Combined effects of habitat modification on trait composition and species nestedness in river invertebrates

Combined effects of habitat modification on trait composition and species nestedness in river invertebrates
of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institutionand sharing with colleagues.Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third partywebsites are prohibited.In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further informationregarding Elsevier’s archiving and manuscript policies areencouraged to visit:  Author's personal copy Combined effects of habitat modification on trait composition and speciesnestedness in river invertebrates Stefano Larsen * , S.J. Ormerod Catchment Research Group, Cardiff School of Biosciences, Cardiff University, Cardiff CF10 3US, UK  a r t i c l e i n f o  Article history: Received 26 May 2009Received in revised form 21 June 2010Accepted 6 July 2010Available online 10 August 2010 Keywords: InsectsMacroinvertebratesLand-useScaleSedimentsStreamsNestedness a b s t r a c t Changes in catchment land-use and sedimentation have large ecological effects on rivers, but there is lim-ited transferable understanding of the consequences for river conservation. In the Usk river system(Wales, UK), we assessed whether catchment-scale change in land-use and patch-scale sedimentation(i) affected organisms with specific life-history traits and (ii) resulted in nested assemblages with spe-cies-poor sites occupied mostly by sub-sets of organisms from richer sites.Reaches in catchments converted to agriculture had nested species assemblages characterised bygreater representation of organisms with small body size, shorter life cycle and effective dispersal capac-ity. In contrast, richer sites in semi-natural catchments supported taxa with longer life cycles. Patch-scalesedimentation was also accompanied by nested patterns in which depauperate patches supported taxamostly with shorter life cycles, small size and detrital feeding habits. Sediment-free patches were richerand characterised by larger taxa, poor dispersers and predators. Trait diversity was reduced by habitatmodification at both scales.We conclude that habitat modification in this river catchment has led to the systematic drop-out at twodifferent scales of specific groups of organisms with particular trait character. Large-scale agriculturalintensification appears to have removed larger, longer-lived invertebrates that probably require stableconditions, and we advocate further studies to appraise whether such organisms are at risk more globallyfrom land-use conversion. This river case study is one of the first to combine nestedness analysis withbiological trait assessment and it might help in developing transferable methods to predict the conserva-tion impacts of land-use change.   2010 Elsevier Ltd. All rights reserved. 1. Introduction Biodiversity loss and impairment are occurring at unprece-dented rates driven by climate change, habitat modification andhabitat loss (Travis, 2003). Although such effects are recognisedwidely in terrestrial habitats, less attention has been paid to fresh-waters, where extinction rates are rapid (Strayer and Dudgeon,2010; Clausnitzer et al., 2009). These systems are stressed directly,for example from pollution, sediment release, water abstraction,impoundment and introduced species, but also indirectly fromthe conversion of river catchments for agriculture or other use(Waters, 1995; Richter et al., 1997; Allan, 2004). Such conversiontypically affects river ecosystems through altered energy flux,hydrology, hydrochemistry, thermal regimes and habitat availabil-ity (e.g. Manel et al., 2000). Changes in sediment delivery are alsoinvolved (Lemly, 1982; Newcombe and MacDonald, 1991; Waters,1995; Gayraud et al., 2002; Parkhill and Gulliver, 2002; Greig et al.,2005), but these can arise both from catchment-wide erosion orfrom local discharges, bank-erosion or livestock encroachment inthe riparian zone (Walling et al., 1999). Distinguishing the conser-vation effects on rivers of sedimentation and wider land-usechange is therefore challenging (Larsen et al., 2009, in press).Despite long-standing efforts to quantify the biological impair-mentof streams andrivers,there is still limited transferableunder-standing of the types of organisms at risk, and of the conservationconsequences. This is partly because ecological assessment infreshwaters has focused dominantly on the taxonomic identity of affected organisms rather their ecological character (e.g. Hellawell,1986; Rosenber and Resh, 1993; Davy-Bowker et al., 2005). Ap-proaches restricted to taxonomic identity not only limit regionalcomparison of stressor effects and conservation impacts, but theyalso limit mechanistic understanding of biological responses (Gay-raud et al., 2003). A possible alternativeis to appraise the biologicaltraits of vulnerable organisms. For example, extinction pronenessappears to be related to life-history specialization, body massand reproductive rate (Duffy, 2003; Kotiaho et al., 2005; Oldenet al., 2007). Impaired sites can acquire distinct trait structure 0006-3207/$ - see front matter    2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.biocon.2010.07.006 *  Corresponding author. Present address: Leibniz-Institute of Freshwater Ecologyand Inland Fisheries (IGB), Mueggelseedamm 310, 12587 Berlin, Germany. Tel.: +4930 641 81607. E-mail address: (S. Larsen).Biological Conservation 143 (2010) 2638–2646 Contents lists available at ScienceDirect Biological Conservation journal homepage:  Author's personal copy among organisms as environmental constraints in any given loca-tion progressively ‘filter’ out types of species whose resourcerequirements are not met (Southwood, 1977; Usseglio-Polateraand Tachet, 1994; Richards et al., 1997; Duffy, 2003). Such trait-based approaches to understanding environmental change havebeen applied particularly to European freshwaters, where informa-tion on invertebrate life-history is readily available (Statzner et al.,2001b). So far, however, most applications have focused on biolog-ical monitoring, pollution assessment and restoration (Gayraudet al., 2003; Statzner et al., 2007; Horrigan and Baird, 2008; Paillexet al., 2009; Tullos et al., 2009). Applications to species loss and riv-er conservation are scarcer (Bonada et al., 2007; Dolédec and Statz-ner, 2008; Townsend et al., 2008).A complementary method of appraising the effects of environ-mental change on species loss involves the assessment of ‘nested-ness’ in species composition. This effect occurs where organisms atspecies-poor sites form a sub-set of those at richer sites (Atmar andPatterson, 1993; Rodriguez-Girones and Santamaria, 2006). Perfectnesting occurs where rarer species are exclusive to species-richlocations, while reduced nesting occurs where rare species are dis-tributed more evenly (Wright and Reeves, 1992; Wright et al., 1998a). While nesting can result naturally from colonization andextinction processes over large timescales, it also arises from hu-man disturbance, habitat alteration and behaviour-related patchuse (Fernandez-Juricic, 2002; Summerville et al., 2002; Hylanderet al., 2005; Larsen et al., in press). Nestedness also reveals conser-vation risks because of the way that (i) rare species are restrictedinto small numbers of locations and (ii) species ‘drop-out’ system-atically along gradients of environmental change or degradation.Although sometimes criticised (e.g. Donlan et al., 2005), this meth-od has been applied to appraise conservation effects in several eco-system types (Fernandez-Juricic, 2002; Schouten et al., 2007;Horsak and Cernohorsky, 2008). However, as with species traitstudies, there have been few applications to river conservationand impairment. Studies involving the combined use of these ap-proaches are scarce in any ecosystem.In this paper, we assess the effects of land-use and sedimenta-tion on trait-dependent sensitivity in a Welsh river catchmentand appraise how such effects affect species nesting (Fleishmanet al., 2007). We hypothesised that, because species differ in theirtolerance to modification, any systematic drop-out of species with‘sensitive’ traits at sites affected by intensive land-use or sedimen-tation would generate nested patterns (Cutler, 1991; Atmar andPatterson, 1993; Heino et al., 2009); both these effects would indi-cate conservation ramifications associated with the loss of traits orspecies under modified conditions. We expected that catchmentland-use and localised sedimentation would generate nested spe-cies assemblages respectively at the reach and patch-scales: (i)by favouring taxa with relatively rapid population growth and in-creased resilience where natural land-use has been converted topasture, for example because of increased temperature, habitatsimplification, increased runoff and increased nutrient concentra-tion (Doledec et al., 2006; Dolédec and Statzner, 2008; Townsendet al., 2008) and (ii) by favouring taxa with resilience traits in siteswith shifting and unstable patches of fine sediment, where effectson feeding and behavioural character should also arise (Richardset al., 1997; Rabeni et al., 2005; Townsend et al., 2008). A furtherprediction was that burrowing and detrital-feeding organismsshould be more common in sediment-affected patches while filterfeeders and grazing organisms should be impaired (Rabeni et al.,2005; Larsen et al., in press; Larsen and Ormerod, 2010; Table 1). We examined whether any nesting was accompanied by changesinoverall trait diversityat sites affected by land-useintensityor lo-cal sediment deposition. 2. Methods  2.1. Study area The Usk, in south-central Wales (UK), is listed as a Special Areaof Conservation under the EuropeanHabitats Directive (92/43/EEC)and holds one of the most important game fisheries in England andWales. The catchment is suitable for appraising combined land-useand sedimentation effects because ionic composition and geologyare relatively homogeneous across the catchment (e.g. total oxi-dised nitrogen 0.65–1.58 mg/L, hardness 49.5–146 mg/L), whileurbanization is negligible. Moreover, land-use varies from semi-natural upland vegetation to increasingly intensive pastoral (sheepgrazing) grassland over a sandstone dominated catchment that ispotentially sensitive to sediment mobilisation (Larsen et al., 2009).  2.2. Survey design The overall survey design, study locations and analytical ap-proaches have been described by Larsen et al. (2009), who alsodemonstrated statistical independence among the survey reaches.Briefly, a reach-scale survey was conducted in summer 2006 to as-sess catchment and reach-scale land-use influences on inverte-brates assemblages. It involved 32 catchment reaches over analtitudinal range of 190–400 m a.s.l. along 18 s-order tributariesdraining semi-natural upland vegetation (bracken, heath, moor-land, acid grassland) and improved pasture grassland. Deciduouswoodland never exceeded 34% of any catchment. While coveringthe main land-uses, reaches were selected to minimize variationsin depth, flow velocity, stream width and particle size distribution(Appendix A). On each of fourteen streams, two reaches were sam-pled, while four streams had samples from one reach, dependingon accessibility and stream length.In 2007, a patch-scale survey assessed within-reach effects of sediment deposition on invertebrate assemblages at locations  Table 1 Predicted effects of land-use intensification and patch-scale sedimentation on biological traits in the Usk catchment, Wales. The expectation was that nested species assemblageswould result either from the gain (+) or loss (  ) of each trait indicated. See text for more details and references. Impact Trait Category Mechanism/rationaleCatchment grasslands Maximal size Small-medium (+) Rapid growth and increased resilience inmore frequently disturbed habitatsLife-cycle duration <1 year (+)No. generation/year >1 (+)Fine sediments Maximal size Small-medium (+)Life-cycle duration <1 year (+)No. generation/year >1 (+)Feeding habits Scraper (  ) Affected by algae smotheringFilterers (  ) Clogging of filtering devicesDetrital feeders (+) Feed on fine materialLocomotion Burrowers (+) Adapted to fine substrataTemporarily attached (  ) Lack of stable surfaces S. Larsen, S.J. Ormerod/Biological Conservation 143 (2010) 2638–2646   2639  Author's personal copy nested within 12 previously sampled reaches on eight streams.Stream patches ( c  . 3m 2 ) represented both upland and pasturegrassland while covering a wide range of fine sediment cover. Onlyriffle-glide habitats were included. According to availability, threeto six patches per reach were sampled in a   15 m long section,for a combined total of 56 patches (see Larsen et al. (2009) forselection methods).  2.3. Environmental variables – reach scale Land-use data were extracted from an existing GIS land-coverlayercreatedbytheCountrysideCouncilforWales(CCW,2002)de-rivedfrom 1:10,000 maps.For everyreach, land-usewas calculatedfor the whole catchment area as well as for a 150 m wide buffer oneachsideofthestreamupto1 kmupstream.Catchmentareaswerederivedfrom 10  10 m resolution DigitalElevation Map of the Uskcatchment(CCW2002).Variationamongmacroinvertebratespeciesassemblagesatthisscaleismostlyrelatedtocatchmentandriparianland-use, namely the percentage of pasture grassland (Larsen et al.,2009). Because riparian (150 m buffer) and catchment land-use arecorrelatedclosely,weusedonlycatchmentimprovedgrasslandasaprimary descriptor of land-use intensity. While reaches drainingsemi-natural vegetation tended to be at higher altitudes, all siteswerewithin210 mofeachother,anddistancebetweentworeachesin a stream was always less than 8 km. In addition, there were nosignificant linear correlations between land-use and natural attri-butesofthesitessuchaswidth,currentvelocity,streamorder,totalhardness and other measures of acid–base status that might con-found any land-use effects on organisms (all  r  s  < 0.5,  P   > 0.1).Monthly concentrations of nitrate (mg/L), phosphate (mg/L),Biochemical Oxygen Demand (BOD mg/L), water hardness asCaCO 3  (mg/L) and pH were available for 12 reaches from the UKEnvironment Agency’s Water Management Information System(WMIS) based on standard methods (Standing Committee of Ana-lysts, 1981, 1979, 1987, 1992). Means for each determinand werecalculated for the year antecedent to invertebrate sampling.Although incomplete, the chemical data from streams drainingboth pasture grassland and semi-natural upland vegetation al-lowed assessment of the extent to which water quality might influ-ence invertebrates. Most reaches without chemical data werehigher altitude locations likely to have the lowest nutrient concen-trations (Clews and Ormerod, 2009).  2.4. Environmental variables – patch-scale Patch-sale environmental data were collected in the field.Water depth and flow velocity were measured in each patch, anddeposited fine sediment (<2 mm) in each sampling site (c. 3m 2 )was measured using a circular quadrat (300 cm 2 ) where the per-centage of stream bed covered by fine particles was estimated in5% increments (Zweig and Rabeni, 2001). Ten observations of sed-iments immediately around Surber samplersused for invertebrateswere averaged. In addition, the amount of suspendable sedimentfrom the stream bed was measured at 49 of the patches sampled,with four measurements made around each Surber location (Lar-sen et al., 2009). A metal box (24  18 cm) was pushed into thesubstratum and sediments from  c.  2 cm depth were entrained inthe measurable volume of water. Suspended sediments withinthe range of 0.063–1 mm were filtered from 1 l of water, dried,weighed and calculated as g/m 2 . Organic material from the col-lected sediments was reduced by washing and decanting.  2.5. Invertebrate sampling Reach-scale survey  (June–July 2006). Benthic macroinverte-brateswerecollectedusingkick-samples of threeminutesdurationacross riffles and glides (typical < 55 cm/s) over a 10 m reach. Weused a standard hand-net (mesh 0.9 mm, 25  25 cm) to mimictypical biological monitoring (Wright et al., 1998b). This methodis in wide use, and effectively detects a large proportion of streamorganisms present in any give reach while providing sufficient pre-cision to detect differences among sites and times (Bradley andOrmerod, 2002). Patch-scale survey  (June–July 2007). A Surber sampler (0.16 m 2 ;0.44 mm) was used to assess smaller-scale variations in composi-tion and to cover the defined area of stream bed where sedimentcharacter was also assessed. The two collection methods, althoughdiffering in spatial extent, provided similar information aboutassemblage composition (Larsen et al., 2009). The difference intiming could affect interpretation, but methodological evaluationssuggest this to be unlikely: in the 12 reaches where invertebrateswere collected in subsequent years by both kick-samples andaggregated Surber samples, assemblage composition (DetrendedCorrespondence Analysis Axis 1) was highly inter-correlated be-tween the 2 years ( r   = 0.86,  n  = 12;  P   < 0.001; (Larsen et al., 2009).Other data from upland Welsh streams also show high inter-an-nual persistence in composition between successive years exceptfollowing major climatic variation (Bradley and Ormerod, 2001;Ormerod & Durance, 2009).  2.6. Invertebrate biological traits The functional composition of all invertebrate samples was de-fined using 51 categories of 11 biological traits (Table 2) based onavailable information (Richoux, 1994; Tachet et al., 1994, 2000;Usseglio-Polatera, 1994; Usseglio-Polatera and Tachet, 1994).Trait information was collected for a total of 50 taxa, mostly gen-era, but Oligochaeta and Hirudinea were omitted because of incomplete information. Fuzzy coding was used to determinethe affinity of each taxon for categories (Chevenet et al., 1994)that ranged between 0 and 3, or 0 and 5. Affinity scores were re-scaled as proportions (sum = 1) for each taxon, while group-wideaverages were used for taxa identified at coarser level (e.g. Dip-tera families). Taxon  trait-categories at each site were multi-plied by the log(  x  + 1) abundance to produce a site  traitabundance matrix, and abundance-weighted trait profiles con-verted to frequency distributions for each trait (Dolédec et al.,2000; Archaimbault et al., 2005).  2.7. Data analysis Trait diversity (TD) in each sample for both surveys was calcu-lated as the average Simpson diversity ( S  ) across all traits: TD  ¼  S   ¼  1   D i ;  where  D i  ¼ X ð n  j = N  Þ 2 with  D i  = Dominance index of trait  i ;  n  j  = relative abundance of traitcategory  j ;  N   = total abundance of all trait categories. The average of  S   across all traits was calculated to account for the lack of indepen-dence among traits (e.g. Bêche and Resh, 2007). In addition, becausetraits can have multiple categories (e.g. feeding traits are dividedinto filterer, grazers, predators, etc.) we also assessed how diversitywithin traits varied across locations using Simpson’s diversityindex. Also, correlation (Pearson product-moment) between overalltrait diversity and taxonomic richness was examined at the twoscales (Micheli and Halpern, 2005; Bêche and Resh, 2007).At the patch-scale, relationship between sediment measures(cover and suspendable sediments) and trait diversity was as-sessed using General Estimating Equations (GEEs). This approachwas used because multiple patches per reach were sampled inthe 2007 survey and samples could not be treated as independent 2640  S. Larsen, S.J. Ormerod/Biological Conservation 143 (2010) 2638–2646   Author's personal copy in conventional regression analyses. Where data are collected inclusters and within-cluster correlation is expected, GEEs accountfor the lack of statistical independence between samples by adjust-ing regression coefficients and variance to avoid spurious correla-tions (see Zorn, 2001; Larsen et al., 2009). This approach is avaluable tool to analyse spatially and temporally correlated dataof the type that are common in ecological research (Larsen et al.,2009). GEEs models were run using  R  (Ihaka and Gentleman,1996) using the program  geeglm  from the library  geepack  (Halekohet al., 2006).To quantify the level of nestedness across communities in bothsurveys, we used the binary-matrix nestedness temperature calcu-lator (Atmar and Patterson, 1993; Rodriguez-Girones and Santam-aria, 2006). The temperature method is relatively insensitive tomatrix size, and also correlates well with other metrics (Rodri-guez-Girones and Santamaria, 2006). BINMATNEST works on thespecies presence/absence matrix, re-ordering rows and columnsto maximizematrix nestedness to calculate a temperature (rangingover 0–100   C) which reflects the matrix deviation from an idealnested structure; perfectly nested matrices with rare taxa in richlocations have  T   = 0   C while totally random matrices have T   = 100   C. The statistical significance of the observed temperaturewas assessed using a Monte-Carlo approach involving comparisonwith simulated temperatures of 400 random generated matrices.In the more conservative null-model III used here, the probabilityof a cell being occupied equals the average probabilities of occu-pancy of its row and column. This null model is considered themost reliable as it is less sensitive to species richness and occur-rences (Heino et al., 2009). The order with which sites are sortedin the maximally packed matrix can then be compared with inde-pendent correlates to assess the likely determinants of nestedness.We used Spearman-rank correlation to evaluate the influence of land-use and sediment measures on the nested patterns of com-munities from the reach and patch-survey respectively (Townsendand Hildrew, 1994; Schouten et al., 2007).To assess whether nestedness in macroinvertebrate assem-blages was mediated by their biological traits, the ranking of sitesin the maximally packed matrix was related to the proportions of each trait category using Spearman-rank correlations. In order totest predictions outlined in the Introduction, we focused on traitsdescribing size, life cycle, dissemination potential, resistance, feed-ing habits and locomotion. We corrected alpha values (=0.05) bydividing by the number of categories within each trait (e.g. if a gi-ven trait has six categories,  a  = 0.008). Results from this analysiswere checked by GEEs to control for independence among sites,but there were no departures from standard Spearman-rankcorrelations. 3. Results  3.1. Reach-scale trends The percentage of improved grassland in the catchments stud-ied ranged from 0% to 64%. Phosphate never exceeded 0.04 mg/L,while mean nitrate concentrations were in the range 0.65–1.58 mg/L; both nitrate ( r  s  = 0.69,  P   < 0.05) and BOD ( r  s  = 0.81, P   < 0.01) increased moderately at intensified grassland sites. How-ever, sediment cover (averaged from reach- and patch-scale mea-sures) varied independently of land-use, implying that sedimentand land-use effects were distinct (see Larsen et al., 2009).Overall trait diversity declined with increasing cover of inten-sive grassland ( r   =  0.46;  p  = 0.008). There were also trends in indi-vidual trait diversity, including maximal size ( r   =  0.7;  p  < 0.001),life-cycle duration ( r   =  0.36;  p  = 0.04) and resistance forms( r   =  0.35;  p  = 0.04). Interestingly trait diversity per reach wasnot related to taxonomic richness (Fig. 2;  p  = 0.6).Assemblages had a significantly nested sub-structure( T   = 36.4  ;  p  < 0.001; Fig. 1). Site ranking in the maximally packedmatrix (i.e. the species-by-site matrix when ordered by BINMAT-NEST to reflect nesting) was significantly correlated with grass-land intensity ( r  s  = 0.43;  p  = 0.01), suggesting that land-use was apotential cause of the nesting pattern. Reaches that were lessnested were characterised by higher representation of longer lifecycles (<1 generation/year; Table 3), while reaches at the nestedend of the matrix were characterised by higher representation of small body sizes (5–10 mm), short life cycles (<1 year) and goodlarval dispersal. In other words, changing trait pattern was relatedto the formation of nested assemblages as land-use varied amongsites.  Table 2 Biological trait and categories considered in the present study. Traits were assigned atthe genus level. Trait CategoriesMaximal size 2.5–5 mm5–10 mm10–20 mm20–40 mmLife-cycle duration <1 year>1 yearNo. of potential generations per year <11>1Dissemination potential <10 m10–100 m100 m to 1 km>1 kmAquatic stage EggLarvaNymphAdultReproduction OvoviviparityIsolated eggs, freeIsolated eggs, cementedClutches, fixedClutches, freeClutches, in vegetationClutches, terrestrialResistance forms Eggs, statoblastsCocoonsDiapauseNo resistance formsFood Fine sedimentsDetritus < 1 mmDetritus > 1 mmLiving microphytesLiving macrophytesDead animal < 1 mmLiving microinvertebratesLiving macroinvertebratesFeeding habits Deposit feederShredderScraperFiltererPredatorsRespiration TegumentGillPlastronAerialLocomotion FlierFull water swimmerCrawlerBurrowerInterstitialTemporarily attached S. Larsen, S.J. Ormerod/Biological Conservation 143 (2010) 2638–2646   2641


Mar 15, 2018

02 - EE_aula02

Mar 15, 2018
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!