Utilization of carbon sources in a northern Brazilian mangrove ecosystem

at SciVerse ScienceDirect

Estuarine, Coastal and Shelf Science 95 (2011) 447e457

Contents lists available

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier .com/locate/ecss

Utilization of carbon sources in a northern Brazilian mangrove ecosystem

Tommaso Giarrizzo a,*, Ralf Schwamborn b, Ulrich Saint-Paul c

a Laboratório de Biologia Pesqueira - Manejo dos Recursos Aquáticos, Universidade Federal do Pará (UFPA), Av. Perimetral 2651, Terra Firme, 66040170 Belém, Pará, Brazilb Zoology Dept., Universidade Federal de Pernambuco, 50730-540 Recife, Brazilc Leibniz Center for Tropical Marine Ecology (ZMT), Fahrenheitstr. 6, 28359 Bremen, Germany

a r t i c l e i n f o

Article history:Received 19 August 2010Accepted 19 October 2011Available online 3 November 2011

Keywords:stable isotopesmangroveestuaryfood websisotope fractionationmixing modelnorthern BrazilCuruçá estuary0�100S47�500W

* Corresponding author.E-mail addresses: [email protected], tgiarrizzo@g

0272-7714/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.ecss.2011.10.018

a b s t r a c t

Carbon and nitrogen stable isotope ratios (13C and 15N) and trophic level (TL) estimates based on stomachcontent analysis and published data were used to assess the contribution of autotrophic sources to 55consumers in an intertidal mangrove creek of the Curuçá estuary, northern Brazil. Primary producersshowed d13C signatures ranging between �29.2 and �19.5& and d15N from 3.0 to 6.3&. The wide rangeof the isotopic composition of carbon of consumers (�28.6 to �17.1&) indicated that different auto-trophic sources are important in the intertidal mangrove food webs. Food web segregation structures theecosystem into three relatively distinct food webs: (i) mangrove food web, where vascular plantscontribute directly or indirectly via POM to the most 13C-depleted consumers (e.g. Ucides cordatus andzooplanktivorous food chains); (ii) algal food web, where benthic algae are eaten directly by consumers(e.g. Uca maracoani, mullets, polychaetes, several fishes); (iii) mixed food web where the consumers usethe carbon from different primary sources (mainly benthivorous fishes). An IsoError mixing model wasused to determine the contributions of primary sources to consumers, based on d13C values. Modeloutputs were very sensitive to the magnitude of trophic isotope fractionation and to the variability in 13Cdata. Nevertheless, the simplification of the system by a priori aggregation of primary producers allowedinterpretable results for several taxa, revealing the segregation into different food webs.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The knowledge of the relative contribution of autotrophicsources to a given food web is fundamental for both theoretical andpractical reasons. First, this information is essential for under-standing the relationship between fauna and flora of differentenvironments. Second, the identification of which autotrophicsources sustain the secondary production is often vital for thedevelopment of conservation priorities and effective managementpolicies of ecosystems (e.g. Connolly et al., 2005).

Knowledge of food sources for consumers are particularlyimportant in mangrove ecosystems, which have long been recog-nized as important nursery grounds for several fish and penaeidshrimp. Although food webs in mangroves have received consid-erable attention, they remain poorly understood.

While early investigations based on stomach content analysis(Odum and Heald, 1972) concluded that mangrove detrital materialconstitutes an important food source for many aquatic organisms,more recent stable isotope studies have questioned these results,

mail.com (T. Giarrizzo).

All rights reserved.

showing that ingested mangrove material is not always assimilatedefficiently by consumers and that other primary producers such asphytoplankton and microphytobenthos are important sources forconsumers (e.g. Newell et al., 1995; Christensen et al., 2001;Bouillon et al., 2002). Less conspicuous primary producers (e.g.microalgae and small filamentous green algae) that have lowbiomass, but high turnover rates and palatability may be moreimportant in the food web than evident large macrophytes(Wiedemeyer and Schwamborn, 1996; Bouillon et al., 2002; Alfaroet al., 2006). More specifically, stable isotope ratios are naturaltracers that can be highly useful to assess the relative contributionsof given primary sources in food webs, especially when detritus,microphagous food chains, or organisms that triturate their food(e.g. many crustaceans) are important (Fry, 2006).

Several previous studies have combined stable isotopes in theinterpretation of stomach contents for selected species (Beaudoinet al., 1999; Schwamborn and Criales, 2000; Hadwen et al., 2007).Also, isotopic analyses of species aggregated into trophic guildshave been used in estuarine food web studies (Martinetto et al.,2006; Winemiller et al., 2007; Armitage and Fourqurean, 2009;Wilson et al., 2009). However, a combination of stomach contentand stable isotope data into an ecosystem-wide food web modelhas not yet been attempted.

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457448

The transfer of carbon and nitrogen through the food webresults in a change in isotope ratios due to fractionation (F), with analteration per trophic level (TL) ranging between�2.7 and 3.9& ford13C and between�2.4 and 7.8& for d15N (Post, 2002; Schwambornet al., 2002; McCutchan et al., 2003; Sweeting et al., 2007; Cautet al., 2009; Travis et al., 2010). Estimates of F typically are basedon paired isotopic measurements of diet and consumer, usuallyunder artificial laboratory conditions (e.g. Buchheister and Latour,2010; Elsdon et al., 2010). There is still no quantitative informa-tion on the possible existence of continuous isotope enrichment ona food web scale.

In contrast to many mangrove forests in the Indo-Pacific (e.g.Bouillon et al., 2004; Abrantes and Sheaves, 2010), Caribbean (e.g.Nagelkerken and van der Velde, 2004), and the semi-arid north-eastern Brazil (Schwamborn et al., 2002), the north Brazilian coastis covered by vast undisturbed mangrove forests without connec-tion to other nearshore macrophytic habitats (e.g. seagrass beds,salt marshes or macroalgae beds). Brazil has one of the largestmangrove areas of the world (Aizpuru et al., 2000). There are stillno data on stable isotope signatures of food webs in northernBrazilian mangrove forests, in spite of their huge biomass andextension, and their evident biogeochemical, ecological, and socio-economic importance.

Themain objective of this study is to determine the contributionof autotrophic sources for a food web in an intertidal creek in theCuruçá mangrove estuary (Northern Brazil) through stable isotope

Fig. 1. Map of the Curuçá estuary, northern Brazil. a: Location of the Curuçá estuary at the mMap of the macrotidal creek, indicating the sampling site.

analysis. Furthermore, this work evaluates the effect of F and TL onthe potential source contributions to consumers estimated by Iso-Error mixing models (Phillips and Gregg, 2001).

2. Material and methods

2.1. Study area

This study was carried out in an intertidal mangrove creeklocated in the inner part of the Curuçá estuary, approximately160 km from Belém, northern Brazil (0�100S, 47�500W) (Fig. 1). TheCuruçá estuary is surrounded by extensive mangrove forests(116 km2) divided by a complex network of branching intertidalcreeks. The mangroves are flooded by semidiurnal tides with anamplitude of up to 5 m. At neap tide, the sampled creek floods anddrains a mangrove surface of approximately 20,000 m2, dominatedby Rhizophora mangle and Avicennia germinans. There are nofreshwater inputs to this creek other than direct rainfall, i.e. there isno measurable input from local rivers or from the Amazon(Giarrizzo and Krumme, 2009). Due to the macrotides, high rainfall(mean annual rainfall is 2526 mm), and high turbidity (Secchidepth usually less than 60 cm), there are no seagrass beds, mac-roalgal beds, or coral reefs in this region (Giarrizzo and Krumme,2008). Therefore, mangroves, benthic and planktonic microalgae,and mangrove macroalgae are the only relevant primary producersin this system.

outh of the Amazon. b: Position of the macrotidal creek within the Curuçá estuary. c:

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457 449

2.2. Sampling design

Potential food sources and consumers were collected in May2004 (rainy season) in a macrotidal mangrove creek. Samplingeffort was concentrated in the intertidal area, with an extension ofapproximately 20,000 m2 (Fig. 1) during one week to reducechanges in environmental conditions. Generally, each biologicalsample was composed of a pool of several specimens of the samespecies of about the same length. Three samples were taken foreach autotrophic source and most consumers.

2.3. Sample collection and treatment

Samples of fresh live (green) and senescent (yellow) mangroveleaves were collected from representative trees of the two mostabundant mangrove species (Rhizophora mangle and Avicenniagerminans). Samples of epiphytic macroalgae (Bostrychia sp., Cat-enella sp. and Enteromorpha sp.) were sampled at low tide byscraping the trunks and roots of R. mangle using scissors andpincers.

Benthic microalgae were collected at low tide by gently scrapingthe visible mats of benthic diatoms on the sediment surface, wherethey formed a conspicuous greenish-brown layer.

Suspended particulate organic matter (POM) was collectedduring ebb tide at the inlet of the intertidal creek from approx.0.5 m depth below thewater surface. A total of approx. 500ml werepassed through a 63 mm screen to remove large particles andzooplankton, and then filtered on pre-combusted Whatmann GF/Fglass fiber filters. Each filter was then freeze-dried and stored ina clean glass vial.

At low tide, samples ofmuddy sediment were taken at the banksof the intertidal creek by using a PVC tube corer (internal diameter:30 mm; length: 70 mm). The upper 1 cm was removed from thecores. Five PVC corer samples were pooled in a clean plastic bag,stored on ice and transported to the laboratory.

Mesozooplankton samples were taken at the inlet of the creekduring ebb tide, using a plankton net (mouth diameter 0.32 m;mesh size 300 mm). Benthic invertebrates were collected atdifferent locations in the intertidal mangrove banks at lowtide, using a PVC tube corer (internal diameter: 10 cm; length:10 cm). Sediment samples were washed through a sieve(0.25 mm mesh) with clean seawater. In the laboratory, gutcontents of polychaetes were removed prior to analysis, to avoidcontamination.

Epibenthos were collected manually from intertidal banks (e.g.fiddler crabs Uca maracoani) and from the mangrove forest bottom(e.g. Ucides cordatus) or picked off mangrove trees (e.g. Littorinaanguilifera) and below submerged fallen branches (e.g. Eurytiumlimosum).

Samples of 31 fish and 3 shrimp species were collected in theintertidal creek using a fyke net set at the creek inlet during hightide and sampled at ebb tide. These key species were chosen dueto their abundance and biomass in the system and their socio-economic importance (Giarrizzo and Krumme, 2007). Due topotential differences in carbon sources for adults of transitoryfish species the sampling was restricted when possible to thosespecies or cohorts that are known to spend substantial periods oftime confined in the estuary. For the fish samples, white muscletissue (w1e2 g) was taken from immediately below the anteriorend of the dorsal fin. For shrimp, a sample of abdomen muscletissue was taken after exoskeleton and digestive tract had beenremoved. Pectoral muscle sample was taken of a single deadspecimen of a spotted sandpiper Actitis macularia and of a fishingbat Noctilio leporinus found entangled in the fyke net.

2.4. Laboratory analyses of samples

Benthic microalgae were filtered onto pre-combusted What-mann GF/F glass-fibre filters, briefly rinsed with distilled water,then freeze-dried and stored in clean glass vials until furtheranalysis. Light microscopic analysis showed that the benthicmicroalgae were dominated by three diatoms species (Navicula sp.,Cylinderutheca closterium and Pleurosigma angulatum) with relativeabundance between 40 and 70%.

For the benthic consumers their gut contents were removedunder a dissecting microscope to avoid the confounding effect ofrecently ingested food in stable isotopic analysis.

All biological samples were cleaned and rinsed with deionizedwater to remove any attached debris, dried in an air-circulatingoven at 60 �C to constant weight, pulverized to a fine homoge-neous powder using ball mill grinder and stored in clean pre-combusted (450 �C, 4 h) glass vials and held in a desiccator prior toisotopic analysis.

The sediment samples were dried at 60 �C to constant weight,sieved through a 0.42 mm mesh to remove fragments of mangrovematerial and mollusk shells, powdered with a pestle and mortarand stored in clean glass vials until further analysis.

2.5. Stable isotopes

To remove all inorganic carbonates and standardize sampletreatment (Jacob et al., 2005), subsamples of all samples for d13Canalysis were treatedwith 0.2ml of 0.1NHCl without rinsing beforetheywere dried at 50 �C for 12 h. Subsamples for d15N analysis werenot acidified because this would result in a 15N enrichment(Pinnegar and Polunin, 1999).

Carbon and nitrogen isotope ratios were determined using a CNanalyzer (ThermoFinnigan 1112 Series e Flash EA) interfaced witha mass spectrometer (ThermoFinnigan MAT Deltaplus). Stableisotope ratios were expressed in standard delta (d) notation (Fry,2006).

Two laboratory working standards (homogenized sedimentfrom a nearby creek) were run for every seven samples. Theanalytical precision of the instruments, assessed by the standarddeviation of 39 measurements of the working standard, was�0.11& and �0.66& for carbon and nitrogen isotope ratios,respectively. A total of 189 samples were successfully analyzed forstable isotopes.

2.6. Data analysis

Prior to analysis, trophic levels (TL) were assigned to eachconsumer type, based on stomach content analysis or publisheddata (TL ¼ 1 for plants, 2 for herbivores, 3 or higher for carnivores).TL values of most fish species were taken from the FishBase data-base (Froese and Pauly, 2007). Stomach contents were analyzed forselected fish species (Table 4). TL was calculated for each species kaccording to the following equation (Cortés, 1999):

TLk ¼ 1þ �SPj � TLj

�

where Pj is the weight contribution of prey category j and TLj is thetrophic level of each prey category j.

2.7. Autotrophic source modeling

Stable isotope mixing models were built to determine thepotential source (primary producers) contributions to any givenmixture (consumers). We have not included d15N in the mixingmodel because of overlapping 15N values of primary sources.

Table 1Mean and standard deviation of carbon and nitrogen stable isotope values ofprimary producers and particulate and sedimentary organic matter collected in theCuruçá estuary, northern Brazil. ACR: acronym; n: number of samples.

Species/taxon ACR n d13C (&) d15N (&)

MangrovesAvicennia germinans (senescent) AviS 3 �27.8 � 0.4 5.8 � 0.1Avicennia germinans (live) AviL 3 �29.0 � 1.1 5.4 � 0.2Rhizophora mangle (senescent) RhiS 3 �28.7 � 1.5 3.5 � 0.6Rhizophora mangle (live) RhiL 3 �28.2 � 1.0 3.8 � 0.2

RhodophytesBostrychia sp. Bost 3 �29.2 � 0.1 4.7 � 0.1Catenella sp. Cate 3 �25.6 � 0.3 5.7 � 0.5

ChlorophytesEnteromorpha sp. Ente 3 �19.5 � 0.3 6.3 � 0.1

Benthic algae Balg 3 �21.5 � 0.8 3.0 � 1.1Particulate organic matter POM 3 �28.2 � 1.7 7.1 � 0.4Sedimentary organic matter SOM 3 �25.4 � 0.2 4.5 � 0.2

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457450

Furthermore, fractionation (Fd15N) between trophic levels can beextremely strong and variable among the consumers and we lackedinformation about the Fd15N value of each consumer. Furthermore,stable isotope mixing models results vary substantially if the cor-rected d15N values are changed even minimally (Connolly et al.,2005). A priori aggregation (sensu Phillips et al., 2005) was usedto simplify the system and to distinguish mangrove vs algae inputs.In this case, we combined d13C values of benthicmacroalgae (exceptBostrychia sp.) with the epiphytic algae Enteromorpha sp. torepresent the “algae” source, and d13C values of a mixture of greenand yellow mangrove leaves to represent the “mangrove” source,by calculating the arithmetic means of these sources. Mixingmodels were built using the IsoError software (version 1.04;Phillips and Gregg, 2001), under the assumption that only twocarbon sources (mangroves and algae) are determining the d13Cvalues of consumers. The inputs for IsoError were the d13C values ofmean, standard deviation, and number of samples measured foreach source and consumer. The output generated by the set ofIsoError equations provides estimate contributions for each source(0e100%), standard errors for these contribution estimates, andapproximate 95% confidence intervals for source contributions,considering error propagation.

Prior to running the IsoError mixing model, d13C of consumerswas corrected (d13Ccorr) for fractionation (F) and TL, using thefollowing equation:

d13Ccorr ¼ d13C��F �

�TL � 1

��:

The models were run twice, assuming (i) no adjustment of d13C,i.e. F ¼ 0 (Peterson and Fry, 1987) and (ii) taxon-specific F values

Table 2Mean and standard deviation of carbon and nitrogen stable isotope values of mes-ozooplankton and benthic invertebrates collected in the Curuçá estuary, northernBrazil. ACR: acronym; n: number of samples; N: number of pooled individuals ineach sample. For consumers, trophic levels (TL) are given.

Species/taxon ACR F TL n (N) d13C (&) d15N (&)

MesozooplanktonChaetognata Chae 0.6 3.2 3 (200) �23.8 � 0.5 10.5 � 0.3Calanoid copepodsPseudodiaptomus marshi Cope 0.6 2 3 (600) �25.5 � 1.2 8.0 � 0.7

Brachyuran zoeae Zoea 0.6 e 2 (175) �28.6 � 0.8 3.8 � 0.9Benthic invertebratesPolychaetesCapitellidae Capi 0.6 2 2 �19.4 � 0.1 7.8Nereidae Nere 0.6 2 2 �22.2 � 3.6 8.5 � 0.4Unid. Polychaetes Poli 0.6 2 2 �20.4 � 0.2 7.9 � 0.1

Nemertea Neme 0.6 3 1 �20.9 9.1

according to consumer type, based on extensive literature reviews(McCutchan et al., 2003; Caut et al., 2009) and selected case studies(Parker et al., 1989; Schwamborn et al., 2002; Yokoyama et al.,2005). Thus, we used F values of 0.6, 1.5, 0.9, and 1.2& per TL forinvertebrates, fish, birds, and predatory mammals, respectively. Forherbivores, a single taxon-specific F value was used to correct d13Cvalues. For carnivores, fractionation occurs at various steps alongthe food chain. Therefore, for carnivores, we used the weightedaverage fractionation Fw to account for the potentially variablefractionation along the food chain sustaining a given predator,based on taxon-specific fractionation values for predator (Fpred) andprey (Fprey):

Fw¼�Fpred�

�1=TLpred�1

��þ�Fprey�

�TLpred�2=TLpred�1

��:

3. Results

3.1. Stable isotope results

3.1.1. Primary producersThe variation of the isotopic composition of primary producers

showed a relatively wide range for d13C (more than 11&) anda narrow range for d15N (approx. 5&) (Fig. 2A). Mean� SD d13C andd15N values formangrove leaves were�28.4� 0.5& and 4.6� 1.1&,respectively (Table 1). Benthic microalgae were isotopically distinctfrom other primary producers and were enriched in d13C(�21.5 � 0.8) and the most depleted in d15N (3.0 � 1.1). Epiphyticmacroalgae showed a high variation in d13C (�24.8 � 4.9&;range: �29.2 to �19.5&) and a little overlap in d15N (5.6 � 0.8&;range: 4.7e6.3&). The rhodophyte Bostrychia sp. was the most 13C-and 15N-depleted macroalga, while the chlorophyte Enteromorphasp. was most enriched for both isotopes. d13C values of Bostrychiasp. overlapped with the range of mangrove leaves and POM(Fig. 2A), and were thus not included in the primary source “algae”for mixing model calculations.

3.1.2. Particulate and sediment organic matterd13C of suspended POM overlapped with mangrove leaves,

showing that mangrove detritus was the dominant carbon sourcefor POM. Suspended POM was 15N-enriched (mean d15N:7.1 � 0.4&) in relation to all primary sources (Table 1; Fig. 2A). d13Cand d15N of sediment showed a narrow range, with valuesbetween �25.5 and �25.2&, and 4.3 and 4.6&, respectively(Fig. 2A). The isotopic composition of sediment was close to that ofmangrove leaves, suggesting that mangroves are themain source ofsedimentary carbon within the system.

3.1.3. Mesozooplankton and benthic infaunaMesozooplankton showed a narrow range of d13C, with averages

of �28.6& for brachyuran crab zoeae, �25.5 for copepods,and �23.8& for chaetognaths (Table 2, Fig. 2B). In contrast, d15N ofzooplankton was highly variable, with 3.8& for brachyuran crabzoeae, 8.0& for copepods, and 10.5& for chaetognaths. Mean d13Cvalues for benthic infauna fell in a narrow range between �22.2&for Nereidae and �19.4& for Capitellidae. d15N values were highestfor Nemertea (9.1&) and lowest for Capitellidae (7.8&).

3.1.4. EpibenthosThe overall range of isotope values for benthic invertebrates

(from �25.3& to �15.3& for of d13C and from 3.5& to 10.8& ford15N) was larger than the range of primary producers (Table 3,Fig. 2C). 50% of all invertebrate consumers exhibited a narrowvariation of d13C values with a range between�22.5& and�20.2&.

Table 3Mean and standard deviation of carbon and nitrogen stable isotope values of epibenthic consumers collected in the Curuçá estuary, northern Brazil. ACR: acronym; TL: trophiclevel; n: number of samples; N: number of pooled individuals in each sample; L: body length.

Species/taxon ACR F TL n (N) L (cm) d13C (&) d15N (&)

AlpheidaeAlpheus sp. Alph 0.6 2.5 3 (2) 4.4 � 0.7 �22.5 � 0.3 8.2 � 0.7

CirripediaBalanidae Bala 0.6 2.0 3 (8) 0.7 � 0.1 �22.2 � 0.3 10 � 0.3

GastropodaThais coronata Tcor 0.6 3.1 3 (5) 3.2 � 0.2 �21.5 � 0.1 7.5 � 0.6Littorina anguilifera Lang 0.6 2.0 3 (6) 1.5 � 0.2 �25.3 � 0.2 3.5 � 0.5

GrapsidaeAratus pisonii Apis 0.6 2.0 3 (4) 1.7 � 0.2 �22.1 � 0.2 6.7 � 0.2Goniopsis cruentata Gcru 0.6 3.0 3 (1) 3.6 � 0.4 �18.9 � 0.8 9.2 � 0.7

HeteropteraVeliidae Veli 0.6 3.2 1 (15) 0.5 � 0.1 �21.6 8.2

IsopodaCirolanidae Ciro 0.6 3.0 3 (5) 1 � 0.2 �20.3 � 0.3 10.8 � 0.4Spaeromidae Spae 0.6 2.0 3 (11) 0.6 � 0.1 �24.9 � 0.3 6.2 � 0.1

OcypodidaeUca maracoani Umar 0.6 2.0 3 (9) 2.6 � 0.3 �17.5 � 0.7 7 � 0.2Ucides cordatus Ucor 0.6 2.0 2 (2) 4.8 � 2.5 �24.8 � 1 5 � 0.1

PalaemonidaeMacrobrachium rosenbergii Mros 0.6 3.1 2 (1) 14.3 � 2.3 �22.7 � 1.9 8.7 � 1Macrobrachium surinamicum Msur 0.6 3.1 3 (5) 7.9 � 2.7 �20.2 � 0.3 10.8 � 0.2

PenaeidaeLitopenaeus schmitti Lsch 0.6 3.1 3 (5) 10.7 � 0.9 �21.8 � 0.6 9.4 � 0.1

PorcellanidaePetrolistes armatus Parm 0.6 2.5 3 (2) 0.8 � 0.1 �22.5 � 0.2 8.3 � 0.2

PortunidaeCallinectes bocourti Cboc 0.6 3.0 3 (3) 5.7 � 0.3 �20.7 � 0.4 9.3 � 0.4

XanthidaeEurytium limosum Elim 0.6 3.2 3 (1) 2.1 � 0.5 �18.5 � 0.9 10 � 0.3

Note: length data for crabs are carapace length; otherwise length is total body length.

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457 451

The variation d15N of epibenthos was considerable, but d15N valueswithin taxa were similar, except Clibanarius sp. (SD: �1.0&).

3.1.5. VertebratesThe wide range of d13C of vertebrates indicated different

mixtures of organic matter sources. d13C values clearly increasedwith TL (Table 4, Fig. 2D). Top predators, such as the benthi-piscivorous jenfish Epinephelus itajara, the piscivorous cutlassfishTrichiurus lepturus, the piscivorous needlefish Strongylura timucu-were the most 13C-enriched fish species, with d13C values of �17.9to �17.4&. On the other hand, planktivores of the familiesEngraulidae and Clupeidae (Cetengraulis edentulus excluded) werethe most 13C-depleted fish species with d13C (�25.0 to �23.2&).The only exception was the filter-feeding anchovie C. edentulus,whichwas highly d13C-enriched (d13C:�17.6� 0.2&) (Fig. 2D). d15Nvalues of vertebrates also clearly increased with TL (Fig. 2D). d15N offish varied between 7.8& for the filter-feeding C. edentulus and12.3& for the zoobenthivorous grunt Genyatremus luteus. Thepiscivorous bat Noctilio leporinus and the benthivorous sandpiper(Actitis macularia) showed relatively low d13C and d15N values(�23.9, 9.4 and �18.0, 9.3&, respectively), as compared to toppredatory fish (Table 4).

3.2. Mixing model results

The calculation of source contributions and confidence intervalsfor source contributions using IsoError with “mangrove” and“algae” (benthic algae and Enteromorpha sp.) as endmember sour-ces yielded a wide range of mangrove contributions, ranging fromzero to 100% mangrove carbon (Fig. 3). The application of taxon-specific F values per TL considerably increased the estimatedmangrove contribution for most taxa (Fig. 3). For example, theaverage mangrove contribution for the white shrimp Litopenaeusschmitti increased from 16% (95% confidence interval: 0e32%

mangrove carbon) to 32% (95% confidence interval: 16e48%mangrove carbon) when increasing the estimated F from zero to0.6& TL�1. For the pisci-benthivorous fish Cynoscion acoupa,average mangrove contribution increased even more, from 15 to58% (95% confidence interval: 50e66% mangrove carbon) whenusing F ¼ 1.2& TL�1 instead of F ¼ 0.

The IsoError model yielded useful ranges of mangrove contri-bution for many taxa (Fig. 3), such as for the fiddler crab Ucamaracoani, where IsoError calculations yielded zero mangrovecarbon (i.e., 100% algae), independently of the applied F value.Other taxa may be considered as based on both sources, such as thecalanoid copepods, where IsoError yielded average mangrovecontributions ranging from 63% (95% confidence interval:23e100%; with F ¼ 0) to 71% (95% confidence interval: 31e100%;with F ¼ 0.6& TL�1). Considering F, copepod tissue was thuspredominantly (71%) composed of mangrove carbon, with a minorcontribution of algal sources.

4. Discussion

4.1. Isotope signatures of sources and consumers

The large d13C range of consumers (�28.6 to �15.3&) suggeststhat different carbon sources are important in the Curuçámangrovefood web, and thus there is a segregation into distinct food webs.Our findings suggest that the ecosystem is segregated into threedistinct groups of food webs (mangrove, algal and mixed foodwebs):

1) In the mangrove food web, vascular plants contribute directlyor indirectly via POM to the most 13C-depleted consumers (e.g.Ucides cordatus) and zooplanktivores. Here, mangrove carbonenters the food web indirectly as POM that is assimilated bycalanoid copepods and is transferred to higher TLs by the

Table 4Mean and standard deviation values of carbon and nitrogen stable isotope values of vertebrate consumers collected in the Curuçá estuary, northern Brazil.

Species/taxon ACR F TL n (N) L (cm) d13C (&) d15N (&)

FishAchiridaeAchirus lineatus Alin 0.9 3.7 3 (3) 18.5 � 5.7 �19.5 � 1.3 9.9 � 0.4

AnablepidaeAnableps anableps* Aana 1.5 2.1 3 (10) 16.1 � 1.3 �18.6 � 0.3 8.3 � 0.5

AriidaeCathorops sp. Cath 1.0 3.5 3 (10) 14 � 1.7 �20.5 � 0.3 11.4 � 0.2Sciades herzbergii* Sher 1.0 3.5 5 (10) 20.4 � 1.4 �21.3 � 0.2 11.5 � 0.1

BatrachoididaeBatrachoides surinamensis* Bsur 1.2 3.7 3 (1) 27.6 � 2.3 �19.7 � 1 9.5 � 0.7

BelonidaeStrongylura timocu Stim 1.5 4.5 3 (2) 41.5 � 3.6 �17.4 � 0.3 11 � 0.5

CarangidaeOligoplites saurus Osau 1.0 3.5 3 (5) 4.5 � 0.7 �21.5 � 0.3 11.7 � 0.3Selene vomer Svom 0.9 3.7 3 (1) 13.5 � 2.3 �20.6 � 2.2 11.8 � 0.2

CentropomidaeCentropomus pectinatus* Cpec 1.2 3.6 3 (3) 17.3 � 1.2 �19.5 � 0.6 10.2 � 0.4

ClupeidaeRhinosardinia amazonica Rama 1.0 3.4 3 (5) 7.1 � 0.6 �24 � 1.3 9.6 � 0.4

EngraulidaeAnchoa hepsetus Ahep 1.0 3.2 3 (7) 4.4 � 0.3 �24.9 � 0.4 11.4 � 0.4Anchovia clupeoides Aclu 1.0 3.4 3 (5) 9.7 � 1.5 �25.0 � 0.4 11.3 � 0.3Anchoviella lepidentostole Alep 1.0 3.1 3 (5) 8.4 � 0.7 �23.2 � 0.4 11.5 � 0.2Centengraulis edentulusy Cede 1.5 2.1 3 (5) 8.7 � 0.7 �17.6 � 0.2 7.8 � 0.2Pterengraulis atherinoides* Path 1.2 3.8 3 (3) 13 � 1.6 �21.3 � 0.7 11.1 � 0.1

GerreidaeDiapterus auratus Daur 1.1 3.0 3 (7) 12.2 � 2 �20.5 � 0.9 11.5 � 0.8

HaemulidaeGenyatremus luteus Glut 1.0 3.5 3 (5) 8.1 � 0.9 �22.1 � 0.5 12.3 � 0.2

HemirhamphidaeHyporhamphus roberti Hrob 1.0 3.5 2 (1) 11.6 � 0.3 �21.4 � 0.1 11.3 � 1

LutjanidaeLutjanus jocu* Ljoc 1.2 3.8 3 (5) 15.6 � 4.4 �19.5 � 1.5 11.7 � 0.6

MugilidaeMugil curemay Mcur 1.5 2.1 3 (4) 16.5 � 2.9 �18.5 � 0.2 8.2 � 0.3Mugil incilisy Minc 1.5 2.0 3 (2) 27.5 � 4.5 �18.6 � 0.5 8.5 � 0.2

OphichthidaeOphichthus parilus Opar 1.0 3.5 2 (1) 41.4 � 12.6 �21.6 � 3.2 9.8 � 0.3

SciaenidaeCynoscion acoupa Caco 1.2 3.7 3 (5) 12.1 � 2.6 �21.8 � 0.3 11.2 � 0.2Stellifer microps Smic 1.0 3.5 3 (4) 12.8 � 4 �21.5 � 0.4 10.7 � 0.6

SerranidaeEpinephelus itajara* Eita 1.2 3.7 2 (1) 25.5 � 1.8 �17.9 � 0.5 10 � 0.7

TetraodontidaeColomesus psittacus* Cpsi 1.0 3.3 9 (10) 16.9 � 6.6 �20.0 � 0.8 9.9 � 0.4Sphoeroides testudineus Stes 1.0 3.4 3 (3) 11 � 2.7 �19 � 1.0 9.9 � 0.6

TrichiuridaeTrichiurus lepturus Tlep 1.5 4.3 3 (2) 47.3 � 14.4 �17.7 � 0.6 11.2 � 0.3

BatsNoctilonidaeNoctilio leporinus Nlep 1.4 4.3 1(1) �23.9 9.4

BirdsScolopacidaeActitis macularia Amac 0.7 3.2 1(1) �18.0 9.3

ACR: acronym; TL: trophic level; n: number of samples; N: number of pooled individuals in each sample; L: body length.* TL calculated according to the equation of Cortés (1999).y mobile species.

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457452

ingestion of these invertebrates by chaetognaths and zoo-planctivorous small pelagic fishes, such as the engraulidsAnchovia clupeoides, Anchoviella lepidentostole, Anchoa hepsetus,and the clupeid Rhinosardinia amazonica. These results haveimportant implications on the trophic role of mangroves intropical estuarine and marine ecosystems since thesuspension-feeding Engraulidae that represent 9% of the totalfish biomass in this aera (Giarrizzo and Krumme, 2007),migrate during some part of their life cycle between the estuaryand coastal regions, thereby transferring nutrients and energyto adjacent ecosystems. Thus, although these species do nothave a direct commercial value in this region, their

participation in the food web is valuable through the conver-sion of planktonic biomass into forage for important commer-cial piscivorous fish species. The ubiquity of mangrove carbonin this system, as evidenced by predominance of mangrovecarbon in SOM and POM, certainly contributes to the formationof the mangrove food web;

2) In the algal food web, benthic algae and green algae Enter-omorpha sp. enter the food web directly trough the grazingimpact of some consumers (e.g. Uca maracoani, mullets, poly-chaetes, several fish species). These algae have potentially highpalatability with fast digestibility and high nutritional value(Montgomery and Gerking, 1980). Benthic algae have been

Fig. 2. Mean values of d13C and d15N of (A) potential primary producers, (B) mesozooplankton and benthic invertebrates, (C) epibenthos and shrimps and (D) vertebrate consumersin an intertidal mangrove creek of the Curuçá estuary, northern Brazil. Boxes in dashed lines indicate the standard deviation of primary producer groups. Refer to Tabless 1e4 foracronyms of primary producers and consumers.

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457 453

shown to play a central role in a Malaysian mangrove food web(Newell et al., 1995). The highly enriched d13C values of fiddlercrab U. maracoani and the hermit crab Clibanarius sp. areconsistent with literature data on these and other depositfeeders which forage on the sediment surface (e.g. Rodelli et al.,1984; Bouillon et al., 2002; Hsieh et al., 2002). Although thed13C values of these consumers fell outside of range of primaryproducers, these should not be considered as isotopic outliers

because, as previously noted by Bouillon et al. (2002), thesedata suggest a strong selectivity for 13C-enriched carbon sour-ces such as microphytobenthos;

3) In the mixed food web, consumers use the carbon fromdifferent primary sources (mainly benthivorous fishes). Thewide range of feeding preferences of epibenthic and benthicinvertebrates justified the intermediate d13C signatures ofbenthophagous and bentho-ichthyophagous fishes, such as

Fig. 3. Percent contribution (mean � 95% confidence interval - C.I.) of mangrove carbon to consumers calculated by the IsoError mixing model, with taxon-specific F values pertrophic level (F > 0, on the right), and without considering trophic fractionation (F ¼ 0, on the left). Primary sources used were mangrove leaves with �28.42 � 1.03& PDB, and“Algae”, a composed value of benthic microalgae and chlorophytes, with �20.63 � 1.19& PDB.

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457454

Sciades herzbergii, Cathorops sp., Colomesus psittacus andBatrachoides surinamensis. Previous studies showed that theirdiet consists of a wide variety of benthic organisms, thus dis-playing opportunistic feeding behavior (Giarrizzo and Saint-Paul, 2008; Giarrizzo et al., 2010).

4.2. Mixing model results

The outputs of any mixing model are likely to be sensitive to theF values used. Nevertheless, few isotope studies have provideda sensitivity analysis to test the effect of F (Schwamborn et al., 2002;Connolly et al., 2005; Inger et al., 2006). However, Connolly et al.(2005) and Inger et al. (2006) assessed the effect of F on thecontributions of each feasible source to consumer nutrition justfocusing on a central value, without considering error propagation.

It has been widely assumed that the trophic shift in d15N isapproximately 3.4& and 1& or less for d13C, (e.g. Vander Zandenand Rasmussen, 2001; McCutchan et al., 2003), but actually notconsidering the TLs of given consumers. However, the trophic shiftcan be highly variable (e.g. taxon, environment, tissue) and smallerrors in estimates of F can result in large errors in estimates ofsource contributions to consumers (McCutchan et al., 2003; Cautet al., 2009). Then, if no specific F values are available, the use ofa unique F value (e.g. zero) can induce a misinterpretation of theresults of the model. Another source of error that may jeopardizethe usefulness of mixing model results is that most studies did notcorrect the isotope values of consumers for F, probably because theylacked information on the TLs of consumers (e.g. Benstead et al.,

2006). In the present study, mixing model outputs were verysensitive to the magnitude of F, especially for higher TLs. Further-more, estimates of source contributions were less precise whenthere was a high variability in consumer d13C (i.e. high variability insource utilization), small sample size, andwhen average d13C valueswere intermediate between both sources. Intra-specific variabilityin source utilization within populations is actually much higherthan represented in our model, especially when considering thatsamples for isotope measurements were composed of many indi-viduals (e.g. up to 600 individuals for zooplankton). Furtherdetailed studies, considering spatial, temporal and behavioralaspects are necessary to understand variability in source utilizationfor selected species. Rather than measuring intra-specific vari-ability for each taxon, the objective of using IsoError in this studywas to assess the precision of the mixing model, considering vari-ability in samples of sources and consumers, thus allowing anecosystem-wide assessment of source utilization in the Curuçámangroves.

A priori aggregation of several sources into two groups should beconsidered only when isotopic signatures of clustered sources arenot significantly different, and sources are ecologically related, sothe combined source group has at least some functional signifi-cance (Phillips et al., 2005). A successful aggregation has clearlybeen achieved in this study, in spite of the exclusion of one algalgroup. The rhodophyte Bostrychia sp. is obligatorily associated withmangrove roots, the inclusion or not of these algae into the group“mangrove” did not affect our model results due to its close isotopicresemblance to mangrove leaf carbon. The group “algae” may also

Table

5Su

mmaryof

d13Cva

lues

(&dev

iation

from

thePD

Bstan

dard)an

dpercentman

grov

eco

ntribution

sto

orga

nic

carbon

(bold,inparen

theses)from

selected

previou

slypublished

studies.

Studyarea

POM

&SO

MZo

oplankton

Ben

thos

Fish

esCom

men

tReferen

ce

Itam

aracáestuary,

Brazil

POM:�2

4.9�

0.4(66±7%

);SO

M:�2

5.1�

0.5(69±9%

)Cop

epod

s:�2

2.0�

0.3

(13to

40%);

Decap

ods:

�22to

�18.0(0

to40

%)

Oyster:

�23.2(53to

62%);

Shrimps:

�19.7to�1

8.2(0%)

Sardine:

�20.2(0

to7%

)Man

grov

eco

ntribution

sto

consu

merscalculatedwith

F¼

0an

dF¼

1.5,

simple

two-sourcemixingeq

uations

Schwam

born

etal.,20

02

Subtropical

Atlan

tic

andCaribbe

anSO

M:�1

5.6

Polych

aetes:

�21.0�

0.1(50%

);Isop

od�1

9.8(38%

);Bluecrab

s�1

6.1(4%)

Red

earsardine�1

4.6(30%

)Yellowfinmojarra:

�12.5�

1.9(18%

)

Man

grov

eco

ntribution

sto

consu

merscalculatedwith

F¼

1,simple

two-source

mixingeq

uations

Kieck

buschet

al.,20

04

CarolineIslands,

Micronésia

Bluecrab

s:�2

3.7to

�22.0

(0to

67%)

Mulle

ts:�1

8.9to

�10.4(0

to7%

);Grouper:�1

5.0to

�14.7(0%)

Ran

geof

man

grov

eco

ntribution

calculated

withIsoS

ourceusingd1

3Can

dd3

4S.

Noco

rrection

sfortrop

hic

fraction

ation.

Ben

stea

det

al.,20

06

Curuçá

estuary,

Brazil

POM:�2

8.2�

1.7(97

%);

SOM:�2

5.4�

0.2(61%

)Cop

epod

s:�2

5.5�

1.2

(63to

71%);

Chae

togn

atha:

�23.8�

0.5(41to

57%)

Shrimps:

�22.7to

�20.2

(0to

43%);

Bluecrab

s:�2

0.7�

0.4

(1to

16%);

Polych

aetes:

�22.2to

�19.4

(0to

27%);

Isop

ods:

�20.3to

�24.9

(0to

63%)

Zoop

lanctivorou

san

chov

y:�2

5.0to

�23.2(33to

87%);

Phytop

lanktivorou

ssardine:

�17.6�

0.2(0%);

Mulle

ts:�1

8.6to

�18.5(0%);

Grouper:�1

7.9�

0.5(0

to7%

)

Man

grov

eco

ntribution

sto

consu

merscalculatedby

IsoE

rror

mixingmod

elwithF¼

0an

dF¼>

0taxo

n-specificFva

lues

Presen

tstudy

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457 455

be considered to be a significant functional unit, since filamentousgreen algae and the microphytobenthos community sampled onthe surface of mud flats have several characteristics in common:small size, fast digestibility, high primary production, and theirspatial closeness. One apparent drawback of this approach is thatthe group “algae” has a 13C signature that is very close to estuarineand marine phytoplankton (“marine” endmember in biogeo-chemical models), another potentially important primary source inthis system (Schwamborn et al., 1999, 2002). However, thisapparent shortcoming of our model is only an artificial problem ofterminology, since the same pennate diatom species that dominatethe microphytobenthos on mud flats (mainly at low tide)contribute to the estuarine phytoplankton communities(Dummermuth, 1997) and to the “phytoneuston” surface layers(Newell et al., 1995).

4.3. Comparison with previous studies

In contrast to the previous studies compiled in Table 5, our studywas carried out in a mangrove forest without a connection to othernearshore macrophytic habitats, such as seagrass beds or saltmarshes. The consumers of the “mangrove food web” detected inthe Curuçá estuary showed to be relatively 13C-depleted, whencomparing with similar consumers in other mangrove ecosystems(e.g. copepods, shrimps). The relative 13C-enrichment of consumersfound in mangroves in the Subtropical Atlantic, Caribbean region,and Micronesia might be related to the availability of other sourcesof organic matter, such as seagrass and marine phytoplankton.Accordingly, the mangrove contribution calculated by mixingmodels was highest for the consumers from the Curuçá and Ita-maracá estuaries (Brazil) and for some consumers (e.g. isopods)associated with the mangrove habitat of the Subtropical Atlanticand Caribbean region (Kieckbusch et al., 2004). These resultssuggest that mangrove carbon is an important source forconsumers in a habitat where other primary sources with highernutritional value are not available or where the spatial habitat useby the consumers (e.g. benthic macroinvertebrate communities) islimited to areas close to the mangrove forest.

Although Benstead et al. (2006) and Kieckbusch et al. (2004)assessed the use of primary sources for several consumers, thezooplankton component was not incorporated in the analysis.Accordingly, they did not discuss the potential role ofzooplankton as a trophic link between primary sources andhigher trophic levels. The zooplankton with its small size, highnutritional value and ubiquitous distribution may have animportant function in the aquatic food web and represents one ofthe most important food items for several benthic (e.g. porcelaincrabs and barnacles) and pelagic consumers (e.g. small pelagicfishes). The relatively high d15N of copepods, that is passed on tothe zoplanktivorous food chains, could be related to the elevatedd15N of POM.

The importance of zooplankton as a food source for the earlyjuveniles of several commercially important fish species wasdemonstrated in Northern Brazil (Krumme and Liang, 2004).

Furthermore, only few earlier mixing models for mangrovefood webs did consider any metabolic fractionation (Schwambornet al., 2002; Kieckbusch et al., 2004), and when doing so,a constant F value was applied to all consumers rather thanconsidering fractionation per trophic levels, although it is quiteobvious that fractionation happens at each step within the foodweb (McCutchan et al., 2003). Furthermore, earlier studies did notconsider the effect of error propagation for the uncertainty asso-ciated to mixing model estimates. Our study showed that forseveral taxa, ignoring the effects of uncertainty and fractionationwould have led to erroneous assessments of carbon sources for

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457456

several consumers. Also, this is the first to analyze the componentsof the variability (carbon source and trophic level) in such data fora mangrove food web.

Further development of mixing models may be necessary formore complex multi-source and multi-isotope analysis. Also, thereis an increasing need for new models and statistical methods forthe interpretation of compound-specific stable isotope measure-ments (e.g. in fatty and amino acids), a new, promising tool for foodweb analysis (McClelland andMontoya, 2002; Schmidt et al., 2006).A further, hitherto poorly considered aspect is the temporal(seasonal an interannual) isotope dynamics of food webs, whichrequires time series sampling for sources and consumers. Yet, ourstudy has shown that in spite of the well-known limitations andpitfalls of stable isotope measurements in bulk samples, our simpletwo-source mixing model, considering fractionation per TL, yieldedinterpretable results for several key taxa. Also, this approachallowed us to show that this apparently homogenous ecosystemclearly supports a strongly segregated food web.

Acknowledgments

We thank F. Arnour, D. Monteiro, A. Jesus, E. Lameira andB. Almeida for their assistance in sample collection and treatment.Thanks to J. Souto, A. Peres, J. Martinelli and P. Almeida for theidentification of invertebrates and benthic algae. Thanks toD. Phillips, J. Benstead and B. Fry for helpful comments on isotopefractionation and mixing models. Special thanks to D. Dasbach andM. Birkicht for training and assistance in sample preparation andmass spectrometry. We also acknowledge the detailed and usefulcomments from the anonymous referees that helped to signifi-cantly improve the manuscript. We thank the Brazilian agencyIBAMA for permit numbers 02001.005636/2004-03. This work wasfunded by the Brazilian-German bilateral MADAM project -Mangrove Dynamics and Management. T. Giarrizzo acknowledgesfinancial support by the Brazilian National Council for Technolog-ical and Scientific Development (CNPq grant 303958/2003-0).

References

Abrantes, K., Sheaves, M., 2010. Use of a d13Ced15N relationship to determineanimals’ trophic positions in a tropical Australian estuarine wetland. AustralEcology 35, 96e103.

Aizpuru, M., Achard, F., Blasco, F., 2000. Global assessment of cover change of themangrove forest using Satellite imagery at medium to high resolution. In: EECResearch Project N� 15017-1999-05 FIED ISP FR. Joint Research Center, Ispra(Italy).

Alfaro, A.C., Thomas, F., Sergent, L., Duxbury, M., 2006. Identification of trophicinteractions within an estuarine food web (northern New Zealand) using fattyacid biomarkers and stable isotopes. Estuarine, Coastal and Shelf Science 10,271e286.

Armitage, A., Fourqurean, J., 2009. Stable isotopes reveal complex changes introphic relationships following nutrient addition in a coastal marine ecosystem.Estuaries and Coasts 32, 1152e1164.

Beaudoin, C.P., Tonn, W.M., Prepas, E.E., Wassenaar, L.I., 1999. Individual speciali-zation and trophic adaptability of northern pike (Esox lucius): an isotope anddietary analysis. Oecologia 120, 386e396.

Benstead, J.P., March, J.G., Fry, B., Ewel, K.C., Pringle, C.M., 2006. Testing isosource:stable isotope analysis of a tropical fishery with diverse organic matter sources.Ecology 87, 326e333.

Bouillon, S., Koedam, N., Raman, A.V., Dehairs, F., 2002. Primary producerssustaining macroinvertebrate communities in intertidal mangrove forests.Oecologia 130, 441e448.

Bouillon, S., Moens, T., Overmeer, I., Koedam, N., Dehairs, F., 2004. Resource utili-zation patterns of epifauna from mangrove forests with contrasting inputs oflocal versus imported organic matter. Marine Ecology Progress Series 278,77e88.

Buchheister, A., Latour, R.J., 2010. Turnover and fractionation of carbon and nitrogenstable isotopes in tissues of a migratory coastal predator, summer flounder(Paralichthys dentatus). Canadian Journal of Fisheries and Aquatic Sciences 67,445e461.

Caut, S., Angulo, E., Courchamp, F., 2009. Variation in discrimination factors (D15Nand D13C): the effect of diet isotopic values and applications for diet recon-struction. Journal of Applied Ecology 46, 443e453.

Christensen, J.T., Sauriau, P.G., Richard, P., Jensen, P.D., 2001. Diet in mangrove snails:preliminary data on gut contents and stable isotope analysis. Journal of Shell-fish Research 20, 423e426.

Connolly, R.M., Hindell, J.S., Gorman, D., 2005. Seagrass and epiphytic algae supportthe nutrition of a fisheries species, Sillago schomburgkii, in adjacent intertidalhabitats. Marine Ecology Progress Series 286, 69e79.

Cortés, E., 1999. Standardized diet compositions and trophic levels of sharks. ICES.Journal of Marine Science 56, 707e717.

Dummermuth, A. 1997. Primärproduktion des Phytoplanktons und des Mikrophy-tobenthos in einer nordbrasilianischen Mangrove (BragancaBragança, Pará).M.Sc. thesis, Bremen University.

Elsdon, T.S., Ayvazian, S., McMahon, K.W., Thorrold, S.R., 2010. Experimental eval-uation of stable isotope fractionation in fish muscle and otoliths. MarineEcology Progress Series 408, 195e205.

Froese, R., Pauly, D. (Eds.), 2007. FishBase WWW Page. http://www.fishbase.org(accessed 30.05.07.).

Fry, B., 2006. Stable Isotope Ecology. Springer. 308 p.Giarrizzo, T., Krumme, U., 2007. Spatial differences and seasonal cyclicity in the

intertidal fish fauna from four mangrove creeks in a salinity zone of the Curuçáestuary, North Brazil. Bulletin of Marine Science 80, 739e754.

Giarrizzo, T., Krumme, U., 2008. Heterogeneity in intertidal fish fauna assemblagesalong the world’s longest mangrove area in northern Brazil. Journal of FishBiology 72, 773e779.

Giarrizzo, T., Krumme, U., 2009. Temporal patterns in the occurrence of selectedtropical fishes in mangrove creeks: implications for the fisheries managementin north Brazil. Brazilian Archives of Biology and Technology 52, 679e688.

Giarrizzo, T., Saint-Paul, U., 2008. Ontogenetic and seasonal shifts in the diet of thepemecou sea catfish Sciades herzbergii (Siluriformes: Ariidae), from a macrotidalmangrove creek in the Curuçá estuary, Northern Brazil. Revista Biologia Tropical56, 861e873.

Giarrizzo, T., Krumme, U., Wosniok, W., 2010. Size-structured migration and feedingpatterns in the banded puffer fish Colomesus psittacus (Tetraodontidae) fromnorth Brazilian mangrove creeks. Marine Ecology Progress Series 419, 157e170.

Hadwen, W.L., Russell, G.L., Arthington, A.L., 2007. Gut content- and stable isotope-derived diets of four commercially and recreationally important fish species intwo intermittently openestuaries.Marine and FreshwaterResearch58, 363e375.

Hsieh, H.L., Chen, C.P., Chen, Y.G., Yang, H.H., 2002. Diversity of benthic organicmatter flows through polychaetes and crabs in a mangrove estuary: d13C andd34S signals. Marine Ecology Progress Series 227, 145e155.

Inger, R., Ruxton, G.D., Newton, J., Colhoun, K., Robinson, J.A., Jackson, A.L.,Bearhop, S., 2006. Temporal and intrapopulation variation in prey choice ofwintering geese determined by stable isotope analysis. Journal of AnimalEcology 75, 1190e1200.

Jacob, U., Mintenbeck, K., Brey, T., Knust, R., Beyer, K., 2005. Stable isotope food webstudies: for standardized sample treatment. Marine Ecology Progress Series287, 251e253.

Kieckbusch, D.K., Koch, M.S., Serafy, J.E., Anderson, W.T., 2004. Trophic linkagesamong primary producers and consumers in fringing mangroves of subtropicallagoons. Bulletin of Marine Science 74, 271e285.

Krumme, U., Liang, T.H., 2004. Tidal-induced changes in a copepod-dominatedzooplankton community in a macrotidal mangrove channel in northern Bra-zil. Zoological Studies 43, 404e414.

Martinetto, P., Teichberg, M., Valiela, I., 2006. Coupling of estuarine benthic andpelagic food webs to land-derived nitrogen sources in Waquoit Bay, Massa-chusetts, USA. Marine Ecology Progress Series 307, 37e48.

McClelland, J.W., Montoya, J.P., 2002. Trophic relationships and the nitrogenisotopic composition of amino acids in plankton. Ecology 83, 2173e2180.

McCutchan Jr., J.H., Lewis Jr., W.M., Kendall, C., McGrath, C.C., 2003. Variation in trophicshift for stable isotope ratios of carbon, nitrogen, and sulphur. Oikos 102, 378e390.

Montgomery, W.L., Gerking, S.D., 1980. Marine macroalgae as foods for fishes: anevaluation of potential food quality. Environmental Biology of Fishes 5,143e153.

Nagelkerken, I., van der Velde, G., 2004. Relative importance of interlinkedmangroves and seagrass beds as feeding habitats for juvenile reef fish ona Caribbean island. Marine Ecology Progress Series 274, 143e151.

Newell, R.I.E., Marshall, N., Sasekumar, A., Chong, V.C., 1995. Relative importance ofbenthic microalgae, phytoplankton, and mangroves as sources of nutrition forpenaeid prawns and other coastal invertebrates from Malaysia. Marine Biology123, 595e606.

Odum, W.E., Heald, E.J., 1972. Trophic analyses of an estuarine mangrove commu-nity. Bulletin of Marine Science 22, 671e738.

Parker, P.L., Anderson, R.K., Lawrence, A.A., 1989. A d13C e d15N tracer study ofnutrition in aquaculture: Pennaeus vannamei in a pond growth system. In:Rundel, P.W., Ehleringr, J.R., Nagy, K.A. (Eds.), Stable Isotopes in EcologicalResearch, pp. 30e77.

Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Annual Review ofEcology and Systematics 18, 293e320.

Phillips, D.L., Gregg, J.W., 2001. Uncertainty in source partitioning using stableisotopes. Oecologia 127, 171e179.

Phillips, D.L., Newsome, S.D., Gregg, J.W., 2005. Combining sources in stable isotopemixing models: alternative methods. Oecologia 144, 520e527.

Pinnegar, J.K., Polunin, N.V.C., 1999. Differential fractionation of d13C and d15Namong fish tissues: implications for the study of trophic interactions. FunctionalEcology 13, 225e231.

Post, D.M., 2002. Using stable isotopes to estimate trophic position: models,methods, and assumptions. Ecology 83, 703e718.

T. Giarrizzo et al. / Estuarine, Coastal and Shelf Science 95 (2011) 447e457 457

Rodelli, M.R., Gearing, J.N., Gearing, P.J., Marshall, N., Sasekumar, A., 1984. Stableisotope ratios as a tracer of mangrove carbon in Malaysian ecosystems. Oeco-logia 61, 326e333.

Schmidt, K., Atkinson, A., Petzke, K.J., Voss, M., Pond, D.W., 2006. Protozoans asa food source for Antarctic krill, Euphausia superba: Complementary insightsfrom stomach content, fatty acids, and stable isotopes. Limnology and Ocean-ography 51, 2409e2427.

Schwamborn, R., Criales, M.M., 2000. Feeding strategy and daily ration of juvenilepink shrimp (Farfantepenaeus duorarum) in a South Florida seagrass bed.Marine Biology 137, 139e147.

Schwamborn, R., Ekau, W., Voss, M., Saint-Paul, U., 1999. Stable isotope compositionof particulate organic matter and zooplankton in northeast Brazilian shelfwaters. Archive of Fishery and Marine Research 47, 201e210.

Schwamborn, R., Ekau, W., Voss, M., Saint-Paul, U., 2002. How important aremangroves as a carbon source for decapod crustacean larvae in a tropicalestuary? Marine Ecology Progress Series 229, 195e205.

Sweeting, C.J., Barry, J.T., Barnes, C., Polunin, N.V.C., Jennings, S., 2007. Effects ofbody size and environment on diet-tissue d13C fractionation in fishes. Journal ofExperimental Marine Biology and Ecology 352, 165e176.

Travis, S.E., Ayvazian, S., McMahon, K.W., Thorrold, S.R., 2010. Experimental eval-uation of stable isotope fractionation in fish muscle and otoliths. MarineEcology Progress Series 408, 195e205.

Vander Zanden, M.J., Rasmussen, J.B., 2001. Variation in d15N and d13C trophicfractionation: implications for aquatic food web studies. Limnology andOceanography 46, 2061e2066.

Wiedemeyer, W.L., Schwamborn, R., 1996. Detritus derived from eelgrass andmacroalgae as potential carbon source for Mytilus edulis in the Kiel Fjord:a preliminary carbon isotopic study. Helgoländer Meeresuntersuchungen 50,409e413.

Wilson, R., Chanton, J., Lewis, G., Nowacek, D., 2009. Combining organic mattersource and relative trophic position determinations to explore trophic struc-ture. Estuaries and Coasts 32, 999e1001.

Winemiller, K.O., Akin, S., Zeug, S.C., 2007. Production sources and food webstructure of a temperate tidal estuary: integration of dietary and stable isotopedata. Marine Ecology Progress Series 343, 63e76.

Yokoyama, H., Tamaki, A., Harada, K., Shimoda, K., Koyama, K., Ishihi, Y., 2005.Variability of diet-tissue isotopic fractionation in estuarine macrobenthos.Marine Ecology Progress Series 296, 115e128.