Stream habitat and rainbow trout (Oncorhynchusmykiss) physiological stress responses tostreamside clear-cut logging in British Columbia

Eric Mellina, Scott G. Hinch, Edward M. Donaldson, and Greg Pearson

Abstract: The impacts associated with streamside clear-cut logging (e.g., increased temperatures and sedimentation,loss of habitat complexity) are potentially stressful to stream-dwelling fish. We examined stream habitat and rainbowtrout physiological stress responses to clear-cut logging in north-central British Columbia using 15 streams divided intothree categories: old growth (reference), recently logged (clear-cut to both banks 1–9 years prior to the study), and sec-ond growth (clear-cut 25–28 years prior to the study). We used plasma cortisol and chloride concentrations as indica-tors of acute stress, and interrenal nuclear diameters, impairment of the plasma cortisol response, and trout conditionand length-at-age estimates as indicators of chronic stress. No statistically significant acute or chronic stress responsesto streamside logging were found, despite increases in summertime stream temperatures (daily maxima and diurnalfluctuations) and a reduction in the average overall availability of pool habitat. Our observed stress responses were ap-proximately an order of magnitude lower than what has previously been reported in the literature for a variety of dif-ferent stressors, and trout interrenal nuclear diameters responses to the onset of winter were approximately five timesgreater than those to logging. The overall consistency of our results suggests that the impacts of streamside clear-cutlogging are not acutely or chronically stressful to rainbow trout in our study area.

Résumé : Les impacts de la coupe à blanc en bordure des ruisseaux, tels une augmentation de la température et de lasédimentation ou une perte de complexité de l’habitat, sont une source potentielle de stress pour les poissons qui y vi-vent. Les autuers ont examiné les effets de la coupe à blanc sur les habitats de ruisseaux et les stress physiologiquesde la truite arc-en-ciel dans le centre nord de la Colombie-Britannique, en comparant 15 ruisseaux divisés en trois caté-gories : forêt ancienne (témoin), coupe récente (coupe à blanc jusqu’aux deux rives, 1–9 ans avant l’étude) et secondevenue (coupe à blanc 25–28 ans avant l’étude). Ils ont utilisé la concentration de cortisol et de chlore plasmatiquecomme indicateurs de stress aigu et les diamètres nucléaires inter-rénaux, l’altération de la réponse du cortisol plasma-tique et des estimations de l’état des truites et de la longueur–âge comme indicateurs de stress chroniques. Aucunstress aigu ou chronique statistiquement significatif n’a été observé en réaction à la coupe en bordure des ruisseaux,malgré des hausses de la température estivale des ruisseaux (maximums quotidiens et fluctuations diurnes) et une dimi-nution générale de la disponibilité moyenne des habitats de fosse. Les réactions de stress qu’ils ont observées étaientinférieures d’environ un ordre de magnitude à celles précédemment rapportées dans la littérature pour une variétéd’agents stressants différents tandis que la réaction des diamètres nucléaires inter-rénaux de la truite à l’arrivée del’hiver était environ cinq fois plus grande que dans le cas de la coupe. La cohérence de l’ensemble de leurs résultatsindique que les impacts de la coupe à blanc en bordure des ruisseaux ne provoquent pas de stress aigus ou chroniqueschez la truite arc-en-ciel dans leur aire d’étude.

[Traduit par la Rédaction] Mellina et al. 556

Introduction

Streamside timber harvesting has been shown to have avariety of potentially complex spatial and temporal impactson the physical and biological components of small streamecosystems (Hicks et al. 1991), with the most severe nega-tive impacts to stream habitat and fish populations expected

when riparian reserve buffer strips are absent and when in-stream work (e.g., falling and yarding, stream crossing, re-moval of large organic debris (LOD)) is conducted (Slaneyet al. 1977a; Hicks et al. 1991). These impacts may placeadditional physiological costs onto, and consequently be stress-ful to, resident stream-dwelling fish, as they attempt to accli-mate to any habitat alterations. Stress can be defined as the

Can. J. For. Res. 35: 541–556 (2005) doi: 10.1139/X04-202 © 2005 NRC Canada

541

Received 20 August 2004. Accepted 15 November 2004. Published on the NRC Research Press Web site at http://cjfr.nrc.ca on17 March 2005.

E. Mellina1 and S.G. Hinch. Department of Forest Sciences, The University of British Columbia, 3041-2424 Main Mall,Vancouver, BC V6T 1Z4, Canada.E.M. Donaldson. West Vancouver Laboratory, Department of Fisheries and Oceans, 4160 Marine Drive, West Vancouver, BCV7V 1N6, Canada.G. Pearson. Canfor, Box 254, Takla Road, Fort St. James, BC V0J 1P0, Canada.

1Corresponding author (e-mail: mellina@interchange.ubc.ca).

suite of physiological responses that occur when organismsattempt to adjust to environmental alterations by maintainingor reestablishing homeostasis, and it is often used as an indi-cator of overall health (Wedemeyer et al. 1984). Forexample, short-term negative effects associated with streamsidelogging include increases in stream temperature and in tem-perature fluctuations, as well as concomitant reductions indissolved oxygen concentrations (Brownlee et al. 1988; Hickset al. 1991), and these have been shown to be stressful tofish (Mazeaud et al. 1977; Strange et al. 1977; Thomas et al.1986). Furthermore, logging practices that do not retain ri-parian buffer strips and that involve in-stream work havebeen shown to lead to increases in sediment delivery tostreams (Slaney et al. 1977a; Hicks et al. 1991), and highlevels of suspended sediment have also been reported as be-ing stressful to fish (Redding and Schreck 1987). In compar-ison with these short-term impacts, long-term negative effectsassociated with streamside logging include the loss of futureLOD recruitment and a concomitant reduction in streambankstability and habitat complexity, once existing LOD degradesand loses functionality (Hicks et al. 1991). Although a re-generation of streamside deciduous vegetation can mitigatelong-term elevations in stream temperature (Gregory et al.1987), reductions in habitat complexity could decrease theamount of cover available to stream-dwelling fish (Fauschand Northcote 1992) and could consequently be stressful tofish (Pickering et al. 1987). Therefore, the presence and se-verity of any logging-related impacts may be detectable byexamining the stress responses of fish, an approach that, toour knowledge, has not previously been attempted in the sci-entific literature.

Stress responses are generally grouped into acute andchronic categories (Donaldson et al. 1984), with acute stress(e.g., as a result of handling and confinement, or episodic in-creases in stream temperature or in levels of suspended sedi-ment) being typically short term and usually resulting in areturn of the organism to a “normal” physiological state. Bycontrast, chronic stress (e.g., as a result of long-term expo-sure to a chemical pollutant or to a lack of overhead cover)is typically ongoing and usually does not allow the organismto return to a “normal” physiological state. Three levels ofstress responses have also been identified that affect succee-dingly higher levels of biological organization: a primary re-sponse involving the production of hormones (e.g., cortisol),a secondary response involving blood and tissue alterationsas a reaction to the production of hormones (e.g., a reductionin the number of white blood cells and an increase in bloodglucose and lactate concentrations), and a tertiary responseinvolving individuals and populations and which may ulti-mately lead to a general decline in population numbers (e.g.,as a result of decreased immune function and disease resis-tance, reduced feeding or food conversion efficiency, and re-duced fecundity through reduced growth; Mazeaud et al.1977; Wedemeyer et al. 1984; Gregory and Wood 1999).

We examined stream habitat and rainbow trout(Oncorhynchus mykiss) physiological stress responses tostreamside clear-cut logging in north-central British Colum-bia, and chose the following acute and chronic stress indica-tors (encompassing all three stress response levels) to assessthe direct effects of streamside logging on trout health: plasmacortisol and chloride concentrations, interrenal nuclear diam-

eters (IRND), and trout condition indices and length-at-ageestimates. Cortisol, a hormone linked to the mobilization ofenergy reserves, has been shown to increase rapidly (e.g., 3–30 min) in the presence of a variety of different stressors,and is commonly used as an indicator of acute stress in fish(Donaldson et al. 1984). Similarly, plasma chloride concen-trations (another hematological indicator of acute stress) havebeen shown to decrease in stressed freshwater fish as a re-sult of water uptake and a consequent dilution of the bloodand its ions (Wedemeyer et al. 1984). Interrenal cells, on theother hand, are specialized cells located in the anterior kid-ney of fish and are involved in the production of the hor-mone cortisol. The presence of a chronic stressor results in ademand for increased cortisol production, and a concomitantenlargement of these cells and their nuclei, and increases inIRND have been shown to be sensitive histological indica-tors of chronic stress in fish (Brown et al. 1984; Donaldsonet al. 1984). An increase in plasma cortisol concentration isan example of a primary stress response, whereas changes inplasma chloride concentrations and in IRND are examplesof secondary stress responses. Furthermore, whereas plasmacortisol concentrations are typically used as acute stress indi-cators, their lack of response to a known acute stressor (termedan “impaired response”) can also be used to assess whether afish is chronically stressed (Hontela et al. 1992). For exam-ple, a chronically stressed fish that is unable to further raiseits plasma cortisol levels when subjected to an additionalacute stressor (such as handling) exhibits an impaired stressresponse. Lastly, we used the condition index of fish (the ratioof mass/length3; Cone 1989) and length-at-age estimates aschronic stress indicators (and as examples of tertiary responses)because of the potential for stress to alter feeding patterns,increase metabolic rates, and decrease growth (Wedemeyer etal. 1984; Barton and Schreck 1987).

Because alterations to a stream’s physical environment result-ing from streamside clear-cut logging practices are potentiallystressful to stream-dwelling fish, we make the following pre-dictions: (i) In the short term (<9 years after timber harvest-ing), the number and volume of pools will not be affected bystreamside logging, but stream temperatures and rainbowtrout densities and biomass will increase relative to streamswith forested riparian zones (throughout this paper we usethe terms “streamside” and “riparian” interchangeably, andwe define them as a zone immediately adjacent to, and ex-tending 30 m perpendicular from, each stream bank; BritishColumbia Ministry of Forests and British Columbia Ministryof Environment 1995). By contrast, in the long term (>25 yearsafter logging), stream temperature increases will be miti-gated, and the number and volume of pools, as well as troutdensities and biomass, will decrease relative to streams withforested riparian zones. (ii) In response to logging-relatedhabitat changes, rainbow trout from streams with logged ri-parian zones (whether short or long term) will show elevatedplasma cortisol and decreased plasma chloride concentra-tions (acute stress responses), as well as increased IRNDs,impairment of their cortisol stress response systems, and de-creased condition and length at a given age (chronic stressresponses), relative to trout from streams with forested ripar-ian zones. (iii) Lastly, we test the validity of using IRND asan indicator of chronic environmental stress by examiningthe interrenal response of rainbow trout to the onset of winter

© 2005 NRC Canada

542 Can. J. For. Res. Vol. 35, 2005

conditions, a known stressor. For example, the physiologicaladjustments that accompany the onset of winter in temperateregions have been reported to be stressful to stream-dwellingfish as they acclimate to rapidly changing environmentalconditions (Cunjak 1988), and we therefore predict that troutwill show enlarged IRNDs in early winter relative to mea-surements made in late summer and early fall.

Materials and methods

Study site descriptionA comparative survey comprising 15 streams located within

the Sub-boreal Spruce bio-geoclimatic zone in the interiorplateau region of British Columbia (Fig. 1; Farley 1979) wasused to test our primary hypotheses. This region has a topog-raphy that is largely dominated by moderate hillside gradientsand by a climate that is generally classified as continental,with an average annual precipitation of approximately 50–100 cm falling primarily as snow between November andMarch, and rain between April and October (Slaney et al.1977a; Macdonald et al. 1992). The region is dominated byglaciolacustrine and sandy glaciofluvial soils, and the highest

stream discharges occur in spring as a result of snowmelt,and the lowest discharges typically occur during winter andduring the month of August (which is also typically the hot-test month of the year; Brownlee et al. 1988).

The 15 streams were chosen based on similarities in theirphysical characteristics and on the presence of stream-dwellingrainbow trout. For example, all streams were relatively small(bank-full widths between 2 and 5 m) and low gradient(2%–5%), with moderately incised hillsides (2%–64%) andelevations ranging from approximately 750 to 1150 m (Ta-ble 1). The 15 streams were dispersed among five main wa-tersheds throughout north-central British Columbia, all ofwhich form part of the Fraser River and Peace River drainages.Drainage areas for the study streams ranged from 128 to1790 ha (Table 1). To examine whether stream habitat andtrout density, biomass, and stress responses were associatedwith any temporal effects of streamside logging, the 15 streamswere divided into three logging categories: “old growth”,“recently logged”, and “second growth”. Old-growth streams(n = 6) served as reference sites to assess responses in systemsnot affected by clear-cut logging. Recently logged streams (n =5) were clear-cut to both stream banks 1–9 years prior to our

© 2005 NRC Canada

Mellina et al. 543

Fig. 1. Map showing general sampling locations in the central interior of British Columbia, Canada. Box “A” encompasses the Taklaand Nation River watersheds, and box “B” encompasses the Willow, Bowron, and Salmon River watersheds.

study, whereas second-growth streams (n = 4) were clear-cutto both stream banks 25–28 years prior to our study (Table 1).

Logging practices around the treatment streams involvedthe construction of a main haul road, followed by clear-cutting and the extraction of trees using tractor and rubber-tireskidders along skid trails to landings situated near the mainroad (Slaney et al. 1977a). The principal tree species thatwere harvested comprised pine (Pinus spp.), spruce (Piceaspp.), and fir (Abies spp.). Logging was carried out duringsummer and winter months and was halted during the springand autumn seasons because of heavy rains leading to poorroad conditions (Slaney et al. 1977a). Cutblock sizes rangedfrom 41 to 1790 ha, representing between 13% and 94% oftheir respective drainage areas (Table 1). Standard, opera-tional clear-cutting practices that were current at the time oflogging were used, so that riparian buffer strips were not re-tained, and in-stream falling and skidding of trees, the delib-erate removal of LOD, stream crossings, and in-stream workwith machines were common occurrences (Slaney et al. 1977a;D. Stevenson, British Columbia Ministry of Water, Land andAir Protection, 4051-18th Avenue, Prince George, BC V2N1B3, personal communication, 2001).

Stream physical characteristics and pool habitatvariables

All field work for the comparative survey was conductedduring late summer and early autumn 1996 (for simplicity,this period is collectively referred to as “summer”). Streamstudy section lengths ranged from approximately 70 to 600 m,and were determined by cutblock boundaries or by the dis-tance covered when the required number of fish for the as-sessment of stress responses was obtained. The followingstream physical characteristics were measured according tothe methods outlined in Ralph (1990) and the British ColumbiaMinistry of Forests (1996): canopy cover, stream and hillsidegradients, bank-full width and height, aspect, discharge, andthe b-axis of the largest particle moved by flowing water

(termed “D”) found within the study section thalwegs (Ta-ble 1). D was used as a surrogate for stream power, which isa measure of a stream’s ability to transport material. Fivemeasurements of D and bank-full width and height, andthree measurements of canopy cover and stream and hillsidegradients, were taken at equidistant locations along the studysections and averaged. Aspect and discharge were measuredat a single point along each study section on the day sam-pling took place. At the time of sampling, a temperature log-ger (range –5 to 35 °C, resolution 0.2 °C) was also installedin each stream to record hourly stream temperatures, but be-cause of the timing of our field work, we lack temperaturerecords for the entire summer with which to make compari-sons among our logging categories. However, the temperatureloggers were left in the streams until November 1997, andwe therefore used data covering the period from 1 June to 30September 1997 (which encompasses the warmest period ofthe year as well as the months during which our samplingtook place in 1996) and assumed that these temperature pat-terns were representative of those occurring during the sum-mer of 1996. Temperature data from streams Airport andSaw were not available because of equipment failures. Thedaily maximum temperature and daily temperature fluctua-tion (defined as the daily maximum – daily minimum) werecalculated for each stream and for each day, and averagedwithin each category over the aforementioned period.

The number and dimensions of pools were measured ineach of the 15 streams to examine the impacts of streamsidelogging practices on this habitat feature, which has the ad-vantage of being biologically important to trout (as areas forfeeding, rearing, and refuge) and can be measured independ-ently of stage (Bowlby and Roff 1986; Lisle 1987). All streamshad predominantly pool-riffle channel morphologies, and poolswere distinguished from riffles as areas of slower velocitywith relatively deeper depths and with substrates often-timescomposed of finer materials (Ralph 1990). The number ofpools within each study reach were counted and standardized

© 2005 NRC Canada

544 Can. J. For. Res. Vol. 35, 2005

Streama WatershedYearlogged

Drainagearea (ha)

Arealogged(ha) Aspect

Elevation(m)

Db

(cm)

Streamgradient(%)

Hillsidegradient(%)

Bank-fullwidth(m)

Bank-fullheight(cm)

Discharge(L·s–1)c

Canopycover(%)d

717.5 (O) Takla na 1378 na NE 781 34 2.8 2.5 2.5 80 81 60.1Leo1 (O) Takla na 1790 na W 824 50 4.3 63.7 4.7 89 46 72.5RB4 (O) Takla na 1405 na E 839 55 3.0 28.5 2.4 53 39 79.8Airport (O) Takla na 809 na SW 775 60 3.0 4.0 3.6 50 35 87.9Muskeg (O) Salmon na 346 na S 891 25 2.0 14.8 2.4 48 23 72.9Willow6 (O) Willow na 845 na NW 964 30 5.0 47.0 3.2 62 93 72.3SK7 (R) Takla 1995 664 89 NE 940 45 3.7 8.5 3.4 58 70 27.8727 (R) Takla 1991 419 132 E 763 40 5.1 25.8 2.5 42 29 13.2Salmon1 (R) Salmon 1995 604 188 SW 1136 16 3.3 13.8 2.8 58 31 49.418 Mile (R) Bowron 1987 692 650 W 963 40 4.7 48.7 3.1 54 54 1.8T2 (R) Nation 1996 128 41 E 843 10 2.6 6.8 2.5 33 38 0.3Thursday (S) Willow 1971 1768 929 N 841 35 4.7 41.5 4.4 67 95 13.9Saw (S) Bowron 1971 1142 709 S 1006 55 3.0 10.0 4.5 86 205 68.6Holonauh (S) Willow 1971 1695 906 W 962 18 2.0 3.5 3.8 55 64 42.6Willow5 (S) Willow 1968 1155 890 W 864 30 4.0 21.2 3.5 53 108 57.8

Note: na, not applicable.aO, R, and S represent old-growth, recently logged, and second-growth streams, respectively.bD represents the size of the largest particle moved by flowing water.cDischarge represents a single measurement taken at the time of sampling.dCanopy cover represents the portion of the stream shaded by riparian vegetation.

Table 1. List of the physical characteristics measured at each of the 15 study streams.

by dividing by the total area of the study section (bank-fullwidth × length sampled × 100). Within each study section,10 pools were systematically chosen (2 pools at each of thesites where bank-full width was measured), and the residualpool volumes were measured following the methods outlinedin Lisle (1987) and averaged. Lastly, for each stream westandardized the total available pool habitat within the studyreaches by multiplying the number of pools by the averagepool volume, and dividing this estimate by the total area ofthe study section.

Rainbow trout density and standing cropbiomass

Rainbow trout were collected from each stream using abackpack electroshocker (with voltage settings ranging from400 to 500 V and a frequency of 50–60 Hz) and a single-pass removal method. Because our sampling protocol wasconstrained by the requirements of the stress work describedbelow and which required that half of the captured fish besacrificed within 1 min of capture, we were unable to employa more rigorous method to assess fish densities and biomass(such as multiple-pass removal; Fausch and Northcote 1992).Stream-specific trout density and standing crop biomass esti-mates were standardized by dividing each estimate by the totalavailable pool habitat (defined previously) within the streamstudy reaches. We justify using the total available pool habi-tat (rather than the area of stream sampled) to standardizeour density and biomass estimates, because pools provideenergetically favorable areas for trout feeding and rearing(Bowlby and Roff 1986), and they have been shown to bethe preferred stream habitat for rainbow trout when othersalmonid species are rare or absent (Muhlfeld et al. 2001).Furthermore, because seasonal changes in microhabitat selec-tion have been observed in rainbow trout, pool habitat maylimit overwinter abundance and survival in stream salmonids(Cunjak 1996; Solazzi et al. 2000). Our standardization there-fore provides an estimate of trout density and standing cropbiomass as a function of preferred or critical habitat avail-ability and not total available area.

Rainbow trout physiological stress responsesRainbow trout collected from each stream were separated

into two groups of approximately equal numbers. The firstgroup (termed “sacrificed immediately”) was sacrificed within1 min of capture to prevent an endocrine response caused byelectroshocking (Sumpter et al. 1986). The second group(termed “handled”) was kept confined in a bucket for 40 minand subjected to additional handling and confinement stress,and this time period was judged to be sufficient for the elici-tation of an acute cortisol stress response resulting fromthese three stressors (electroshocking, handling, and confine-ment; Strange and Schreck 1978; Woodward and Strange1987). The first group (sacrificed immediately) was used todetermine if any habitat alterations resulting from streamsidelogging were acutely stressful to trout, whereas the secondgroup (handled) was used to determine if the trout cortisol re-sponse system was impaired (chronic response). Trout weremeasured (fork length) and weighed prior to being sacri-ficed, and otoliths were used to classify fish into age classesfor the length-at-age comparisons following the methods of

Chilton and Beamish (1982). Fish were killed by a blow tothe head, followed by a severing of the caudal peduncle.Blood from each fish was collected using heparinized capil-lary tubes, separated into its constituent components using aportable centrifuge, frozen on dry ice, and later transferredto a freezer and kept at –20 °C. Fish kidneys were preservedin the field in Bouin’s fixative for later analyses of IRND.

In the laboratory, plasma cortisol concentrations were mea-sured using cortisol radioimmunoassay kits (Incstar Corp.,Stillwater, Minnesota), and plasma chloride concentrationswere determined amperometrically using a digital chloridometer(Haake Buchler Instruments, Inc., Saddlebrook, New Jersey).A maximum of four plasma replicates per fish were obtainedfor the hematological determinations, although in some casesthe amount of blood collected from the smaller fish resultedin only a single sample being available for each test.

Histological preparations to examine interrenal nuclei wereconducted following the methods outlined in Brown et al.(1984) and are summarized below: after removal, trout headkidneys were washed in ethanol, acetone, and xylene beforebeing embedded in paraffin for sectioning. Sections were cutat 7-µm intervals and mounted on microscope slides afterstaining with Harris’ hematoxylin and eosin. A minimum offour slides were prepared for each fish to ensure that a suffi-cient number of interrenal cells would be available. A meanIRND value for each fish was obtained by averaging 30round nuclear diameters measured using video microscopy(Zeiss light microscope and a Cohu solid state video camera)and a computer program for measurements in micrometres(Bioscan Optimas, version 3.14, Edmonds, Washington). Toreduce possible biases in measurements, microscope slideswere coded to obscure the identities of fish and their streams,and all diameters were measured by the same operator.

Lastly, we examined trout IRND responses to the onset ofwinter conditions using the same sampling protocols for fishcapture and histological preparations described previouslyfor the comparative survey. In early December 1996 (“win-ter”), 19 trout were captured from stream Leo1 (old growth)and 12 trout from stream T2 (recently logged) in reaches up-stream of, and adjacent to, the study sections sampled duringthe summer. Stream temperature at the time of sampling wasalso recorded.

Data analysisTo ensure that any differences in trout stress levels among

our three logging categories could be attributed to the log-ging treatments and not to natural variation in underlyingstream physical characteristics, we used discriminant analysisbased on drainage area, stream and hillside gradients, bank-full width and height, and D to determine if any differencesexisted in these characteristics among our categories.

For all analyses involving analysis of variance (ANOVA),stream averages were calculated for each response variable,and streams were considered to be the sampling units tominimize pseudoreplication (Hurlbert 1984). Normal proba-bility plots were examined to ensure that all variables metthe assumptions of normality, and the pool habitat, troutdensity and biomass, and hematological (plasma cortisol andchloride concentration) response variables all required log10transformations to ensure variance homogeneity. However,because the overall conclusions remained unchanged whether

© 2005 NRC Canada

Mellina et al. 545

or not log10-transformed variables were used, for clarity allfigures involving these variables were constructed using theoriginal, untransformed data. For the hematological responsevariables, trout blood samples from streams Airport, SK7,and T2 were accidentally thawed and had to be discarded.Therefore, data from only 12 streams (5 old growth, 3 recentlylogged, and 4 second growth) were used in the analyses in-volving cortisol and chloride stress responses. Data from all15 streams were used in the analyses involving IRND, troutcondition, and length-at-age comparisons for the age-1 class,whereas data from 13 streams (5 old growth, 5 recentlylogged, and 3 second growth) were used for the age-2 classcomparisons. Our age-class analyses were restricted to thesetwo age classes, because they were the only ones from whichwe collected sufficient numbers of fish within each loggingcategory. Lastly, stream physical characteristics and trout den-sity, biomass, and physiological stress response averages foreach stream are included in the Appendix (Tables A1 and A2).

One-way ANOVA with unequal cell sizes followed byTukey a posteriori tests for effects were used to assess if dif-ferences in canopy cover, pool habitat, and trout density andbiomass were evident among the three logging categories.We determined whether trout length was correlated with thehematological and IRND stress response variables using lin-ear regression analysis. Two-way ANOVA with unequal cellsizes was used to simultaneously test for differences in troutplasma cortisol concentrations among the three logging cate-gories (using only trout that were sacrificed immediately toassess whether streamside harvesting was acutely stressful tofish), and within each logging category (between trout thatwere sacrificed immediately and those that were handled toassess impairment of the cortisol response system). Becauseresponses in plasma chloride concentrations and interrenalnuclei occur on the order of hours and days, respectively(Woodward and Strange 1987; Donaldson et al. 1984), thesacrificed and handled fish were grouped together withineach logging category for analyses involving these two indi-cators. One-way ANOVA with unequal cell sizes followedby Tukey a posteriori tests for effects were used to evaluatedifferences in plasma chloride concentrations, IRND, andlength-at-age among the three logging categories. For the as-sessment of IRND responses to the onset of winter condi-tions, data analysis involved grouping the trout IRND datafrom the two streams according to season, followed by one-way ANOVA (using streams as the sampling units) to testfor differences between mean summer and winter values.Post hoc power analyses were conducted using the statisticalpackage G*Power (Erdfelder et al. 1996) and an α level of0.05, when probability (P) values for the ANOVA resultswere >0.05. However, our statistical power estimates representapproximations because of the imbalance in our experimentaldesign (Erdfelder et al. 1996). Differences in trout conditionamong our three logging categories were assessed using anal-ysis of covariance (ANCOVA) on stream-averaged log10-transformed length and mass data (Cone 1989).

To allow us to compare the magnitude of the stress re-sponses observed in our study streams with those reported inthe literature, for each hematological and IRND responsevariable, we calculated the difference (termed “effect size”; %)between the mean value for our reference category and thosefor our logged (recently and second growth) categories. Effect

size was calculated as [(mean value logged – mean value oldgrowth)/mean value old growth] × 100, and a positive (nega-tive) difference indicates an increase (decrease) relative tothe old-growth category. Only those trout that were sacri-ficed immediately were used in our calculations involvingplasma cortisol concentrations, as our primary interest lay inassessing the impacts from streamside logging and not theeffects from electroshocking, handling, and confinement. Simi-lar calculations were made using results from previous stud-ies in the stress literature that examined salmonid cortisol,chloride, and IRND stress responses to a variety of differentstressors (Table 2). For those studies assessing stress re-sponses to different doses, the minimum and maximum ef-fect sizes were also calculated, whereas only the maximumreported effect sizes were calculated for those studies assess-ing stress responses to a single dose or over time.

Results

Stream physical characteristics, pool habitat, and troutdensity and biomass

The discriminant analysis revealed no obvious groupingstructure among the three logging categories (Wilks’ λ =0.20, P = 0.25), with an overall correct classification rate of53% using a jackknifed procedure. This supports our as-sumption that the streams were relatively similar in terms oftheir general physical characteristics (drainage area, bank-full width and height, stream and hillside gradients, and D).

Average canopy cover levels (±1 SE) were 74.3% (3.8),18.6% (9.3), and 45.7% (11.9) for the old-growth, recentlylogged, and second-growth streams, respectively, with signifi-cant differences (ANOVA, P < 0.03) between the old-growthand treatment streams, but not between the two treatmentcategories (P = 0.11). Whereas stream temperatures duringour late summer and early autumn sampling period in 1996ranged from 3.4 to 9.6 °C, the average maximum daily tem-peratures for the period from 01 June to 30 September 1997were 12.7 °C for the old-growth streams, 14.1 °C for therecently logged streams, and 13.9 °C for the second-growthstreams (Fig. 2a). Average daily fluctuations over the same timeperiod were 2.2, 5.4, and 3.1 °C for the old-growth, recentlylogged, and second-growth categories, respectively (Fig. 2b).Using ANOVA, no significant differences were found amongthe three logging categories with respect to the number ofpools (P = 0.12, power approx. 0.42), the average pool volume(P = 0.09, power approx. 0.46), or the total pool habitatpresent within the study reaches (P = 0.44, power approx.0.17; Fig. 3a). Qualitative trends in these three pool habitatvariables were not consistent among categories: recently loggedstreams had the greatest number of pools but the lowest poolvolumes, whereas the reverse trend was seen in the second-growth streams. However, the overall trend was a reductionof approximately 30% in the total available pool habitat inthe logged streams relative to the old-growth category (Fig. 3a).No significant differences were found among the loggingcategories with respect to rainbow trout density (P = 0.98,power approx. 0.05) and standing crop biomass (P = 0.40,power approx. 0.19; Fig. 3b). In contrast with the pool habitatvariables, the qualitative trends in trout density and biomasswere relatively consistent, with the logged streams (recent andsecond growth) showing increases of between 5% and 269%

© 2005 NRC Canada

546 Can. J. For. Res. Vol. 35, 2005

© 2005 NRC Canada

Mellina et al. 547

Old

grow

thR

ecen

tly

logg

edS

econ

dgr

owth

Str

esso

rsan

dst

ress

resp

onse

s(e

ffec

tsi

ze)

repo

rted

from

the

lite

ratu

re

Str

ess

indi

cato

rM

ean

Mea

nE

ffec

tsi

ze(%

)M

ean

Eff

ect

size

(%)

Str

esso

rS

tudy

a

Min

imum

effe

ctsi

ze(%

)

Max

imum

effe

ctsi

ze(%

)

Cor

tiso

l(n

mol

·L–1

)43

.02(

18.1

5)49

.95(

9.75

)+

1664

.01(

21.6

3)+

49H

andl

ing

and

conf

inem

ent

1+

4700

2+

460

3+

4100

+21

000

4+

432

+71

25

+32

75In

crea

ses

inw

ater

tem

pera

ture

2+

250

6+

66ns

Sus

pend

edso

lids

7+

250n

s+

1950

Aci

difi

cati

on8

+12

3ns

+62

46E

lect

rosh

ocki

ng1

+83

6E

xpos

ure

toco

pper

9+

100n

s+

4300

Exp

osur

eto

herb

icid

es10

–32n

s+

2187

Exp

osur

eto

land

fill

leac

hate

11+

148

+20

350

Exp

osur

eto

pulp

mil

lef

flue

nt12

+49

4+

1340

0E

xpos

ure

toth

ean

tibi

otic

Kan

amyc

in13

+80

ns+

1362

Chl

orid

es(m

mol

·L–1

)13

8.11

(2.2

4)13

5.39

(3.3

8)–2

136.

92(1

.17)

–0.9

Han

dlin

gan

dco

nfin

emen

t1

–19

–76

Ele

ctro

shoc

king

1–4

nsE

xpos

ure

topu

lpm

ill

effl

uent

14–3

ns–1

5H

ooki

ngst

ress

15–6

ns–1

3IR

ND

( µm

)5.

56(0

.04)

5.67

(0.0

8)+

25.

61(0

.07)

+1

Aci

difi

cati

on8

+3n

s+

10E

xpos

ure

tohe

rbic

ides

10+

3ns

+17

Exp

osur

eto

land

fill

leac

hate

11+

8+

26E

xpos

ure

toth

ean

tibi

otic

Kan

amyc

in13

+11

+22

Exp

osur

eto

hydr

ocar

bons

16+

11O

nset

ofw

inte

r17

+9.

5

Not

e:V

alue

sin

pare

nthe

ses

repr

esen

tst

anda

rder

ror

ofth

em

ean.

For

rece

ntly

logg

edan

dse

cond

-gro

wth

stre

ams,

the

diff

eren

ces

(eff

ect

size

)be

twee

nth

eha

rves

ted

and

old-

grow

th(c

ontr

ol)

cate

go-

ries

are

also

incl

uded

.A

posi

tive

(neg

ativ

e)di

ffer

ence

indi

cate

san

incr

ease

(dec

reas

e)re

lativ

eto

the

cont

rol

cate

gory

.V

alue

sfo

rpl

asm

aco

rtis

olco

ncen

trat

ions

are

only

for

trou

tth

atw

ere

sacr

ific

edim

med

iate

ly(i

.e.,

not

subj

ecte

dto

hand

ling

stre

ss),

whe

reas

valu

esfo

rch

lori

des

and

inte

rren

alnu

clea

rdi

amet

ers

(IR

ND

)ar

efo

ral

lca

ptur

edfi

sh(s

acri

fice

dim

med

iate

lyan

dsu

bjec

ted

toha

ndlin

gst

ress

).St

ress

ors

and

stre

ssre

spon

ses

(rep

orte

din

term

sof

effe

ctsi

ze)

from

liter

atur

est

udie

sus

ing

salm

onid

sas

test

orga

nism

sar

eal

soin

clud

edfo

rco

mpa

rativ

epu

rpos

es.

For

thos

est

udie

sas

sess

ing

stre

ssre

spon

ses

todi

ffer

ent

dose

s,th

em

inim

uman

dm

axim

umef

fect

size

sw

ere

also

calc

ulat

ed,

and

nsre

pres

ents

effe

ctsi

zes

that

wer

ere

port

edas

bein

gst

atis

tical

lyno

nsig

nifi

cant

atth

eP

=0.

05le

vel.

Onl

yth

em

axim

umre

port

edef

fect

size

sw

ere

calc

ulat

edfo

rth

ose

stud

ies

asse

ssin

gst

ress

resp

onse

sto

asi

ngle

dose

orov

ertim

e.a 1,

Woo

dwar

dan

dSt

rang

e19

87;

2,St

rang

eet

al.

1977

;3,

Bar

ton

etal

.19

80;

4,St

rang

ean

dSc

hrec

k19

78;

5,Su

mpt

eret

al.

1986

;6,

Tho

mas

etal

.19

86;

7,R

eddi

ngan

dSc

hrec

k19

87;

8,B

row

net

al.

1984

;9,

Don

alds

onan

dD

ye19

75;

10,

McB

ride

etal

.19

81;

11,

McB

ride

etal

.19

79;

12,

Dye

and

Don

alds

on19

74;

13,

McB

ride

etal

.19

75;

14,

And

erss

onet

al.

1988

;15

,W

ydos

kiet

al.

1976

;16

,B

rand

etal

.20

00;

17,

McL

eay

1975

.

Tab

le2.

Lat

esu

mm

eran

dea

rly

autu

mn

mea

nva

lues

for

rain

bow

trou

the

mat

olog

ical

and

inte

rren

alnu

clea

rdi

amet

er(I

RN

D)

stre

ssre

spon

ses

tost

ream

side

clea

r-cu

tlo

ggin

ggr

oupe

dby

logg

ing

cate

gory

.

in both of these response variables relative to the referencecategory (Fig. 3b).

Rainbow trout physiological stress responses, condition,and length-at-age: comparative survey

Trout captured from our streams ranged in fork lengthfrom 4.9 to 21.8 cm, with the majority of fish comprisingthe age-1 and age-2 classes (of the 481 trout we captured,only 24 age-0, 30 age-3, 4 age-4, and 1 age-5 trout were en-countered). Linear regression analyses revealed that lengthwas not significantly correlated (P > 0.22 for all analyses)

with any of the hematological (plasma cortisol and chlorideconcentrations) and IRND stress response variables. With re-spect to trout plasma cortisol responses to streamside clear-cut logging, the two-way ANOVA revealed no significantdifferences (P = 0.46, power approx. 0.14) in mean cortisolconcentrations of trout that were sacrificed immediately amongthe three logging categories (Fig. 4a), despite increases (effectsizes) of 16% and 49% in recently logged and second-growthstreams, respectively, relative to the reference streams (Ta-ble 2). Furthermore, significant differences (ANOVA, P <0.001) were found between trout that were sacrificed immedi-ately and those that were handled within each logging category(Fig. 4a), suggesting that the cortisol response systems werenot impaired. The interaction term in this analysis (handled orkilled trout × logging category) was not significant (P = 0.64).

Using stream averages for all captured fish, one-way ANOVArevealed no significant differences in trout plasma chlorideconcentrations (P = 0.72, power approx. 0.10; Fig. 4b) orIRND values (P = 0.40, power approx. 0.19; Fig. 4c) among

© 2005 NRC Canada

548 Can. J. For. Res. Vol. 35, 2005

Fig. 2. (a) Average summertime (1 June – 30 September) dailymaximum temperatures and (b) daily temperature fluctuations(daily maximum – minimum) during 1997, grouped according tologging category (black line, forested, n = 5 streams; brokenline, recently logged, n = 5 streams; grey line, second growth,n = 3 streams). The daily data were calculated for each streamand averaged within each category over the aforementioned period.

Fig. 3. (a) Pool habitat and (b) rainbow trout density and standingcrop biomass responses to streamside clear-cut logging. Data aregrouped by logging category (striped bar, forested, n = 6 streams;unfilled bar, recently logged, n = 5 streams; grey bar, secondgrowth, n = 4 streams). In panel (b), a logarithmic scale was usedfor the y-axis to clarify the density patterns. Mean estimates werecalculated using stream averages, and error bars represent SE.

the three logging categories. Decreases (effect sizes) of 0.9%–2% in chloride concentrations, and increases of 1%–2% inIRND values, were observed in trout from the logged streamsrelative to the reference streams (Table 2). The observed effectsizes in our hematological and IRND stress responses werealso generally an order of magnitude lower than those re-ported in the salmonid stress literature for a variety of differ-ent stressors, and none of the studies we surveyed reportedsignificant differences when effect sizes approximated thosewe observed in our study (Table 2).

With respect to the effects of streamside timber harvestingon trout condition, significant relations (ANCOVA, P < 0.001)were found between log10-transformed stream-wide averagesfor trout length and mass within each logging category, withcoefficients of determination (R2 values) of 0.97, 0.99, and0.99 for old-growth, recently logged, and second-growthstreams, respectively (Fig. 5a). Furthermore, the regressionsfor all three logging categories were coincident (ANCOVA,P = 0.14 for the length × treatment interaction term and P =0.15 for the treatment effect), suggesting that there were nodifferences in trout masses across all measured lengths amongthe three categories. Lastly, ANOVA results for the length-at-age analyses revealed no significant differences in averagelengths among the three logging categories for age-1 (P = 0.16,power approx. 0.53) and age-2 (P = 0.60, power approx.0.15) trout, despite increases (effect sizes) of between 10%and 17% for both age classes in the logged categories rela-tive to the reference category (Fig. 5b).

Rainbow trout physiological stress responses: interrenalnuclear diameter responses to winter conditions

Stream temperatures at the time of winter sampling were0.6 °C for stream Leo1 and 0.5 °C for stream T2, comparedwith 5.5 °C and 8 °C for each stream, respectively, duringthe summer sampling period. Furthermore, one-way ANOVArevealed that trout interrenal nuclei from these two streamswere significantly larger (P < 0.001) in early winter (meanIRND = 6.03 µm, SE = 0.02) when compared with samplescollected during the summer (mean IRND = 5.42 µm, SE =0.01), with a difference (effect size) of approximately 11%between the two seasons. This effect size was comparable tothose reported for salmonids exposed to a variety of otherstressors (acidification, hydrocarbons, landfill leachates, an-

© 2005 NRC Canada

Mellina et al. 549

Fig. 4. Rainbow trout acute and chronic hematological andinterrenal nuclear diameter (IRND) stress responses to streamsideclear-cut logging grouped according to logging category. In panel(a), comparisons of plasma cortisol concentrations were madebetween logging categories (using only trout that were sacrificedimmediately to assess the occurrence of acute stress) and withinlogging categories (between trout that were sacrificed immediatelyand those that were handled to assess impairment of the cortisolresponse system). In panels (b) and (c), mean plasma chlorideconcentrations and IRND were compared across logging catego-ries using data from all captured trout (sacrificed immediatelyand handled). In all panels, mean values for each logging categorywere calculated using stream averages (with the number of streamswithin each category indicated along the x-axis), and error barsrepresent standard errors of the mean. Asterisks denote significantdifferences at the P < 0.001 level among the indicated comparisons.All other comparisons were nonsignificant at the P = 0.05 level.

tibiotics, and herbicides; Table 2). Although our seasonalIRND test comprised an old-growth and a recently loggedstream (which may have confounded the results), significantincreases (ANOVA, P < 0.001) of approximately 10%–11%in winter IRND values were also observed in each streamrelative to summer values.

Discussion

Over the time frame during which we conducted our sum-mer sampling, the overall consistency of our results suggestsa lack of both acute and chronic environmental stressors act-ing on rainbow trout cortisol, chloride, and IRND stress re-

sponse systems, as well as on growth (approximated by troutcondition and length-at-age), in our logged study streams.Based on previous studies conducted in our study region, weassumed that stream habitat alterations resulting from clear-cut logging practices that did not retain riparian buffer stripsand that involved in-stream work to have been severe enoughto be stressful to stream-dwelling trout. For example, Slaneyet al. (1977a) found that these practices resulted in extensivealterations to stream channels, including postlogging chronicincreases in suspended sediment loads (leading to the depo-sition of suspended solids in spawning and rearing areas,thereby reducing their availability to fish), episodic increasesof 4–9 °C in summer stream temperatures, and the creation

© 2005 NRC Canada

550 Can. J. For. Res. Vol. 35, 2005

Fig. 5. (a) Rainbow trout length – mass relations and (b) length-at-age for age-1 and age-2 classes, grouped according to old-growth,recently logged, and second-growth categories. In panel (a), log10-transformed length and mass relations were used to assess trout con-dition between logging categories, and each data point represents a stream-wide average. In panel (b), mean trout length-at-age esti-mates were calculated using stream averages, and numbers embedded in the histogram bars represent the number of streams (with thetotal number of trout in parentheses) within each logging category (old growth; recently logged; second growth). Trout ages were de-termined through otolith analysis, and all error bars represent SE.

of permanent obstructions to fish migration. These same log-ging practices also resulted in reductions in subgravel dis-solved oxygen levels and in benthic invertebrate standingcrop biomass (which can potentially reduce salmonid growth;Slaney et al. 1977b; Brownlee et al. 1988). With respect tothe potential for these logging-related impacts to be stressfulto fish, Redding and Schreck (1987) found significant in-creases in steelhead trout and coho salmon plasma cortisolconcentrations following exposure to suspended topsoil atconcentrations that were similar to those reported by Slaneyet al. (1977b) following streamside logging. Furthermore,Strange et al. (1977) found increases of approximately 200%in plasma cortisol concentrations in cutthroat trout exposedto temperature fluctuations of 10 °C, and Pickering et al.(1987) reported signs of chronic stress and reductions ingrowth in salmonids that were denied access to overheadcover. Although we found reductions in the overall avail-ability of pool habitat as well as increases in stream temper-atures in our logged streams, these changes (and any otherhabitat alterations we did not measure) were likely not se-vere enough to elicit a stress response in rainbow trout. Forexample, average daily maximum stream temperatures inour logged streams during the summer following our fieldwork ranged from approximately 11 to 18 °C, which arewell below the reported lethal temperatures of approximately25 °C for rainbow trout (Jobling 1981). Because stressfulwater temperatures are often observed in streams at lowerlatitudes than those of our study areas (Beschta et al. 1987),we speculate that summer temperatures in our streams maynot reach stressful levels for this species (even after streamsidelogging has occurred; Slaney et al. 1977b). However, thesame forecast does not necessarily hold true for a specieslike bull trout (Salvelinus confluentus), which prefer temper-atures below 15 °C and are generally less tolerant of warmwater (Bonneau and Scarnecchia 1996). Ultimately, the re-quirements of the fish species of interest will determine whetherpostlogging habitat alterations will likely be stressful.

The inconsistency in the patterns we observed in the num-ber and volume of pools may be related to the impacts oflogging on a variety of stream processes. For example,streamside logging and the deliberate cleaning of in-streamdebris can promote higher and more powerful flows, leadingto increased scour and consequently deeper pools, as well asto the creation of new pools (Slaney et al. 1977a; Lestelleand Cederholm 1984; Bilby and Ward 1991). Conversely, re-ductions in pool size may be caused by the infilling of existingpools with additional sediment that is frequently generatedby recent streamside harvesting activities (Slaney et al. 1977a;Grant et al. 1986). However, we observed increases in troutdensity and biomass in our treatment streams despite theoverall reduction in pool habitat, and this may be related tothe impacts of logging on stream productivity. For example,increases in light and nutrient levels, with concomitant in-creases in primary and secondary productivity leading to in-creased fish production, have been reported as short-term benefitsof clear-cut logging (Beschta et al. 1987). The average canopycover in our recently logged and second-growth streams wereapproximately 74% and 34% lower, respectively, than in theold-growth streams, and the observed increases in trout den-sity and biomass may therefore have resulted from increasedlight levels reaching the treatment streams, even 25–28 years

after logging took place. Therefore, over the time frame en-compassed by the timber harvesting regimes around ourtreatment streams, the benefits of increased light levels mayhave outweighed the reductions in available pool habitat,and the loss of habitat may translate into future reductions introut density and biomass once light levels reaching the streamsbecome attenuated (Sedell and Swanson 1984).

Whereas we had predicted that long-term increases instream temperature would be mitigated by the regenerationof streamside deciduous vegetation, we observed continuedhigh temperatures and diurnal fluctuations in our second-growth streams, despite canopy cover levels that were twicethose over our recently logged streams. This pattern may bedue to reductions in wind speed and evaporation, energybudget components that contribute to stream cooling, result-ing from a low, deciduous canopy (which typically occurswithin the first few decades following logging and whichdominated our second-growth streams) when compared witha high, coniferous cover (Hewlett and Fortson 1982).

Although the lack of significant differences in our hema-tological (cortisol and chloride) and IRND stress indicatorsamong logging categories may stem from our small samplesizes and the relatively high natural variability in these re-sponse variables (which contributed to the low power of ourstatistical tests), had our reported effect sizes been statisticallysignificant their biological significance would have remainedquestionable. The changes in our stress responses relative toour reference streams were modest when compared with thestudies we surveyed from the stress literature, and even thoughthe majority of the stressors in our literature survey were notrelated to clear-cut logging (e.g., handling and confinement,electroshocking, exposure to herbicides) we consider the com-parisons to be valid. For example, the 50% increase in plasmacortisol concentration observed in trout from our second-growth streams is relatively small considering the sensitivityof this indicator and the levels to which this hormone can beelevated (e.g., up to 21 000%; Barton et al. 1980). Further-more, Barton et al. (1980), Strange et al. (1977), and Reddingand Schreck (1987) reported diurnal variations in plasmacortisol concentrations of approximately 35%, 70%, and 66%,respectively, in nonstressed juvenile rainbow trout, values thatare similar to those found in our treatment streams (16%–49%).Our observed increases (859%–1222%) in plasma cortisolconcentrations as a result of the electroshocking, handling,and confinement test we applied to the handled trout werealso of the same order of magnitude as previously reportedin the literature for these same stressors, and all of the stud-ies we surveyed reported nonsignificant effects (i.e., the fishwere not stressed) when the magnitude of their cortisol,chloride, and IRND effect sizes approximated ours. Thesecomparisons therefore provide strong support that any im-pacts from clear-cut logging around our streams may nothave been stressful to trout in a biologically meaningful way.Lastly, the results of our seasonal IRND stress response testdemonstrate that this chronic stress indicator can be used toreveal the presence of a known environmental stressor (theonset of winter conditions), and further suggest that accli-mating to rapidly changing environmental conditions duringearly winter in temperate regions was more stressful to rainbowtrout than were the effects of streamside logging. McLeay(1975), in a study conducted in coastal British Columbia,

© 2005 NRC Canada

Mellina et al. 551

also reported increases of approximately 10% in IRND val-ues of stream-dwelling juvenile coho salmon in early Decemberrelative to early September, with differences in seasonal wa-ter temperatures that were similar to those in our study.

Given the lack of stress responses at the primary (cortisol)and secondary (chloride and IRND) levels, it is perhaps notsurprising that no significant effects were detected at the tertiary(condition and length-at-age) level. For example, Gregory andWood (1999) reported reductions in rainbow trout appetite,growth, condition, and food conversion efficiency after cortisollevels were chronically elevated by 200%, a level not reachedby trout that were sacrificed immediately in our logged streams.Furthermore, although growth in fish is often density de-pendent (Diana et al. 1991), the similarity in trout conditionamong our logging categories, combined with the observed10%–17% increases in length for the age-1 and age-2 classesin our logged streams, were likely not a result of density-dependent effects given the trout densities in our streams.

Our inability to detect statistically significant stress re-sponses may also stem from other factors. For example,chronic stress may lead to alterations in distribution pat-terns as well as in migration and reproductive behaviours(Wedemeyer et al. 1984), and because we restricted oursampling to nonreproductive periods and did not measuremovement patterns, we were limited in our ability to exam-ine these additional responses. Furthermore, acute stresscan be caused by brief ecological events (such as increasedsuspended sediment following rain storms or episodic in-creases in stream temperature) and be measurable only dur-ing very restricted periods that were missed by our samplingtimes. We realize our classification of streams into threebroad logging categories does not identify any particularstressor, and we assume that the rubric of “logging” inte-grates all associated impacts. However, our approach mayobviate the need to measure a wide range of biotic andabiotic factors or to sample at specific times, because thestress responses of fish should integrate any direct, negativeeffects from logging (whatever they might be) and allowfor an assessment of their physiological consequences. Forexample, if environmental stressors such as high suspendedsediment levels were present in our logged streams butwere missed during our sampling times, our cortisol, chloride,and IRND results suggest that any effects from such stress-ors were likely short lived and that they were not translatedto the tertiary stress response level (trout growth responses).

In conclusion, despite reductions in the availability of poolhabitat and increases in daily maximum stream temperaturesand in diurnal fluctuations, the overall consistency of our he-matological, IRND, and growth results, combined with themagnitude of our observed effect sizes relative to other stud-ies in the stress literature, suggest that the impacts resultingfrom the streamside clear-cut logging practices conducted 1–28 years ago around our study streams were not acutely orchronically stressful for rainbow trout during the time oursampling took place. Our seasonal IRND results also suggestthat this chronic stress indicator can be used to reveal thepresence of a known environmental stressor (the onset ofwinter conditions) and that acclimating to winter conditionsin temperate regions is more stressful to trout than are theshort- or long-term consequences of logging. Although awide variety of other stress indicators exist but were not

tested in our study (e.g., blood cell counts, plasma protein,fatty acid concentrations, and the expression of heat shockproteins; Wedemeyer et al. 1990; McKinley et al. 1993; Lundet al. 2002), the suite of indicators we used is relativelycomprehensive and encompasses all three stress responselevels. Lastly, the differences in climate, topography, soils,forest cover, and logging methods that exist between interiorand coastal regions (Slaney et al. 1977a) raise the questionof whether logging-related impacts are similar between thesetwo regions. For example, the more moderate hillsides in in-terior regions, combined with drier soils, are less prone toerosion and landslides (Carlson et al. 1990), and may helpmitigate habitat degradation due to debris torrents (Chamberlinet al. 1991). Environmental degradation resulting from clear-cut logging in interior regions may therefore be less severe,and consequently less stressful to stream-dwelling fish, andwe encourage future, similar studies that encompass differ-ent geographic regions and different fish species to enablecomparisons to be made with our results.

Acknowledgements

We thank the following people for their help, advice, andsupport in the design, execution, and discussions of thisstudy: Peter Baird, Thaddeus Seidler, Jim Houston, DaveBarnes, Dennis Martens, Bob Gordon, Ken Morton, RichElson, Nick Leone, Steve Macdonald, Erl MacIsaac, HerbHerunter, Ryan Galbraith, Lars Flemming Pedersen, KerryMullen, Vivian Magnusson, Dawna Brand, Dave Stevenson,Don Cadden, Tom Johnston, the Berg family, and the staff atthe CANFOR Fort St. James office. Financial support to EMwas provided by a post-graduate scholarship from the Natu-ral Sciences and Engineering Research Council of Canada(NSERC) and by a grant to S.G.H. from Forest RenewalBritish Columbia.

References

Andersson, T., Forlin, L., Hardig, J., and Larsson, A. 1988. Physio-logical disturbances in fish living in coastal water polluted withbleached kraft pulp mill effluents. Can. J. Fish. Aquat. Sci. 45:1525–1536.

Barton, B.A., and Schreck, C.B. 1987. Metabolic cost of acutephysical stress in juvenile steelhead. Trans. Am. Fish. Soc. 116:257–263.

Barton, B.A., Peter, R.E., and Paulencu, C.R. 1980. Plasma cortisollevels of fingerling rainbow trout (Salmo gairdneri) at rest, andsubjected to handling, confinement, transport, and stocking. Can.J. Fish. Aquat. Sci. 37: 805–811.

Barton, B.A., Schreck, C.B., and Fowler, L.G. 1988. Fasting anddiet content affect stress-induced changes in plasma glucose andcortisol in juvenile chinook salmon. Prog. Fish-Cult. 50: 16–22.

Beschta, R.L., Bilby, R.E., Brown, G.W., Holtby, L.B., andHofstra, T.D. 1987. Stream temperature and aquatic habitat: fish-eries and forestry interactions. In Streamside management: for-estry and fishery interactions. Edited by E.O. Salo and T.W.Cundy. College of Forest Resources, University of Washington,Seattle, Wash. pp. 191–232.

Bilby, R.E., and Ward, J.W. 1991. Characteristics and function oflarge woody debris in streams draining old-growth, clear-cut,and second-growth forests in southwestern Washington. Can. J.Fish. Aquat. Sci. 48: 2499–2508.

© 2005 NRC Canada

552 Can. J. For. Res. Vol. 35, 2005

Bonneau, J.L., and Scarnecchia, D.L. 1996. Distribution of juve-nile bull trout in a thermal gradient of a plunge pool in GaniteCreek, Idaho. Trans. Am. Fish. Soc. 125: 628–630.

Bowlby, J.N., and Roff, J.C. 1986. Trout biomass and habitat rela-tionships in southern Ontario streams. Trans. Am. Fish. Soc.115: 503–514.

Brand, D.G., Fink, R., Bengeyfield, W., Birtwell, I.K., andMcAllister, C.D. 2002. Salt water-acclimated pink salmon fry(Oncorhynchus gorbuscha) develop stress-related visceral lesions after10-day exposure to sublethal concentrations of the water-solublefraction of North Slope crude oil. Toxicol. Pathol. 29: 574–584.

British Columbia Ministry of Forests 1996. Channel assessmentprocedure field guidebook. Forest Practices Code of British Co-lumbia. British Columbia Ministry of Forests, Victoria, B.C.

British Columbia Ministry of Forests and British Columbia Ministryof Environment. 1995. Riparian management area guidebook.Forest practices code of British Columbia. British ColumbiaMinistry of Forests and British Columbia Ministry of Environ-ment, Victoria, B.C.

Brown, S.B., Eales, J.G., Evans, R.E., and Hara, T.J. 1984.Interrenal, thyroidal, and carbohydrate responses of rainbow trout(Salmo gairdneri) to environmental acidification. Can. J. Fish.Aquat. Sci. 41: 36–45.

Brownlee, M.J., Shepherd, B.G., and Bustard, D.R. 1988. Some ef-fects of forest harvesting on water quality in the Slim Creek wa-tershed in the central interior of British Columbia. Can. Tech.Rep. Fish. Aquat. Sci. No. 1613.

Carlson, J.Y., Andrus, C.W., and Froehlich, H.A. 1990. Woody de-bris, channel features, and macroinvertebrates of streams withlogged and undisturbed riparian timber in northeastern Oregon.Can. J. Fish. Aquat. Sci. 47: 1103–1111.

Chamberlin, T.W., Harr, R.D., and Everest, F.H. 1991. Timber har-vesting, silviculture, and watershed processes. Am. Fish. Soc.Spec. Publ. 19: 181–205.

Chilton, D.E., and Beamish, R.J. 1982. Age determination methodsfor fishes studied by the groundfish program at the Pacific Bio-logical Station. Can. Spec. Publ. Fish. Aquat. Sci. 60: 1–102.

Cone, R.S. 1989. The need to reconsider the use of condition indi-ces in fishery science. Trans. Am. Fish. Soc. 118: 510–514.

Cunjak, R.A. 1988. Physiological consequences of overwinteringin streams: the cost of acclimitization? Can. J. Fish. Aquat. Sci.45: 443–452.

Cunjak, R.A. 1996. Winter habitat of selected stream fishes andpotential impacts from land-use activity. Can. J. Fish. Aquat.Sci. 53(Suppl. 1): 267–282.

Diana, J.S., Dettweiler, D.J., and Lin, C.K. 1991. Effect of Niletilapia (Oreochromis niloticus) on the ecosystem of aquacultureponds, and its significance to the trophic cascade hypothesis.Can. J. Fish. Aquat. Sci. 48: 183–190.

Donaldson, E.M., and Dye, H.M. 1975. Corticosteroid concentra-tions in sockeye salmon (Oncorhynchus nerka) exposed to lowconcentrations of copper. J. Fish Res. Board Can. 32: 533–539.

Donaldson, E.M., Fagerlund, U.M., and McBride, J.R. 1984. Aspectsof the endocrine stress response to pollutants in salmonids. In Con-taminant effects on fisheries. Edited by V.W. Cairns, P.V. Hodson,and J.O. Nriagu. John Wiley & Sons, New York. pp. 213–221.

Dye, H.M., and Donaldson, E.M. 1974. A preliminary study ofcorticosteroid stress response in sockeye salmon to a componentof kraft pulp mill effluent dehydroabietic acid. Fish. Mar. Ser.Res. Dev. Tech. Rep. 461.

Erdfelder, E., Faul, F., and Buchner, A. 1996. GPOWER: a generalpower analysis program. Behav. Res. Methods Instrum. Comput.28: 1–11.

Farley, A.L. 1979. Atlas of British Columbia. University of BritishColumbia Press, Vancouver.

Fausch, K.D., and Northcote, T.G. 1992. Large woody debris andsalmonid habitat in a small coastal British Columbia stream.Can. J. Fish. Aquat. Sci. 49: 682–693.

Gibbons, D.R., Meehan, W.R., Koski, K.V., and Merrell, T.R. 1987.History of studies of fisheries and forestry interactions in south-eastern Alaska. In Streamside management: forestry and fisheryinteractions. Edited by E.O. Salo and T.W. Cundy. College of ForestResources, University of Washington, Seattle, Wash. pp. 297–329.

Grant, J.W.A., Englert, J., and Bietz, B.F. 1986. Application of amethod for assessing the impact of watershed practices: effectsof logging on salmonid standing crops. North Am. J. Fish. Man-age. 6: 24–31.

Gregory, S.V., Lamberti, G.A., Erman, D.C., Koski, K.V., Murphy,M.L., and Sedell, J.R. 1987. Influence of forest practices onaquatic production. In Streamside management: forestry and fish-ery interactions. Edited by E.O. Salo and T.W. Cundy. Collegeof Forest Resources, University of Washington, Seattle, Wash.pp. 233–256.

Gregory, T.R., and Wood, C.M. 1999. The effects of chronic plasmacortisol elevation on the feeding behaviour, growth, competitiveability, and swimming performance of juvenile rainbow trout.Physiol. Biochem. Zool. 72: 286–295.

Hewlett, J.D., and Fortson, J.C. 1982. Stream temperature under aninadequate buffer strip in the southeast Piedmont. Water Res.Bull. 18: 983–988.

Hicks, B.J., Hall, J.D., Bisson, P.A., and Sedell, J.R. 1991. Re-sponses of salmonids to habitat changes. Am. Fish. Soc. Spec.Publ. 19: 483–518.

Hontela, A., Rasmussen, J.B., Audet, C., and Chevalier, G. 1992.Impaired cortisol stress response in fish from environments pol-luted by PAHs, PCBs, and mercury. Arch. Environ. Contam.Toxicol. 22: 278–283.

Hurlbert, S.H. 1984. Pseudoreplication and the design of ecologi-cal field experiments. Ecol. Monogr. 54: 187–211.

Jobling, M. 1981. Temperature tolerance and the final preferendum-rapid methods for the assessment of optimum growth tempera-tures. J. Fish Biol. 19: 439–455.

Lestelle, L.C., and Cederholm, C.J. 1984. Short-term effects of or-ganic debris removal onresident cutthroat trout. In Proceedings,fish and wildlife relationships in old-growth forests symposium.Edited by W.R. Meehan, T.R. Merrell, Jr., and T.A. Hanley.American Institute of Fishery Research Biologists, Asheville,N.C. pp. 131–140.

Lisle, T.E. 1987. Using residual depths to monitor pool depths in-dependently of discharge. USDA For. Serv. Pac. Southwest For.Range Exp. Stn. Res. Note PSW-394.

Lund, S.G., Caissie, D., Cunjak, R.A., Vijayan, M.M., and Tufts,B.L. 2002. The effects of environmental heat stress on heat-shock mRNA and protein expression in Miramichi Altlantic salmon(Salmo slara) parr. Can. J. Fish. Aquat. Sci. 59: 1553–1562.

Macdonald, J.S., Scrivener, J.C., and Smith, G. 1992. The Stuart-Takla Fisheries/Forestry Interaction project: study descriptionand design. Can. Tech. Rep. Fish. Aquat. Sci. No. 1899.

Mazeaud, M.M., Mazeaud, F., and Donaldson, E.M. 1977. Primaryand secondary effects of stress in fish: some new data with ageneral review. Trans. Am. Fish. Soc. 106: 201–212.

McBride, J., Strasdine, G., and Fagerlund, U.H.M. 1975. Acutetoxicity of Kanamycin to steelhead trout (Salmo gairdneri). J.Fish. Res. Board Can. 32: 554–558.

McBride, J.R., Donaldson, E.M., and Derksen, G. 1979. Toxicityof landfill leachates to underyearling rainbow trout (Salmogairdneri). Bull. Environ. Contam. Toxicol. 23: 806–813.

© 2005 NRC Canada

Mellina et al. 553

McBride, J.R., Dye, H.M., and Donaldson, E.M. 1981. Stress re-sponse of juvenile sockeye salmon (Oncorhynchus nerka) to thebutoxyethanol ester of 2,4-dichlorophenoxacetic acid. Bull. En-viron. Contam. Toxicol. 27: 877–884.

McKinley, R.S., Singer, T.D., Ballantyne, J.S., and Power, G. 1993.Seasonal variation in plasma nonesterified fatty acids of lakesturgeon (Acipenser fulvescens) in the vicinity of hydroelectricfacilities. Can. J. Fish. Aquat. Sci. 50: 2440–2447.

McLeay, D.J. 1973. Effects of a 12-hr and 25-day exposure to kraftpulp mill effluent on the blood and tissues of juvenile cohosalmon (Oncorhynchus kisutch). J. Fish Res. Board Can. 30:395–400.

McLeay, D.J. 1975. Variations in the pituitary–interrenal axis andthe abundance of circulating blood-cell types in juvenile cohosalmon, Oncorhynchus kisutch, during stream residence. Can. J.Zool. 53: 1882–1891.

Meehan, W.R. (Editor). 1991. Influences of forest and rangelandmanagement on salmonid fishes and their habitats. Am. Fish.Soc. Spec. Pub. 19.

Muhlfeld, C.C., Bennett, D.H., and Marotz, B. 2001. Summer hab-itat use by Columbia River redband trout in the Kootenai Riverdrainage, Montana. North Am. J. Fish. Manage. 21: 223–235.

Murphy, M.L., and Hall, J.D. 1981. Varied effects of clear-cut log-ging on predators and their habitat in small streams of the Cas-cade Mountains, Oregon. Can. J. Fish. Aquat. Sci. 38: 137–145.

Pickering, A.D. 1981. Stress and fish. Academic Press, New York.Pickering, A.D., Griffiths, R., and Pottinger, T.G. 1987. A compari-

son of the effects of overhead cover on the growth, survival, andhaematology of juvenile Atlantic salmon, Salmo salar L., browntrout, Salmo trutta L., and rainbow trout, Salmo gairdneri Rich-ardson. Aquaculture, 66: 109–124.

Ralph, S.C. 1990. Timber/fish/wildlife stream ambient monitoringfield manual. Centre for Streamside Studies, University ofWashington, Seattle, Wash. TFW-16E-90–004.

Redding, J.M., and Schreck, C.B. 1987. Physiological effects oncoho salmon and steelhead of exposure to suspended solids.Trans. Am. Fish. Soc. 116: 737–744.

Salo, E.O., and Cundy, T.W. (Editors). 1987. Streamside manage-ment: forestry and fishery interactions. Institute of Forest Re-sources, University of Washington, Seattle, Wash.

Schreck, C.B., Whaley, R.A., Bass, M.L., Maughan, O.E., andSolazzi, M. 1976. Physiological responses of rainbow trout (Salmogairdneri) to electroshock. J. Fish Res. Board Can. 33: 76–84.

Sedell, J.R., and Swanson, F.J. 1984. Ecological characteristics ofstreams in old-growth forests of the Pacific Northwest. In Pro-ceedings, fish and wildlife relationships in old-growth forestssymposium. Edited by W.R. Meehan, T.R. Merrell, Jr., and T.A.Hanley. American Institute of Fishery Research Biologists,Asheville, N.C. pp. 9–16.

Slaney, P.A., Smith, H.A., and Halsey, T.G. 1977a. Physical alter-ations to small stream channels associated with streamside log-

ging practices in the central interior of British Columbia. BritishColumbia Ministry of Environment, Fish and Wildlife Branch,Fisheries Management Report No. 31.

Slaney, P.A., Halsey, T.G., and Smith, H.A. 1977b. Some effects offorest harvesting on salmonid rearing habitat in two streams inthe central interior of British Columbia. British Columbia Min-istry of Environment, Fish and Wildlife Branch, Fisheries Man-agement Report No. 71.

Solazzi, M.F., Nickelson, T.E., Johnson, S.L., and Rodgers, J.D.2000. Effects of increasing winter rearing habitat on abundanceof salmonids in two coastal Oregon streams. Can. J. Fish. Aquat.Sci. 57: 906–914.

Strange, R.J., and Schreck, C.B. 1978. Anesthetic and handlingstress on survival and cortisol concentration in yearling chinooksalmon (Oncorhynchus tshawytscha). J. Fish. Res. Board Can.35: 345–349.

Strange, R.J., Schreck, C.B., and Golden, J.T. 1977. Corticoidstress response to handling and temperature in salmonids. Trans.Am. Fish. Soc. 106: 213–218.

Sumpter, J.P., Dye, H.M., and Benfey, T.J. 1986. The effects ofstress on plasma ACTH, α-MSH, and cortisol levels in salmonidfishes. Gen. Comp. Endocrinol. 62: 377–385.

Thomas, R.E., Gharrett, J.A., Carls, M.G., Rice, S.D., Moles, A.,and Korn, S. 1986. Effects of fluctuating temperature on mortal-ity, stress, and energy reserves of juvenile coho salmon. Trans.Am. Fish. Soc. 115: 52–59.

Wedemeyer, G. 1973. Some physiological aspects of sublethal heatstress in the juvenile steelhead trout (Salmo gairdneri) and cohosalmon (Oncorhynchus kisutch). J. Fish. Res. Board Can. 30:831–834.

Wedemeyer, G.A., McLeay, D.J., and Goodyear, C.P. 1984. As-sessing the tolerance of fish and fish populations to environmentalstress: The problems and methods of monitoring. In Contami-nant effects on fisheries. Edited by V.W. Cairns, P.V. Hodson,and J.O. Nriagu. John Wiley & Sons, New York. pp. 163–196.

Wedemeyer, G.A., Barton, B.A., and McLeay, D.J. 1990. Stressand acclimation. In Methods for fish biology. Edited by C.B.Schreck and P.B. Moyle. American Fisheries Society, Bethesda,Md. pp. 451–490.

Woodward, C.C., and Strange, R.J. 1987. Physiological stress re-sponses in wild and hatchery reared rainbow trout. Trans. Am.Fish. Soc. 116: 574–579.

Wydoski, R.S., Wedemeyer, G.A., and Nelson, N.C. 1976. Physio-logical response to hooking stress in hatchery and wild rainbowtrout (Salmo gairdneri). Trans. Am. Fish. Soc. 105: 601–606.

Appendix A

Appendix appears on following page.

© 2005 NRC Canada

554 Can. J. For. Res. Vol. 35, 2005

© 2005 NRC Canada

Mellina et al. 555

Stream

Studysectionlength(m)

No. pools·100 m–2

Mean poolvolume (m3)

Total poolhabitat(m3·100 m–2)

Trout density(no. fish·(m3 ofpool habitat)–1)

Trout biomass(g·(m3 of poolhabitat)–1)

Avg. dailymaximumtemperature(°C)

Avg. dailytemperaturefluctuation(°C)

717.5 110 2.9 5.1 14.8 0.9 9.5 11.6 1.8Leo1 128 2.7 1.3 3.3 2.5 20.1 15.6 2.4RB4 275 5.4 1.5 8.0 1.6 8.3 11.7 2.1Airport 250 4.3 1.2 5.2 0.4 3.0 — —Muskeg 280 7.0 1.4 9.5 1.3 4.2 11.2 2.4Willow6 293 5.5 0.6 3.5 2.6 13.5 13.6 2.5SK7 275 4.1 0.9 3.8 0.5 2.9 11.4 2.2727 590 5.6 0.8 4.7 0.4 14.2 14.4 4.9Salmon1 230 5.3 0.5 2.4 4.7 22.2 13.6 3.918 Mile 620 1.4 0.3 0.4 7.0 112.0 18.0 10.7T2 67 13.0 1.1 14.3 0.9 19.2 12.4 5.2Thursday 275 2.0 0.9 1.7 2.3 21.4 12.2 3.4Saw 237 1.1 1.5 1.7 2.9 18.8 — —Holonauh 200 1.7 7.1 12.2 0.5 3.7 15.7 3.3Willow5 226 4.5 1.3 6.0 0.8 7.0 13.6 2.6

Note: The average summertime (1 June – 30 Septemper 1997) daily maximum temperature and daily temperature fluctuation (daily maximum – mini-mum) for all streams except Airport and Saw are also included.

Table A1. Study section length, number of pools, mean pool volume, total pool habitat, and rainbow trout density and biomass esti-mates for each of the streams sampled in this study.

© 2005 NRC Canada

556 Can. J. For. Res. Vol. 35, 2005

No.

ofca

ptur

edtr

out

Tro

utw

etm

ass

(g)

Tro

utfo

rkle

ngth

(cm

)C

orti

sol

(nm

ol·L

–1)

Str

eam

Tota

lS

HM

ean

Max

.M

in.

Mea

nM

ax.

Min

.S

HC

hlor

ides

(mm

ol·L

–1)

IRN

D( µ

m)

Len

gth

atag

e1

(cm

)L

engt

hat

age

2(c

m)

717.

559

2930

13.9

79.2

4.7

10.5

(0.3

)19

.07.

536

.50

(6.3

4)68

4.78

(84.

84)

140.

70(2

.63)

5.64

(0.0

4)8.

0(0

.4)

9.3

(0.2

)L

eo1

5829

2912

.253

.23.

710

.2(0

.3)

16.7

7.0

22.1

3(2

.03)

581.

49(4

7.85

)14

2.46

(2.0

8)5.

43(0

.03)

;6.

01(0

.08)

8.7

(0.3

)11

.6(0

.4)

RB

430

1515

11.4

34.7

4.6

10.2

(0.3

)15

.07.

920

.68

(2.1

0)53

5.07

(31.

63)

141.

78(3

.19)

5.51

(0.0

6)8.

6(0

.1)

10.8

(0.2

)A

irpo

rt16

88

7.4

24.1

3.6

8.3

(0.3

)12

.56.

7—

——

5.50

(0.0

8)7.

9(0

.2)

11.5

(1.1

)M

uske

g31

1615

5.5

10.9

3.1

8.0

(0.2

)9.

76.

621

.14

(2.6

5)37

2.04

(30.

15)

130.

94(5

.61)

5.69

(0.0

4)8.

0(0

.2)

—W

illo

w6

2914

1512

.530

.82.

610

.0(0

.4)

14.4

6.5

114.

65(3

3.05

)67

1.20

(63.

78)

134.

82(1

.81)

5.56

(0.0

6)9.

4(0

.3)

12.4

(0.3

)S

K7

116

56.

111

.64.

08.

0(0

.3)

10.0

6.9

——

—5.

60(0

.08)

7.8

(0.3

)10

.0(-

)72

761

3229

21.8

45.9

3.6

12.2

(0.2

)15

.96.

659

.25

(16.

26)

490.

34(2

5.87

)14

1.19

(2.0

4)5.

74(0

.04)

10.5

(0.3

)13

.6(0

.2)

Sal

mon

131

1615

7.1

11.7

3.8

8.6

(0.2

)10

.56.

930

.46

(2.9

0)51

0.10

(32.

08)

129.

50(5

.74)

5.84

(0.0

4)8.

5(0

.2)

10.4

(-)

18M

ile

2312

1124

.167

.42.

912

.0(0

.7)

18.3

6.2

60.1

3(2

0.54

)70

0.10

(67.

83)

135.

47(1

.86)

5.75

(0.0

6)10

.2(0

.2)

14.3

(1.1

)T

221

1110

21.1

54.6

6.5

12.2

(0.4

)16

.08.

0—

——

5.41

(0.0

6);

6.04

(0.0

7)10

.3(0

.7)

12.6

(0.4

)T

hurs

day

3015

1510

.545

.04.

79.

3(0

.4)

16.4

7.5

23.6

6(4

.51)

466.

43(2

8.06

)13

3.57

(2.6

0)5.

50(0

.06)

8.3

(0.1

)11

.8(1

.1)

Saw

2111

1029

.312

9.8

10.4

12.6

(0.7

)21

.89.

646

.43

(12.

64)

791.

42(9

3.73

)13

7.07

(1.4

3)5.

80(0

.08)

12.1

(0.5

)—

Hol

onau

h30

1515

10.6

59.6

1.4

8.3

(0.7

)17

.24.

912

4.58

(52.

97)

626.

97(5

2.71

)13

8.21

(4.1

8)5.

55(0

.06)

10.1

(0.5

)15

.1(1

.6)

Wil

low

530

1515

10.7

65.2

3.6

9.4

(0.4

)17

.96.

561

.36

(19.

35)

571.

01(7

3.79

)13

8.82

(1.8

1)5.

61(0

.06)

8.9

(0.3

)9.

9(1

.4)

Not

e:St

ream

aver

ages

for

the

sum

mer

hem

atol

ogic

alan

din

terr

enal

nucl

ear

diam

eter

(IR

ND

)st

ress

resp

onse

vari

able

s,as

wel

lm

ean

leng

th-a

t-ag

efo

rth

eag

e-1

and

age-

2cl

asse

s,ar

eal

soin

clud

ed.

Num

bers

ofca

ptur

edtr

out

and

plas

ma

cort

isol

conc

entr

atio

nsar

efu

rthe

rdi

vide

din

togr

oups

repr

esen

ting

trou

tth

atw

ere

sacr

ific

edim

med

iate

ly(S

)an

dth

ose

that

wer

eha

ndle

d(H

).Fo

rst

ream

sL

eo1

and

T2,

the

seco

nden

trie

sin

the

IRN

Dco

lum

nre

pres

ent

mea

nva

lues

for

IRN

Dco

llect

eddu

ring

the

win

ter

seas

on.

Bla

nkce

llsde

note

stre

ams

for

whi

chda

taw

ere

not

avai

labl

e,an

dva

lues

inpa

ren-

thes

esre

pres

ent

SE.

No

vari

ance

estim

ates

wer

eca

lcul

ated

for

mea

nle

ngth

-at-

age

for

the

age-

2cl

ass

inst

ream

sSK

7an

dSa

lmon

1be

caus

eon

lya

sing

lefi

shw

asca

ptur

edin

this

age-

clas

sin

each

ofth

ese

stre

ams.

Tab

leA

2.To

tal

num

ber

and

mas

san

dfo

rkle

ngth

mea

sure

men

tsof

trou

tca

ptur

edin

each

stre

am.