J. N. Am. Benthol. Soc., 2000, 19(1):145-157© 2000 by The North American Benthological Society

Food resources of stream macroinvertebrates determined bynatural-abundance stable C and N isotopes and a

15N tracer addition

PATRICK J. MuLHOLLAND1, JENNIFER L. TANKZ, DIANE M. SANZONE3,WILFRED M. WOLLHEIM", BRUCE J. PETERSON*, JACKSON R. WEBSTERS, AND

JUDY L. MEYER3

1Environmental Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge,Tennessee 37831-6036 USA. E-mail: mulhollandpj@ornl.gov

department of Natural Resources and Environmental Sciences, University of Illinois, Urbana,Illinois 61801 USA

institute of Ecology, University of Georgia, Athens, Georgia 30602-2602 USA^Ecosystems Center, Marine Biological Laboratory, Woods Hole, Massachusetts 02543 USA^Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg,

Virginia 24061-0406 USA

Abstract. Trophic relationships were examined using natural-abundance 13C and 15N analyses anda 15N-tracer addition experiment in Walker Branch, a Ist-order forested stream in eastern Tennessee.In the 15N-tracer addition experiment, we added 15NH4 to stream water over a 6-wk period in earlyspring, and measured 15N:UN ratios in different taxa and biomass compartments over distance andtime. Samples collected from a station upstream from the 15N addition provided data on natural-abundance 13C:12C and 15N:14N ratios. The natural-abundance 15N analysis proved to be of limitedvalue in identifying food resources of macroinvertebrates because 15N values were not greatly differentamong food resources. In general, the natural-abundance stable isotope approach was most usefulfor determining whether epilithon or detritus were important food resources for organisms that mayuse both (e.g., the snail Elimia davaeformis), and to provide corroborative evidence of food resourcesof taxa for which the 15N tracer results were not definitive. The 15N tracer results showed that themayflies Stenonema spp. and Baetis spp. assimilated primarily epilithon, although Baetis appeared toassimilate a portion of the epilithon (e.g., algal cells) with more rapid N turnover than the bulk poolsampled. Although Elimia did not reach isotopic equilibrium during the tracer experiment, applicationof a N-turnover model to the field data suggested that it assimilated a combination of epilithon anddetritus. The amphipod Gammarus minus appeared to depend mostly on fine benthic organic matter(FBOM), and the coleopteran Anchytarsus bicolar on epixylon. The caddisfly Diplectrona modesta ap-peared to assimilate primarily a fast N-tumover portion of the FBOM pool, and Simuliidae a fast N-tumover component of the suspended particulate organic matter pool rather than the bulk poolsampled. Together, the natural-abundance stable C and N isotope analyses and the experimental 15Ntracer approach proved to be very useful tools for identifying food resources in this stream ecosystem.

Key words: stream macroinvertebrates, food resources, N cycling, tracer addition, stable isotope,'5N, 13C

Over the past 2 decades, stable isotope ratios against the heavier isotope during assimilationhave been used increasingly to identify food at each higher trophic level. Isotope tracer ad-web linkages and organic matter sources in eco- dition experiments involve addition of elementssystems. In particular, 13C:12C and 15N:14N ratios highly enriched in the heavy isotope (e.g., 15N)appear to be most useful for identifying trophic and detection of changes in isotope ratios of or-relationships (Peterson and Fry 1987, KJing ganic matter in response to tracer addition once1994). isotopic equilibrium is achieved.

Stable isotope food web investigations are Both stable isotope approaches have beengenerally of 2 types. Natural-abundance stable used to determine trophic relationships inisotope studies rely on variation in the isotope streams. Rounick et al. (1982) used variations inratios of elements in plants from different en- natural abundance of stable C isotopes to showvironments or different photosynthetic path- that allochthonous sources of organic C sup-ways and relatively constant fractionation ported invertebrate production in forested

145

146 P. J. MULHOLLAND ET AL. [Volume 19

streams, whereas autochthonous C sources weremore important food resources for invertebratesin streams in clearcut catchments. Similarly, Ro-senfeld and Roff (1992) used natural-abundance13C measurements to show a dependence of in-vertebrates on autochthonous sources of organicmatter in unforested and forested streams inspring in southern Ontario. Hamilton et al.(1992) used natural-abundance 13C and 15N mea-surements to determine that the predominantfood resource for many invertebrates and fishesin the Orinoco River floodplain was microalgaerather than the more abundant and conspicuousaquatic vascular plants. Angradi (1994) usednatural-abundance 13C and 15N measurements toshow that there were 3 trophic levels in the low-er Colorado River, with fish being supported byautochthonous production in 1 tributary but ri-parian or upland organic matter inputs in othertributaries. In a variation of the natural-abun-dance approach, Peterson et al. (1993) usedshifts in the fractionation of N stable isotopesresulting from N fertilization to alter the stableisotope ratios of algae and, in turn, documentthe increased trophic importance of epilithic al-gae in the Kuparuk River, Alaska, food web af-ter fertilization.

The use of stable isotope tracer addition ex-periments in food web studies is a more recentdevelopment. Hall (1995) added 13C-labeled ac-etate to a forested headwater stream to identifythe role of bacteria as a food resource for inver-tebrates. Peterson et al. (1997) used 15N contentsof insects during a 15NH4 tracer addition exper-iment in the Kuparuk River to show the differ-ences in food utilization between grazers andfilter-feeders. Hall et al. (1998) used a similar15NH, addition approach in Hugh White Creek,North Carolina, and found evidence for selec-tive assimilation of microbial versus detrital Nby several invertebrate detritivores. Hughes etal. (2000) added 15NO3 to an oligohaline reachof the Parker River estuary in Massachusetts todetermine the relative importance of planktonicand benthic primary producers and detritus de-rived from the surrounding marsh in fueling thefood web. Last, Wollheim et al. (1999) showedhow a mass balance model of N flow could beused with field 15N tracer results to more clearlyidentify food web relationships.

In this study, we used both natural-abun-dance stable C and N isotope and experimental15N tracer addition approaches to identify food

web relationships in Walker Branch, a Ist-orderforested stream in eastern Tennessee. We showhow this combination of stable isotope ap-proaches provides additional and more defini-tive information on trophic relationships thaneither approach alone. This experiment was partof the Lotic Intersite Nitrogen Experiment(LINX), a large multi-site project on N cyclinginvolving identical I5N tracer additions tostreams throughout the United States.

Study Site

The study was conducted during early spring1997 in a 125-m reach of the West Fork of WalkerBranch (hereafter referred to as Walker Branch),a Ist-order, forested stream in the Ridge andValley geophysical province of eastern Tennes-see (lat 35°58'N, long 84°17'W). The climate istypical of the humid Appalachian region of thesoutheastern USA, with a mean annual temper-ature of 14.5°C and mean annual precipitationof 140 cm. Although precipitation (mostly asrain) is distributed relatively evenly throughoutthe year, stream baseflow discharge is highlyseasonal because of high rates of evapotranspi-ration during the growing season. The WestFork of Walker Branch drains a 38.4 ha catch-ment at the Oak Ridge National EnvironmentalResearch Park. The vegetation is dominated byoak (Quercus spp.), hickory (Carya spp.), red ma-ple (Acer rubrum), and yellow-poplar (Lirioden-dron tulipifera). The catchment is underlain bydolomite, and the stream arises from severalsprings discharging water from the bedrockaquifer (Mulholland 1992). The stream gradientis ~0.035 m/m. Stream water is moderately al-kaline (alkalinity of 2-3 meq/L) and the pH isslightly basic (usually 8.0-8.3).

At the time of the year when this study wasconducted, primary production is at the annualmaximum in Walker Branch, although total me-tabolism is dominated by heterotrophy with P:R ratios <0.5 (Marzolf et al. 1994, Mulhollandet al. 2000). Invertebrate biomass is dominatedby the pleurocerid snail Elimia davaeformis(~95% of invertebrate biomass), although sev-eral insect and other macroinvertebrate taxa canbe numerically abundant in some seasons (New-bold et al. 1983, J. G. Smith, ORNL, unpublisheddata).

2000] STREAM MACROINVERTEBRATE FOOD RESOURCES USING STABLE ISOTOPES 147

Methods

Beginning on 1 April 1997 and continuing for6 wk, a solution of I5N-enriched NH4C1 (10%15N) was pumped from a stream-side carboyinto a constricted section of the stream toachieve rapid mixing. The 15N addition was de-signed to achieve a 50% increase in the 13N:UNratio in the dissolved NH4 pool (based on anassumed discharge rate of 10 L/s and a NH4

concentration of 3 (xg N/L). The I5N additionresulted in a very small increase in the concen-tration of NH4 in stream water (-0.05 p,g N/L).The 15N input was maintained at a rate of ~0.20mg 15N/h throughout the 6-wk period, exceptfor a 12 to 16 h period on 28 to 29 April whenthe pump malfunctioned and the supply of 1SNto the stream was interrupted. The actual I5Naddition rate was determined based on the vol-ume remaining in the carboy when the 15NH4

solution was replenished each week.Samples of water, organisms, and different

types of benthic organic matter were collectedfor 15N measurements at a station ~10 m up-stream from the 15N input and used for natural-abundance analysis of 15N and I3C. A small wa-terfall (~0.5 m in height) 2 m upstream fromthe 15N input prevented upstream movement ofmost organisms from the I5N-enriched reach tothe station used for natural-abundance analysis.Samples for 15N were also collected at stationslocated at distances of 10 m, 25 m, 50 m, 75 m,and either 100 m or 125 m downstream fromthe 15N addition at approximately weekly inter-vals during the addition (at about mid-morningon each sampling date). Because of the largenumber of samples collected and the high ana-lytical costs, the samples at all stations were an-alyzed for I5N only on the last sampling date(day 42) of the experiment. Samples collected at25 m on each sampling date were analyzed for15N to provide a time course of 15N content foreach sample type Some samples from other sta-tions and dates were analyzed for 15N to provideadditional data for spatial and temporal analy-ses.

We collected epilithon, bryophytes, leaves,epixylon, fine benthic organic matter (FBOM,<1 mm), suspended participate organic matter(SPOM), and macroinvertebrates at each station.Epilithon was scraped from the surface of 4 to6 randomly chosen rocks at each station using astiff-bristled brush and washed into a small vol-

ume of water. Epilithon slurries were filteredthrough precombusted glass fiber filters (What-man GF/F) within 2 h of collection. Bryophyteswere removed from rock surfaces at 4 to 6 lo-cations at each station. The distal ends of eachfrond (~1 cm) were removed and only this ma-terial was used for 15N analysis. Leaves andpieces of wood were randomly collected from 4to 6 locations at each station. The outer surfaceof wood was removed by scraping, collected,and the remainder discarded. Thus, our 15N val-ues of wood represent epixylon (wood biofilmonly). FBOM was collected by suctioning sur-face material from 4 to 6 areas of fine sedimentaccumulation at each station. FBOM slurrieswere filtered through precombusted glass fiberfilters (Whatman GF/F) within 2 h of collection.One sample of SPOM was collected at each sta-tion by filtering 2 to 4 L of streamwater throughprecombusted glass fiber filters (Whatman GF/F). Samples of epilithon, bryophytes, leaves,epixylon, and FBOM collected from all areas ateach station were combined to form 1 compositesample of each type for isotope analysis.

Approximately 20 to 30 £. dauaeformis (3-5mm shell width) were collected randomly ateach station, allowed to clear their guts over-night in the laboratory, and the soft tissue wasremoved after 30 s in a microwave oven. We col-lected other common macroinvertebrates (gen-erally 5-20 individuals at each station) by kicksampling with a D-frame net positioned justdownstream and by hand-picking organismsfrom rocks and detritus. Some macroinvertebratetaxa varied greatly in abundance over the ex-periment and we were able to collect individu-als of these taxa only on some dates and at somestations. Organisms were held overnight in thelaboratory to clear guts prior to further process-ing. Individuals of each taxon collected at eachstation were combined to form 1 compositesample

All samples were dried at 60°C for at least 2d and, with the exception of the samples on fil-ters, were ground using a Wiley mill. Subsam-ples were analyzed for 13C:I2C and I5N:I4N ratiosusing a Europa Model 20/20 isotope ratio massspectrometer at the Ecosystems Center, MarineBiological Laboratory, Woods Hole, Massachu-setts. All isotope ratios are expressed as either813C or 8I5N values (units of %>) according tothe following equation:

148 P. J. MULHOLLAND ET AL. [Volume 19

S'3C or S'5N = [(Rsample/R5tandard) - 1] X 1000

where R^ ,̂. is the 13C:12C or 15N:'4N ratio of thesample and R^,^ is the 13C:12C or 15N:14N ratioof a standard (PeeDee belemnite carbonate for8'3C, atmospheric N for 515N). Usually only 1subsample of each homogenized sample per sta-tion per date was analyzed. Where replicatesubsamples were analyzed, the mean of the rep-licate values was used. Analysis of replicates for40 samples indicated a mean difference in 815Nof 0.50%o, although 2/3 of the replicates differedby <0.5%o.

The 813C and 815N values of samples collectedfrom the station upstream from the 15N additionpoint were considered natural-abundance val-ues and used in the natural-abundance analy-ses. In the experimental 15N tracer analyses, allS15N values for samples collected at stationsdownstream from the 15N addition point werecorrected for natural-abundance levels of 15N bysubtracting the 815N value of similar samplescollected upstream from the 15N addition. Thus,all 815N values determined for the 15N tracer ex-periment at stations downstream from the 15Ninput represent only the tracer I5N content of thesample (i.e., they do not include naturally oc-curring 15N).

Results

Natural-abundance 15N and I3C analysis

The natural-abundance 15N analysis, based onan expected trophic I5N enrichment (i.e., in-crease in 815N values of an organism comparedwith its food) of +2 to +4%o (Ehleringer et al.1986, Peterson and Fry 1987, Keough et al.1996), provided only a limited amount of defin-itive information (Fig. 1A). Only epilithon andleaves had 815N values that differed by morethan 3%o, so it was difficult to discriminateamong other potential food resources. Our re-sults suggested that epilithon may be the pri-mary food resource for £. davaejbrmis but not forother taxa. Among the other presumed scrapers,Baetis spp. and Stenonema spp. appeared to beassimilating at least some detritus (probablyFBOM) because their natural-abundance 815Nvalues were lower than would be expected froma diet solely of epilithon. The natural-abundance815N values for the liverwort Porella pinnata andepilithon were very close, and thus Porella could

a>IoQ.

2 H

H

(A)

Psephenus

2 -

oa.o

/v

ra CO

-40 -38 -36 -34 -32 -30 -28 -26 -24

FIG. 1. Natural-abundance 815N (A) and 813C val-ues (B) of primary consumer taxa and their potentialfood resources in Walker Branch. Expected ranges inconsumer 815N and 813C values based on consumptionof each individual food resource are indicated by thehorizontal lines below the consumers. These lines re-flect the mean and SD of the isotope ratios for thatfood resource, and trophic isotopic enrichments of +2to +4%o for 15N and +l%o for 13C (for organisms rel-ative to their food indicated by the dotted lines). Hor-izontal lines through data points indicate 1 SD (wherelarge enough to be seen, except for epixylon, Porellapinnata, Baetis spp., and Anchy tarsus bicolor, for whichonly 1 sample was analyzed). FBOM = fine benthicorganic matter, SPOM = suspended particulate organ-ic matter.

not be ruled out as a possible food resource forElimia and the mayflies as well.

Leaves had the lowest natural-abundance815N values of all potential food resources anddid not appear to be a primary resource for anyconsumer taxa. The amphipod shredder Gam-marus minus and the coleopteran shredder An-chytarsus bicolor had natural-abundance 8 I5N val-

2000] STREAM MACROINVERTEBRATE FOOD RESOURCES USING STABLE ISOTOPES 149

ues consistent with a diet primarily of FBOM orepixylon rather than leaves.

The natural-abundance 15N analysis showedseveral surprises. Natural-abundance 815N val-ues for the coleopteran scraper Psephenus herrickiwere considerably lower than those expectedfrom a diet of either epilithon or FBOM, as con-stituted in bulk samples of these resources. Sim-uliidae had considerably lower natural-abun-dance 815N values than expected from a diet pri-marily of the bulk SPOM pool, the presumedfood resource for these filterers. The natural-abundance 815N values for SPOM and FBOMdiffered by 1.4%o, suggesting that SPOM wasnot simply the result of entrainment of theFBOM pool that we sampled (surface materialthat could be readily suctioned off the bottom).These data suggested that SPOM included sub-stantial amounts of a more highly I5N-enrichedpool, such as the epilithon.

The natural-abundance '3C analysis was pri-marily focused on determining whether thedominant scraper taxa were dependent on epi-lithon or detritus. Based on this analysis, Elimiaand Stenonema appeared to use a combinationof detritus and epilithon, although the natural-abundance 8I3C values of Baetis and Psephenuswere lower than expected from a diet of epili-thon, detritus, or a combination of both (Fig.IB). The data are consistent with a diet consist-ing primarily of Porella for these taxa, althoughPsephenus was usually found only on the under-sides of rocks.

15N tracer addition experiment

Data from the 15N tracer addition experimentprovided more definitive information on foodweb relationships than did the natural-abun-dance stable C and N isotope analyses. If or-ganisms of a specific taxon were feeding pri-marily on a particular food resource and if theirbody N turned over at relatively rapid rates (onthe order of several days), then we expect theirtracer 815N values to generally follow the tracer815N values of their food over time and spaceEarly in the experiment, organisms would likelyhave tracer 815N values considerably lower thantheir food if their N pools had not reached iso-topic equilibrium. Later in the experiment, or-ganisms should have tracer 815N values closer totheir food resources, unless their N turnoverrates are quite slow (weeks or greater). Thus,

tracer 815N values of small taxa with rapidgrowth rates (e.g., mayflies, simuliids) would beexpected to correspond closely with the tracer815N values of their food resources, but tracer815N values of larger taxa or those with lowgrowth rates (e.g., Elimia) may not.

The time course tracer 8I5N profiles at 25 m(Fig. 2A) and the longitudinal 815N profiles onday 42 (Fig. 2B) suggested that Stenonema spp.and Baetis spp. depended on epilithon as theirprimary food resource, although similarity inthe 815N profiles for epilithon and Porella did notrule out the latter as a food resource for theseorganisms. The generally higher tracer 815N val-ues for Baetis than epilithon over time and spacealso suggested that this taxon assimilated a por-tion of epilithon that acquires NH4 from watermore rapidly than the bulk epilithon sampled(e.g., actively growing algal or bacterial cells).The generally lower tracer 815N values for Sten-onema than epilithon over time at 25 m (Fig. 2A)suggested that this organism may also consumesome detritus, although this pattern was not ob-served in the longitudinal profile (Fig. 2B). Sur-prisingly, the temporal and longitudinal tracer815N profiles for Elimia and Psephenus followedthose of leaves and FBOM more closely than ofepilithon, the presumed food resource for thesescrapers (Fig. 2A,B). However, Elimia is a rela-tively large organism (average mass of nearly 1mg ash-free dry mass [AFDM] per individualfor the samples collected) with a relatively longlife cycle. Thus, tracer 815N levels for Elimia mayhave lagged well behind those of its food be-cause of its slow N turnover rate.

The ratio of tracer 8I5N in an organism to trac-er 8I5N in its food resource at the same time andstation should be equal to 1.0 if the organism Nis in isotopic equilibrium with its food (the ef-fect of trophic fractionation is eliminated be-cause we consider only the tracer 15N by sub-tracting background values as described in themethods). Organism:epilithon tracer 815N ratiosfor scrapers over time suggested that only Sten-onema was at isotopic equilibrium with respectto epilithon by the end of the 42-d experiment(Fig. 3A). SfCTi0ne/H«:epilithon tracer 815N ratiosappear to converge on ~ 1.0 over time, suggest-ing that bulk epilithon was the primary foodresource for this taxon. Boefts:epilithon tracer815N ratios >1 later in the experiment suggestedthat Baetis assimilated a more highly 15N-labeledportion of the epilithon (eg., actively growing

150 P. J. MULHOLLAND ET AL. [Volume 19

r

Z£us

120

110

100

90

80

70

60

50

40

30

20

10

0

(A) Scrapers and potentialfood sources (25 m)

10 15 20 25 30 35 40 45

Day

180-

160 -

140 -

120-

100 -

80-

60 -

40 -

20 -

0-

(B) Scrapers and potentialfood sources (day 42)

—•— Epilithon-•-• Porella

A Leaves-T- FBOM-©-• Stenonema-e-- Baetis-A-- Elimia-'v— Psephenus

-a

20 40 60 80

Distance from ̂

100 120 140 160

input (m)

FIG. 2. Time course profiles at the 25 m station (A) and longitudinal profiles on day 42 (B) of tracer 5'5Nvalues for scrapers and their potential food resources. Time course data were available for Baetis spp. andPsephenus herricki for days 7, 28, and 42 only. The day 42 value for Baetis is the average between the stations at10 m and 50 m because the sample from 25 m was lost. Each data point represents only 1 subsample ofcomposited material and has an uncertainty of approximately ±0.5%o (see methods).

algal cells on the epilithon surface) than thebulk pool sampled (Fig. 3B). It was not clearwhether the time-course trend in Eh'mia:epili-thon tracer 8I5N ratio was approaching an as-ymptote by the end of the experiment becausethe values remained relatively low throughoutthe study (Fig. 3C). If so, this would suggestthat Elimia fed primarily on a food resource(eg., detritus) that was much less highly 15N-labeled than the epilithon. Time course Psephen-us:epilirhon tracer 815N ratios suggested thatthis organism had not reached isotopic equilib-rium with respect to epilithon by the end of theexperiment (Fig. 3D).

A model of tracer 815N in Elimia was devel-oped based on a N turnover rate for this taxon(0.003 /d) determined from previous measure-ments of NH4 excretion and N content (Mul-holland et al. 1991, P. J. Mulholland, unpub-lished data) and the tracer 8:5N values of poten-tial food resources measured during this study(Fig. 4). The model predicted the tracer 815N ofElimia over time at the 25 m station assumingassimilation of epilithon, leaves, and FBOM sep-arately. Comparison of model predictions with

actual measurements of Elimia tracer 515N sug-gested that this organism relied on a combina-tion of epilithon and detritus, particularly laterin the experiment. Because leaf detritus was rel-atively scarce at this time of year (spring), ourresults probably reflected assimilation of epili-thon and FBOM, roughly in equal proportions.

Among presumed detritivore taxa, the tem-poral tracer 815N profile for G. minus indicatedthat this taxon was at or near isotopic equilib-rium by the end of the experiment and that itsprimary food resource was more likely FBOMthan leaves (Fig. 5A). The sharp declines in 515Nof FBOM and leaves on day 22 at 25 m was theresult of the storm 2 d earlier, which transport-ed more highly enriched materials downstreamand deposited unlabeled materials from up-stream and the channel margins in the experi-mental reach. The 815N of leaves and FBOM in-creased again in the period following the stormas the newly deposited material became labeled.The longitudinal tracer 815N profiles for Gam-marus were in better agreement with those ofepixylon than FBOM or leaves (Fig. 5B). Gam-marws:detritus tracer 815N ratios suggested that

2000] STREAM MACROINVERTEBRATE FOOD RESOURCES USING STABLE ISOTOPES 151

(A) Stenonema2.5 -

2.0 -

1.5 -

1.0

0.5 -

0.0

(B) Saet/s

J

30 -|

(C) Elimia

.- .

(D) Psephenus

0 10 20 30 40 50 0 10 20 30 40 50Day Day

FIG. 3. Organism:epilithon 815N ratios for scrapersover time during the 15N tracer experiment. Histo-gram heights represent means of values from all sam-pling stations on a particular date for which epilithontracer 515N values were >5%o (n = 2-5 per date). Ver-tical error bars represent 1 SD.

FBOM was the most likely primary food re-source for this taxon and that it appeared to beat approximate isotopic equilibrium with FBOMby the end of the experiment, although variationwas large reducing the certainty of this assess-ment (Fig. 6A,B).

The longitudinal 815N profile for A. bicolor wasmore similar to that for epixylon than for otherdetritus pools (Fig. 5B) and the Anchytarsus:ioodS15N ratio at the end of the experiment was clos-er to 1 for the case of epixylon than leaves (Fig.6C,D). Merritt and Cummins (1996) indicatedthat larvae in this family are long-lived (up to 3y), however, and it is likely that the tracer 815Nvalues for this organism would lag well behindthose of its food (organism food ratios <1throughout the experiment). Unfortunately, wehave no information on N turnover rate in An-chytarsus and cannot evaluate its food resourcesusing a model similar to that used for Elimia.

The longitudinal tracer S15N profile for theSimuliidae was distinctly hump-shaped andsimilar to that for SPOM with maxima observedat 25 to 40 m (Fig. 7). The hump-shaped longi-tudinal profile in SPOM 815N is the result ofgradual deposition of unlabeled SPOM from

20 40Day

FIG. 4. Values of 515N for Elimia ciavaeformis at the25 m station over time. Solid lines are model simula-tions based on consumption of single food resources(epilithon, leaves, fine benthic organic matter [FBOM])and points are measured values.

upstream, generation of highly labeled SPOM inthe upper part of the experimental reach via re-suspension of FBOM and sloughing of epilithonor direct labeling of SPOM, and gradual replace-ment of highly labeled SPOM by entrainedFBOM and epilithon farther downstream.Among the potential food resources, onlySPOM clearly had this longitudinal profileshape. Further, Simuliidae tracer 815N valueswere considerably more enriched than SPOMthroughout the study reach, suggesting assimi-lation of a portion of the SPOM that takes upNH4 from water more rapidly than the bulk ma-terial (e.g., attached microbes, algal cells).

The values and shape of the longitudinal trac-er 8'5N profile for the caddisfly Diplectrona ma-desta were intermediate between those forSPOM and FBOM (Fig. 7). Our results suggest-ed that either Diplectrona used a combination ofFBOM and SPOM, or that it assimilated a por-tion of the FBOM pool that cycled N more rap-idly than the bulk material as constituted in oursamples.

Discussion

Comparison of approaches

Each of the 2 types of analysis used to inferfood web relationships (natural-abundance sta-

152 P. J. MULHOLLAND EX AL. [Volume 19

30 -i

25 -

20 -

fin15 -

10 -

5 -

(A) Detritivores and potentialfood sources (25 m)

0 5 10 15 20 25 30 35 40 45

Day

(B) Detritivores and potentialfood sources (day 42)

—•— Leaves-•- FBOM—A— Epixylon-©-• Gammarus-S— Anchytarsus

a

0 20 40 60 80 100 120 140 160

Distance from 15NH4 input (m)

FIG. 5. Time course profiles at the 25 m station (A) and longitudinal profiles on day 42 (B) of tracer 815Nvalues for the detritivores Gammarus minus and Anchytarsus bicolor (longitudinal profile only) and their potentialfood resources. Each data point represents only 1 subsample of composited material and has an uncertaintyof approximately ±0.5%o (see methods).

ble C and N analysis and 15N tracer experi-ments) has limitations, and the combination ofapproaches provided a more complete picturethan either alone Natural-abundance stable iso-tope analysis relies on substantial separation be-tween potential food resources (generally atleast 3-4%o) and this was true only for epilithonand leaves for 815N, and epilithon and all detri-tus compartments for 8I3C. Further, trophic en-richment in natural-abundance 815N values isvariable, depending on an organism's age, size,and nutritional status; values as low as +2 andas high as +4%o have been reported in differentstudies (Minagawa and Wada 1984, Ehleringeret al. 1986, Peterson and Fry 1987, Keough et al.1996). Last, an organism may assimilate a por-tion of the food resource that has a different sta-ble isotope ratio than the bulk material sampled(e.g., bacterial or algal cells that have different8I5N or 8I3C values than the material as awhole). The strengths of the natural-abundancestable isotope approach are that it integratesover time and that organisms are likely to be ator near isotopic equilibrium with respect totheir food resources, assuming there have been

no recent shifts in physical or chemical condi-tions that alter the isotopic content or relativeabundance of food resources.

The natural-abundance 15N and I3C analysestaken together appeared to show that £. clavae-Jbrmis and Stenonema spp. assimilated a mixtureof epilithon and detritus (more likely FBOMthan leaves). These analyses also showed thatBaetis spp. assimilated either a more highly 15N-enriched portion of the epilithon or a mixture ofepilithon and Porella. There was disagreement inthe natural-abundance analyses as to food re-sources for P. herricki, with the I5N analysis sug-gesting detritus and the 13C analysis a mixtureof epilithon and Porella. The natural-abundance15N analysis also suggested that G. minus and A.bicolor fed mostly on FBOM and/or epixylonrather than on leaves. Gammarus is known toshred leaves and was commonly found in leafaccumulations in Walker Branch; however, ouranalysis suggested that leaves were not its pri-mary food resource Our results for Anchytarsuswere in agreement with Merrirt and Cummins(1996), who indicated that rotting wood is itsprimary food resource

2000] STREAM MACROINVERTEBRATE FOOD RESOURCES USING STABLE ISOTOPES 153

(A) Gammarus:leaves

(8) GammaruFBOM

ii

2.0 -

1.5 -

1.0 -

0.5 -

n n -

(C) Anchytarsus:leaves

I

(D) Anchytarsus:epixylon

0 10 20 30 40 50 0 10 20 30 40 50

Day Day

FIG. 6. Organism:detritus S15N ratios for detriti-vores over time during the 15N tracer experiment. His-togram heights represent means of values from allsampling stations on a particular date for which de-tritus tracer 815N values were >5%o (n = 2-4 per date).Vertical error bars represent 1 SD. Only day 42 valueswere determined for Anchytarsus.

The strength of the experimental 15N tracerapproach for determining food web relation-ships is that there is likely to be greater sepa-ration in isotope ratios between potential foodresources. Food resources of primary consumerscan acquire different amounts of tracer 15N as aresult of differential use of the various forms ofdissolved N in water (NHV NOV dissolved or-ganic N), only 1 of which (NH4) was labeledwith tracer I5N. Different food resources alsomay have different levels of tracer 15N becausethe proportion of actively cycling N may differ.For example, the much lower tracer 815N levelsof detritus compared with epilithon and Porelladuring the latter portions of our tracer experi-ment likely reflected the large pool of N of ter-restrial origin within detritus. Another strengthof the tracer 15N approach is that it can some-times identify when a food resource is exploitedby a consumer in a manner different from theway in which it was sampled. For example, epi-lithon usually consists of many layers of livingand dead algal cells as well as excreted mucilageand adsorbed organic matter (Lock et al. 1984).

3

40 -

35 -

30 -

25 -

-•- SPOM-•- FBOM-G- Simuliidae-a— Dip/ectrona

0 20 40 60 80 100 120 140 160Distance from 15NH4 input (m)

FIG. 7. Longitudinal profiles (day 42) of tracer 615Nvalues for the collector/filterers Simuliidae and Di-plectrona modesta and their potential food resources(suspended particulate organic matter [SPOM], finebenthic organic matter [FBOM]). Time series sampleswere not collected for these taxa. Each data point rep-resents only 1 subsample of composited material andhas an uncertainty of approximately ±0.5%o (seemethods).

Tracer 15N is primarily incorporated into the liv-ing components of the epilithon (and recentlyexcreted organic compounds), which are likelyto be concentrated at or near the upper surface.Thus, tracer 15N is not likely to be distributedhomogeneously within the epilithon matrix, andscrapers that remove and assimilate primarilyliving cells in the upper layers would becomemore highly labeled than the bulk material sam-pled.

Herbivores

Our experimental I5N tracer analyses sug-gested that Stenonema spp. and Baetis spp. de-pend primarily on epilithon as their primary re-source, although as with the natural-abundanceanalysis, our tracer results do not rule out thepossibility that Baetis may also consume Porella.Results from a I3C tracer experiment in ColdSpring, North Carolina (Hall 1995), and I5Ntracer experiments in Hugh White Creek, NorthCarolina (Hall et al. 1998), Upper Ball Creek,North Carolina (J. L. Tank and co-workers, un-published data), and the Kuparuk River, Alaska(Peterson et al. 1997, Wollheim et al. 1999) have

154 P. J. MULHOLLAND ET AL. [Volume 19

also shown that these taxa feed primarily onepilithon. Stenonema in Cold Spring and UpperBall Creek and Baetis in the Kuparuk Riverreached tracer 615N levels that were substantial-ly greater than those for the bulk epilithon bydie end of the respective experiments, suggest-ing assimilation of a more metabolically active,fast N-turnover pool (e.g., living algal or bacte-rial cells) within the bulk epilithon. In the Ku-paruk River study, Wollheim et al. (1999)showed that when the epilithon were separatedinto fast N-turnover (diatoms) and slow N-turn-over (detritus) pools, the longitudinal 815N pro-file for Baetis agreed more closely with that ofthe fast N-turnover epilithon pool than the bulkepilithon. In Walker Branch, there was evidencefor assimilation of fast-turnover N in the epili-thon by Baetis but not by Stenonema (Fig. 2). Ourresults suggested that Baetis may feed primarilyon the surface of the epilithon whereas Steno-nema appears to consume the bulk epilithon ma-terial in Walker Branch.

Differences in the way Stenonema exploits theepilithon as a food resource in different streamsmay be related to the quality of the epilithon. Inheavily shaded streams, such as Cold Springand Upper Ball Creek, much of the epilithonmay be detrital material and Stenonema maypreferentially consume the more nutritious com-ponent of the epilithon (e.g., algal or bacterialcells). In Walker Branch where light levels dur-ing the period of this experiment (early spring)were relatively high (12-15 mol quanta m~2 d~',Mulholland et al. 2000), the bulk epilithon ap-peared to be of higher quality, as indicated byconsiderably higher chlorophylhAFDM ratios(3.2 x IQ-3, Mulholland et al. 2000), comparedwith Upper Ball Creek (0.47 x 1Q-3, J. L. Tank,unpublished data). Thus, the epilithon may havebeen used more completely as a food resourceby Stenonema in Walker Branch.

Detritivores

The tracer 15N results for G. minus and A. bi-color were largely consistent with the natural-abundance 15N results, suggesting that Gamma-rus feeds primarily on FBOM and Anchytarsuson epixylon, at least in early spring. Gammarushas been described as a collector-gatherer inWalker Branch (Newbold et al. 1983), but wefound it mostly in leaf packs and thought, apriori, that it might be primarily a shredder.

Gammarus sometimes may function as a shred-der in Walker Branch, but our results suggestedthat it relied on FBOM as its primary food re-source in spring.

For D. tnodesta, the natural-abundance and ex-perimental 15N tracer analyses together sug-gested that FBOM was its primary food, al-though neither analysis alone was very clear.Merritt and Cummins (1996) classified this tax-on as both a collector-gatherer and filterer. As-suming that FBOM was the primary food of Di-plectrona, the experimental 15N tracer analysissuggested that this taxon may digest and assim-ilate disproportionately the microbes associatedwith FBOM (or selectively consume portions ofthe FBOM that are higher in microbial cells) be-cause its tracer 815N values during the experi-ment were consistently higher than those forbulk FBOM.

The experimental 15N tracer analysis also sug-gested that the presumed scraper P. herricki(Merritt and Cummins 1996) is actually a detri-tivore, in agreement with the natural-abundance15N analysis (but not the 13C analysis). In astream in middle Tennessee, Smith (1978) foundthat detritus was a considerable proportion ofthe diet of Psephenus, based on analysis of stom-ach contents. Alternatively, Psephenus may feedon epilithon, and our 15N tracer results failed toshow this relationship because this organismmay have a low N turnover rate, despite itssmall size. Psephenus has a relatively long lifecycle in Walker Branch (~2 y, J. G. Smith, per-sonal communication), and the time course trac-er 815N values for Psephenus also suggested thatit may not have been in isotopic equilibriumwith its food resource by the end of the exper-iment.

Filterers

The experimental 15N tracer analysis provedconsiderably more informative than the natural-abundance 15N analysis for the food resourcesof Simuliidae. The natural-abundance 15N anal-ysis suggested that FBOM might be a food re-source for black flies, but not SPOM. The 15Ntracer experiment, however, clearly showed thatSPOM was the primary food resource for theseorganisms, and that they assimilated a portionof the SPOM pool that was more highly labeledwith tracer 15N (e.g., free-living or particle sur-face microbes, algal cells sloughed from the epi-

2000] STREAM MACROINVERTEBRATE FOOD RESOURCES USING STABLE ISOTOPES 155

lithon) than the bulk pool. Black fly larvae con-sume a disproportionate share of bacteria in theseston load of several streams (Edwards andMeyer 1987, Hershey et al. 1996). Mulholland etal. (2000) focused on N cycling and showed thatmicrobes associated with benthic detritus hadtracer 615N values 2-4 times higher than thebulk detrital material and it is likely that thiswas true of SPOM as well.

Use of modeling with the tracer 15N approach

A weakness of the tracer 15N approach is thatthe 8I5N values of consumers will lag behindthose of their food, with the lag period being afunction of the rate of N turnover within theconsumer. Thus, the experimental 15N tracer ap-proach is most useful for determining food re-sources of consumers that have high N turnoverrates (on the order of several days). We believethat this was the case at least for Stenonema andBaetis, based on time course plots of organism:food tracer 815N ratios (Fig. 3A,B), and for G.minus, based on its time course tracer 815N val-ues (Fig. 5A). The tracer 15N approach can alsobe very useful for organisms with slower Nturnover rates if N turnover rate is known orcan be estimated, and a model of expected 615Nover time developed for assimilation of each po-tential food resource (Wollheim et al. 1999). Wewere able to show that Elimia assimilated a mix-ture of epilithon and detritus, probably mostlyFBOM, using this approach (Fig. 4). This resultwas consistent with the natural-abundance UCanalysis, but conflicted with the natural-abun-dance 15N analysis, which suggested that Elimiaassimilated primarily epilithon. Previous studieshave shown that Elimia feeds on epilithon,FBOM, and leaves (Mulholland et al. 1985a, Ro-semond et al. 1993), and its primary diet maydepend on which of these resources is mostplentiful at a particular time. The importance ofepilithon as a food resource for Elimia may begreater in the spring when light levels are high-est and algal productivity is greatest (Marzolfet al. 1994), whereas the importance of leaves islikely greater in late autumn when they are mostabundant (Mulholland et al. 1985b).

Bryophytes as a food resource

Our 15N results did not rule out the possibilitythat P. pinnata may be a food resource for some

macroinvertebrates in Walker Branch becausethe natural-abundance and tracer 815N valuesfor Porella and epilithon were similar. The simi-larity in the tracer 815N values did not appearto be the result of epiphytes because the 8'5Nvalues for Porella were determined on entirefronds and the large amount of N in Porella tis-sues (3.7% of dry mass, Mulholland et al. 2000)would preclude high tracer 815N values for thebulk material if only epiphytes were labeled. Wehave observed no physical evidence of feedingon Porella and suspect that, despite its highabundance (Steinman and Boston 1993) and im-portance in uptake of NH4 from water (Mulhol-land et al. 2000), it is not a major food resourcefor macroinvertebrates in Walker Branch. How-ever, it is intriguing that the natural-abundance13C analysis suggested that Psephenus and Baetismay consume Porella. It is possible that Baetismay consume a component of the epilithon thatis more similar in 13C content to Porella than tothe bulk epilithon pool, as suggested by the 15Ntracer results. Nonetheless, the possible role ofPorella as a food resource in Walker Branch andelsewhere deserves further study.

Acknowledgements

Melanie Carter, Norm Leonard, and RamieWilkerson assisted with field sampling andsample processing. Kris Tholke performed themass spectrometry analyses. John Smith provid-ed unpublished data on macroinvertebrateabundance from previous years in WalkerBranch. We thank Walter Hill, John Smith, ChrisPeterson, and 2 anonymous reviewers for theirhelpful comments on earlier versions of themanuscript. The research was supported by agrant from the Ecosystems Program, NationalScience Foundation, to Virginia Tech (DEB-9628860). The research was conducted at theOak Ridge National Environmental ResearchPark, Environmental Sciences Division, Office ofBiological and Environmental Research, US De-partment of Energy under contract DE-AC05-96OR22464 with Lockheed Martin Energy Re-search Corp. Publication 4959, EnvironmentalSciences Division, Oak Ridge National Labora-tory.

Literature Cited

ANGRADI, T. R. 1994. Trophic linkages in the lowerColorado River: multiple stable isotope evidence.

156 P. J. MULHOLLAND ET AL. [Volume 19

r

Journal of the North American Benthological So-ciety 13:479^98.

EDWARDS, R. T., AND J. L. MEYER. 1987. Bacteria as afood source for black fly larvae in a blackwaterriver. Journal of the North American Benthologi-cal Society 6:241-250.

EHLERINGER, J. R., P. W. RUNDEL, AND K. A. NAGY.1986. Stable isotopes in physiological ecology andfood web research. Trends in Ecology and Evo-lution 1:42-45.

HALL, R. Q 1995. Use of a stable carbon isotope ad-dition to trace bacterial carbon through a streamfood web. Journal of the North American Ben-thological Society 14:269-277.

HALL, R. Q, B. J. PETERSON, AND J. L. MEYER. 1998.Testing a nitrogen-cycling model of a foreststream by using a nitrogen-15 tracer addition.Ecosystems 1:283-298.

HAMILTON, S. K., W. M. LEWIS, AND S. J. SIPPEL. 1992.Energy sources for aquatic animals in the OrinocoRiver floodplain: evidence from stable isotopes.Oecologia 89:324-330.

HERSHEY, A. E., R. W. MERRITT, M. C. MILLER, AND J.S. McCREA. 1996. Organic matter processing bylarval black flies in a temperate woodland stream.Oikos 75:524-532.

HUGHES, J. E., L. A. DEEGAN, B. J. PETERSON, R. M.HOLMES, AND B. FRY. 2000. Nitrogen flowthrough the food web in the oligohaline zone ofa New England estuary. Ecology 81:433-452.

KEOUGH, J. R., M. E. SIERSZEN, AND C. A. HAGLEY.1996. Analysis of a Lake Superior coastal foodweb with stable isotope techniques. Limnologyand Oceanography 41:136-146.

KLING, G. W. 1994. Ecosystem-scale experiments: theuse of stable isotopes in freshwaters. Pages 91-120 in L. A. Baker (editor). Environmental chem-istry of lakes and reservoirs. Advances in Chem-istry. American Chemical Society, Washington,DC.

LOCK, M. A., R. R. WALLACE, J. W. COSTERTON, R. M.VENTULLO, AND S. E. CHARLTON. 1984. River epi-lithon: toward a structural-functional model. Oi-kos 42:10-22.

MARZOLF, E. R., P. J. MULHOLLAND, AND A. D. STEIN-MAN. 1994. Improvements to the diumal up-stream-downstream dissolved oxygen changetechnique for determining whole-stream metab-olism in small streams. Canadian Journal of Fish-eries and Aquatic Sciences 51:1591-1599.

MERRITT, R. W, AND K. W. CUMMINS. 1996. An intro-duction to the aquatic insects of North America.3rd edition. Kendall/Hunt, Dubuque, Iowa.

MINAGAWA, M., AND E. WADA. 1984. Stepwise enrich-ment of 15N along food chains: further evidenceof the relation between d!5N and animal ageGeochimica et Cosmochimica Acta 48:1135-1140.

MULHOLLAND, P. J. 1992. Regulation of nutrient con-

centrations in a temperate forest stream: roles ofupland, riparian, and instream processes. Lim-nology and Oceanography 37:1512-1526.

MULHOLLAND, P. J., J. W. ELWOOD, J. D. NEWBOLD, ANDL. A. FERREN. 1985a. Effect of a leaf-shreddinginvertebrate on organic matter dynamics andphosphorus spiralling in heterotrophic laboratorystreams. Oecologia 66:199-206.

MULHOLLAND, P. J., J. W. ELWOOD, J. D. NEWBOLD, L.A. FERREN, AND J. R. WEBSTER. 1985b. Phosphorusspiralling in a woodland stream: seasonal varia-tions. Ecology 66:1012-1023.

MULHOLLAND, P. J., A. D. STEINMAN, A. V. PALUMBO,J. W. ELWOOD, AND D. B. KIRSCHTEL. 1991. Roleof nutrient cycling and herbivory in regulatingperiphyton communities in laboratory streams.Ecology 72:966-982.

MULHOLLAND, P. J., J. L. TANK, D. M. SANZONE, W. M.WOLLHEIM, B. J. PETERSON, J. R. WEBSTER, AND J.L. MEYER. 2000. Nitrogen cycling in a foreststream determined by a I5N tracer addition. Eco-logical Monographs (in press).

NEWBOLD, J. D., J. W. ELWOOD, R. V. O'NEILL, AND A.L. SHELDON. 1983. Phosphorus dynamics in awoodland stream ecosystem: a study of nutrientspiralling. Ecology 64:1249-1265.

PETERSON, B. J., M. BAHR, AND G. W. KLING. 1997. Atracer investigation of nitrogen cycling in a pris-tine tundra river. Canadian Journal of Fisheriesand Aquatic Sciences 54:2361-2367.

PETERSON, B. J., AND B. FRY. 1987. Stable isotopes inecosystem studies. Annual Review of Ecologyand Systematics 18:293-320.

PETERSON, B. J., B. FRY, L. DEEGAN, AND A. HERSHEY.1993. The trophic significance of epilithic algalproduction in a fertilized tundra river ecosystem.Limnology and Oceanography 38:872-878.

ROSEMOND, A. D., P. J. MULHOLLAND, AND J. W. EL-WOOD. 1993. Top-down and bottom-up control ofstream periphyton: effects of nutrients and her-bivores. Ecology 74:1264-1280.

ROSENFELD, J. S., AND J. C. ROFF. 1992. Examination ofthe carbon base in southern Ontario streams us-ing stable isotopes. Journal of the North AmericanBenthological Society 11:1-10.

ROUNICK, J. S., M. J. WlNTERBOURN, AND G. L. LYON.1982. Differential utilization of allochthonous andautochthonous inputs by aquatic invertebrates insome New Zealand streams: a stable carbon iso-tope study. Oikos 39:191-198.

SMITH, J. G. 1978. The life histories, production ratesand feeding habits of selected benthic macroin-vertebrates of Spring Creek, Overton County,Tennessee. MS Thesis, Tennessee TechnologicalUniversity, Cookeville, Tennessee.

STEINMAN, A. D., AND H. L. BOSTON. 1993. The eco-logical role of aquatic bryophytes in a woodland

2000] STREAM MACROINVERTEBRATE FOOD RESOURCES USING STABLE ISOTOPES 157

to27 *? analysis °fogicai Society 12.17-26. nal of ̂ Nor^ American Bentholoeical Society

WOLLHEIM, W. M., B. J. PETERSON, L. A. DEECAN, M. 18:199-219 ™'°gicai societyBAHR, J. E. HOBBIE, D. JONES, W. B. BOWDEN AE^HERSHEY, G K KLING, AND M. C. M.LLER. ^ceived: 5 October im1999. A coupled held and modehng approach for ^cepted. 1 February 2m

L.