BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, researchlibraries, and research funders in the common goal of maximizing access to critical research.

Evaluation of two timesaving techniques for processing benthic invertebratesamples for estimating secondary productionAuthor(s): Jaynie M. Stephenson, Geneviève Carr, Uta Gruenert, and Antoine MorinSource: Journal of the North American Benthological Society, 26(4):611-619. 2007.Published By: The Society for Freshwater ScienceDOI: http://dx.doi.org/10.1899/06-096.1URL: http://www.bioone.org/doi/full/10.1899/06-096.1

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, andenvironmental sciences. BioOne provides a sustainable online platform for over 170 journals and books publishedby nonprofit societies, associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance ofBioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiriesor rights and permissions requests should be directed to the individual publisher as copyright holder.

J. N. Am. Benthol. Soc., 2007, 26(4):611–619� 2007 by The North American Benthological SocietyDOI: 10.1899/06-096.1Published online: 11 September 2007

Evaluation of two timesaving techniques for processing benthicinvertebrate samples for estimating secondary production

Jaynie M. Stephenson1

Ottawa–Carleton Institute of Biology, Department of Biology, University of Ottawa,30 Marie Curie, Ottawa K1N 6N5 Canada

Genevieve Carr2

United Nations Environment Programme, Global Environment Monitoring System/Water Programme,351 St Joseph Boulevard, 8th Floor, Gatineau, Quebec K1A 0H3 Canada

Uta Gruenert3

Leibniz Institute of Freshwater Ecology and Inland Fisheries, Muggelseedamm 310, 12587 Berlin, Germany

Antoine Morin4

Ottawa–Carleton Institute of Biology, Department of Biology, University of Ottawa,30 Marie Curie, Ottawa K1N 6N5 Canada

Abstract. We compared the accuracy, precision, and efficiency of 2 timesaving techniques to a standardlaboratory procedure for processing benthic samples used to estimate secondary production. In the coarse-sieve technique, production was quantified from invertebrate biomass collected in a 2-mm sieve andcorrected for retention probabilities of individual organisms. In the sieve-fractionated technique, productionper sieve was quantified from the total biomass and average body mass of organisms collected in ageometric series of 9 sieves. Production estimates for the entire assemblage (coarse-sieve: r2 ¼ 0.96, sieve-fractionated: r2¼ 0.99, n¼ 57) and for individual taxa (coarse-sieve: r2¼ 0.84, n¼ 308; sieve-fractionated: r2¼0.99, n ¼ 544) were strongly related to production estimates derived from standard processing techniques.The analytical error introduced by coarse-sieve (entire assemblage: residual mean square [RMS] ¼ 0.014,individual taxa: RMS¼ 0.001) and sieve-fractionated (entire assemblage: RMS¼ 0.070, individual taxa: RMS¼ 0.0004) techniques was insignificant relative to the variance of production estimates among replicates(entire assemblage¼0.150, individual taxa¼0.329). Coarse-sieve and sieve-fractionated techniques required,on average, only 18% and 63% of the time required for standard processing, respectively, but coefficients ofvariation for production means of the entire assemblage differed by ,1% between standard and timesavingtechniques. These timesaving techniques for processing benthic samples increase the feasibility of studies ofsecondary production that require multiple sampling sites and dates, and thus, should further ourunderstanding of the mechanisms that control stream productivity.

Key words: secondary production estimates, coarse-sieve, sieve-fractionated, retention, method, effi-ciency, accuracy, precision.

Benthic invertebrates transfer energy and matter

from instream primary producers or allochthonous

organic matter to top predators (Wallace and Webster

1996). Therefore, quantifying and understanding ener-gy flow in lotic systems requires accurate measure-ments of biomass and productivity of benthicinvertebrates (Benke 1993). Moreover, secondary pro-duction estimates are an essential component ofeffective fisheries management (Wallace and Webster1996) and can provide a more accurate measure ofbenthic response to disturbance than indices of

1 E-mail addresses: jstep081@uottawa.ca2 genevieve.carr@gemswater.org3 uta.gruenert@kwilke.de4 amorin@uottawa.ca

611

community structure (Lugthart and Wallace 1992),which sometimes are used as surrogates for benthicproduction when assessing stream health (Rosenbergand Resh 1993, Wallace et al. 1996). However,relatively few secondary production studies haveestimated production of entire benthic communitiesat multiple streams, sites, or years along environmen-tal gradients (e.g., Krueger and Waters 1983, Lugthartand Wallace 1992, Wallace et al. 1999, Morin et al. 2001,Shieh et al. 2002, 2003, Carlisle and Clements 2003).

The scarcity of community production studies atmultiple streams, sites, or years is probably theconsequence of the high costs of measuring inverte-brate production. For instance, quantifying annualproduction of a single species at just 1 site typicallyrequires at least 36 samples (3 replicates 3 12 dates, ifsamples are collected monthly). Describing the re-sponse of annual production to only 1 environmentalvariable requires at least 3 to 5 streams (i.e., 108–150samples) to assess even minimally a gradient in thevariable of interest. Assuming 8 h to process eachsample using standard processing techniques (JMS,unpublished data), an estimated 864 to 1200 laboratoryprocessing hours would be required to assess howproduction of a single species responds to 1 environ-mental variable. Production estimates of multiplespecies require even more laboratory processing hours.

Various methods for calculating secondary produc-tion have been proposed to reduce the costs ofobtaining production estimates. The size–frequencymethod (Hynes and Coleman 1968, Hynes 1980) wasdeveloped to estimate production of an entire asyn-chronous invertebrate community to eliminate the needfor time-consuming taxonomic identifications, althoughthe method is applied more commonly to individualspecies than to communities (Benke 1993). Further-more, annual production estimates for semivoltine ormultivoltine species must be corrected for cohortproduction intervals, a procedure that requires knowl-edge of life-cycle lengths (Benke 1979). Recent empiricalmodels have been proposed to circumvent making thedirect measurements of growth rates required forinvertebrate production estimates (Banse and Mosher1980, Huryn and Wallace 1986, Hauer and Benke 1987,1991, Morin et al. 1988, Stites and Benke 1989, Benkeand Parsons 1990, Huryn 1990, Morin and Bourassa1992, Morin and Dumont 1994). These developments inthe computational components of production estimateshave reduced the need for field data on growth rates,but the high cost of processing samples to estimatenumbers and sizes of organisms remains a problem.

Techniques to reduce laboratory processing time ofbenthic samples are desirable if they do not affect the

quality of the data, or if savings in time and moneyoutweigh reductions in precision and accuracy. Morinet al. (2004) developed a coarse-sieve processingtechnique and showed that biomass estimates forentire assemblages were not significantly differentfrom those based on invertebrates retained in 1-mmsieves and corrected for invertebrates lost through 1-mm sieves. Ramsay et al. (1997) showed that a sieve-fractionated technique allowed accurate measurementsof biomass and size distributions from the average sizeof a small number of invertebrates and the totalnumber of invertebrates retained in a geometric seriesof sieves. Both of these techniques could be used tostreamline processing of samples for estimating pro-duction. However, whether the savings of suchstreamlined techniques are sufficient to offset the lossof precision and accuracy of the resulting productionestimates is not obvious. Moreover, a risk of failing todetect some taxa exists when coarse-sieve techniquesare used, and that risk has not yet been quantified.

We assessed the accuracy (i.e., the magnitude ofsystematic bias), the precision (i.e., the added variabil-ity), and the efficiency (i.e., the reduction in processingtime) of 2 techniques used to reduce laboratoryprocessing time of benthic invertebrate samples whenestimating secondary production of entire communitiesand individual taxa. We compared the performance ofthe 2 timesaving techniques to the standard laboratoryprocedure and quantified the risk of failing to detect taxawhen using only coarse (�2-mm mesh aperture) sieves.

Methods

Field and laboratory procedures

Six replicate cobbles (8–25 cm in diameter) werecollected from the riffle areas of 10 headwater streamsin southeastern Ontario, Canada, during July andAugust 2001 (n ¼ 57 [6 cobbles 3 10 streams � 3damaged samples]). Each cobble sample was placed ina plastic bag with 95% ethanol and stored in a darkrefrigerator (68C) upon returning to the laboratory.Three cobbles were not used because sample bagsleaked during transport from the field to the labora-tory. Invertebrates from the samples were collected ona geometric series of 9 sieves with mesh aperturesranging from 0.063 to 16 mm. The stacked sieves wereplaced in a sink with the largest sieve (16 mm) at thetop, smaller sieves in order of decreasing size beneaththe top sieve, and the finest sieve (0.063 mm) on thebottom. The contents of each sample bag were pouredcarefully into the 16-mm sieve, and the cobble wasgently brushed over the series of stacked sieves toremove attached material. The sample material on thetop sieve was gently washed with tap water until all

612 [Volume 26J. M. STEPHENSON ET AL.

material finer than 16 mm passed through the topsieve. The material remaining in the top sieve wasplaced in a sample jar. The top sieve was thenremoved, and the finer fractions remaining in thestack of sieves were processed similarly. Invertebratescollected in each sieve were sorted, assigned to 1 of 22taxonomic groups (Turbellaria, Gastropoda, Ephemer-optera [Baetidae, Heptageniidae, Leptophlebiidae,Tricorythidae, Caenidae], Coleoptera [Elmidae,Psephenidae], Diptera [Tipulidae, Simuliidae, Chiro-nomidae], Plecoptera [Perlidae], Trichoptera [Philopo-tomatidae, Hydropsychidae, Polycentropodidae,Limnephilidae, Leptoceridae, Helicopsychidae], Lepi-doptera [Pyralidae], Isopoda, and Amphipoda), andmeasured along their central axis to the nearest 0.01mm using a digital analysis system. The cleanedcobble was retained and its surface area was deter-mined from the mass of aluminum foil required tocover it completely (Reice 1980).

Production estimates

Standard processing technique.—To estimate the accu-racy and precision of coarse-sieve and sieve-fractionatedapproaches for processing benthic samples, productionestimates obtained from samples processed using ourtimesaving techniques were compared to productionestimates obtained using a standard laboratory process-ing technique in which all invertebrates collected in a 63-lm sieve were sorted, identified, and measured in thelaboratory. Invertebrate dry mass (M, mg) was calcu-lated from taxon-specific (i.e., family or order) length–mass regressions (Benke et al. 1999). Mass-specificgrowth rates (g, /d) were estimated for the averagewater temperature (T) across all sites (208C) at the timeof collection using an empirical model developed forstream invertebrates (Morin and Dumont 1994):

log10ðgÞ ¼ �2:09� 0:27ðlog10½M�Þ þ 0:025T: ½1�

Order-level taxon-specific models were used for Dip-tera, Ephemeroptera, Plecoptera, and Trichoptera.Equation 1 was used to estimate growth rates whencalculating taxon-specific production for all other orders(Morin and Dumont 1994). Using empirical models toestimate growth rates is not a standard method forestimating benthic invertebrate production. However,we defined standard production for comparative purpos-es as production that was quantified from samples thatwere processed using standard laboratory techniques.Standard production (P, mg m�2 d�1) for each samplewas calculated as

P ¼Xn

i¼1

Migi

A; ½2�

where Mi is the dry mass (mg) of each invertebrate, gi isthe mass-specific growth rate of each invertebratecalculated from equation 1, A is the surface area of thecobble (m2), and n is the total number of invertebratescollected from the sample.

Coarse-sieve and sieve-fractionated processing tech-niques.—The invertebrates retained in all sieves withmesh apertures �2 mm (2-mm sieve) were used forestimates based on the coarse-sieve technique. It wasassumed that invertebrates caught in sieves withapertures .2 mm would have been retained in a 2-mm sieve. A sieve-retention probability model (Morinet al. 2004) was used to account for organisms lostthrough the 2-mm sieve. Production per cobble (P1)was calculated for the entire assemblage and forindividual taxa from the 22 taxonomic groups listedabove as

P1 ¼Xn

i¼1

Migi

Api; ½3�

where pi is the probability that the ith organism isretained in a 2-mm sieve and n is the number ofinvertebrates retained in a 2-mm sieve. We determinedp for each invertebrate retained in a 2-mm sieve by firstcalculating logit p (Morin et al. 2004) as

logit p ¼ �2:84þ 5:8ðlog10½RL�Þ� 3:18ðlog10½RL�Þ3 log10ðmesh sizeÞ; ½4�

where relative length (RL) is the ratio of invertebratebody length (mm) to the 2-mm sieve mesh aperture,and then calculating p as e(logit p)/(1 þ e[logit p]). Forsimplicity, the general sieve-retention model (equation4) was used rather than taxon-specific sieve-retentionmodels when quantifying taxon-specific productionbecause Morin et al. (2004) found that taxon-specificmodels increased the explained variation in sieveretention by ,2%.

The invertebrates in each sieve were used forestimates based on the sieve-fractionated technique.Production per sample was estimated from biomass (Bj,mg/m2) and average body mass (Mj, mg) for each ofthe j sieve fractions. For comparative purposes, thesame samples were used to calculate production usingstandard and timesaving processing techniques. There-fore, the dry mass of each sieve fraction could not bedetermined directly. Instead, the number of inverte-brates in each sieve fraction (nj) was counted, and totaldry mass/sieve (Bj) was estimated to obtain averagebody mass/sieve fraction (Mj). Using Mj negates theneed to identify and measure individual invertebratesunless the method is used to obtain taxon-specificproduction, in which case individual identificationswould be required, but individual measurements of

2007] 613TIMESAVING TECHNIQUES FOR PROCESSING BENTHIC SAMPLES

body length would not. It was assumed that the sum ofindividual dry masses obtained from length–weightregressions was equivalent to the mass that would havebeen obtained by weighing the organisms retained in asieve after drying. The average growth rate/sieve (gj)was determined from Mj using equation 1 (Morin andDumont 1994). Taxon-specific growth-rate modelsfrom Morin and Dumont (1994) were used to calculategj when quantifying sieve-fractionated production ofindividual taxa. Production was quantified for eachsieve, and the total production of each sample (P2) wasdetermined as the sum of production from the 9 (¼m)sieves as

P2 ¼Xm

j¼1

Bjgj: ½5�

Sample processing time

The laboratory processing time of each techniquewas estimated from a random subset of replicatesamples from each site (26 of the 57 processed cobbles).Total processing time included sorting and measuringthe lengths of all individuals retained in a cobblesample (standard processing technique) or in a 2-mm-sieve subsample (coarse-sieve technique). The timerequired to sieve each sample was assumed to benegligible. For the sieve-fractionated technique, onlythe time spent sorting invertebrates from each sievefraction per cobble sample was included in totalprocessing time. The time required to sieve, dry, andweigh the organisms collected in each sieve fractionwas assumed to be negligible. The time required toprocess a given number of invertebrates using stan-dard, coarse-sieve, and sieve-fractionated techniqueswas predicted using regression analysis. Analysis ofcovariance was used to compare the slopes and y-intercepts among the 3 regressions.

Accuracy and precision of timesaving approaches

The accuracy of the 2 timesaving techniques wasdetermined by comparing P (equation 2) to P based oncoarse-sieve fractions and sieve-fractionated subsam-ples (equations 3, 5) for the 57 cobbles. The relativeanalytical precisions of each technique were comparedby examining the residual variance between P andcoarse-sieve P1 or sieve-fractionated P2, i.e., theresidual mean squares (RMS) in regressions predictingP from P based on coarse-sieve and sieve-fractionatedtechniques. The analytical error introduced by thesetechniques for processing benthic samples was quan-tified relative to the natural spatial variance of P by

comparing the RMS from approaches with thevariance of P among replicates.

Probability of detecting taxa in sieve fractions

The probability that at least 1 individual of a taxonwas retained in a sieve fraction when the entire samplecontained n individuals of the taxon was calculated as

Retention probability ¼ 1�Yn

i¼1

ð1� piÞ; ½6�

where pi is the probability that each individual i isretained by the sieve as predicted by the sieve-retention probability model of Morin et al. (2004;equation 4). The validity of this estimate of taxaretention probability was assessed by comparing thecalculated values from equation 6 to the observeddetections in the 2-mm-sieve fraction for each taxonfound on individual cobbles.

Results

We collected 15,406 invertebrates from 57 cobblesamples. The number of invertebrates collected persample ranged from 49 to 1485, and per-sampleestimates of P ranged from 4.8 to 691.5 mg m�2 d�1.

Accuracy and precision for total production

Values of P for entire assemblages from coarse-sievefractions and sieve-fractionated subsamples werestrongly related to P (coarse-sieve: r2 ¼ 0.96, sieve-fractionated: r2 ¼ 0.99; Fig. 1A, B). The regressionpredicting P from P1 did not differ significantly fromthe 1:1 line between P and P1 (coarse-sieve: F1,56¼ 1.83,p ¼ 0.18; Fig. 1A). However, P2 consistently overesti-mated P by ;5% (sieve-fractionated: F1,56¼456.37, p ,

0.001; Fig. 1B).P1 was significantly less precise (coarse-sieve: RMS¼

0.014; Fig. 1A) than P2 (sieve-fractionated: RMS ¼0.001; Fig. 1B) (F55,55 ¼ 14, p , 0.001). Nonetheless,sampling error caused by the natural spatial variationin P was significantly larger (variance among repli-cates ¼ 0.150) than the analytical error (i.e., RMS)introduced by either timesaving technique (coarse-sieve: F47,55¼ 10.71, p , 0.001; sieve-fractionated: F47,55

¼ 150, p , 0.001).

Accuracy and precision for production of individual taxa

Values of P for individual taxa obtained with the 2timesaving techniques were strongly correlated with P(coarse-sieve: r2¼ 0.86, n¼ 308; sieve-fractionated: r2¼0.99, n ¼ 544; Fig. 2A, B), although small-bodied taxa,especially rare ones, often were not detected by the

614 [Volume 26J. M. STEPHENSON ET AL.

coarse-sieve technique. No statistically significant bias

was introduced by either timesaving technique, and

the fitted regressions did not significantly differ from

the 1:1 line between P and P estimated using the 2

timesaving techniques (coarse-sieve: F1,306¼ 63.71, p ,

0.001; sieve-fractionated: F1,542¼ 164.68, p , 0.001).

The coarse-sieve technique (RMS ¼ 0.065) was

significantly less precise (F306,542 ¼ 162.5, p , 0.001;

Fig. 2A) at estimating P than the sieve-fractionated

technique (RMS ¼ 0.0004; Fig. 2B). However, spatial

variance in individual taxon production (variance

among replicates ¼ 0.371) was .23 that for total

production and significantly larger than the analytical

error introduced by either timesaving technique

(coarse-sieve: F306,393 ¼ 5.71, p , 0.001; sieve-fraction-

ated: F393,542 ¼ 927.5, p , 0.001).

Efficiency of coarse-sieve and sieve-fractionated techniquesand precision of production means

Coarse-sieve and sieve-fractionated approaches re-quired significantly less processing times for a givennumber of invertebrates than the standard processingtechnique (Fig. 3), and y-intercepts varied significantlyamong the 3 regressions (F2,74 ¼ 107, p , 0.001).However, slopes of the regressions were not signifi-cantly different (F2,72¼0.2, p¼0.84). On average, the 2-mm sieve retained 18% of the invertebrates that werepresent on a cobble. Therefore, the coarse-sieveapproach required sorting and measuring 18% of thetotal number of invertebrates per sample, and theapproach reduced laboratory processing time by 82%.The sieve-fractionated approach completely eliminatedthe need to measure individuals and required onlysorting of invertebrates, and the approach reducedlaboratory processing time by 36%. Both approaches

FIG. 1. Comparison of standard production (P) estimatesfor entire communities with estimates (P) obtained fromcoarse-sieve fractions (P1) (A) and sieve-fractionated sub-samples (P2) (B) collected at 10 sites (n ¼ 57 cobbles). Thedashed line represents the 1:1 line between P and P, and thesolid line is the regression through the data.

FIG. 2. Comparison of standard production (P) estimatesfor individual taxa with estimates (P) obtained from coarse-sieve fractions (P1; n ¼ 308) (A) and sieve-fractionatedsubsamples (P2; n¼ 544) (B). The dashed line represents the1:1 line between P and P, and the solid line is the regressionthrough the data.

2007] 615TIMESAVING TECHNIQUES FOR PROCESSING BENTHIC SAMPLES

substantially reduced the number of hours required toprocess 6 replicate cobbles (standard: 33 h 6 6 [SE],coarse-sieve: 6 h 6 1, sieve-fractionated: 21 h 6 4).However, the average coefficients of variation forproduction means varied among the 3 approaches by�1% (standard: 30% 6 4, coarse-sieve: 31% 6 4,sieve-fractionated: 30% 6 4).

Detection of taxa in the coarse-sieve fraction

The 57 cobbles yielded 544 opportunities to deter-mine whether a taxon present on a cobble was presentin the coarse (�2-mm)-sieve fraction. When comparingthe probabilities of detecting taxa in a 2-mm sieve withpredicted detection probabilities using the model ofMorin et al. (2004), the lowess curve fitted to these datawas quite similar to the 1:1 line between actual andpredicted detection probabilities (Fig. 4). On average,each cobble harbored 9.54 taxa, whereas the coarse-sieve fraction contained only an average of 5.4 taxa percobble. Taxa that were missing from the coarse-sievefraction typically were small (average body length ¼1.8 mm, range: 0.4–8.2 mm) and not numerous(average total number on the cobble ¼ 4.8, range: 1–62). The proportion of taxa detected in the coarse-sievefraction increased with increasing length and numberof individuals and was predicted accurately by thesieve-retention probability model of Morin et al. (2004;Fig. 5).

Discussion

Total assemblage production

The analytical error of the coarse-sieve approachwas significantly greater than that of the sieve-

fractionated approach, indicating that the error asso-ciated with estimating biomass lost through coarsesieves was greater than the error introduced byestimating average growth rates for each sievefraction. Indeed, sieve-retention probabilities for cor-recting biomass estimates obtained from the coarse-sieve approach overestimated the amount of biomassthat passed through 2-mm sieves by as much as 612%and underestimated biomass that passed through 2-mm sieves by as much as 65%. However, averagegrowth rates predicted from average size classes persieve using sieve-fractionated techniques overestimat-ed average production:biomass ratios per sieve by, atmost, 63%.

The systematic overestimation of production usingsieve-fractionated approach for the entire assemblageresults from the shape of the size distribution oforganisms in sieve fractions. The distributions ofindividual sizes in each sieve fraction were right-skewed and approached a lognormal distribution. Thisright-skewed distribution was typical of our samplesand is what was observed by Morin et al. (2004; fig. 6in Morin et al. 2004). Most of the biomass found in asieve was constituted of relatively few large individ-

FIG. 3. Hours required to process a given number ofinvertebrates (N) to obtain production estimates (n ¼ 26cobbles from a subset of sampling sites) using standard,coarse-sieve, and sieve-fractionated laboratory processingtechniques.

FIG. 4. Proportion of observations for which a taxonpresent on a cobble was detected in the 2-mm-sieve fractionas a function of the detection probability in the 2-mm sievepredicted using the model of Morin et al. (2004). Circlescorrespond to the presence (1) or absence (0) of a taxon in the2-mm sieve when it was present in the cobble sample (n ¼544). The solid line is the lowess fit (tension ¼ 0.3) throughthe observed data, and the dashed line is the 1:1 line betweenpredicted and actual detection probabilities.

616 [Volume 26J. M. STEPHENSON ET AL.

uals that had, because of their size, a much lower

mass-specific growth rate than most of the smaller

individuals in the sieve. However, the growth rate

used to calculate production for the sieve fraction is

based on average individual mass and does not take

into account the mixture of sizes of organisms retained

in a sieve. The consequence is that biomass-weighted

growth rate is overestimated, and thus, the production

estimate for the sieve fraction is overestimated.

Correction of the bias would require measuring the

individuals retained in a sieve (which would defeat the

purpose of the sieve-fractionated approach), using

more sieves, or making assumptions about the

parameters of the size distributions of the individuals

in a sieve. Given the small bias observed here (;5%),

we doubt that correcting the bias would be worth the

effort. However, further reducing the number of sieves

used to save more time should be avoided because

doing so would increase the inaccuracy of the sieve-

fractionated technique. Nonetheless, the analytical

error introduced by both timesaving techniques was

insignificant relative to the natural spatial variance of

standard production estimates within each site. The

coarse-sieve technique and, to a lesser extent, the sieve-

fractionated technique resulted in less variation in

production estimate means for a given amount of

laboratory processing time than production meansobtained from standard processing techniques.

Production of individual taxa

Sieve-fractionated production estimates of individ-ual taxa are less biased and more precise thanproduction estimates for the entire assemblage. Asthe number of individuals in each sieve fractiondecreases, the sieve-fractionated estimate convergestoward the standard production estimate because, atthe limit where each sieve contains only 1 individual,the calculations become the same. The probability ofobserving a heavily right-skewed size distribution in asieve fraction decreases and the bias described above islower when a single taxon is considered rather thanthe entire assemblage because there are fewer individ-uals in the single taxon. Further reductions inprocessing times might be achieved without introduc-ing objectionable bias by reducing the number ofsieves used when few individuals are present. How-ever, we suspect that such time savings would be smallbecause samples containing few individuals often canbe processed very quickly.

For moderately abundant taxa, the coarse-sieveapproach can be used to estimate production almostas precisely as the standard processing technique inmuch less time. For less common or smaller taxa, werecommend the use of a mesh aperture ,2 mm todecrease the probability that a taxon is not detected.Obviously, if no individuals of a rare or small taxon areretained in a coarse sieve, then the production estimateby the coarse-sieve technique will be 0. The simplesolution is to use a finer-mesh sieve for small and raretaxa if they must be detected. The appropriate meshsize can be determined from the size of the organismsand the number of individuals expected in replicatesamples (Fig. 5) because the sieve-retention probabilitymodel of Morin et al. (2004) can be used to estimate theprobability that a taxon present in a sample will bedetected in the �2-mm sieve fraction (Fig. 4).

Applications and limitations

The coarse-sieve approach provided accurate andprecise estimates of standard production with lesseffort than sieve-fractionated techniques. Therefore,we recommend the coarse-sieve approach for process-ing benthic samples to estimate secondary productionof moderately abundant taxa, functional feedinggroups, or the entire assemblage at multiple sites anddates. The coarse-sieve technique should be especiallyuseful in large-scale studies that assess how secondaryproduction responds to a gradient of disturbance orother environmental factors (e.g., Morin et al. 2001).

FIG. 5. Probability that a taxon represented by nindividuals in a sample will be detected in the �2-mm-sievefraction. Probabilities (p) for n ¼ 1 are calculated from thesieve-retention probability model of Morin et al. (2004). Forlarger values of n, they were calculated as 1 � (1 � p)n.

2007] 617TIMESAVING TECHNIQUES FOR PROCESSING BENTHIC SAMPLES

Coarse-sieve techniques for processing benthic sam-ples will make quantitative measures of food avail-ability for fish consumption more feasible. Suchmeasures might be used rather than the indices ofbenthic invertebrate community structure that aretypically used in lotic bioassessments (Rosenberg andResh 1993). Discrepancies that can arise betweenbioassessments based on secondary production whensamples are collected in sieves with different meshsizes have limited the utility of secondary productionestimates in bioassessments (Bonada et al. 2006).However, coarse-sieve techniques provide secondaryproduction estimates for the entire sample, therebyavoiding the problem of mesh sizes. Precision ofcoarse-sieve production estimates for small taxa willbe low unless fine-mesh sieves are used, but the sieve-fractionated approach would be a cost-effective alter-native and would save 1 /

3 of the processing timerequired by traditional methods.

Our assessment of the 2 timesaving techniques wasbased on instantaneous production estimates from asingle sampling date. Seasonal or annual productionestimates integrate several instantaneous productionestimates over time, and our general conclusionsregarding the relative merits of the different techniquesshould hold when using the instantaneous growthmethod to estimate annual production (Clarke et al.1946, Ricker 1946, Allen 1949). Moreover, the coarse-sieve technique also could reduce the significantlaboratory processing time required to estimate annualproduction using increment-summation (Waters 1977),removal-summation (Waters 1977), Allen (1951) curve,and size–frequency (Hynes and Coleman 1968, Hynes1980) methods by correcting for the density, dry mass,and biomass lost through a 2-mm sieve with retentionprobabilities (Morin et al. 2004).

Investigators using the coarse-sieve and sieve-fractionated techniques could quantify the influenceof 1 environmental variable on invertebrate produc-tion accurately and precisely in 151 to 210 and 540 to750 processing hours, respectively. These times aremuch less than the 864 to 1200 h required withstandard laboratory procedures. Time saved by pro-cessing benthic samples with coarse-sieve and sieve-fractionated techniques could be used to increase thenumber of sampling sites, dates, and replicates inproduction studies to gain better understanding of themechanisms that govern secondary production instreams. The substantial savings associated with theseapproaches could increase the feasibility of studies thatassess how eutrophication, deforestation, invasivespecies, channel alterations, and other anthropogenicimpacts to freshwater habitats affect secondary pro-duction.

Acknowledgements

This study was supported by a Natural Sciences andEngineering Research Council Grant of Canada(RGPIN-42799-2001) to AM. We thank Y. Yao for herhelp with sample processing and data entry. We alsothank J. Holomuzki, M. R. Whiles, and 2 anonymousreferees for their time and for valuable suggestionsthat greatly improved this paper.

Literature Cited

ALLEN, K. R. 1949. Some aspects of the production andcropping of fresh waters. Report, 6th Science Congress,1947. Transactions of the Royal Society of New Zealand77:222–228.

ALLEN, K. R. 1951. The Horokiwi Stream. A study of a troutpopulation. New Zealand Marine Department FisheriesBulletin 10:1–238.

BANSE, K., AND S. MOSHER. 1980. Adult body mass and annualproduction/biomass relationships of field populations.Ecological Monographs 50:355–379.

BENKE, A. C. 1979. A modification of the Hynes method forestimating secondary production with particular signif-icance for multivoltine populations. Limnology andOceanography 24:168–171.

BENKE, A. C. 1993. Edgardo Baldi Memorial Lecture: conceptsand patterns of invertebrate production in runningwaters. Verhandlungen der Internationalen Vereinigungfur theoretische und angewandte Limnologie 25:15–38.

BENKE, A. C., A. D. HURYN, L. A. SMOCK, AND J. B. WALLACE.1999. Length–mass relationships for freshwater macro-invertebrates in North America with particular referenceto the southeastern United States. Journal of the NorthAmerican Benthological Society 18:308–343.

BENKE, A. C., AND K. A. PARSONS. 1990. Modeling black flyproduction dynamics in blackwater streams. FreshwaterBiology 24:167–180.

BONADA, N., N. PRAT, V. H. RESH, AND B. STATZNER. 2006.Developments in aquatic insect biomonitoring: a com-parative analysis of recent approaches. Annual Reviewof Entomology 51:495–523.

CARLISLE, D. M., AND W. H. CLEMENTS. 2003. Growth andsecondary production of aquatic insects along a gradientof Zn contamination in Rocky Mountain streams. Journalof the North American Benthological Society 22:582–597.

CLARKE, G. L., W. T. EDMONDSON, AND W. E. RICKER. 1946.Mathematical formulation of biological productivity.Ecological Monographs 16:336–337.

HAUER, F. R., AND A. C. BENKE. 1987. Influence of temperatureand hydrograph on black fly growth rates in asubtropical river. Journal of the North American Bentho-logical Society 6:251–261.

HAUER, F. R., AND A. C. BENKE. 1991. Rapid growth of snag-dwelling chironomids in a blackwater river: the influenceof temperature and discharge. Journal of the NorthAmerican Benthological Society 10:154–164.

HURYN, A. D. 1990. Growth and voltinism of lotic midge

618 [Volume 26J. M. STEPHENSON ET AL.

larvae: patterns across an Appalachian Mountain basin.Limnology and Oceanography 25:339–351.

HURYN, A. D., AND J. B. WALLACE. 1986. A method forobtaining in situ growth rates of larval Chironomidae(Diptera) and its application to studies of secondaryproduction. Limnology and Oceanography 31:216–222.

HYNES, H. B. N. 1980. A name change in the secondaryproduction business. Limnology and Oceanography 25:778.

HYNES, H. B. N., AND M. J. COLEMAN. 1968. A simple methodof assessing the annual production of stream benthos.Limnology and Oceanography 13:569–573.

KRUEGER, C. C., AND T. F. WATERS. 1983. Annual production ofmacroinvertebrates in three streams of different waterquality. Ecology 64:840–850.

LUGTHART, G. J., AND J. B. WALLACE. 1992. Effects ofdisturbance on benthic functional structure and produc-tion in mountain streams. Journal of the North AmericanBenthological Society 11:138–164.

MORIN, A., AND N. BOURASSA. 1992. Modeles empiriques de laproduction annuelle et du rapport P/B d’invertebresbenthiques d’eau courante. Canadian Journal of Fisheriesand Aquatic Sciences 49:532–593.

MORIN, A., N. BOURASSA, AND A. CATTANEO. 2001. Use of sizespectra and empirical models to evaluate trophicrelationships in streams. Limnology and Oceanography46:935–940.

MORIN, A., M. CONSTANTIN, AND R. H. PETERS. 1988. Allometricmodels of simuliid growth rates and their use ofestimation of production. Canadian Journal of Fisheriesand Aquatic Sciences 45:315–324.

MORIN, A., AND P. DUMONT. 1994. A simple model to estimategrowth rate of lotic insect larvae and its value forestimating population and community production.Journal of the North American Benthological Society13:357–367.

MORIN, A., J. STEPHENSON, J. STRIKE, AND A. G. SOLIMINI. 2004.Sieve retention probabilities of stream benthic inverte-brates. Journal of the North American BenthologicalSociety 23:383–391.

RAMSAY, P. M., S. D. RUNDLE, M. J. ATTRILL, M. G. UTTLEY, P. R.

WILLIAMS, P. S. ELSMERE, AND A. ABADA. 1997. A rapid

method for estimating biomass size spectra of benthic

metazoan communities. Canadian Journal of Fisheries

and Aquatic Sciences 54:1716–1724.

REICE, S. R. 1980. The role of substratum in benthic

macroinvertebrate microdistribution and litter decom-

position in a woodland stream. Ecology 61:580–590.

RICKER, W. E. 1946. Production and utilization of fish

populations. Ecological Monographs 16:373–391.

ROSENBERG, D. N., AND V. H. RESH. 1993. Freshwater

biomonitoring and benthic macroinvertebrates. Chap-

man and Hall, New York.

SHIEH, S. H., J. V. WARD, AND B. C. KONDRATIEFF. 2002. Energy

flow through macroinvertebrates in a polluted plains

stream. Journal of the North American Benthological

Society 21:660–675.

SHIEH, S. H., J. V. WARD, AND B. C. KONDRATIEFF. 2003.

Longitudinal changes in macroinvertebrate production

in a stream affected by urban and agricultural activities.

Archiv fur Hydrobiologie 157:483–503.

STITES, D. L., AND A. C. BENKE. 1989. Rapid growth rates of

chironomids in three habitats of a subtropical blackwater

river and their implications for P:B ratios. Limnology

and Oceanography 34:1278–1289.

WALLACE, J. B., S. L. EGGERT, J. L. MEYER, AND J. R. WEBSTER.

1999. Effects of resource limitation on a detrital-based

ecosystem. Ecological Monographs 69:409–442.

WALLACE, J. B., J. W. GRUBAUGH, AND M. R. WHILES. 1996. Biotic

indices and stream ecosystem processes: results from an

experimental study. Ecological Applications 6:140–151.

WALLACE, J. B., AND J. R. WEBSTER. 1996. The role of

macroinvertebrates in stream ecosystem function. An-

nual Review of Entomology 41:115–139.

WATERS, T. F. 1977. Secondary production in inland waters.

Advances in Ecological Research 10:91–164.

Received: 11 September 2006Accepted: 11 June 2007

2007] 619TIMESAVING TECHNIQUES FOR PROCESSING BENTHIC SAMPLES