutliers. For each trial,the mean xenoma intensityscore was calculatedfor each treatmentseparately on allsample days using SAS 8(SAS InstituteInc., Cary, NC, USA version8). The means were graphed and analysed us

ing repeated measures analysisof variance(ANOVA). Fig.1. Kaplan-Meier survivor curves for the three temperature

Because ofthe low replication,itwas assumed thatthere groups(℃)post exposure(PE)to a chronic low dose was equal correlationacross time forthe ANOVA calcu of Loma salmonae spores in a rainbow trout popula lations in both the temperature and flow rate tion. trials.Bonferroni corrections were made for allwithin time periodcomparisons.

Results

During the temperatureand flow ratetransmission trials,97.4and 92.3%of the naiveRBT developed bran chialxenomas and were subsequentlyidentified as dis ease positivefor L. salmonae, respectively.None of the non-exposed control fish developed branchial xenomas ineither trial and thereforewere not included in the data analysis.On day 21 PE, the originalhighly in fectiousfish were removed from the treatmenttanks and theirgills were examined for branchialxenomas. No xenomas were detectedon any of these fishfor both the temperature and flow rate trial,suggesting that all xenomas rupturedand releasedspores during the 3 Fig.2. Kaplan-Meier survivor curves for the three flow rate weeks cohabitationperiod. groups(L/min)post exposure(PE)to a chronic low dose of Loma salmonae spores in a rainbow trout population. Survival curves

There were no significant differences(all p>0.15) between replicate cold or warm tanks in the temperature to the normal and high flow tanks(p=0.006)(Fig. trial so data were combined to generate warm, moderate 2). The median survivaltime forthe fishin a low flow and cold treatment groups. The Kaplan-Meier survival tank was 31 days, whereas itwas 35 days forboth the curves were plotted for each temperature group(Fig. normal and high flow tanks. The survivalprobabilities 1). Fish held at 19℃ had significantly reduced survival forthe normal and high flowtanks were not significantly throughout the entire trial(p=0.004). These fish were differentfrom each other. the more likely to develop xenomas when compared to the moderate and cold water groups. The median sur Proportionalhazards model vival time was 33.5 days for the warm water fish , For both the temperatureand flow ratetrials, the whereas it was 39 days for the moderate and coldwater hazard curves in allthe treatmentgroups had a mono fish. tonicallyincreasing Weibull distribution. Subsequent to

In the flow rate trial, once again the within group this,a parametricproportional hazards model was fit to replicates were not significantly different from each other each data set to determinethe relationshipamongst the

(all p>0.15)and were combined to generate low, nor hazard functions.The estimated hazard function for mal and high flow rate treatment groups. The Kaplan each temperaturegroup was plottedin Fig.3 with the Meier survival curves revealed that the low flow - tanks warm water fish always having the highest hazard had significantly lower probability of surviving compared function.From the proportionalhazards model, on any Survival analysis of Loma salmonae transmission 109

observations. The assumption that the hazard ratio

was constant over time was examined and there were no

evidence of a violation for either temperature or the flow

rate trials (p •„ 0.10). Shared frailty models were devel-

oped and produced similar estimates for the proportional

hazards for both models with the tank effect estimators

near zero. Consequently it was not necessary to add

tank effects to the model. Finally, examination of the

Cox-Snell residuals did not reveal any outliers or influen-

tial observations. The proportional hazards model as

described above for each trial was considered appropri-

ate.

Xenoma intensity index

For both trials, the mean xenoma intensity score for

Fig. 3. Hazard function for each temperature group (•Ž) post each tank was calculated on all sample days (Fig. exposure (PE) to a chronic low dose of Loma 5). For the temperature trial, the level of disease salmonae spores in rainbow trout population.

Fig. 4. Hazard function for each flow rate group (Umin) post exposure (PE) to a chronic low dose of Loma salmonae spores in rainbow trout population.

given day, a fish in cold water has a hazard ratio for de- veloping branchial xenomas of 0.80 compared with a fish in warm water. Also, on any give day, a fish in moder- ate water has a hazard ratio for developing xenomas equal to 0.68 of that of a fish in warm water. Similar procedures were completed on the data generated from the flow rate trial and the hazard curves for these groups are plotted in Fig. 4 with the low flow rate group having the highest hazard. The hazard curve for the normal and high flow rates are indistinguishable from each Fig. 5. The mean xenoma intensity for a Loma salmonae other. From the proportional hazards model, on any infection with a low dose cohabitation infection model given day a fish in either a normal or high flow tank had a beginning at day 21 post exposure (PE). The top hazard ratio for developing xenomas equal to 0.69 com- graph shows the xenoma scores for the three different temperature groups, cold (blue), moderate (black) and pared to a fish in a low flow tank. warm (red). The bottom graph shows the xenoma Once a proportional hazards model was developed scores for the three different flow rate groups, low for each trial, the model assumptions were tested, the (blue), normal (black) and high (red). For both possibility of tank effect was evaluated and the residuals graphs, the solid and dotted lines show the replicates were examined for possible outliers or influential in each group. 110 J.A. Becker, D. J. Speare and I. R. Dohoo peaked twice at day 35 and day 46 with littlevariation moderate and coldwater fish in the cohabitation among the replicatetanks at alltemperatures. The model. The fish held at 11℃ were expected to have a xenoma scores were consistentlyhigher in the warm lagged onset time because of the delayed parasite water tanks with the combined overallwarm water mean development reported by Beaman et al.(1999). score equal to 0.772. The overallmean score for the A simple interpretation for this event was a differ moderate and coldwater tanks was 0.634 and 0.661, ence in the xenoma resolution time and subsequent respectively. The repeated measures ANOVA was spore release in the infectious fish at different completed with only mean xenoma scores up to day 50 temperatures. The large pool of infectious RBT were

PE because afterthis time the curves overlap and are held at 15℃ during their entire course of infection not differentfrom each other. Up to day 50 PE, there including the exposure point, incubation period and dis was a marginally significantdifference in the mean ease onset. Once these fish were induced into the tem xenoma intensityscores across temperatures(p= perature trial, they were placed into either 11℃,15℃ or 0.061). 19℃. The immediate temperature change may have For the flow rate trial,the xenoma intensitycurves caused the xenomas to rupture more quickly for the graduallyincreased to a peak at approximately day 45 11℃ group and conversely may have delayed the rup PE and then began to decline. The repeated measures ture for the 19℃ group. Therefore, the differential

ANOVA revealed there were was littlewithin group varia xenoma rupture times would have obscured the lag tion amongst the replicates. Once again, the ANOVA period. A further trial to specifically investigate the data set only used mean score values up to day 45 PE possibility of differential xenoma rupture times with because after this time there was no differences immediate temperature changes is required to support detected amongst the curves. Up to day 45 PE, there this proposed theory. were significantdifferences in the mean xenoma scores To our knowledge, this is the first report measuring amongst the differentflow rates(p=0.015). The low the effect of flow rate on the transmission potential of L. flow tanks had consistentlyhigher mean xenoma scores salmonae. From this study, fish in a low flow tank had a with a combined overallmean score of 0.824 whereas greater risk of developing microsporidial gill disease the moderate and high flow overallmeans were 0.693 when compared to fish at normal and high flow rates. and 0.706, respectively. Under normal RBT rearing conditions in a circular tank

(body weight>15 g), the flow rate should not exceed two exchanges per hour(Sedgwick,1990;Ross et al., Discussion 1995;Pennel and Barton,1996). Therefore, the flow Survival analysis was an excellenttool to identify rates for this study were set at one(low), two(normal) differencesin the transmission potentialof L. salmonae and three(high)exchanges per hour, assuming constant under differenttemperature and flow rate regimes. mixing. Anecdotally, increasing flow rates has been

Additionally,a simple experimental design was used to generally recommended for both disease and stress gather the requireddata forthe survivalanalyses. This reduction in salmonid culture for many years. Presum statisticaltechnique is relativelynew in the analysis of ably, the fewer water exchanges per hour will increase fish health studies. A recently study used survival the contact time the fish have with a pathogen thus analysis to effectivelyquantify the effectof fish density increase the probability of disease, as was the situation and number of infectiousfish during an laboratory based in this study. challenge to infectiouspancreatic necrosis(Bebak-Will For two species of gill parasites infecting wild eels iams et al.,2002). For our study, the proportionalhaz (Anguilla rostrata)in Atlantic Canada, Barker and Cone ard models developed for each trialrevealed that warm (2000)found that increasing flow rate over 5 cm per water and low flow rates are riskfactors which enhance second significantly reduced the abundance of the the development of branchialxenomas in RBT exposed parasites. The reduced level of disease was contrib to L. salmonae spores. uted to the quickened clearance of the free-swimming Temperature has a well-definedrole in the develop stages of the pathogens(Barker and Cone,2000). ment and transmissionof L. salmonae withinthe permis Although microsporidians do not have a free-swimming sible temperature range(Beaman,1998). Using the stage, the infective spore themselves are free-floating in high dose oral exposure model, Beaman et al.(1999) the water column. Therefore, the use of high flow determined that as water temperature increased, the regimes removes the infective spores from the water mean number of days untildisease onset decreased. more quickly and thus the fish have a reduced probability

Similar resultswere found in this cohabitationinfection of contacting a spore and developing microsporidial gill model with the warm water having the least number of disease. Additionally, increased flowing water was days to disease onset. However, unlikethe oral expo found to be an effective treatment for ichthyophthiriasis sure model, there does not appear to be a differencein (ich)caused by Ichthyophthirius multifiliis in channel cat the time to onset and subsequent survivalcurves forthe fish Ictalurus punctatus(Bodensteiner et al.,2000). Survival analysis of Loma salmonae transmission 111

Compared to over 99% mortalityat the base flow rate of (Microspora)development in rainbow trout. J. Aquat. .5 exchanges per hour, it was determined that 2.5 0 Anim. Health,11,237-245. Bebak-Williams, J., P. E. McAllister,G. Smith and R. Boston exchanges per hour greatlyreduced ich-relateddeaths (2002):Effectof fishdensity and number of infectiousfish to less 10% and 4.5 exchanges per hour, deaths were on the survivalof rainbow troutfry, Oncorhynchus mykiss reduced to 7%and the pathogen was removed from the (Walbaum), during epidemics of infectious pancreatic raceway(Bodensteiner et al.,2000). The use of a flow necrosis. J. Fish Dis.,25,715-726. ing water treatment insteadof a chemical treatment(e.g. Becker, J. A., D. J. Speare, J. Daley and P. Dick(2002):Effects formalin)was advantageous because itis non-toxic,safe of monensin dose and treatment time on xenoma reduction in microsporidialgill disease in rainbow trout,Oncorhynchus for allspecies, easily applied and does not require any mykiss(Walbaum). J. Fish Dis.,25,673-680. government approvals(Bodensteiner et al.,2000). Bodensteiner, L. R., R. J. Sheehan, P. S. Wills, A. M. Similarto ich,the treatments availablefor L. salmonae Brandenburg and W. M. Lewis(2000):Flowing water:an are extremely limited(Becker et al.,2002). The inclu effectivetreatment for ichthyophthiriasis.J. Aquat. Anim. sion of increasing flow rates in a health management Health,12,209-219. Cleves, M. A., W. W. Gould and R. G. Gutierrez(2002):An plan for microsporidialgill disease could reduce the introductionto survivalanalysis using Stata. Stata Corpo transmission of the pathogen in a high density sea ration,College Station,Texas. cage. Additionally, the potential reduction in L. Georgiadis, M. P., I.A. Gardner and R. P. Hedrick(2001):The salmonae transmission through flow rate manipulation role of epidemiology in the prevention,diagnosis, and con trolof infectiousdiseases in fish. Prev. Vet. Med.,48, would be essential ifthe aquaculture industry moves 287-302. towards land-based rearingfacilities. Gutierrez,R. G.(2002):Parametric frailtyand shared frailtysur Fish health in aquaculture is mainly population vivalmodels. Stata J.,1,22-44. based medicine;therefore epidemiologic methods and Hedrick, R. P.(1998):Relationships of the host, pathogen, and tools are a natural choice for addressing fish-disease environment:implicationsfor diseases of culturedand wild fishpopulations.J. Aquat. Anim. Health,10,107-111. problems(Georgiadis et al.,2001). Itis well known that Kent, M. L.,S. C. Dawe and D. J. Speare(1995):Transmission occurrence of infectiousdisease in a populationdepends of Loma salmonae(Microsporea)to Chinook salmon in on interactionsamong a pathogen, the host and the seawater. Can. Vet. J.,36,98-101. environment. Realisticexperimental challenge models Kleinbaum, D. G.(1996):Survival analysis:a selflearning text. can provide valuable data for identifyingpotential risk Springer-Verlag,New York, pp.4-82. Magor, B. G. (1987):Firstreport of Loma sp.() factors for disease and these resultscan be used to in juvenile coho salmon(Oncorhynchus kisutch)from specificallydesign an epidemiologicalstudy to further Vancouver Island, BritishColumbia. Can. J. Zool.,65, investigatethose factors. The development of a com 751-752. prehensive health management plan for controlling Pennell, W. and B. A. Barton(Editors).(1996):Principles of microsporidial gilldisease caused by L. salmonae salmonid culture. Elsevier,New York. Ramsay, J. M., D. J. Speare, J. G. Sanchez and J. Daley requiresa collaborativeeffort amongst culturists, veteri (2001):The transmission potential of Loma salmonae narians, epidemiologists and others in the fish health (Microspora)in the rainbow trout,Oncorhynchus mykiss community. (Walbaum), is dependent upon the method and timing of exposure. J. Fish Dis.,24,453-460. Reno, P. W.(1998):Factors involved in the dissemination of Acknowledgments disease in fish populations. J. Aquat. Anim. Health,10, 160-171. The authors wish to thank the staffof the Aquatic Ross, R. M., B. J. Watten, W. F. Krise and M. N. DiLauro Facilityat the AtlanticVeterinary College for monitoring (1995):Influence of tank design and hydraulicloading on the tank systems. This research was funded by Project the behaviour, growth and metabolism of rainbow trout Loma, a StrategicGrant from the National Sciences and (Oncorhynchus mykiss). Aquacult. Eng.,14,29-47. Sedgwick, S. D. (1990):Trout farming handbook. Fishing Engineering Research Council of Canada(DJS). News Book, London. Shaw, R. W., M. L. Kent and M. L. Adamson(1998):Modes of References transmission of Loma salmonae(Microsporidia). Dis. Aquat. Org.,33,151-156. Barker, D. E. and D. K. Cone(2000):Occurrence of Ergasilus Shaw, R. W., M. L. Kent and M. L. Adamson(2000):Innate sus celestis(Copepoda)and Pseudodactylogryrus anguillae ceptibilitydifferences in Chinook salmon Oncorhynchus (Monogenea)among wild eels(Anguillarostrata)in relation tshawytscha to Loma salmonae(Microsporidia). Dis. to stream flow,pH and temperature and recommendations Aquat. Org.,43,49-53. for controllingtheir transmission among captive eels. Speare, D. J.,H. J. Beaman, S. R. M. Jones, R. J. F. Markham Aquaculture,187,261-274. and G. J. Arsenault(1998a). Induced resistancein rain Beaman, H. J.(1998):The effectsof temperature on the devel bow trout,Onchorynchus mykiss(Walbaum), to gilldisease opment of Loma salmonae and resistanceto re-infectionin associated with the microsporidian gillparasite Loma rainbow trout(Oncorhynchus mykiss). M. Sc. thesis,Uni salmonae. J. Fish Dis.,21,93-100. versityof Prince Edward Island,Charlottetown, PE. Speare, D. J.,G. J. Arsenault and M. A. Buote(1998b):Evalua Beaman, H. J.,D. J. Speare and M. Brimacombe(1999):Regu tionof rainbow troutas a model for use in studies on patho latory effects of water temperature on Loma salmonae genesis of the branchial microsporidian Loma salmonae. 112 J.A. Becker, D. J. Speare and I. R. Dohoo

Contemp. Top. Lab. Anim. Sci.,37,55-58. by in situhybridization and immunohistochemistry.J. Fish Speare,D. J.,J. Daley,R. J. F. Markham, H. J. Beaman and J. Dis.,21,345-354. G.Sanchez(1998c):Loma salmonae associatedgrowth Speare, D. J.,H. J. Beaman and J. Daley(1999):Effect of water suppressionin rainbow trout(Oncorhynchusmykiss)oc temperature manipulation on a thermal unit predictive curs duringearly-onset xenoma dissolutionas determined model for Loma salmonae. J. Fish Dis.,22,277-283.