Journal of Sea Research 86 (2014) 86–96

Contents lists available at ScienceDirect

Journal of Sea Research

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

An abundant small sized fish as keystone species? The effect ofPomatoschistus microps on food webs and its trophic role in twointertidal benthic communities: A modeling approach

Moritz Pockberger a,b,⁎, Florian Kellnreitner a,b, Harald Ahnelt b, Ragnhild Asmus a, Harald Asmus a

a Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Wadden Sea Station Sylt, Hafenstrasse 43, D-25992 List, Germanyb University of Vienna, Department of Theoretical Biology, Althanstrasse 19, A-1090 Vienna, Austria

⁎ Corresponding author at: Alfred Wegener InstituteWadden Sea Station Sylt, Hafenstrasse 43, D-25992 List, G

E-mail address: moritz.pockberger@awi.de (M. Pockbe

1385-1101/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.seares.2013.11.008

a b s t r a c t

a r t i c l e i n f o

Article history:Received 26 December 2012Received in revised form 18 June 2013Accepted 13 November 2013Available online 22 November 2013

Keywords:Pomatoschistus micropsFood webEcological network analysisSeagrassArenicola marinaWadden Sea

Ecological network analysis (ENA) was used to study the effects of Pomatoschistus microps on energy transportthrough the foodweb, its impact on other compartments and its possible role as a keystone species in the trophicwebs of an Arenicola tidal flat ecosystem and a sparse Zostera noltii bed ecosystem. Three ENAmodels were con-structed: (a) model 1 contains data of the original foodweb from prior research in the investigated area by Bairdet al. (2007), (b) an updated model 2 which included biomass and diet data of P. microps from recent sampling,and (c) model 3 simulating a foodweb without P. microps. A comparison of energy transport between the differ-ent models revealed thatmore energy is transported from lower trophic levels up the food chain, in the presenceof P.microps (models 1 and 2) than in its absence (model 3). Calculations of the keystone index (KSi) revealed thehigh overall impact (measured as εi) of this fish species on food webs. In model 1, P. micropswas assigned a lowKSi in the Arenicola flat and in the sparse Z. noltii bed. Calculations in model 2 ranked P. microps first forkeystoneness and εi in both communities, theArenicolaflat and the sparse Z. noltii bed. Taken together, our resultsgive insight into the role of P. micropswhen considering a whole food web and reveal direct and indirect trophicinteractions of this small-sizedfish species. These resultsmight illustrate the impact and importance of abundant,widespread species in food webs and facilitate further investigations.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The common goby Pomatoschistus microps occurs in intertidal nearshore areas, estuaries, lagoons, and brackish waters along the Atlanticcoast from Norway to Morocco, in the North Sea including theWaddenSea, in the Baltic Sea, and in the western part of the Mediterranean Sea(Miller, 1975, 2004;Vorberg andBreckling, 1999). It is a short-lived spe-cies (up to two years) and reaches a maximal total length of 64 mm(Ahnelt, 1991;Miller, 1975, 1986, 2004). As an omnivorous, opportunis-tic fish species (e.g. Polte et al., 2005), it feeds on as many as sixtydifferent prey species including zooplankton e.g. calanoid, cyclopoid co-pepods, veliger larvae, meiofauna e.g. harpacticoid copepods, ostracodsand macrofauna e.g. small crustaceans, polychaetes, bivalvia, insectsand even small teleosts (del Norte-Campos and Temming, 1994;Kellnreitner et al., 2012; Leitão et al., 2006; Pihl, 1985; Zander andHartwig, 1982). Small-sized fish species such as P.microps are an impor-tant prey for larger predators (Behrends, 1985; Dänhardt et al., 2011;Hall et al., 1998; Kubetzki and Garthe, 2003; Moore and Moore, 1976;

for Polar and Marine Research,ermany.rger).

ghts reserved.

Ntiba and Harding, 1993). Although small sized, common gobies arethemselves predators and by their often high abundances possiblyexert a high impact on population growth of their prey (Maes et al.,2003; Pihl, 1982; Zander and Hartwig, 1982). Coll and Libralato(2012) and Cury et al. (2000) point out that small sized fish speciescan have an impact on population size of species in lower andhigher trophic levels due to their middle trophic position within thefood web. According to these publications a “wasp-waist” flow controloccurs when small sized fish species are the dominant component inthe mid-trophic level. Additionally, small-sized fish species are impor-tant competitors for other species feeding on the same resource(Del Norte-Campos, 1995; Del Norte-Campos and Temming, 1994;Magnhagen and Wiederholm, 1982a, 1982b; Pihl, 1985). Therefore, ifabundant, they are potentially important components of ecosystemsand may fulfill the role of a keystone species (Coll and Libralato, 2012).

The concept of keystone species was formulated by Paine (1969)and later adapted by Power et al. (1996)whodefined a keystone speciesas having a large impact on its community, compared to its relativelylow abundance and biomass. Davic (2003) proposed an alternativedefinition: “Any species identified a priori as biomass-dominant withinits occupied functional group should be recognized as a potential key-stone species”. Coll and Libralato (2012) suggested a further definitionof “keystoneness” that considered abundant species and their high

87M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

overall impact on a system. Although the common goby is not commer-cially exploited, it may play an important role in foodwebs and in ener-gy flow within and between ecosystems due to its high abundance(Miller, 1975, 1986, 2004; Polte et al., 2005) and broad diet spectrum(Ehrenberg et al., 2005; Evans, 1984; Herrmann et al., 1998). Whilefeeding activity and diet of P. microps have been intensively studied,the analysis of empirical data by ecological network analysis (ENA) toinvestigate the effects of a small-sized fish species' presence/absenceconsidering a whole food web has not yet been done.

Wehypothesize thatmore energy is transported up the food chain inthe presence of P.microps than in its absence, as this species feeds inten-sively on lower trophic levels and serves as an energy source for highertrophic levels. We tested this for two different benthic communities, anArenicola sandflat, representing the largest community in the studyarea, and a sparse seagrass bed, forming the largest continuous intertid-al plant canopies in the Sylt Rømø Bight (SRB) within the GermanNorth Sea. Given the high abundance and the broad diet spectrum ofP. microps, we expect to find a high overall impact of this species in theinvestigated tidal areas, and accordingly, a high keystone index (KSi).

Specifically, our study aims to investigate the influence of the com-mon goby on energy flow and trophic interactions using empiricaldata analyzed by the routines of ENA. A second aim is to calculate theimpact of P. microps on other taxa within the food web. Finally, thekeystoneness of this abundant and small-sized species in intertidalfood webs is investigated and discussed.

2. Materials and methods

2.1. Study site

The SRB (54° 52′–55° 10′N, 8° 20′–8° 40′E) is a part of the WaddenSea, which extends along the coast of the southeastern North Seareaching from the Netherlands to Denmark. The bight itself is situatedbetween the islands of Sylt (Germany) and Rømø (Denmark) and com-prises an area of 404 km2, including 135 km2 of tidal flats (Asmus andAsmus, 2000). The tidal range inside the bight is up to 2 m and salinityranges between 24 and 32 (Martens and Van Beusekom, 2008). Twocauseways connecting the islands with the mainland inhibit the move-ment of water around the islands and direct exchange with adjacenttidal basins. Between the islands there is only one tidal inlet of 2 kmwidth and 40 m depth (Bayerl and Köster, 1998). A detailed descriptionof the SRB is given in Gätje and Reise (1998). For a detailed descriptionof the Arenicola flat and the sparse Zostera noltii bed used in this studysee Baird et al. (2007).

2.2. Sampling and laboratory procedures

P. microps was sampled monthly from August until December 2006and September 2008 until May 2009, and sampling locations werechosen randomly at every sampling event in an Arenicola flat and asparse Z. noltii bed using a beach seine. The net was 12 m long and2.50 m high, with a mesh size of 6 mm in the wings and 4 mm in thecod end. To obtain biomass and size data, all individuals were counted,weighted (to the nearest mg) and measured using a calliper (TL, totallength, to the nearest 0.1 mm), and bigger catches were subsampledmonthly. For gut content analyses, TL of randomly subsampled fishwas measured and specimens were separated into 10 mm size classes.In total, contents of 180 whole digestive tracts of P. microps (21.7 to58.7 mm TL) were analyzed. Prey taxa were identified to the lowestpossible taxonomic unit.

To standardize our data so that it could be compared to that of Bairdet al. (2007), it was necessary to determine dry and ash weight for allprey items found in investigated P. microps to the nearest mg. Gutcontents were dried for 5 days at 50 °C and subsequently combustedat 500 °C for 5 h. Ash free dry weight (AFDW) was calculated and

converted into carbon units (Heerkloss and Vietinghoff, 1981;Herrmann et al., 1998).

2.3. Data processing

Foodwebmodels constructed by Baird et al. (2007) for the Arenicolaflat and the sparse Z. noltii bed were compared with those constructedfor this studywhich included recentlymeasured biomass and consump-tion of P.microps. Foodwebmodelswere constructedwith 56 living and3 non-living compartments. Names and values for these compartmentsare given in Supplement 1.Models were based on biomass, imports, ex-ports, respiration and the flow of carbon (mg C m−2 day−1) betweenthe compartments using ENA (NETWRK 4.2a). Carbon contents wereobtained from the literature for certain components of the food webe.g. sediment POC (Baird et al., 2007) copepods that were not species-specific (Hickel, 1975), harpacticoid copepods (Hickel, 1975), cyprislarvae (Lucas et al., 1979), Foraminifera and Ostracoda (Widbom,1984). Mean individual weight of every prey component was calculat-ed. Consumption was estimated after Winberg (1956). Respirationrates (Eq. (1)) were computed for different size classes of P. micropsand different temperatures using coefficients after Fonds and Veldhuis(1973).

R ¼ ea�T � k �Wb ð1Þ

where R = respiration, temperature dependent coefficient a = 0.1(Ege and Krogh, 1914), T = temperature in °C, k and b are temperaturedependent coefficients given by Fonds and Veldhuis (1973) andW = the wet weight in g of individuals.

Production (Del Norte-Campos, 1995; Winberg, 1956) was estimat-ed by calculating the increase of weight per day (Eq. (2))

ΔB ¼ 10d−1� logWt2− logWt1ð Þ−1h i

� 100 ð2Þ

where ΔB gives the daily increase of weight in percent, and d gives thetime interval in days (Wt2 the weight after 1 day − Wt1 startingweight). Increase in weight was estimated by converting daily growthrate of 0.25 mm (Del Norte-Campos, 1995) to weight units on thebasis of current empirical length–weight relationships published byKellnreitner (2007) (Eq. (3)). This relationship provides the most con-temporary weight data of P. microps in the SRB to this investigation.

W ¼ 1:2751 � 10−6 � TL3:5037 ð3Þ

where W is the wet weight in g and TL is total length in mm.Consumption rates were calculated in mg C m−2 day−1 (Heerkloss

and Vietinghoff, 1981) taking into account different size classes andabundances of every sampling month (Eq. (4)).

C ¼ 1 � A−1 ΔBþ Rð Þ ð4Þ

where C = consumption, A = the absorption coefficient of 0.8, ΔBgives the daily increase of weight in percent and R the respiration(Herrmann et al., 1998; Winberg, 1956).

To calculate the consumption of each ingested prey item the relativeproportion of the particular prey item on the total consumption wasestimated. Finally, the monthly consumption rates of each ingestedprey species considering different size classes were summed up andthe arithmetic mean was calculated for the total sampling period.Consumption rates of P. microps per month are given in Supplement 2.

2.4. Ecological network analysis

2.4.1. Lindeman spineA Lindeman spine is a concatenated form of a food web,

transforming a complex network of trophic transfers into a linear food

Table 1Compartment names in alphabetical order (bold characters used for network analysis andtheir properties, standard characters used for model 1 and model 2 in Arenicola flat andsparse Zostera noltii (Z. noltii) bed). Values indicate consumption of compartments byPomatoschistus microps in mg C m−2 day−1. + indicates prey item found, − indicatesprey item not found in the diet of P. microps.

Compartment nameproperties

Model 1Arenicolaflat

Model 2Arenicolaflat

Model 1Sparse Z. noltiibed

Model 2Sparse Z. noltiibed

Arenicola marina − 1.87 − 0.66Bivalvia n.d. − 0.22 − −Capitellidae n.d. − − 0.31 −Carcinus maenas − 0.78 − 0.29Crangon crangon − 3.52 − 2.38Gammarus locusta 0.03 − − −Gammarus spec. − 0.76 − 0.56Meiobenthos 0.15 0.62 2.32 1.17Copepoda harpacticoid − + − +Copepoda n.d. − + − +Foraminifera − + − +Ostracoda − + − +

Nephtys sp. − 1.20 − −Nereididae n.d. − 7.48 − −Hediste diversicolor + + + −Nereis sp. − + − −

Oligochaeta − − 0.31 −Pygospio elegans 0.04 − − −Scoloplos armiger 0.12 − 0.15 −Small polychaetes 0.06 2.82 − 0.52Eteone longa + − + −Eteone picta − + − +Malacoceros sp. + − + −Pectinariidae n.d − + − +Polycheata n.d. − + − +Scolelepis foliosa + − + −Sphaerosyllis hystrix − + − +Spio filicornis + − + −

Small crustacea − 10.31 − 12.09Amphipoda n.d. − + − +Cumaceasum n.d. − + − +Decapoda n.d. − + − +Hyalidae n.d. − + − +Isopoda n.d. − + − +Leptomysis arenosa − + − +Microprotus maculatus − + − +Mysidacea n.d. − + − +Nebalidae n.d. − + − +Paramysis arenosa − + − +Praunus sp. − + − +

Zooplankton − 0.59 − 1.09Cypris larvae − + − +Polychaeta larvae − + − +

Table 2Values indicate the feeding activity in % of Pomatoschistus microps at different trophiclevels of the food web and the effective trophic position calculated as the weightedaverage of feeding activity.

Trophic level Model 1Arenicola flatFeedingactivity(%)

Model 2Arenicola flatFeedingactivity(%)

Model 1Sparse Z. noltiibedFeedingactivity(%)

Model 2Sparse Z. noltiibedFeedingactivity(%)

1 − − − −2 − − − −3 81.00 56.50 87.80 67.004 18.40 39.30 12.20 30.105 0.55 4.06 − 2.776 N0.01 0.14 − 0.097 − N0.01 − N0.01Effective trophicposition

3.20 3.48 3.12 3.36

88 M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

chainwith discrete trophic levels. It shows the imports, exports, respira-tion, amount of energy passed to the next level, and the returns to thedetrital pool of each trophic level. Furthermore, the efficiency andamount of energy transferred in a system is quantified (Baird et al.,2004; Wulff et al., 1989). Trophic efficiency is defined as the ratio be-tween input and output of a trophic level in percent (Wulff et al.,1989). Here the Lindeman spine is used to elucidate the possible influ-ence of P. microps on trophic interactions and energy transport in thefood web of the tidal flat. To construct these simplified food chainsthree models for an Arenicola flat and three for a sparse Z. noltii bedwere calculated. In total six models were computed.

I. Model 1 contained the original data of the food web constructedby Baird et al. (2007), sampled in the framework of an interdisci-plinary research project (SWAP 1991–1994). Samples tocalculate consumption rates were collected from April toNovember (Herrmann et al., 1998). To analyze gut contents,individuals of P. microps were sampled in subtidal areas insideand intertidal areas outside the SRB, at the estuaries of the riversWeser, Eider and Tümlauer diked land (Del Norte-Campos, 1995;Del Norte-Campos and Temming, 1994).

II. In model 2, the biomass and energy budget of P. microps waschanged to recent values derived from the 2006–2009 field sam-pling program. Due to the more comprehensive analysis of dietand consumption of P. microps a few additional compartmentswere included in the two subsystems. These compartmentswere zooplankton, Hediste diversicolor, Gammarus spec., smallcrustaceans and Crangon crangon for both communities. Further-more, there were small polychaetes in the Arenicola flat andCarcinus maenas in the sparse Z. noltii bed. Biomass, consumption,respiration, production and egestion of these additional compart-ments were taken from networks of other subsystems of the SRB(Baird et al., 2007) and treated as imports. Input data to thenetworks and to the Lindeman spines are given in Supplement 1.

III. Model 3 represents a foodwebwithout P. microps. This was achievedby removing biomass and trophic interactions of P. microps frommodel 2. It was constructed to illustrate the effect of the absence ofP. microps on energy transport in the food web. Furthermore itreveals the effect of additional compartments, added in model 2, onenergy transport.

2.4.2. Effective trophic positionThe effective trophic position (ETP) is a weighted average of trophic

steps of a living compartment to the primary producers or non-livingcompartments (Wulff et al., 1989). The ETP for P. microps in models 1and 2 was calculated using the program NETWRK 4.2a.

2.4.3. Mixed trophic impactThe mixed trophic impact analysis described by Ulanowicz and

Puccia (1990) gives the impact of biomass change of one compartmenton the biomass of other compartments. For example the net impact (qij)of compartment i on compartment j is calculated (Eq. (5)) (Ulanowiczand Puccia, 1990):

qij ¼ gij− f ij ð5Þ

where gij = the fraction of compartment j's diet comprised by compart-ment i, and fij the fraction of j's total production that is consumed bycompartment i.

The net impact is estimated for the direct and indirect interactionsbetween functional compartments in a food web and displayed in animpactmatrix (Ulanowicz and Puccia, 1990). The effect of one compart-ment on all other compartments in a given food web and the overalleffect (εi), can be calculated from this matrix (Gasalla et al., 2010).Values of the impact matrix, calculated in model 1 and model 2 wereused to construct figures which illustrate the positive and negative

effects of P. microps impacting other compartments and vice versa. Forcalculations, the routines of the program Ecopath with Ecosim version6 (EwE) (Christensen et al., 2008) were used.

89M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

2.4.4. Keystone index (KSi)To estimate the keystoneness of P. microps in model 1 and model 2,

the KSi proposed by Libralato et al. (2006) was calculated using theprogram EwE (Eq. (6)).

KSi ¼ log εi � 1−pið Þ½ � ð6Þ

where εi = overall effect of compartment i in the food web calculatedby the mixed trophic impact analysis, and pi = the contribution of thebiomass of compartment i to the total biomass of the food web.

High values of keystoneness are calculated for compartments with alow biomass proportion but a high εi in a food web. These compart-ments are ranked according to εi (a value of 1 being highest) closer to1 and with highest values of KSi (Gasalla et al., 2010; Libralato et al.,2006). All ENA indices and food web models were calculated mass-balanced.

3. Results

3.1. Trophic structure

Whereas model 1 revealed that mainly meiobenthos and differentpolychaetes were consumed, our analyses of digestive tract contentsused to calculate model 2 showed a broad spectrum of prey itemsfrom various compartments (Table 1). Model 1 indicatedmajor feedingactivities at level 3, whilst model 2 showed a more balanced feedingactivity on various trophic levels (Table 2). The feeding activity ofP. microps was highest at trophic level 3, especially in model 1 (81%

Fig. 1. Lindeman spines for Arenicola flat of model 1: contains data of the food web constructedaccording to field samplings and gut content analysis. Model 3: biomass of P. microps and tropfood web without P. microps.

and 87.8%, respectively for the Arenicola flat and for the sparse Z. noltiibed) and presented decreasing values at higher trophic levels. Inmodel 1, feeding at trophic level 4 was about 62% to 75% smaller thanat trophic level 3 and below 1% at higher trophic levels, while inmodel 2 the decrease in feeding activity from trophic levels 3 to 4 wasabout 17% to 36%. Overall, this led to a higher average trophic level ofP. microps in model 2 for the Arenicola flat (3.48) and for the sparseZ. noltii bed (3.36) (Table 2).

3.2. Energy transport and impact on food webs

Our Lindeman spines show that trophic efficiencies and energytransport through the food chain are higher when P. microps is presentin the food web. Model 1 showed that energy is transported up to the6th and the 5th for the Arenicola flat (Fig. 1) and the sparse Z. noltiibed (Fig. 2), respectively. Beyond level 3, trophic efficiencies do not ex-ceed 1%. Lindeman spines of model 2 show an increased energy trans-port compared to models 1 and 3. In model 2, energy is transportedup to trophic level 7 in the Arenicola flat and 8 in the sparse Z. noltiibed. Values of trophic efficiencies below 1% were only found betweenthe highest trophic levels in the sparse Z. noltii bed. Lindeman spinesconstructed for model 3 (in which P. micropswere artificially removed)gave higher values for energy transport and trophic efficiencies thanmodel 1, but lower values than model 2.

The mixed trophic impact analysis revealed the direct and indirectimpacts of P. microps on other compartments and vice versa for model1 (relatively low abundance of P. microps) and model 2 (relativelyhigh abundance of P. microps).

by Baird et al. (2007). Model 2: biomass and diet of Pomatoschistus microps is exchangedhic interactions between P. microps and other compartments are removed, representing a

90 M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

3.2.1. P. microps as an impacting compartmentAnalyses of model 1 gave lower impact values of P. microps on fewer

compartments compared to model 2 (Fig. 3). Impact values calculatedinmodel 1 for anArenicola flat ranged between−0.18 and 0.06, and be-tween−0.49 and 0.04 for a sparse Z. noltii bed (Fig. 3). Impact values formodel 2 were between −0.79 and 0.53 and −0.79 and 0.56 for anArenicola flat and a sparse Z. noltii bed, respectively (Fig. 3). In theArenicola flat no negative impact stronger than−0.20 could be detectedinmodel 1, while formodel 2, nine such distinct effects could be record-ed (Fig. 3). These effects were mainly related to species directlyconsumed by P. microps, such as C. maenas, C. crangon, Zooplankton,H. diversicolor, Gammarus spec., Arenicola marina and small polychaetes.Indirect negative effectswere due to interspecific (Merlangiusmerlangus)and intraspecific competition.

In the Z. noltii bed, only six compartmentswere impacted negatively,exceeding an impact value of −0.20 in model 1. However, in model 2,ten negatively impacted compartments could be detected. In bothmodels Capitellidae, Oligochaeta and Scoloplos armiger were directlyand negatively impacted by P. microps due to predation. In model 1,meiobenthos also contributed to this group, and in model 2 C. crangon,zooplankton, Gammarus spec, H. diversicolor, small crustaceans andA. marina contributed additionally to this group.

3.2.2. P. microps as an impacted compartmentIn both models 1 and 2 P. microps seems to be positively influenced

via direct and indirect effects. In the Arenicola flat, prey compartments

Fig. 2. Lindeman spines for sparse Zostera noltii (Z. noltii) bed of model 1: contains data of themicrops is exchanged according to field samplings and gut content analysis. Model 3: biomassmoved, representing a food web without P. microps.

such as S. armiger and meiobenthos exerted direct effects strongerthan 0.20 on P. microps in model 1. In model 2, small crustaceans andother prey components like small polychaetes and C. crangon influencedP. microps positively (Fig. 4). These prey groups were also observed ingut content analyses.

In the seagrass bed, meiobenthos (directly used as a food resource)and sediment POC (indirectly) influenced P. microps positively inmodel 1, while in model 2 small crustaceans had the strongest positiveinfluence. M. merlangus revealed high values of direct negative impact(−0.44) on P. microps in model 1 and in model 2 (−0.61), underliningits role as a direct predator on P. microps in the seagrass bed. C. maenasalso was an important predator on P. microps (−0.35 in model 1 and−0.34 in model 2) (Fig. 4). Other strong negative impacts were due tointraspecific competition and to indirect effects exerted by the proso-branch Hydrobia ulvae.

3.3. Keystoneness

Keystone indices for P. micropswere lower in model 1 than inmodel2. In model 1, P. microps is ranked at position 18 in the Arenicola flat(KSi −0.46, εi 0.27) (Table 3) and at position 4 in the sparse Z. noltiibed (KSi −0.08, εi 0.9) (Table 4). Calculations in model 2 rankedP. microps first regarding keystoneness and εi, in the Arenicola flat (KSi0.26, εi 1) (Table 3) and the sparse Z. noltii bed (KSi 0.30, εi 1)(Table 4) indicating P. microps as a keystone species with the highestrelative impact on the food web of all compartments.

food web constructed by Baird et al. (2007). Model 2: biomass and diet of Pomatoschistusof P. microps and trophic interactions between P. microps and other compartments are re-

Fig. 3.Mixed trophic impact of Pomatoschistus microps as an impacting compartment illustrated for model 1 and model 2 in Arenicola flat and sparse Zostera noltii (Z. noltii) bed.

91M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

4. Discussion

4.1. Trophic structure

The addition to the diet composition of P. microps, obtained from di-rect sampling directly in the Arenicola flat and the sparse Z. noltii bed, toour network analyses of model 2, revealed new trophic interactions andallowed a reassessment of its trophic position. The updated trophic levelof the common goby lies between levels 3 and 4, due to its ingestion ofzooplankton, small crustaceans and small polychaetes. This result is inaccordance with studies in other locations on gut contents of P. microps(e.g. Dornboos and Twisk, 1987; Evans, 1984; Leitão et al., 2006; Pihl,1985; Zander and Hartwig, 1982). Differences in feeding activity andconsumption levels between this study and earlier investigationswithin

the SRB (Baird et al., 2004; Del Norte-Campos and Temming, 1994;Herrmann et al., 1998; Kellnreitner et al., 2012) are likely due to the cur-rent focus on intertidal communities, the Arenicola flat and the seagrassbed i.e. only digestive tracts of fish from these habitats were analyzed.

4.2. Energy transport and impact on food webs

Alterations to the food web depending on our different models areillustrated by the Lindeman spines. Specifically, the import of additionalcompartments ingested by P. microps into model 2 resulted in higherenergy transport and higher trophic efficiencies compared to model 1.The import of these additional compartments from other subsystemsrepresents the passive or active migration of certain compartmentsbetween different subsystems. Species in intertidal areas are rarely

Fig. 4. Mixed trophic impact of Pomatoschistus microps as an impacted compartment illustrated for model 1 and model 2 in Arenicola flat and sparse Zostera noltii (Z. noltii) bed.

92 M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

restricted to one subsystem as they are influenced by the tidal cycle andseasonal changes or are swirled up and drifted from the sediment bystrong currents and wave action resulting in lateral energy transfer be-tween habitats (Armonies, 1989, 1994; Baird et al., 2007; Boddeke,1976; Hampel and Cattrijsse, 2004; Hunter and Naylor, 1993). Never-theless in the presence of P. microps (model 2), more carbon istransported in the food chain resulting in the highest values of trophicefficiencies and energy. The increased energy transport might giveevidence for bottom up effects through the presence or absence ofabundant small sized species in Wadden Sea food webs. High valuesfor energy transport in the presence of P. microps are based on its highabundances and intensive feeding on various prey compartments,causing high turnover rates by small benthic fish in intertidal areas(Pihl, 1985; Zander and Hartwig, 1982).

Mixed trophic impact analysis was used to investigate the impact ofP. microps on other compartments in the food web. This analysis pointsout direct and especially indirect impacts, which are not readily apparentfrom empirical data sets on diet composition alone (Ulanowicz andPuccia, 1990). Our results of mixed trophic impact analysis indicate littleinfluence of P. microps on higher trophic levels via direct bottom up ef-fects. Here, our results show that P. microps affects only a few predatorcompartments positively. Although P. microps is described as an impor-tant food and energy resource to other species (Behrends, 1985;Dornboos and Twisk, 1987; Healy, 1972; Pihl, 1982), only few speciesfeeding on P. microps are reported for the SRB (Kellnreitner et al., 2012;Scheiffarth and Nehls, 1997). The lack of positive impacts on predatorspecies in the SRB may be based on the high availability of C. crangonas an alternative prey to the same predators. The strongest negative

Table 3Compartments of the Arenicola flat food web ranked according to keystone index (KSi) and overall effect (εi) in model 1 and model 2.

Rank Model 1Arenicola flatCompartment name

KSi εi Rank Model 2Arenicola flatCompartment name

KSi εi

1 Phytoplankton 0.11 1.00 1 Pomatoschistus microps 0.26 1.002 Nephtys sp. N0.1 0.83 2 Phytoplankton 0.11 0.723 Small polychaetes N0.1 0.69 3 Microphytobenthos N0.1 0.604 Pomatoschistus minutus −0.05 0.68 4 Nephtys sp. N0.1 0.615 Other birds −0.06 0.66 5 Small polychaetes N0.1 0.546 Somateria mollissima −0.12 0.58 6 Hediste diversicolor N0.1 0.547 Macoma baltica −0.17 0.56 7 Gammarus spec. −0.08 0.478 Crangon crangon −0.19 0.50 8 Somateria mollissima −0.14 0.409 Microphytobenthos −0.21 0.48 9 Macoma baltica −0.16 0.4210 Arenicola marina −0.22 0.56 10 Arenicola marina −0.17 0.4411 Meiobenthos −0.22 0.47 11 Cerastoderma sp. −0.20 0.6512 Calidris alpina −0.23 0.44 12 Calidris alpina −0.25 0.3113 Cerastoderma sp. −0.25 0.92 13 Small crustaceans −0.28 0.3114 Sediment bacteria −0.32 0.38 14 Sediment bacteria −0.29 0.2915 Scoloplos amiger −0.35 0.36 15 Meiobenthos −0.29 0.2916 Pluvialis apricaria −0.42 0.29 16 Macroalgae −0.33 0.2617 Haematopus ostralegus −0.43 0.29 17 Crangon crangon −0.40 0.2218 Pomatoschistus microps −0.46 0.27 18 Pluvialis apricaria −0.42 0.2119 Larus canus −0.68 0.16 19 Scoloplos amiger −0.44 0.2120 Limosa lapponica −0.71 0.15 20 Mya arenaria −0.54 0.1621 Chroicocephalus ridibundus −0.76 0.13 21 Haematopus ostralegus −0.68 0.1222 Pygospio elegans −0.82 0.12 22 Larus canus −0.68 0.1223 Calidris canutus −0.84 0.11 23 Carcinus maenas −0.74 0.1024 Lanice conchilega −0.89 0.10 24 Calidris canutus −0.85 0.0825 Numenius arquata −0.89 0.10 25 Limosa lapponica −0.92 0.0726 Hydrobia ulvae −1.00 0.08 26 Pomatoschistus minutus −0.92 0.0727 Zooplankton −1.02 0.07 27 Chroicocephalus ridibundus −0.94 0.0628 Mya arenaria −1.04 0.07 28 Zooplankton −0.95 0.0629 Merlangius merlangus −1.09 0.06 29 Lanice conchilega −0.97 0.0630 Freeliving bacteria −1.83 N0.1 30 Corophium volutator −0.99 0.0631 Capitellidae −2.28 N0.1 31 Hydrobia ulvae −0.99 0.0632 Pleuronectes platessa −3.38 N0.1 32 Other birds −1.00 0.06

33 Littorina littorea −1.03 0.0534 Pygospio elegans −1.03 0.0535 Numenius arquata −1.14 N0.136 Freeliving bacteria −1.86 N0.137 Merlangius merlangus −2.23 N0.138 Capitellidae −2.32 N0.139 Pleuronectes platessa −3.61 N0.1

93M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

impact on P. microps was exerted by the predators M. merlangus andC. maenas in the seagrass bed.M. merlangus is well known as a predatorof P. microps (Hamerlynck and Hostens, 1994; Herrmann et al., 1998).However, P. microps is rarely found in gut contents of C. maenas(Ropes, 1968), probably due to its feeding technique of shredding thefish to small pieces before ingestion.

High positive impacts by prey compartments of P. microps (e.g. smallcrustaceans and small polychaetes) reveal direct bottom up effects inour studies' food web. Meiobenthos is ingested directly, and indirect bot-tom up effects derive from sediment bacteria and microphytobenthosthat serve as energy resources of compartments ingested by P.microps. In-terestingly, the negative impacts of A. marina, Nephtys spec. and H. ulvae,on P. micropswere found to be quite strong. These components are rarelyfound in digestive tracts of P. microps and might cause indirect negativeeffects via competition for food with prey species of P. microps. OurLindeman spines confirm increasing energy transport in a food web inthe presence of P. microps, although earlier publications assume low pre-dation pressure on abundant benthic fish in the SRB (Kellnreitner et al.,2012; Scheiffarth and Nehls, 1997). Energymight be transported via indi-rect pathways. Mixed trophic impact analysis revealed that some energyis passed to higher trophic levels via direct ingestion. However, the mainimpacts on P. microps are through indirect effects, such as intraspecificcompetition for food. For example, according to Christensen et al.(2008), a compartment can affect itself negatively due to competitionfor resources. The high intraspecific competition found in our study

might be based on the relatively low predation impact on P. microps inthe SRB.

We derived our results by analyzing the studied food web with theENA method. However, Baird et al. (2009) identify limitations of sucha food webmodel. First themodel should represent an energetic steadystate condition, which is rarely possible in many studies due toincomplete empirical data for estimations of energy budgets. Secondspecies mainly of lower trophic levels (e.g. bacteria, phytoplankton,microphytobenthos) are assembled into functional compartments dueto difficulties of taxonomical identification and measuring energeticsof a single species. Yet, in spite of the above mentioned limitations,ENA results were derived from highly articulated networks of the SRBecosystem based on reliable empirical data (Baird et al., 2009).

Calculations of food web models represent interactions betweenspecies or compartments, which can be uncovered using ENA consider-ing all dominant compartments of a foodweb, revealing the strength ofdirect and especially indirect interactions on a large scale (Heymanset al., 2007; Libralato et al., 2006). Also, certain system indices calculatedby ENA can be used to compare ecosystems of different sizes and geo-graphic locations (Christian et al., 2005; Monaco and Ulanowicz,1997). These indices are also used to assess the status or health of anecosystem (Mageau et al., 1995). A purely experimental approach ismostly restricted to a few species in a food web and thus interactionsbetween species are reduced to those that are selected for the specialexperiment (Bender et al., 1984; Wootton et al., 1996). An experiment

Table 4Compartments of the sparse Zostera noltii (Z. noltii) bed food web ranked according to keystone index (KSi) and overall effect (εi) in model 1 and model 2.

Rank Model 1Sparse Z. noltii bedCompartment name

KSi εi Rank Model 2Sparse Z. noltii bedCompartment name

KSi εi

1 Microphytobenthos N0.1 1.00 1 Pomatoschistus microps 0.30 1.002 Merlangius merlangus N0.1 0.99 2 Merlangius merlangus 0.27 0.933 Carcinus maenas −0.05 0.98 3 Carcinus maenas 0.14 0.684 Pomatoschistus microps −0.08 0.90 4 Small crustaceans 0.06 0.605 Other birds −0.09 0.89 5 Phytoplankton N0.1 0.556 Cerastoderma sp. −0.09 0.99 6 Hediste diversicolor N0.1 0.557 Phytoplankton −0.10 0.87 7 Microphytobenthos N0.1 0.488 Meiobenthos −0.11 0.86 8 Hydrobia ulvae −0.07 0.599 Calidris alpina −0.11 0.84 9 Gammarus spec. −0.08 0.4210 Arenicola marina −0.17 0.90 10 Cerastoderma sp. −0.09 0.4411 Crangon crangon −0.20 0.68 11 Calidris alpina −0.12 0.3812 Anas penelope −0.22 0.66 12 Arenicola marina −0.14 0.4313 Hydrobia ulvae −0.23 0.96 13 Macrophytes −0.19 0.4214 Macrophytes −0.28 0.77 14 Meiobenthos −0.23 0.3015 Sediment bacteria −0.32 0.53 15 Anas penelope −0.29 0.2616 Calidris canutus −0.40 0.43 16 Crangon crangon −0.32 0.2417 Macoma baltica −0.41 0.43 17 Calidris canutus −0.36 0.2218 Pluvialis apricaria −0.53 0.32 18 Sediment bacteria −0.36 0.2219 Phyllodocidae −0.59 0.28 19 Macoma baltica −0.39 0.2120 Numenius arquata −0.66 0.24 20 Pluvialis apricaria −0.50 0.1621 Chroicocephalus ridibundus −0.68 0.23 21 Mya arenaria −0.51 0.1622 Capitellidae −0.68 0.23 22 Phyllodocidae −0.51 0.1523 Larus canus −0.71 0.21 23 Other birds −0.54 0.1424 Scoloplos amiger −0.72 0.21 24 Larus canus −0.54 0.1425 Oligochaeta −0.82 0.17 25 Small polychaetes −0.63 0.1226 Branta bernicla −0.99 0.11 26 Numenius arquata −0.68 0.1027 Mya arenaria −1.03 0.10 27 Capitellidae −0.70 0.1028 Pomatoschistus minutus −1.11 0.08 28 Scoloplos amiger −0.72 0.1029 Platichthys flesus −1.62 N0.1 29 Chroicocephalus ridibundus −0.73 0.0930 Littorina littorea −1.86 N0.1 30 Corophium volutator −0.88 0.0731 Freeliving bacteria −2.04 N0.1 31 Zooplankton −0.95 0.0632 Pleuronectes platessa −3.16 N0.1 32 Oligochaeta −0.96 0.06

33 Pomatoschistus minutus −0.97 0.0534 Littorina littorea −1.03 N0.135 Branta bernicla −1.06 N0.136 Platichthys flesus −1.62 N0.137 Freeliving bacteria −1.97 N0.138 Pleuronectes platessa −3.17 N0.1

94 M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

on artificial communities might reveal direct and indirect interactions,but the conclusions on the system level might not be relevant, due tothe exclusion of species (Wootton, 1994). Following these argumentsthe ENA approach provides a useful tool to give insights of direct and in-direct interactions of P. microps with other compartments in intertidalsubsystems and provides a promising technique to integrate empiricaland experimental investigations.

4.3. Keystoneness

The results of our investigation show that, P. microps is an importantkeystone species with a high overall effect on other compartments inArenicola flat and sparse Z. noltii bed habitats. Various investigationshave focused on the consumption and impact of epibenthic macrofaunaon communities. For instance, Reise (1979) andGee et al. (1985) carriedout cage experiments, ex- and including common gobies, with bothstudies observing no significant alteration of prey compartmentsthrough ingestion of P. microps in enclosed areas. Gee et al. (1985) sug-gest problems employing cage experiments as a potential explanationfor this result: P. microps showed a different diet composition than indi-viduals sampled from the outside of cages, due to the exclusion of pos-sible prey items. Also, food competitors of P. microps, for example smallC. maenas and C. crangon cannot be excluded from the cages. Finally thetidal feeding rhythm of individuals might have been disturbed due toenclosure in tidal areas. Ehrenberg et al. (2005) estimate consumptionof gobies to represent 15–30% of the total benthos production and up

to 50–100% of macrofaunal production (excluding bivalvia) in Balticsubtidal systems. High consumption rates are also found in other areas(Pihl, 1985; Zander and Hartwig, 1982) supporting the high overalleffect calculated in this investigation.

Can P. microps be defined as a keystone species for the investigatedintertidal areas? Keystoneness in a food web is not clearly predictable,although compartments of lower trophic levels appear to be importantfor energy transport in shallow coastal ecosystems (Libralato et al.,2006). Due to its high abundance and accordingly to high biomass thekeystoneness of P. microps seems contradictory to the original definitionof a keystone species proposed by Paine (1969) and Power et al. (1996).Coll and Libralato (2012) point out high keystone indices for abundantsmall pelagic fish due to their high overall impact, suggesting potentialproblems in the weighting of biomass proportion in the keystone indexof Libralato et al. (2006), but nevertheless stating that the index remainsa useful tool to estimate the overall effect. Two types of keystone speciesmight be distinguished using the overall effect (Coll and Libralato,2012): key compartments with high overall effect but low biomass(Paine, 1969; Power et al., 1996) and compartments with high overalleffect due to their high biomass (Coll and Libralato, 2012; Pirainoet al., 2002). Our results suggest that P. microps might be defined as akey species in both investigated communities based on our results forkeystoneness (KSi values) and its high overall effect due to high bio-mass. This conclusion should, however, be regarded with care as thedefinition of a keystone species has become broader (Davic, 2003),and experiments removing P. microps from a system on a large scale

95M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

are missing and seem hard to realize. Nevertheless, our results revealthe high overall impact of the common goby in the food web of theWadden Sea, resulting in a high-energy uptake from lower trophiclevels and an increased transport up the food chain.

5. Conclusion

This ENA investigation shows that energy transport to higher trophiclevels is distinctly enhanced in the presence of a dense population ofP. microps. The mixed trophic impact analysis reveals complex trophicinteractions, particularly, a large impact of P. microps on lower trophiclevels and vice versa. Especially, a dense population of P. microps exertsdistinct intraspecific competition on itself. Finally, our calculations sug-gest that the keystone role of P. microps in shallowwater habitats mightvary according to the local density of its population, resulting in highconsumption and hence a high overall impact. Being commonly highlyabundant, it is to be expected that the species has such a role all overits geographic range in other ecosystems. Therefore it would be very in-teresting to study the importance of P. microps and other abundantsmall-sized omnivore species in food web models and include theseresults in coastal management plans.

Acknowledgments

The first two authors were partly financed by the Short ResearchStays Abroad program (KWA) by the Research Services and Internation-al Relations (FSIB) of the University of Vienna. We are grateful to AlfredResch, ReimerMagens and PaulWalter for technical advice and their ex-perience in designing sampling gear that made sampling proceduresmuch easier. Wewish to thank Dan Baird for advice in network analysisand Lisa Shama for improving the manuscript. We are much obliged tothe anonymus reviewers revising the manuscript.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.seares.2013.11.008.

References

Ahnelt, H., 1991. Some rare fishes from the Western Mediterranean Sea. Ann. Naturhist.Mus. Wien. B 92, 49–58.

Armonies, W., 1989. Meiofaunal emergence from intertidal sediment measured in thefield: significant contribution to nocturnal planktonic biomass in shallow waters.Helgol. Mar. Res. 43, 29–43.

Armonies, W., 1994. Drifting meio- and macrobenthic invertebrates on tidal flats inKönigshafen: a review. Helgol. Mar. Res. 48, 299–320.

Asmus, H., Asmus, R., 2000. Material exchange and food web of seagrass beds in the Sylt-Rømø Bight: how significant are community changes at the ecosystem level? Helgol.Mar. Res. 54, 137–150.

Baird, D., Asmus, H., Asmus, R., 2004. Energy flow of a boreal intertidal ecosystem the Sylt-Rømø Bight. Mar. Ecol. Prog. Ser. 279, 45–61.

Baird, D., Asmus, H., Asmus, R., 2007. Trophic dynamics of eight intertidal communities of theSylt-Rømø Bight ecosystem northern Wadden Sea. Mar. Ecol. Prog. Ser. 351, 25–41.

Baird, D., Fath, B.D., Ulanowicz, R.E., Asmus, H., Asmus, R., 2009. On the consequences ofaggregation and balancing of networks on system properties derived from ecologicalnetwork analysis. Ecol. Model. 220, 3465–3471.

Bayerl, K., Köster, R., 1998. Morphogenesis of the List tidal basin. In: Gätje, C., Reise, K. (Eds.),Ökosystem Wattenmeer-Austausch-Transport- und StoffumwandlungsprozesseSpringer, Berlin-Heidelberg, pp. 26–29.

Behrends, G., 1985. Food selection of common seals (Phoca vitulina L) in the coastalshallows of Schleswig Holstein (W Germany). Z. Jagdwiss. 31, 3–14.

Bender, E.A., Case, T.J., Gilpin,M.E., 1984. Perturbation experiments in community ecology:theory and practice. Ecology 65, 1–13.

Boddeke, R., 1976. The seasonal migration of the brown shrimp Crangon crangon. Neth.J. Sea Res. 10, 103–130.

Christian, R.R., Baird, D., Luczkovich, J., Johnson, J.C., Scharler, U.M., Ulanowicz, R.E., 2005.Role of network analysis in comparative ecosystems ecology of estuaries. In:Belgrano, A., Scharler, U.M., Dunne, J., Ulanowicz, R.E. (Eds.), Aquatic Food Webs:An Ecosystem Approach. Oxford University Press, pp. 25–40.

Christensen, V., Walters, C.J., Pauly, D., Forrest, R., 2008. Ecopath with Ecosim version 6User Guide. Lensfest Ocean Futures Proj. 2008.

Coll,M., Libralato, S., 2012. Contributions of foodwebmodeling to the ecosystemapproachto marine resource management in the Mediterranean Sea. Fish Fish. 13, 60–88.

Cury, P., Bakun, A., Crawford, R.J.M., Jarre, A., Quiñones, R.A., Shannon, L.J., Verheye, H.M.,2000. Small pelagics in upwelling systems: patterns of interaction and structuralchanges in “wasp-waist” ecosystems. ICES J. Mar. Sci. 57, 603–618.

Dänhardt, A., Fresemann, T., Becker, P.H., 2011. To eat or to feed? Prey utilization ofcommon terns Sterna hirundo in the Wadden Sea. J. Ornithol. 152, 347–357.

Davic, R.D., 2003. Linking keystone species and functional groups: a new operationaldefinition of the keystone species concept. Cons Ecol 7 (response 11).

Del Norte-Campos, A.G.C., Temming, A., 1994. Daily activity feeding and rations in gobiesand brown shrimp in the northern Wadden Sea. Mar. Ecol. Prog. Ser. 115, 41–53.

Del Norte-Campos, A.G.C., 1995. Ecological studies on the coexistence of the brown shrimpCrangon crangon L and the gobies PomatoschistusmicropsKrøyer and P.minutus Pallasin shallow areas of the GermanWadden Sea. (PhD Thesis) University of Hamburg.

Dornboos, G., Twisk, F., 1987. Density growth and annual food consumption of gobiid fishin the saline lake Grevelingen the Netherlands. Neth. J. Sea Res. 21, 45–74.

Ege, R., Krogh, A., 1914. On the relation between temperature and the respiratoryexchange in fishes. Int. Rev. Gesamten Hydrobiol. Hydrogr. 1, 48–55.

Ehrenberg, Z., Hansson, S., Elmgren, R., 2005. Sublittoral abundance and food consumptionof Baltic gobies. J. Fish Biol. 67, 1083–1093.

Evans, S., 1984. Energy budgets and predation impact of dominant epibenthic carnivoreson a shallow soft bottom community at the Swedish west coast. Estuar. Coast. ShelfSci. 18, 651–672.

Fonds, M., Veldhuis, C., 1973. The oxygen consumption of four Pomatoschistus species(Pisces Gobiidae) in relation to water temperature. Neth. J. Sea Res. 7, 376–386.

Gasalla, M., Rodrigues, A.R., Postuma, F.A., 2010. The trophic role of squid Loligo plei as akeystone species in the South Brazil Bight ecosystem. ICES J. Mar. Sci. 67, 1413–1424.

Gätje, C., Reise, K. (Eds.), 1998. Ökosystem Wattenmeer-Austausch-Transport- undStoffumwandlungsprozesse. Springer-Verlag, Berlin-Heidelberg, p. 570.

Gee, J.M., Warwick, R.M., Davey, J.T., George, C.L., 1985. Field experiments on the role ofepibenthic predators in determining densities in an estuarine mudflat. Estuar.Coast. Shelf Sci. 21, 429–448.

Hall, A.J., Watkins, J., Hammond, P.S., 1998. Seasonal variation in the diet of harbor seals inthe south-western North Sea. Mar. Ecol. Prog. Ser. 170, 269–281.

Hamerlynck, O., Hostens, K., 1994. Changes in the fish fauna of the Oosterschelde estuary— a ten-year time series fyke catch. Hydrobiologia 282 (283), 497–507.

Hampel, H., Cattrijsse, A., 2004. Temporal variation in feeding rhythms in a tidal marshpopulation of the common goby Pomatoschistus microps (Kroyer, 1838). Aquat. Sci.66, 315–326.

Healy, M.C., 1972. On the population ecology of the common goby in the Ythan estuary.J. Nat Hist. 6, 133–145.

Heerkloss, R., Vietinghoff, U., 1981. Biomasseäquivalente planktischer und benthischerOrganismen in den Darß-Zingster Boddengewässern. Math – Nat - R, 4/5.Wiss ZtschrWilhelm-Pieck Univ, Rostock 31–36.

Herrmann, J.P., Jansen, S., Temming, A., 1998. Consumption of fish and decapod crustaceansand their role in the trophic relations of the Sylt-RømøBight. In: Gätje, C., Reise, K. (Eds.),Ökosystem Wattenmeer-Austausch-Transport- und Stoffumwandlungsprozesse.Springer, Berlin-Heidelberg, pp. 437–462.

Heymans, J.J., Guenette, S., Christensen, V., 2007. Indicators of ecosystem status in the Gulfof Alaska. Ecosystems 10, 488–502.

Hickel, W., 1975. The mesozooplankton in the Wadden Sea of Sylt (North Sea). Helgol.Mar. Res. 27, 254–262.

Hunter, E., Naylor, E., 1993. Intertidal migration by the shore crab Carcinus maenas. Mar.Ecol. Prog. Ser. 101, 131–138.

Kellnreitner, F., 2007. Populationsdynamik und Produktivität der StrandgrundelPomatoschistus microps und der Sandgrundel Pomatoschistus minutus (Teleostei:Gobiidae) der Sylt-Rømø Bucht (Deutschland). (MS Thesis) University of Vienna.

Kellnreitner, F., Pockberger, M., Asmus, H., 2012. Seasonal assemblage and feeding guildstructure of fish species in a boreal tidal basin. Estuar. Coast. Shelf Sci. 108, 97–108.

Kubetzki, U., Garthe, S., 2003. Distribution diet and habitat selection by four sympatricallybreeding gull species in the south-eastern North Sea. Mar. Biol. 143, 199–207.

Leitão, R., Martinho, F., Neto, J.M., Cabral, H., Marques, J.C., Pardal, M.A., 2006. Feedingecology, population structure and distribution of Pomatoschistus microps (Krøyer,1838) and Pomatoschistus minutus (Pallas, 1770) in a temperate estuary, Portugal.Estuar. Coast. Shelf Sci. 66, 231–239.

Libralato, S., Christensen, V., Pauly, D., 2006. A method for identifying keystone species infood web models. Ecol. Model. 195, 153–171.

Lucas, M.I., Walker, G., Holland, D.L., Crisp, D.J., 1979. An energy budget for the freeswimming and metamorphosing larvae of Balanus balanus (Crustacea: Cirripedia).Mar. Biol. 55, 221–229.

Maes, J., de Brabandere, L., Ollevier, F., Mees, J., 2003. The diet and consumption of dominantfish species in the upper Scheldt estuary, Belgium. J Mar Biol Assoc. U.K. 83, 603–612.

Mageau, M.T., Costanza, R., Ulanowicz, R.E., 1995. The development and initial testing of aqualitative assessment of ecosystem health. Ecosyst. Health 1, 201–213.

Magnhagen, C., Wiederholm, A.M., 1982a. Habitat and food preferences of Pomatoschistusminutus and P. microps (Gobiidae) when alone and together: an experimental study.Oikos 39, 152–156.

Magnhagen, C., Wiederholm, A.M., 1982b. Food selectivity versus prey availability: astudy using the marine fish Pomatoschistus microps. Oecologia 55, 311–315.

Martens, P., Van Beusekom, J.E.E., 2008. Zooplankton response to a warmer northernWadden Sea. Helgol. Mar. Res. 62, 67–75.

Miller, P.J., 1975. Age structure and life span in the common goby, Pomatoschistus microps.J Zool (Lond.) 177, 425–448.

Miller, P.J., 1986. Gobiidae. In: Withehead, P.J.P., Bauchat, M.L., Hureau, J.C., Nielsen, J.,Tortonese, E. (Eds.), Fishes of the North-eastern Atlantic and the Mediterranean. 3.UNESCO, Paris, pp. 1019–1085.

96 M. Pockberger et al. / Journal of Sea Research 86 (2014) 86–96

Miller, P.J., 2004. Pomatoschistus microps (Krøyer 1838), Gobiidae 2. In: Miller, P.J. (Ed.),The Freshwater Fishes of Europe. vol. 8/II. AULA-Verlag, Wiesbaden, pp. 293–330.

Monaco, M.E., Ulanowicz, R.E., 1997. Comparative ecosystem trophic structure of three USmid-Atlantic estuaries. Mar. Ecol. Prog. Ser. 161, 239–254.

Moore, J.W., Moore, I.A., 1976. The basis of food selection in some estuarine fishes. Eels,Anguilla anguilla (L), whiting, Merlangius merlangius (L), sprat, Sprattus sprattus (L)and stickleback, Gasterosteus aculeatus (L). J. Fish Biol. 9, 375–390.

Ntiba, M.J., Harding, D., 1993. The food and the feeding habits of the long rough dabHippoglossoides platessoides (Fabricius 1780) in the North Sea. Neth. J. Sea Res. 31,189–199.

Paine, R.T., 1969. A note on trophic complexity and community stability. Am. Nat. 929,91–93.

Pihl, L., 1982. Food intake of young cod and flounder in a shallow bay on the Swedishwestcoast. Neth. J. Sea Res. 15, 419–432.

Pihl, L., 1985. Food selection and consumption of mobile epibenthic fauna in shallowmarine areas. Mar. Ecol. Prog. Ser. 22, 169–179.

Piraino, S., Fanelli, G., Boero, F., 2002. Variability of species' role in marine communities:change of paradigms for conservation priorities. Mar. Biol. 140, 1067–1074.

Polte, P., Schanz, A., Asmus, H., 2005. The contribution of seagrass beds (Zostera noltii) tothe function of tidal flats as a juvenile habitat for dominant mobile epibenthos in theWadden Sea. Mar. Biol. 147, 813–822.

Power, M.E., Tilman, D., Estes, J.A., Menge, B.A., Bond, W.J., Mills, L.S., Daily, G., Castilla, J.C.,Lubchenco, J., Paine, R.T., 1996. Challenges in the quest for keystones. Bioscience 46,609–620.

Reise, K., 1979. Moderate predation on meiofauna by the macrobenthos of the WaddenSea. Helgol. Mar. Res. 32, 453–465.

Ropes, J.W., 1968. The feeding habits of the green crab, Carcinus maenas. FishB-NOAA 67,183–203.

Scheiffarth, G., Nehls, G., 1997. Consumption of benthic fauna by carnivorous birds in theWadden Sea. Helgol. Mar. Res. 51, 373–387.

Ulanowicz, R.E., Puccia, C.J., 1990. Mixed trophic impacts in ecosystems. Coenoses 5, 7–16.Vorberg, R., Breckling, P., 1999. Atlas der Fische im Schleswig-Holsteinischen

Wattenmeer. In: Vorberg, R., Breckling, P., Wurche, B. (Eds.), Schriftenreihen desNationalparks Schleswig Holsteinische Wattenmeer Heft 10. WestholsteinischeVerl.-Anst. Boyens, Heide, p. 178.

Widbom, B., 1984. Determination of average individual dry weights and ash-free dryweights in different fractions of marine meiofauna. Mar. Biol. 84, 101–108.

Winberg, G.G., 1956. Rate of metabolism and food requirements of fishes. J. Fish Res.Board Can. Transl. Ser. 194, 202.

Wootton, J.T., 1994. Predicting direct and indirect effects: an integrated approach usingexperiments and path analysis. Ecology 75, 151–165.

Wootton, J.T., Parker, M.S., Power, M.E., 1996. Effects of disturbance on river food webs.Science 273, 1558–1561.

Wulff, F., Field, J.G., Mann, K.H. (Eds.), 1989. Network analysis in marine ecologymethodsand application. Coastal and Estuarine Studies, 32. Springer, Heidelberg, p. 284.

Zander, C.D., Hartwig, E., 1982. On the biology and food of small-sized fish fromNorth andBaltic Sea areas IV. Investigations on an eulittoral mud flat at Sylt island. Helgol. Mar.Res. 35, 47–63.