Effect of riparian vegetation on diatom assemblages in headwater streams under different land uses

Science of the Total Environment 475 (2014) 234–247

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Effect of riparian vegetation on diatom assemblages in headwater streams underdifferent land uses

Daša Hlúbiková, Maria Helena Novais, Alain Dohet, Lucien Hoffmann, Luc Ector ⁎Public Research Centre-Gabriel Lippmann, Department of Environment and Agro-biotechnologies (EVA), 41 Rue du Brill, 4422 Belvaux, Luxembourg

H I G H L I G H T S

• Diatoms in headwaters with different riparian cover and land use were compared.• We assessed diatom assemblage structure, structure of diatom guilds and IPS index.• Diatoms at impacted sites were similar regardless of the status of riparian cover.• Diatoms assemblages were mainly driven by urbanization and nutrients.• Riparian vegetation did not buffer impacts of the catchment land use on diatoms.

⁎ Corresponding author. Tel.: +352 470261 421; fax:E-mail address: [email protected] (L. Ector).

0048-9697/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.scitotenv.2013.06.004

a b s t r a c t
a r t i c l e i n f o
Article history:Received 31 December 2012Received in revised form 20 April 2013Accepted 2 June 2013Available online 29 June 2013

Editor: Christian EW Steinberg

Keywords:Riparian bufferHeadwatersDiatom guildsBenthic diatomsLand use

Differences in the structure of diatom assemblages in headwaterswith contrasting shading conditions and differ-ent land use in the buffer zone and upper catchmentwere studied in order to evaluate the influence of the lack ofriparian vegetation on the biofilm. The objective was to ascertain whether a riparian buffer canmitigate the neg-ative influence of human induced disturbance and pollution on diatom assemblages in headwaters. Four streamswere selected in order to maximize the differences in the land cover and minimize other environmentalgradients. Multivariate statistics, different comparative and permutation tests and correlations were applied tocompare the diatom assemblages, the Specific Polluosensitivity Index (IPS) and the diatom ecological guilds(low profile, high profile and motile) among the sites studied and to evaluate their responses to disturbances.The analysis showed that lowprofile diatoms typically dominated in forested headwaterswith limited resources,whilst assemblages at impacted sites showed awider range of growth forms. In unimpacted streams, the diatomassemblages were influenced by temperature, pH, conductivity and calcium, as usually reported for oligotrophicstreams with high natural disturbance due to fast current and shading. In both shaded and unshaded impactedstreams, the importance of nutrients and land use disturbance, especially urbanization, prevailed. This trendwasalso reflected by the IPS index that showed consistently lower values at impacted sites, correlating most signif-icantly with nutrients. The diatom species composition as well as diatom guilds at impacted sites were similar,regardless of the presence or absence of riparian vegetation, and were significantly influenced by seasonalchanges.Our results indicate that diatoms react sensitively to alterations of the water environment in headwaters, in-duced by anthropogenic activities, and these impacts are not buffered by an intact riparian zone. Diatomsclosely reflected land use practices in the upper catchment regardless of the buffer zone status.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The importance of the role of riparian vegetation in determining thestructure and function of stream ecosystems has long been recognized(Vannote et al., 1980; Knight and Bottorff, 1984; Richardson andDanehy, 2007). Headwater streams are greatly influenced by riparianvegetation since they function as processors of organic matter comingfrom the watershed (Cummins and Spengler, 1978). The riparian

+352 470264.

rights reserved.

interface regulates stream temperature (Hetrick et al., 1998; Moore etal., 2005; Studinski et al., 2012) and is reported to function as a filter,buffer and stabilizer (Keller and Swanson, 1979; Knight and Bottorff,1984; Sabater et al., 2003). These effects are particularly strong in for-ested headwaters (Studinski et al., 2012). Headwater streams andtheir riparian areas differ from downstream reaches in a number of fun-damental ways that shape their characteristic biological communities(Richardson and Danehy, 2007). These streams are characterised bysmall channel size, closed canopy resulting in reduced light conditions,higher input rates of organic matter and low primary production. Thepredominance of organic matter favours detritus-based communities

http://dx.doi.org/10.1016/j.scitotenv.2013.06.004

mailto:[email protected]

http://dx.doi.org/10.1016/j.scitotenv.2013.06.004

http://www.sciencedirect.com/science/journal/00489697

235D. Hlúbiková et al. / Science of the Total Environment 475 (2014) 234–247

(Richardson and Danehy, 2007), whilst the biomass of primary pro-ducers, such as algae, is limited due to reduced light (Sabater et al.,1998; Greenwood and Rosemond, 2005). Thus, the removal of riparianvegetation reduces inputs of detrital organic matter and increases lightavailability to the stream community, which not only increases the po-tential for primary producers (Sabater et al., 1998), but also increasesthe summer water temperature (Richardson, 2004) and changes thewater quality and quantity (Knight and Bottorff, 1984). Many of thestudies carried out to date have indicated that riparian zones providesome degree of protection (Richardson and Danehy, 2007). The mostcommonapproach to protect streams fromagricultural or forestry prac-tices is to use some form of riparian buffer to preserve some of thestream-riparian functions (Richardson, 2004). In this context, headwa-ter streams are rather understudied, in part because they do not repre-sent a management concern. However, intact headwater streams arecrucial to the functioning of river systems (Meyer and Wallace, 2001)and they have been shown to be critical sites in river networks for pro-cesses such as nutrient uptake and retention (Peterson et al., 2001).

Catchment land use is often recognized as oneof themost significantstressors of stream ecosystems. There are several principal mechanismsby which land use influences stream ecosystems: sedimentation, nutri-ent enrichment, contaminant pollution, riparian disturbance andhydrologic alterations (Allan, 2004). Thus different land use patternslead to changes in water chemistry (Johnson et al., 1997; Ometo et al.,2000), hydrology and physical habitat conditions (Dunne and Leopold,1978; Roth et al., 1996), which eventually decrease the biological integ-rity of streams. Despite the abundance of literature on the effects of landuse on stream ecosystems (Leland and Porter, 2000; Pan et al., 2004;Newall and Walsh, 2005; Binckley et al., 2010; Studinski et al., 2012),quantifying the relationships between land use and the biological integ-rity of these habitats remains challenging. Headwater streams are par-ticularly vulnerable to changing land use and non-point sourcepollutants, as small-order streams have a greater contribution of water-shed area to stream area compared with larger streams (Selby et al.,1985). “They may experience greater nutrient inputs than largerstreams owing to atmospheric deposition, saturation of terrestrial eco-systems, or mobilization from soils from the surrounding catchment”(Greenwood and Rosemond, 2005, see also Selby et al., 1985).

Stream periphyton is, contrary to secondary producers, directlyinfluenced by changes related to light and nutrients availability. Inheadwaters, the benthic algal communities are typically dominated bydiatoms (Cantonati, 1998; Greenwood and Rosemond, 2005; Danehyet al., 2007; Niedermayr and Schagerl, 2010). Benthic diatoms havelong been recognized as reliable indicators of organic pollution, eutro-phication and general pollution (Van Dam et al., 1994) and could betherefore considered as the only applicable indicator among primaryproducers in headwaters to detect impacts of stream ecosystem alter-ations. In this context, the focus of our work was to evaluate whetherthe impact of land uses such as urbanization and pasture on headwaterstreams can bemitigated by an intact forested buffer zone andwhetherdiatoms reflect this buffer effect. Since such land use practices arereported to lead to significant nutrient enrichments of river systems(Leland and Porter, 2000; Rhodes et al., 2001; Inwood et al., 2005), dia-toms should sensitively reflect all related changes in the stream.

Benthic diatoms are seen as reliable indicators of the impacts of dif-ferent land use practices on stream ecosystems (Pan et al., 2004;Newall and Walsh, 2005; Hering et al., 2006; Walker and Pan, 2006;Walsh and Wepener, 2009). Most commonly, the effect of land uses ondiatom assemblages is related to specific water quality variables suchas nutrient concentrations, pH (Zampella et al., 2007) or salinity (Blinnand Bailey, 2001). The majority of these studies are typically based onthe calculation of different diatom indices of water quality, which havebeen developed to assess pollution in rivers (Coste in Cemagref, 1982;Kelly and Whitton, 1995; Coring et al., 1999; Rott et al., 1999, 2003).The diatom indices calculation is usually based on the specific sensitivityof species to general or specific pollution (or to nutrient enrichment in

general) and the species abundance. Among the metrics, the SpecificPolluosensitivity Index (IPS) (Coste in Cemagref, 1982) is currently themost common diatom-based metric applied to the ecological status as-sessment of running waters in Europe (Kelly et al., 2009). Compared toother diatom indices, IPS was developed from a large database and in-volves large number of diatom taxa in the calculation, whose ecologicalcharacteristics are being regularly updated based on recent data fromthe monitoring networks. However, the high number of species andtheir unstable and fast changing taxonomy, which results in differentspecies concepts being adopted by different authors, brings the accuracyof purely diatom metric-based assessment into question. Therefore, an-other assessment approach was recently proposed, which classifies dia-tom genera into three diatom guilds (low profile, high profile, motile)based on their growth form and potential to tolerate nutrient limitationand physical disturbance (Passy, 2007) and is currently being tested(Berthon et al., 2011; Gottschalk and Kahlert, 2012; Rimet andBouchez, 2012). The assignment to the different guilds not only refersto the tolerance of species to nutrient supply, but also provides informa-tion on the structure of the diatom biofilm and resistance to disturbance.The low profile guild is favored in nutrient-poor and high physicaldisturbance habitats; the high profile guild reaches a maximum innutrient-rich sites and in conditions of low flow disturbance; and themotile guild increases along the nutrient gradients and decreases alongthe disturbance gradient (Passy, 2007).

In our study, we aimed to evaluate the indicative potential of epilithicdiatoms in headwaters with different riparian cover and different degreeof land use practices in the catchment and the buffer zone. The specificobjectives were (1) to evaluate howmuch the differences in the ripariancover influence diatom communities in headwater streams with differ-ent land use areas; (2) whether the presence of riparian vegetation andforestry in the buffer zone of a stream might act as a barrier against thenegative influence of land use in the catchment (urbanization, pastureand crop land) on benthic diatoms and (3) whether diatom indicatorssuch as diatom guilds and the IPS index have comparable indicativepower across different degrees of nutrient enrichment and disturbance.This was undertaken by elucidating how water chemistry is changedby different land use patterns in the upper catchment and the bufferzone, whether the presence of riparian vegetation influences these rela-tionships and, subsequently, how the diatom community structure, theproportion among the diatom guilds and the IPS index reflect thesechanges and impacts on aquatic system. The relative importance of theeffects of land uses in the upper catchment and the buffer zone onwater chemistry and diatoms was explored.

2. Methods

2.1. Site Selection and Description

This research was conducted in the Attert, Ernz noire and Syre riverbasins in Luxembourg (Fig. 1). Four streams were selected in the west-ern highlands eco-region of Luxembourg, with contrasting shading con-ditions and land use. The streams were selected to represent threedifferent types of conditions depending on the anthropogenic impactand riparian cover: unimpacted forested (shaded) conditions, impactedforested (shaded) conditions and impacted open (unshaded) condi-tions. The Schwaartzbaach stream represented unimpacted forestedconditions, it occurs in an intact forested area with no or minimalanthropogenic influence in the buffer zone and upper catchment. TheConsdorferbaach stream represented impacted forested (shaded)conditions and it occurs in an intact forested area similar to theSchwaartzbaach, but with intensive land use in the upper catchment.The riparian vegetation cover of both forested streams is composed ofnative deciduous woody vegetation, leading to overall shading of 85–100% of the stream channel, depending on the daily period. TheSauerbaach and Hemeschbaach streams represented impacted open(unshaded) conditions; they occur in areas with intensive land use in

Fig. 1. Location of the streams and sampling sites in Luxembourg.

236 D. Hlúbiková et al. / Science of the Total Environment 475 (2014) 234–247


the upper catchment and buffer zone and have very scarce or absent ri-parian vegetation and open river channel. The catchment land use prac-tices at all impacted streamswere dominated by pasture, crop land andurbanization; the buffer zone of the two open streams (Sauerbaach,Hemeschbaach) was intensively used for pasture. The percentages ofdifferent land uses in the buffer and the catchment zone of the fourstreams are summarized in Table 1. The streams were selected inorder to minimize the possible natural variability within the differentconditions as much as possible. In order to maximize the differences inthe landuse and to eliminate additional source of variation in the dataset,the experimental headwater streams were selected in geologically andmorphologically homogeneous catchments. Marls, limestones, dolo-mites and alluvials compose the lithology of the area and give a calcare-ous nature to the substrate, similar sediment granulometry and a highwater mineralization to the four streams investigated. The streams se-lected belong to the small and mid-altitude stream type in the Gutland(Ferréol et al., 2005). These are generally low-to-middle gradientstreams flowing through U-shaped valleys (upper course) and moremeandering lower reaches. Due to the calcareous geology of this region,pH values range between 7.0 and 8.7 (Table 1). The meso-lithal micro-habitats are predominant on the sampling sites and the mineral sub-strates are characterized by a significant proportion of fine-to-mediumgrain-sized sediments. Organic substrates generally range between 10%and 50% and comprise considerable amounts of coarse particulate organ-ic matter (e.g. fallen leaves in pools). Additional biotic substrates arefloating riparian vegetation, fine roots of woody riparian vegetationand deposits of fine particulate organic matter. Forested (shaded)streams (Schwaartzbaach and Consdorferbaach) lie in floodplains, dom-inated by native deciduous woody vegetation (mainly Fagus sylvatica L.,Carpinus betulus L. and Quercus petraea (Matt.) Liebl.) and non-native

Table 1Mean values and standard deviation of the environmental variables. B: buffer zone. C: catchmmarked (based on Kruskal-Wallis test).

Schwaartzbaach Cons

(forested unimpacted FS1) (fore

Variable Unit Mean/SD Range Mea

Elevation * m a.s.l. 286 ± 29 248/327 323Distance to source m 2204 ± 1202 491/3794 2810Catchment area upstream km2 4.9 ± 3.2 0.6/8.7 4.9Slope ** m 3.7 ± 1.8 2.5/7.2 6.4Forestry B ** % 99 ± 2 95/100 91Crop land (Crop B) ** % 1 ± 2 0/5 1Pasture B (Past B) ** % 0 ± 0 0/0 6Urbanization (Urban B) * % 0 ± 0 0/0 1Total disturbance B (Disturb B) ** % 1 ± 2 0/5 8Land Use Index B (LUI B) ** % 0.0 ± 0.0 0.0/0.1 0.1Forestry C ** % 80 ± 6 74/91 22Crop land C (Crop C) ** % 11 ± 4 3/14 38Pasture C (Past C) ** % 6 ± 1 4/8 28Urbanization C (Urban C) ** % 0 ± 0 0/0 9Total disturbance C (Disturb C) ** % 17 ± 5 7/22 75Land Use Index (LUI C) ** % 0.3 ± 0.1 0.1/0.4 1.4Temperature-1 month average (t-1 m) °C 7.79 ± 3.10 9.54Temperature-2 months average (t-2 m) °C 8.89 ± 2.69 9.74Temperature-3 months average (t-3 m) °C 9.80 ± 1.95 9.99pH - 8.0 ± 0.0 8.0/8.0 8.3Conductivity (Cond) ** μS/cm 558 ± 115 145/699 804Oxygen (O2) % 94 ± 12 85/103 95Phosphates (PO4

-3) ** mg/l 0.05 ± 0.04 0.00/0.13 1.06(PO4-P) * mg/l 0.01 ± 0.01 0.00/0.04 0.25Chlorides (Cl-) ** mg/l 21 ± 10 12/43 62Nitrites (NO2-N) * mg/l 0.00 ± 0.01 0.00/0.02 0.08Nitrates (NO3-N) ** mg/l 3.5 ± 1.5 1.3/7.6 7.2Sulphates (SO4

-2) ** mg/l 18 ± 2 15/23 33Ammonium (NH4-N) * mg/l 0.00 ± 0.01 0.00/0.04 0.26Potassium (K+) ** mg/l 1.4 ± 0.3 1.0/2.1 4.9Magnesium (Mg2+) ** mg/l 38 ± 4 32/45 7Calcium (Ca2+) ** mg/l 65 ± 6 56/77 118Total organic carbon (TOC) ** mg/l 3.8 ± 3.2 0.9/12.2 3.8

coniferous forests (mainly Picea abies (L.) H. Karst.). Open streams(Hemeschbaach and Sauerbaach), except in their most upper course,are characterized by open river channel with no or only scarce woodyriparian vegetation (mainly Fraxinus excelsior L., Alnus glutinosa (L.)Gaertn. and Salix spp.).

Five sampling sites were selected on each stream, situated along alongitudinal gradient, at about 1–2 km intervals, with the first sam-pling site situated near to the spring source. The sampling sites didnot contain any lentic facies and were characteristic by similar hydro-dynamic conditions. The streams with distribution of sites are repre-sented on Fig. 1.

2.2. Sampling Design and Sample Processing

Sampling was carried out in spring and autumn 2010 and 2011. Intotal, 20 sites were sampled during each sampling campaign; eachsite was sampled four times during the whole survey. The samplingwas carried out during hydrologically stable period in order to mini-mize the possible differences in physical disturbance such as differentdischarge conditions in the sites studied. The Consdorferbaach (im-pacted forested) and Sauerbach (impacted open) were sampledtwice in autumn 2010 due to extremely low discharge to ensure thecomparability of samples within the dataset.

Environmental parameters, such as, pH, concentration of dissolvedoxygen, oxygen saturation and specific conductivity, were measured insitu with the Hach HQ40d portable meter (Wissenschaftlich-TechnischeWerkstätten GmbH, WTW). Water samples for physico-chemical analy-seswere gathered simultaneouslywithdiatomsampling, in order to char-acterize the general water chemistry, nutrient content and degree oforganic pollution. All variables are listed in Table 1. Water temperature

ent. Variables significantly different among streams with *p b 0.05 and **p b 0.001 are

dorferbaach Hemeschbaach Sauerbaach

sted impacted FS2) (open impacted IS1) (open impacted IS2)

n/SD Range Mean/SD Range Mean/SD Range

± 50 252/378 317 ± 38 268/362 304 ± 28 275/345± 2332 207/5843 2548 ± 2105 232/5774 3179 ± 2298 470/6450± 4.3 0.3/10.6 5.8 ± 5.2 0.5/13.6 5.9 ± 5.0 0.3/13.2± 1.6 4.9/8.9 2.7 ± 0.7 1.5/3.5 2.4 ± 2.2 0.6/6.2± 14 64/100 14 ± 13 0/30 12 ± 21 0/53± 3 0/7 7 ± 10 0/24 27 ± 23 0/47± 10 0/25 70 ± 11 60/89 58 ± 39 0/100± 1 0/3 8 ± 16 0/40 1 ± 2 0/4± 14 0/35 85 ± 13 70/100 87 ± 21 47/100± 0.2 0.0/0.5 1.2 ± 0.6 0.7/2.2 1.2 ± 0.3 0.9/1.5± 15 3/37 34 ± 14 27/60 53 ± 27 27/88± 15 23/60 23 ± 7 17/33 16 ± 11 1/26± 4 22/33 37 ± 13 16/50 25 ± 16 5/42± 3 5/13 2 ± 2 0/4 2 ± 2 0/4± 17 59/96 62 ± 14 35/70 44 ± 29 6/71± 0.4 1.1/1.9 0.9 ± 0.2 0.5/1.1 0.7 ± 0.4 0.1/1.1± 0.54 9.35 ± 2.78 8.61 ± 3.55± 1.28 9.60 ± 3.14 9.84 ± 3.06± 2.18 9.93 ± 3.92 10.15 ± 3.55± 0.1 8.2/8.4 7.8 ± 1.1 7.0/8.5 8.3 ± 0.2 7.9/8.7± 209 588/1305 671 ± 522 121/2750 905 ± 149 727/1381± 5 82/99 80 ± 22 17/97 97 ± 11 72/119± 1.16 0.02/4.84 0.60 ± 0.83 0.01/2.78 0.26 ± 0.24 0.00/0.83± 0.37 0.00/1.58 0.20 ± 0.27 0.00/0.91 0.09 ± 0.08 0.00/0.27± 58 15/230 40 ± 50 4/239 32 ± 9 22/55± 0.15 0.00/0.50 0.07 ± 0.12 0.00/0.49 0.04 ± 0.04 0.00/0.15± 3.6 1.6/13.8 5.5 ± 3.8 0.0/13.3 2.7 ± 1.6 0.7/7.4± 11 15/51 30 ± 36 6/170 166 ± 69 41/291± 0.50 0.00/1.78 0.14 ± 0.31 0.00/1.29 0.19 ± 0.43 0.00/1.74± 1.3 1.8/6.9 3.6 ± 1.8 0.6/6.5 3.0 ± 0.7 2.1/4.4± 10 4/48 34 ± 15 7/49 27 ± 18 6/49± 20 76/166 66 ± 31 12/157 148 ± 21 100/188± 1.1 2.2/6.1 8.4 ± 3.7 4.0/16.6 3.4 ± 0.9 2.0/5.3


was continuously measured (at 20 min intervals over the course of atwo-year period) by data loggers (Onset HOBO pendant data logger,model UA-002-64, Onset computer corporation, Pocasset, MA) placed ateach sampling point, in order to detect differences in water temperaturedepending on the canopy cover during the different seasons.

Benthic diatoms were sampled following standard methods (CEN13946: European Committee for Standardization, 2003), which con-sisted of scrubbing a minimum of five stones (cobbles if possible)using a toothbrush. Stones used for sampling were found in the mainflow with fast current, completely submerged and in stable conditionsin order to minimize the variation caused by hydrological conditionsand related physical disturbance. With regard to the light conditions,the forested sites had the riverbed shadedwithout anywell lighted sec-tions thus the substrate sampled occurred in shaded conditions duringmost of the day; the open sites had the riverbed completely unshadedand the stones were well exposed to the sunlight.

Prior to any treatment, diatom samples were checked under aLeica DMRX optical microscope with 40x objective in order to esti-mate the ratio of live/dead frustules. The estimation of dead frustulesproportion was always below 10%.

Diatom samples were further treated using the hot hydrogen per-oxide method, according to CEN 14407 (European Committee forStandardization, 2004), to obtain clean frustule suspensions. Aftereliminating the organic matter from the diatom suspension, dilutedHCl was added into the samples to remove the calcium carbonateand the oxidised samples were rinsed with deionised water by decan-tation of the suspension several times. Permanent slides weremounted using Naphrax©. On average, 400 valves were counted oneach slide in random transects with a Leica DMRX optical microscopewith 100× oil immersion objective, equipped with a Leica DC 500digital camera. The identification was based on reference floras(e.g. Krammer and Lange-Bertalot, 1986, 1988, 1991a,b), as well ason recent bibliography including series of Diatoms of Europe,Iconographia Diatomologica and relevant scientific papers.

A list of taxawith relevant quantitative datawas compiled fromeachslide and the counts were converted into relative abundance and usedto calculate the Specific Polluosensitivity Index (IPS) using the softwareOMNIDIA version 5.3 (Lecointe et al., 1993, http://omnidia.free.fr/omnidia_english.htm). All taxa encountered were assigned to an eco-logical guild according to Passy (2007) and Berthon et al. (2011): lowprofile, high profile and motile guilds.

2.3. Data Treatment and Statistical Analysis

2.3.1. Environmental VariablesIn total, 27 explanatory variables were considered for statistical anal-

ysis, comprising data onwater chemistry, land use in the buffer zone andtheupper catchment, geographical and catchment information (Table 1).The water chemistry included physico-chemical data such as tempera-ture, pH, conductivity, oxygen, Cl−, K+,Mg2+, Ca2+, sulphates, nutrients(phosphates, nitrates, nitrites) and total organic carbon (Table 1). Thetemperature values were averaged for the periods of one, two andthree months prior to diatom sampling for purposes of further analysis.The land use parameters comprised the percentage of land use practicesoccurrsing in the area studied, e.g. pasture, urbanization, forestry andcrop land. The forestry category included deciduous native forest, conif-erous native and non-native forest andmixed native forest. The land useparameters were treated separately, as well as expressed in a total dis-turbance, summarizing the negative land use practices (percentage ofcrop land, urbanization, pasture) and in a global “Land Use Index” (LUI)according to Feld (2004). The land use was calculated separately forthe upper catchment (upstream of the sampling point) as well as forthe buffer zone defined as a 100 mwide buffer that extended 400 m up-stream and 100 m downstream from each sampling (modified fromPella et al., 2004) (see Table 1). The status of riparian vegetation wasfor purposes of statistical analysis represented by percentage of forestry

and total disturbance in the buffer zone, which are closely linked to thestatus of canopy cover on the river banks. All environmental variableswere standardized and log-transformed before statistical analysis. Ap-propriate tests for normality (Shapiro W-test, p b 0.05) were conductedusing STATISTICA 6.0 (StatSoft Inc., 2001).

First, a Principal Component Analysis (PCA) (Goodall, 1954) wasconducted onwater chemistry data and land use in order to distinguishdifferent chemical environments at the sites and explore the relation-ships between environmental variables among the sites studied usingSTATISTICA 6.0 (StatSoft Inc., 2001). Data sets were further treated inorder to explore the seasonal and temporal variability of environmentalvariables. Despite trying several transformations, nitrates and ammoni-um did not follow normal distributions and non-parametric statisticswere therefore applied to further environmental data analysis. Anon-parametric Kruskal-Wallis one-way analysis of variance by rankswas performed to evaluate which of all the explanatory variables(land-use in the buffer zone and upper catchment, water chemistry,geographical and catchment data) differed between the forestedunimpacted (Schwaartzbaach), forested impacted (Consdorferbaach)and open impacted (Hemeschbaach and Sauerbaach) streams. In addi-tion, the non-parametric Mann–Whitney U test was applied to investi-gatewhichwater chemistry parameters differed between the two yearsand consequently between the two sampling seasons. Both Kruskal–Wallis one-way ANOVA by ranks and Mann–Whitney U test wereperformed using STATISTICA 6.0 (StatSoft Inc., 2001). Afterwards, theenvironmental data were tested for correlations in order to identifythe relationships between and among the water chemistry andland-use from the buffer zone and the upper catchment. Spearman cor-relations were calculated using STATISTICA 6.0 (StatSoft Inc., 2001).Strong relationships (or correlations) were reported when the correla-tion coefficients were above 0.5 (p b 0.01). Based on the Spearman cor-relations, the Correlation Index (CI: Blanco et al., 2007) was calculatedin order to evaluate the significance and strength of correlations. CI cal-culation was based on the formula proposed by Blanco et al. (2007):CI = [∑ρs2 S)].n−2, where CI is the correlation index for a given vari-able; ρs is Spearman's correlation coefficient; S is the number of statis-tically significant correlations at p b 0.05 and n is the number ofvariables evaluated. CI ranges from 0 to 1, indicating the theoreticalminimum and maximum relationship between variables. CI was calcu-lated for the different land uses from the buffer zone and the uppercatchment in order to compare the relevance of the land uses in relationto the water chemistry.

2.3.2. Diatom CommunityOnly diatom taxa reaching a relative abundance of more than 2%

in at least one sample were included in the statistics (65 taxa intotal). Species data were arcsin square root transformed prior to anystatistical analysis to approximate normal distribution.

The effect of environmental conditions on diatoms was evaluatedon the level of the species composition, the diatom guilds and the IPSindex. The diversity of diatom communities and the relationships be-tween the diatom communities and environmental variables wereexplored by multivariate statistics. Diatom data was first analysedusing a Detrended Correspondence Analysis (DCA, Hill and Gauch,1980) to determine the length of the gradient in the species data.DCA with detrending by segments and downweighting of rare taxagave a gradient length of 3.5 SD, indicating a marginal gradient be-tween linear and unimodal. Since diatoms are reported to have ratherunimodal than linear response to environmental gradients (Potapovaet al., 2004), we further used a constrained ordination, Canonical Cor-respondence Analysis (CCA) to relate diatom assemblage structure toall predictor environmental variables and to explore the relationshipsamong and between species and the environment (ter Braak andVerdonschot, 1995). Manual selection and the Monte Carlo permuta-tion test (499 runs) were used to reduce the environmental variablesto those correlating significantly with the derived axes, at a cut-off

http://omnidia.free.fr/omnidia_english.htm

http://omnidia.free.fr/omnidia_english.htm


point of P = 0.05. Hill's scaling was selected with inter-sample dis-tances. All variables selected by forward selection followed normaldistribution; therefore we assumed that the tests of significancecould be reliably used to support the interpretations. With regard tothe environmental parameters involved in the analysis, in additionto water chemistry and land use practices, we also included generalsite descriptors such as distance from the source, elevation andstream order. Both Detrended Correspondence Analysis and Canoni-cal Correspondence Analysis were performed using CANOCO 4.5 (terBraak and Šmilauer, 2002).

The statistical differences in the species composition between thestreams and seasons (taking into account the results of the CanonicalCorrespondence Analysis) were further tested using Analysis of Similar-ity (ANOSIM; Bray-Curtis distancemeasure, 999 permutations). ANOSIMis a randomization-basedmethod of multivariate analysis and it is main-ly used to test of significant differences between two ormore apriory se-lected groups, based on any distance measure (Clarke, 1993). Thesamples from each stream were separated by season (8 groups of siteswere created) in order to test the spatial and seasonal differences inthe species composition. The groups that did not significantly differbased on the R statistics and Bonferroni corrected p values of theANOSIMweremerged for further analysis. Consequently, One-way Anal-ysis of Similarity Percentages (SIMPER, Bray–Curtis distance measure)based on species contribution was used to identify the diatom speciesthat primarily provided discrimination between groups created afterANOSIM. SIMPER is a simple method for assessing which taxa are pri-marily responsible for anobserved difference between groups of samplesbased on Bray–Curtis similarity measure (Clarke, 1993). Three groups ofsampleswere created a priory for SIMPER analysis based on the results ofANOSIM and CCA (forested unimpacted, spring impacted (forested andopen) and autumn impacted (forested and open)). The taxa presentedcontributed up to 50% of dissimilarity between the groups. BothANOSIM and SIMPER analysis were carried out using PRIMER_Version5.2.0 (Clarke and Gorley, 2001).

Associations between the diatom species composition data matrixand the environmental data matrix (all land uses and water chemis-try variables) were further examined using the BEST procedure inPRIMER Version 5.2.0 (Clarke and Gorley, 2001). This analysis iden-tifies variables and combinations of variables that best explain thepattern in the biological data by maximizing a rank correlation be-tween their respective resemblance matrices. Due to the large num-ber of environmental variables, the BVSTEP algorithm was applied,based on a stepwise search of the variables and employing both for-ward selection and backward elimination. It calculates the Spearmancorrelation coefficients between the rank similarity matrices of biota

Fig. 2. PCA plot showing the water chemistry gradients (a), land use gradients among the simpacted sites, solid line: forested unimpacted sites, uncircled: open impacted sites.

and environment (Clarke and Ainsworth, 1993). The BVSTEP proce-dure results in a list of variables, which have the highest correlationto the biotic pattern. Bray–Curtis similarity was employed for similar-ity matrices of species data. The analysis was employed to the wholedata set and then separately to the different seasons and to three dif-ferent groups of sites depending on the impact and type of ripariancover (forested unimpacted, forested impacted and open impacted).

2.3.3. Diatom Guilds and IPS IndexDiatom guilds distribution among different sites was evaluated by

plotting the species (grouped according to their assignment to the di-atom guild) in the CCA ordination space. Further, the mean relativeabundances of diatom guilds in different streams during differentsampling campaigns were graphically compared in order to evaluatetheir spatial and seasonal variability. Values of the IPS index at differ-ent streams and different years were compared using box-plots creat-ed in the Sigma Plot software ver. 11.0 (Systat Software Inc., 2008).Diatom guilds (represented by sum of relative abundances of speciesbelonging to the three guilds calculated for every site) and the IPSindex were further correlated by means of Spearman correlationswith the environmental matrix (land-use practices in the bufferzone and upper catchment and water chemistry) in order to comparetheir indicative potential. Based on the results of Spearman correla-tions, the Correlation Index (CI: Blanco et al., 2007, describedabove) was calculated again in order to compare the significanceand strength of correlations and to evaluate their indication power.

3. Results

3.1. Relationships Between Land Use and Water Chemistry

The PCA ordination carried out to explore the variation of environ-mental variables showed that themost significant gradients in the envi-ronmental data were represented by organic and inorganic pollution(Fig. 2a). The first PCA axis accounted for a 23% variance in the chemicaldata and corresponded to the gradient of nutrients; the second axisexplained 16% of variance and significantly correlated with sulphatesand TOC.When projected with land use parameters, the gradient of nu-trients corresponded to the percentage of land use disturbance in theupper catchment, mainly to urbanization and crop land, and was nega-tively correlated to forestry in the upper catchment (Fig. 2b). The siteswere distributed in the ordination space along this gradient, the forest-ed unimpacted sites occurring along the forestry factor, whilst those im-pacted, regardless of the presence or absence of riparian vegetation,were placed along the factors representing increasing nutrients and

ites studied (b) and distribution of sites along the gradients (c): dashed line: forested

image of Fig.�2


disturbance (Fig. 2c). The second axis was associated with a gradient ofsulphates and conductivity, negatively correlated with TOC, whichshowed to be related to the buffer zone land use (Fig. 2b).

With regard to the results of the Spearman correlations betweenland use parameters from the buffer zone and the upper catchment, sig-nificant positive relationships (rs > 0.5) were detected between thepasture in the catchment and pasture and total disturbance in the bufferzone and negatively correlated with buffer forestry (Table 2). The com-parison of correlation coefficients between land uses andwater chemis-try revealed that the buffer zone was in general less significantlycorrelated with water chemistry than the land use practices from theupper catchment, indicating that the influence of the buffer zone isless important in determining water chemistry (Table 3). CI of theland use practices from the upper catchment reached consistentlyhigher values than the buffer zone and the strongest correlationpower was obtained for urbanization (Table 4). This difference in thecorrelation power of the buffer zone and catchment land uses wasmost obvious in the case of phosphates and potassium, which did notcorrelate with land use in the buffer zone at all, whilst their correlationcoefficients with the land use in the upper catchment were significantfor all the land use types tested and were among the strongest in thedataset (Table 3). The upper catchment was clearly associated withmost of the parameters tested; the most significant correlations werecalculated for nutrients and potassium. With regard to the differentland uses, phosphates, nitrates, nitrites and ammonium tend toassociate most strongly with increasing urbanization and pasture inthe catchment. In general, catchment urbanization reached the highestcorrelation coefficients within the land use parameters tested, some-thing also reflected in the Correlation Index. The buffer zone land usetended to be more closely linked to inorganic pollution, namely to sul-phates, calcium and TOC; in particular, the TOC was associated with in-creasing pasture and sulphates, with increasing urbanization anddecreasing forestry. Furthermore, strong positive correlation wasdetected between conductivity and urbanization in the buffer zone.Also, the results proved the existence of a close relation of the bufferzone disturbance and stream temperature. The percentage of forestryand general disturbance that represent the status of canopy cover inthe buffer zone (linked to the presence of riparian vegetation) wasshown to correlate best with water temperature (one month average).Correlations of average temperature values of two and three monthswere weaker and less significant.

With regard to the variability of the environmental variables amongthe different types of sites grouped according to the riparian cover typeand impact (3 groups tested: forested impacted, forested unimpactedand open impacted), the Kruskal–Wallis one-way ANOVA by ranksproved that land use and water chemistry, both in the buffer zone andin the upper catchment, were significantly different (see in Table 1

Table 2Spearman correlation coefficients between land uses in the buffer zone (B) and the upper catch

Crop B Pasture B Urban B Forestry B Disturb B LUI B

Crop B 1 0.40* ns −0.43* 0.43* 0.52*Pasture B 0.40* 1 ns −0.54** 0.54** 0.51*Urban B ns ns 1 −0.58** 0.58** 0.60*Forestry B −0.43* −0.54** −0.58** 1 −1** −0.9Disturb B 0.43* 0.54** 0.58** −1** 1 0.96*LUI B 0.52** 0.51** 0.60** −0.96** 0.96** 1Forestry C ns −0.44* ns 0.30* −0.30* nsCrop C ns 0.35* ns ns ns nsPasture C ns 0.50** ns −0.55** 0.55** 0.47*Urban C ns ns ns ns ns nsDisturb C ns 0.49** ns −0.35* 0.35* 0.25*LUI C ns ns ns ns ns nsLow profile ns ns −0.28* 0.34* −0.34* −0.2High profile ns ns ns ns ns nsMotile ns ns 0.31* −0.43* 0.43* 0.34*IPS ns ns −0.36* 0.46** −0.46** −0.3

the variables that were significantly different with p b 0.05* andp b 0.001**). Among all the variables, only the distance to source, catch-ment area upstream of the site, temperature (t-1, t-2, t-3 averages), pHand oxygen saturation did not differ between groups, indicating that nat-ural variability between the different types of sites is very low. Elevation,urbanization in the buffer zone, PO4\P, nitrites and ammonium weresignificantly different between the groups with p b 0.05 and all theother variables were significantly different with p b 0.001 (Table 1).Temperatures were slightly higher at impacted forested and open loca-tions compared to shaded unimpacted sites, yet the differences weresmall for the whole dataset and not significant (see Table 1). Therewere, however, significant seasonal differences in temperature detectedusing the non-parametric Mann–Whitney U test. The comparison be-tween the two seasons of impacted sites (forested and open) showedthat they differed in the t-2, t-3 average temperatures and PO4-P withp b 0.001 and in the pH, oxygen saturation, phosphates and potassiumwith p b 0.05. The seasonality in the forested unimpacted sites wasreflected by values of temperature (t-2, t-3 months average), PO4-Pand potassium with p b 0.001 and phosphates and TOC with p b 0.05.The annual differences showed to be significant in case of temperature,oxygen, phosphates and nitrites (p b 0.001).

3.2. Diatom Assemblages Structure and Variability in Relation toEnvironmental Parameters

A total of 215 specific and infraspecific diatom taxa were identifiedin the 82 samples. Out of them only 65 taxa reached a relative abun-dance of more than 2% in at least one sample. The most abundant taxaand occurring at more than 50% of all sites included: Achnanthidiumlineare W. Smith, A. minutissimum (Kützing) Czarnecki, Amphorapediculus (Kützing) Grunow in Schmidt et al., Cocconeis euglyptaEhrenberg, Reimeria sinuata (W. Gregory) Kociolek & Stoermeremend. Sala, Guerrero & Ferrario, Navicula cryptotenella Lange-Bertalotin Krammer & Lange-Bertalot, N. lanceolata (C. Agardh) Kützing,N. tripunctata (O.F. Müller) Bory, Rhoicosphenia abbreviata (C. Agardh)Lange-Bertalot, Planothidium frequentissimum (Lange-Bertalot) Lange-Bertalot and P. lanceolatum (Brébisson) Lange-Bertalot.

The CCA revealed that sites’ distribution in the ordination space wasgreatly influenced by the land uses and season. The forested unimpactedsites were clearly separated from the impacted ones (both forested andopen) and samples were further grouped according to the season(Fig. 3). The ordination diagram showed that the diatom communities’structure at impacted sites, regardless of the presence or absence of ri-parian vegetation, was similar and much more influenced by seasonalchanges. This tendency was further confirmed by ANOSIM, used to eval-uate the differences in the community structure between the differentstreams and seasons (8 groups of sites). The most significant differences

ment (C) and diatoms (expressed as diatom guilds and IPS index). *p b 0.05, **p b 0.001.

Forestry C Crop C Pasture C Urban C Disturb C LUI C

* ns ns ns ns ns ns* −0.44* 0.35* 0.50** ns 0.49** ns* ns ns ns ns ns ns6** 0.30* ns −0.55** ns −0.35* ns* −0.30* ns 0.55** ns 0.35* ns

ns ns 0.47** ns 0.25* ns1 −0.89** −0.83** −0.74** −0.99** −0.91**−0.89** 1 0.66** 0.74** 0.86** 0.93**

* −0.83** 0.66** 1 0.56** 0.85** 0.69**−0.74** 0.74** 0.56** 1 0.72** 0.87**−0.99** 0.86** 0.85** 0.72** 1 0.88**−0.91** 0.93** 0.69** 0.87** 0.88** 1

5* 0.41* −0.36* −0.55** −0.40* −0.41* −0.36*ns ns ns −0.26* ns −0.26*−0.55** 0.48** 0.67** 0.54** 0.56** 0.49**

9* 0.56** −0.49** −0.70** −0.59** −0.56** −0.53**

Table 3Spearman correlation coefficients between the land use in the buffer zone (B) and the upper catchment (C), diatom communities (expressed as diatom guilds, IPS index) and waterchemistry. *p b 0.05, **p b 0.001.

t-1 m t-2 m t-3 m pH Cond O2 PO4-3 PO4-P Cl- NO2-N NO3-N SO4

-2 NH4-N K+ Mg2+ Ca2+ TOC

Crop B ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns nsPasture B ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns 0.39*Urban B 0.29* ns ns ns 0.56** ns ns ns ns ns ns 0.59** 0.27* ns ns 0.57** nsForestry B −0.46** −0.28* −0.25* ns −0.38* ns ns ns ns ns ns −0.37* −0.26* ns −0.35* −0.34* −0.32*Disturb B 0.46** 0.28* 0.25* ns 0.38* ns ns ns ns ns ns 0.37* 0.26* ns 0.35* 0.34* 0.32*LUI B 0.39* 0.27* ns ns 0.36* ns ns ns ns ns ns 0.37* ns ns 0.31* 0.34* 0.35*Forestry C −0.34* ns ns ns −0.33* ns −0.45* −0.25* ns −0.27* −0.38* ns ns −0.53** ns −0.25* nsCrop C 0.33* ns ns ns 0.35* ns 0.39* ns ns ns 0.40* ns ns 0.51** ns 0.26* nsPasture C 0.43* ns ns ns 0.26* ns 0.54** 0.31* ns 0.45** 0.32* ns 0.40* 0.56** ns ns 0.31*Urban C 0.26* ns ns ns 0.39* ns 0.63** 0.40* 0.31* 0.38* 0.46** 0.29* 0.35* 0.71** −0.47** 0.39* nsDisturb C 0.35* ns ns ns 0.29* ns 0.43* ns ns 0.27* 0.37* ns ns 0.52** ns ns nsLUI C 0.28* ns ns ns 0.32* ns 0.55** 0.32* 0.29* 0.30* 0.49** ns ns 0.64** −0.33* 0.28* nsLow profile −0.45* ns ns ns ns ns −0.29* ns ns ns ns −0.26* −0.40* ns ns ns nsHigh profile ns ns −0.27* ns ns 0.38* −0.38* ns ns −0.40* ns ns ns −0.34* ns ns nsMotile 0.50** ns ns ns 0.26* ns 0.41* ns ns 0.34* ns 0.30* 0.48** 0.32* ns ns nsIPS −0.47** ns ns ns −0.30* ns −0.55** −0.30* ns −0.45** ns −0.27* −0.52** −0.47** ns −0.30* ns


were identified between the forested unimpacted sites and impactedsites (both forested and open) (p b 0.001) proving that the communitystructure of impacted sites with an intact riparian zone is similar toopen impacted sites and significantly different from forested unimpactedsites. We further confirmed significant seasonal differences betweenspring and autumn samples from impacted sites (p b 0.001), whilst theforested sites (both unimpacted and impacted) did not differ significant-ly between the two seasons (Table 5). Based on these results of diatomcommunity structure, three groups of samples were created: forestedunimpacted, impacted from spring (open and forested) and impactedfrom autumn (open and forested) for purposes of SIMPER analysis. Thediatom species that mostly accounted for the dissimilarity between for-ested unimpacted and impacted sites included oligotraphentic speciessuch as A. lineare and Gomphonema elegantissimum E. Reichardt &Lange-Bertalot, and pollution tolerant taxa such as Mayamaea permitis(Hustedt) Bruder & Medlin and Navicula gregaria Donkin (the last twooccurring only at impacted sites). Other taxa identified by SIMPERanalysis differed mainly in their abundance (Table 6).

With regard to the relationships between the diatom assemblagesand the explanatory variables, the CCA showed that land use in theupper catchment (crop land, pasture and urbanization) together withphosphates, ammonium, potassium, TOC, pH, temperature, magnesiumand slope were the best predictors of diatom community composition

Table 4Correlation Index values showing the correlation powerof selected variables. CI for land use practices was calcu-lated from Spearman correlation coefficients with waterchemistry; for diatom guilds and IPS the land use prac-tices in the buffer and catchment zone as well as thewater chemistry were taken into account.

Land use CI

Forestry B 0.036962Crop B 0.006858Pasture B 0.004391Urban B 0.022022Disturb B 0.036964LUI* B 0.024943Forestry C 0.038541Crop C 0.039322Pasture C 0.071014Urban C 0.097907Disturb C 0.028082LUI C 0.065380

Diatoms CILow profile 0.047539High profile 0.017061Motile 0.097731IPS 0.122214

(Fig. 3). The first axis of the CCA biplot represented the gradient ofhumandisturbance and clearly determined the separation of unimpactedsites from sites with intensive land use in the upper catchment. Further-more, it separated the two sampling seasons, whilst the separation wasmore pronounced at impacted sites. The spring impacted sites were dis-tributed in the ordination space along the gradient of ammonium andTOC, whilst the autumn sites were related to potassium and phosphates.Besides, the differentiation of impacted sites could also be related to tem-perature. The spring impacted sites were distributed along the values ofaverage temperatures of one month prior to diatom sampling; the au-tumn sites reflected the values over three months indicating that thehigher summer temperatures might have influenced the species compo-sition at impacted sites also in autumn (Fig. 3). The first two axes of theCCA accounted for only 19% of explained variance and therefore partialCCAs were carried out for the separate seasons, which significantly in-creased the explanatory power of the analysis (27% for spring samplesand 23% for autumn samples). The CCA on spring samples led to a clearseparation of both impacted forested and open sites from unimpactedones (Fig. 4b). The unimpacted sites were distributed along the forestrygradient contrary to impacted sites that grouped together along distur-bance indicators such as urbanization, crop land and pasture. The mostimportant environmental variables determining the distribution of bothforested and open impacted sites in spring were nitrates, phosphates,

Fig. 3. CCA site biplot showing the most significant environmental gradients that deter-mine the species composition at sites. Square: spring samples; circle: autumn samples.Solid line: forested unimpacted sites; dashed line: impacted forested and open sites.

image of Fig.�3

Table 5R statistics of analysis of similarities (ANOSIM) with p b 0.001 between diatom assemblages of the four streams separated by season. F/unimp (forested, unimpacted):Schwaartzbaach (11 spring samples, 11 autumn samples), F/imp (forested impacted): Consdorferbaach (10 spring samples, 10 autumn samples), O1/imp (open impacted):Hemeschbaach (10 spring samples, 10 autumn samples), O2/imp: (open impacted): Sauerbaach (10 spring samples, 10 autumn samples).

spring autumn

F/unimp F/imp O1/imp O2/imp F/unimp F/imp O1/imp O2/imp

spring F/unimp 0.713 0.697 0.389 ns 0.742 0.858 0.814F/imp ns ns 0.569 ns ns nsO/imp ns 0.675 0.409 0.618 0.614O/imp 0.359 0.333 0.494 0.364

autumn F/unimp 0.489 0.643 0.562F/imp ns nsO1/imp nsO2/imp


TOC and source distance. The selected variables explained 18% of the totalvariance in the species data. In autumn, only landuse parameters anddis-tance to source, together with calcium seemed to influence the variationin diatom assemblages, whilst other water chemistry variables were notrelevant (Fig. 5b). Impacted sites weremostly driven by land use param-eters; the forested sites followed the gradient of urbanization and streamorder.

Similarly to CCA, the BVSTEP conducted on all samples showed thatthe percentage of pasture (in the catchment and buffer zone) and ur-banization in the upper catchment, temperature (2 and 3-months aver-ages), potassium and TOC (0.530, rho > 0.95; delta rho b 0.001) werebest explaining the variation in the diatom community structure.Contrary to CCA, the nutrients were not significant. The analysisconducted for the separate seasons resulted in similar composition ofbest-explaining variables in autumn; although the diatom communitydata variation was driven by the same set of variables, temperaturewas irrelevant (0.543; rho > 0.95; delta rho b 0.001). Contrary to that,nutrients became relevant in explaining the species composition inspring, where pasture and urbanization in the upper catchment togeth-er with phosphates (0.547; rho > 0.95; delta rho b 0.001) were identi-fied as best explaining variables. Diatom community structure atforested unimpacted sites was best explained by forestry and pasturein the upper catchment, temperature (1 and 3-months averages), pH,conductivity and Ca (0.417; rho > 0.95; delta rho b 0.001). This con-trasts with the impacted sites (both forested and open), where diatomassemblages varied mostly due to disturbance (urbanization in theupper catchment and pasture in the buffer zone) and pollutants (phos-phates, ammonium, potassiumand TOC) aswell as temperature (0.569;rho > 0.95; delta rho b 0.001).

3.3. Diatom Guilds and IPS Index

The species were further plotted according to their assignment tothe different diatom ecological guilds on the CCA diagram in order to

Table 6Results of SIMPER analysis showing the mean relative abundance of taxa (and their corregroups of sites. Taxa presented contributed cumulatively up to 50% of the dissimilarity. Dopen) = 79.67%, forested unimpacted vs. impacted autumn sites (forested and open) = 78and open) = 74.41%. F: forested, O: open, L: low profile guild, H: high profile guild, M: mo

Taxon Code(Guild) Me

For(N

Achnanthidium lineare W. Smith ACLI(L) 0.Gomphonema elegantissimum E. Reichardt & Lange-Bertalot GELG(L) 0.Amphora pediculus (Kützing) Grunow APED(L) 0.Achnanthidium minutissimum (Kützing) Czarnecki ADMI(L) 0.Rhoicosphenia abbreviata (C. Agardh) Lange-Bertalot RABB(H) 0.Mayamaea permitis (Hustedt) Bruder & Medlin MPMI(M) 0Navicula lanceolata (C. Agardh) Kützing NLAN(M) 0.Navicula gregaria Donkin NGRE(M) 0Average similarity within the group 47.

visualize the distribution of diatom guilds among the sites in relationto the environmental data (Figs. 4a and 5a). The analysis showed thatforested headwaters with limited resources, such as light and nutri-ents, were typically dominated by low profile diatoms, whilst the as-semblages at impacted sites, both forested and open, showed a widerrange of growth forms represented by motile and low profile guilds,indicating that resources availability at these sites is higher (Figs. 4aand 5a). The proportion of diatom guilds was further compared be-tween different streams and sampling seasons and the results con-firmed the CCA outcomes. The forested unimpacted sites weredominated by low profile diatoms with no seasonal differences inthe guilds proportion (Fig. 6). At impacted sites, the low profile wasoften dominant, but the proportion of the motile guild was muchmore noticeable or even prevailed. Furthermore, the guilds structureat impacted sites, regardless of the presence or absence of ripariancover, displayed a distinct seasonal variability. The spring samplesshowed to have higher proportion of the motile and high profileguild at both forested and open impacted sites (Fig. 6).

The diatom guilds were significantly correlated with temperature,conductivity, oxygen, phosphates, nitrites and ammonium, althoughthe correlation power of low profile andmotile guilds wasmuch higherthan that of the high profile guild (Tables 3 and 4). Furthermore, bothlow profile and motile guilds, as well as the IPS, were significantly cor-related to all types of land uses tested; although theyweremore strong-ly associated with the upper catchment than with the buffer zone(Table 2). The highest correlation among the data set was detected be-tween IPS and pasture in the upper catchment and strong negativecorrelationswere also reached between IPS and phosphates and ammo-nium (Tables 2 and 3). On the other hand, strong positive correlationswere detected between the IPS and forestry, both in the buffer zoneand upper catchment. Further, IPS significantly correlated with, nitrites,temperature (t-1 average), disturbance in the buffer zone and uppercatchment and urbanization in the upper catchment. The IPS indexhas further clearly differentiated between unimpacted and impacted

sponding guild assignment) mainly responsible for the differences between the threeissimilarity between groups: forested unimpacted vs. impacted spring (forested and.32% and impacted spring sites (forested and open) vs. impacted autumn sites (forestedtile guild.

an relative abundance

ested unimpacted= 22)

Impacted (F + O) spring(N = 30)

Impacted (F + O) autumn(N = 30)

34 0.02 0.0517 0.02 0.0407 0.12 0.3613 0.22 0.1002 0.02 0.03

0.10 0.0101 0.06 0.01

0.06 0.0249% 36.94% 40.65%

Fig. 4. CCA species (a) and site (b) biplots of spring samples showing the most important predictors of diatom community structure and the distribution of the most important speciesamong the diatom guilds (a): asterisk: low profile guild; empty circle: motile guild; rectangle: high profile guild; and distribution of samples (b): circle: unimpacted forested, square:impacted forested, triangle: impacted open.


sites, but showed similar values for impacted sites regardless of thepresence or absence of riparian vegetation (Fig. 7a, b). Furthermore,this result was constant for the two years of the study (Fig. 7a, b),discarding a possible annual variability.

4. Discussion

4.1. Relationships Between Land Uses and Water Chemistry

In our study, the riparian buffer zone was represented by the per-centage of woodland in a 100 m wide buffer zone that extended400 m upstream and 100 m downstream from each sampling site.The land use characteristics in this area of thewatershedwere generallyfound to have a less significant influence on water chemistry than thedeterminants from the upper catchment. However, among the positiverelationships found between the buffer zone and water chemistry, weproved that the percentage of forests in the buffer zone was closely re-lated to stream temperature. These findings confirm the key role of ri-parian vegetation in determining the temperature regime in streams,which is in line with many other studies (Hetrick et al., 1998; Mooreet al., 2005; Studinski et al., 2012). Furthermore, we also confirmedthat the intensive pasture and deforestation in the buffer zone are

Fig. 5. CCA species (a) and site (b) biplots of autumn samples showing the most important preamong the diatom guilds (a): asterisk: low profile guild; empty circle: motile guild; rectangle:impacted forested, triangle: impacted open.

related with the increase of TOC levels in streams (Franzluebbers andStuedemann, 2002). On the other hand, we failed to prove any associa-tion of riparian cover with nutrients (nitrates, nitrites, phosphates),which contradicts some previous studies that identified riparian zonesas key regulators of nitrogen inputs (Sliva and Williams, 2001; Sabateret al., 2003; Dodds and Oakes, 2006) or phosphorus dynamics (Slivaand Williams, 2001). Mayer et al. (2007) suggested that besides thesoil type, subsurface hydrology and subsurface biogeochemistry, themost important factor affecting themanaging of nitrogen inwatershedsis the buffer width. According to their review (Mayer et al., 2007), widebuffers (>50 m) more consistently remove significant portions of ni-trogen entering a riparian zone than narrow ones. In this context, the100 mwide buffer zonewe assessed should be effective enough. There-fore the lack of any relation between the buffer zone and nutrients ismost likely caused by other spatial and/or temporal limitations. Thespatial limitations involve variations in the point and diffuse pollutiontypical for urban land use, which is most closely linked to nutrient en-richment among the different land uses of the impacted sites studiedand thus we might have failed to detect some pollution pathways. Thetemporal limitations lie in the water quality data that is subject to fluc-tuations typical for headwaters. This could not be taken into accountsince it was gathered only during the diatom sampling in the absence

dictors of diatom community structure and the distribution of the most important specieshigh profile guild; and distribution of samples (b): circle: unimpacted forested, rectangle:

image of Fig.�4

image of Fig.�5

Fig. 6. Proportion of the three diatom guilds among the different streams during the different seasons and years. Mean relative abundances are compared. Light grey bar: low profileguild, black bar: high profile guild, dashed grey bar: motile guild.


of discharge data. Moreover, the nutrient dynamics were not studied indetail. Furthermore, Meynendonckx et al. (2006) stated that an under-standing of diffuse source contributions to instream nutrient concentra-tions cannot be based solely on an examination of land use variables, andsuch an approachmay bemisleading for the significance of human activ-ities, since instream nutrient concentrations are largely regulated by soildrainage. In spite of the fact that these relationships were not taken intoconsideration, we succeeded in proving that the nutrient concentrationswere closely related to catchment land use. In particular, the concentra-tions of phosphates, nitrates and potassium closelymirrored the intensi-ty of urbanization in the catchment, whilst buffer urbanization seemedto be more closely related to conductivity, sulphates and calcium.These results suggest that urban land use is themost important predictorof water quality variability (see also Osborne andWiley, 1988; Sliva andWilliams, 2001).

We further confirmed that catchment land usewas, in general, moreclosely related to water chemistry and prevailed over the influence ofthe buffer zone practices (similarly to Sliva and Williams, 2001 orMeynendonckx et al., 2006), despite the fact that the total disturbancein the buffer zone among the sites studied was greater (see Table 1).Nevertheless, distinguishing the impacts of land uses in the riparianzone from those in the entire catchment can be difficult because theymight be highly correlated (Dodds and Oakes, 2006). Similarly,according to Hunsaker and Levine (1995), in manywatersheds impact-ed by anthropogenic land use modifications, riparian zone characteris-tics may simply reflect dominant catchment land cover types; thustheir influencemust overlap.With regard to our results, the only signif-icant relationship between the buffer and catchment land use was con-firmed for pasture and forestry, whilst catchment urbanization, whichwas most closely related to water chemistry, was not associated with

Fig. 7. Comparison of IPS values between the studied streams in 2010 (a) and 2011 (b). FSIS1: Hemeschbaach (open impacted); IS2: Sauerbaach (open impacted). The gray bars in egray bar shows the median, the whiskers mark 1.5× interquartile range, circles show outli

any other anthropogenic disturbance from the buffer zone. This findingproves that catchment urbanization alone seemed to be the most im-portant determinant of water quality, especially in the sense of nutri-ents, regardless of the buffer zone characteristics. However, thetemperature, calcium, sulphates, conductivity and organic pollutionwere linked to both, catchment and buffer zone practices. Therefore, de-spite the dominant influence of the catchment land use, using bothcatchment and buffer landscape variables in headwaters are advisablein predicting water quality.

4.2. Diatom Assemblages' Structure, Variability and Indicative Potential

On the basis of the diatom community structure, the species essen-tially responsible for the differentiation between forested unimpactedand impacted were A. lineare and G. elegantissimum. Achnanthidiumlineare has only recently been studied in detail and it seems to inhabitoligotrophic streams and small rivers requiring low nutrient content(Van de Vijver et al., 2011). As a dominant species in unimpacted head-waters in our study, it exhibited a high tolerance to light limitation, lownutrient content and dischargefluctuations typical for headwaters, sim-ilarly to the group of A. minutissimum (Berthon et al., 2011). These con-ditions disfavour erect or large species, favouring instead small taxa thatmay be able to resist the extreme flow events and recolonize rapidly(Berthon et al., 2011), which is typical for the low profile guild thatdominated at all unimpacted forested sites. However, this species wasalso found in very low abundance at impacted open sites, contributingto the relatively high proportion of the low profile guild at these sites.This might simply be because the species was washed down from theupper reaches or spring sources not affected by human activities andthus also occurred in the inventories of polluted sites. Dead frustules

1: Schwaartzbaach (forested unimpacted); FS2: Consdorferbaach (forested impacted);ach box plot show the 25th and 75th percentiles of the data, the black bar inside theers between 1.5× and 3.0× interquartile range.

image of Fig.�6

image of Fig.�7


might occur in considerably high abundances in diatom assemblages insmall streams (Gillett et al., 2009) and can therefore significantly influ-ence the species richness and the bioassessment results. Nevertheless,in our study, the diatom samples contained less than 10% of deadfrustules, which we considered as negligible. Achnanthidium linearedominated at unimpacted sites in both spring and autumn sampleswith G. elegantissimum as subdominant species. This is also reportedto occur in oligotrophic streams and rivers (Ivanov et al., 2007;Morales et al., 2007; Angeli et al., 2010; Spitale et al., 2012).

The variability of diatom species distribution in the headwaters stud-ied was driven by twomain determinants, season and land use. The sea-sonal changes were reflected in the species composition and detected byCCA and also influenced the diatomguilds distribution. Nevertheless, theseasonal effectwasmore pronounced at impacted sites that were clearlyseparated based on different seasons, unlike at the unimpacted forestedsites, whichweremuch less affected by seasonality. High seasonal varia-tion in diatom assemblages in streams influenced by urbanization wasalso reported by Walker and Pan (2006). In algae, such seasonal differ-ences in species composition are often linked to light availability(Greenwood and Rosemond, 2005; Tornés and Sabater, 2010) and dis-charge (Rosemond, 1994; Biggs and Smith, 2002) in headwaters, whichare typically spring-fed. Seasonal differences in the flow regime can dis-turb the algal community physically and affect the species compositionthrough ion and nutrient concentrations. The seasonality at unimpactedsites should be most likely driven by seasonal differences in light condi-tions and/or by the leaf cover. However, our results showed that the sea-sonal variation at these sites was rather weak. On the other hand, themore distinct seasonality in species distribution at impacted sites, bothforested and open, could not be linked to light conditions since they sig-nificantly differ in the riparian cover, despite displaying similar seasonalpatterns. Therefore, the strong seasonal effect on diatom communities inboth types of impacted sitesmust have resulted from changes in nutrientconcentrations, since phosphates and potassium concentrations weresimilar across the impacted sites (both forested and open) and displayedthe same seasonal variation according to the Mann–Whitney test. Thisvariation that was not detected at unimpacted sites due their low nutri-ent content is most likely linked to seasonal changes in the discharge re-gime. The clear seasonal differences in the diatom guilds compositionmust have reflected seasonal differences in the discharge, which resultedin higher nutrient concentrations in spring, favouring motile and nutri-ent tolerant taxa at impacted sites. Compared to literature data, similarly,during a study of algae in relation to stream conditions in spring-fed ol-igotrophic streams in Ontario, Sherwood et al. (2000) detected only fewseasonal differences in algal communities, while the vastmajority of spe-cies did not display any seasonal variability that could be related tomea-sured stream conditions. In oligotrophic forested headwaters, it has beenreported that the most important controls of diatom assemblages arelight availability, reported as a primary limiting factor by Tornés andSabater (2010), grazing (Power et al., 1988; Wellnitz et al., 1996) andflow (Poff et al., 1990; Danehy and Bilby, 2009), whereas nutrientsmay be more relevant at sites where the canopy cover is low (Hill andKnight, 1988; Hill and Harvey, 1990). According to our results, wewould hypothesize that the influence of nutrients (or relevant pollut-ants) on diatom assemblages in headwaters prevails over the naturallimiting factors, in conditions where the nutrient supply is elevateddue to anthropogenic activities in the watershed regardless of the char-acter of the buffer zone. The increased levels of nutrients then shapethe diatom community favouring nutrient tolerant taxa, as shown inour study, regardless of the light availability (or lack of it) and status ofriparian vegetation. According to the results of the multivariate analysis,the diatom species composition clearly responded to nutrient content atimpacted sites, both forested and open, andwe did not detect any differ-ences in relation to the different riparian cover. On the contrary,Greenwood and Rosemond (2005) found that chronic nutrient enrich-ment at moderate concentrations had only little detectable effect onbenthic algal composition in two headwater streams with intact tree

canopies and the seasonal effects greatly prevailed over the nutrients en-richment. However, compared to our headwaters, their streams wereslightly acidic and the diatom assemblages were dominated by acido-philic and aerophilic taxa, so the community was apparently driven bylower pH and dessication rather than nutrients.

Nevertheless, the stream temperature that is directly linked to thepresence or absence of riparian vegetation also influenced the diatomspecies composition at impacted sites. Temperature proved to be signif-icant in shaping the diatom community according to forward selectionand Monte-Carlo testing in the CCA analysis and, furthermore, all dia-tom indicators significantly correlatedwith average temperature valuesof one month prior to sampling. On the ordination diagram, spring im-pacted samples were grouped along the one month temperature aver-age and the autumn samples along the three month temperatureaverage, which would suggest that diatom assemblages responded tohigher temperatures rather than to lower.Moreover, temperature influ-ence was also confirmed by the BVSTEP procedure. Therefore, we can-not entirely exclude that temperature also contributed to significantseasonality in the diatom community structure at impacted sites. How-ever, it is a paradox to suggest that temperature contributes similarly tothe seasonality of both forested and open sites, since they are exposedto completely different light conditions. Nevertheless, the measuredstream temperature during the tested months did not differ betweenthe forested and open impacted sites. This puts the influence of lightconditions in question and confirms the primary role of pollutants,which caused the overlap of the impacted forested and open sites, re-gardless of the presence of riparian buffer. This also corresponds tothe results of Danehy et al. (2007), who compared periphyton structureand biomass in headwaters in Oregon (USA) with different density ofDouglas–Fir in the buffer zone. The authors only detected differencesin the biomass and diatom species richness between the clear-cut andmature sites, but other assemblage metrics as well as morphologicalguilds with dominance of prostrate and erect species were found to besimilar regardless of the presence or absence of riparian vegetation.

We further confirmed that diatom assemblages in the studied head-waters were very closely linked to the degree of land use in the uppercatchment rather than the buffer zone. The correlation coefficients be-tween the diatom guilds and IPS and land use practices in the uppercatchment were the highest among the whole set tested, and the IPS aswell as the motile guild were strongly correlated with all land use prac-tices from the upper catchment. These relationships were stronger thanbetween diatoms andwater chemistry, indicating that diatoms integratethe multiple effects of all environmental stressors linked to negative im-pacts of land uses on stream biota and do not only respond to alteredwater chemistry. Further, asmentioned above, thewater chemistrymea-surements are temporally limited, since the data in our study was onlygathered along with diatom sampling. The close relationship of diatomcommunities to land use practices in the streams studied is probably am-plified by the specific physical attributes (such as narrow wetted areaand distinct discharge fluctuations) and small size of headwaters,which causes the water channel and biological communities to bestrongly influenced by their surrounding catchment (see Vannote et al.,1980). The most significant measurable impacts of the land uses on dia-toms were high nutrient concentrations, especially phosphates and po-tassium. The diatom communities at impacted open and forested sitesdisplayed similar patterns and responded to similar variables. Amongthe land uses, diatoms were most significantly related to urbanizationin the catchment,most likely due to its close associationwith phosphatesand potassium.We thus confirmed the outcomes of other studies show-ing that diatom assemblages respond sensitively to urban impacts onstream conditions (Newall and Walsh, 2005; Walker and Pan, 2006).

All indicators used, diatom guilds and IPS, sensitively reflected thenutrient enrichment and general degradation caused by human activ-ities in the upper catchment, in spite of the intact buffer zone. Amongdiatom guilds, the motile guild was the best indicator. This also corre-sponds to the findings of Stenger-Kovács et al. (2013), who tested the


applicability of diatom guilds in streams in Hungary. In our study, themotile guild reached a similar correlation power to IPS and appearedto be very sensitive to nutrient enrichment, as well as to temperaturevariation. It also reflected the variation in the different catchmentland use practices, especially urbanization. These results are in linewith previous tests carried out in rivers and streams (Berthon et al.,2011; Stenger-Kovács et al., 2013) and demonstrated that diatomguilds can be a very useful and effective tool in bioassessment studies,since they can provide a very precise image of the status of water en-vironment without deep and detailed taxonomical knowledge ofdiatoms.

5. Conclusions

This study indicates the high potential of diatoms as biomonitors ofenvironmental degradation in headwaters induced by land use prac-tices, especially related to nutrient enrichment. Our data showed thatchanges in diatom species composition at impacted sites were similarregardless of the presence or absence of riparian vegetation and showeddistinct seasonal variability. Diatom assemblages sensitively reflectedchanges in the water environment caused by land use practices in theupper catchment and the intact buffer zonedid notmitigate the impactsof anthropogenic activities. The diatomguildswere found as potentiallyreliable proxies for diatom-based assessment, although the IPS indextogether with the community structure reflected the pollution effectin the studied headwaters more precisely. However, the temporalchanges of diatom guilds distribution across the sites studied were notexamined in detail. This study was carried out on a small scale andthere were also several determinants of diatom species compositionin headwaters that were not surveyed, such as grazing and particularlydischarge. This remains open to additional research.

Conflict of interest statement

I certify that there is no conflict of interest with any financialorganization regarding the material discussed in the manuscript.

Acknowledgements

We gratefully acknowledge Delphine Collard for the treatment andpreparation of diatom samples and water chemistry analysis, LionelL'Hoste for his valuable support and participation in the fieldwork,François Barnich for his help with water chemistry analysis, ArnaudCours and Aina Martínez Useros for their assistance with the fieldworkand Lindsey Stokes and Dr. Angélica Oliveira (Universidade Federal deSanta Maria, Brasil) for language editing. Last but not least we wouldlike to thank the three anonymous reviewers for their valuable com-ments and suggestions to improve the manuscript.

This study was financially supported by the National Research Fundof Luxembourg (CO9/SR/13 — AQUACOM project).

References

Allan JD. Landscape and riverscapes: the influence of land use on river ecosystems.Annu Rev Ecol Evol Syst 2004;35:257–84.

Angeli N, Cantonati M, Spitale D, Lange-Bertalot H. A comparison between diatom as-semblages in two groups of carbonate, low–altitude springs with different levelsof anthropogenic disturbances. Fottea 2010;10:115–28.

Berthon V, Bouchez A, Rimet F. Using diatom life-forms and ecological guilds to assessorganic pollution and trophic level in rivers: a case study of rivers in south-easternFrance. Hydrobiologia 2011;673:259–71.

Biggs BJF, Smith RA. Taxonomic richness of stream benthic algae: effects of flood distur-bance and nutrients. Limnol Oceanogr 2002;47:1175–86.

Binckley CA, Wipfli MS, Medhurst RB, Polivka K, Hessburg P, Salter RB, et al. Ecoregionand land-use influence invertebrate and detritus transport from headwaterstreams. Freshw Biol 2010;55:1205–18.

Blanco S, Bécares E, Cauchie HM, Hoffmann L, Ector L. Comparison of biotic indices forwater quality diagnosis in the Duero Basin (Spain). Arch Hydrobiol Suppl 2007;161:267–86.

Blinn DW, Bailey PCE. Land-use influence on stream water quality and diatom commu-nities in Victoria, Australia: a response to secondary salinization. Hydrobiologia2001;466:231–44.

Cantonati M. Diatom communities of springs in the Southern Alps. Diatom Res 1998;13:201–20.

Cemagref. Etude des méthodes biologiques d’appréciation quantitative de la qualitédes eaux. Lyon: Ministère de l’Agriculture, Cemagref, Division Qualité des Eaux,Pêche et Pisciculture; 1982 [218 pp.].

Clarke KR. Non-parametric multivariate analysis of changes in community structure.Aust J Ecol 1993;18:117–43.

Clarke KR, Ainsworth MA. Method of linking multivariate community structure to en-vironmental variables. Mar Ecol Prog Ser 1993;92:205–19.

Clarke KR, Gorley RN. PRIMER. 755 Plymouth, UK: PRIMER-E Ltd.; 2001.Coring E, Schneider S, Hamm A, Hofmann G. Durchgehendes Trophiesystem auf der

Grundlage der Trophieindikation mit Kieselalgen. Koblenz: Deutscher Verbandfür Wasserwirtschaft und Kulturbau e. V; 1999 [219 pp.].

Cummins KW, Spengler GL. Stream ecosystems. Water Spectr 1978;10:1–9.Danehy RJ, Bilby RE. Periphyton and macroinvertebrate assemblage responses to flow

regime in spring-fed headwaters. Verh Internat Verein Limnol 2009;30:1210–4.Danehy RJ, Chan SS, Lester GT, Langshaw RB, Turner TR. Periphyton and macroinvertebrate

assemblage structure in headwaters bordered by mature, thinned, and clearcutdouglas-fir stands. For Sci 2007;53:294–307.

Dodds WK, Oakes RM. Controls on nutrients across a prairie stream watershed: landuse and riparian cover effects. Environ Manage 2006;37:634–46.

Dunne T, Leopold LB. Water in environmental planning. San Francisco: W.H. FreemanCo.; 1978 [818 pp.].

European Committee for Standardization. The European Standard. Water quality. Guid-ance standard for the routine sampling and pretreatment of benthic diatoms fromrivers. CEN/TC 230. EN 13946. Comité Européen de Normalisation; 2003.

European Committee for Standardization. The European Standard. Water quality. Guid-ance standard for the identification, enumeration and interpretation of benthic di-atom samples from running waters. CEN/TC 230. EN 14407. Comité Européen deNormalisation; 2004.

Feld CK. Identification and measure of hydromorphological degradation in Central Eu-ropean lowland streams. Hydrobiologia 2004;516:69–90.

Ferréol M, Dohet A, Cauchie HM, Hoffmann L. A top down approach for the developmentof a stream typology based on abiotic parameters. Hydrobiologia 2005;551:193–208.

FranzluebbersAJ, Stuedemann JA. Particulate andnon-particulate fractions of soil organic car-bon under pastures in the Southern Piedmont USA. Environ Pollut 2002;116:S53–62.

Gillett N, Pan Y, Parker C. Should only live diatoms be used in the bioassessment ofsmall mountain streams? Hydrobiologia 2009;620:135–47.

Goodall DW. Objective methods for the classification of vegetation. III. An essay in theuse of factor analysis. Aust J Bot 1954;2:304–24.

Gottschalk S, Kahlert M. Shifts in taxonomical and guild composition of littoral diatomassemblages along environmental gradients. Hydrobiologia 2012;694:41–56.

Greenwood JL, Rosemond AD. Periphyton response to long-term nutrient enrichmentin a shaded headwater stream. Can J Fish Aquat Sci 2005;62:2033–45.

Hering D, Johnson RK, Kramm S, Schmutz S, Szoszkiewicz K, Verdonschot PFM. Assess-ment of European streams with diatoms, macrophytes, macroinvertebrates andfish: a comparative metric-based analysis of organism response to stress. FreshwBiol 2006;51:1757–85.

Hetrick NJ, Brusven MA, Meehan WR, Bjornn TC. Changes in solar input, water temper-ature, periphyton accumulation, and allochthonous input and storage after canopyremoval along two small salmon streams in southeast Alaska. Trans Am Fish Soc1998;127:859–75.

Hill MO, Gauch HEJ. Detrended correspondence analysis: an improved ordination tech-nique. Vegetation 1980;42:47–58.

Hill WR, Harvey BC. Periphyton responses to higher trophic levels and light in a shadedstream. Can J Fish Aquat Sci 1990;47:2307–14.

Hill WR, Knight AW. Nutrient and light limitation of algae in two northern Californiastreams. J Phycol 1988;24:125–32.

Hunsaker CT, Levine DA. Hierarchical approaches to the study of water quality in rivers.Bioscience 1995;45:193–203.

Inwood SE, Tank JL, Bernot MJ. Patterns of denitrification associated with land use in 9Midwestern headwater streams. J N Am Benthol Soc 2005;24:227–45.

Ivanov P, Kourteva E, Mancheva A. Diatom taxonomic composition of streams andsmall rivers in the Strouma basin, SW Bulgaria. Phytol Balcan 2007;13:293–305.

Johnson LB, Richards C, Host GE, Arthur JW. Landscape influences on water chemistryin Midwestern stream ecosystems. Freshw Biol 1997;37:193–208.

Keller EA, Swanson FJ. Effects of large organic material on channel form and fluvial pro-cesses. Earth Surf Process 1979;4:361–80.

Kelly MG, Whitton BA. The Trophic Diatom Index: a new index for monitoring eutro-phication in rivers. J Appl Phycol 1995;7:433–44.

Kelly M, Bennett C, Coste M, Delgado C, Delmas F, Denys L, et al. A comparison of na-tional approaches to setting ecological status boundaries in phytobenthos assess-ment for the European Water Framework Directive: results of an intercalibrationexercise. Hydrobiologia 2009;621:169–82.

Knight AW, Bottorff RL. The importance of riparian vegetation to stream ecosystems. In:Warner RE, Hendrix KM, editors. California Riparian Systems. Ecology, Conservation,and Productive Management. Berkeley: University of California Press; 1984. p. 160–7.

Krammer K, Lange-Bertalot H. Bacillariophyceae 1. Teil: Naviculaceae. In: Ettl H, GerloffJ, Heynig H, Mollenhauer D, editors. Süsswasserflora von Mitteleuropa, 2/1.Stuttgart: Gustav Fischer Verlag; 1986. p. 1–876.

Krammer K, Lange-Bertalot H. Bacillariophyceae 2. Teil: Bacillariaceae, Epithemiaceae,Surirellaceae. In: Ettl H, Gerloff J, Heynig H, Mollenhauer D, editors. Süsswasserfloravon Mitteleuropa, 2/2. Stuttgart: Gustav Fischer Verlag; 1988. p. 1–596.

http://refhub.elsevier.com/S0048-9697(13)00656-6/rf0005







































































































Krammer K, Lange-Bertalot H. Bacillariophyceae 3. Teil: Centrales, Fragilariaceae,Eunotiaceae. In: Ettl H, Gerloff J, Heynig H, Mollenhauer D, editors. Süsswasserfloravon Mitteleuropa, 2/3. Stuttgart: Gustav Fischer Verlag; 1991a. p. 1–576.

Krammer K, Lange-Bertalot H. Bacillariophyceae 4. Teil: Achnanthaceae. KritischeErgänzungen zu Navicula (Lineolatae) und Gomphonema. GesamtliteraturverzeichnisTeil 1–4. In: Ettl H,GärtnerG,Gerloff J, HeynigH,MollenhauerD, editors. Süsswasserfloravon Mitteleuropa, 2/4. Stuttgart: Gustav Fischer Verlag; 1991b. p. 1–437.

Lecointe C, Coste M, Prygiel J. OMNIDIA: software for taxonomy, calculation of diatomindices and inventories management. Hydrobiologia 1993;269/270:509–13.

Leland HV, Porter SD. Distribution of benthic algae in the upper Illinois River basin inrelation to geology and land use. Freshw Biol 2000;44:279–301.

Mayer PM, Reynolds SK, McCutchen MD, Canfield TJ. Meta-analysis of nitrogen removalin riparian buffers. J Environ Qual 2007;36:1172–80.

Meyer JL, Wallace JB. Lost linkages and lotic ecology: rediscovering small streams. In:Huntly NJ, Levin S, editors. Ecology: achievement and challenge. Oxford: BlackwellScience; 2001. p. 295–317.

Meynendonckx J, Heuvelmans G, Muys B, Feyen J. Effects of watershed and riparianzone characteristics on nutrient concentrations in the River Scheldt Basin. HydrolEarth Syst Sci Discuss 2006;3:653–79.

Moore RD, Spittlehouse DL, Story A. Riparian microclimate and stream temperature re-sponse to forest harvesting: a review. J Am Water Resour Assoc 2005;41:813–34.

Morales EA, Vis ML, Fernández E, Kociolek JP. Epilithic diatoms (Bacillariophyta) fromcloud forest and alpine streams in Bolivia, South America II: A preliminary reporton the diatoms from Sorata, Department of La Paz. Acta Nova 2007;3:680–96.

Newall P, Walsh CJ. Response of epilithic diatom assemblages to urbanization influ-ences. Hydrobiologia 2005;532:53–67.

Niedermayr R, Schagerl M. Structuring factors of the phytobenthos community along amountain headwater (Kalkalpen National Park, Austria). Fundam Appl Limnol2010;177:93–104.

Ometo JP, Martinelli LA, Ballester MV, Gessner A, Krusche AV, Victoria RL, et al. Effectsof land use on water chemistry and macroinvertebrates in two streams of thePiracicaba river basin, Southeast Brazil. Freshw Biol 2000;44:327–37.

Osborne LL, Wiley MJ. Empirical relationships between land use/cover and streamwater quality in an agricultural watershed. J Environ Manage 1988;26:9–27.

Pan Y, Herlihy A, Kaufmann P, Wigington J, Van Sickle J, Moser T. Linkages amongland-use, water quality, physical habitat conditions and lotic diatom assemblages:a multi-spatial scale assessment. Hydrobiologia 2004;515:59–73.

Passy SI. Diatom ecological guilds display distinct and predictable behavior along nutri-ent and disturbance gradients in running waters. Aquat Bot 2007;86:171–8.

Pella H, Chandesris A, Wasson JG. Constitution d'un système d'information à référencespatiale dans le contexte de la Directive Cadre européenne sur l'eau. Ingénieries.Eau Agric Territ 2004;40:11–20.

Peterson BJ, WollheimWM, Mulholland PJ, Webster JR, Meyer JL, Tank JL, et al. Control ofnitrogen export from watersheds by headwater streams. Science 2001;292:86–90.

Poff NL, Voelz NJ, Ward JV, Lee RE. Algal colonization under four experimentally-controlledcurrent regimes in a high mountain stream. J N Am Benthol Soc 1990;9:303–18.

Potapova MG, Charles DF, Ponader KC, Winter DM. Quantifying species indicator valuesfor trophic diatom indices: a comparison of approaches. Hydrobiologia 2004;517:25–41.

Power ME, Stewart AJ, Matthews WJ. Grazer control of algae in an Ozark mountainstream: effects of short-term exclusion. Ecology 1988;69:1894–8.

Rhodes AL, Newton RM, Pufall A. Influences of land use on water quality of a diverseNew England watershed. Environ Sci Technol 2001;35:3640–5.

Richardson JS. Meeting the conflicting objectives of stream conservation and land usethrough riparian management: another balancing act. In: Scrimgeour GJ, Eisler G,McCulloch B, Silins U, Monita M, editors. Forest-land-fish conference II-Ecosystemstewardship through collaboration, April 26–28, 2004, Edmonton, Alberta, Proc.Forest-Land-Fish Conf. II; 2004. p. 1–6.

Richardson JS, Danehy RJ. A synthesis of the ecology of headwater streams and their ripar-ian zones in temperate forests. For Sci 2007;53:131–47.

Rimet F, Bouchez A. Life-forms, cell-sizes and ecological guilds of diatoms in Europeanrivers. Knowl Manag Aquat Ecosyst 2012;406:01.

Rosemond AD. Multiple factors limit seasonal variation in periphyton in a foreststream. J N Am Benthol Soc 1994;13:333–44.

Roth NE, Allan JD, Erickson DL. Landscape influences on stream biotic integrity assessedat multiple spatial scales. Landsc Ecol 1996;11:141–56.

Rott E, Pipp E, Pfister P, Van Dam H, Ortler K, Binder N, et al. Indikationslisten fürAufwuchsalgen in österreichischen Fließgewässern. Teil 2: Trophieindikationsowie geochemische Präferenz, taxonomische und toxikologische Anmerkungen.Wien: Wasserwirtschaftskataster herausgegeben vom Bundesministerium f.Land- u. Forstwirtschaft; 1999 [248 pp.].

Rott E, Pipp E, Pfister P. Diatommethods developed for river quality assessment in Aus-tria and a cross-check against numerical trophic indication methods used in Eu-rope. Algol Stud 2003;110:91–115.

Sabater S, Gregory SV, Sedell JR. Community dynamics and metabolism of benthic algaecolonizing wood and rock substrata in a forest stream. J Phycol 1998;34:561–7.

Sabater S, Butturin A, Clement JC, Burt T, Dowrick D, Hefting M, et al. Nitrogen removalby riparian buffers along a European climatic gradient: patterns and factors of var-iation. Ecosystems 2003;6:20–30.

Selby DA, Ihnat JM, Messe JJ. Effects of subacute cadmium exposure on a hardwatermountain stream microcosm. Water Res 1985;19:645–55.

Sherwood AR, Rintoul TL, Müller KM, Sheath RG. Seasonality and distribution ofepilithic diatoms, macroalgae and macrophytes in a spring-fed stream system inOntario, Canada. Hydrobiologia 2000;435:143–52.

Sliva L, Williams D. Buffer zone versus whole catchment approaches to studying landuse impact on river water quality. Water Res 2001;35:3462–72.

Spitale D, Lencioni V, Cantonati M. Relative importance of space and time in determin-ing the biotic structure in the upper part of spring-fed streams. Freshw Sci2012;31:586–98.

StatSoft Inc. STATISTICA for Windows (Computer program manual). Tulsa; 2001.Stenger-Kovács C, Lengyel E, Crossetti LO, Üveges V, Padisák J. Diatom ecological guilds

as indicators of temporally changing stressors and disturbances in the smallTorna-stream, Hungary. Ecol Indic 2013;24:138–47.

Studinski JM, Hartman KJ, Niles JM, Keyser P. The effects of riparian forest disturbanceon stream temperature, sedimentation, and morphology. Hydrobiologia 2012;686:107–17.

Systat Software Inc. Sigma Plot for Windows Version. 11.0. Dundas Software Ltd.; 2008ter Braak CJF, Šmilauer P. CANOCO reference manual and CanoDraw for Windows

user's guide: software for canonical community ordination (Version 4.5). Ithaca,New York; 2002.

ter Braak CJF, Verdonschot PFM. Canonical correspondence analysis and related multi-variate methods in aquatic ecology. Aquat Sci 1995;57:255–89.

Tornés E, Sabater S. Variable discharge alters habitat suitability for benthic algae andcyanobacteria in a forested Mediterranean stream. Mar Freshw Res 2010;61:441–50.

Van Dam H, Mertens A, Sinkeldam J. A coded checklist and ecological indicator valuesof freshwater diatoms from The Netherlands. Neth J Aquat Ecol 1994;28:117–33.

Van de Vijver B, Ector L, Beltrami ME, de Haan M, Falasco E, Hlúbiková D, et al. A criticalanalysis of the type material of Achnanthidium lineare W. Sm. (Bacillariophyceae).Algol Stud 2011;136/137:167–91.

Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing CE. The river continuumconcept. Can J Fish Aquat Sci 1980;37:130–7.

Walker CE, Pan Y. Using diatom assemblages to assess urban stream conditions.Hydrobiologia 2006;561:179–89.

Walsh G, Wepener V. The influence of land use on water quality and diatom communitystructures in urban and agriculturally stressed rivers. Water SA 2009;35:579–94.

Wellnitz TA, Rader R, Ward JV. Light and a grazing mayfly shape periphyton in a RockyMountain stream. J N Am Benthol Soc 1996;15:496–507.

Zampella RA, Laidig KJ, Lowe RL. Distribution of the diatoms in relation to land use andpH in blackwater coastal plain streams. Environ Manage 2007;39:369–84.
























































































































Effect of riparian vegetation on diatom assemblages in headwater streams under different land uses

Documents

Transcript of Effect of riparian vegetation on diatom assemblages in headwater streams under different land uses