ARTICLE

Macrophyte Structural Complexity Influences SpiderAssemblage Attributes in Wetlands

Eduardo Ribeiro Cunha & Sidinei Magela Thomaz &

Roger Paulo Mormul & Emanuel Giovani Cafofo &

Alexandre Bragio Bonaldo

Received: 7 August 2011 /Accepted: 12 January 2012 /Published online: 10 February 2012# Society of Wetland Scientists 2012

Abstract Macrophytes increase structural complexity inaquatic ecosystems and their emergent structures providehabitats for spiders. We sampled spiders in three species ofmacrophytes (Eichhornia azurea, Eichhornia crassipes andLimnobium laevigatum) and measured five traits indicative ofstructural complexity: horizontal structure, vertical structure,plant height, macrophyte richness and macrophyte biomass.We tested the hypothesis that spider density and diversityincrease and guild composition changes along a gradient ofstructural complexity provided by macrophytes. Vertical andhorizontal structure and macrophyte richness covaried withthe attributes of spider assemblages. However, vertical struc-ture was an important habitat-complexity trait because itcovaried with all spider attributes investigated, positivelyaffecting density and taxa richness and changing guild com-position. Our findings suggest that the increase in macrophytestructural complexity, primarily through vertical structure,provides additional habitat and microhabitat variability. Theseforms of variability may increase the availability of prey and

shelter and even reduce intraguild predation between spiders,thereby potentially increasing density and taxa richness.Moreover, an increase in vertical complexity provides avail-able structures for web attachment, favoring web-buildingspiders and consequently affecting guild composition. Thus,habitat structure plays an important role in structuring spiderassemblages and specific traits may mediate changes in par-ticular attributes of spider assemblages.

Keywords Guild composition . Habitat structure .

Associated fauna . Spider diversity

Introduction

Habitat complexity consists of the arrangement of the physicalelements in the environment in a way that provides support fororganisms (McCoy and Bell 1991). The pioneers who deter-mined the effect of habitat complexity on organisms wereMacArthur and MacArthur (1961), who described the rela-tionship between foliage height diversity and bird speciesdiversity. This relationship suggests that the habitat variabilityproduced by variation in the height of the foliage in treesinfluences the diversity of available habitats for birds, a resultthat in turn leads to increased species diversity. Furthermore,variation in habitat structural complexity is known to affectcommunity structure by changing several community attrib-utes (MacArthur and MacArthur 1961; McCoy and Bell1991). For example, an increase in habitat complexity altersmicroclimatic characteristics and ensures higher levels ofhabitat variability, prey availability, foraging sites, shelterand breeding microhabitats (Halaj et al. 1998; McNett andRypstra 2000; Romero and Vasconcellos-Neto 2005; Finkeand Denno 2006).

Aquatic macrophytes play an important role in structur-ing habitats in shallow coastal brackish waters and inlandaquatic ecosystems. These plants vary in their level of

Electronic supplementary material The online version of this article(doi:10.1007/s13157-012-0272-1) contains supplementary material,which is available to authorized users.

E. R. Cunha (*) : S. M. Thomaz : R. P. MormulPrograma de Pós-graduação em Ecologia de AmbientesAquáticos Continentais,Universidade Estadual de Maringá. Av. Colombo, 5790,87020-900 Maringá, PR, Brazile-mail: edurcunha@gmail.com

E. G. CafofoPrograma de Pós Graduação em Zoologia,Convênio Universidade Federal do Pará / Museu Paraense EmílioGoeldi. Av. Perimetral, 1901,66077-830 Belém, PA, Brazil

A. B. BonaldoLaboratório de Aracnologia, Museu Paraense Emílio Goeldi,Universidade Federal do Pará. Av. Perimetral, 1901,66077-830 Belém, PA, Brazil

Wetlands (2012) 32:369–377DOI 10.1007/s13157-012-0272-1

complexity and include a variety of life forms that occuralong depth gradients, providing habitat for aquatic organ-isms and consequently influencing the attributes of theirassociated assemblages. Most studies focusing on the rela-tionship between the associated fauna and the structuralcomplexity of macrophytes investigate aquatic organisms,such as fish and aquatic invertebrates (Jeffries 1993; Peliciceand Agostinho 2006; Thomaz et al. 2008). Nevertheless,these plants also provide emergent structures that supportterrestrial or semi-aquatic organisms such as spiders.

Spider assemblages also respond to the habitat complexityfactors that change species distributions and composition(Robinson 1981; Souza and Martins 2005; Gonçalves-Souzaet al. 2011). In general, spider abundance and diversity in-crease with habitat structural complexity, which covaries withthe availability of prey and of protection from predation (Halajet al. 1998; McNett and Rypstra 2000; Souza and Martins2005) and additionally, reduces intraguild predation and can-nibalism (Finke and Denno 2004; Finke and Denno 2006).Furthermore, changing patterns of guild composition in re-sponse to habitat complexity may arise from a combination offactors. Web-building spiders may be favored by an increasein available structures for web attachment. Conversely, wan-dering spiders may need less structured habitats to detect theirprey more easily (e.g., Robinson 1981).

Although habitat structural complexity involve severaldifferent characteristics of physical structures and their dispo-sition (see Tokeshi and Arakaki 2012), only a few traits mayactually influence the attributes of assemblages. Horizontaland vertical structures, structure height, biomass and evenplant richness are considered important determinants of thecommunity attributes of spider assemblages (Robinson 1981;Balfour and Rypstra 1998; Brandt and Lubin 1998; Halaj etal. 1998; Schaffers et al. 2008). However, most relationshipsbetween structural traits and spider attributes are describedfrom terrestrial ecosystems.

Pioneering studies in freshwater and salt marsh ecosys-tems investigated the relationships between the structuralcomplexity of macrophytes and spiders. For example, pre-vious studies analyzed the response of food web interactionsand habitat complexity using the presence and absence ofthatch (e.g., Finke and Denno 2006; Langellotto and Denno2006). A few other studies investigated the role played byaquatic macrophytes in assemblage structure and used onlya categorical classification of habitat complexities and com-parisons between macrophyte species with very distinctivearchitectures (Döbel et al. 1990; Raizer and Amaral 2001).Such approaches make it difficult to identify the traits ofhabitat complexity that may actually influence spider assem-blage structure in these ecosystems. Moreover, becausehabitat complexity integrates traits that vary among differentplants, an alternative approach for assessing the relationshipof habitat complexity to spider assemblages is to consider

gradients of physical structures. In addition, different traitsof plant structural complexity do not act in isolation (e.g.,Thomaz et al. 2008) and an analysis of habitat complexityshould therefore consider the simultaneous effects of multi-ple traits.

We tested the hypothesis that spider density and diversityincrease and guild composition changes with an increase inthe habitat structure provided by aquatic macrophytes. Ourassumption was that the increase in habitat structural com-plexity resulting from the increasing density of leaves ofaquatic macrophytes ensures a greater availability of habitat,which, in turn, affects spider density and diversity. Addition-ally, we assumed that changes in structural complexity wouldaffect distinct spider guilds differently (i.e., would change theproportion of web-building to wandering spiders). We alsoaimed to identify the macrophyte traits related to habitatstructural complexity that best explain the attributes of spiderassemblages in aquatic ecosystems. To test our hypothesis, weselected five surrogates of habitat structure that show variationand used these traits as predictor variables. The data necessaryto test this hypothesis were furnished by three species ofaquatic macrophytes with a similar architecture but differentcomplexities (Eichhornia azurea (Sw.) Kunth, Eichhorniacrassipes (Mart. Solms.) and Limnobium laevigatum (Humb.and Bonpl. ex Willd.) Heine.

Methods

Study Area

Our samples were collected in a lateral channel of the UpperParaná River, Brazil (22°48′43.44″S; 53°22′40.92″W), whichis nearly 2 km long and 30–90 m wide. The littoral zone iswell developed and supports large stands of E. azurea, E.crassipes and L. laevigatum. In addition to these species,several other free-floating (Salvinia auriculataAubl., Salviniaminima Baker, Salvinia biloba Raddi and Pistia stratiotes L.)and emergent macrophytes (Polygonum spp. and grasses) arecommon. Riparian vegetation is established along the entirechannel. This vegetation is dominated by diverse arborealspecies (e.g., Cecropia pachystachya Trécul, Inga uruguensisHook. & Arn. and Triplaris americana L.; Souza et al. 1997(Fig. S1A)).

Data Collection

Field samples were collected in April and July 2009from 7:00 am to 11:00 am. We selected three species ofmacrophytes (E. azurea, E. crassipes and L. laevigatum;Fig. 1a; Fig. S1a) with similar emergent architecturesbut different numbers and sizes of leaves. We collectedspiders in stands dominated by each species. Although

370 Wetlands (2012) 32:369–377

other species of macrophytes also occurred, their densi-ties were always low (see below).Ten replicate samplingunits (1 m×1 m), five in each month of sampling, weretaken per macrophyte species. Each sampling unit wascollected from an isolated stand (at distances from ca.30 to 1,000 m apart) to ensure independence amongsamples. Furthermore, the water surrounding the macro-phyte stands contributed to the isolation of assemblagesfrom distinct stands (Denno et al. 2004). The shoresides (right or left) sampled were randomized to mini-mize the influence of the time of day and the positionof the stand in the channel on our results.

To collect the spiders, we used a sampling device adaptedfrom Raizer and Amaral (2001) (Fig. 1b). Our device con-sisted of two quadrats constructed of 1 m×1 m PVC tubesand connected to each other with 20 cm PVC tubes. Thequadrat walls (20 cm in height) were covered by nylonnetting to prevent escape by the spiders. A nylon bag with-out a bottom (1.4 m long) was attached to this quadrat. Afterall macrophytes and the visible spiders had been removedfrom the inside of the quadrat (see below), this bag waspushed into the water, its lower end was closed and thequadrat was pulled out of the water. The material remaininginside the quadrat (plant debris, together with small spidersthat were difficult to see in the field) was preserved in

bottles for sorting in the laboratory. To catch spiders thatwere escaping by climbing the quadrat walls, we used analcohol spray (96%). The spiders collected in the field andthose collected with the plant debris and plant biomass werepreserved in 96% alcohol.

For each quadrat, we measured five macrophyte traits(Fig. 1a). These traits described only the emergent plantstructures and were used as predictor variables to explainspider abundance, richness and guild structure. The fivetraits were determined as follows:

(1). Horizontal structure was estimated as the mean num-ber of leaves (foliar lamina or petiole) touching a 1 mstick held horizontally and adjacent to the water sur-face in the center of the quadrat. This measurementwas repeated twice. The positions of the stick for thetwo measurements were perpendicular to each other.

(2). Vertical structure was estimated as the mean numberof times that leaves (foliar lamina or petiole) touched astick inserted vertically in the quadrat. This measure-ment was repeated five times at random positionsinside the quadrat.

(3). Plant height was measured at the same five points usedto estimate vertical structure and we considered themean value.

Fig. 1 The three species ofmacrophyte sampled forspiders. (a) Schematicillustration showing the similararchitecture of the three species.From left to right: E. azurea, E.crassipes and L. laevigatum.Three traits of habitat structuralcomplexity are shown.Continuous lines with closedcircles represent horizontalstructure measurements (3touches), continuous lines withopen circles represent verticalstructure measurements (3touches) and continuous lineswith tick marks represent plantheight measurements (15 cm).(b) Schematic illustration ofsampling device used to collectspiders on macrophytes. Thedevice was constructed fromPVC tubes and nylon nets

Wetlands (2012) 32:369–377 371

(4). Macrophyte richness was represented by the numberof macrophyte species inside the quadrat.

(5). Macrophyte biomass was determined as the dryweight of all emergent plant structures inside thequadrat. It was obtained after drying the plant materialin an oven at 70°C to a constant weight.

Data Processing

The spiders were identified to the lowest possible taxonomiclevel. The immature spiders were identified only to thefamily level because a more precise identification requiresmature individuals. All spiders were deposited in the MuseuParaense Emílio Goeldi collection.

Spider densities were expressed as individuals m−2. Forspecies richness, we first considered the total number of adultsidentified at the species level. Our second estimate of diversitywas taxa richness, defined as the sum of species richness andthe number of families identified only from juveniles (assum-ing that each sampled individual of a family thus identifiedmust represent at least one additional species). Only the taxarichness estimate of diversity was used in the analyses ofrelationships with habitat structural complexity. To minimizethe effect of sampling effort (i.e., the number of individuals inthe sample) as a confounding factor in taxa richness measure-ments (Gotelli and Colwell 2001), the value of taxa richnesswas adjusted before any analysis. To perform this adjustment,we used nonlinear regression to fit a Michaelis-Menten equa-tion to the taxa-richness relationship of the sampled individu-als (Fig. S2). The residuals in this regression analysisrepresented the variation in diversity not explained by sam-pling effort. These residuals, standardized to positive valuesby adding 3 to each residual, were used to represent theadjusted taxa richness.

The spiders were also classified into two simple huntingguilds (wandering and web-building spiders) according to thefamilies to which they belonged (Uetz et al. 1999). Using thisclassification, we estimated the percentage of each guild in asample. We used this information to create an index of guildcomposition. The value of this index is the difference betweenthe proportion of wandering spiders and the proportion ofweb-building spiders. This value ranges from −1 to 1. Morenegative and more positive values of the index represent ahigher proportional number of web-building and wanderingspiders, respectively. Values of the index near zero representan equivalent proportional guild composition.

We used the spider density, the adjusted taxa richness andthe guild composition index as response variables and thehorizontal structure, vertical structure, plant height, macro-phyte richness and macrophyte biomass as explanatory var-iables. Before the analyses, all variables except taxa richnessand the guild composition index were ln (x+1)-transformed

to produce normality. One sample was excluded from theanalysis as an outlier because no spiders were found in thesample. Thus, the value of guild composition index couldnot therefore be obtained for this sample.

Data Analyses

We used the data on adult individuals to construct speciesaccumulation curves with a rarefaction algorithm (R Devel-opment Core Team 2008). These curves enabled us to assesswhether our sample size was sufficiently large to reveal thetotal richness. To test our hypothesis of increase in spiderabundance and richness and of compositional changes withincrease in habitat structural complexity, we used linearregressions. However, because we had five descriptors ofhabitat complexity, we used the Akaike Information Crite-rion (AICC, a second-order criterion corrected for smallsample sizes) to select the best potential explanatory modelsubsets (see Johnson and Omland (2004) and Burnham andAnderson (2002) for additional information). These subsetswere selected based on the value of Δ AICC, which wasfixed at a maximum of four (Burnham and Anderson 2002).From the subsets thus selected, we only retained for interpre-tation those subsets that reached significance for all parame-ters of the model (intercept and beta, α00.05). The nullhypothesis would not be rejected if no relationship wasdetected. We used SAM v. 4.0 to calculate the value of AICC.STATISTICA 7.1 (StatSoft Inc 2005) was used for the regres-sion analyses.

Results

We collected a total of 579 spiders (but only 87 adults)belonging to 23 morphospecies and 13 families. The dom-inant morphospecies were Coleosoma sp1 (17 individuals),Agalenocosa sp1 (15), Salticidae sp1 (11), Actinosoma pen-tacanthum (Walckenaer 1842) (9), Tetragnatha sp1 (7) andOtoniela sp1 (6). The other morphospecies were only rep-resented by singletons (i.e., only one individual was found).The spider densities varied between 4 and 57 individualsm−2 and taxa richness varied from 1 to 8 taxa m−2. Neitherthe species accumulation curves constructed with individu-als collected in each macrophyte species nor the accumula-tion curve constructed with the total number of individualsreached an asymptote (Fig. 2).

In agreement with our expectations, all predictor varia-bles varied among the three plant species (Fig. 3). Thelowest trait values for horizontal and vertical structure (rep-resented by the number of leaves touching a stick) werefound in the stands dominated by E. azurea. Both variableswere 1.5 times greater in the stands dominated by E. cras-sipes than in the stands dominated by E. azurea, but the

372 Wetlands (2012) 32:369–377

values of these variables were 4.8 and 1.8 times greater,respectively, in the stands dominated by L. laevigatum. Theplant height and macrophyte biomass were lowest in L.laevigatum and higher in E. azurea (1.9 and 1.3 timeshigher, respectively) and E. crassipes (2.6 and 1.6 timeshigher, respectively). However, a consideration of all theresults revealed a gradient of values for the plant traits as awhole (Fig. 3).

Macrophyte richness was lowest in the L. laevigatumstands. Increasingly higher values were found in the E.azurea (1.6 times higher) and E. crassipes stands (2.1 timeshigher) (Fig. 3). Species that also contributed to macrophyterichness were Salvinia spp. (S. auriculata, S. minima, S.biloba), Oxycaryum cubense (Poepp. and Kunth) Palla,Azolla sp., Paspalum repens P.J. Bergius, Cyclanthera sp.

and Commelina sp. The biomass of these species was al-ways less than 12% of the total and therefore indicated thatthe three species selected were truly dominant.

In all, 31 possible models were generated for each spiderresponse variable. The spider density was positively relatedto macrophyte structural complexity and each of the two bestmodels retained only one variable, namely, vertical structureor horizontal structure (Table 1; Fig. 4a and b). Although theeffect of horizontal structure on spider density was not statis-tically significant, we decided to retain the horizontal struc-ture model because it was marginally significant (β1: p00.09; Table S1) and has ecological meaning. Spider taxarichness (adjusted for spider abundance) also increased withplant complexity and the best model explaining this variableincluded only vertical structure (Table 1; Fig. 4c). Guildcomposition also changed significantly with habitat structure.Although wandering spiders dominated throughout the gra-dient of habitat complexity, the proportion of web-buildingspiders increased with increasing habitat complexity. Thebest model explaining this attribute included only verticalstructure and macrophyte richness as explanatory variables(Table 1; Fig. 4d and e). Despite the findings of statisticalsignificance, the explanatory capacity of the models variedfrom low to moderate, as shown by the values of R2. Accord-ing to these values, the models explained 7–10%, 20% and39% of the total variation in spider density, taxa richness andthe guild composition index, respectively (Table 1).

Discussion

Our principal results showed that the structure of the spiderassemblages associated with the aquatic macrophytes isrelated with a gradient of structural complexity furnishedby these plants. Our sampling design, including three

Fig. 2 Individual-based species accumulation curves for spiders asso-ciated with the aquatic macrophytes E. azurea (open circles), E. cras-sipes (open triangles) and L. laevigatum (open squares. The largercurve with confidence intervals (closed diamonds) represents the totalnumber of individuals sampled

Fig. 3 Mean values±standard deviation of the traits used to charac-terize habitat structure for the three aquatic macrophyte species E.azurea (EA), E. crassipes (EC) and L. laevigatum (LL). The followingtraits are represented: horizontal structure—number of leaves touchinga 1 m stick adjacent to the water surface (a); vertical structure—

number of leaves touching a 1 m stick perpendicular to the watersurface (b); plant height—above-water plant height (c); macrophytebiomass—dry macrophyte biomass m−2 (d); macrophyte richness—number of macrophyte species m−2 (e)

Wetlands (2012) 32:369–377 373

macrophyte species with similar architecture but varying inthe quantity and size of horizontal and vertical structures,allowed us to infer the effects of different habitat-complexitytraits (vertical and horizontal structure and macrophyterichness) on the attributes of spider assemblages. Overall,our findings demonstrated that vertical structure may beconsidered the most important explanatory trait for struc-tural complexity because this trait was present in the

principal models explaining spider abundance, richnessand guild composition.

Spider density was explained by the positive effects ofvertical and horizontal structures. These variables representthe number of macrophyte leaves available for spider colo-nization and the extent to which these leaves are clumpedand intertwined to provide overlapping layers of availablehabitats. The preference of spiders for more complex hab-itats may reflect the suitability of microclimatic character-istics and the availability of habitats and prey items, all ofwhich are related to the increased physical complexity. Sucha relationship was previously reported by other authors forspiders in terrestrial ecosystems (Balfour and Rypstra 1998;Halaj et al. 1998; McNett and Rypstra 2000; Souza andMartins 2005) and even for aquatic macroinvertebrates (e.g.,Taniguchi et al. 2003; Thomaz et al. 2008). For example,spiders may be limited by microclimatic characteristics, suchas solar radiation and temperature, which can be amelioratedby an increase in the complexity of physical structures (e.g.,Riechert and Gillespie 1986; Uetz 1991). Such increasingcomplexity may also provide greater prey numbers, due to anincrease in foraging microhabitats per unit area (Dibble et al.1996; Halaj et al. 1998), a greater number of refugees frompredators (Gunnarsson 1990; Heck and Crowder 1991; Jeffries1993; Souza and Martins 2005) or may even provide specificfeatures required for particular species’ reproductive activity(e.g., Romero and Vasconcellos-Neto 2005).

Table 1 Models selected to explain the density, taxa richness andguild composition of spiders associated with aquatic macrophytes.Model selection was based on the Akaike Information Criterion(AIC). The independent variables used in the analysis were verticalstructure, horizontal structure, plant height, macrophyte richness andmacrophyte biomass

Selected model R2adju F p

Spider density

ln (SD)01.567+0.871 * ln (VS) 0.10 4.18 0.05

ln (SD)01.953+0.287 * ln (HS) 0.07 3.10 0.09

Taxa richness

TR0−0.393+2.262 * ln (VS) 0.20 8.11 <0.01

Guild composition

GC02.120+0.647 * ln (VS)+0.515 *ln (MR)

0.39 10.11 <0.01

SD Spider density, VS Vertical structure, HS Horizontal structure, GCGuild composition, MR Macrophyte richness, TR Taxa richness

Fig. 4 Relationships between spider assemblage attributes and habitatstructural complexity in aquatic macrophytes. (a) Spider density (indi-viduals m−2) and vertical structure (mean number of leaves touching a1 m stick perpendicular to the water surface); (b) spider density(individuals m−2) and horizontal structure (mean number of plant

structures along a 1 m horizontal transect); (c) taxa richness adjustedto abundance (taxa m−2) and horizontal structure; (d) guild composi-tion (proportion of wandering spiders minus the proportion of web-building spiders) and vertical structure; (e) guild composition indexand macrophyte richness (species m−2)

374 Wetlands (2012) 32:369–377

Antagonistic interactions between predators may alsoexplain the observed reductions in spider densities along adecreasing gradient of habitat structural complexity. Spi-ders, which show intraguild predatory and cannibalisticbehaviors, may experience negative effects on their densityin less structurally complex habitats (Finke and Denno2006; Langellotto and Denno 2006). Such negative effectsare well known in salt marsh ecosystems, where experimen-tal studies demonstrated that increases in habitat complexityreduce intraguild predation and cannibalism due to an in-crease in the availability of refuges. Our results do notprovide evidence supporting this mechanistic explanation.However, previous studies in aquatic ecosystems stronglysuggest that this mechanism is responsible for the observedreductions in spider densities with decreasing habitat com-plexity (Langellotto and Denno 2004; Finke and Denno2006; Langellotto and Denno 2006).

Spider taxa richness also increased with habitat complexity.The vertical structure shown by the macrophytes was thespecific trait associated with this increase. One can argue thattaxa richness could be indirectly affected by the number ofspiders sampled (a common confounding factor in diversitymeasurements; see Gotelli and Colwell 2001) because thisrichness measure is correlated with vertical structure andspider density. However, we controlled the effects of abun-dance on taxa richness and thereby allowed direct inferencesto be drawn about the relationship between species richnessand habitat complexity. The observed increase in spider taxarichness could therefore be associated directly with the in-crease in vertical structure. The clumping of leaves causes anincrease in vertical structure that produces greater structuralcomplexity over a vertical profile and provides stratifiedlayers of different microhabitats. Such microhabitat differen-tiation influences the diversity of available habitats in a givenarea, which can in turn increase species diversity (MacArthurand MacArthur 1961). This pattern is generally evident inother assemblages and is also well known to structure spiderassemblages in terrestrial ecosystems (Riechert and Gillespie1986; Uetz 1991; Castro and Wise 2009).

Macrophyte richness was an additional factor expected toinfluence spider taxa richness. Although our sampling de-sign was focused on stands of macrophytes that were strong-ly dominated by each of the three preselected species, othermacrophyte species were present at low densities. In fact,plant species composition has been identified as an impor-tant explanatory variable for spider assemblage structure interrestrial ecosystems (Beals 2006; Schaffers et al. 2008).This effect occurs because different plant species may havetheir own particular architecture. An increase in macrophyterichness should increase habitat variability and thereforeaffect spider taxa richness (Schaffers et al. 2008). However,our findings did not confirm this relationship. It is possiblethat those additional macrophyte species present were not

sufficiently abundant to cause a significant change in plantcomposition and, therefore, to affect spider assemblages.Future studies artificially creating multispecies stands ofmacrophytes with no single dominant species should bedesigned to investigate this relationship in aquatic ecosys-tems more effectively.

Spider richness (23 morphospecies) was lower than thevalues reported in the literature (33 and 34 species found byRaizer and Amaral (2001) and Döbel et al. (1990), respec-tively). However, these differences may result from highersampling effort and/or from a substantially greater diversityof habitats sampled by these other authors. The speciesaccumulation curves constructed for our study showed thatthe actual number of spider species found in our samplesremains far from its asymptotic value. The validity of thisfinding is particularly evident if we consider that our datawere collected from only three dominant species of macro-phytes colonizing a relatively small channel in the UpperParaná River. The detection of additional spider species isexpected if samples from other aquatic habitats of thisregion are included, which has ca. 153 species of macro-phytes (Ferreira et al. 2011).

Spider composition also changed with habitat complexitydue to an increase in the proportion of web-building spiders inmore complex patches of macrophytes. These changes indi-cated that web-building spiders are more successful in com-plex than in less complex habitats. Indeed, complex habitatshave more structural elements available for web attachment.Additionally, more complex habitats may also decrease pre-dation pressure because the mobility of web builders is gen-erally reduced (Askenmo et al. 1977; Finke and Denno 2006).Due to this characteristic, they are more susceptible to preda-tion in habitats where they are more easily detected. More-over, an increase in habitat structural complexity producesmore favorable microclimatic conditions. For example, theprotection of webs from wind destruction favors the settle-ment of web-building spiders (Riechert and Gillespie 1986;McNett and Rypstra 2000). In contrast, wandering spiders,which depend on hunting behavior, may have an advantage inless structured habitats. This argument at least applies to visualpredators, whose prey is more easily detected by sight in suchhabitats (Robinson 1981). However, although our results onthe relationship between changes in guild composition andhabitat complexity were consistent with the findings of terres-trial studies, results contradicting this relationship are knownfrom aquatic ecosystems. For example, Raizer and Amaral(2001) and Döbel et al. (1990) showed a limiting effect ofincreasing structural complexity on web-building spiders.Thus, although the relationship between guild compositionand habitat complexity may not be linear beyond a certainthreshold, the gradient of habitat complexity included in thecurrent study was not sufficient to identify a limiting effect onweb-building spiders.

Wetlands (2012) 32:369–377 375

The significant relationships between habitat complexityand the attributes of spider assemblages found by our studyand by other investigators (e.g., Döbel et al. 1990; Raizer andAmaral 2001) provide evidence of the important role playedby the physical structure of emergent macrophytes on inverte-brate communities. In fact, these findings indicate that such astructuring role is not restricted to underwater communities,such as those of fish and aquatic macroinvertebrates. Theserelationships therefore show that macrophytes may be a keytarget for conservation and management actions in aquaticenvironments because the structuring effects of macrophytesinclude both above-water and underwater fauna. Furthermore,because a single plant can furnish both above-water and un-derwater habitat structure (through its emergent and sub-merged structures, respectively), it may potentially play a keyrole in facilitating trophic-network interactions between thetwo environments. Spiders inhabiting above-water habitatsmay therefore prey on underwater organisms (e.g., aquaticinsects, crustaceans, fish, tadpoles) and may in turn serve asprey for birds (Askenmo et al. 1977; Akamatsu et al. 2004;Marczak and Richardson 2007). The same spiders mayfeed on terrestrial insects and may be eaten by fishes(Luz-Agostinho et al. 2006; Novakowski et al. 2008; Abilhoaet al. 2010). Thus, aquatic macrophytes may play an importantrole in the interchanges between underwater and above-waterfood webs.

In conclusion, our results support our hypothesis that agradient of increasing habitat structure provided by threespecies of aquatic macrophytes positively affects the abun-dance and taxa richness of spiders and changes the compo-sition of spider guilds. Additionally, our findings suggestthat the distinctions between different traits of habitat com-plexity may contribute to a better understanding of thechanges in different attributes of spider assemblages. How-ever, vertical structure should, in general, be highlightedas an important explanatory trait due to its role in addingnew layers of vertical habitat and thereby increasing thedensity and variability of microhabitats. Because ourstudy was limited to an observational approach, we rec-ommend further experiments to highlight the drivingmechanisms that shape the patterns of spider assemblagestructure along complexity gradients. We also encouragestudies aimed at understanding the role of macrophyte-colonizing spiders in exchanges between underwater andabove-water trophic networks and the role played bycomplexity in mediating this process.

Acknowledgements We would like to thank all our contributors andreviewers. We thank Thaísa S. Michelan, Juliana C. de Oliveira,Camila F. de Souza, Rosemara Fugi, Heloísa B. A. Evangelista, CristianeA. Umetsu and Cássia R. Ceole for their assistance with fieldwork. Weare also indebted to Sebastião Rodrigues and Alfredo Soares da Silva forfield support; David Candiani, Nancy L. M. Hung, Adalberto J. dosSantos, Antônio Brescovit, Daniele Polotow Geraldo and Regiane S.

Ferreira for sorting support; Katya E. Kovalenko for insightful sugges-tions and Luiz Carlos Gomes for statistical discussions. We additionallythank Núcleo de Pesquisas em Limnologia Ictiologia e Aquicultura forproviding facilities for the development of this study. ERC is extremelygrateful to ‘Coordenadoria de Aperfeicoamento de Pessoal de Nível Supe-rior’ (CAPES) for a Master’s degree scholarship. SMT and ABB areespecially thankful to the Brazilian Council of Research (CNPq) forcontinuous funding through Research Productivity grants.

References

Abilhoa V, Vitule JRS, Bornatowski H (2010) Feeding ecology ofRivulus luelingi (Aplocheiloidei: Rivulidae) in a coastal AtlanticRainforest stream, southern Brazil. Neotrop ichthyol 8:813–818

Akamatsu F, Toda H, Okino T (2004) Food source of riparianspiders analyzed by using stable isotope ratios. Ecol Res 19:655–662

Askenmo C, von Brömssen A, Ekman J, Jansson C (1977) Impact of somewintering birds on spider abundance in spruce. Oikos 28:90–94

Balfour RA, Rypstra AL (1998) The influence of habitat structure onspider density in a no-till soybean agroecosystem. J Arachnol26:221–226

Beals ML (2006) Understanding community structure: a data-drivenmultivariate approach. Oecologia 150:484–495

Brandt Y, Lubin Y (1998) An experimental manipulation of vegetationstructure: consequences for desert spiders. Isr J Zool 44:201–216

Burnham KP, Anderson DR (2002) Model selection and multimodelinference: a practical information – theoretic approach. Springer,New York

Castro A, Wise D (2009) Influence of fine wood debris on spiderdiversity and community structure in forest leaf litter. BiodiversConserv 18:3705–3731

Denno RF, Mitter MS, Langellotto GA, Gratton C, Finke D (2004)Interactions between a hunting spider and a web-builder: conse-quences of intraguild predation and cannibalism for prey suppres-sion. Ecol Entomol 29:566–677

Dibble ED, Killgore KJ, Harrel SL (1996) Assessment of fish plantinteractions. Am Fish Soc Symp 16:357–372

Döbel HG, Denno RF, Coddington JA (1990) Spider (Araneae) com-munity structure in an intertidal salt marsh: effects of vegetationstructure and tidal flooding. Environ Entomol 19:1356–1370

Ferreira FA, Mormul RP, Thomaz SM, Pott A, Pott VJ (2011) Macro-phytes in the Upper Paraná River floodplain: checklist and com-parisons with other large South American wetlands. Rev BiolTrop 59:541–556

Finke DL, Denno RF (2004) Predator diversity dampens trophic cas-cades. Nature 429:407–410

Finke DL, Denno RF (2006) Spatial refuge from intraguild predation:implications for prey suppression and trophic cascades. Oecologia149:265–275

Gonçalves-Souza T, Almeida-Neto M, Romero GQ (2011) Bromeliadarchitectural complexity and vertical distribution predict spiderabundance and richness. Austral Ecology 36:476–484

Gotelli NJ, Colwell RK (2001) Quantifying biodiversity: proceduresand pitfalls in the measurement and comparison of species rich-ness. Ecol Lett 4:379–391

Gunnarsson B (1990) Structure and the abundance and size distributionof spruce-living spiders. J Anim Ecol 59:743–752

Halaj J, Ross DW, Moldenkel AR (1998) Habitat structure and prey avail-ability as predictors of the abundance and community organization ofspiders in western Oregon forest canopies. J Arachnol 26:203–220

Heck KLJ, Crowder LB (1991) Habitat structure and predator–preyinteractions in vegetated aquatic systems. In: Bell SS, McCoy ED,

376 Wetlands (2012) 32:369–377

Mushinsky HR (eds) Habitat structure: the physical arrangementof objects in space. Chapman and Hall, London, pp 281–299

Jeffries M (1993) Invertebrate colonization of artificial pondweeds ofdiffering fractal dimension. Oikos 67:142–148

Johnson JB, Omland KS (2004) Model selection in ecology andevolution. Trends in Ecology and Evolution 19:101–108

Langellotto GA, Denno RF (2004) Responses of invertebrate naturalenemies to complex-structured habitats: a meta-analytical synthe-sis. Oecologia 139:1–10

Langellotto GA, Denno RF (2006) Refuge from cannibalism incomplex-structured habitats: implications for the accumulationof invertebrate predators. Ecol Entomol 31:575–581

Luz-Agostinho KDG, Bini LM, Fugi R, Agostinho AA, Júlio HF Jr(2006) Food spectrum and trophic structure of the ichthyofauna ofCorumbá reservoir, Paraná river Basin, Brazil. Neotropical Ich-thyology 4:61–68

MacArthur RH, MacArthur JW (1961) On bird species diversity.Ecology 42:594–598

Marczak LB, Richardson JS (2007) Spiders and subsidies: results fromthe riparian zone of a coastal temperate rainforest. J Anim Ecol76:687–694

McCoy ED, Bell SS (1991) Habitat structure: the evolution and diver-sification of a complex topic. In: Bell SS, McCoy ED, MushinskyHR (eds) Habitat structure: the physical arrangement of objects inspace. Chapman and Hall, London, pp 3–27

McNett BJ, Rypstra AL (2000) Habitat selection in a large orb-weaving spider: vegetational complexity determines site selectionand distribution. Ecol Entomol 25:423–432

Novakowski GC, Hahn NS, Fugi R (2008) Diet seasonality and foodoverlap of the fish assemblage in a pantanal pond. Neotropichthyol 6:567–576

Pelicice FM, Agostinho AA (2006) Feeding ecology of fishes associ-ated with Egeria spp. patches in a tropical reservoir, Brazil.Ecology of Freshwater Fish 15:10–19

R Development Core Team (2011) R: A language and environment forstatistical computing. R Foundation for Statistical Computing,Vienna, Austria. Available from: http://www.Rproject.org/

Raizer J, Amaral MEC (2001) Does the structural complexity ofaquatic macrophytes explain the diversity of associated spiderassemblages? J Arachnol 29:227–237

Riechert SE, Gillespie RG (1986) Habitat choice and utilization in webbuilding spiders. In: Shear WA (ed) Spiders webs, behavior, andevolution. Stanford University Press, Stanford, pp 23–48

Robinson JV (1981) The effect of architectural variation in habitat ona spider community: an experimental field study. Ecology 62:73–80

Romero GQ, Vasconcellos-Neto J (2005) The effects of plant structureon the spatial and microspatial distribution of a bromeliad-livingjumping spider (Salticidae). J Anim Ecol 74:12–21

Schaffers PA, Raemakers IP, Sýkora KV, ter Braak CJF (2008) Arthro-pod assemblages are best predicted by plant species composition.Ecology 89:782–794

Souza ALT, Martins RP (2005) Foliage density of branches and distri-bution of plant-dwelling spiders. Biotropica 37:416–420

Souza MC, Cislinski J, Romagnolo MB (1997) Levantamento florí-stico. In: Vazzoler AEAM, Agostinho AA, Hahn NS (eds) Aplanície de inundação do alto rio Paraná. EDUEM, Maringá, pp343–368

StatSoft Inc. (2005) STATISTICA (data analysis software system),version 7.1. Available from: http://www.statsoft.com

Taniguchi H, Nakano S, Tokeshi M (2003) Influences of habitatcomplexity on the diversity and abundance of epiphytic inverte-brates on plants. Freshwat Biol 48:718–728

Thomaz SM, Dibble E, Evangelista LR, Higuti J, Bini LM (2008)Influence of aquatic macrophyte habitat complexity on inverte-brate abundance and richness in tropical lagoons. Freshw Biol53:358–367

Tokeshi M, Arakaki S (2012) Habitat complexity in aquatic systems:fractals and beyond. Hydrobiologia. doi:10.1007/s10750-011-0832-z

Uetz GW (1991) Habitat structure and spider foraging. In: Bell SS, McCoyED, Mushinsky HR (eds) Habitat structure: the physical arrangementof objects in space. Chapman and Hall, London, pp 325–341

Uetz GW, Halaj J, Cady AB (1999) Guild structure of spiders in majorcrops. J Arachnol 27:270–280

Wetlands (2012) 32:369–377 377