Comparing trophic structure of a subtropical bay as estimated from mass-balance food web model and stable isotope analysis

Ecological Modelling 312 (2015) 175–181

Contents lists available at ScienceDirect

Ecological Modelling journal homepage: www.elsevier.com/locate/ecolmodel

Comparing trophic structure of a subtropical bay as estimated from mass-balance food web model and stable isotope analysis Jianguo Du a,b,∗ , William W.L. Cheung b , Xinqing Zheng a , Bin Chen a , Jianji Liao a , Wenjia Hu a a b

Third Institute of Oceanography, State Oceanic Administration, Daxue Road 178, Xiamen 361005, PR China Changing Ocean Research Unit, The University of British Columbia, Vancouver, British Columbia V6T 1Z4, Canada

a r t i c l e

i n f o

Article history: Received 26 January 2015 Received in revised form 22 May 2015 Accepted 23 May 2015 Keywords: Xiamen Bay Ecopath Stable isotope Trophic level Food web

a b s t r a c t The trophic structure of a subtropical bay in Taiwan Strait was analyzed by using two methods: mass balance modeling (Ecopath) and stable isotopic analysis (SIA). Trophic levels (TLs) of main functional groups estimated from the two methods were compared. The Ecopath model was built based on the fishery resource survey in Xiamen Bay in 2009. Specifically, data on species composition, biomass, mortality rates, diet composition and fisheries catches were obtained from the survey in and around the bay. The model consisted of 26 functional groups, including plankton, benthos, fish, cephalopods, shrimps, crabs and marine mammals. TLs of the main functional groups were estimated to be between 2.89 (cephalopods) and 3.94 (congers), with an average of 3.11. Trophic transfer efficiencies from levels II to V were 12.8%, 19.2%, 19.7% and 12.1%, respectively. Catfish (Tachysurus sinensis and Netuma thalassina) and the fisheries have major trophic impacts on most functional groups in the Xiamen Bay ecosystem. Total system throughput was estimated to be 411 t km−2 year−1 . TLs derived from isotopic analysis were highly correlated with those estimated from Ecopath (Linear regression: R2 = 0.696, n = 23, p < 0.001). On an average, Ecopath underestimated TLs of the functional groups by about 12.2% compared to those estimated from SIA, with TLs from Ecopath being slightly higher at low TLs and lower at high TLs. This studies support value of using both stable isotopes and Ecopath methods to analyze this food web. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Trophodynamic models and stable isotope analysis (SIA) are commonly used approaches to quantify trophic levels in marine food webs (Nilsen et al., 2008). One of the modeling approaches that have been most widely used is Ecopath with Ecosim (Christensen et al., 2008, 2014; Downing et al., 2012). Ecopath allows the application of network analysis to calculate effective trophic levels (TLs) for ecosystem functional groups based on carbon flow between groups and food web relationships derived mainly from laboratory feeding experiments and analysis of stomach contents. SIA has become a standard approach for understanding trophic interactions (Peterson and Fry, 1987; Post, 2002). Carbon and nitrogen stable isotope ratios, in particular, have been shown to be a valuable source of information to understand animals’ sources of diet (e.g., Papiol et al., 2012). By comparing TLs value from Ecopath with TLs estimated from ␦15 N SIA, it is possible to validate the

∗ Corresponding author. Tel.: +86 592 2195539; fax: +86 592 2191929. E-mail addresses: [email protected], j.du@fisheries.ubc.ca (J. Du). http://dx.doi.org/10.1016/j.ecolmodel.2015.05.027 0304-3800/© 2015 Elsevier B.V. All rights reserved.

network model (Deehr et al., 2014). Disagreement in such comparison could indicate an incompletely specified ecosystem model or unaccounted variation in stable isotope concentration due to environmental, metabolic or species-specific parameters (Dame and Christian, 2008). Several recent studies have shown good agreement between TLs calculated from ␦15 N values and those from food web models (Dame and Christian, 2008) although some studies also show that TLs estimated from Ecopath models are different from ␦15 N-based TLs (Polunin and Pinnegar, 2000; Milessi et al., 2010). For example, Nilsen et al. (2008) shows that TLs from Ecopath are slightly lower than isotope-based estimates and vice versa. Here, using Xiamen Bay marine ecosystem as a case study, we aim to compare TLs of key functional groups estimated from SIA and Ecopath trophic dynamic modeling. Several quantitative studies on the trophic ecology of Xiamen Bay have been conducted (Huang et al., 2006, 2008; Zhang and Huang, 2009; Du et al., 2012; Liao et al., 2014). However, the system has not been described using a trophic model. Thus, in this study, we develop an Ecopath model for Xiamen Bay marine ecosystem and analyze its trophic structure. We also estimate the TLs of key ecosystem groups using stable isotope

176

J. Du et al. / Ecological Modelling 312 (2015) 175–181

Xiamen Bay is a subtropical area, located from 24◦ 21 to 24◦ 32 N and from 118◦ 05 to 118◦ 13 E in the southeast coast of China bordering the Taiwan Strait to the east (Fig. 1). The area covers about 1260 km2 , with water depth ranging from 5 to 31 m. There are 31 islands in the Bay. The outflows of Jiulong River (Compiling Committee of Records of China Bays, 1993) greatly influence plankton dynamics, composition and function of the Bay’s food web (Huang et al., 2008). Moreover, Xiamen Bay supports a rich fauna while it is subjected to a variety of anthropogenic activities such as fishing and shipping (Lu et al., 1998; Zhang and Huang, 2009; Du et al., 2012; Liao et al., 2014).

An Ecopath model representing the state of Xiamen Bay marine ecosystem in 2009 was developed. The model area is 1260 km2 . The biomass flows within this area was assumed to be closed. We aggregated major species occurring in Xiamen Bay into 26 functional groups according to their ecological characteristics. Input parameters were based on data collected from the fishery resources surveys in 2009, fishery statistics, published literature and reports on diet compositions and FishBase (www.fishbase.org). Specifically, biomass estimates for most functional groups were based on survey data and published literature, while P/B was based on estimated total mortality rates and other publications. Q/B of fish and non-fish groups was based on data from FishBase and published literature. Catch is based on Fujian statistical yearbook (Yang, 2009). Diet composition matrix of the model is based on survey data and published literature (Yang, 2001a,b; Huang et al., 2008; Cheng et al., 2009). As biomass and diet composition are most uncertain among the input parameters, they were adjusted to achieve mass-balance of the model.

2.2. Mass-balanced trophic dynamic model

2.3. Stable isotopes processing

Ecopath with Ecosim (EwE) is one of the most widely used food web modeling and network analysis tools for marine, estuaries and freshwater ecosystems around the world. Ecopath was first developed to model coral reef ecosystems (Polovina, 1984). It was then further developed into the Ecopath with Ecosim software package (Christensen and Pauly, 1992, Christensen et al., 2005). This study used EwE version 6.4 that was available at http://www.ecopath. org/ Ecopath model is based on the mass balance of energy that goes into and out of the modeled ecosystem. To reduce complexity of the model, species with similar ecological characteristics are usually aggregated into functional groups. The model assumes that total amount produced or consumed by a group is equal to the amounts that go out of the group through predation and fishing mortalities, migrations and biomass accumulations:

The SIA was performed on 23 species/groups, including marine mammals, cartilaginous fish, bony fish, mollusks, crustaceans, benthos and zooplankton. All the muscles samples were freeze-dried at −40 ◦ C, ground into powder and sieved with 120 meshes. The ␦15 N signals of samples were measured using an isotope ratio mass spectrometer (IRMS) attached to a Flash EA1112 HT Elemental analyzer. The ratio of 15 N/14 N was detected by IRMS, and then compared with the international standards (Pee Dee Belnite and atmospheric N2 ), after which ␦15 N was calculated using the following equation:

approach. We then compare the trophic position and trophic levels estimated from the two methods. 2. Materials and methods 2.1. Study area

Bi (P/B)i EEi = Yi +

n 

ı15 N(%0 ) =

(Rsample − Rstandard ) Rstandard

× 1000

(2)

The detection limits were 0.2‰ for ␦15 N. R represents the 15 N/14 N.

2.4. Comparison between Ecopath and isotope results Bj (Q/B)j DCij + Bi BAi + Ei

(1)

j=1

where, B is biomass, P the production, EE the ecotrophic efficiency, Y the fishery catch, i and j are prey and predator groups, Q is consumption, DC the diet composition, BA the biomass accumulation and E the net immigration. The model achieves mass balance by solving Eq. (1) simultaneously for all functional groups in the model. Thus, one of the input parameters B, P/B, Q/B or EE for each functional group should be left to be estimated by the model.

TLs of each organism were estimated according to Post (2002): 15

TLconsumer = TLbasis +

15

(␦ Nconsumer − ␦ Nbasis ) TEF

(3)

where, TLbasis is the trophic position of a primary consumer used to estimate the TLs of other consumers in the food web (Vander Zanden and Rasmussen, 1999; Post, 2002), and is assumed to be equal to 2. ␦15 Nconsumer is the value measured for consumers. ␦15 Nbasis is ␦15 N of organisms that are herbivores and are sessile or have very limited mobility. In the present study, Ruditapes philippinarum was identified as the most important species that makes up the total organic matter at the base of the food web. TEF is the ␦15 N trophic enrichment factor for the difference between a source and its consumer, and 3.4 parts per thousand (ppt) is the average ␦15 N enrichment per trophic level (Vander Zanden and Rasmussen, 2001). TLs estimated by the Ecopath model were plotted against the corresponding TLs estimated by SIA and their correlation was tested using the Spearman-rank correlation coefficient test (Zar, 1999). 3. Results 3.1. Ecopath outputs and sensitivity

Fig. 1. Study area of the Xiamen Bay.

The basic input such as biomass, landings, P/B, Q/B, diet composition and estimated outputs like TLs, mortality rates and ecotrophic efficiency from the Xiamen Bay model in 2009 are summarized in Table 1 and Table 2. The flow diagram of the balanced trophic model

Table 1 Basic input and estimated outputs (bold) parameters for the functional groups in the Xiamen Bay model. (P/B: production–biomass ratio; Q/B: consumption–biomass ratio). Species or pooled taxa

Marine mammals Bambooshark Fanray Stingray Congers Catfish Big head croaker Japanese seabass Anchovy Bombay-duck Silver pomfret Belanger’s croaker Terapon Lattice blaasop Gobiidae Bartail flathead Tongue sole Flounder Small fishes Shrimps Crabs Cephalopods Benthos Zooplankton Phytoplankton Detritus

Sousa chinensis Chiloscyllium plagiosum Platyrhina sinensis Dasyatis spp. Muraenesox cinereus Tachysurus sinensisNetuma thalassina Collichthys lucidus Lateolabrax japonicus Coilia mystus Harpadon nehereus Pampus argenteusPsenopsis anomala Johnius spp. Therapon spp. Takifugu spp. Odontamblyopus spp.Oxyurichthys spp.Trypauchen spp.Chaeturichthys spp. Platycephalus spp. Cynoglossus spp. Pleuronichthys spp. Lepidotrigla spp.Inimicus spp.Bostrychus spp.Xenocephalus spp. Metapenaeus spp.Exopalaemon spp. Charybdis spp. Loligo spp.Octopus spp. Ruditapes spp.

0.00402[2] 0.00216[2] 0.00209[2] 0.00516[2] 0.510[2] 0.00119[2] 0.00300[2] 0.0842[2] 0.00357[2] 0.000402[2] 0.0157[2] 0.000183[2] 0.000575[2] 0.0122[2] 0.000180[2] 0.047617[2] 0.000217[2] 0.0268[2] 0.0676[2] 0.0840[2] 0.0299[2]

Biomass(t km−2 year−1 )

P/B(year−1 )

Q/B(year−1 )

Trophiclevel

Ecotrophicefficiency

Fishingmortality

0.00806 0.0201 0.00920 0.00888 0.00240 0.468 0.00138 0.00268 0.0526 0.00225 0.000345 0.0114 0.000169 0.000268 0.0102 0.0000767 0.0261 0.0000997 0.0353 0.0919 0.0930 0.0582 0.537 0.369 0.922 43[1]

0.050[1] 0.480[3] 0.380 0.380 2.300[1] 1.620[3] 2.420[3] 2.400 2.200[3] 2.410[3] 2.120 2.010[3] 2.410 2.500[1] 2.980[3] 2.98 3.550[5] 3.550[5] 2.300[1] 3.900[1] 3.900[1] 3.000[1] 5.000[1] 40.000[1] 200.000[1]

30.000 [1] 17.400[4] 5.700[4] 9.600[4] 4.400[4] 3.800[4] 19.200[4] 4.000[4] 26.200[4] 11.800[4] 10.100[4] 8.700[4] 9.600[4] 16.000[4] 13.600[4] 5.200[4] 8.300[4] 11.500[4] 24.000[1] 27.000[1] 15.000[1] 10.000[1] 20.000[1] 160.000[1]

3.848 3.422 3.476 3.468 3.941 3.158 3.728 3.610 3.048 3.810 3.327 3.666 3.735 3.421 3.094 3.400 3.115 3.293 2.902 2.906 2.920 2.892 2.100 2.000 1.000 1.000

0.0870 0.975 0.960 0.973 0.983 0.999 0.851 0.849 0.996 0.985 0.970 0.975 0.972 0.964 0.981 0.955 0.996 0.910 0.985 0.991 0.994 0.997 0.875 0.438 0.295 0.103

0.000 0.200 0.235 0.235 2.150 1.090 0.865 1.118 1.600 1.590 1.164 1.380 1.085 2.141 1.200 2.350 1.826 2.178 0.758 0.735 0.904 0.514 0.000 0.000 0.000

Notes: [1]—Cheng et al. (2009); [2]—Yang (2009); [3]—Huang (2011); [4]—FishBase; [5]—Liu et al. (2014).

Table 2 Diet composition of functional groups in the Xiamen Bay model. Prey\predator

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Sum

Marine mammals Bambooshark Fanray Stingray Congers Catfish Big head croaker Japanese seabass Anchovy Bombay-duck Silver pomfret Belanger’s croaker Terapon Lattice blaasop Gobiidae Bartail flathead Tonguesole Flounder Small fishes Shrimps Crabs Cephalopods Benthos Zooplankton Phytoplankton Detritus

0 0.0006 0.0006 0.0006 0.001 0.161 0.001 0.01 0.04 0.0005 0.001 0.02 0 0.0003 0 0 0.06 0 0.07 0.03 0.04 0.4 0.099 0.064 0 0 1

0.0001 0.015 0.003 0.003 0 0.08 0 0 0 0 0 0 0 0 0.001 0 0.079 0.0003 0.02 0.04 0.05 0.05 0.6446 0.014 0 0 1

0 0 0 0 0 0.08 0 0 0.005 0 0 0 0 0 0.072 0 0 0 0.05 0.15 0.031 0.038 0.559 0.015 0 0 1

0 0 0 0 0 0.08 0 0 0.005 0 0 0 0 0 0.062 0 0 0 0.05 0.15 0.031 0.038 0.584 0 0 0 1

0 0 0 0 0 0.1 0.01 0 0 0 0 0.041 0 0 0.065 0 0.22 0 0.069 0.067 0.092 0.236 0.05 0.05 0 0 1

0 0 0 0 0 0.09 0 0 0.003 0 0 0 0 0 0 0 0 0 0.001 0.04 0.1 0.01 0.38 0.251 0 0.125 1

0 0 0 0 0 0.05 0 0 0 0 0 0 0.008 0 0.002 0 0.002 0 0.1 0.51 0.106 0 0.02 0.202 0 0 1

0 0 0 0 0 0.073 0.006 0.005 0.105 0 0 0.006 0 0 0.062 0 0 0 0 0.127 0.15 0.038 0.323 0.105 0 0 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.004 0.049 0 0 0 0.947 0 0 1

0 0 0 0 0.001 0.16 0.04 0 0.1075 0.06 0.0025 0.0002 0 0 0.0765 0.001 0 0 0.0745 0.1429 0.01 0.008 0.306 0.0095 0 0 1

0 0 0 0 0 0 0.05 0 0 0 0 0 0 0 0 0 0 0 0.1 0.117 0.048 0 0 0.685 0 0 1

0 0 0 0 0 0.02 0 0 0.09 0.0005 0 0 0 0 0 0 0 0 0.01 0.2058 0.345 0.0032 0.326 0 0 0 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.325 0 0 0 0.1 0.15 0.15 0 0.15 0.125 0 0 1

0 0 0 0 0 0 0 0 0.01 0 0 0 0 0 0 0 0 0 0.018 0.296 0.2765 0.03 0.0135 0.2 0 0.156 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.02 0 0 0 0.005 0.0455 0 0 0.7645 0.115 0 0.05 1

0 0 0 0 0 0 0 0 0.011 0 0 0.01 0 0 0.075 0 0.067 0 0 0.21 0.018 0.078 0.3 0.14 0 0.091 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.006 0 0 0 0 0.153 0.02 0.015 0.676 0 0 0.13 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.045 0.01 0.014 0 0 0.215 0.0563 0.042 0.4977 0 0 0.12 1

0 0 0 0 0 0 0 0 0.003 0 0 0 0 0 0 0 0 0 0 0 0.02 0 0.4 0.417 0.08 0.08 1

0 0 0 0 0 0 0 0 0 0 0 0.0005 0 0 0 0 0 0 0.0007 0 0 0 0.05 0.8488 0.05 0.05 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.01 0.003 0 0.509 0.335 0.003 0.14 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.01 0.02 0 0 0.05 0.78 0.09 0.05 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.1 0.1 0.8 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.9 0.1 1

177

No.

J. Du et al. / Ecological Modelling 312 (2015) 175–181

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

landing(t km−2 year−1 )

Group name

178

J. Du et al. / Ecological Modelling 312 (2015) 175–181 Congers 4

Bombay-duck

Big head croaker

Terapon

Marine mammals Fanray

Japanese seabass

Stingray

Bambooshark

Flounder

Crabs

Shrimps

Tonguesole

Gobiidae

Anchovy

Catfish Small fishes

Bartail flathead

Lattice blaasop

Silver pomfret

3

Trophic level

Belanger's croaker

Cephalopods

2

Benthos

1

Zooplankton

Detritus Phytoplankton

Fig. 2. Trophic flow diagram of the balanced trophic model of Xiamen Bay. The components of the system are structured along the vertical axis according to their trophic level. The area of each circle is proportional to the biomass of each function group.

is shown in Fig. 2. Three top predators are identified at Xiamen Bay: congers, marine mammals and bombay duck (TLs of 3.94, 3.85 and 3.81, respectively). Other high level consumers are fish such as catfish and flounder (TLs from 3.0 to 3.8). Primary consumers are small fishes and invertebrates (TLs from 2.0 to 3.0). The benthos and catfish are the most representative group of primary and second consumers (Table 1). The parameter values of this model were mostly estimated from regional and local data (75.6% of 121 parameters), indicating a relatively higher reliability of using this model to represent the Xiamen Bay ecosystem.

Table 3 Absolute (t km−2 ) and relative (%) distribution of biomass, and transfer efficiencies (%) among discrete trophic levels of the Xiamen Bay model.

3.2. Structural and trophic network analyses

Groups with TL ≥ 3 represent about 31.2% of the total biomass in the ecosystem (Table 3). When aggregating biomass flows of different functional groups into seven trophic levels, level I–III contributes more than 94.8% of the total biomass flows in the ecosystem. Flows

TLs of the major exploited groups in Xiamen Bay are between 2.89 (cephalopods) and 3.94 (congers), with an average of 3.11.

Trophic level

Biomass

Transfer efficiencies

I II III IV V VI VII

0.905 (33.018) 0.981 (35.791) 0.712 (25.977) 0.136 (4.962) 0.00681 (0.249) 0.000106 (0.00387) 0.000001 (0.0000365)

12.8 19.2 19.7 12.1 5.9 5.6

Fleet1

Phytoplankton

Positive Detritus

Zooplankton

Benthos

Crabs

Cephalopods

Shrimps

Small fishes

Flounder

Bartail flathead

Tonguesole

Lattice blaasop

Gobiidae

Terapon

Belanger's croaker

Silver pomfret

Bombay-duck

Anchovy

Negative Marine mammals Bambooshark Fanray Stingray Congers Catfish Big head croaker Japanese seabass Anchovy Bombay-duck Sil ver pomfret Belanger's croaker Terapon Lattice blaasop Gobiidae Bartail flathead Tonguesole Flounder Small fishes Shrimps Crabs Cephalopods Benthos Zooplankton Phytoplankton Detritus Fleet1

Fig. 3. Mixed trophic impact assessment of the Xiamen Bay ecosystem.

Impacting group

Big head croaker

Japanese seabass

Catfish

Stingray

Congers

Fanray

Bambooshark

Marine mammals

Impacted group

Trophic level from Ecopath

J. Du et al. / Ecological Modelling 312 (2015) 175–181

4.5 4.0 3.5 3.0

y=0.784+0.762x 2 r =0.6962, n=23 p<0.001

2.5 2.0 1.5 1.0

theoretical value

0.5

179

and the world, suggesting certain level of ubiquity of these ecosystem properties across ecosystems. Specifically, total system throughput is about 411 t km−2 year−1 , near to a subtropical coastal lagoon in Uruguay (451 t km−2 year−1 in Milessi et al., 2010), but just about one fifth of the southern East China Sea (Li et al., 2010b). In addition, estimated mean transfer efficiency of Xiamen Bay ecosystem is 16.9%, very close to 16.3% of the East China Sea (Cheng et al., 2009), higher than Kuosheng Bay in Northern Taiwan (6.5% in Lin et al., 2004), also higher than Daya Bay (8.9% in Wang et al., 2005) and Beibu Gulf (9.1% in Chen et al., 2011) in South China Sea, indicating that the system has high potential for adaptation and resilience capacity.

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Trophic level from SIA Fig. 4. Linear regression between trophic levels estimated from Ecopath model and stable isotope analysis for Xiamen Bay ecosystem (solid line). The broken line represents a 1:1 relationship.

originating from primary producers and detritus are combined to estimate trophic transfer efficiency (TTE) by trophic level, TTE for levels II to VII are 12.8%, 19.2%, 19.7%, 12.1%, 5.9% and 5.6%, respectively (Table 3). The average TTE for the whole Xiamen Bay ecosystem is 16.9%, within the theoretical range of 10-20% (Odum et al., 1971). Besides the higher TL groups like marine mammals and the lower TL groups like planktons, the EE of most groups is over 0.85 (Table 3), which indicated that large proportion of productions from these groups was transferred directly to their predators or to the fisheries. While the EE of marine mammals is very low (EE = 0.087) as their predation mortality is low. Catfish and the fisheries exert the largest trophic impacts on most other functional groups in Xiamen Bay ecosystem (Fig. 3). 3.3. Comparison of TLs estimates between Ecopath and SIA TLs estimated from the Ecopath model were highly and positively correlated with those estimated from ␦15 N values (Fig. 4, R2 = 0.6962, n = 23, p < 0.001), suggesting a good agreement in TL estimates from both methodologies. However, overall, TLs estimated from Ecopath were slightly higher than those from SIA for low TL functional groups and lower for high TL functional groups. On average, TLs estimated from Ecopath were 12.2% lower than estimates from SIA. 4. Discussion 4.1. Trophic structure and trophic impact assessment of Xiamen Bay ecosystem This study provides a quantitative description of the Xiamen Bay ecosystem and allows us to compare it with nearby ecosystems. Estimated benthos biomass in Xiamen Bay (0.537 t km−2 year−1 ) is lower than East China Sea (3.66 t km−2 year−1 in Cheng et al., 2009) and Kuosheng Bay in Northern Taiwan (8.79 t km−2 year−1 in Lin et al., 2004), because the components of benthos function group inshore and offshore are different, and the benthos value of Kuosheng Bay was from Caribbean coral reef, which is a quite different ecosystem. Estimated TL of catches (3.11) is similar to southern East China Sea (Li et al., 2010a,b), but is a little lower than other bay or open sea in the East China Sea region (e.g. 3.32 in Lin et al., 2004; 3.32 in Cheng et al., 2009), because the catfish (TL = 3.16) contributed about 56.62% of the total catch in Xiamen Bay, while the hailteil (TL = 3.88) dominated in East China Sea region. Estimated trophic transfers in the Xiamen Bay ecosystem are comparable to similar types of ecosystems in the western Pacific

4.2. Comparison of the Ecopath and stable isotope approaches Our results revealed a clear correlation between the TLs estimated by the Ecopath model and ␦15 N values, indicating that both methodologies are useful to determine the trophic levels of marine organisms in Xiamen Bay food web. This study provides support to previous studies that TLs estimated from Ecopath model and SIA are complementary, for example, significant and close-to-one positive correlation were found between TLs estimated by Ecopath and those derived from SIA in Prince William Sound (R2 = 0.97 in Kline and Pauly, 1998), in a high latitude fjord ecosystem (R2 = 0.72 in Nilsen et al., 2008), in a subtropical lagoon (R2 = 0.82, n = 14 in Milessi et al., 2010), and in Bay of Biscay continental shelf food web (R2 = 0.72, n = 16 in Lassalle et al., 2014). There was also a clear correlation between the TLs estimated by Ecopath and the ␦15 N values in Mediterranean marine food web (R2 = 0.48, n = 24 in Navarro et al., 2011). Furthermore, Kline and Pauly (1998) also applied their isotope data to another ecosystem, the precision was still good though was reduced. Therefore, the isotope data obtained from our study will be valuable also for the future modelling of other subtropical ecosystems. Also, the strong relationship between trophic levels estimated by the Ecopath model and the SIA highlights that the diet information used in the Ecopath model was accurate (Coll et al., 2006). The biases in TLs estimated in Ecopath model relative to those estimated from SIA agrees with previous estimates. For example, analysis of 14 species in Laguna de Rocha suggested a 13.5% lower in TLs estimated from Ecopath model (Milessi et al., 2010) while analysis of 16 species in Bay of Biscay suggest a <13% underestimation of TLs (Lassalle et al., 2014). The convention of using detritus and discarded material as TL 1 has probably resulted in lower model TL estimates relative to those calculated from SIA (Nilsen et al, 2008; Navarro et al, 2011). In contrast, in one case, the values of TLs from Ecopath overestimated 8% on average to those estimated by stable isotope analysis in 13 species in the Mediterranean Sea (Polunin and Pinnegar, 2000). This shows that the bias is not consistent across ecosystems. In addition, our results showed that TL values from Ecopath were slightly lower at high TLs and higher at low TLs compared with from stable isotope. Similar conclusion was drawn from a study of a fjord ecosystem in North Norway (Nilsen et al., 2008). Thus, although TLs estimates from Ecopath model and SIA are broadly correlated, there are systematic biases between TLs that vary between ecosystems. Therefore, it is recommended that TL estimation be cross-validated using trophodynamic models and SIA whenever possible. There are several potential reasons for the differences between the two approaches. One reason is that, though SIA reflects timeintegrated diet information, the results are strongly affected by temporal and spatial sampling (e.g., seasons and sites), while Ecopath model synthesizes available data to provide an average representation of the ecosystem. Moreover, TLs derived from SIA are based on two critical assumptions, the average ␦15 N enrichment between the predator and prey, and the choice of herbivorous

180

J. Du et al. / Ecological Modelling 312 (2015) 175–181

organisms at the base of food web to set reference values (Fry, 2006). Though 3.4 ppt used in this study is a robust average for ␦15 N enrichment, it can vary between 0.5 to 5.5 ppt (Vander Zanden and Rasmussen, 2001; Post, 2002). Long-lived large bivalves can integrate temporal diet variation in the isotopic signatures, and this makes them suitable as baselines in ecosystems. The average large bivalve TL of 2.0, with Ruditapes spp. as baseline, is similar to the baseline level selected for Pecten maximus (Pinnegar et al., 2002) and Calanus finmarchicus (Nilsen et al., 2008). However, large variation in TL estimates for filtering bivalves show that robust baseline assumptions are difficult to make for this group of species (Lesage et al., 2001; Tamelander et al., 2006; Nilsen et al., 2008). Decision in the design of functional group structures in the Ecopath model may be a source of uncertainty, although there is a trade-off between simplicity and complexity. Aggregating species into “function group” is a way to reduce the complex of food web. The definition of groups like species aggregation will affect the model performance, specifically on the TL of each group. Although effort is made to design functional group structure that is representative of trophodynamics, there will be within group variability, like trophic ontogeny. A simple solution to the variability of dealing with trophic ontogeny is to divide each group into two sub-groups representing the juvenile and adult stage, especially for top predators (Walters et al., 1997). In this study, each function group is conformed according to species traits that are most relevant to define their ecological role. These traits include individual size, diet, habitat and phylogeny. Thus use of species’ traits in designing model structure help to reduce uncertainty (Milessi et al., 2010). Some input parameters are relatively more uncertain because they were either based on educated guess or, and based on estimates for other ecosystems which are similar to Xiamen Bay. For instance, parameters for phytoplankton, zooplankton and detritus are based on data from the southern East China Sea model and East China Sea model (Li et al., 2010b; Cheng et al., 2009). Moreover, some parameters, like the P/B of some fish groups are based on the author’s previous study in 2006, but the fish biomass is from the survey in 2009, and the diet information of some fish is from the 1980s and 1990s. Furthermore, there are seasonal changes in fish biomass in Xiamen Bay (Liao et al., 2014). Such seasonal dynamics may have strong influences on the Xiamen Bay ecosystem, which may not be captured using an annual-averaged Ecopath model. Further studies by developing seasonal sub-models and performing dynamic simulations at seasonal and longer time scale may allow us to examine these assumptions in more details. The results of the study can be used for management and conservation of the bay. The high EE of the present Xiamen Bay ecosystem suggests that the system is instable. Average EE is usually less than 0.5 in a stable ecosystem. Therefore, the low maturity and stability of the current ecosystem may render the system more sensitive to global changes. Higher sensitivity to environmental changes may lead to larger variations in stock abundance, and therefore, the fishery catches. As several species in Xiamen Bay are already severely depleted (Lu et al., 1998; Du et al., 2012), the increase in fluctuation or reduced stability may greatly affect the conservation and restoration of these species. Meanwhile, catfish and the fisheries exert major trophic impacts on most groups in Xiamen Bay ecosystem (Fig. 3). Increasing catfish biomass exerts negative impacts on most other groups, like lattice blaasop, belanger’s croaker, congers and Japanese seabass. Moreover, if fishing effort increases, biomass of most groups would decrease except shrimps and crabs. The negative impacts are more prominent in higher TLs. Such trophic responses are similar to those estimated in the East China Sea (Cheng et al., 2009; Li et al., 2010b), which may negatively affects the conservation and restoration of Xiamen Bay ecosystem. In summary, this study synthesized the best available scientific knowledge using the Ecopath modeling approach to describe the

Xiamen Bay marine ecosystem. Our results show a clear correlation between TLs estimated by Ecopath and derived from SIA, which suggested both methodologies are useful to determine the trophic position of marine species in food web. However, Ecopath underestimated TLs of the functional groups compared to those estimated from SIA, with TLs from Ecopath being slightly higher at low TLs and lower at high TLs. Therefore, it is recommended that TL estimation be cross-validated using trophodynamic models and SIA whenever possible. Acknowledgments The authors are grateful to Dr. Villy Christensen in the University of British Columbia Fisheries Centre for his help in developing the Ecopath model, to Dr. Xijie Yin in the Third Institute of Oceanography, State Oceanic Administration that assisted in the laboratory work. And we thank all anonymous reviewers for their constructive comments which greatly improved this paper. JGD is a visiting scholar supported by China Scholarship Council (no. 201309660067), acknowledges financial support from National Natural Science Foundation of China (no. 31101902), Scientific Research Foundation of Third Institute of Oceanography, SOA (no. 2011006), Natural Science Foundation of Fujian Province (no. 2012J05074 and 2014J01127) and Public Science and Technology Research Funds Projects of Ocean of China (no. 201305030-4). WWLC acknowledges funding support of the National Sciences and Engineering Research Council of Canada (no. RGPIN 41819812) and Nippon Foundation—University of British Columbia Nereus Program. References Chen, Z.Z., Qiu, Y.S., Xu, S.N., 2011. Changes in trophic flows and ecosystem properties of the Beibu Gulf ecosystem before and after the collapse of fish stocks. Ocean. Coast. Manag. 54, 601–611. Cheng, J.H., Cheung, W.W.L., Pitcher, T.J., 2009. Mass-balance ecosystem model of the East China Sea. Prog. Nat. Sci. 19, 1271–1280. Christensen, V., Pauly, D., 1992. ECOPATH II—a software for balancing steady-state ecosystem models and calculating network characteristics. Ecol. Model. 61, 169–185. Christensen, V., Walters, C.J., Pauly, D., 2005. Ecopath with Ecosim: A User’s Guide. Fisheries Centre, University of British Columbia, Vancouver. Christensen, V., Walters, C.J., Pauly, D., Forrest, R., 2008. Ecopath with Ecosim Version 6: User Guide. University of British Columbia, Vancouver (Lenfest Ocean Futures Project). Christensen, V., Coll, M., Piroddi, C., Steenbeek, J., Buszowski, J., Pauly, D., 2014. A century of fish biomass decline in the ocean. Mar. Ecol. Prog. Ser. 512, 155–166. Coll, M., Palomera, I., Tudela, S., Sarda, F., 2006. Trophic flows, ecosystem structure and fishing impacts in the South Catalan Sea, Northwestern Mediterranean. J. Marine Syst. 59, 63–96. Compiling Committee of Records of China Bays, 1993. Records of China Bays, 8th Fascicule. China Ocean Press, Beijing, China (in Chinese). Dame, J.K., Christian, R.R., 2008. Evaluation of ecological network analysis: validation of output. Ecol. Model. 210, 327–338. Deehr, R.A., Luczkovich, J.J., Hart, K.J., Clough, L.M., Johnson, B.J., Johnson, J.C., 2014. Using stable isotope analysis to validate effective trophic levels from Ecopath models of areas closed and open to shrimp trawling in Core Sound, NC, USA. Ecol. Model. 282, 1–17. Du, J.G., Liu, Z.H., Yu, X.G., Xu, Z.C., Hu, W.J., Chen, B., Ma, Z.Y., Lin, J.L., 2012. Fish species diversity and trophic level in the Jiulong Estuary. J. Trop. Oceanogr. 31, 76–82 (in Chinese with English abstract). Downing, A.S., Van Nes, E.H., Janse, J.H., Witte, F., Cornelissen, I.J., Scheffer, M., Mooij, W.M., 2012. Collapse and reorganization of a food web of Mwanza Gulf, Lake Victoria. Ecol. Appl. 22, 229–239. Fry, B., 2006. Stable Isotope Ecology, 1st ed. Springer, New York, pp. 308. Huang, L.M., 2011. Study on Fishery Resources And Fish Diversity In Minjiang River Estuary And Jiulong River Estuary And Their Adjacent Waters. Ocean University of China, China (Ph.D. thesis, in Chinese with English abstract). Huang, L.M., Zhang, Y.Z., Pan, J.J., Wu, Y.J., Cui, Y.X., 2008. Food web of fish in Xiamen eastern waters. J. Oceanogr. Taiwan Strait 27, 64–73 (in Chinese with English abstract). Huang, Z.G., Hong, R.B., Zhang, L.F., 2006. Study of species in Xiamen Bay. J. Xiamen Univ. 45, 10–15 (in Chinese with English abstract). Kline, T.C., Pauly, D., 1998. Cross-validation of trophic level estimates from a massbalance model of Prince William Sound using 15N/14N data. In: Fishery Stock Assessment Models. Alaska Sea Grant College Program, AK-SG-98-01.

J. Du et al. / Ecological Modelling 312 (2015) 175–181 Lassalle, G., Chouvelon, T., Bustamante, P., Niquil, N., 2014. An assessment of the trophic structure of the Bay of Biscay continental shelf food web: comparing estimates derived from an ecosystem model and isotopic data. Prog. Oceanogr. 120, 205–215. Lesage, V., Hammill, M.O., Kovacs, K.M., 2001. Marine mammals and the community structure of the Estuary and Gulf of St Lawrence, Canada: evidence from stable isotope analysis. Mar. Ecol. Prog. Ser. 210, 203–221. Li, R., Han, Z., Cheng, H.Q., Jiang, H., 2010a. A preliminary study on biological resource energy flows based on the Ecopath model in the East China Sea. Resour. Sci. 32, 600–605 (in Chinese with English abstract). Li, Y.K., Yu, N., Chen, L.Q., Chen, Y., Feng, D.X., 2010b. Ecological modeling on structure and functioning of southern East China Sea ecosystem. Prog. Fish. Sci. 31, 30–39 (in Chinese with English abstract). Liao, J.J., Zheng, X.Q., Du, J.G., Chen, B., Ma, Z.Y., Hu, W.J., 2014. Biodiversity and trophic level characteristics of fishes captured by set nets in Tong’an Bay, Xiamen. Biodivers. Sci. 22, 624–629 (in Chinese with English abstract). Lin, H.J., Shao, K.T., Hwang, J.S., Lo, W.S., Cheng, I.J., Lee, L.H., 2004. A trophic model for Kuosheng Bay in northern Taiwan. J. Mar. Sci. Technol. 12, 424–432. Liu, Y.W., Xue, Y., Wei, B.F., Ren, Y.P., 2014. An estimation of population parameters for five major fish species in the Haizhou Bay, Yellow Sea. J. Fish. Sci. China 21, 125–133 (in Chinese with English abstract). Lu, Z.B., Yan, Y.M., Du, Q., 1998. Variation of fishery resources and estimation of suitable fishing efforts in Xiamen coastal waters. J. Oceanogr. Taiwan Strait 17, 309–316 (in Chinese with English abstract). Milessi, A.C., Danilo, C., Laura, R.G., Daniel, C., Javier, S., Rodríguez-Gallegod, L., 2010. Trophic mass-balance model of a subtropical coastal lagoon, including a comparison with a stable isotope analysis of the food-web. Ecol. Model. 221, 2859–2869. Navarro, J., Coll, M., Louzao, M., Palomera, I., Delgado, A., Forero, M.G., 2011. Comparison of ecosystem modelling and isotopic approach as ecological tools to investigate food webs in the NW Mediterranean Sea. J. Exp. Mar. Biol. Ecol. 401, 97–104. Nilsen, M., Pedersen, T., Nilssen, E.M., Fredriksen, S., 2008. Trophic studies in a highlatitude fjord ecosystem-a comparison of stable isotope analyses (␦13 C and ␦15 N) and trophic-level estimates from a mass-balance model. Can. J. Fish. Aquat. Sci. 65, 2791–2806. Odum, E.P., Odum, H.T., Andrews, J., 1971. Fundamentals of Ecology. Saunders, Philadelphia.

181

Papiol, V., Cartes, J.E., Fanelli, E., Rumolo, P., 2012. Food web structure and seasonality of slope megafauna in the NW Mediterranean elucidated by stable isotopes: relationship with available food source. J. Sea Res. 77, 53–69. Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Annu. Rev. Ecol. Syst. 18, 293–320. Pinnegar, J.K., Jennings, S., O’Brien, C.M., Polunin, N.V.C., 2002. Long-term changes in the trophic level of the Celtic Sea fish community and fish market price distribution. J. Appl. Ecol. 39, 377–390. Polovina, J.J., 1984. Model of a coral reef ecosystem I. The ECOPATH model and its application to French Frigate Shoals. Coral Reefs 3, 1–11. Polunin, N.V.C., Pinnegar, J.K., 2000. Trophic-level dynamics inferred from stable isotopes of carbon and nitrogen. In: Briand, F. (Ed.), Fishing Down the Mediterranean Food-Webs? CIESM Workshop Series, vol. 12, pp.69–72. Post, D.M., 2002. Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83, 703–718. Tamelander, T., Renaud, P.E., Hop, H., Carroll, M.L., Ambrose, W.G., Hobson, K.A., 2006. Trophic relationships and pelagic-benthic coupling during summer in the Barents Sea Marginal Ice Zone, revealed by stable carbon and nitrogen isotope measurements. Mar. Ecol. Prog. Ser. 310, 33–46. Vander Zanden, M.J., Rasmussen, J.B., 1999. Primary consumer ␦13 C and ␦15 N and the trophic position of aquatic consumers. Ecology 80, 1395–1404. Vander Zanden, M.J., Rasmussen, J.B., 2001. Variation in ␦15 N and ␦13 C trophic fractionation: implications for aquatic food web studies. Limnol. Oceanogr. 46, 2061–2066. Walters, C., Christensen, V., Pauly, D., 1997. Structuring dynamic models of exploited ecosystems from trophic mass-balance assessments. Rev. Fish. Biol. Fish. 7, 139–172. Wang, X.H., Du, F.Y., Qiu, Y.S., Li, C.H., Huang, H.H., Sun, D.R., Jia, X.P., 2005. Study on the ecosystem model of Daya Bay I. A preliminary approach on energy flow model. South China Fish. Sci. 1, 1–8 (in Chinese with English abstract). Yang, H.C., 2009. Fujian Statistical Yearbook. China Statistics Press, Beijing, China (in Chinese). Yang, J.M., 2001a. A study on food and troohic levels of Baohai Sea invertebrates. Modern Fish. Inform. 16, 10–19 (in Chinese). Yang, J.M., 2001b. A study on food and troohic levels of Baohai Sea fish. Modern Fish. Inform. 16, 10–19 (in Chinese). Zar, J.H., 1999. Biostatistical Analysis, 4th ed. Prentice-Hall, Englewood Cliffs. Zhang, Y.Z., Huang, L.M., 2009. A study on fish species diversity and community structure in eastern waters of Xiamen. J. Trop. Oceanogr. 28, 66–76 (in Chinese with English abstract).