CSIRO PUBLISHINGMarine and Freshwater Research, 2014, 65, on control on stream invertebrate diversity:is periphyton architecture more important than biomass?Jonathan D. Tonkin A,B,D, Russell G. Death A and José Barquı́n CAAgriculture and Environment – Ecology (PN-624), Massey University, Private Bag 11-222,Palmerston North, New Zealand.BPresent address: Department of River Ecology and Conservation, Senckenberg ResearchInstitute and Natural History Museum, Clamecystrasse 12, Gelnhausen 63571, Germany.CEnvironmental Hydraulics Institute, University of Cantabria, C/ Isabel Torres no 15,Parque Cientı́fico y Tecnológico de Cantabria, 39011, Santander, Spain.DCorresponding author. Email: [email protected] There is little consensus on the form of the periphyton biomass–macroinvertebrate diversity relationship instreams. One factor that these relationships do not account for is the growth form of primary producers. We (1) examinedthe periphyton biomass–macroinvertebrate diversity relationship in 24 streams of Cantabria, Spain, in July 2007, and(2) determined whether this relationship was underpinned, and better explained, by specific responses to the growth formof the periphyton community. We hypothesised that macroinvertebrate diversity would be a log-linear function ofperiphyton biomass and would respond differently to two coarse divisions of the periphytic community; i.e. positively to%cover of non-filamentous algae and negatively to %cover of streaming filamentous algae. There was no relationshipbetween benthic periphyton biomass and macroinvertebrate diversity in these streams but, as predicted, this relationshipwas underpinned by responses to the growth form of periphyton community. Generally, macroinvertebrate diversityresponded positively to %cover of non-filaments and negatively to %cover of streaming filaments, although results werevariable. These findings suggest that periphyton biomass–macroinvertebrate diversity relationships in streams can beunderpinned by interactions with specific growth forms of periphyton. We suggest that further research is required todevelop robust thresholds of %cover of filamentous algae cover that would benefit managers wishing to minimise negativeeffects of eutrophication on stream communities.Additional keywords: algae, biomonitoring, Cantabria, changepoint, diatom, filamentous, macroinvertebrate, rapidassessment, river management, Spain, thresholds.Received 16 October 2013, accepted 10 December 2013, published online 16 June 2014IntroductionAlthough the relationship between productivity and diversity isa central theme in ecological research (Abrams 1995; Mittelbachet al. 2001), we are far from reaching a consensus on the form ofthe relationship either empirically or theoretically for bothproducers and consumers (e.g. Mittelbach et al. 2001; Adleret al. 2011). Differences in the observed patterns may be a resultof several factors, including the spatial scale of observation(Chase and Leibold 2002; Tonkin and Death 2013), disturbance(Huston 1994), history of community assembly (Fukami andMorin 2003) and differences between ecosystems and organisms studied (Mittelbach et al. 2001).Compared with lentic systems and indeed most other environments, few studies have specifically investigated whetherhigher productivity or, in fact, standing crop of algal biomassleads to greater diversity in lotic systems. The few to lookspecifically at this periphyton standing crop–invertebrate diversity relationship in streams have found both unimodal (DeathJournal compilation Ó CSIRO 2014and Zimmermann 2005) and log-linear (Death 2002; Tonkinet al. 2013) increases in diversity with productivity. Primaryproducers are principally periphytic algae in streams, whichvary greatly in their growth form and include prostrate, stalkedand filamentous forms (Hoagland et al. 1982; Steinman andMcIntire 1986), all of which respond differently to environmental conditions and grazing. However, typically, biomonitoringinvolves assessing periphyton biomass using either chlorophylla and/or ash-free dry weight (AFDW) and although these twomeasures are often highly correlated, they do not always respondin the same way to environmental conditions (Biggs and Hickey1994; Feminella and Hawkins 1995). Detailed assessment ofperiphyton community has been less widely used as an index forbiomonitoring environmental conditions (Pan et al. 1996; Hillet al. 2000). Typically, the focus of these assessments of bioticintegrity has been diatoms (Kelly and Whitton 1995; Pan et al.1996), although Whitton and Kelly (1995) advocated the use ofthe full community of plants including bryophytes.www.publish.csiro.au/journals/mfr
Stream periphyton–invertebrate relationshipsNot only do various growth forms of periphytic algaerespond differently to environmental conditions, but they canprovide diverse habitat and resources for higher trophic levels(Dudley et al. 1986). Different periphyton growth forms canalso fulfil different functional roles in benthic communities(Steinman et al. 1992). When levels of periphyton reach greaterdensities and epilithic films such as diatoms are replaced bymacroalgae such as filamentous green algae, interactionsbetween grazers and periphyton can shift from simple plant–herbivore interactions to more complex relationships. As well asproviding food for a few specialist taxa, macroalgae can bothprovide and remove habitat and compete for space with invertebrates. Dudley et al. (1986) classed invertebrates into thosenegatively affected by macroalgae because of competition forspace, positively affected because of habitat provision, andpositively affected by food provision. This can be reflected inthe typical shift from pollution (nutrient)-sensitive taxa associated with thin periphytic films, to pollution-tolerant taxa andfilamentous-algae growth forms often associated with nutrienteutrophication (Suren et al. 2003).We set out to (1) test the response of stream invertebratediversity metrics, often used in biomonitoring, to periphytonbiomass (assessed as chlorophyll a) and (2), because biomassmeasurements do not account for variation in the growth form ofprimary producers, to examine whether this link can be betterexplained by underlying responses to different growth forms ofperiphyton categorised coarsely into two major groups (i.e. allnon-filamentous films and mats and streaming filamentous0Marine and Freshwater Research819green algae). We also use a common stream-specific metric,%EPT (Ephemeroptera, Plecoptera and Trichoptera), to assesswhether this metric is more sensitive to environmental gradientsin streams than are simple invertebrate diversity measures.Percentage EPT is commonly used in stream bioassessmentbecause, as a result of their sensitivity, EPT taxa often respond inpredictable manners to changes in environmental conditions(Lenat 1988). As a result of previous work in streams (Death2002; Tonkin and Death 2012; Tonkin et al. 2013), we hypothesise that invertebrate diversity, including richness and rarefiedrichness, will increase logarithmically with increasing periphyton biomass, but we predict that this relationship will be underpinned by particular responses to different growth forms ofperiphyton. Specifically, because of the view that diatoms,which make up a large part of non-filamentous periphytic filmand mat growth forms, are considered favourable food andhabitat to many stream invertebrates and filamentous algaecan be poor habitat for many (but not all, e.g. Dudley et al.1986; Power 1990) invertebrates (Suren and Riis 2010),invertebrate diversity will respond positively to %cover ofnon-filamentous and negatively to %cover of streaming filamentous algae.Materials and methodsStudy sitesTwenty-four streams were sampled in the Cantabria region,Northern Spain, on one occasion in July 2007 (Fig. 1, Appendix 1).Kilometers5 1020Fig. 1. Map showing the location of 24 streams in Cantabria, Spain, sampled in July 2007. Latitudes andlongitudes are given in Appendix 1.
820Marine and Freshwater ResearchCantabria is a mountainous and coastal region, with both characteristics playing an important role in determining climate andriver morphology. Near the northern coast, valleys are below400 m asl, whereas the Cordillera Cantábrica, a west to eastrunning mountain range, reaches more than 2600 m asl in thesouth-west of the region. These steep inland valleys run northward and contain short ‘flashy’ rivers with high erosive power.Cantabria has a humid oceanic temperate climate, with anaverage annual temperature of 148C and an average annualprecipitation of ,1200 mm. Rainfall is regularly distributedthroughout the year, being heaviest in winter and spring. Stormsoccur in any season, and snow is common from late autumn toearly spring on the mountain ranges (for a more detaileddescription see Barquı́n et al. 2012). Land use surroundingsampling sites varied from Atlantic deciduous forest consistingpredominantly of oak (Quercus spp.) and European beech(Fagus spp.) to pasture and small urban settlements.Sites were selected from the following six river catchments:Rı́o Besaya, Rı́o Saja, Rı́o Pas, Rı́o Pisueña, Rı́o Nansa and Rı́oEbro. To account for local variation in factors such as geologyand land use, sites were selected in pairs a priori within each ofthe six catchments, so that one low- and one high-productivitysite in close geographic proximity were sampled. Becausethese were selected before sampling, productivity estimatesfor the selection of a priori high- and low-productivity streamswere based on one-off visual estimates of periphyton, which aredetailed below. All sites were cobble-bottom streams. Altitudeof the sites ranged from 163 to 1061 m asl and average channelwidth ranged from 1.9 to 30.7 m (Appendix 1). Riparian canopycover ranged from 1% to 80% cover (Appendix 1).Physicochemical variablesSeveral physical and chemical variables were measured onceduring the time of macroinvertebrate sampling. Depth and watervelocity were recorded with a Marsh–McBirney flowmate current meter (Marsh-McBirney, Frederick, MD) at five equidistantpoints along the thalweg. Conductivity, temperature, dissolvedoxygen concentration and percentage saturation, and pH weremeasured using a YSI 556 MPS meter (YSI Inc., YellowSprings, Ohio, USA). A 250-mL unfiltered water sample wascollected at each site and kept in the dark and on ice duringtransport, for later analysis of nitrate (NO 3 : cadmium-reduction:molybdatemethod)and ammoniamethod), phosphate (PO3 4(NH3: salicylate method), calculated using a Beckman CoulterDU Series 700 UV/Vis Scanning Spectrophotometer (BeckmanCoulter Inc., Brea, CA). Substrate size composition was assessed by sampling 100 stones using the Wolman walk methodology (Wolman 1954), and then converting these measurementsto a substrate-size index following Jowett and Richardson(1990). Substrate heterogeneity was assessed using the Shannondiversity index, whereas bed stability was measured using thebottom component of the Pfankuch stability index (Pfankuch1975). Finally, percentages of riparian vegetation and canopycover were visually estimated over a ,50-m reach.Biological collectionsPeriphyton biomass was measured by extracting chlorophyll afrom five stones (mean area: 60 cm2) collected randomly fromJ. D. Tonkin et al.riffles within the same ,50-m reach at each site. These werekept cool and in dark, before being frozen and taken back to thelaboratory. Chlorophyll a and phaeophytin were extracteddirectly from the stones by using 90% acetone at 58C for 24 h inthe dark. Absorbances were read on a Beckman Coulter DUSeries 700 UV/Vis Scanning Spectrophotometer and convertedto pigment concentration following Steinman and Lamberti(1996). Stone surface area was estimated from axial dimensionsfollowing Graham et al. (1988) and then halved to correct for theproportion of the stone available for periphyton growth.We calculated two other metrics to assess periphyton communities within riffles along the sampling reach, namely,%cover of all non-filamentous algae cover (i.e. periphyticfilms and mats) and %cover of streaming filamentous algae(i.e. clearly identifiable filamentous algae to the naked eye).These were visually assessed along three randomly locatedtransects across the entire width of the stream bed within rifflesalong the sample reach, using modified rapid assessment protocols from the New Zealand Stream Periphyton MonitoringManual (Biggs and Kilroy 2000). We assessed coverage ofthe two periphyton categories on every stone directly beneaththe transect line across the width of the streams. Where substratewas finer than ,10 mm, we did not examine the periphytoncover. Where coverage was uncertain between bare substrateand thin films, we also felt the rock surfaces with our hands toexamine coverage. We grouped all algal forms into the twocoarse groups of non-filaments (includes diatoms and all othercrustose, prostrate and stalked algae growth forms, as well asnon-streaming filamentous algae) and streaming filaments, withthe remaining being classified as having no cover. Even thoughthere is some potential for observer bias using this method,observations were made by the same person at all 24 sites.Recent research has highlighted that, given appropriate training,variability in estimates using visual assessment approachesmay not be a major problem (Kilroy et al. 2013).Five 0.1-m2 500-mm-mesh Surber samples were collected atrandom from riffles at each site and were preserved in 10%formalin in the field. In the laboratory, the samples were washedthrough 500-mm and 1-mm Endecott sieves before being identified and counted to the lowest possible taxonomic level.Invertebrates were mostly identified to morpho-species; however, where possible morpho-species were identified usingavailable keys (e.g. Tachet et al. 2000).The number of animals per 0.1 m2 (density) was calculatedfor each individual sample and averaged per site, as was thenumber of taxa (richness). Rarefied richness (ES[N]) wascalculated for 261 individuals, which was the lowest averagenumber of animals at a site. Rarefaction accounts for the passiveincrease in the number of taxa collected with increasing numberof individuals collected (Hurlbert 1971). This, in effect, standardises sites by predicting richness per a set number of animalsrather than a set area. The final community metric used was themean percentage of Ephemeroptera, Plecoptera and Trichoptera(EPT) animals per sample.Statistical analysisAll analyses were performed using R version 2.15.2 (R CoreTeam 2013). First, to assess any clear linkages between
Stream periphyton–invertebrate relationshipsMarine and Freshwater Research821Table 1. Pearson’s correlation coefficients for periphyton and macroinvertebrate community metrics against physicochemical variables collectedfrom 24 streams of Cantabria, Spain, July 2007DO, dissolved oxygen; OH cover, overhead cover; ES(261), rarefied taxonomic richness, calculated for 261 individuals; SI, size index; hetero., heterogeneity.***P , 0.0001, **P , 0.01, *P , 0.05 (significances are after correcting for false discovery tureDOOH coverVelocityDepthWidthSubstrate SISubstrate hetero.SlopeNitratePhosphateAmmoniaNo. of taxa (N)Log(N)ES(261)%EPT animalsChlorophyll a%Non-filaments%Streaming filaments%Bryophytes 0.090.27 0.07 0.170.110.270.170.18 0.260.02 0.06 0.070.34 0.40 0.19 0.290.170.200.080.50 0.110.18 0.13 0.29 0.31 0.12 0.440.27 0.170.120.210.51 0.140.16 0.13 0.3126.96.36.1990.30 0.120.050.15 0.160.41 0.39 0.25 0.43 0.090.10 0.03 0.31 0.07 0.040.040.310.05 0.230.05 0.09 0.09 0.26 0.17 0.39 0.240.000.390.64* 0.290.43 0.16 0.57 0.07 0.27 0.380.19 0.060.100.050.140.17 0.07 0.15 0.460.500.04 0.190.440.210.530.07 0.15 0.10 0.13 0.16 0.49 0.16 0.010.130.75*** 0.340.220.00 0.27 0.05 0.29 0.290.22 0.090.250.100.26 0.18 0.090.25 0.19 0.060.02 0.06 0.310.13 0.050.040.250.130.09 0.18 0.09physicochemical variables and both periphyton and invertebrates, we correlated all invertebrate, periphyton and physicochemical variables using Pearson’s correlation coefficient withthe rcorr() function in the R package ‘Hmisc’. We adjustedP-values for multiple comparisons by using the false discoveryrate method (Benjamini and Hochberg 1995) with the p.adjust()function in the R ‘stats’ package.We used linear regression to examine relationships amongperiphyton biomass, %cover of non-filaments, %cover ofstreaming filaments and invertebrate metrics, using the lm()function in the ‘stats’ package in R. Where required, welog(x þ 1)-transformed data to remove heteroscedasticity.Where both linear and quadratic regressions were run, we usedAkaike’s information criterion (AIC) to select the best-fittingcurve. We also regressed the three dominant taxa against thethree periphyton metrics. Where thresholds were apparent in theresponse of invertebrate metrics to %cover of streaming filamentous algae, we tested these using the cpt.mean() procedurein the R package ‘changepoint’ (Killick and Eckley 2011). Weused Bayesian information criterion (BIC) and ‘at most onechange’ (AMOC) to select the location of single change-pointsif present.To visually assess the multivariate structure of the macroinvertebrate community, we performed a non-metric multidimensional scaling (NMDS) ordination using the metaMDS()function in the Vegan package (Oksanen et al. 2011). We usedBray–Curtis distances and limited the number of NMDS axes totwo. To examine different influences of the three periphytonmetrics (chlorophyll a, %cover of non-filaments, and %cover ofstreaming filaments), we fitted smooth-surface thinplate splinesusing the ordisurf() function in Vegan. This uses generalisedadditive models (GAMs) to overlay a smoothed responsesurface, which allows a more detailed interpretation than doesa simple linear vector. More specifically, it enables non-lineareffects of the three periphyton metrics on macroinvertebratecommunity structure to be examined visually.ResultsPhysicochemical variables and periphytonPeriphyton biomass, assessed as mean chlorophyll a, ranged from2.58 to 15.35 mg cm 2, with a mean s.e. of 5.8 0.7 mg cm 2.Chlorophyll a was positively correlated with %cover ofstreaming filamentous algae (r ¼ 0.63, P ¼ 0.039). Percentagecover of non-filaments ranged between 8.3% and 95.0%, with amean of 48.3 5.6% cover, and %cover of streaming filamentsaveraged 24.0 6.1%, ranging between 0.0% and 91.7%.Conductivity ranged from 68 to 402 mS cm 1 (Appendix 1).Conductivity was positively correlated with chlorophyll a and%cover of streaming filamentous algae (Table 1).Invertebrate community compositionMean taxonomic richness ranged between 9.6 and 21.0 taxaper 0.1 m2 with a mean s.e. of 15.6 0.6 taxa per 0.1 m2 andthe number of individuals averaged 928.2 406.7 individualsper 0.1 m2, ranging between 261.4 and 10194.2 taxa per 0.1 m2.Percentage EPT ranged between 11.5% and 83.6%, with a meanof 56.8 3.5%.Ephemeroptera was the most abundant family, making upbetween 12.6% and 75.2% of animals at each site, with anaverage (mean s.e.) of 47.9% 3.2% of the communitycomposition, followed by Diptera (25.9% 4.1%; range 3.1–83.2%), Coleoptera (5.1% 0.7%), Trichoptera (4.6% 0.8%)and Plecoptera (3.6% 0.8%). These patterns were largely dueto the dominance of three individual taxa. Baetis spp., onaverage, made up 41.4% 3.0% of the community, followedby Prosimulium spp. (17.8% 3.7%), and Echinogammarusspp. (7.2% 2.1%).In response to gradients of periphyton cover, %Prosimulium spp. was lowest at intermediate levels of %coverof non-filaments (F2,21 ¼ 7.39, P ¼ 0.004, R2 ¼ 0.41, y ¼ 55.45 2.09x þ 0.02x2), whereas it was not related to %cover ofstreaming filaments (F2,21 ¼ 2.94, P ¼ 0.075, R2 ¼ 0.22,
822Marine and Freshwater ResearchJ. D. Tonkin et al.y ¼ 19.48 0.79x þ 0.01x2). Percentage Baetis spp. peaked atintermediate levels of %cover of non-filaments (F2,21 ¼ 6.25,P ¼ 0.007, R2 ¼ 0.37, y ¼ 11.88 þ 1.61x 0.02x2), whereas itwas also not related to %cover of streaming filaments (F2,21 ¼2.53, P ¼ 0.10, R2 ¼ 0.19). Moreover, % Echinogammarus spp.did not respond to %cover of either non-filaments (F1,22 ¼ 1.49,P ¼ 0.24, R2 ¼ 0.06) or streaming filaments (F1,22 ¼ 1.98,P ¼ 0.17, R2 ¼ 0.08).Ordination on log(x þ 1)-transformed invertebrate data produced a reasonable fit, with a stress of 0.19 (Fig. 2). OverlayingGAM-fitted smooth surfaces for each of the three periphytonmetrics indicated three different effects on the structure ofmultivariate invertebrate community (Fig. 2). Streaming filamentous algae loaded negatively on NMDS 1, whereas nonfilaments cover exhibited a clear negative loading on NMDS 2(Fig. 2). However, the influence of chlorophyll a on the structureof invertebrate community was more non-linear, with thestrongest gradient on NMDS 1 but the lowest value situatedcentral in ordination space.Chlorophyll a0.20 0.2 0.60.4NMDS2Density and diversity patternsThe only aspect of the invertebrate community to respond tochlorophyll a was the number of individuals, which increasedmonotonically with increasing biomass; however, althoughsignificant, the explained variance was low (Fig. 3, Table 2).Taxonomic richness, rarefied richness and %EPT animals werenot related to chlorophyll a (Fig. 3, Table 2).Density and diversity measures exhibited opposingresponses to the two growth forms of periphyton measured.Taxonomic richness and rarefied richness increased log-linearlywith increasing substrate cover of non-filaments, but the numberof individuals was not related to %cover of non-filaments(Fig. 3, Table 2). The percentage of EPT animals appearedmore sensitive to higher percentage cover of non-filamentousalgae, and peaked strongly at intermediate levels and declined athigher levels of %cover of non-filamentous algae (Fig. 3,Table 2). However, this was largely dependent on one site whichhad 95% cover. Removing this site strengthened the fit andaltered the relationship to a quadratic increase (F2,20 ¼ 9.77,P ¼ 0.001, R2 ¼ 0.49, y ¼ 24.36 þ 1.36x – 0.01x2).Taxonomic richness was not linearly related to %cover ofstreaming filamentous algae; however, density of individualsexhibited a quadratic increase with increasing %cover (Fig. 3,Table 2). Both rarefied richness and %EPT animals respondednegatively to streaming filamentous algae, exhibiting a curvilinear decline with an increasing cover of streaming filamentousalgae (Fig. 3, Table 2). However, removing the site with 92%cover of streaming filamentous algae removed any relationshipwith %EPT (F2,20 ¼ 0.41, P ¼ 0.67, R2 ¼ 0.04). Changepointanalysis indicated that taxonomic richness exhibited a thresholdresponse to increasing streaming filaments at 40% cover, with adrop in mean richness from 16.27 taxa below and including40% cover to 13.60 taxa above 40% cover. Rarefied richnessexhibited a similar threshold response, with a drop from 15.32 to11.73 taxa above 40% cover of streaming filaments. Changepoint analysis did not return a significant threshold response of%EPT animals to the cover of streaming filamentous algae,despite %EPT being considerably lower at the last data point of92% cover.0.4 0.4 0.200.20.4 0.200.20.400.20.4% Non-filaments0.20 0.2 0.60.4 0.4% Streaming filaments0.20 0.2 0.6 0.4 0.2NMDS1Fig. 2. Non-metric multidimensional scaling (NMDS) ordination onlog(x þ 1)-transformed invertebrate-community data collected from 24streams in Cantabria, Spain, July 2007. Individual plots display overlaidsmooth-surface thin-plate splines using generalised additive models(GAMs) for the three periphyton metrics. Numbers on the splines representthe values of the periphyton metric. 2D stress ¼ 0.19.DiscussionThere was no relationship between periphyton biomass andinvertebrate diversity in the present study. Recent studies instream communities have found log-linear increases in diversitywith periphyton biomass (e.g. Death 2002; Tonkin and Death2012; Tonkin et al. 2013), which, along with the belief thatstream-wide competitive exclusion does not often materialise at
Stream periphyton–invertebrate relationshipsMarine and Freshwater Research823No. of taxa 0.1 m 22015105(b)(c)(d )(e)(f )(g)(h)(i )(j)(k)(a)Log10 no. animals 0.1 m 204321Rarefied richness2015105% EPT animals75502500510Chlorophyll a (µg cm 2)150(l )255075% Non-filaments0255075% Streaming filamentsFig. 3. Mean ( 1 s.e.) (a–c) taxonomic richness, (d–f ) number of animals, (g–i) rarefied richness (ES(261)), and (j–l) %EPT animals as a function of (a, d, g, j)chlorophyll a, (b, e, h, k) %cover of non-filamentous algae, and (c, f, i, l) %cover of streaming filamentous algae in 24 streams of Cantabria, Spain, July 2007.Grey area represents 95% confidence interval of the regression line. Vertical shaded area (light grey), between two dotted lines, on c and i represent thesignificant threshold point identified using changepoint analysis. Regression equations are given in Table 2.high periphyton biomass in streams, led us to predict that thislog-linear trend would occur in these Spanish streams. However,no clear link was evident between periphyton biomass and anyof the metrics used.Relationships with periphyton growth formAlthough invertebrate communities did not respond clearly tochanges in periphyton biomass, the growth form of the periphyton community was important in determining diversity
824Marine and Freshwater ResearchJ. D. Tonkin et al.Table 2. Results for regression analysis for taxonomic richness, number of animals, rarefied richness and %EPT animals against periphytonmetrics for 24 streams in Cantabria, Spain, July 2007Degrees of freedom for linear and log-linear models are 1,22 and for quadratic models 2,21. AIC ¼ Akaike’s information criterion for the selection of the bestmodel among linear, log-linear and quadratic curves. Lowest values represent the best modelParameterF (AIC)PR2Equation 2Chlorophyll a (mg cm )No. of taxaLog(no. of animals)ES(261)%EPT animalsFilms and mats cover (%)No. of taxaQuadraticLog(no. of animals)ES(261)Quadratic%EPT animalsQuadraticFilamentous algae cover (%)No. of taxaLog(no. of animals)QuadraticES(261)Quadratic%EPT 0020.1940.0400.049Non-significanty ¼ 2.43 þ 0.037xNon-significantNon-significant9.62 (42.8)4.91 (43.1)0.6212.77 (42.3)6.38 0.3190.0270.3670.3780.1440.471y ¼ 7.55 þ 2.2 ln(x)y ¼ 10.7 þ 0.19x – 0.002x2Non-significanty ¼ 5.73 þ 2.37 ln(x)y ¼ 9.08 þ 0.21x – 0.0016x2Non-significanty ¼ 15.35 þ 1.98x – 0.02x21.5912.30 (4.18)13.71 ( 3.21)5.54 (118.1)4.22 670.3590.5660.2010.2870.1430.285Non-significanty ¼ 2.51 þ 0.006xy ¼ 2.59 – 0.011x þ 0.0002x2y ¼ 15.45 – 0.043xy ¼ 14.93 þ 0.059x – 0.001x2Non-significanty ¼ 57.93 þ 0.58x – 0.01x2patterns. There has been extensive research on the effects ofgrazers on algal communities, and this top-down control hasbeen the central focus of periphyton–invertebrate communityrelationships (Hillebrand 2009). However, of note is the fact thatthe majority of this research has focused solely on the control ofperiphyton biomass and not on different growth forms. Ourresults have demonstrated that focusing simply on biomass ispotentially masking important underlying relationships. Otherthan grazer-specific responses (e.g. Gresens and Lowe 1994;Maasri et al. 2008), the bottom-up effects of algal assemblageson invertebrate communities has received little direct attentionin streams (but see Dudley et al. 1986; Koksvik and Reinertsen2008). Nonetheless, it is clear that the growth form of periphytonhas strong influences on the structure of stream benthic communities (Dudley et al. 1986; Koksvik and Reinertsen 2008),and grazing communities can in fact grow at different ratesdepending on the dominant algal growth form (Feminella andResh 1991).In the present study, although variation was evident in theshape of relationships, general patterns suggest that there wereopposing influences of the two main growth forms detected.Namely, %cover of non-filamentous algae exerted a positiveresponse and %cover of streaming filamentous algae a negativeresponse on invertebrate diversity. Differences were mainlydue to changes in the densities of the three dominant taxa,namely, the blackfly larvae, Prosimulium spp., the mayfly,Baetis spp., and the amphipod, Echinogammarus spp. Moreover, the response to the dominant growth forms appeared to behighly species specific depending on feeding habits, such asfavouring more palatable epilithic films or drift-feeding onfilamentous algal cells.Percentage of non-filaments was the best predictor of diversity, with both taxonomic and rarefied richness increasing loglinearly as cover increased. This mirrors the hypothesis we set ofa log-linear curve for the relationship between periphytonbiomass and diversity that several recent studies have found inbenthic communities (e.g. Death 2002; Tonkin and Death 2012).Although diatoms are just one of the groups of algae thatcomprise our ‘non-filaments’ classification, they are likely tomake up a large component of these films and mats. Diatoms arethe most important food source for a high proportion of benthicinvertebrates, because grazers tend to be able to assimilatediatoms better than other algal taxonomic classes (Lambertiet al. 1989).The percentage of EPT animals can respond to shifts inperiphyton biomass (Tonkin et al. 2009); however, we found nosuch relationship in the present study. Percentage EPT didrespond to the growth form of periphyton, declining at thehighest levels of %cover of both filaments and non-filaments;however, these trends were influenced by individual sites at theend of the spectrum of cover. The decline at higher levels ofcover of non-filaments may be due to the fact that more palatableforms of periphytic films, such as d
measured using a YSI 556 MPS meter (YSI Inc., Yellow Springs, Ohio, USA). A 250-mL unfiltered water sample was collected at each site and kept in the dark and on ice during transport,forlateranalysisofnitrate(NO 3:cadmium-reduction method), phosphate (PO 4 3: molybdate method) and ammonia