Prone Lakesa As Means To

Home , Hypoxia in fish

This articleThis protectedis by copyright. All rights reserved. 10.1002/tafs.10176 differences between this version and the Version Record.of Please cite this article as Acceptedthrough the copyediting, typesetting, pagination and proofreading process, which may to lead articleThis has be ArticleLake Winter Survival, HabitatUse, and Hypoxia Tolerance of Montana ArcticGrayling inWinterkill a Revised TAFS

Articletype : Article - 2018 - 0135 en accepted for publication and undergone peer full review but has not been U.S. Fish and WildlifeService, Bozeman Fish Technology Center,

Departmentof Wildlife, Fish and Conservation Biology 4050 4050 BridgerCanyon Road, Bozeman, Montana 59715 Fish and WildlifeEcology and Management Program, 730 ½ North Montana Road, Post OfficePost Box 173460, Bozeman, Montana 59717 Montana Department Fish,of Wildlifeand Parks Ecology Department, Montana State University, Michael N. Davis* and Thomas E. McMahon

Museum Wildlifeof and Fish Biology Molly A.H. Webb and Jason Ilgen E. 1 Shields1 Avenue, Davis, CA 95616 University California,of Davis Matthew E. Jaeger Kyle A. Cutting Alan T.

Hitch Dillon, Montana 59725

doi: - Prone This articleThis protectedis by copyright. All rights reserved. cover,ice macrophyte hypoxiaof the consumption of DO photosyntheticproduction ( persistent snowand icecover, which lowof oxygen, hypoxia,or in is winter common a phenom availability (Greenbank1945; Agbeti & Smol 1995; Hasler et al.2009 Ice Introduction population study periodicallyoccur in the lakeand winters. However, winter conditions more severethan those observed during our two sufficient suitable habitat > 4 mg/L was availa response to DO high toleranceto acute hypoxia exposureand exhibit a physiological and behavioral stress oxygenated habitatduring cover.ice Our study demonstrated t (0.97 in 2014, 0.95 in 2015) andtelemetered that fish selected deeper (>1m), more 4 > mg/L cover, field study we 0.75 mg/L for adults and 1.50 response to dissolved oxygen (DO)levels hypoxia.to Inlaboratory the and assess the physiological tolerance, behavioral response, combined laboratory and telemetry study to documentextent hypoxiaof over two winters ArcticG survival dueto winter hypoxialimiting isa factor for a rare,adfluvi overwintering fish populations. In thisstudy we testedhypothesis the that low overwinter Abstract. mdavi *Presentaddress:

Accepted- Article covered lakescan exhibit pronounced spatial and temporal variability indissolved oxygen (DO) [email protected] ranging from persistent near rayling ( - in the epilimnion

Winter hypoxia in shallow, ice . (Mathias and Barica 1980; Bertilsson al. et 2013)

observed dynamic DO concentrations U.S. Fish and WildlifeService, Red Rock LakesNational WildlifeRefuge Thymallus arcticus concentrations Stillwater Sciences, 279 Cousteau Place, Davis, California 95618. Email:

abundance,and presence absenceor inletof streams havebeen shown to

by 27650B South Valley Road, Lima, Montana 59739

. R microbia Mathias and Barica 1980). , adiotelemetry we observed a significantbehavioral and physiological –

< 1.96 mg/L for juveniles temperatures at 1 of

4 4 mg/L. Although hypoxia presentwas inparts of ) in ) - l anoxic conditions nearbottom the decomposition of prevents or greatly limits habiting thus - < covered lakes cana be significantlimiting factor for

4 4 mg/L and determined indicated adult winter survival rate was high winterkill could stilla be lim ble in the lake epilimnion inboth study Upper Red Rock

(DO <1

organic matt If ice cover persists forsufficient a duration,

enon in thesesystems andfrom results . -

and winter survival in relation 10 mg/L) Lake depth, duration

Lake, Montana atmospheric hatArctic G al pop acute e ; Leppi al. et 2016 r

may depleteDO to thepoint during winter ice

to DO concentrations iting factor ulation of native 24 rayling a have

- gas exchangea – hr LD

3°C. In the . usedWe a 50

and thickness the lake,

for the values of ). A condition -

year

nd of This articleThis protectedis by copyright. All rights reserved. Laboratory Study Methods examinethe extent, timing, and severity hypoxiaof in UpperRed Rock Lake. and field measurements to estimate overwinter survival, Arcticfor Grayling. We used laboratory experimentsto define stressful and lethal DO concentrations la this population (Vincent 1962; Mogen 1996; Gangloff overwintersurvival due to winter hypoxia hasbeen hypothesized aspotential a limiting factor for 2017) undergonesignificant declines in abundanceand distribution (Nelson 1954; Kaya 1992 genetically distinct from otherMontana population inUpperRed Rock Lakeis the only significantremaining n relict species occupying onlyfraction a of theirhistoric range(Liknes Gould& 1987; Kaya 1992). The winter hypox purposeThe of thisstudy towas evaluatethe response of Arctic Grayling ( during winter temporal e study design cana powerful be for tool on acute responseto hypoxia and estimation of hypoxia th (Magnuson al. et 1985; Danylchuck Tonn& 2003). Laboratory studies al. 2011)or field studies examining changesin distribution and survival inrelation to DO availability tolerancestudies (Petrosky & Magnuson 1973; Furimsky etal. 2003; Landman et al. 2005; Poulsen Past studies theon effects of in differences inspeciescomposition inlakes hi with h 1979; Nilsson 2007), byor reducing oxygen demand via reduced activity (Kramer 1987). via mechanisms suchincreased as ventilation rate and increasedblood perfusion theof gi al. 1985). or migration out of lakes into shiftsin distribution (Magnuson (Whitmoreet al. 1960; Skjaeraasen al. et 2008; Poulsenal. et 2011),resulting invertical horizontalor ice Hughes1973), which inturn, cana be factor major limiting species abundanceand persistencein fishes. Severe Obtaining oxygen during in spring (Barica Mathias& 1979). initiation atmosphericof re al. et 2016). influencethe degree of winter hy ypoxia ypoxia tolerance varies widely among species (Davis 1975; Landman etal. 2005), which can result boratory and field

Accepted Article- characteriz covered lakes(Danylchuk & Tonn 2003). Fish may respond to hypoxiaavoidi by . Previous studies havedocum

Fish also exhibit a physiological responseto xtent withinsystem, a and associated fish behavioral and survival responses to low DO In northernmany temperate lakes, the oxygen ia inUpper Red RockLake, Montana. Montana Arctic Grayling area rare, now glacial

in natural conditions ing or prolonged hypoxia can be a major causeof winter mortality (Greenbank 1945;

the extentand severity hypoxiaof inthe wild. Acombined laboratory and field investigations winter hypoxia - aeration and photosyntheticDO production associated with icebreak winter comparatively well

et al.1985; Kramer 1987; andEby poxia (Mathias and Barica 1980;

( (Hasler et al.2009). sensu ented instances of very low winter DOin the lakeand poor hypoxia havegenerally been either laboratoryhypoxia

identifying acute responses to winterhypoxia, its spatial and ArcticG can place Hasleret al. 2009) to test the winter hypoxia hypothesis rayling (Peterson & Ardren 2009), and has - oxygenated streams for the winter (Magnuson et high gh hypoxia winter risk (Tonn & Magnuson 1982). 1996). In1996). this study, weused complementary winter

demands on the physiology of overwintering

resholds, whereas fieldstudies have focused

characterize winter habitat use, and - limiting period only ends w hypoxia by increasing oxygen uptake

Guenther and Hubert1991; Leppi

Crowder 2002; Hasleret al. 2009), ative,adfluvial population, is havefocused primarily on Thymallusarcticus

ngareas hypoxic ; ith the Warren etal. Winter lls (Booth ) to

up et

This articleThis protectedis by copyright. All rights reserved. tanks was monitored with two electronic DOprobes (YSI Professional Plus, Yellow Spring, OH), one fishreact did not to observerpresence when mirrors were in coveredalso a with one water 60 cm diameter). Clear, rigid 0.3 chilled, deoxygenated water recirculated was to six circular 100 insulatedL testtanks (45 cm deep pump;jet Xylem, RyeBrook, NY) usedwas to recir CA),optimally tuned for performanceat low watertemperatures. A hp 1.5 external pump (Goulds insulated reservoir. Thebrine solution chilledwas a byhp 1.5 Deltastar chiller (Aqualogic, San Diego, water through a stainless steel heat exchanger coil submerged in chilled brinesolution ins Landman al. et 2005). deoxygenationof is responses to hypoxia, asso to reducethe possibility anadditiveof effect of nitrogen supersaturation nitrogen supersaturation vacuum column ableedervia valve theon column the packed column water by exposureto a high vacuum fromvacuum a generator (Air 5 circulating lethal DOthresholds under simulated winterconditions. water was constructed to assess behavioral and physiological res apparatus Test and procedure. – feed. trout were initially fed astarter diet(Otohime Beta, Nisshin FeedCo. Ltd. Tokyo), followedpelleted by supplied with 10 incubator (MariSource,Fife, WA), transferred toflow a G embryos inMay 2014 from Tooele, UT). Juvenile Arctic by 8˚C 2009) the sameupper Missouri drainage River Upper as Red Rock La Red Rock Lakepopulation insufficientwas to allowfor collection of adults. fluvial W),24” using short length stage

- rayling from UpperRed Rock Lake. Fertilized embryos werehatched inflow a 70 mm cm

Accepted Articlewell for - - . diameter specific stressful and lethal DOthresholds. Adult ; 201; surface to simulate ice lake and prevent testfish from jumping out of the tank. Tanks were Big HoleRiver origin, and wereused in place of Red Rockgrayling since the sizeof the Upper

a minimum of Adult water and Hypoxia Hypoxia tolerance of Arctic G

and acclimated at these conditions for ~ waterthrough a custom F – ry were transferred to a 3.0 x m 1.0 m x 1.0m flow

ArcticG 328 g) werecollected from Axolotl Lake, near Ennis, Montana (45 plasticrandom pa

iscommon a treatment hatcheryof well water risk at of –

fed d 12˚C . -

term used in hypoxia experiments ( Desired DO test concentrations wereachieved by rayling

2 2 months whichbe might presentwhen themore common nitrogen Watertemperatures as lowas 1˚C wereachieved pumping by deoxygenated aily abeltvia feeder with 3.5 mm floating ClassicTrout diet (Skretting, - well way mirror

(~ h) 1 (Marking 1987 adults spawning in Grayling

were transported to the Bozeman Fish Technology andCenter held at water -

priorto testing A specialized apparatus chilling for and deoxygenating recirculating - gill net sets. Axolotl cm cking (Koch , as determined from pre

- to enhancegrowth , positioned slightly horizontal, off built, sealed, packed -

used in thickpoly rayling was tested inboth adults and juvenilesto determinelife ) . We

testing (age 0, 147 - Glitsch, Wichita, KS) carbonate covers werepositioned justbelow the used this ap Red RockCreek,

in3.0 a x m 1.0 m x 1.0 m flow 2 culate water through thesystem see also Landman and mo to prior testing Lake

untildesired the test size achievedwas - ArcticG through, 150 round L tank hatchingafter and -

Deoxygenated water column G proach to deoxygenate water inour study - rayling are a broodstockpopulation of trial measurements place ke rayling - – a main spawning tributary fo throughat tank 8˚C

ponse to hypoxia and determine (Kaya 1992; Peterson and Ardren 162 mm cylinder . DO was - . The DO concentration inthe test Vac, Seymour, CT)

. to shield fish from observers ( adjusting thevacuum n =

van den Heuvel 2003 gas bubble disease from ) were) obtained as ( 30.5 x 91.4 cm 50, TheBig River Hole in is

removed f - was 246 - throughs tank during testing of .  through vertical 14’ 2”N, D in egassing attained – . filtration method

afterreaching 60 324 mm total During

rom cascading attached to ) 111 idean filled with using a

by trials, r Arctic pressure  upplied ; .

52’ F ry ; x This articleThis protectedis by copyright. All rights reserved. exposureswere conduct hr exposures were conducted mortality concentrations included therange levelsof DO bracketinglethal a threshold between 0% and 100% were thenheld theat 2009; Barreto & Volpato 2011).At the end theof stepwise reduction DO period eachof trial, fish surfacein search conclusion of each Both measurements were duringmade a 2 majority of the observation timewas spent ab beats/min; Flintet al. 2015)and behavioral response assessedwas recording by whether the Physiological response to DOreduction was measured by counting ventilation rate(opercular ( at withinpacked the column theFor hour, first level. thenwas induced viastepwise a reduction in DO over 9 a the final tes test temperaturew reduced by trials4 juveniles with (n 12).= Afterintroduction fish tes suchlow atest temperature. At 3˚C, weperformed trials4 with adults trials, only tanks3 were tested at a time sincea reduced volume waterof requiredwas to maintain netted were tested attwo common winter temperatures of 1˚C and At3˚C. the start aof trial, fish were Behavioral and physiological responsesto progressive hypoxia injuvenile and adult acute responsesto hypoxia relatively with shortacclimation periods. when testperiods risk of pump failurefrom over temperature could only be maintained for relatively shortperiods as initial testing showed ahigh However,limitation a theof test apparatus was that thecombination veryof low DO and temperature high precision level of ( apparatus delivered a rangeof temperatures by regulating chiller output usingdigital a temperaturecontroller (Aqualogic, San Di concentrations remained within 0.1 mg/L of target concentration. Water temperature was adjusted the tanks to th positioned in the linesupplying waterto thetanks, the other placed in the linereturning waterfrom +

0.1 mg/L).

Accepted Article each progressivelyof lower levels DO of 10.0, 6.0, 4.0, 2.0, 1.0 and, inadult trials onl Stepwisereduction in accomplished DO was ted) and trials6 with juveniles (n= 36). At 1˚C, performedwe trials4 with adults (n= 12)and haphazard

(actual mortality lossor equilibrium)of

1˚C/h t temperaturefor

environmental e chiller/vacuum column assembly. During alltests, inflowand outflow DO

of higherof levels DO are common responses to hypoxia among fishes (Hasler et al. ly from holding tanks (8˚C)and placed indi by adjustingvolume the of water cir DO DO reducedwas extended 2 as - hr exposureperiod. Increased ventilation rate and movementto the water

achieved target +

ed 2.0 at and 1.0 mg/L atboth test temperatures and also at 4.0 mg/L at 3˚C 0.1˚C, ; ; insecond the hour, DO was held constant. Stepwisereduction occurred

beyond beyond a fewdays. Thus, we werelimited DO concentration for 24 h determine to the acut conditions present inU

a heating at and 1.0 0.5 mg/L both at test temperatures. For juvenile

. Once the test temperaturewas reached, + 24

0.10 mg/L DO,), enabling simulate usto -

h holding period gra and dually to the next lowesttarget level for - minute observation period for each test fish atthe (1 ice buildice ove the midline theof tank or below(coded 0 1).or - 7

of testfish to test tanks, watertemperatures were observed in initial, short ˚C) using a . pperRed Rock Lake andconcentrations DO Atthe end of acclimation, culating through the chiller - up

– series of 2 on the on chillerunit (reducing efficiency)

11 h11 testperiod, depending final DO on vidually into the6 testtanks. For the 1˚C - h

- (n = per 6 trial 4 x trials = long DO reduction periods to

duringwinter. the

the - fish weremaintained at testing short for term trials. adults For

by increasing the vacuum

(0.5 lowDO and low e lethal threshold. Test progressivehypoxia -

until the 1˚C or 3˚C 10 mg/L) Arctic G ego, CA). The y, 0.5 mg/L s,24 -

term rayling with a - hr , 24 24

. - This articleThis protectedis by copyright. All rights reserved. long winters,typically rema Red Rock Lakeexhibits many characteristics typical ofwinterkill a Rock Lake and Red Rock River. Grayling are not known inhabit to the lowerlake during throughlake single a outletconnects that Upper Rock Red Laketo the shallowermuch Lower Red contributing the largest Rock Lakes National WildlifeRefuge. lakeThe is fedby five streams, with Red RockCreek wetlands with predominantly emergentsedge ( Missouri RiverBasin of southwest Montana (Figure1 Study site. Field Study version 3.0.2 (R DevelopmentCore Team 2014). mortality, LT ArcticGrayling acute lethal concentration DO (LC appliedWe generalized linear models (Kerr Meador& 1996)to thesurvival data calculateto the a) support,respectively, as a plausible model (Burnham Anderson& <2, 2 (ΔAICc). Following recommendations Burnhamof and Anderson (2002), models with ΔAICc values differencebetween the information criteria valuefor the top model and that of every other model the lowest tovalue bethe best and sample size( Comparison of the random effect,and DO, watertemperature applied test ranktransformation precluded use of mean ventilation rate compared to the control DO level (10 mg/L). the ventilation data, fish ventilation rate as the r concentrationof DO ventilation rate data were normal distribution and juvenile datarequiring ranktransformation analysis.Data J trials were conducted during (December winter to March)and adult trials during spring (May to abovewas used, except thatthe final exposureperiod lengthened was mg/L (2 trials, lethal DOthreshold for adults was also determined in separate 96 level for calculation Theonly. additional 4.0 mg/L Accepted Articleune) usingnatural, the seasonal photoperiod pres

DOconcentrations including DO, watertemperature, and theinteraction andof DO water temperature - 10, . Surfacing

and >10, were considered to havesubstantial, some, and essentially no evidenceof -

UpperRed RockLake lies inthe high elevation (2,030 Centennial m) Valley in the Upper 50 AICc - ) for adultsfor ) during96 the n

at each test temperature, and b)thechronic exposurethreshold time (time to 50% all models performedwas using Akaike’sinformation criterion adjusted for small

V = 12). For the96 at each testtemperature usedwe ; Burnham; Anderson& 2002; Compton etal. 2002). Weconsidered the model with entilation

time

of theof acute lethal threshold.

and temperaturelevel

at data was volume surfaceof water (Gangloff 1996 which

examined separately. Adult

rate andom andom effect,and ining ice –

3˚C treatment for juvenileswas necessary to achieve a data adult - examined using mixed effects logisticregression individual with as hr 50 ) (Davison ) al. et 1959; Wagner et al.2001) for juvenile and adult

- evaluated the plausibility of other models based on the chronic test, the same stepwise DO reduction protocol describe covered from October to were Arctic Grayling mixed mixed effects - h trials. All first

was assessed using DO DO and , assessed Carex

and lifestage Testfish wereused only in once trials. ent the at time theof trials. data analys ANOVA with orthogonal contrasts models

). The 893 halake is surroundedextensive by exhibited )

lining wetland margins, andwithin lies Red water ventilation rate

for normality,for . Instead, a one temperature

April. Depth isshallowand rather (adult v e

a significant difference( mixed effects s w ; Warren et al. - - pronelake. Thelake experiences 2002). hr exposure trials at 3˚C and 1.0 ere For juvenile ventilation data, the ; thus,; s. juvenile)

responseto conducted with adult data exhibiting a from 24

as fixed effects

- adult and juv way re mode 2017

to to 96 h as fixed effects. using peated measures all combinations ls ). Flows exitthe with and identified

100% survival . winter. Upper p

program R enile

Juvenile was < . Chronic

For adult individual 0.05)in

This articleThis protectedis by copyright. All rights reserved. theirice), live/dead s sampled. Fish werelocated to within 10 m (confirmed by relocation test of transmittersunder the during daylighthours cover.ice To locatefish, two surveyors, each a with telemetry Telemetry surveys coincided lakewide with DOsampling, occurring every 2 recovery inholding pens anterior to the pelvicgirdle usingshielded the needle technique(Ross Kleiner& 1982) body weights. During implantation, fish wereanesthetized with MS (Lotek MCFT2 and implanted with radio transmittersfitted with mortality (motion)sensors and external antennas adult ArcticGrayling (261 with radiotelemetry. In Radiotelemetry. sens created by measuring depths to prior winter a with boat using3 a covariates of ice thickness, start eachof sampling day usingrecommended the procedure with water electronic probe DO (YSI Professional Plus, YellowSpring, OH). The DO probe was calibrated atthe nearest ± 0.1 10 safely 1996). differences between DO concentration inthe lake and inareas and stream mouth (5 sites) limnetictypes. Sample locations w Random Points intool ArcGIS (ver. 10.2; ESRI, Redlands, CA)and included both mid 20 sampling occasions spread across twoseparatewinters, 8 occasions in2013 ice measurements were made every 2 Lake Sampling. Oncorhynchusmykiss T L lota 1996). lakeThe supportsan abundantfish ass ( uniform, a with maximum depth of only 2 m. There areabundant rooted macrophytes Potamogeton ongnose rout ( 14 -

Accepted Article- cm up (October)until travel lake on ice became unsafe prior to break up (April). This resulted in 14 ), or or (Garmin, Olathe, KS) along transects spaced 150 apart.m - 15. Fifteen15. samplesites were generated randomlyfor each sampling occasion using the Create

W - sampled inday, a given frequent severe winter conditions. Holes weredrilled inice the with a Salvelinu

diameterhand augereach at sample location and(mg/L) DO and watertemperature (to Samplesize determined was by the maximum number sitesof that could bereliably hite - m D - ace ( long extendablemeasuring rod. Ahigh S o ucker( - C) ), uniform a on bottom comprised of mud,and peat, detritus (Gangloff 1996; Mogen 3 - s fontinalis

BM, 11 x 43mm, Newmarket, Ontario, Canada). Tag weight g)(8 was 1 - To assess theextent, duration, and annual variability inwinterhypoxia, DO Rhinichthyscataractae were

Survival and distribution Arcticof Grayling during ice winter cover was measured Catostomus commersonii tatus noted telemetryvia signal, and coordinates recorded using handheld GPS ) (Gangloff) 1996). on theon ice innorth measured at 1

September eachof study year ~

(~30 min) – ), and

snow depth,snow and water depth werealso measured at each sample site 431mm; 220 C utthroat , tagged fish werereleased backinto the lake. - 4 4 weeks - m

), and

intervals from the ice - – south transects acrosslake the until thelake entire was

- 765 g)were captured using short R during the winters of 2013 ainbow M emblage,including nativeArctic Grayling, ), ottled L ongnose - resolution bathymetric map theof lakewas T routhybrids ( S culpin ( 1 1 month prior to surfaceice formation, 49 -

mounted Intelliducertransom mount S ucker ( -

waterin erestratified to detect anticipated receiver,skied 100 Cottus bairdi <

300 m streamof mouths (Gangloff Catostomuscatostomus Oncorhynchus clarkii bouvieri x - 222, and tags positioned terface to thebottom using an - 14 and 2014 - saturated air. Lake habitat – ), and nonnative - -

term 14 and occasions 6 in 4 4 weeks during winter

– (~1 h)(~1

- 300 m lake (10 sites) - 15 from15 before . Following – B

gillnet sets 3.6 % of urbot (

), apart

and B rook Lota This articleThis protectedis by copyright. All rights reserved. Acceptedhabitat. Thesecategories werechosen based resultson of the laboratory study thatdemonstrated measurements werecategorized as either unsuitable (0 depth for each sampling occasion during the midwinter DO minima bothof winters. For DO, DO pro furtherWe analyzed winter habitatselection of habitatselection wereperformed using AIC. Hosmer Lemeshow& 1989). Comparisons theof different paired logistic regression models of nearest lake sampling pointwere used to assign habitatcovariate values (Bresl sampling occasion. For both fish and random locations, habitatcovariate measurementsfrom the compare h Winter habitatselection was assessed in two ways. First,paired logistic regression used was to (Pollocket al. 1989). their transmitter signal wentundetected and notwas detected subsequenton sampling occasions surgery recovery period (~ 1 mo determinethe potential effect hypoxiaof and winterconditions survivalon and to allowfor a post focused curves (Kaplan & Meier 1958) using bimonthly data onnumber the taggedof fish alive dead.or We Overwinter survival for each the of two study years was estimated using Kaplan approximations of the spatial variation in ( values using linear simple regression. Therewas strong a association between predicted and measured DO Articledays afterinitial field measurements(10 sites persample date, ranging f assessedcomparing by predicted DOto anindependentset of DO measurements conducted 1 20 m weighted interpolation (IDW)tool in ArcMap Spatial Analyst based interpolated on rasterdatasets of generated for each sampling occasion during the ice coverperiod using the inversedistance winter icecover. studies and wetherefore assumed they would notbe likely toselected be by coincided with levels that elicited physiological stress and behavioral avoidance in laboratory our completeice cover(mean DO 2.0< mg/L in 2013 inlake both winters. lowerThe meter of the water column was consistently hypoxicsoon after from the analysis.Data assessing habitat selection. sampling occasion ( usingsame the protocol as described above at a random subset 10 fishof relocation (Garmin GPSmap 60csx,Olathe, KS). DO, water temperature,and habitatcovariates were measured portion to availability for two different categories of DO and twodifferent categories lakeof

x 20x gridm cells of DOconcentration. Accuracy predictedof from DO spatial mapping was r

our analysisour theon periodfrom date completeof icecover to lake

= 0.85,= upper1 m of the water column were usedaswas it the only well abitatcovariates between fish locations and randomly generated locations each on -

Ateach sample site, DO at and0 1 m were averaged and inputinto ArcGIS. Only data n

Maps of spatia = 120; ~

one p - third the of total number of fish relocations persample forperiod) usein

< 0.01),< indicating that maps generated using IDW provided good

. l distribution of DO concentration inthe upper meter were

prior to lake iceformation). Fish w

DO present DO inthelakeat each sampling occasion.

Arctic Grayling - 14, 3.0< mg/L in2014 -

4 4 mg/L) s or

by assessing iffish usedhabitatin erecensored from analysis if uitable(> 4.0mg/L) winter - rom 0.5 15) - - oxygenated zone inthe wide rebound DO to .

ArcticGrayling DO levels inthis zone ow& Day 1980; – –

Meiersurvivorship 10.8 mg/L DO) s during each

during –

2 - This articleThis protectedis by copyright. All rights reserved. observation time positivelywas correlated with the juvenilelifestage and negatively correlated with models having lowsupport ( indicated the model with both water temperature and lifestagewas plausible,most with all atperiod) 4.0 mg/L from about 40 sec(33% of observation <4at mg/L.In contrast, juvenile the midline increase averaged 28 120of in s the upperhalf of testtanks. Adult the interaction between DOand watertemperature significantly influenced juvenileventila v water temperature as fixed effectsbest fit data the mixed effects linear adultmodels of ventilation rateindicated that beats/min atrate 10.0 mg/L was juvenile ventilation rate at 3 Grayling increased furtherat 2 rate At 1 beats/min butincreased averaged 50 beats/min at Physiological Laboratory Study Results avoidance categoryby each sample date, with intervals werealso calcula sampling occasion (Zar 1999). Standardized selection indiceswith Bonferroni 95% confidence were usedto test whether wasuse inproportion to availability for DOand depth categories for each locations and compared useversus availability using theaforem ArcGIS. We then divided available lake depths into categories (0 F calculateduse as the proportion relocationsof made within each DO category (Manly al. et 2002). occasion during the midwinter DO minima were then plotted on maps of DO availability and habitat category quantifiedwas for eachsampli increased stress at levels DO below 4 mg/L DO(see below). proportionalThe area of each DO Acceptedentilation rate negatively correlated with both and DO water temperature. Article depth, or wegenerated first alakedepth raster usingTriangular the Irregular Network intool ° at 1 C, ArcticGrayling adult

° was significantly higherat and4 2 mg/L compared to control levels in mean time above the midlineat 4 mg/L (114 sec). C ,

, respectively ( was similar at bytelemetered fish + at 3°C was 34 ventilation ratewas and Behavioral Response to Hypoxia. by ~ %

increasefrom 10% .

Comparison all of stayed below the tank midlineduring

mg/L. ~

~ at4 mg/L.At2 mg/L,

(8 (8 and 1 22%lower for juvenilesthan 3 at 50 (42%sec of observation period)at 10 and 6 mg/L and decreased slightly 10 ted toshowthe level of selection for each category DO and depth thecontrol

° ΔAICc 10).> The odds of rising to the surfacefor a majority theof and 6 mg/L butincreased 20% Within C increased from control levels (~ 7%) at 4.0 mg/L. At1

control) (Manly etal. 2002). 6%increase from surf ~ 12%higher than

the two indices >1 indicating preference acing behavior DO level of

mixed effects regression logistic models of surfacing behavior period)at 10 and 6 mg/L to ng occasion using ArcGIS. Fish relocations each at sampling , and temperaturegroups

mean mean increased furtherto beats/min75 10.0 control -

ventilation rate At 3 at (all other models had Δ was much pronouncedmore 1°C,at increasing However, during 3 mg/L. Ventilation rate was similar at 6 m ° ° C at

C, ventilation rate of adult ° ) C, and increased significantly to 81 and 87 ,

3°C trials3°C at 4.0 and 2.0 mg/L. at 4 mg/L compared to control levels the control DO. Asat3 For juvenile - entioned protocol. Chi , 1 1 andm 1> m)and plotted fish ventilation rate the ~ increased markedly to , and indices indicating<1 except at1 mg/L when fish 100 sec(83% of observation 1 °C trials, mo Arctic del AICc > 10 .

Watertemperature and Similar to adults, including

the

Comparison allof at 1 mg/L (Figure2) Grayling

of adultof re wasre a Arctic Grayling ° ° ), C, ventilation C, v

with -

square tests DO and DO , entilation time above

Arctic 66

t sharp ion rate. other other g/L , and

This articleThis protectedis by copyright. All rights reserved. the upper 1 m proportional volume of hypoxicareas was 31%of inlateFebruary, followeda by decline to 0%by the end March.of In contrast, the upper DO minima period. column rema in 2015 ( parts oflake, the particularlyin 2014, when overall DO concentrations were significantly lowerthan period, low DO areas were concentrated, and persistently so, concentration in DO between the two study winters(Figures interpolation DO maps showed the expansion and contraction of hypoxicareas as as well differences winters increase throughout the icecover per ice. DOconcentrations wereoften hypoxic (0 4.3of mg/L in thewinter of 2013 and across sample sites. Following iceformation, lake in 2013 and 2014, respectively, and was uniform throughoutthe lake with little variation with depth Lake and cover ice absentwas by mid occasions. Icebreak up began inthefirst week Aprilof in2014 and the last week of March in 2015 near the throughout the winter except for intermittent, small November 25 (2014)and grew to andExtent timing ofhypoxia. Field Study a test, adults held from 1 mg/L and at 0% 1 mg/L. Juveniles at 3 varied with temperature. Juvenile survival at and1 3 juvenile temperature 10 of to 1 mg/L Lethal t DO water temperature. f  t C (75% vs. 100%). JuvenileLC

Accepted Articleer h 72 -

wide averageDO concentration 0.75 1 1 m theof water column .

ArcticGrayling

p beginning March in 27 2014 and February 16 in2015 mouth

exposure; the calculated LT – < 0.01). In contrast,in both years, areas near inletstreams and the upper 1 m of the water

hresholds.

1.21 mg/L higher . The calculated LC ined underthe ice , however 0%of fish s

at1 mg/L 3 at

of Elkof Springs, Rock, Red Shambow and Grayling creeks relatively In winter2014, the proportional volume theof hypoxic(

Surviva hadhigher a lethal DOconcentration than adults and the lethal response well throughoutthe lake by mid iod. than the Complete ice cover formed on the lakeby November 15 (2013) and by l adultof ArcticGrayling  50 50 C had 100%survival at 24 h underthe ice - - oxygenated during thewinter, even duringJanuary the - April both of years. was 1.96was mg/L at 3 for for DO DO concentrations in the 14 and 5.9 mg/L in2014 ~

0.5 0.5 m inthick both Ice winters. cover uniformwas across the lake survived both both temperatures in October  50 C hadslightly a lower lethal level

was 48.3was hours.

1%in winter 2015, and high levels DO werepresent at - steadily expanded from 7%in ea 4 4 mg/L) inhypolimnetic the layer below 1 m DO DO concentrations 0.5of mg/L prior iceformation to was 9.0 mg/L and 10.2 mg/L calculated for 

 ice C and 1.50 mg/L at 1 C was C was 100% at DO concentrations of 10 to 4 - wid at 1 1 at and 3

- - (1 and 3

free areas of open water February. - e averagee DOdeclined to minimum levels 15 within 40 , > 1 deepm

58% survival at48 h survival at 2 mg/L than those tested at 3,

, in theshallow northwest and southern abouta month

4 adults.For thechronic exposure   ). C) C was 100%at DOconcentrations During the midwinter DO minima

was 0.75 mg/L. In contrast, hypolimnetic layerbegan to - 50 days after onset of lake

during many sampling  C . Juvenile LC <

4 4 mg/L)zone in the rly Januarypeakto its

prior to ice in bothoff at either and 0% survivaland 0% ( <

100 - February m 50

2 ranged ) present in This articleThis protectedis by copyright. All rights reserved. foundWe hypoxic that winter concentrations DO developed in UpperRed Rock Lake,but substantial Discussion depth category was notsignifi February27, 2014 sampling occasion ( depths mg/L study winters, whereas othersampling occasion comprised 69% availableof habitat and fish exhibited significantselection for DO > number fish in of each DO category significantlywas differentfrom expected ( 1.73, ArcticGrayling increase in selection (Table 2). resulted inaincrease 30% in selection, while every increase0.5 m indepth resulted in63% a bothof depth and DO inhabitatselection. the lake. positiveand s received supportincluded DOboth and depthcovariates. as Coefficientsfor DO and depth were among models inwhich all coefficients weresignificant (Compton al. et 2002). and depth model ( we Habitat u in 1.0) winter 2014 and 0.95 (95%CI: 0.88 survived the approximate 5 month overwinterperiod iceof cover and hypoxia: 0.97 CI:(95% 0.92 occa unsuccessful attempts to relocate the fewcensored fish in (1 2014, 2 in2015)by conducting in 38 winter 2015. Werelocat using datafrom individuals alive the at Overwinter survival. northeastcorner and itsassociated inlet tributaries. 2015, fish werenotably absent from inletof tributaries, especiallyElk Springs Creek. Duringlast the two sampling occasions in winter moved frequently and over substantial (>2 km)distances. Fish tended to aggregate near the mouths sampling day and between sampling occasions indicated fish did not remain in generally had the lowest DOconcentrations. Repeated relocations of individual fish during a outlet inthe n (Figures3, 4 Fish distribution.

Acceptedre the only models with strong or Article sional telemetry surveys intributary streams within p habita

= 0.19 > 1.0 deepm

Depth and DO werenot collinear ( se.

t - ). However,across all sampling occasions, very fewfish were relocated nearthe lake

available. AICc model comparison yielded four models with ignificant, indicating – orthwestcorner of the lake. This lake section characterizedwas by depthsshallow and

used DO categories of

0.87), except0.87), theon February 27, 2014 sampling occasion when the observed -

Generally,  AICc = was1.55) con -

on all on sampling occasions ( Kaplan

Grayling showed significant selection for areas theof lake - Arctic Grayling Meier Meier survival estimates thefor overwinter period were calculated ed 89 cantly different from expected

Arctic Grayling

in theupper 1 m –

the southwest corner of the lake and wereconcentrated inthe somesupport as plausible models (Table 1). A combined DO

100 % of < 

time of complete icecover each year, 35 in winter 2014 and 4 and4 > 4.0 mg/L inproportion to availability ( 2

sidered = 0.85, –

Odds ratios indicated t r

were widely distributed across th 1.0)in winter = - telemetered

 4.0 mg/L(Figure 0.15, s had p the plausiblemost 2

selected deeper, more oxygenated areas within

= when0.4) the obser = 5.25 , thelowest proportion observed during the two

n a

= 105;= much highermuch proportion

~ –

2015 (Figure

500 m 22.91, fish at (Figure p 

< 0.01),< indicating animportantrole 6 AICcvalues < 10 upstream p ). On this occasion, DO 4> mg/L hat every 1.0 mg/L increase inDO

each sampling occasion and made

< 0.01 as it hadas it the lowest 6 ). 5

ved number of fish ineach ). –

of of the Nearly lake. all fish 0.02), except theon e lakee in both winters 

fixed locations but All models that 2

, indicating these = 6.09,= with maximum

lakearea  2

 = p AICcvalue

0.03 = 0.01),

> 4 –

–

This articleThis protectedis by copyright. All rights reserved. Our laboratory testing revealed Arctic Grayling exhibit stre lethal DOconcentration for thislife stage. non winter testing. temperatures (Cossins et al. 1977), results.Full acclimation to low temperaturescan a be slow process, taking up to determined. hypoxia, and fishes( juveniles, study demonstrated that adult 1.0at mg/L This differenceis al. 2005)are likely to indicatehigher a using a experimental m Taylor& 2004). Lastly, the discrepancy inlethal DO thresholds could beresult the of differencesin they maybetter be adapted to the winter hypoxia common to high latitudeArctic systems (Stamford reflect speciesdifferences inhypoxia tolerance. The test temperatures(Clarke Johnston& 1999). Observed differences inlethal thresholds could also closely sim than the and1 3 experiments differences differences inhypoxia tolerancebetween adult salmonids include Possible explanations for why our LC identified by McLeay etal. (1983)and similar to that of juvenile rainbow trout(Landman et al.2005). – 10 of reported 24 1.3 mg/L at6 0.75 mg/L at and1 3 Laboratory resultsshowed that Arctic contributing factors enabling overwintersurvival Arcticof Grayling hypoxia andability the to preferentially selectoxygen numbers study winters. We also found evidence no Arctic that Grayling emigrated from the lake inlarge unexpectedly high wintersurvival adultof Arctic areas highof DOpersisted undereven complete A limitation of laboratory our

Accepted Article - C for juvenile winter photoperiods –

e.g., independent temperatureof 20 > 24> Winter photoperiod hasbeenshown to reduce to seekwinter refuge  supporting ulated winter C (Doudoroff & Shumway 1970; Alabaster Lloyd& 1982). The LC

) compared) to our 24 in experimental methods. - Nilsson Ostlund& - h hr lethal oxygen threshold of 1.5 at watertemperatures > 10  howArctic Grayling respond to long term, chronic exposureto low DO We alsoWe used relatively short acclimation periods, which could have influencedalso

C establishedFeldmeth by and Eriksen when tested exposuretime (Downing Merkens& 1957; Doudoroff & Shumway 1970; Landman et ethods, especially exposure timeused defineto the acutelethal th  C used in this study. lowThe watertemperatures used in this study, which more illustratedthe by high ArcticGrayling  and thus test our fish werelik thetheory thatlarge a body size resultsin a greatertolerance to hypoxia in C

for adult :

conditions, likely a) the extremely low watertemperatures used inour experiments,b) , which could , though they had been acclimated to 8 - survival Nilsson 2008). Arcti in

- ArcticGrayling in s our h exposuretest (100% mortality at 0.5 mg/L) tributaries. Study resultsfurther indicated that a high tolerance to

c Graylingc (Evans 1 50 Mo

Grayling tests wa lethal threshold than the 24 values are lower thanleth the therefore er lethaler threshold inour 96 tudy is similar tolethal the threshold (1.5 mg/L at 5  st p st C (Davis 1975; Landman et al. 2005), considerablyhigher

Arctic G reduced metabolic oxygen demand compared higherto

984 aststudies hypoxia of tolerancehave

–

havea highertolerance to severe hypoxia than s that we only evaluated have a

2.0 2.0 mg/L for salmonids at is

) Grayling ice cover.Contrary to expectations, we observed have , and adult our trials were slightly rayling and other salmonid species elyfully not acclim Holarctic (1978) andlower is than the commonly

- high rich winter habitat may be important also also metabolic oxygen demands

lower than the critical oxygen minima of ( > ss responses to lowDO concentrations at toleranceto influenced

95% of tagged individuals) over the two

distribution of Arctic  . - C for at least2 prior mo to - h

h al DO thresholds for most other

expos

ated to exposure test our estimatesour of minimum acute acute responses to winter acclimation 50

ure run inthespring during

of of 1.50 the very hypoxia. The .

timein study. our

Furthermore, our 20 d longer20 or reshold. Studies conducted Grayling was (100% mortality of fish of , –

temperatures and

low winter 1.96 mg/L at 1

not

c) during LC

 suggests C) 50 of of

This articleThis protectedis by copyright. All rights reserved. under out the ice d stream mouths were often ice layer of the lakeremained well oxygenated dueto input of oxygen parts oflake the in past winters 1985; Terzheviket al.2009),and that very low DO concentrations havebeen regularlyrecorded in eutrophic, long period iceof cover; Barica & Mathias 1979; Mathias & Barica 1980; Babin & Prepas Red Rock Lakehas all the common attributes highof winterkill risk (sh lowwas during the two study years. the upper spatial variability in DO, thelarge proportional volume(69 winter period. Despitethe presence hypoxi of high rate survival awas result a of high availability appear toa be direct result of their high toleranceto low DO. Rather, our results suggest that the veryThe high overwinter survival we observed in may similarly avoid DO habitatassess selection juvenileof the well that depth is animportant habitatfeature for overwintering depth of 2 m In contrastto DO, suggestingpreference that for DO 4> mg/L becamepronounced more when DO was limiting.most Grayling with DO> 4 mg/L made it themost access to a wide range of available DO (0 additional beh Overwintering ArcticGrayling displayedpreference a for more oxygen in relation to winterhypoxia wouldbea fruitful for area additional research temperature survival may be more sensitive t thereafter. ratherthan hypoxia acclimation, since fishmost exhibiting this physiological consequences of sustained stress under extremehypoxia exposure (Schreck 2010), ventilation rate with decreasing DO. suggestWe this reflected aninability to copewith the 1.0 mg/L, especially injuveniles,which contrasted with the morecommon trend increasingof survival Arcticof Grayling indicating that behavioral states results in a reduced probability of survival (Davis 1975; Schreck al. et 1997) Environmental conditions thatcause significantdepartures from baseline physiological and increased surfacing time marks the individuals attempt to extract more oxygen from the hypoxic environment (Kramer 1987), while movement. ArcticGrayling well above minimum lethal levels.As DO concentrations dropped to 4 mg/L, a Acceptedandensity Article -

oxygenated zone alongice the exhibited significantselection for DO > 4 mg/L when extent hypoxiaof was its at greatest, meter theof water column throughout both study winters indic Moreover,interaction the between lowtemperature and suggests DO juvenile that Such increases in ventilation rate signal theonset of the physiological stress responseas higher ), yetexhibited significantselection for in winterkill avioral responsethat likely aids survival underwinter hypoxia.

DO concentration DO of exhibited - waterinterface without mixing with bottom layers(Bergmann Welch& 1985 Arctic Grayling temperature this inputof would allowed have this oxygen <

both - 4 4 m . alsoWe observed a rather unexpected depression in ventilation rate near pronecondition g/L. - increased ventilation rate free and always well (Gangloff 1996). We hypothesizethat someportions theof upper han adults t

had acces

ArcticGrayling initiation of commonlyutilized range DO by transmitt 4 The lackThe extensiveof low DOwas surprising given that Upper

mg - water interfacewas greatest inareas > 1 deep. m We did not / – L shouldL be consideredcritical a threshold s

. Further exploration juvenileof distribution and survival 10 mg/L), however, anabundance of available habitat s to a narrowrange availableof o o the combination of very lowDO and very low water c conditionsc behavioral avoidance Arctic Grayling inUpper Red RockLake

in this study, butlaboratory resultssuggest they of suitableof DOlevels > 4 mg/L throughout the -

oxygenated, and that the sites

and increased incidenceof surface

with lake alonglake the bottom and considerable - 100%) and persistence DO>of 4 mg/L in Arctic Grayling resp - depths > 1 m. rich stream water, given that onse lost onse equilibrium t allow,low volume, highly - o lowo DO rich habitatinlake, the an ated that

lake . since the thickness of ered -

Arctic Grayling rich water to spread comparatively lower dultand juvenile depths ( (Hasler et al.2009).

We hypothesizeWe fish.

winterkill risk for Arctic winter maximum did not soon

, had ; adult This articleThis protectedis by copyright. All rights reserved. Accepted Institutional Animal Careand Committee Use Protocol 2013 manuscript. logi acknowledge the staff theof Red Rock Lakes National WildlifeRefuge, particularly Bill West, for their Paige Maskill, JasmineCutter, Cody Deane,Emma Freeman Halvorson, Luke Holmquist, Bateman, Lucas Clark,Rob Tracy El theirfor importanthelp in conducting the field and lab investigations: Parks and studyThe was funded from generous contributions from Montana Department of Fish, Wildlifeand Acknowledgements ArticleBarica etal. 1983; Prowse Stephenson& 1986). increaseto readily measurehypoxic zonedevelopment and assessneed the for employing preemptiveactions hypoxic zone usedin studyour could be employed in other winterkill Red Rock Lake. The method quantif of adequate depth and ArcticGrayling. We also identified key habitat variables, emphasizingimportance the of both exposureto levels DO levels on overwinter survival and habitatuse infield investigations. Our results demonstrated that requirementsfor A assessingfor risk posed by an environmental stressor. We were ableto identify minimum winter DO coupledThe study design of laboratory and field investigations proved to be aneffectiveapproach extensive epilimnetic oxygen depletion than we obs colder winter (2015 possibility high of winterkillis likely risk to periodically occur climaticconditions Upperat RedRock Lake showhigh a degree interannualof variability depletion Bertilsson etal. 2013 stical support.

in small, high elevation Wyoming reservoirs when inflowing water was present. the U.S. Fish and WildlifeService

DO levels underice surface to prevent extensive winterkill (Barica & Mathias 1979; Thestudy was performed under the auspices theof Montana State University

Rev rcticGrayling laboratory inour experimentsand directly evaluate theeffect of DO -

16) presenceof ). < iews by Similarly,

4 4 mg/L are likely to result in a reduced probab than two our study winters(2013 WyattCross

Guentherand Hubert highly ication theof - oxy

, National RefugeSystem and by anonymous refereeshelped improvethe genated areas to overwintering ArcticG

erved in studyour dynamic expansion and contraction of the (1991)foun

- 14 14 and 2014 , -

and Nikki Diedrich. gratefullyWe 18. (Davis 201 am, Glenn Boltz, CharlotteMarshall,

d . We thank the

a lowerrisk winterof - prone lakesa as means to

ility of Jon Rees, David Dockery, Leif (Warren al. et 201 - 6 15) ) .DO surveys found much more winter rayling inUpper following people survival for during a much

and the 7 ). DO DO However,

This articleThis protectedis by copyright. All rights reserved. AcceptedBurnham, K.P., and D.R. Anderson. 2002. Breslow,N.E., and N.E. Day. Statistical 1980. methods inresearch. cancer Volume the I: analysis of Booth, J.H. 1979. The effectsof oxygen supply,epinephrine, and acetylcholine ondistribution the of Bertilsson, S.A. Burgin, C. Bergmann, M.A., and H.E. Welch. Spring1985. meltwater mixing insmall arcticlakes. Barreto, R.E., and Volpato. G.L. Ventilation2011. rates indicatestress Barica, J., J. Gibson, and W. Howard. 1983. Feasibility snowof clearing to impr Article Barica, J., and J.A. Mathias. 1979. Oxygen depletion and winterkillin risk small prairielakes under Babin, J., and Prepas. E.E. 1985. Modelling winter oxygen depletion ratesin ice Alabaster J.S., and R. Lloyd. 1982. Waterquality criteria freshwater for fish. Food and Agricul Agbeti, M.D., and J.P.Smol. References information case flowblood introutgills. lake Lennon, A.Shade,and R. L. Smyth. 2013. Journal Fisheries of and Aquatic Sciences Journal of Biosciences 36:851 1531. conditions ina winterkill lake. extended ice cover. lakeszone in Canada. Organization theof United Nations, Butterworths, Boston, MA. 234. biological characteristics in two temperate lakes during cover. ice Hydrobiologia

s. L

- control studies. International Agency for Research Cancer, on Lyon, France.

imnology and Oceanography - theoretic approach. Springer

C. Carey, S.B.Fey, H

1995. Winter limnology: a comparison physical,of chemical and Journal of the FisheriesBoard Canadaof

Canadian Journal of

Journal Experimental of Biology -

85 Canadian Journal Fisheriesof and Aquatic Sciences 5.

Model selection and multimodel inference: a practical

58:1998 - P. Grossart,P. L. Grubisic, M. I. Jones,D. G.

42:1789 -

The under‐iceThe microbiome seasonallyof frozen Verlag, New York. Fisheries and Aquatic Sciences - 2012.

- 1798.

83:31

36:980 - coping stylesin Nile tilapia. - 39.

- 986. ovedissolved oxygen - covered temperate

42: 239

Canadian

Kirillin, J. T.

304:221 -

249. 40:1526 ture

- -

This articleThis protectedis by copyright. All rights reserved. Accepted O. 1984. D. Evans, L. Eby, Downing, K. Doudoroff, P., and D. Davison, R. Davis, M. D. ArticleDavis, J. Danylchuk, A. Friedland J. M. A.R., Cossins, Compton, B. Clarke, A., and N.

A., and L.

in the pumpkinseed. in and Aquatic Sciences 59:952 context species fishof inlowtensions dissolved of oxygen. Anna FisheriesTechnical Report 86:1 oxygen requirements coldof Lakes National Wildlife Refuge. Master’s thesis. Mo species:review. a Society winterkillon fatheadof minnowsin boreal lakes. Transactions ofAmerican the Fisheries 120:109 Physiology Comparative of Journal membranes. synaptic their of composition acid fatty and viscosity the and goldfish of adaptations temperature ( tele C. 1975. Minimal dissolved oxygen requirementsof aquatic life with Clemmysinsculpta

C., W. ost fish. ost

2016. Winter survival and habitatas limiting factors for Arctic Grayling Red at Rock M., andM., J. W., J.

J., and W.

- 132:289

dependentshifts inbehavioral avoidancethresholds. B. Crowder. Hypoxia 2002.

M. Johnston.M. 1999. Scaling of metabolicrate with body mass and temperature in P. Breese,P. C.

M. Rhymer,M. and M. McCollough. 2002. Habitat selection by wood turtles

Journal Animalof Ecology Temperatureindependence of annualcyclethe ofstandard metabolism

L. Shumway.L.

C. Merkens. 1957.influence The of temperature theon survival of several

- M. Tonn. 298.

Journal ofFisheries the Research Board Canada of ): anapplication paired logisticof regression. Ecology

er, and C. L. Prosser. 1977. 1977. Prosser. L. C. and er,

E. Warren, and P. Doudoroff. 1959. Experiments ondissolved the Transactionsofthe AmericanSocietyFisheries 113:494

2003. Natural2003. disturbances and fish: local and regional influences

1970. Dissolved oxygen requirements freshwaterof fishes. FAO - - 965 water fishes. water Sewage and Industrial Wastes 31:950 - 291. .

- based habitatcompression inthe NeuseRiver Estuary:

68:893 - 905.

Correlations between behavioral behavioral between Correlations ntana State University, Bozeman. ls of Appliedls of Biology 45:261

Canadian Journal Fisheriesof - 121.

32:2295 emphasis Canadian on

83:833 - 2332. - - 843. 966.

- - 512. 267.

This articleThis protectedis by copyright. All rights reserved. Kaya, C. Kaplan, E. Hughes,G. Hosmer, D. Hasler, C. Gu J. Greenbank, Limnologicalconditions1945. inice Gangloff, M. Furimsky, M., S. Flint,N., M. Feldmeth, C.R., and C.

Acceptedenther, P. andM., A. W. Hubert.1991. Factors influencing dissolved oxygen concentrations during Article

in the of studies and laboratory experiments. Physiological and Biochemical Zoology oxygen on winterhabitat selection by largemouth bass: anintegration of field biotelemetry winter insmall Wyoming reservoirs. GreatBasin Naturalist 51:282 winter Bozeman, Montana. Ro 1075. release” angling tournaments. Transactions of the American Fisheries Society responses to hypoxia in juvenile tropical Australian freshwaterfish. Theoretische und Ungewandte Limnologie drainage of Montana's Big River.VerhandlungenHole InternationaleVereinigung fur M. M. 1992. Review ofdecline the and status fluvialof Arctic

T., C. L., and P. Meier. Nonparametric1958. estimation from incomplete observations. Montana. Proceedings of the Montana Academy of Sciences ck Lakes National WildlifeRefuge, Montana. Master’s thesis. Montana State University,

M. M. 1973. Respiratory responses to hypoxia infish.

W., and S.Lemeshow. 1989. Appli R. Crossland, and R.

M. 1996.

American Statistical Association

- D. Suski,D. C.

kill of fish. J. Cooke,C.

Winterhabitat and distribution Arcticof

H. Eriksen. 1978. Ahypothesis to explain the distribution

Ecological Monographs 15:343 D. Hanson,D. S.

D. Suski,D. Y. Wang, and B.

largemouth bass and smallmouth bass:implications for “live G. Pearson. 2015. Sublethal

J. Cooke,and B. ed logistic regression. Wiley, New York, New York,USA.

53:457

20:2040

Marineand FreshwaterResearch

L. Tufts.2003. Respiratory and circulatory - - 481. coveredlakes,especially relatedas to

L. Tufts. L. - 392.

- 2044.

effects fluctuatingof hypoxia on

American Zoologist 13: 475

Grayling

2009. The influenceof dissolved

Grayling

1992:43

in Upper Red Rock Lake,Red - , 285. Thymallusarcticus -

70. of nativeof troutin a

82:143

66:293 132:1065 - 489. -

152. Journal - - 304.

This articleThis protectedis by copyright. All rights reserved. AcceptedMcLeay, D. Mathias, J. Marking, L. L. 1987. Evaluation gassupersaturationof treatment equipment at fish hatcheries in Manly, B. Magnuson, J. Liknes, G.A.,and W. ArticleLeppi, J. C.,C. Arp,D. and S.M. Whitman. Landman, M. Landman, M. Kramer, D. Kerr, D.

1241:96p. laboratory and field studies. Canadian Technical Reportof Fisheries and Aquatic Sciences Grayling ( lakes. Michigan and Wisconsin. ProgressiveFish BusinessMedia, Dordrecht. selection by animals: statistical design and analysis f Fishes adaptations fishesof in a northern Wisconsin winterkill Environmental lake. Biology of fluvial Arctic 473. Arcticlakes using morphology and landscape metrics. Environmental Management57:463 FreshwaterResearch 39:1061 freshwaterfish and invertebratesto acutehypoxia. oxygen in freshwateraquaria. models. R., and J.

L. L. 1987. Dissolved oxygen and fish behavior. A., and J. J., A.

L., L., L. McDonald, Thomas, D. T. L. McDonald, and W.

J., M. J., and M. J., A.

Canadian Journal of Fisheriesand Aquatic Sciences 14:241

Environmental Toxicology and Chemistry J. Knox,J.

P. P. Meador. Thymallusarcticus

L. Beckel,L. K.Mills, and S.

R. van d

Barica. Factors 1980. controlling oxygen depletion inice Grayling R. Gould. - 250.

R. van denHeuvel. 2

G. Malick, I. en Heuvel, and N. Ling. Relative2005. sensitivities of common

1996. 1996. Modeling doseresponse using generalized linear

(

Thymallusarcticus 1987. The1987. distribution, habitatand population characteristics of ) of) short

Water Research 37:4337 1067 K.Birtwell, G.

2016. 2016. Predicting late winter dissolved oxygen levels in

. B. Brandt. Surviving1985. winter hypoxia: behavioral

003. An improved system for the control dissolvedof - term exposure to Yukon placer mining sediments: ) in) Montana. Northwest Science 61:122 - Culturist 49:208

Hartman, and C.

Environmental Biology of Fishes

15:395

New Zealand Journal of Marineand or field studies. Springer Science& -

4342. P. Erickson. 2002. -

37:185 401. - 212.

L. L. Ennis. 1983. Effects on Arctic

- 194. - covered

Resource

18:81 - 129. -

92.

- This articleThis protectedis by copyright. All rights reserved. AcceptedSchreck, C. Ross, M. Prowse, T. Poulsen, S. Pollock, K. Petrosky, B. ArticlePeterson, Nilsson, G. Nilsson, Nelson, P. Mogen, J. ComparativeEndocrinology1653:549 frequency transmitters in fish. receipts and the oxygen deficitin temperate lakes. Atmosphere progressivehypoxia. rainbowof trout ( studies: the staggered entry design. 133. bluegill to oxygen concentrations under simulated winterkill conditions. Fisheriesand Aquatic Sciences 66:1758 Arctic Reviews Experimental Biology tricolor Master’sthesis. Montana State University, Bozeman, Montana. G.

J., and C.

D. T. Status1996. and biology of the spawning population of Red RockL

H. 1954. historyLife and management theof American

H., S.

D., andD., R. E., and S.Östlund E. 2007. Gill remodeling infish B. 2010. Stress and fish reproduction: the roles of B., L.

R., and J.

P., P., and W. Grayling ) inMontana.

R. Winterstein, C. F. Jensen, K. 83:173

F. Kleiner. Shielded 1982.

L. Stephenson.L. 1986. The relationship between winterlake radiationcover,

( R. Ardren. 2009. Ancestry, population structure, and conservation genetics of - Magnuson. 1973. Behavioral responses of northern pike, yellowperch and Thymallus arcticus 189. Oncorhynchusmykiss

‐

Journal Fish of Biology

The JournalThe of Wildlife Management 18:324 Nilsson. 2008. Does size matterfor hypoxia tolerancein fish? S. Nielsen, H. Malte, K. Aarestrup, and J. 210:2403

M. Bunck,M. and P.

The ProgressiveFish - 2409. – ) in) theupper Missouri River, USA. a newfashion or anancient secret? Journal of -

The Journal of WildlifeManagement 53:7 needle technique surgicallyfor implanting radio -

556. - ) presented) with 1774.

D. Curtis. 1989. Survival analysi 79:969

- - 979. Culturist 44:41 allostasis and hormesis.

a choice normoxiaof and stepwise Grayling

C. Svendsen. Behaviour 2011.

- Ocea - 342.

- ( 43. Thymallussignifer

Canadian Journal of n

akes Arctic 24:386

Copeia 1973:124 s intelemetry

General and - - 15. 403.

Grayling

Biological - - . This articleThis protectedis by copyright. All rights reserved. Accepted Zar, J. Whitmore, C. Warren, J. Wagner,E. Vincent,R. W. Tonn, ArticleTerzhevik,A., S.Golosov, N. Palshin, A. Mitrokhov, R. Zdorovennov, G. Zdorovennova,and I.Zverev Stamford, M. Skjæraasen, J. Schreck, C.

H. 1999. Biostatistical analysis. Society centrarchid fishes to low oxygen concentrations. Lakes National Wildlife Refuge,Lima, MT. Centennial Valley Arctic 61:434 trout to extremes intemperature,salinity, and hypoxia. Michigan, Grayling assemb covered LakeVendyurskoe, Russia. Some2009. features of the thermal and dissolved oxygen structure inboreal, shallowice Thymallus. arcticus haemoglobin genotype. Hypoxic2008. avoidance behaviour in cod ( aquaculture. CambridgeUniversity Press, New York. K. Iwama, A.

M., and J.

M., M., M.

E. 1962. Biogeographical and ecological factors contributing todecline the of Arctic B., B. J., R.

D., and E. M., C. M.,

E., T. Nilsen,E., T. J. -

lagesin northern Wisconsin lakes. 444. 89:17 ) in) North America: divergenc

E. Arndt, and Brough. M. 2001. Comparative tolerance fourof stocks of cutthroat L. Olla,L. and M.

Thymallus arcticus E. Jaeger A. Gilham, K. Ann Arbor.

Molecular Ecology

J. Magnuson. 1982. Patterns inthespecies composition and richness fish of

E. Warren, and P. Doudoroff.

D. D. Pickering, J. - 26.

B. Taylor. 2004. Phylogeographical lineages Arcticof

J. Meager, N.

W. Davis. 1997. Behavioral responses to stress. Pages145 Grayling Journal Experimental of MarineBiology and Ecology

P. Sumpter,P. and C. :

Michigan and Montana. Doctoral dissertation. University of PrenticeHall, UpperSaddle River,New Jersey.

13:1533

Cutting, Bateman,L. T. Paterson, and M. Duncan. 201 adaptive managementproject annual report.

A. Herbert, O. Moberg, V. Tronci, and A.

Aquatic Ecology e, origins and affinitieswith Eurasian - 1549.

1960. Avoidance reactions of salmonid and

Ecology 63:1149 Gadus morhua

B. Schreck, editors. Fish

Transactions ofAmerican the Fisheries

43:617

Western North American Naturalist

- L.): the effect temperatureof and - 1166. 627.

Grayling

stressand healthin

358:70 (

Thymallus V. Salvanes. Red Rock –

- 170 77. 7 .

in - G. .

This articleThis protectedis by copyright. All rights reserved. Accepted ratios werecalculated. Grayling habitatselection. Valuesin parenthesesare the DO and depth incrementsfor which odds Table2. +icesnow DO +depth DO +depth Article +temp +icesnow DO +depth +temp DO +depth thickness iceof and snow ( habitatselection. Model variables includedissolved oxygen ( Table1. AICc,ΔAICc and model coefficients for paired logist

Coefficients and odds ratios for best

Model

Variable depth

icesnow

545.7200 545.0293 544.8249 543.4767 AICc

) and) water temperatur

Coefficient

0.0201 0.3591 2.2433 1.5526 1.3482 0.0000 Δ AICc -

supported paired logistic regression model Arcticof

Tables

0.3673 0.3591 0.3755 0.3705 1.432 (1 mg/L) 1.432 (1 2.732 (0.5 m) 2.732 (0.5 DO Odds ratio Odds

e (

ic regression models Arcticof G

DO temp ), lake), depth (

0.0193 0.0201 0.0211 0.0219

depth

1.908 1.294

95% CI depth – –

icesnow 3.789 1.585 0.0088 0.0064

), the), combined

rayling 0.3638 0.3957 temp

This articleThis protectedis by copyright. All rights reserved. Accepted 1 m) Figure Standarized6. selection indicesfor two levels of DO (< and4 > 4 mg/L) and d 14; 2 fish in2014 Grayling inwinter 2013 ArticleFigure Kaplan5. sampling date inUpperRed RockLake, winter Figure 4. sampling Figure 3. response to different concentrations DO at 1 Figure 2. ( Mean streams and Figure Upper1. Red RockLake shown with smoothed bathymetric contours and location inletof List of for each sampling date F igures

Interpolated maps dissolved of oxygen concentrations and observed fish locations by Interpolated maps dissolved of oxygen concentrations and observed fish locations by date inUpperRed RockLake, the

- + Meier survival probabilities with 95% lake outlet. -

15). 15). 95% confidenceintervals) ventilation rate adultof and juvenile grayling in

- 14 14 and 2014

for winters 20 - 15. Cross15. hatches indicate censored individuals fish (1 in2013 winter 2013 13  - 14 14 and 2014 C and 3 2014 - - 15. 14.  C confidenceintervals for radio

testtemperatures

- 15 . Dotted lineindicates no selection. .

epth (< epth (< 1 andm > - tagged Arctic

Ventilation Rate (beats/min) Ventilation Rate (beats/min) 100 120 100 120 20 40 60 80 20 40 60 80 0 0

10 10

6 6

DO (mg/L) DO (mg/L)

4 4

2 2

1 3   Juveniles Adults C C Juveniles Adults

Survival Probability

Survival Probability 0.75 0.80 0.85 0.90 0.95 1.00 0.75 0.80 0.85 0.90 0.95 1.00

0 0

Days 60 Days

2013 100 60 2014

- 14 - 15

120

Depth

Selection Index Selection Index - - 1/9/14 1/9/14 1/9/14 1/30/14 2/13/14 2/27/14 1/5/15 1/19/15 1/19/15 1/5/15 2/27/14 2/13/14 1/30/14 1/9/14 2.5 0.5 1.5 0.5 0.5 0.5 1.5 2.5 0 1 2 0 1 2

Dissolved Oxygen 1/30/14 1/30/14 0.5

0.5

2/13/14 2/13/14 1

2/27/14 2/27/14 1.5

1.5

1/5/15 1/5/15

2.5

1/19/15 1/19/15 3

2/2/15 2/2/15 3.5

3.5

4+ mg/L 0 - 4 mg/L 4.5

1+ meter 0 4.5

- 1 meter