ARTICLE

Seasonal and spatial fluctuations in Oncorhynchus trout diet in atemperate mixed-forest watershed1

Judith L. Li, William J. Gerth, Richard P. Van Driesche, Doug S. Bateman, and Alan T. Herlihy

Abstract: To examine seasonal and spatial factors affecting prey consumption by Oncorhynchus trout, we examined trout dietfrom mainstem and tributary sites at Hinkle Creek, Oregon. Benthic invertebrate densities were similar across seasons and didnot differ between tributaries and the mainstem. Fluctuations in diet followed seasonal changes in invertebrate sizes andabundances. Average prey biomass consumed was positively correlated with fish size. Consumption rates were high in springand summer but fell significantly in fall when fewer and smaller prey were eaten. A switch in consumption from 36% terrestrialprey biomass in spring to 85% in summer coincided with an increase in terrestrial prey size and a decrease in benthic prey size.Location within the watershed also affected prey consumption. Despite similarities in diet composition, tributary trout con-sumed somewhat more biomass than trout in the mainstem but grew relatively slower. Because stream fishes such asOncorhynchus trout feed opportunistically on varied prey, studies incorporating multiple seasons and stream types are importantto understanding energy exchanges between terrestrial and aquatic ecosystems.

Résumé : Afin d’examiner les facteurs saisonniers et spatiaux qui ont une incidence sur la consommation de proies par lestruites Oncorhynchus, nous nous sommes penchés sur le régime alimentaire de truites dans des sites du bras principal etd’affluents du ruisseau Hinkle, en Oregon. Les densités d’invertébrés benthiques étaient semblables d’une saison a l’autre et nevariaient pas entre les affluents et le bras principal. Les fluctuations du régime alimentaire suivaient des variations saisonnièresde la taille et de l’abondance des invertébrés. La biomasse de proies moyenne consommée était positivement corrélée a la tailledes poissons. Les taux de consommation étaient élevés au printemps et a l’été, mais baissaient de manière significative al’automne, période où les proies consommées étaient plus petites et moins nombreuses. La biomasse consommée est passée de36 % a 85 % de proies terrestres du printemps a l’été, ce changement coïncidant avec une augmentation de la taille des proiesterrestres et une diminution de la taille des proies benthiques. L’emplacement dans le bassin versant avait également uneincidence sur la consommation de proies. Bien que la composition de leurs régimes alimentaires ait présenté des similitudes, lestruites des affluents consommaient plus de biomasse que les truites du bras principal, mais leur croissance était moins rapide.Parce que les poissons d’eau courante comme les truites Oncorhynchus se nourrissent de manière opportuniste de proies variées,les études qui intègrent plusieurs saisons et types de cours d’eau sont importantes pour la compréhension des échangesénergétiques entre les écosystèmes terrestres et aquatiques. [Traduit par la Rédaction]

Fisheries biologists and fishermen have long known that sal-monids feed on a combination of benthic invertebrates, emergentaquatic adults, and terrestrial invertebrates (Dimick and Mote1934; Allen 1951; Edwards and Huryn 1995). As mobile trout growfrom juvenile to adult stages, their behavior and preferences forhabitat and food resources will change, just as the types andsources of prey are changing (Northcote 1997; Bramblett et al.2002). The phenologies of lotic insects tend to flux synchronously,leading to seasonal peaks of abundance and emergence of adults(Ward and Stanford 1982; Sweeney 1984). The timing of these bio-logical fluxes, reflecting physiological and genetic responses tophysical conditions, varies with annual and regional patterns ofhydrology, temperature, and other climate-related phenomena(Hogg and Williams 1996; Harper and Peckarsky 2006; Li et al.2011). Similarly, terrestrial insects become more active and possi-bly more abundant in warm seasons of the year in associationwith plant phenologies and terrestrial life history patterns. Theways in which these fluxes occur likely provide critical resourceoptions for opportunistic fishes (Nakano and Murakami 2001;

Kawaguchi and Nakano 2001); however, many salmonid diet stud-ies have been restricted to late summer or autumn (Syrjänen et al.2011), and the number of empirical studies examining seasonalfluctuations are few (Wipfli and Baxter 2010).

Terrestrial invertebrates potentially represent cross-ecosystemcontributions from multiple sources. Ground-dwelling inverte-brates such as millipedes, some beetles, or spiders might bewashed in from the stream edge, while plant-dwelling terrestrialsmay fall directly from adjacent riparian vegetation. Others derivefrom upstream and arrive in surface drift. As many terrestrialinsects are winged and mobile, they may be dispersed by wind orfly in from an even wider range. The variety of terrestrial preyconsumed by fish reflects this diversity of cross-system origins.

While salmonids feed opportunistically on the surface, in thewater column, and benthos (Syrjänen et al. 2011; Allan et al. 2003),they can also show selectivity for prey. They are visual predators,and their size-selectivity makes larger, more visible prey mostvulnerable (Allan 1981). Social status can confer advantages tolarger, dominant fish that outcompete smaller fish for prime feed-

Received 8 November 2015. Accepted 6 July 2016.

J.L. Li, W.J. Gerth, and A.T. Herlihy. Department of Fisheries and Wildlife, Oregon State University, 104 Nash Hall, Corvallis, OR 97331, U.S.A.R.P. Van Driesche. Department of Integrative Biology, Oregon State University, Corvallis, OR 97331, U.S.A.D.S. Bateman. Department of Forest Engineering, Oregon State University, Corvallis OR 97331, U.S.A.Corresponding author: William J. Gerth (email: william.gerth@oregonstate.edu).1This article is part of the special issue “Cross-ecosystem resource subsidies: from land to water and back again”.Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

1

Can. J. Fish. Aquat. Sci. 00: 1–8 (0000) dx.doi.org/10.1139/cjfas-2015-0520 Published at www.nrcresearchpress.com/cjfas on 27 July 2016.

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

ing locations (Nakano et al. 1999). These advantages can result inlarger fish catching the larger prey (Syrjänen et al. 2011). In concertwith seasonal changes in aquatic and terrestrial prey, changes introut consumption may also reflect availability of the largest prey.

For the delivery of terrestrial invertebrates, close proximity ofriparian canopy and vegetation has been shown to be importantin a great diversity of streams around the world; the highest infallrates are often in deciduous canopies (Mason and MacDonald1982; Chloe and Garman 1996; Syrjänen et al. 2011). Some of thehighest rates of terrestrial prey consumption by salmonids havebeen reported in temperate deciduous forests (Wipfli 1997;Kawaguchi and Nakano 2001; Romero et al. 2005), particularly byOncorhynchus trout species (see review by Syrjänen et al. 2011). Inthe Pacific Northwest, many Douglas-fir forests that were cut inthe mid-20th century have grown back to 40- to 60-year-old stands.Riparian trees in those forests are often a mix of conifers withdeciduous alders and maples. At the time of our study, HinkleCreek watershed in the southern Cascade Mountains of Oregonhad such a forest with relatively homogenous riparian canopies.

Our objective was to examine seasonal and spatial factors af-fecting prey consumption by coastal cutthroat trout, Oncorhynchusclarkii clarkii, and steelhead, Oncorhynchus mykiss irideus, in HinkleCreek. We hypothesized that the types of prey consumed wouldfluctuate with seasonal changes in abundance and size of preyfrom aquatic and terrestrial sources. Secondly, we hypothesizedthat the amount of prey consumed would be positively associatedwith increasing fish size. In our spatial comparison between trib-utaries and the mainstems, we hypothesized that warmer temper-atures in the mainstems would result in higher consumption andgreater fish growth in those mainstem sites.

Methods

Study sitesHinkle Creek is a tributary stream in the Umpqua River drain-

age, located in the foothills of the Cascade Mountains of south-western Oregon, USA. The land within the Hinkle watershed isentirely privately owned and managed for timber production; a50- to 60-year-old second growth Douglas-fir forest was dominantthroughout the watershed at the start of this study. The watershedis composed of two sub-basins (North Fork and South Fork; Fig. 1)and drains 19.4 km2. For this fish diet study, there were six main-stem sampling sites (four on South Fork, two on North Fork) andfive sites on fish-bearing tributaries (three in South Fork, two inNorth Fork). Colleagues also working in the Hinkle watershed(Kibler et al. 2013) provided water temperature data from foursites (two mainstem and two tributary) from which data werecollected year-round during the course of our study.

Field collection and sample processingWe collected trout for stomach contents by electrofishing, ex-

cept in spring in 2004 when fish were collected by hook and line.Salmonids ≥ 80 mm fork length were categorized as either coastalcutthroat trout (O. clarkii clarkii) or steelhead (O. mykiss irideus). MostO. mykiss in this study were anadromous as determined by obser-vation and tagging, so we refer to them as “steelhead”, althoughwe recognize that resident forms, i.e., rainbow trout, can co-occur(McMillan et al. 2012; Tattam et al. 2013). Differentiating thesetrout species based on external characteristics was not reliable forsmaller individuals; therefore, trout < 80 mm fork length wereconsidered as a single group of “unknown species.” In 2004, 71%were cutthroat trout, 16.3% were steelhead, and 13% were un-known. In 2005, all trout ≥ 80 mm fork length in the tributariesand 90% in the mainstem were coastal cutthroat; the remaining10% of mainstem trout were steelhead. Fish were collected only atsites where there was a strong likelihood of obtaining 20 fish fordiet analyses; because previous fish sampling indicated low trout

densities at certain locations, we did not collect fish diet samplesin all of the tributaries (Fig. 1).

We compared diet between seasons in 2004 at five mainstemsites (Fig. 1); diet samples were collected on 28 April – 4 May(spring), 30 August (summer), and 14 and 22 October (fall). Spatialdifferences were examined explicitly during the next year, in thespring of 2005 (2–5 May) between tributary and mainstem sites.Ranges of distances between tributaries and closest mainstemsites were from 387 to 826 m in North Fork and from 606 to 1706 min South Fork (Fig. 1).

After fish at least 60 mm in fork length were captured, theywere anesthetized, and fork length and wet mass were recorded.Fish stomach contents were gently flushed out into a large metaltray using a squirt bottle filled with water. Contents of the traywere poured through a funnel containing a small coffee filter thatretained the entire sample. After stomach samples were collected,fish were held in buckets with stream water until they recuper-ated and then were returned to the stream. Optimally, 20 fish persite, or as many fish as could be collected when densities were low,were sampled for diet at each site per date. Monte Carlo simula-tions were run to explore the number of fish needed to attainvarying levels of precision in predicting prey consumption. Theseconsisted of 1000 simulations for composites of 54 spring, 83 sum-mer, and 68 fall mainstem samples.

Invertebrate prey from stomach contents preserved in 70% eth-anol were identified and enumerated in the laboratory. Each preyitem was measured so that length–mass regressions could be usedto determine biomass per prey item (Hodar 1996; Benke et al.1999; Sabo et al. 2002). If invertebrate bodies were not whole, headwidths were used in regressions based on whole organisms in gutsamples to calculate body lengths. Identifications were made tothe finest level possible, typically family or order. Prey organismswere also classified according to their sources: benthic (meaningbottom or stream water column), flying adults of aquatic, or ter-restrial. We omitted a few fish from calculations of diet becausewe did not have suitable length–mass regression coefficients for afew types of prey organisms in the diet of those fish (4% of fish in2004, 6% of fish in 2005) or because their stomachs were empty (5%of fish in 2004, <1% of fish in 2005).

Benthic invertebrates were collected prior to fish sampling ateach sampling site in 2004 and 2005 (Fig. 1). Sampling for seasonalcomparisons occurred in mainstem sites on 29 April, 10–12 Au-

Fig. 1. Map of Hinkle Creek watershed, Oregon, USA, showinginvertebrate and fish sampling sites. Fish and invertebrates werecollected at mainstem sites in 2004 and 2005 (gray circles) and attributary sites in 2005 (black diamonds); black crosses indicatewhere year-round water temperature data were recorded.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

2 Can. J. Fish. Aquat. Sci. Vol. 00, 0000

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

gust, and 11 October 2004. For comparisons between tributariesand mainstem sites, benthic samples were taken from 5–12 April2005. Sites were 100 m long in mainstems and 30 m long in trib-utaries. Six randomly located samples per site were collected withSurber nets. Benthic invertebrate samples were processed in thelab by combining the six stream-bottom samples (total area sam-pled = 0.54 m2) from each site and subsampling randomly with agridded sieve (Caton 1991) to get a minimum count of 500 organ-isms. Invertebrates were identified to the lowest practical taxo-nomic level and enumerated. In general, insects were identified togenus except for chironomids and ceratopogonids, which wereonly identified to subfamily or tribe. Snails were identified to genus.Other non-insects were identified to coarser taxonomic levels.

Additional fish sampling was also performed to assess troutgrowth in mainstems and tributaries. Single-pass electrofishing(Bateman et al. 2005) was conducted annually during late-summer, low-discharge periods (15 August – 15 September) to cap-ture trout in all fish-bearing portions of the stream network. Uponcapture, fish were anesthetized, and fork length and wet masswere recorded. All trout ≥ 100 mm were surgically implanted witha 23 mm half-duplex (HDX) passive integrated transponder (PIT)tag (Texas Instruments, Inc., Dallas, Texas, USA) following proce-dures described by Bateman and Gresswell (2006). Subsequently,all trout ≥ 100 mm captured were scanned with an Allflex (AllflexUSA, Inc., Dallas, Texas, USA) handheld PIT-tag scanner, and anyuntagged trout ≥ 100 mm found were implanted with PIT tags.After handling, all fish were allowed to recover (defined by up-right swimming) in an aerated bucket of stream water and thenwere returned to their location of capture. Growth was calculatedfor PIT-tagged fish that were recaptured in subsequent annualsampling events. We used data from 2003–2005 to compare trib-utary and mainstem growth; data from all mainstem fish wereused, but we only included tributary fish data from tributarieswhere we collected fish diet samples. In addition, through the useof gate readers at tributary junctions, we were able to monitor fishmovement year-round. Relatively few tagged fish moved betweentributaries and mainstems. Only growth data from fish that livedexclusively in mainstems or tributaries were compared. Growthdata from more mobile fish that moved between tributaries andmainstems were excluded, because it would be difficult to inter-pret how habitat affected growth for these individuals.

AnalysesDifferences in gut content mass among seasons and between

tributary and mainstem sites were analyzed using analysis of co-variance (ANCOVA). We ran two separate one-way ANCOVAs usingmixed models, one for season as the fixed effect (using only main-stem 2004 data) and one for tributary versus mainstem as thefixed effect (using only spring 2005 data). In each model, fish masswas used as the covariate; site location was treated as a randomeffect. Fish mass and gut content mass were both skewed andwere log10-transformed for this and all subsequent statistical anal-yses.

Overall differences in diet composition patterns were examinedwith nonmetric multidimensional scaling (NMS) ordination. Twoseparate ordinations were run to examine seasonal (2004) andspatial (spring 2005) patterns. Prey biomass (log-transformed) wasused for these analyses, in which prey were categorized by orderand source (benthic, adult aquatics, terrestrial). To identify inver-tebrate prey orders and sources distinctive for fish consumptionin the seasonal analysis, we calculated indicator values (IV)(Dufrene and Legendre 1997). These values quantified fidelity of aprey category to a season (Bij) and specificity of that prey categoryto a season (Aij) using the equation IVij = Aij × Bij × 100.

In contrast to other analyses in which we used data on a per fishbasis, to compare changes in prey size across seasons and by preytype, we used biomass data from all individuals of each prey typein each season, without considering which fish they came from.

Seasonal differences in average biomass per prey type were testedwith a two-way ANOVA using log-transformed biomass data. Logtransformation corrected highly skewed data distributions.

Relative growth rates of tributary versus mainstem trout(change in length (mm) / initial length (mm) / time between sam-pling events (years)) were analyzed using a mixed model one-wayANOVA. Site location within tributary or mainstem was treated asa random effect.

Differences in benthic invertebrate densities between seasons(in 2004) were tested using a one-way analysis of variance(ANOVA). To compare spring 2005 benthic densities in tributariesversus mainstems, a t test was used. For all of these analyses, datawere log-transformed to normalize data.

Results

Seasonal variationInitially, our goal for sampling was to capture 20 fish at each

site. We were able to obtain diet from an average of only 14.7 fishper site. According to Monte Carlo simulations, a composite totalof only 20 fish for the entire mainstem would have been needed inspring or summer to estimate the proportion of terrestrial preyconsumed (standard deviation ± 5%); a composite size of morethan 25 fish would have been required in fall. These outcomessuggest that our seasonal sampling provided robust estimates ofprey composition.

In general, the amount of prey biomass consumed increasedwith individual fish biomass (ANCOVA: F[1,189] = 38.47, p < 0.001;Fig. 2a). Fish collected in spring weighed more (median mass =21.3 g) than those collected in summer and fall (median masses of13.6 g and 11.0 g, respectively). The wider ranges of fish sizes andsmaller median sizes of fish sampled for diet in summer and fallresulted from young-of-the-year trout too small to be sampled inspring that became sufficiently large to be sampled later in theyear. After accounting for this trend in fish size, prey consump-tion declined significantly in the fall (back-transformed leastsquare mean (LSM) = 4.95 mg) compared with spring and summer(back-transformed LSM = 33.43 mg and 33.64 mg, respectively)(ANCOVA: F[2,189] = 59.99, p < 0.001). There was no significant dif-ference between prey biomass consumed in spring and summer.

Average masses of benthic invertebrates, adult aquatics, andterrestrial invertebrates found in trout diet differed significantlybetween seasons (ANOVA: F[2,4619] = 22.99, p < 0.001; Fig. 3) andbetween prey types (ANOVA: F[2,4619] = 210.93, p < 0.001); a signifi-cant interaction between these factors revealed that differencesbetween prey types varied with season (ANOVA: F[4,4619] = 62.89,p < 0.001). Notably, benthic invertebrates were smaller than otherprey types in spring and summer, while average masses of adultaquatics and terrestrial invertebrates overlapped in those seasons.Benthic prey were larger in spring than in summer or fall. Terres-trial prey were largest in summer and smallest in fall when adultaquatics were significantly larger than terrestrial or benthic in-vertebrates.

Trout diet switched in relative proportions of prey types from amixture of benthics (mean = 45% of biomass, 67% of abundance),terrestrials (mean = 36% of biomass, 19% of abundance), and adultaquatics (mean = 15% of biomass, 10% of abundance) in spring toprimarily terrestrials in summer (mean = 85% of biomass, 62% ofabundance) (Fig. 4a). Aquatic adult biomass comprised a smallproportion of prey biomass overall. The switch in diet composi-tion, driven by high proportions of terrestrial prey in summer,was also apparent in NMS analysis (Fig. 5a). In the fall, whenoverall consumption rates were low, almost half of the diet wasmade up of terrestrial prey (mean = 48% of biomass).

Significant seasonal differences in prey composition were gen-erally described by indicator species analysis (Table 1). Aquaticswere important in the spring when benthic and adult Ephemer-optera, Plecoptera, and Diptera and benthic Trichoptera were in-

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

Li et al. 3

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

dicative of diet. Terrestrial Coleoptera and Diptera were alsocharacteristic of spring samples. Terrestrial Hymenoptera weregood indicators in summer (indicator variable (IV) = 61); 892 antscomprised 45% of terrestrial prey biomass consumed that season.Indicator values for Homoptera, Hemiptera, and Isoptera in sum-mer were lower than for Hymenoptera, in part, because they wereconsumed by only a portion of the fish in summer.

Examining biomass per taxa in each season revealed other con-tributors to diet among non-indicator taxa, particularly less abun-dant but heavier prey. While the high numbers of ants wereimportant in summer diet, less numerous, larger ants in thespring (average = 12.1 mg, n = 36) made up 38% of total terrestrialprey. In summer, larval Plecoptera contributed 27.5% of benthicbiomass eaten, and adult Trichoptera (n = 30) comprised 86% oftotal adult aquatics. Forty-three caddisflies made up 97% of adultaquatics in fall. Small Collembola (n = 59) were only fairly goodindicators of fall samples (IV = 25.7) when overall consumptionwas much lower than in other seasons. Benthic Ephemeroptera,Trichoptera, and Diptera comprised 85% of benthic prey biomassconsumed in fall when benthic invertebrates constituted 39% ofprey biomass eaten. Occasionally big, uncommon prey (e.g., gerridwater striders) in the summer and Diplopoda (millipedes) andOrthoptera (namely cave crickets) in the fall contributed dispro-portionately large portions of total biomass consumed. Con-versely, small invertebrates such as aquatic Diptera in spring (n =407) and terrestrial Collembola in fall constituted very small por-tions of total biomass consumed.

Spatial variationAverage prey consumed increased with fish size in spring 2005

(Fig. 2b). We noted that fish in tributaries tended to consume morethan fish of the same size in the mainstem (Fig. 2b), and as a group,tributary trout weighed less than trout in the mainstem (medianmasses = 9.6 g and 12.9 g, respectively). After accounting for thetrend in fish size, fish in the tributaries tended to consume morethan those in in the mainstem sites (back-transformed LSM =23.07 mg and 10.08 mg, respectively) (ANCOVA: F[1,134] = 4.22,p = 0.06). However, relative growth rates of trout at tributary sites

Fig. 2. Relationships between diet biomass of individual troutversus trout biomass compared (a) across seasons, 2004, and(b) between tributary and mainstem sites, spring 2005.

Fig. 3. Comparison of prey type sizes in trout gut samples fromspring, summer, and fall of 2004. Graph is the interaction plotrelated to two-way ANOVA showing mean values and 95% Tukeyconfidence intervals. Note: y axis is log-scale.

Fig. 4. Comparisons of average percent benthic, adult aquatic,terrestrial, and unknown prey biomass consumed per fish at(a) mainstem sites in spring, summer, and fall of 2004 and(b) tributaries and mainstem sites in spring 2005.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

4 Can. J. Fish. Aquat. Sci. Vol. 00, 0000

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

(LSM = 0.178 mm·mm−1·year−1) were somewhat lower than atmainstems (LSM = 0.249 mm·mm−1·year−1) (ANOVA: F[1,124] = 4.75,p = 0.06; Fig. 6). Benthic invertebrates comprised >60% of preybiomass consumed in both tributary and mainstem sites (Fig. 4b),and there was considerable diet overlap among fishes in the twostream types (NMS; Fig. 5b).

Mean daily water temperatures were, on average, 1.5 °C warmerin mainstems than in tributaries during summer (Fig. 7a). How-ever, temperature differences between mainstems and tributarieswere less pronounced at other times of the year. As a result, aver-age cumulative degree-days accumulated over a year at tributarysites (3315.6) differed by about 6% from those accumulated atmainstem sites downstream (3515.0). In both mainstems and trib-utaries, mean daily temperatures (Fig. 7a) and 7-day moving aver-ages of the daily maximum water temperatures (Fig. 7b) remainedbelow 18 °C.

Benthic invertebratesIn samples collected contemporaneously with trout diet, ben-

thic densities were slightly higher in summer (mean = 2678·m−2)than in spring (mean = 2155·m−2) and fall (mean = 1980·m−2), butthere were no significant differences between seasons (ANOVA:F[2,15] = 2.3, p = 0.14). In spring 2005, benthic invertebrate densitieswere slightly, but not significantly, higher in the mainstems

Fig. 5. Nonmetric multidimensional scaling (NMS) ordination of preybiomass consumed by individual trout. Prey per fish is represented byinvertebrate orders and prey types (benthic, adult aquatic, terrestrial,unknown). Two axes of three-dimensional ordinations are shown.Vectors indicate strong correlations (r < 0.6) with fish gut metrics.(a) Spring, summer, and fall, 2004; axis 1, R2 = 0.219; axis 2, R2 = 0.195;axis 3, R2 = 0.298; total, R2 = 0.712. (b) Mainstem + tributaries, spring2005; axis 1, R2 = 0.309; axis 2, R2 = 0.338; axis 3, R2 = 0.146; total, R2 = 0.792.

Table 1. Indicator values (IV) for invertebrates consumed by trout inthree seasons (p < 0.01).

Spring Summer Fall

Order IV Order IV Order IV

Benthic Ephemeroptera 58.7Trichoptera 52.4Diptera 44.1Plecoptera 34.7

Adult Ephemeroptera 58.6

Aquatic Plecoptera 31.5Diptera 20.6

Terrestrial Coleoptera 41.0 Hymenoptera 61.0 Collembola 25.7Diptera 40.5 Homoptera 30.9

Hemiptera 25.4Isoptera 22.9

Note: Indicator values (IV) can range from 0 to 100, where a value of 100expresses perfect indication (taxon is found in all samples from that season anddoes not occur in any samples from other seasons).

Fig. 6. Annual relative growth rates (2003–2005) for recaptured, pit-tagged, age 1+ trout in Hinkle Creek comparing fish that reared intributaries versus mainstems. The relatively few fish movingbetween stream types were excluded. Numbers within each boxare sample sizes.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

Li et al. 5

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

(mean = 2559·m−2) than in the tributaries (mean = 2040·m−2) (t =1.25, p = 0.23).

DiscussionCutthroat trout and steelhead at Hinkle Creek were opportunis-

tic predators, eating prey from aquatic and terrestrial sourcesthroughout the year. By collecting an average of 15 diet samplesper site, we were able to overcome the high variability in fish dietsand develop robust measures of prey consumption as confirmedby Monte Carlo simulations. In our study, diet biomass was posi-tively related to increasing fish size. The seasonal dynamics ofprey availability and size were key to fluctuations in fish diet,particularly for fish switching from a mixed diet in spring toterrestrial prey in the summer and for fish consuming dramati-cally less prey in the fall.

Invertebrate patterns of phenology were apparent in our sea-sonal comparisons of fish diet. Typical life history patterns ofbenthic invertebrates were detected among benthic prey con-sumed. In spring, as many taxa prepared to emerge, benthic lar-vae were larger than at other times of the year. Terrestrialinvertebrates were largest in the summer, and greater activitylevels likely contributed to their increased availability to fish. Byautumn, diversity and numbers of terrestrial prey consumed de-clined dramatically, when benthic invertebrates remained small.Seasonal changes among invertebrates translated into what wasavailable and most vulnerable to fish predation.

Terrestrial prey were important components of trout diet in allseasons at Hinkle Creek, increasing from spring to summer (i.e.,from 36% to 85% of biomass consumed). In Japan, consumption bysalmonids switched from very low to high levels of terrestrial prey

Fig. 7. Hinkle Creek water temperatures from two mainstem and two tributary sites: (a) mean daily water temperatures from 4/1/2004(month/day/year) through 3/31/2005; and (b) 7-day moving averages of the daily maximum water temperatures during summer 2004.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

6 Can. J. Fish. Aquat. Sci. Vol. 00, 0000

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

(Nakano and Murakami 2001; Kawaguchi and Nakano 2001), coin-ciding asynchronously with a dramatic decrease in benthicavailability. At Hinkle Creek, trout also switched to higher con-sumption of terrestrial invertebrates in summer even thoughbenthic invertebrate densities did not change from spring to sum-mer. A decrease in benthic prey abundance was not likely theprimary mechanism for the switch; rather, preference for larger,easily detectable terrestrial prey probably played a role. Benthicswere significantly smaller in the summer, and trout selectedlarger terrestrial prey, particularly ants and termites. Althoughthis study was not designed to assess the rate of terrestrial infall,the fish response suggests that a seasonal pulse of terrestrial preywas likely. Greater visibility of terrestrial insects in the drift, es-pecially patchily distributed larger individuals, and greater easeof capture with less energetic cost likely favored terrestrial preyconsumption. Adult aquatics were among the largest prey in anyseason but were never the dominant prey consumed. Thus webelieve availability, in addition to size and visibility, also played arole in trout selection.

One of the strongest seasonal signals detected in our study wasa dramatic decrease in prey eaten by trout in the fall. Water tem-peratures for the October 2004 sampling were comparable withthose in early summer and were not extremely warm or cold.Other abiotic factors might have been in play such as reductionsin suitable habitat prior to fall rains. However, the very small sizeof prey consumed (e.g., Collembola and small Ephemeroptera) incombination with the lack of large taxa that had been available inother seasons suggest that resource scarcity was likely. Fish mor-tality at Hinkle Creek was also high in the fall (Berger andGresswell 2009). In coastal streams of the Pacific Northwest, foodcan be limiting to cutthroat trout in the summer (Boss andRichardson 2002), but in the southern Cascades, trout weighedless and ate significantly less. Prey appeared to be most lim-ited in the critically dry autumn. Fish did eat benthic mayflies andtrue flies in the fall, but terrestrial prey (including spiders, occa-sional millipedes, and cave crickets) comprised almost 50% of thediet during that season. Terrestrial invertebrates were mortalitiesresulting from infall to the stream; therefore, predation by fishhad little or no effect on terrestrial communities. Conversely, theenergetic resources provided by terrestrial prey appeared to becritical for fish survival in the autumn.

Experiments in Japan suggested that terrestrial subsidies in thesummer could buffer effects of fish predation, allowing benthicpopulations to recover or grow while fish switch to terrestrialprey (Nakano et al. 1999; Kawaguchi et al. 2003; Baxter et al.2005). Because benthic abundances in summer at Hinkle did notdiffer from spring, trout switching to terrestrial prey may result ina similar buffering effect. Reduced trout predation on aquaticprey may increase availability to other aquatic insectivores(Baxter et al. 2005) such as salamanders, riparian spiders (Marczakand Richardson 2007), and birds (Wright et al. 2000).

To understand apparent contradictions between relativelyhigher consumption rates and slower growth rates in Hinkle trib-utaries, we considered differences in habitat availability and theenergetics of foraging. The proportions of age-0 fish lavaged attributary and mainstem sites were relatively the same (18% and16%, respectively); thus, it was unlikely that differences in ageclasses explained differences in size. Spring through fall watertemperatures at Hinkle Creek occurred within a 10–18 °C range,providing a large scope for growth (Dwyer and Kramer 1975) andminimizing stress from excessively warm water temperatures(Boyd and Sturdevant 1997). Though cumulative degree-days intributaries were only 200 degree-days lower than those at mains-tem sites, when temperature differences were greatest in mid-summer, digestion rates might have been lower in tributaries,leading to lower growth rates. Models developed by Railsback andRose (1999) for O. mykiss predicted that growth is more affected byfood consumption than temperature, but at Hinkle Creek tribu-

taries, trout sizes and growth rates were lower than in the main-stem, despite moderately higher rates of food consumption. Webelieve that bioenergetics models (sensu Arismendi et al. 2013;Penaluna et al. 2015) could help us determine if small temperaturedifferences, particularly in the summer, were important to growth.

Other habitat characteristics can also affect trout growth(Rosenfeld et al. 2000). During an enclosure study in British Co-lumbia, Canada, age-0 and older cutthroat trout had highergrowth rates in pools relative to riffles; age-0 fish experiencedpositive growth in riffles, whereas older fish tended to lose mass(Rosenfeld and Boss 2001). Trout surveyed in Hinkle Creek tribu-taries also lived in higher gradient, swifter reaches where therewere fewer pools for refuge or feeding stations compared with themainstem. These habitat associations suggest that tributary troutwere smaller, potentially because they spent more energy feedingin high-gradient riffles compared with fish downstream. Dietcomposition was similar in tributaries and mainstems, but troutin the tributaries ate more than those in bigger reaches at appar-ently at higher energetic cost. In our springtime comparison, hy-drologic and thermal differences in habitats probably createdgreater demands for prey consumption and may have resulted inrelatively slower growth.

The riparian vegetation in the managed forest of the HinkleCreek watershed, located in the temperate, mesic climate ofsouthern Oregon, was a mix of Douglas-fir, alder, and maple over-story. As in homogeneously deciduous canopies noted elsewhere(Mason and MacDonald 1982; Allan et al. 2003; Syrjänen et al.2011), the forest’s capacity for invertebrate production was high.Differences in riparian diversity and composition, even within aregion, can influence the availability and timing of invertebratelitterfall. In a study of three widely separated coastal Oregonstreams (Romero et al. 2005), sites varied in riparian species, in-cluding Sitka spruce and Douglas-fir conifers, red alder, and big-leaf maple deciduous species. At those sites, the peak in terrestrialprey consumption occurred in the fall, whereas at Hinkle Creek,terrestrial prey clearly contributed more to summer trout diet.Both the warmer, more southern location and more contiguous,homogenous deciduous riparian vegetation at Hinkle likely con-tributed to the higher availability of terrestrial prey in the sum-mer and prey scarcity in the fall. These contrasts suggest that thedynamics of seasonal patterns are likely watershed-specific, andthe timing of prey resource limitations and fish vulnerabilityprobably varies with a wide range of conditions, including cli-mate, landscape morphology, and forest type.

Because Oncorhynchus trout are opportunistic, they consumeboth aquatically and terrestrially derived prey throughout theyear. In our study, summer reductions in benthic prey size coin-cided with timely fluxes of terrestrial invertebrates that probablyprovided a temporal buffer for benthic organisms to grow. Sea-sonal fluctuations in fish consumption were driven by prey abun-dance and prey size, especially as trout switched from benthic toterrestrial prey in summer. Identifying seasonal patterns within awatershed is important not only for describing seasonal highs,but also for site-specific timing of resource lows and criticalstress periods. Behavioral adaptations such as increased foodconsumption in higher gradient, cooler tributaries may notfully compensate for more demanding environmental condi-tions. Multiseasonal, watershed-wide studies provide the mostcomplete approach to understanding the fluctuations and dynam-ics between these fish predators and their prey resources.

AcknowledgementsWe are grateful to the Hinkle Creek landowner, Roseburg For-

est Products, for providing the space and time needed for long-term studies there. Funding was provided by the OregonDepartment of Forestry Fish and Wildlife Program and the Water-shed Research Cooperative at Oregon State University. We espe-cially appreciated the water temperature data provided by

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

Li et al. 7

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.

A. Simmons and A. Skaugset, the technical assistance fromJ. Sobota, D. Hockman-Wert, and I. Arismendi, the sampling assis-tance from K. Whitehead, J. King, A. Olegario, D. Leer, andS. Clark, and the project coordination by A. Skaugset.

ReferencesAllan, J.D. 1981. Determinants of diet of brook trout (Salvelinus fontinalis) in a

mountain stream. Can. J. Fish. Aquat. Sci. 38(2): 184–192. doi:10.1139/f81-024.Allan, J.D., Wipfli, M.S., Caouette, J.P., Prussian, A., and Rodgers, J. 2003. Influ-

ence of streamside vegetation on inputs of terrestrial invertebrates to sal-monid food webs. Can. J. Fish. Aquat. Sci. 60(3): 309–320. doi:10.1139/f03-019.

Allen, K.R. 1951. The Horokiwi Stream: a study of a trout population. New Zea-land Department of Fisheries Bull. No. 10.

Arismendi, I., Johnson, S.L., Dunham, J.B., and Haggerty, R. 2013. Descriptors ofnatural thermal regimes in streams and their responsiveness to change inthe Pacific Northwest of North America. Freshw. Biol. 58: 880–894. doi:10.1111/fwb.12094.

Bateman, D.S., and Gresswell, R.E. 2006. Survival and growth of age-0 steelheadafter surgical implantation of 23-mm passive integrated transponders.N. Am. J. Fish. Manage. 26: 545–550. doi:10.1577/M05-111.1.

Bateman, D.S., Gresswell, R.E., and Torgersen, C.E. 2005. Evaluating single-passcatch as a tool for identifying spatial pattern in fish distribution. J. Freshw.Ecol. 20(2): 335–345. doi:10.1080/02705060.2005.9664974.

Baxter, C.V., Fausch, K.D., and Saunders, W.C. 2005. Tangled webs: reciprocalflows of invertebrate prey link streams and riparian zones. Freshw. Biol. 50:201–220. doi:10.1111/j.1365-2427.2004.01328.x.

Benke, A.C., Huryn, A.D., Smock, L.A., and Wallace, J.B. 1999. Length–mass rela-tionships for freshwater macroinvertebrates in North America with particu-lar reference to the southeastern United States. J. N. Am. Benthol. Soc. 18:308–343. doi:10.2307/1468447.

Berger, A.M., and Gresswell, R.E. 2009. Factors influencing coastal cutthroattrout (Oncorhynchus clarkii clarkii) seasonal survival rates: a spatially continu-ous approach within stream networks. Can. J. Fish. Aquat. Sci. 66(4): 613–632.doi:10.1139/F09-029.

Boss, S.M., and Richardson, J.S. 2002. Effects of food and cover on the growth,survival, and movement of cutthroat trout (Oncorhynchus clarki) in coastalstreams. Can. J. Fish. Aquat. Sci. 59(6): 1044–1053. doi:10.1139/f02-079.

Boyd, M., and Sturdevant, D. 1997. The scientific basis for Oregon’s stream tem-perature standard: common questions and straight answers. Oregon Depart-ment of Environmental Quality, Salem, Oregon.

Bramblett, R.G., Bryant, M.D., Wright, B.E., and White, R.G. 2002. Seasonal use ofsmall tributary and main-stem habitats by juvenile steelhead, coho salmon,and Dolly Varden in a southeastern Alaska drainage basin. Trans. Am. Fish.Soc. 131: 498–506. doi:10.1577/1548-8659(2002)131<0498:SUOSTA>2.0.CO;2.

Caton, L.W. 1991. Improved subsampling methods for the EPA “Rapid Bioassess-ment” benthic protocols. N. Am. Benthol. Soc. Bull. 8: 317–319.

Chloe, W.W., and Garman, G.C. 1996. The energetic importance of terrestrialarthropod inputs to three warm-water streams. Freshw. Biol. 36: 104–114.doi:10.1046/j.1365-2427.1996.00080.x.

Dimick, R.E., and Mote, D.C. 1934. A preliminary survey of the food of Oregontrout. Agricultural Experiment Station, Oregon State Agricultural Collegeand Oregon State Game Commission, Station Bulletin No. 323.

Dufrene, M., and Legendre, P. 1997. Species assemblages and indicator species:the need for a flexible asymmetrical approach. Ecol. Monogr. 67: 345–366.doi:10.1890/0012-9615(1997)067[0345:SAAIST]2.0.CO;2.

Dwyer, W.P., and Kramer, R.H. 1975. The influence of temperature on scope foractivity in cutthroat trout, Salmo clarki. Trans. Am. Fish. Soc. 104: 552–554.doi:10.1577/1548-8659(1975)104<552:TIOTOS>2.0.CO;2.

Edwards, E.D., and Huryn, A.D. 1995. Annual contribution of terrestrial inverte-brates to a New Zealand trout stream. N.Z. J. Mar. Freshw. Res. 29: 467–477.doi:10.1080/00288330.1995.9516680.

Harper, M.P., and Peckarsky, B.L. 2006. Emergence cues of a mayfly in a high-altitude stream ecosystem: potential response to climate change. Ecol. Appl.16: 612–621. doi:10.1890/1051-0761(2006)016[0612:ECOAMI]2.0.CO;2. PMID:16711048.

Hodar, J.A. 1996. The use of regression equations for estimation of arthropodbiomass in ecological studies. Acta Oecol. 17: 421–433.

Hogg, I.D., and Williams, D.D. 1996. Response of stream invertebrates to a global-warming thermal regime: an ecosystem-level manipulation. Ecology, 77:395–407. doi:10.2307/2265617.

Kawaguchi, Y., and Nakano, S. 2001. Contribution of terrestrial invertebrates tothe annual resource budget for salmonids in forest and grassland reaches of

a headwater stream. Freshw. Biol. 46: 303–316. doi:10.1046/j.1365-2427.2001.00667.x.

Kawaguchi, Y., Taniguchi, Y., and Nakano, S. 2003. Terrestrial invertebrate inputsdetermine the local abundance of stream fishes in a forested stream. Ecology,84(3): 701–708. doi:10.1890/0012-9658(2003)084[0701:TIIDTL]2.0.CO;2.

Kibler, K.M., Skaugset, A., Ganio, L.M., and Huso, M.M. 2013. Effect of contem-porary forest harvesting practices on headwater stream temperatures: initialresponse of the Hinkle Creek catchment, Pacific Northwest, USA. For. Ecol.Manage. 310: 680–691. doi:10.1016/j.foreco.2013.09.009.

Li, J.L., Johnson, S.L., and Sobota, J.B. 2011. Three responses to small changes instream temperature by autumn-emerging aquatic insects. J. N. Am. Benthol.Soc. 30: 474–484. doi:10.1899/10-024.1.

Marczak, L.B., and Richardson, J.S. 2007. Spiders and subsidies: results from theriparian zone of a coastal temperate rainforest. J. Anim. Ecol. 76: 687–694.doi:10.1111/j.1365-2656.2007.01240.x. PMID:17584374.

Mason, C.F., and MacDonald, S.M. 1982. The input of terrestrial invertebratesfrom tree canopies to a stream. Freshw. Biol. 12: 305–311. doi:10.1111/j.1365-2427.1982.tb00624.x.

McMillan, J.R., Dunham, J.B., Reeves, G.H., Mills, J.S., and Jordan, C.E. 2012.Individual condition and stream temperature influence early maturation ofrainbow and steelhead trout, Oncorhynchus mykiss. Environ. Biol. Fishes, 93:343–355. doi:10.1007/s10641-011-9921-0.

Nakano, S., and Murakami, M. 2001. Reciprocal subsidies: dynamic interdepen-dence between terrestrial and aquatic food webs. Proc. Natl. Acad. Sci. U.S.A.98: 166–170. doi:10.1073/pnas.98.1.166. PMID:11136253.

Nakano, S., Miyasaka, H., and Kuhara, N. 1999. Terrestrial–aquatic linkages:riparian arthropod inputs alter trophic cascades in a stream food web. Ecol-ogy, 80: 2435–2441. doi:10.1890/0012-9658(1999)080[2435:TALRAI]2.0.CO;2.

Northcote, T.G. 1997. Potamodromy in Salmonidae — living and moving in the fastlane. N. Am. J. Fish. Manage. 17: 1029–1045. doi:10.1577/1548-8675(1997)017<1029:PISAMI>2.3.CO;2.

Penaluna, B.E., Railsback, S.F., Dunham, J.B., Johnson, S., Bilby, R.E., andSkaugset, A.E. 2015. The role of the geophysical template and environmentalregimes in controlling stream-living trout populations. Can. J. Fish. Aquat.Sci. 72(6): 893–901. doi:10.1139/cjfas-2014-0377.

Railsback, S.F., and Rose, K.A. 1999. Bioenergetics modeling of stream troutgrowth: temperature and food consumption effects. Trans. Am. Fish. Soc.128: 241–256.

Romero, N., Gresswell, R.E., and Li, J.L. 2005. Changing patterns in coastal cut-throat trout (Oncorhynchus clarki clarki) diet and prey in a gradient of deciduouscanopies. Can. J. Fish. Aquat. Sci. 62(8): 1797–1807. doi:10.1139/f05-099.

Rosenfeld, J.S., and Boss, S. 2001. Fitness consequences of habitat use for juvenilecutthroat trout: energetic costs and benefits in pools and riffles. Can. J. Fish.Aquat. Sci. 58(3): 585–593. doi:10.1139/f01-019.

Rosenfeld, J., Porter, M., and Parkinson, E. 2000. Habitat factors affecting theabundance and distribution of juvenile cutthroat trout (Oncorhynchus clarki)and coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 57(4): 766–774.doi:10.1139/f00-010.

Sabo, J.L., Bastow, J.L., and Power, M.E. 2002. Length–mass relationships foradult aquatic and terrestrial invertebrates in a California watershed. J. N. Am.Benthol. Soc. 21: 336–343. doi:10.2307/1468420.

Sweeney, B.W. 1984. Factors influencing life-history patterns of aquatic insects.In The ecology of aquatic insects. Edited by V.H. Resh and D.M. Rosenberg.Praeger Scientific, New York. pp. 56–100.

Syrjänen, J., Korsu, K., Louhi, P., Paavola, R., and Muotka, T. 2011. Stream sal-monids as opportunistic foragers: the importance of terrestrial invertebratesalong a stream-size gradient. Can. J. Fish. Aquat. Sci. 68(12): 2146–2156. doi:10.1139/f2011-118.

Tattam, I.A., Ruzycki, J.R., Li, H.W., and Giannico, G.R. 2013. Body size andgrowth rate influence emigration timing of Oncorhynchus mykiss. Trans. AmFish. Soc. 142(5): 1406–1414. doi:10.1080/00028487.2013.815661.

Ward, J.V., and Stanford, J.A. 1982. Thermal responses in the evolutionary ecol-ogy of aquatic insects. Annu. Rev. Entomol. 27: 97–117. doi:10.1146/annurev.en.27.010182.000525.

Wipfli, M.S. 1997. Terrestrial invertebrates as salmonid prey and nitrogensources in streams: contrasting old-growth and young-growth riparian for-ests in southeastern Alaska, U.S.A. Can. J. Fish. Aquat. Sci. 54(6): 1259–1269.doi:10.1139/f97-034.

Wipfli, M.S., and Baxter, C.V. 2010. Linking ecosystems, food webs, and fishproduction: subsidies in salmonid watersheds. Fisheries, 35: 373–387. doi:10.1577/1548-8446-35.8.373.

Wright, K.K., Bruner, H., Li, J., Jarvis, R., and Dowlan, S. 2000. The distribution,phenology and prey of harlequin ducks, Histrionicus histrionicus, in a Cascademountain stream, Oregon. Can. Field-Nat. 114: 187–195.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

8 Can. J. Fish. Aquat. Sci. Vol. 00, 0000

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y O

rego

n St

ate

Uni

vers

ity o

n 08

/25/

16Fo

r pe

rson

al u

se o

nly.