Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast

Fish responses to increasing distance from artificial reefs on the SoutheasternBrazilian Coast

Luciano Neves dos Santos a, Daniel Shimada Brotto b, Ilana Rosental Zalmon c,⁎

a Department of Ecology and Marine Resources, Federal University of Rio de Janeiro State, Rio de Janeiro, Brazilb Department of Health Sciences, Veiga de Almeida University, Rio de Janeiro, Brazilc Centre of Biosciences and Biotechnology, North Rio de Janeiro State University, Campos dos Goytacazes, Brazil

a b s t r a c ta r t i c l e i n f o

Article history:

Received 18 September 2009

Received in revised form 29 December 2009

Accepted 27 January 2010

Available online xxxx

Keywords:

Brazil

Fish assemblages

Generalized additive models

Gillnet

Reef balls

Visual census

Artificial reefs have been deployed worldwide in shallow marine environments to enhance attraction and

capture of fish, but, despite their potential as a fishery management option, few studies addressed whether

fish assemblages will change with distance from a reef. We experimentally assessed the relationship

between fish abundance and richness to increased distance (0, 25, 100, and 300 m) to reef balls deployed on

a flat and homogeneous bottom, 9-m deep, on the north coast of Rio de Janeiro, Southeastern Brazil. A total of

thirty fish species was associated with the reefs and the variation in species composition with sampling

period or technique did not account for changes in fish assemblages with reef distance. Fish abundance and

richness were significantly greater at the nearest distances (0 m and 50 m) to the reefs than at 300 m, but

while the first dependent variable decreased exponentially with reef distance the later diminished linearly.

Fish responses to reef distance were clearly species-specific and the tomtate (Haemulon aurolineatum — the

dominant species that probably used the reefs for shelter purposes) strongly influenced the exponential

decreasing pattern found for the total number of fishes and abundance of Haemulidae. The abundance of the

close congener Haemulon steindachneri, and of Chaetodon striatus, Serranus flaviventris and serranid species as

a group also decreased, but linearly, with reef distance, probably related to variation in home-range.

Acanthurus chirurgus and Scorpaena isthmensis decreased linearly in number with reef distance, probably due

to feeding rather than refuge purposes. The abundance of Isopisthus parvipinnis tended to increase with reef

distance, suggesting that artificial reefs might be unprofitable for those species less dependent on clear water

or structurally complex habitats to obtain refuge or food. Overall, our study indicates that artificial reefs are a

promising management tool to enhance fish capture in the oligotrophic waters and structureless bottoms of

the north coast of Rio de Janeiro, but with varied rates of success, depending on the target species and fishing

distance from the reefs.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Artificial reefs have been deployed worldwide in shallow (b30 mdepth) marine environments to aid in environmental rehabilitation(i.e., providing new suitable habitats for invertebrates and fish) butprimarily to enhance attraction and catch of fish resources (Seamanand Sprague, 1991; Seaman, 2000). Most scientific studies on artificialreefs, however, have been done on the Atlantic and Pacific coasts ofthe United States, in Japan and throughout the Atlantic andMediterranean seashores in Europe (Baine and Side, 2003). Despitethe growing number of applied initiatives, there is still little researchassessing the use of artificial reefs for fisheries in South America, andeven fewer manipulative experimental studies that tested specific

hypotheses (but see Godoy and Coutinho, 2002; Brotto et al., 2006a;Galván et al., 2008).

The global nature of artificial reef research, however, hasfacilitated much scientific debate about their effectiveness. Forinstance, the intense dispute on the attraction versus productiontheme has lasted for decades (Lindberg, 1997; Brickhill et al., 2005),but there is no doubt that artificial reefs may improve the catch of fishresources by redistributing the exploitable biomass, increasing onlythe exploitable biomass but not the total biomass, or increasing boththe exploitable and total biomass (Bortone, 1998). There have beenmany efforts to appraise the extent to which artificial reefs will affectthe surrounding community, but these studies focused largely oninvertebrate communities (Barros et al., 2001; Fabi et al., 2002a;Danovaro et al., 2002; Galván et al., 2008). Despite the appliedinterests, knowledge on how fish assemblages respond to increasingreef distance is restricted to few studies, all performed in the northernhemisphere (Løkkeborg et al., 2002; Soldal et al., 2002; Jordan et al.,2005).

Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

⁎ Corresponding author. Present address: Laboratory of Environmental Sciences,

Avenida Alberto Lamego 2000, Horto-Centro 28013-602, Campos dos Goytacazes, RJ,

Brazil. Tel.: +55 22 2726 1470; fax: +55 22 2726 1472.

E-mail address: [email protected] (I.R. Zalmon).

JEMBE-49051; No of Pages 7

0022-0981/$ – see front matter © 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.jembe.2010.01.018

Contents lists available at ScienceDirect

Journal of Experimental Marine Biology and Ecology

j ourna l homepage: www.e lsev ie r.com/ locate / jembe

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018

Here, we analyze the change in fish assemblages with increasingdistance from artificial reefs deployed along the north coast of Rio deJaneiro, Southeastern Brazil. Our purpose was to experimentallyaddress, through visual censuses and gillnet samplings, whether fishassemblages change with gradual increase in distance (0, 25, 100, and300 m) to reef balls deployed on a flat and homogeneous bottom, 9 mdeep.We predicted that abundance of most species will decrease withreef distance but that the rate of change is species-specific. The use ofartificial reefs as a fishery management tool is also addressed withrespect to their potential to attract and affect the prevalent species.

2. Methods

2.1. Study site

The north coast of Rio de Janeiro (Southeastern Brazil) (Fig. 1) isnaturally depleted of rock or other hard substrates, and is covered byextensive sandy beaches with variable amounts of mud andcalcareous nodules (i.e., rhodolites; Zalmon et al., 2002). This area islocated in a transitional zone between warm and oligotrophic watersof the Brazil Current from the north and cold, nutrient-rich upwellingof the South Atlantic Central Water from the south (Valentin andMonteiro-Ribas, 1993). Primary productivity (chlorophyll a) is low,Secchi depth does not exceed 4 m, and strong bottom currents arecommon (Krohling and Zalmon, 2008). Although dominated byoligotrophic waters and homogeneous bottom relief, the north coastof Rio de Janeiro is often exploited by local inshore artisanal fishermen(Zalmon et al., 2002).

Togetherwith oceanic circulation, the north coast of Rio de Janeiro isalso strongly influenced byweather and freshwater runoff. The outflow

of the Paraíba do Sul River (the largest river in theRio de Janeiro State) isespecially important during the rainy period (December to February),when a turbid (Secchi depth b0.5 m) and polyhaline (18–33 psu)estuarine plume spreads over 15 km north from the river mouth,covering most of the continental shelf up to ca. 10 km distant from theshore (Godoy et al., 2002). This plume does not, however, reach the seabottomduring the rainyperiod, because the local tradewinds lead to theintrusion of clearer and saline bottom waters. During the dry period(April to November), but mostly during winter, the intensity ofsouthwest winds increases, stratification ceases and, consequently,water turbidity significantly increases near the bottom (Godoy et al.,2002).

Since 1996, invertebrate and fish colonization of artificial reefsalong the north coast of Rio de Janeiro have been investigated, toassess the role of man-made structures in management andconservation of local fishery resources (Zalmon et al., 2002; Krohlinget al., 2006). More than 40 fish species have been recorded to beassociated with reefs of different materials and complexities, butmostly in the form of concrete modules, the most effective inattracting and harboring fish (Zalmon et al., 2002). Overall, thedeployment of artificial structures on the homogeneous and struc-tureless bottom of the north coast of Rio de Janeiro is regarded as apromising alternative to mitigate local losses of fishery resources andenhance fish populations (Brotto et al., 2006b).

2.2. Experimental design

Thirty-six prefabricated concrete reef balls® (ca. 1.0 m3; 0.5 ton)were deployed in January 2002 on a flat and homogeneous bottom,9 m deep, and 10 km offshore of the Guaxindiba Beach (21°29′S;

Fig. 1. Geographic location of the north coast of Rio de Janeiro (Southeastern Brazil), where the reef complex was deployed. The spatial arrangement of the reef ball replicates and

sets is also shown.

2 L.N. dos Santos et al. / Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018

41°00′W), northern Rio de Janeiro coast (Fig. 1). Artificial reefs werearranged in sets (following the terminology proposed by Grove et al.(1991)) of three reef balls (ca. 0.5 m distance) and positioned 100 mapart to each other, in a 3×4 reef system configuration that coveredc.a. 60.000 m2 of sea bottom (Fig. 1). The reef balls location wasmarked using global positioning system (GPS). Although the reef ballsets differed in complexity (e.g., presence or absence of lateralcavities) and area covered by epibenthic invertebrates (reef surfacetreated or not with antifouling paint) (see Krohling et al., 2006), suchfactors appeared to have minor effects on the associated fishassemblages (Brotto et al., 2006b), in part, because of fast between-reef homogenization through the attachment of invertebrates(Krohling and Zalmon, 2008).

Artificial reefs were surveyed for associated fishes in November2006 (end of the dry period) and February 2007 (end of the rainyperiod) at four distances from the reefs: 0, 50, 100 and 300 m. At eachsampling period, three reef sets, located on the periphery of the reefsystem were sorted out and surveyed for each of the four distances(following a virtual transect, orthogonal to the reef set, and towardoutside the reef system). Firstly, two SCUBA divers performedstationary visual censuses (following Brotto et al., 2006b), countingall fish found at each reef-distance treatment. Then, bottom gillnets(25×2 m; 40 mm mesh) were installed in each treatment at sunsetand recovered 24 h later. Again, global positioning system (GPS) wasused to obtain an accurate measure of the survey distance in relationto each reef set.

2.3. Data analysis

Non-metric multidimensional scaling (MDS) ordination, performedon log10-transformed abundances and Bray–Curtis dissimilarity mea-sures, were used to detect variation in species composition withsampling treatments (dry×rainy periods; gillnets×visual censuses).The effects of sampling treatments were tested by similarity analysis(ANOSIM)while Bray–Curtis dissimilarities between andwithin groupsof replicates in each sampling treatment were calculated using thesimilarity percentage breakdown procedures (Clarke and Green, 1988;Clarke, 1993). These analyses were performed with the statisticalpackage PRIMER 6.

Fish richness and abundancewereused as descriptors of the changesin fish assemblages with increased reef distance. Permutationalmultivariate analyses of variance (PERMANOVA) were applied formultivariate (the entire community) and univariate (fish richness andabundance) comparisons of fish assemblage attributes among the fourexperimental reef distances. PERMANOVA is a computer program fortesting the simultaneous response of one or more variables to one ormore factors in an ANOVA experimental design on the basis of anydistance measure, using permutation methods (Anderson, 2001). TheBray–Curtis similarity distance was chosen as the basis of allPERMANOVA analysis and data were permutated 4999 times peranalysis, for tests at an α-level of 0.01 (Manly, 1997). Where significantdifferences were found, pair-wise post-hoc comparisons were per-formed under 4999 permutations (see Anderson, 2005 for furtherdetails). Data were log10-transformed for PERMANOVA analyses.

Generalized additive models (GAMs), as available in CANOCO®(V4.5), were also fitted to appraise the response of fish richness andabundance to reef distance. GAMs are an extension of generalizedlinear models that, unlike more conventional regression methods, donot assume a particular functional relationship between the responsevariable and the predictor (distance in our case) (Lepš and Šmilauer,2003). The model complexity of GAMs was chosen by the stepwiseselection procedure using the Akaike information criterion (AIC), alsoavailable in CANOCO® (V4.5). AIC considers not only the goodness offit but also parsimony, penalizing very complexmodels (Burnham andAnderson, 1998).

3. Results

3.1. Species composition and sampling methodology

A total of 30 fish species was recorded at the reef balls using SCUBAand gillnet surveys (Table 1). Data of visual censuses performed inNovember 2006 (dry period) were excluded from the analyses, sinceno fish were recorded because of the very low (b0.5 m) horizontalwater transparency. The tomtate (Haemulon aurolineatum) was by farthe dominant species in the reefs, followed by Genidens genidens,Haemulon steindachneri, Serranus flaviventris and Conodon nobilis,whereas scianids was the most diverse group (10 species). Except forH. steindachneri, there was no other fish species recorded in all threesampling treatments (Table 1). Most species occurred exclusively inone sampling treatment, but a greater number of common specieswas partitioned by gillnet surveys, independently of the samplingperiod, than between visual censuses and gillnet samples performedduring the rainy period.

Table 1

Fish species recorded at artificial reefs in the north coast of Rio de Janeiro, Southeastern

Brazil. Total abundance (pooled values of SCUBA censuses and gillnets samples) for

each species and species occurrence per sampling treatment are shown. RC = rainy

period; visual censuses; RG = rainy period; gillnets; DG = dry period; gillnets.

Fish species Total

abundance

Sampling

RC RG DG

Carcharhinidae

Rhizoprionodon porosus (Poey, 1861) 9 × ×

Dasyatidae

Dasyatis americana (Hildebrand and Schroeder 1928) 1 ×

Clupeidae

Odontognathus mucronatus (Lacépède 1800) 1 ×

Sardinella brasiliensis (Steindanchner, 1889) 1 ×

Ariidae

Bagre marinus (Mitchill, 1815) 5 ×

Genidens genidens (Valeciennes, 1839) 36 × ×

Carangidae

Chloroscombrus chrysurus (Linnaeus, 1766) 4 ×

Holocentridae

Holocentrus adscensionis (Osbeck, 1765) 1 ×

Haemulidae

Conodon nobilis (Linnaeus, 1758) 14 × ×

Haemulon aurolineatum (Cuvier 1829) 408 × ×

Haemulon steindachneri (Jordan and Gilbert, 1882) 32 × × ×

Acanthuridade

Acanthurus chirurgus (Bloch, 1787) 5 ×

Polynemidae

Polydactylus virginicus (Linnaeus, 1758) 1 ×

Centropomidae

Centropomus parallelus (Poey 1860) 1 ×

Serranidae

Diplectrum formosum (Linnaeus, 1766) 1 ×

Mycteroperca acutirostris (Valenciennes, 1828) 3 ×

Serranus flaviventris (Cuvier, 1829) 16 ×

Lutjanidae

Lutjanus jocu (Bloch and Schneider, 1801) 1 ×

Sciaenidae

Ctenosciaena gracilicirrhus (Metzelaar, 1919) 3 × ×

Cynoscion virescens (Cuvier, 1830) 6 ×

Isopisthus parvipinnis (Cuvier, 1830) 6 × ×

Larimus breviceps (Cuvier, 1830) 8 × ×

Menticirrhus americanus (Linnaeus, 1758) 2 ×

Micropogonias furnieri (Desmarest, 1823) 1 ×

Paralonchurus brasiliensis (Steindanchner, 1875) 6 ×

Stellifer brasiliensis (Schultz, 1945) 9 ×

Stellifer rastrifer (Jordan, 1889) 1 ×

Stellifer stellifer (Bloch, 1790) 1 ×

Chaetodontidae

Chaetodon striatus (Linnaeus, 1758) 3 ×

Scorpaenidae

Scorpaena isthmensis Meek and Hildebrand (1928) 6 ×

Total 592 8 14 17

3L.N. dos Santos et al. / Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018

The MDS ordination revealed, however, that fish assemblages,considered as a whole (all 30 species) were similar among mostsampling treatments, except for visual censuses performed in therainy period and 300 m away from the reefs (Fig. 2), in which not evena single fish was recorded. After excluding this treatment from theanalysis, the new MDS ordination showed clear differences in speciescomposition among the other treatments (period and technique;Fig. 2), which were confirmed by global (R=0.39 and 0.87,respectively; P=0.03 for both) and pair-wise (dissimilarity amongtreatments ≥84.8%) ANOSIM tests.

The average similaritywithin gillnet captures during the rainy periodwas 33.6%, with Rhizoprionodon porosus, G. genidens and Chloroscombrus

chrysurus showing the highest cumulative contributions (76.7%),whereas the average similarity was 26.1% for visual censuses performedin the sameperiod,with greatest contributionof S.flaviventris, Scorpaenaisthmensis and H. aurolineatum (94.9%). Mean similarity among gillnetcatches during the dry period was only 18.2%, with Cynoscion virescens,Bagre marinus, Larimus breviceps, Paralonchurus brasiliensis, G. genidensand C. nobilis accounting for 94.9% of the load. Agreeing with previousresults, PERMANOVA also indicated significant differences in fishassemblages (the abundance of all 30 species) among sampling treat-ments (F2, 35=3.71, Pb0.001), inwhich all of themdiffered significantlyfrom each other (PERMANOVA's pair-wise post-hoc tests; Pb0.05).

3.2. Species-specific responses to reef distance

The abundance of all 30 fish species (PERMANOVA; multivariatecomparisons) varied significantly with reef distance (F3, 35=2.40,Pb0.01) and no significant sampling treatment×reef-distance inter-action was found (F6, 35=1.29, P=0.14). Such results indicated thatchanges in fish assemblages with reef distance were not affected bythe period or sampling technique. PERMANOVA's pair-wise, post-hoctests revealed that the fish assemblages at 0 m and 50 m from thereefs were significantly more abundant (Pb0.01 and Pb0.05,respectively) than those located 300 m away.

Fish species richness and abundance changed significantly with reefdistance (PERMANOVA, F3, 35=4.19 and 4.13, respectively; Pb0.01 forboth), but not with sampling treatments (F2, 35≤1.65, P≥0.18 forboth).No significant sampling treatment×reef-distance interactionwasfound for richness (F6, 35=0.73, P=0.71) or abundance (F6, 35=0.97,P=0.48). Richness and abundance changed similarlywith reef distance,butmore noticeably between 0 m and the other distance treatments forabundance (Fig. 3). Richness and abundancewere significantly higher atthe nearest distances (0 m and 50 m) to the reefs than at 300 m apart,with intermediate values at 100 m (PERMANOVA's pair-wise post-hoctests, P≤0.01 for both).

The response curves (GAMs) of fish richness and abundance alsovaried with reef distance (Fig. 4). AIC revealed negative trends for bothabundance and richness, but the first indicated that species richnessdecreased exponentially with reef distance, whereas fish abundancediminished linearly. Response curves (GAMs) of fish abundance withreef distance were selected by AIC for four families and seven species(Fig. 5). Except for Haemulidae, which decreased exponentially, theabundances of Serranidae, Acanthuridae, Chaetodontidae and Scorpae-nidae (the three later groups represented solely byAcanthurus chirurgus,Chaetodon striatus and S. isthmensis, respectively) decreased linearlywith reef distance. AIC selected GAMs for two haemulid species, andalthough negative trendswere detected for both species, the abundanceof H. aurolineatum decreased exponentially with reef distance whereasa linear response was found for H. steindachneri. The abundance ofS. flaviventris also decreased linearly with reef distance, whereas a non-linear and non-monotonic positive trend was selected by AIC forIsopisthus parvipinnis.

4. Discussion

4.1. Species composition and sampling methodology

Fish assemblages associated with our artificial reefs variedsignificantly with sampling technique (visual censuses×gillnets)and seasonal period (dry×rainy). Further, sampling techniquecontributed more than season to data variability in fish surveys.Many authors have argued on the importance of sampling program tocharacterize the fish assemblages associated with artificial reefs(Bohnsack and Sutherland, 1985; Bortone and Kimmel, 1991; Seaman,2000). Here, comparisons with previous studies on artificial reefsperformed on the north coast of Rio de Janeiro provided anopportunity to check the precision of our data.

Zalmon et al. (2002) recorded, through a 2-year gillnet monthlysamplings at concrete and tire reefs located ca. 5 km away from ourexperiment, 45 fish species, of which 14 species were common in bothstudies and 10 species were exclusive of our reefs. Taking into accountthe differences between reef systems (materials, structural complex-ity and size) and sampling effort (24monthly samples of Zalmon et al.

Fig. 2. Non-metric multidimensional scaling (MDS) ordinations of fish composition (all

thirty species; log10-transformed abundances) with sampling treatments, in which

visual censuses data gathered during the rainy period and at 300 m to the reefs are

included in (top-panel) or excluded (bottom-panel) from the analysis. ▲ = gillnets at

dry period;△= gillnets at rainy period;□ = visual censuses at rainy period; numbers

below the symbols correspond to reef distance (m).

Fig. 3. Mean fish richness (grey bars) and abundance (black) recorded at different dis-

tances to the experimental reefs. Vertical lines indicate the standard error.

4 L.N. dos Santos et al. / Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018

(2002) versus two sample periods in our study), the number ofcommon species between these two studies was considerably high(31.1%). Even so, Carcharhinidae, Carangidae, and Triakidae were themost abundant families in Zalmon et al.'s (2002) study in contrast toScianidae and Ariidae in ours, indicating important differences inspecies composition. Such differences are, however, apparently morerelated to changes in fish assemblages throughout time (e.g., there is atime lag of 10 years between these two studies), since Scianidae andAriidae accounted for the dominant families in the more recent Brottoand Zalmon's (2007) gillnetting study, carried out during 2002–2003on the same reef balls we have examined in the present study. Thenumber of common species with Brotto and Zalmon's (2007) studyalso increased to 19, accounting for 41.3% of the forty-six speciesrecorded in that study.

Haemulidae and Serranidae were the most abundant familiesassociatedwith the reef balls, according to the visual census data here.Significant differences between data gathered with visual censusesand gillnets are normally expected, since visual censuses areespecially conceived to record benthic, diurnal and reef-residentfishes whereas gillnets are less selective, also collecting nocturnal,pelagic or transient species (Bortone and Kimmel, 1991). Our results(numerical dominance of Haemulidae and Serranidae) agree, how-ever, with those of Brotto et al. (2006b), who also applied stationaryvisual censuses to describe fish use patterns of the same reef balls weexamined in the present study. Also, nine fish species were commonbetween these two studies, accounting for 42.9% of the 21 speciesrecorded by Brotto et al. (2006a), and with H. aurolineatum as thedominant species common in both survey programs.

Of a total of 24 species recorded trough gillnets, only 29.1% (sevenspecies) were common between dry and rainy seasons. Similartemporal comparisons for visual censuses were, unfortunately,precluded by very low bottom water visibility during the dry period.However, the increased dominance of Scianidae and Ariidae (speciesnot relying on complex habitats or clear waters to find food), thedecreased occurrence of visual predators (Centropomidae, Lutjanidaeand Carangidae), and the presence of small clupeid species exclusivelyduring the dry period might be related to efficiency loss of theartificial reefs in attractingmore demersal and reef-dependent speciestogether with increase in contribution of small-size, transient or moreopportunistically reef-associated species, whenever severe andprolonged periods of turbid waters persist. In summary, the patternof fish species composition and structure found in our surveys aresimilar to those of previous studies on artificial reefs along the northcoast of Rio de Janeiro, indicating that the sampling program wascoherent with the objectives of our work.

4.2. Species-specific responses to reef distance

Overall, fish assemblage abundance and richness were significant-ly higher at distances of 0 m and 50 m from the reefs than thoseassemblages located 300 m apart. Løkkeborg et al. (2002) documen-ted that gillnet catches of gadoids were overall higher within adistance of 110–300 m from oil platforms in the North Sea than atgreater distances, but these results cannot be broaden since theyvaried largely among fish species, platform type and location, andseasons, and because concurrent hydroacoustic and trawling apprai-sals did not corroborate higher fish abundances near the platforms(Soldal et al., 2002). Our results also contrast with those of Jordan etal. (2005), who found a trend of increasing fish abundance andrichness with increased reef distance in South Florida. However,Jordan et al. (2005) tested, in fact, the degree of isolation betweenadjacent reef modules and sets, and the patterns foundwere related toa halo of decreasing density of benthic prey items approaching thereef, as a result of a greater overlap of fish feeding grounds. Therefore,our study is probably the first to demonstrate and model thedecrement of fish abundance and richness with reef distance, at ascale from fifty to hundreds of meters.

While shifts in fish assemblageswith reef distancewere not affectedby sampling technique or period, the responses varied among species.Although not particularly concerning to investigate fish responses toreef distance, Fabi et al. (2002b) recorded greater fish richness anddiversitywithin a 50-m radius of a gas platform in the northern AdriaticSea than in a sand-muddy control site located ca. 6.5 km away, but nosignificant difference was found for fish abundance or biomass. Thesecontrasting results were related to the continuous presence, in theplatform, of many (N=31) but more evenly-distributed obligatory orfacultative reef-dwelling species in opposition to the prevalence, in thecontrol site, of few (N=17) but comparatively more abundant specieswith weak or none relationship with hard substrates. If the findings ofFabi et al. (2002b) are generalized, the extent to which artificial reefsaffect the surrounding fish assemblages would rely on whether theindividual species is more or less dependent on reefs for orientation,shelter, and/or feeding benefits. By applying this hypothesis to ourstudy, the decreasing curves of fish abundance with reef distance(GAMs) will be, thus, steepest for species with high reef-dependency(obligatory reef-dwelling fishes) but more gradual for species lessclosely related to hard substrates (facultative reef-dwelling fishes),whereas no response curve or even an inverted pattern, of increasingfishabundancewith reef distance,will be expected for speciesweakly ornot associated with artificial reefs.

Our experimental reefsweredominatedby tomtates,H. aurolineatum(mostly juveniles recorded through visual censuses during the rainyperiod), which led to the exponential decrease in the total number offish and abundance of haemulids with reef distance. H. aurolineatum

Fig. 4. Response of fish richness and abundance with reef distance (m). Lines are the

generalized additive models selected by the Akaike information criterion.

5L.N. dos Santos et al. / Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018

displayed an aggregate behavior, swimming primarily in large shoals(NN100) very close to the reef balls, but also inhabiting the reef cavitiesor searching for food along reef–sand interfaces (DSB, personal obs.).Few individuals (Nb5) were found 50m away from the reefs and nota single H. aurolineatum was recorded at distances greater than that.Many studies have documented a numerical dominance of juvenileH. aurolineatum at artificial reefs of varied materials and complexity(Godoy and Coutinho, 2002; Cunningham and Saul, 2004; Freitas et al.,2006), in which the intensive use of complex structures by this specieswas largely related to shelter or nursery purposes. Therefore, the highreef-dependency found for H. aurolineatum in this work, accounted forthe occurrence of large shoals closely associated with the reefs and theexponentially decreasing trend of abundance with reef distance, wasprobably related to the use of reef balls as refuge against predation.

This hypothesis is not, however, supported with regard to otherfish species that were abundant enough and showed consistentassociations with reefs to be predicted by GAMs. Even the closecongenerH. steindachneri, despite its greater abundance near the reefsthan at distances 50 or 100 m apart, decreased linearly in numberwith reef distance. A similar pattern was found for S. flaviventris andalso for serranids, in which no fish were recorded 300 m away fromthe reefs. We anticipated that the reef balls also provided importantshelter for H. steindachneri, S. flaviventris and Serranidae, but thesefishes probably have a larger home-range than H. aurolineatum, andtherefore could travel greater distances, on a daily basis, to find food.The importance of home-range on the use of submerged structures by

Haemulidae and Serranidae has been previously noted (Steimle andFigley, 1996; Ferreira et al., 2001; Lindberg et al., 2006).

Despite A. chirurgus and C. striatus being recorded exclusively inthe reefs (distance b1 m), these species were not frequent orabundant enough to allow more profound analyses. Notwithstanding,it is naturally expected that structureless and soft bottoms will be lessprofitable for C. striatus, a closely reef-associated species (Ferreiraet al., 2001), while A. chirurgus, a strict herbivore, may have benefitedfrom the artificial reefs to graze on epilithic macroalgae (Krohlinget al., 2006). The abundance of S. isthmensis decreased linearly withreef distance but apparently in lower rates than the other species thatalso responded negatively to reef distance. Except at 300 m away fromthe reefs, where no individual of this species was detected, there waslittle variation in the abundance of S. isthmensis among the otherdistance treatments. S. isthmensis is a small (often b150 cm totallength) and cryptic species, with many hard spines in the dorsal andanal fins and on preorbital and preoperculum bones. Although thisspecies apparently does not depend on artificial reefs to obtain refugeagainst predators, underestimating counts through visual censusescannot be entirely discarded.

In general, fish richness and abundance diminished with reefdistance and the same trend was observed for most fish species butnot for all. Although very few individuals were recorded (N=6),the number of I. parvipinnis increased with reef distance. Also, noGAM was fitted for many other species, such as G. genidens, C. nobilisand R. porosus, although their abundances were comparable or even

Fig. 5. Relationship of the abundance of fish families and species with reef distance (m). Lines are the generalized additive models selected by the Akaike information criterion.

6 L.N. dos Santos et al. / Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018

greater than to those species negatively related to reef distance. Thus,further research on whether artificial reefs are unprofitable for thosespecies less dependent on clear waters or structurally complexhabitats to obtain refuge or food would be essential.

4.3. Management implications

Managers should evaluate the importance of using varied samplingmethods (visual censuses and gillnets) during at least two differentperiods (dry-turbid water and rainy-clear water) to accurately assessthe fishery resources associatedwith artificial reefs in the north coast ofRio de Janeiro. Further, although fish can be influenced by artificial reefsto a maximum distance of 100 m, the responses are species-specific,with a sharp decrease in the total abundance and richness and theabundance of most species at distances of 50 m or less to the reefs.Therefore, reef setsmustbedeployedwith a spatial configuration so thatthey do not exceed 50 mof distance gap, to enhance their usefulness forconservation or fishery purposes. Also, if artificial reefs will be primarilyused for fishery purposes andmanagers prioritize reef sets of small-size(b5 m2) in a scattered spatial configuration (25–50m apart), thestructures must be spread out, to allow the local inshore fishermen tomaximize their catch by fishing as close as possible to the reefs. Finally,further effort should be employed to investigate how South Atlantic fishreact to reef distancewithin afine scale approach (from fewcentimetersup to 50 m) and whether the size of artificial reefs could influence suchresponses.

Acknowledgements

This work was funded by North Rio de Janeiro State University —

UENF, Brazil (postdoctoral research grant to L.N.S.; code 36193-1),Fundação de Amparo à Pesquisa Carlos Chagas Filho — FAPERJ, Brazil(E26/152.540/2006), and BrazilianAgency for ResearchDevelopment—CNPq (479546/2008-8). We are grateful to MSc. B. P. Masi for divingassistance. [RH]

References

Anderson, M.J., 2001. A newmethod for non-parametric multivariate analysis of variance.Austral Ecol. 26, 32–46.

Anderson, M.J., 2005. PERMANOVA: A FORTRAN Computer Program for PermutationalMultivariate Analysis of Variance. Department of Statistics, University of Auckland,New Zealand.

Baine, M., Side, J., 2003. Habitat modification and manipulation as a management tool.Rev. Fish. Biol. Fish. 13, 187–199.

Barros, F., Underwood, A.J., Lindegarth, M., 2001. The influence of rocky reefs onstructure of benthic macrofauna in nearby soft-sediments. Est. Coast. Shelf Sci. 52,191–199.

Bohnsack, J.A., Sutherland, D.L., 1985. Artificial reef research: a reviewwith recommenda-tions for future priorities. Bull. Mar. Sci. 3, 11–39.

Bortone, S.A., 1998. Resolving the attraction-production dilemma in artificial reefresearch: some yeas and nays. Fisheries 23, 6–10.

Bortone, S.A., Kimmel, J.J., 1991. Environmental assessment and monitoring of artificialhabitats. In: Seaman, W., Sprague, L. (Eds.), Artificial Habitats for Marine andFreshwater Fisheries. Academic Press, San Diego, USA, pp. 117–236.

Brickhill, M.J., Lee, S.Y., Connolly, R.M., 2005. Fishes associated with artificial reefs:attributing changes to attraction or production using novel approaches. J. Fish Biol.67, 53–71.

Brotto, D.S., Zalmon, I.R., 2007. The effect of artificial reef structural complexity andbenthic colonization on gill net fish assemblages. Trop. Oceanography 35, 1–16.

Brotto, D.S., Krohling, W., Zalmon, I.R., 2006a. Fish community modeling agents on anartificial reef on the northern coast of Rio de Janeiro, Brazil. Braz. J. Oceanography54, 205–212.

Brotto, D.S., Krohling, W., Zalmon, I.R., 2006b. Usage patterns of an artificial reef by thefish community on the northern coast of Rio de Janeiro. ICES J. Coast. Res. 39,1277–1281.

Burnham, K.P., Anderson, D.R., 1998. Model Selection and Inference. Springer-Verlag,New York, USA.

Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in communitystructure. Aust. J. Ecol. 18, 117–143.

Clarke, K.R., Green, R.H., 1988. Statistical design and analysis for a ‘biological effects’study. Mar. Ecol. Prog. Ser. 46, 213–226.

Cunningham, P.T.M., Saul, A.C., 2004. Spatial partition of artificial structures by fish atthe surroundings of the conservation unit — Parque Estadual da Ilha Anchieta, SP,Brazil. Braz. Arch. Biol. Technol. 47, 113–120.

Danovaro, R., Gambi, C., Mazzola, A., Mirto, S., 2002. Influence of artificial reefs on thesurrounding infauna: analysis of meiofauna. ICES J. Mar. Sci. 59, 356–362.

Fabi, G., Luccarini, F., Panfili, M., Solustri, C., Spagnolo, A., 2002a. Effects of an artificialreef on the surrounding soft-bottom community (central Adriatic Sea). ICES J. Mar.Sci. 59, 343–349.

Fabi, G., Grati, F., Lucchetti, A., Trovarelli, L., 2002b. Evolution of the fish assemblagearound a gas platform in the northern Adriatic Sea. ICES J. Mar. Sci. 59, 309–315.

Ferreira, C.E.L., Gonçalves, J.E.A., Coutinho, R., 2001. Community structure of fishes andhabitat complexity on a tropical rocky shore. Environ. Biol. Fish 61, 353–369.

Freitas, L.E.L., Feitosa, C.V., Araújo, M.E., 2006. Mangrove oyster (CrassostreaRhizophorae) (Guilding, 1928) farming areas as artificial reefs for fish: a casestudy in the State of Ceará, Brazil. Braz. J. Oceanography 54, 31–39.

Galván, D.E., Parma, A.M., Iribarne, O.O., 2008. Influence of predatory reef fishes on thespatial distribution of Munida gregaria (=M. subrugosa) (Crustacea; Galatheidae)in shallow Patagonian soft bottoms. J. Exp. Mar. Biol. Ecol. 354, 93–100.

Godoy, E.A.S., Coutinho, R., 2002. Can artificial beds of plastic mimics compensate forseasonal absence of natural beds of Sargassum furcatum? ICES J. Mar. Sci. 59,111–115.

Godoy, E.A.S., Almeida, T.C.M., Zalmon, I.R., 2002. Fish assemblages and environmentalvariables on an artificial reef north of Rio de Janeiro, Brazil. ICES J. Mar. Sci. 59,138–143.

Grove, R.S., Sonu, C.J., Nakamura, M., 1991. Design and engineering of manufacturedhabitats for fisheries enhancement. In: Seaman, W., Sprague, L. (Eds.), ArtificialHabitats for Marine and Freshwater Fisheries. Academic Press, San Diego, USA,pp. 109–152.

Jordan, L.K.B., Gilliam, D.S., Spieler, R.E., 2005. Reef fish assemblage structure affected bysmall-scale spacing and size variations of artificial patch reefs. J. Exp. Mar. Biol. Ecol.326, 170–186.

Krohling, W., Zalmon, I.R., 2008. Epibenthic colonization on an artificial reef in astressed environment off the north coast of Rio de Janeiro State, Brazil. Braz. Arch.Biol. Technol. 51, 215–223.

Krohling,W., Brotto, D.S., Zalmon, I.R., 2006. Functional role of fouling community on anartificial reef at the northern coast of Rio de Janeiro state, Brazil. Braz. J.Oceanography 54, 183–191.

Lepš, J., Šmilauer, P., 2003. Multivariate Analysis of Ecological Data Using CANOCO.Cambridge University Press, Cambridge, UK.

Lindberg, W.J., 1997. Can science resolve the attraction-production issue? Fisheries 22,10–13.

Lindberg, W.J., Frazer, T.K., Portier, K.P., Vose, F., Loftin, J., Murie, D., Mason, D.M., Nagy, B.,Hart, M., 2006. Density-dependent habitat selection and performance by a largemobile reef fish. Ecol. Appl. 16, 731–746.

Løkkeborg, S., Odd-Børre, H., Jørgensen, T., Soldal, A.V., 2002. Spatio-temporalvariations in gillnet catch rates in the vicinity of North Sea oil platforms. ICES J.Mar. Sci. 59, 294–299.

Manly, B.F.J., 1997. Randomization, Bootstrap andMonte CarloMethods in Biology (2nd ed).Chapman & Hall, London, UK.

Seaman, W., 2000. Artificial Reef Evaluation with Application to Natural MarineHabitats. CRC Press, Boca Raton, USA.

Seaman, W., Sprague, L., 1991. Artificial habitat practices in aquatic systems. In:Seaman, W., Sprague, L. (Eds.), Artificial Habitats for Marine and FreshwaterFisheries. Academic Press, San Diego, USA, pp. 1–29.

Soldal, A.V., Svellingen, I., Jørgensen, T., Løkkeborg, S., 2002. Rigs-to-reefs in the NorthSea: hydroacoustic quantification of fish in the vicinity of a “semi-cold” platform.ICES J. Mar. Sci. 59, 281–287.

Steimle, F.W., Figley, W., 1996. The importance of artificial reef epifauna to black seabass diets in the Middle Atlantic Bight. North Am. J. Fish. Manage. 16, 433–439.

Valentin, J.L., Monteiro-Ribas, W.L., 1993. Zooplankton community structure on theeast-southeast Brazilian continental shelf (18–23°S latitude). Cont. Shelf Res. 13,407–424.

Zalmon, I.R., Novelli, R., Gomes, M.P., Faria, V.V., 2002. Experimental results of anartificial reef program on the Brazilian coast north of Rio the Janeiro. ICES J. Mar. Sci.59, 83–87.

7L.N. dos Santos et al. / Journal of Experimental Marine Biology and Ecology xxx (2010) xxx–xxx

ARTICLE IN PRESS

Please cite this article as: dos Santos, L.N., et al., Fish responses to increasing distance from artificial reefs on the Southeastern Brazilian Coast,J. Exp. Mar. Biol. Ecol. (2010), doi:10.1016/j.jembe.2010.01.018