• Tidak ada hasil yang ditemukan

Community-level impacts of trawl selectivity in the Eastern Mediterranean Sea assessed using an ecosystem modelling approach

N/A
N/A
MUHAMMAD NAUFAL ALTAAF

Academic year: 2024

Membagikan "Community-level impacts of trawl selectivity in the Eastern Mediterranean Sea assessed using an ecosystem modelling approach "

Copied!
15
0
0

Teks penuh

(1)

Original Article

Community-level impacts of trawl selectivity in the Eastern Mediterranean Sea assessed using an ecosystem

modelling approach

_ Ismet Saygu

1

*, Johanna J. Heymans

2,3

, Clive Fox

2

, Hu¨seyin O ¨ zbilgin

4

, Jacob W. Bentley

5,6

, Ahmet Raif Eryas¸ar

7

, and Go¨khan Go¨kc¸e

1

1Cukurova University, Fisheries Faculty, Adana, Turkey

2Scottish Association for Marine Science, Scottish Marine Institute, Oban, Scotland, UK

3European Marine Board, Wandelaarkaai 7, Oostende 8400, Belgium

4Mersin University, Fisheries Faculty, Mersin, Turkey

5UN Environment World Conservation Monitoring Centre, Cambridge, UK

6Durrell Institute of Conservation and Ecology, School of Anthropology and Conservation, University of Kent, Canterbury, UK

7Recep Tayyip Erdogan University, Vocational School of Technical Sciences, Rize, Turkey

*Corresponding author: tel:þ90 322 338 6084; e-mail:[email protected].

Saygu,_I., Heymans, J. J., Fox, C., O¨ zbilgin, H., Bentley, J. W., Eryas¸ar, A. R., and Go¨kc¸e, G. Community-level impacts of trawl selectivity in the Eastern Mediterranean Sea assessed using an ecosystem modelling approach. – ICES Journal of Marine Science, 77: 2918–2932.

Received 26 February 2020; revised 24 August 2020; accepted 25 August 2020; advance access publication 17 November 2020.

Switching from the traditional 40- or 44-mm diamond mesh codends to 50-mm diamond mesh (D50) or 40-mm square mesh (S40) codends is known to improve the size selection for commercial species and reduce discarding in the Mediterranean demersal trawl fisheries. This change has been recommended in the General Fisheries Commission for the Mediterranean and European Commission regulations. However, ecosystem-based assessment of improved selectivity still remains a challenge, and that is the aim of this study. For this purpose, an Ecopath model was developed and used to initialize dynamic simulations in Ecosim. The simulations changed gear-specific fishing mortality rates to represent the shifting from traditional codend to alternative codends. Our results suggest that the use of D50 or S40 codends would have a positive impact on the ecosystem as a whole and on the stock status of several commercial and non-commercial groups. Mixed trophic impact analysis indicated that, for species, like red mullet, the S40 codend was significantly better than the D50 codend. This information can be useful to policymakers, particularly for the areas where red mullet is the main target species and could inform trawl mesh options in the eastern Mediterranean.

Keywords:bottom trawl fisheries, ecological network analysis, Ecopath with Ecosim, ecosystem-based management, selectivity

Introduction

Fisheries management in the Mediterranean is complicated by the multispecies nature of the bottom trawl fisheries, which consti- tute a substantial portion of overall fishing effort. Achieving in- ternationally recognized targets, such as maximum sustainable yield (MSY), is challenging for fisheries that target multiple spe- cies because achieving the target for one species may mean

missing the target for others (Ulrichet al., 2017;Thorpe, 2019).

In areas such as the North Sea and the Baltic Sea, various solu- tions have been proposed including multispecies MSY (Horbowy and Luzenczyk, 2017) and “pretty good multispecies yield”

(Rindorfet al., 2017;Ulrichet al., 2017). Further extensions have been proposed to include economics, e.g. multispecies maximum economic yield (Hoshinoet al., 2018), and wider social objectives

VC International Council for the Exploration of the Sea 2020. All rights reserved.

For permissions, please email: [email protected]

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(2)

(Kempf et al., 2016). However, in areas such as the Mediterranean, implementation of such approaches may not be practical at present because of the lack of strong fisheries manage- ment. For example, although the trawl fisheries in the Eastern Mediterranean are regulated with a range of technical measures (i.e. area restrictions, effort control, seasonal closures and coastal exclusion zones, codend mesh regulations, and minimum landing sizes for the main commercial species), there are no total allow- able catch restrictions (Papaconstantinou and Farrugio, 2000).

These technical measures have been largely implemented based on single-species stock considerations. Nevertheless, there is a need to move these fisheries towards ecosystem-based manage- ment implying a need for a stronger focus on multispecies inter- actions (Sayguet al., 2020).

Because trawl mesh sizes affect the catchability of both target and non-target species, it is a key consideration in multispecies management approaches. A large number of studies exist on codend selectivity in demersal trawl fisheries, most of which high- light mesh size and shape as important factors (Stewart, 2002;

ICES, 2007). In the Mediterranean, 40- and 44-mm diamond mesh (D40 and D44, respectively) codends are standard but are characterized by poor selectivity resulting in substantial levels of discarding (Sa´nchezet al., 2004;Ordineset al., 2006;Salaet al., 2008;O¨ zbilginet al., 2015;Go¨kc¸eet al., 2016). Switching to 40- mm square mesh (S40) or 50-mm diamond mesh (D50) codends would improve the size selection for commercial species and re- duce bycatch (Sa´nchez et al., 2004; Bahamon et al., 2006;

Guijarro and Massutı´, 2006;Ordineset al., 2006;Lucchetti, 2008;

Sala et al., 2008; Ates¸ et al., 2010; Sala and Lucchetti, 2010;

O¨ zbilginet al., 2012,2015;Tokac¸et al., 2016;Go¨kc¸eet al., 2016).

The General Fisheries Commission for the Mediterranean (GFCM), the authority that makes binding recommendations for fisheries management in the Mediterranean (GFCM, 2007), has therefore recommended these changes and it has been amended by European Commission regulations (E.C, 2006). These regula- tions are mainly designed to improve the selectivity of the main target species such as red mullet (Mullus barbatus), common pan- dora (Pagellus erythrinus), and European hake (Merluccius mer- luccius), but this approach again ignores possible species interactions and the effects of changing codend selectivity on the wider ecosystem.

The Gulf of Mersin is one of the most important bottom trawl fishing grounds in the Eastern Mediterranean due to its wide and productive shelf (Gu¨cu¨ and Bingel, 1995). The bottom trawlers target mainly red mullet, brushtooth lizardfish (Saurida undos- quamis), common pandora, Randall’s threadfin bream (Nemipterus randalli), green tiger prawn (Penaeus semisulcatus), and speckled shrimp (Metapenaeus monoceros) (O¨ zbilgin et al., 2015). In terms of regulations, bottom trawling is prohibited from 15 April to 15 September (Official Gazette, 2016) and banned within 2–3 nm from the coast year-round. Currently, Turkish Fisheries Regulations prescribe a minimum D44 codend for the Mediterranean coast of Turkey (Official Gazette, 2016). As a consequence, commercial 44-mm diamond mesh handmade slack knotted (CD44) codends are used in the region (Eryas¸ar et al., 2014;O¨ zbilginet al., 2015). CD44 codends have poor selec- tivity and consequently a high discard rate (Eryas¸aret al., 2014;

O¨ zbilginet al., 2015), with an estimated 44% of the catches by weight being discarded (Sayguet al., 2020).

The aim of this study was to evaluate the potential effects on the Gulf of Mersin ecosystem of completely switching trawl

codends to those with improved selectivity by using an ecosystem model.

Methods

The main tool used in this study was an Ecopath model of the Gulf of Mersin developed by Saygu (2018) and Saygu et al.

(2020). The model extent covers the commercial bottom trawl fishing grounds in the continental shelf of the Gulf of Mersin (4352 km2), excluding the coastal area where trawling is banned (Figure 1).

Modelling approach

Ecopath with Ecosim version 6.6 (Christensen and Walters, 2004) was used to describe the average Gulf of Mersin ecosystem for 2009–2013 in Ecopath, and how the ecosystem might change over time, with changes in the gear-specific fishing mortality rate in Ecosim. Ecopath uses two master equations to ensure that the masses are balanced in the final model (Christensen and Walters, 2004;Christensen et al., 2005).

Consumption¼productionþrespirationþunassimilated food;

(1) Production¼predation mortalityþfishing mortalityþother

mortalityþbiomass accumulationþnet migration:

(2) This equation can be re-expressed as:

Pi¼YiþBiM2iþEiþBAiþPi ð1EEiÞ; (3) wherePiis the total production of functional group i;Yiis the total fishery catch rate ofi;M2i is the instantaneous predation mortality rate for groupi; Eiis the net migration rate (emigra- tion–immigration) fori;BAiis the biomass accumulation rate for i;EEiis the ecotrophic efficiency of the groupi; and 1EEis the other mortality rate that is not attributed to predation or catches.

EEexplains the proportion of the production that is utilized by the next trophic level through direct predation and fishing (Heymans et al., 2016).

Among the four basic parameters, namely biomass (B), pro- duction/biomass (P/B), consumption/biomass (Q/B), andEE, at least three have to be known or assumed, and one parameter can be estimated by the model. In addition, diets (proportions) and catch data (where appropriate) are needed as inputs for each functional group.

Based on initial parameters inherited from the Ecopath model, Ecosim simulates the food web dynamics over time using a set of differential equations (Christensen and Walters, 2004;

Christensen et al., 2005), expressed as:

dBi

dt ¼ P

Q iX

QjiX

QijþIi ðMiþFiþeiÞ Bi; (4) where dBi/dtis the growth rate of groupiduring timetin terms of its biomassBi; (P/Q)iis the net growth efficiency of groupi;

Qjiis the consumption of groupiover all of its preysj;Qijis the predation of groupiby all of its predatorsj;Iiis the immigration rate of groupi;Miis the non-predation (other) mortality rate of

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(3)

groupi;Fiis the fishing mortality rate of groupi; andeiis the em- igration rate of groupi(Christensen et al., 2005).

Baseline model

The baseline model (GoM-CD44), representing the structure of the ecosystem in the period 2009–2013, was parameterized using catch information for the CD44 codend currently used by bottom trawl fishers in the area. The input data of the baseline model are described in Saygu (2018)and Sayguet al.(2020)and given in Supplementary Table S1. The baseline model consists of 48 func- tional groups with the important commercial species: red mullet, brushtooth lizardfish, and common pandora, split into juvenile and adult age classes. Full information regarding the parameteri- zation of multi-stanza groups is provided in Supplementary Table S2.

TheP/Bratio relates to the turnover rate of a functional group.

For finfish, this is equal to the total mortality (Z, year1), which is equal to fishing mortality (F, year1) plus natural mortality (M, year1), all expressed as instantaneous annual rates (Allen, 1971).

F is easily calculated as the ratio of catch (C) over biomass (B) (F¼C/B) (Christensen and Walters, 2004).

The model was balanced following best practice guidelines (Heymans et al., 2016), ecological and thermodynamic rules (Christensen and Walters 2004; Darwall et al., 2010; Heymans

et al., 2016), and pre-balance diagnostics (Link, 2010; seeSaygu, 2018;Sayguet al., 2020).

Catch estimates

The main data on discards and landings, as well as estimates of gear selectivity for the main commercial species using the covered codend method (Eryas¸ar, 2014; O¨ zbilgin et al., 2015), were obtained from the project “Investigations to Improve Species and Size Selectivity in Mersin Bay Trawl Fisheries” (TUBITAK 1090684), which ran from 2009 to 2013. As part of that project, 182 valid hauls (Figure 1) were carried out using CD44 codend to calculate species-specific landings and discards based on catch per unit effort (CPUE) in the baseline model as described bySaygu et al.(2020). In addition, 23 valid hauls were conducted to esti- mate the size selection in the CD44 codend, 21 valid hauls using D50 codend and 23 valid hauls using S40 codend (Figure 1). All these hauls were conducted with codends fitted with fine-mesh covers (Eryas¸ar, 2014;O¨ zbilginet al., 2015). The resulting selec- tivity information was used to estimate catch values of multi- stanza groups using the selectivity-based catch calculation method described in detail in the baseline model (Saygu, 2018;

Sayguet al., 2020).

The selectivity-based catch calculation method demonstrated in Figure 2 was applied to six commercial species: red mullet, brushtooth lizardfish, common pandora, Randall’s threadfin Figure 1. Modelled area, Gulf of Mersin in the Eastern Mediterranean. Trawl hauls for the catch estimation of CD44 codend (red dots), D50 codend (green diamonds), and S40 codend (blue squares).

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(4)

Figure 2. Flow-diagram for the selectivity-based catch calculations.

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(5)

bream, goatfishes (Upeneus molucensisþUpeneus pori), and axil- lary seabream (Pagellus acarne). Weight, length, and selectivity parameters for these six species are provided inTable 1. Length frequency distributions of fish (FL) entering into the trawl (i.e.

codend þ cover) were calculated by combining all valid trawl hauls and were re-expressed by multiplying species-specific length–weight relationships (5) as weight distributions (6). For D50 and S40 codends, the distributions of weights at length (6) were multiplied with length-dependent retention probabilities (7) to calculate weight-dependent retention probabilities (8).

Rcodþcov, which is the weight of fish entering into the trawl (i.e.

codendþcover), was obtained from the baseline model (Saygu, 2018;Sayguet al., 2020) and multiplied by the percentage of fish in each size class as weight (9) to calculate the estimated retained weight for each size class in t km2year1(10). Finally, (11) was used to calculate the total catch of a species for single stanza func- tional groups (Randall’s threadfin bream, goatfishes and axillary seabream), (12) was used to calculate the catch of juvenile stanzas for red mullet, brushtooth lizardfish, and common pandora, and (13) was used to calculate the adult catch for the species split into stanzas.

For demersal fish groups for which no trawl mesh selectivity data were available, a negative correlation was assumed between CPUE and selectivity performance (Ordines et al., 2006;

Bahamonet al., 2007). Therefore, the catch data of D50 and S40 codends were estimated from CPUE (CPUE-based catch calcula- tion) values based on the 21 and 23 valid hauls in the research trawl surveys respectively (Saygu, 2018; Saygu et al., 2020).

Anglerfishes, pufferfishes, demersal sharks, and “demersal rays and skates” were excluded from this assumption because these species cannot escape from the trawl codend and therefore their catchability does not depend on the type of trawl codend. The different methodological approaches used in the catch estima- tions explained above are detailed inTable 2.

Model structure and simulated codend scenario models The gear-specific fishing mortality rates (F, year1) representing improved trawl selectivity with the D50 and S40 codends were calculated as a ratio between catch (C) and biomass (B) (Christensen and Walters, 2004). Trawl catches (landing and dis- card), the mortality rates of the alternative codends along with the gear-specific fishing mortality rates, and biomasses of the baseline model (GoM-CD44) are given inTable 2.

We developed a novel approach using Ecosim to simulate how the food web might change if demersal trawls switched to D50 and S40 codends (Figure 3). Previous studies have used Ecosim for similar purposes to simulate the impact of bycatch reduction devices (Criales-Hernandez et al., 2006) and im- proved gear selectivity (Celicet al., 2018). In those studies, val- ues in Ecopath (e.g. landings, discards or fleet structure) were changed prior to simulating dynamics over time. This can make interpretation of the results more complicated because the un- derlying model itself has been altered. Instead, we took advan- tage of the recently added capacity to change species-gear catchability rates in Ecosim version 6.6. Using this new feature, it was possible to change the fishing mortality exerted upon functional groups by specific fleets within the temporal simula- tion, without having to alter the baseline fishing mortality val- ues. The baseline Ecopath model was transferred into the Ecosim module as the initial state. Initial gear-specific fishing mortality rates were applied for the first 5 years of the Ecosim simulation. After this 5-year spin-up period, fishing mortality rates were changed to the gear-specific fishing mortality rates of D50 and S40 codends (Table 2). The food web continued to be simulated forward for an additional 10 years under the new fishing mortalities, allowing the food web to shift into a new state. Other Ecosim parameters, which are often adjusted when fitting model simulations to observed data, such as foraging time adjustment rates and vulnerabilities, were left at their de- fault values. New balanced Ecopath models (GoM-D50 and GoM-S40) were then extracted from the last year of the Ecosim simulation (15th year), representing the ecosystem that has shifted into, and stabilized in its new state. Finally, we com- pared the diagnostics for the initial Ecopath model (GoM- CD44) with these extracted Ecopath snapshots (GoM-D50 and GoM-S40 models). This approach has been conceptualized in Figure 3. The input parameters of GoM-D50 and GoM-S40 models are presented inSupplementary Table S3.

Ecosystem and fishing indicators

The following ecological and fisheries indicators were calculated for each Ecopath model and compared to assess the overall impacts of changing trawl codend selectivity on the ecosystem and functional groups:

Mean trophic level of the catch (mTLc) is an indicator of the average trophic level of the total catch (Paulyet al., 1998):

Table 1.Parameters used in catch calculation process.

Species Red mullet Brushtooth lizardfish Common pandora Randall’s threadfin Goatfishes Axillary seabream L–W relationship

Intercept (a) 0.0067 0.0029 0.0109 0.0011 0.0050 0.0090

Slope (b) 3.1789 3.2664 3.0536 3.0610 3.2890 3.1380

Ref. (Ok, 2012) (Ok, 2012) (Ok, 2012) (Ergudenet al., 2010) (Ok, 2012) (Soykanet al., 2015) Selectivity regression parameters (Eryas¸ar, 2014;O¨ zbilginet al., 2015)

D50

v1 5.113 5.528 5.950 2.323 3.856 2.172

v2 0.418 0.186 0.408 0.200 0.181 0.113

S40

v1 11.648 9.154 15.688 8.323 6.285 11.075

v2 0.836 0.374 1.207 0.574 0.385 0.799 Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(6)

mTLc¼ P

iTLiYi

P

iYi

; (14)

where TLiis the trophic level of group iand Yiis the catch of groupi.

The exploitation status of the ecosystem was assessed using several indices: the exploitation rates of the functional groups (E, the proportion of the total mortality ascribed to fishing); the cu- mulative exploitation rate [CumE, (15)]; and the percentage of Primary Production Required to sustain the total catch [PPRc, considering PP and detritus, (16)] normalized by total primary production (PPRc/TPp).

The CumE by different trawl codend was calculated following Agnettaet al.(2019):

CumE¼Xn

i¼1Ei; (15)

whereEis the exploitation rate of functional groupi.

PPRc was calculated following Christensen and Pauly (1993) andPauly and Christensen (1995):

PPRc¼X

Paths

Y Y

Pred;Prey

Qpred

Ppred

DC0Pred;prey

" #

; (16)

whereYis the catch of given group,Pis the production,Qis the con- sumption, andDC0 is the diet composition for each predator–prey interaction in each path, with cycles removed from the diet composi- tions. PPRc/TPp values were used to assess the ecological expense of fisheries in the ecosystem in response to increasing selectivity.

Mixed trophic impact (MTI) analysis was used to analyse the rel- ative direct and indirect impacts of bottom trawling on any func- tional group by applying an hypothetical increase (Ulanowicz and Puccia, 1990). MTI is a form of sensitivity analysis, which shows the relative impact (positive or negative) that a change in the biomass of a given group would have on the biomass of other groups, pro- vided that the trophic structure in the system does not change (Christensen and Pauly, 1992). MTI ranges from1 toþ1, and values close to these limits indicate strong effects. MTI was calcu- lated followingUlanowicz and Puccia (1990):

b c ¼M ½ I ½ Q1

½ ;I (17) whereMis all the MTIs that occur in the food web,Qis the net impact matrix involving all impacts, andIis the identity matrix.

Table 2.Input parameters of the simulated codend scenario models.

# Functional groups

Baseline model

D50 Trawl codend

S40

Trawl codend

B Ftrawl LD50 DD50 FD50 LS40 DS40 FS40 CE

6 Alien shrimps 0.09 0.31 0.012 0.001 0.15 0.010 0.001 0.12 CB

7 Shrimps 0.03 0.45 0.001 0.001 0.04 0.001 0.000 0.02 CB

8 Swimming crab 0.08 0.39 0.000 0.014 0.16 0.000 0.010 0.12 CB

9 Crabs 0.03 0.48 0.000 0.002 0.06 0.000 0.005 0.17 CB

11 Octopuses and cuttlefish 0.05 0.50 0.018 0.002 0.43 0.018 0.005 0.49 CB

12 Squids 0.01 0.34 0.003 0.001 0.44 0.003 0.000 0.35 CB

13 Red mullet (Juvenile) 0.01 1.02 0.009 0.000 0.69 0.004 0.000 0.30 SB

14 Red mullet (Adult) 0.11 0.46 0.042 0.000 0.38 0.041 0.000 0.37 SB

15 Striped red mullet 0.00 0.25 0.000 0.000 0.25 0.000 0.000 0.25 CB

16 Goatfishes 0.01 0.28 0.001 0.000 0.04 0.000 0.000 0.04 SB

17 Anglerfishes 0.00 0.55 0.000 0.000 0.55 0.000 0.000 0.55 NS

18 Common sole 0.01 0.43 0.002 0.000 0.37 0.003 0.000 0.43 CB

19 Flatfishes 0.01 0.48 0.000 0.001 0.04 0.000 0.002 0.17 CB

20 Hake 0.01 0.59 0.006 0.000 0.59 0.006 0.000 0.59 CB

21 Brushtooth lizardfish (Juvenile) 0.00 0.60 0.001 0.000 0.12 0.000 0.000 0.05 SB

22 Brushtooth lizardfish (Adult) 0.04 0.24 0.002 0.000 0.05 0.003 0.000 0.07 SB

23 Pufferfishes 0.01 0.53 0.000 0.005 0.53 0.000 0.005 0.53 NS

24 Small demersal fishes with high discard rate 0.04 0.50 0.001 0.002 0.07 0.001 0.001 0.07 CB

25 Large demersal fishes with high discard rate 0.02 0.39 0.000 0.001 0.03 0.000 0.001 0.05 CB

26 Demersal fishes with high retention rate 0.00 0.46 0.001 0.000 0.26 0.002 0.000 0.46 CB

27 Alien demersal fishes 0.05 0.26 0.000 0.001 0.02 0.000 0.001 0.01 CB

28 Common pandora (Juvenile) 0.00 1.16 0.003 0.000 0.69 0.002 0.000 0.64 SB

29 Common pandora (Adult) 0.03 0.22 0.005 0.000 0.16 0.006 0.000 0.21 SB

30 Randall’s threadfin bream 0.02 0.29 0.003 0.000 0.20 0.002 0.000 0.14 SB

31 Axillary seabream 0.02 0.45 0.004 0.000 0.17 0.004 0.000 0.18 SB

32 Benthopelagic fishes 0.01 0.64 0.000 0.000 0.00 0.000 0.001 0.20 CB

33 Picarels and bogue 0.02 0.62 0.001 0.000 0.04 0.002 0.000 0.11 CB

34 Demersal sharks 0.00 0.46 0.000 0.001 0.46 0.000 0.001 0.46 NS

35 Demersal rays and skates 0.05 0.37 0.000 0.017 0.37 0.000 0.017 0.37 NS

Biomass of the baseline Ecopath model (B); gear specific fishing mortality rates of the baseline model (Ftrawl) and of the D50 and S40 codends (FD50 and FS40, respectively); trawl landings (LD50 and LS40) and discards (DD50 and DS40) for the D50 and S40 codends, respectively, used for F calculations (F=(LþD)/B) and methodological differences in the catch estimations by functional groups (CE): selectivity-based catch calculation method (SB); CPUE-based catch calculation method (CB); assuming non-selectivity (NS).

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(7)

TheQmatrix has elementsqi, which is the differences between the positive effectsdji, the fraction of the preyiin the diet of the predator jand negative effectsfij, the fraction of total consumption ofiused by predatorj(Ulanowicz and Puccia, 1990;Libralato et al., 2006).

The overall impact (ei) derived from the net MTI (mij) was cal- culated as:

ei¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffi Xn

j6¼i

m2ij vu

ut ; (18)

in which the effect of the change in biomass on the group itself (i.e.mii) is not included (Libralato et al., 2006).

Applying the definition given byLibralato et al. (2006), key- stoneness (KSi) (Poweret al., 1996) indicates the ability of a func- tional group with a low biomass to influence other functional groups and therefore quantifies its structural role in the food web. Keystoneness was calculated as follows:

KSi¼log½eið1piÞ; (19) wherepiis the contribution of the group to the total biomass of the system.

Uncertainty analysis

We applied uncertainty analysis to understand the impact of in- put parameter sensitivity on outputs using a Monte Carlo (MC) approach (Kennedy and O’Hagan, 2001). Parameters were assigned confidence intervals based on data origin and pedigree (Christensen and Walters, 2004). The confidence intervals ofB, P/B,Q/B, landings and discards were obtained from the baseline model (Saygu, 2018; Sayguet al., 2020). However, we assigned confidence intervals of 0.1 to allP/Bratios because higher confi- dence intervals (>0.1) generatedP/Qratios outside of the rec- ommended 0.1–0.3 range (Link, 2010). The MC routine built into EwE was used to generate 500 plausible iterations of each GoM model using the EcoSampler plugin to store model results (Steenbeeket al., 2018). To assess the consequences of parame- ter uncertainty, estimates of MTI, mTLc, and PPRc/TPp were extracted for the calculation of 95% confidence intervals.

Outputs and uncertainties from the GoM-CD44, GoM-D50, and GoM-S40 models were then compared. Model results were considered to be significantly different if there was no overlap between the 95% confidence intervals for the indicators (Hines et al., 2018).

Figure 3. Conceptual diagram depicting the steps taken to simulate the impacts of different trawl codends on the Gulf of Mersin ecosystem using EwE.

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(8)

Results

The outputs of the baseline Ecopath model (GoM-CD44) and detailed outputs from the Ecopath models (GoM-D50 and GoM-S40) extracted from Ecosim and used to compare with the initial state (GoM-CD44) are given inSupplementary Table S4.

Fishing impacts

Compared with the baseline model, switching to the alternative codends (GoM-D50 and GoM-S40) decreased trawl landings by 36 and 35%, and trawl discards by 56 and 54%, respectively (Table 3). In relation to the ecosystem indicators, comparison of

the three models showed that the GoM-D50 and GoM-S40 mod- els produced lower CumE estimates when compared to the GoM- CD44 model (Table 3). The mTLc increased from 3.29 to 3.34 in both GoM-D50 and GoM-S40 models and PPRc/TPp decreased as selectivity increased, but the changes in both mTLc and PPRc/

TPp were not significant (Table 3).

The exploitation rates (which include landings and discards, E¼F/Z) of several demersal groups were reduced in both selec- tivity improvement models (Figure 4). Switching from the tradi- tional CD44 codends to more selective D50 or S40 codends resulted in decreases in exploitation rates for juvenile red mullet, juvenile and adult brushtooth lizardfish, and juvenile common Table 3.Fishery indicators for the models

Parameter

Ecopath models

GoM-CD44 GoM-D50 GoM-S40

Total trawl catches (t km2.year1) 0.316 0.174 0.179

Total trawl landing (t km2.year1) 0.176 0.112 0.115

Total trawl discard (t km2.year1) 0.140 0.062 0.064

CumE 12.09 7.79 8.14

mTLc 3.29 (3.27–3.31) 3.34 (3.30–3.36) 3.34 (3.31–3.37)

PPRc/TPp 6.79 (5.24–8.70) 5.41 (4.60–7.78) 5.46 (4.66–8.14)

Cumulative exploitation rate (CumE); mean trophic level of the catch (mTLc); primary production to sustain the catch (PPRc); total primary production (TPp).

Brackets indicate 95% confidence intervals calculated using Monte Carlo approach.

Figure 4. The reduced exploitation rates of several demersal functional groups in the original model GoM-CD44, and the two new models GoM-D50 and GoM-S40: Juvenile (J.); adult (A.).

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(9)

pandora. However, the exploitation rates of the adult stanzas of red mullet and common pandora (Figure 4), “octopuses and cuttlefish”, squids, and common sole (Supplementary Table S4) were not so strongly affected.

The MTI analysis highlighted that switching to more selective trawl codends caused significant positive changes in the mixed trophic effects of bottom trawl fisheries. It shows a decreased im- pact of the fishery on a number of demersal functional groups in- cluding shrimps, swimming crab, crabs, small demersal fishes with high discard rate, large demersal fishes with high discard rate, demersal fishes with high retention rate, alien demersal fishes, axillary seabream, benthopelagic fishes, and “picarels and bogue” (Figure 5).

In the multi-stanza groups, the impact of trawling on juvenile red mullet was significantly reduced (decreased from 0.30 to 0.14) in the use of S40 codend (GoM-S40 model), but there was no significant change when using a D50 codend (GoM-D50 model). In addition, there was no significant change for red mul- let adults in both alternative models (Figure 5). For brushtooth lizardfish, there was a significant change in the impact of trawling on both juvenile and adult stanzas in the GoM-D50 and GoM- S40 models, going from a negative impact to positive for juveniles and reducing the negative impact for adults (Figure 5). While the juvenile stanza of common pandora was slightly positively af- fected by improved trawl selectivity with the D50 and S40

codends (significantly decreased from0.38 to0.27), no signif- icant change was predicted for its adult stanza in either alternative models (Figure 5).

Simulated change in biomass

The simulations in Ecosim, with the changes in gear-specific fish- ing mortality rate, show that shifting from the initial state to the new state representing a switch to trawl codends with improved selectivity increased the biomasses of several functional groups.

Specifically, it increased the biomass of lizardfish, small demersal fishes with high discard rate, large demersal fishes with high dis- card rate, demersal fishes with high retention rate, axillary seab- ream, and “picarels and bogue” when shifting to D50 codend (Figure 6). Shifting to S40 codend increased the biomass of red mullet, lizardfish, small demersal fishes with high discard rate, large demersal fishes with high discard rate, and “picarels and bogue” (Figure 6). However, the codend switches were also asso- ciated with decreases in the biomasses of some other functional groups, such as alien shrimps and Randall’s threadfin bream (Figure 6).

Keystoneness

In the models, no functional groups had keystoneness indices greater than zero, which would have identified them as keystone Figure 5. The Mixed Trophic Impacts (MTI) of bottom trawl fisheries on the different functional groups in the Gulf of Mersin models. Circles indicate means of the MTI values and error bars are 95% confidence intervals calculated from 500 plausible mass balanced models generated using Monte Carlo approach: Juvenile (J.); adult (A.); the baseline model (GoM-CD44); D50 codend scenario model (GoM-D50); S40 codend scenario model (GoM-S40).

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(10)

species according toLibralato et al. (2006). However, common bottlenose dolphin had relatively low biomass but high overall impact and therefore could be considered as a keystone species (Figure 7). In the GoM-D50 and GoM-S40 models, the keystone- ness indices of common bottlenose dolphin were closer to zero compared with the baseline model (Figure 7). In addition, adult brushtooth lizardfish, which had the second highest overall im- pact among predators in the food web and is targeted for trawling in the region, showed an increase in keystoneness in response to a switch to more selective trawl codends (Figure 7).

Discussion

Fishing impacts

In our study, bottom trawl landings and discards were substan- tially reduced by changing trawl codends from commercial 44- mm diamond mesh handmade slack knotted codends (CD44) to 50-mm diamond mesh (D50) or 40-mm square mesh (S40) codends. Such reductions in landings and discards may be antici- pated to have positive impacts at the ecosystem and population levels. Previous studies have identified discarding as a threat to key species, as a waste of resources, and as negatively impacting the food web and ecosystem pathways (Hall et al., 2000; Coll et al., 2008a; Sarda` et al., 2015). Therefore, the European Commission instituted a discard ban under the Common Fishery Policy to promote sustainable fisheries [Regulation (EU) 1380/

2013; EU, 2013]. However, debates around the implementation of the landing obligation suggest that it could actually increase fishing mortalities as landing all catches will include previously

discarded species, a proportion of which might have survived af- ter being thrown back into sea (Sarda`et al., 2015). In addition, discards can be consumed by species such as sea turtles and sea- birds (Sarda`et al., 2015). Therefore, the landing obligation could lead to unintended negative consequences for the ecosystem (Celic et al., 2018) and affect species of conservation interest (Heathet al., 2014). Increasing gear selectivity is also identified as an aim of the reformed CFP and might more effectively reduce discarding, particularly in the Mediterranean (Sarda`et al., 2015;

Celicet al., 2018).

Comparing Ecopath models showed that the CumEs and the exploitation rates of a number of demersal groups decreased when using the more selective trawl gears. Reducing exploitation rates of target and non-target stocks in an exploited ecosystem should result in rebuilding biomass in the long term (Worm et al., 2009;Worm and Branch, 2012). Our MTI results show that more selective gears had less direct and indirect impacts on a number of important species in the area. Therefore, changes in the gear-specific fishing mortality rate cause an increase in bio- mass over time, as shown in our simulations (Figure 6).

Moreover, the more selective trawl fisheries led to slightly higher mTLc values, which have been used to explore overfishing risks within ecosystems (Paulyet al., 1998;Pauly and Palomares, 2005;

Libralatoet al., 2008), and reduced the PPRc/TPp values indicat- ing that primary production was converted more efficiently to the catch (Cury et al., 2005), although the confidence intervals for these predictions were quite large in our simulations.

As a consequence, this study suggests that moving away from traditional CD44 codends and towards more selective trawl Figure 6. Simulated change in the biomasses (relative to that of the initial state with codend CD44) of functional groups when using D50 (red line) and S40 (blue line) codends: Juvenile (J.); adult (A.).

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(11)

fisheries with D50 or S40 codends would have positive impacts on both commercial and non-commercial species and the ecosys- tem at large. This result also supports the case for implementing the legal regulations by the European Commission (E.C, 2006) and the GFCM (GFCM, 2007) requiring the use of S40 or D50 codends.

However, for red mullet, one of the most commercially impor- tant species in the demersal trawl fisheries of the Mediterranean, particularly for Italy, Tunisia, Greece, and Turkey (FAO, 2011), the transition to only S40 codend is advisable. As the MTI analy- sis shows, the transition to a D50 codend would not provide an advantage to juvenile red mullet, although size selectivity studies on red mullet indicate that both S40 and D50 codends are more selective than the small diamond mesh codends traditionally used in the Mediterranean (Salaet al., 2008,2015;Tokac¸et al., 2016) and in the Gulf of Mersin (O¨ zbilginet al., 2015).

As seen inFigure 5, the mixed trophic interactions in our find- ings predict significant positive changes for both juvenile and adult lizardfish in the implementation of D50 or S40 codends in the bottom trawl fisheries, with the S40 codend showing best results for juveniles. This is in agreement with selectivity studies showing that D50 and S40 codends retain less juvenile and adult lizardfish compared with CD44 (O¨ zbilginet al., 2015). Lizardfish is a predatory invasive species and is targeted by trawlers in the Eastern Mediterranean (O¨ zbilginet al., 2015;Go¨kc¸eet al., 2016).

The MTI matrix also predicts that positive impacts for lizardfish under more selective codends should not have unintended conse- quences for commercial bottom trawl species via increased

predation. This is likely due to its diet preferences, as a recent study in the modelled area revealed that more than90% of the liz- ardfish diet is comprised of small pelagic fishes (O¨ zyurt et al., 2017). However, the impacts of lizardfish on the small pelagic fishes and therefore on the food web should be further investi- gated because of the ecological importance of small pelagics in the research area (Sayguet al., 2020).

Given the same levels of fishing effort, the models suggest that the use of D50 or S40 codends in the bottom trawl fisheries of the Gulf of Mersin would reduce overall landings by 36 and 35%, re- spectively. Thus, a switch to more selective mesh nets might be resisted by the industry, at least in the short term.O¨ zbilginet al.

(2015) predicted short-term commercial losses in the Gulf of Mersin of 21% if implementing D50 codends and 17% using S40 codends considering seven commercial species (red mullet, brushtooth lizardfish, common pandora, goldband goatfish, Randall’s threadfin bream, green tiger prawn and speckled shrimp).Eryas¸ar and O¨ zbilgin (2015)also showed a seasonal ef- fect where these commercial losses might increase to 40% towards the end of the fishing season. In comparison, the short-term eco- nomic losses of commercial species in the shallow fishing areas of the Catalan Sea were predicted to be between 12 and 33%

(Bahamon et al., 2006) and 12% in the central Adriatic Sea (Lucchetti, 2008) when using S40 codend instead of commercial diamond codend (D40). The level of commercial losses related to the adoption of more selective trawl gear is therefore affected by fishing grounds, fishing season, catch composition, and landing price, but from the above examples is likely to be in the 17–40%

Figure 7. The keystoneness and the relative overall effect of the functional groups in the models: Brushtooth lizardfish (22); common bottlenose dolphin (46); the baseline model (GoM-CD44); D50 codend scenario model (GoM-D50); S40 codend scenario model (GoM-S40).

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(12)

range. However, providing overall fishing effort can be con- strained, switching to more selective trawls should increase the biomass of commercial species in the longer term, as have been shown in single-species evaluations (Bahamonet al., 2006) and in an ecosystem framework (Collet al., 2008b;Celicet al., 2018) in the wider Mediterranean. Larger stocks are generally considered as a desirable management outcome since they provide an in- creased buffer against over-exploitation or periods of unfavoura- ble environmental conditions. In addition, selective fisheries can be an important solution to compensate for the unfavourable economic and ecological consequences of the landing obligation (Celicet al., 2018). When time series data become available, these aspects should be further explored for the Gulf of Mersin using dynamic ecosystem models coupled to an economics model, to advise on the longer-term bio-economic impacts of changing to more selective fishing gears.

Keystoneness indices

Keystone analysis demonstrated that improved trawl selectivity could cause an increase in the keystoneness of common bottle- nose dolphin. This is likely to be due to the dolphin’s direct nega- tive effects on its prey including small pelagic fishes and squids, its indirect negative effect on the prey’s predators such as lizard- fish, seabirds, and squids, and indirect positive effect on the prey of their prey, such as macrozooplankton, and predators of their prey, such as benthopelagic fishes (predators of macrozooplank- ton) (Saygu, 2018;Sayguet al., 2020). Dolphins have also been identified as a keystone species in other Mediterranean studies (Collet al., 2006;Piroddiet al., 2010). Dolphin populations have been declining in most ecosystems, and it is suggested that they can be used as an indicator of ecosystem changes and fishing im- pact due to their vulnerability to reduced prey availability, which is mainly caused by the increasing fishing pressure on small pe- lagic species (Piroddiet al., 2011). According to a time dynamic ecosystem model created by Collet al.(2008b), the biomass of small pelagic fishes could slightly increase in response to im- proved trawl selectivity. Moreover, dolphin biomass increases un- der different fisheries management scenarios such as fishing closures and a closure of both purse seining and trawling are more efficient than only closing purse seining (Piroddi et al., 2011). Dolphins also increase relatively quickly under non-fishing scenarios (no-take zone) in models of other parts of the Mediterranean (Collet al., 2009;Piroddiet al., 2011). Thus, we predict that the adoption of the D50 or S40 codends would in- crease the keystoneness of dolphins.

The keystoneness of adult lizardfish also increased with the im- plementation of a more selective trawl codend. Its functioning in the food web and the importance to the trawl fisheries have been previously shown in the Eastern Mediterranean (Corrales et al., 2017a,b, 2018;Sayguet al., 2020). Therefore, using more selective codends may raise the role of the lizardfish within the food web by reducing fishing mortality on the juvenile stage in particular.

The keystoneness of lizardfish was also identified on the Israeli coast of the Eastern Mediterranean (Corraleset al., 2017a) where a biomass increase was predicted in response to decreased fishing mortality (Corraleset al., 2018). Future modelling work should test the indirect impacts of lizardfish on the food web to see if an potential increase in lizardfish biomass would have positive or negative ecosystem impacts in the Eastern Mediterranean.

Model assumptions and future work

This study is based on the baseline model of the Gulf of Mersin bySaygu (2018)andSayguet al.(2020)combined with trawl se- lectivity data fromEryas¸ar (2014)andO¨ zbilginet al.(2015). The new models thus include the assumptions from those previous studies. In particular, the catchability coefficient of bottom trawl surveys used in the biomass and catch estimations was assumed to be 1.0. The biomass estimation of demersal groups was calcu- lated using the swept area method, and the wing spread of the trawl net was used instead of the length of the head-rope which is unknown. In addition, post-escape survival rates for demersal species were assumed to be 100%. Post-escape mortality informa- tion plays an important role when evaluating the consequences of selectivity, because without such information, the fishing mortal- ity resulting from trawl fisheries is assumed to be higher (Coll et al., 2008b). It is usually assumed that post-trawl escape survival will be considerably <100% due to physical damage sustained when fish contact the fishing gear (Broadhurst et al., 2006, Du¨zbastılar et al., 2017). It is known that for S40 and D50 codends, post-escape survival rates are higher compared with the CD44 codend, at least for red mullet (Du¨zbastılar et al., 2010, 2017). In our models, we assumed that post-discard survival rates were zero, as there are little data on this aspect from the Mediterranean. Post-discard survival rates should be further in- vestigated and included in future modelling studies.

In our models, the main commercial species were split into two stanzas: juveniles and adults. Although using more than two groups could improve the modelling of trawl selectivity, this would require additional data on diets, energetics and predators of each group, plus the breakdown of these groups in both the landings and discards data. The present status of data from the Gulf of Mersin does not permit such detailed modelling.

We applied the selectivity-based catch calculation method de- scribed in Figure 2 for some commercial species, which were abundant enough in the demersal trawl fisheries to enable selec- tivity estimation. However, for some species, particularly the non-commercial species, selectivity was estimated using assump- tions based on the correlation between biomass index and selec- tivity performance. Therefore, further studies on selectivity, particularly for the non-commercial species, are required.

One of the major limitations to extending this research is the absence of time series data for the Gulf of Mersin. Such data are needed to calibrate dynamic temporal predictions using models such as Ecosim (Christensen and Walters, 2004). The develop- ment of such dynamic models for this region should be encour- aged because they can be used to explore multiple management options including changes in fishing effort, landing obligations, and even the combination of management alternatives such as the combination of effort reduction and selective fisheries.

Moreover, while assessing alternative management options, we also should consider cumulative impacts of multiple stressors such as invasive species and climate change, and the long-term economic impacts of these possible implications.

Conclusions

Our models showed that the implementation of more selective trawl codends in demersal trawl fisheries in the Eastern Mediterranean should reduce discards and the exploitation rates of several important functional groups including commercial spe- cies. In addition, MTI analysis showed that the overall negative

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(13)

impact of demersal trawl fisheries on several commercial func- tional groups, such as shrimps, red mullet, brushtooth lizardfish, demersal fishes with high retention rate, common pandora, and

“picarels and bogue”, and non-commercial functional groups, such as crabs, small demersal fishes with high discard rate, and large demersal fishes with high discard rate, should decrease with the use of D50 or S40 codends. Thus, improved trawl selectivity through these alternative codends would benefit the ecosystem and functional groups in the Eastern Mediterranean and would be expected to lead to biomass increases in some demersal spe- cies. The major caveat is that a reduction in the catches of the commercial species is expected in the short term and this is likely to affect the willingness of fishers to adopt more selective trawls.

Furthermore, if a switch to more selective trawls was made, over- all fishing effort would have to be carefully managed to avoid a compensatory increase.

Our models show that the S40 trawl codend appears to be the better alternative for red mullet than the D50 codend. This type of information can be useful to the policymakers, particularly in the areas where red mullet is main target species and may con- tribute to addressing selectivity options in the Mediterranean.

This study underlines the need to develop a better understand- ing of trawl selectivity in the Eastern Mediterranean. Future efforts should thus be directed at experimental studies of escape and discard survival, but also data collection to generate reliable time series of catches and biomasses. Such data would allow the development of dynamic ecosystem models for the region that can be used to address a wide variety of management questions, including the longer-term effects of changing trawl selectivity within an ecosystem-management context.

Supplementary data

Supplementary material is available at the ICESJMSonline ver- sion of the manuscript.

Acknowledgements

IS would like to address a special thanks to Barıs¸ Saliho_ glu, Ekin Akoglu, Nazlı Demirel, Xavier Corrales, Sinan Mavruk, Fethi Bengil, and Marieke Steuben for their advice. The authors also thank to two anonymous reviewers for their helpful comments.

Funding

The article is based on work in the PhD thesis of I_S, funded by the Scientific and Technological Research Council of Turkey (TUBITAK) on the TUBITAK 2214-A, Programme 2014-2 and supported by the Scottish Association for Marine Science (SAMS) as host institution. In addition, the main data of this study were obtained from the project 109O684 financed by TUBITAK.

Data availability statement

The data underlying this article are available in the article and in its onlinesupplementary material.

References

Agnetta, D., Badalamenti, F., Colloca, F., D’Anna, G., Di Lorenzo, M., Fiorentino, F., Garofalo, G.et al.2019. Benthic-pelagic cou- pling mediates interactions in Mediterranean mixed fisheries: an ecosystem modeling approach. PLoS One, 14: e0210659.

Allen, K. R. 1971. Relation between production and biomass. Journal of the Fisheries Research Board of Canada, 28: 1573–1581.

Ates¸, C., Deval, M. C., Bo¨k, T., and Tosunoglu, Z. 2010. Selectivity of diamond (PA) and square (PE) mesh codends for commercially important fish species in the Antalya Bay, eastern Mediterranean.

Journal of Applied Ichthyology, 26: 465–471.

Bahamon, N., Sarda`, F., and Suuronen, P. 2006. Improvement of trawl selectivity in the NW Mediterranean demersal fishery by us- ing a 40 mm square mesh codend. Fisheries Research, 81: 15–25.

Bahamon, N., Sarda, F., and Suuronen, P. 2007. Potential benefits from improved selectivity in the northwest Mediterranean multi- species trawl fishery. ICES Journal of Marine Science, 64:

757–760.

Broadhurst, M. K., Suuronen, P., and Hulme, A. 2006. Estimating collateral mortality from towed fishing gear. Fish and Fisheries, 7:

180–218.

Celic, I., Libralato, S., Scarcella, G., Raicevich, S., Marceta, B., and Solidoro, C. 2018. Ecological and economic effects of the landing obligation evaluated using a quantitative ecosystem approach: a Mediterranean case study. ICES Journal of Marine Science, 75:

1992–2003.

Christensen, V., and Pauly, D. 1992. ECOPATH II—a software for balancing steady-state ecosystemmodels and calculating network characteristics. Ecological Modelling, 61: 169–185. International Center for Living Aquatic Resources Management, Manila Philippines; the International Council for the Exploration of the Sea, Copenhagen, Denmark; and the Danish International Development Agency, Copenhagen, Denmark.

Christensen, V., and Pauly, D. 1993. Flow characteristics of aquatic ecosystems.InTrophic Models of Aquatic Ecosystems. ICLARM Conference Proceedings, 26, pp. 338–352. Ed. by V. Christensen and D. Pauly. Manila.

Christensen, V., and Walters, C. J. 2004. Ecopath with Ecosim: meth- ods, capabilities and limitations. Ecological Modelling, 172:

109–139.

Christensen, V., Walters, C. J., and Pauly, D. 2005. Ecopath with Ecosim: A User’s Guide. Fisheries Center, University of British Columbia, Vancouver. 154 pp.

Coll, M., Bahamon, N., Sarda`, F., Palomera, I., Tudela, S., and Suuronen, P. 2008b. Improved trawl selectivity: effects on the eco- system in the South Catalan Sea (NW Mediterranean). Marine Ecology Progress Series, 355: 131–147.

Coll, M., Libralato, S., Tudela, S., Palomera, I., and Pranovi, F. 2008a.

Ecosystem overfishing in the ocean. PLoS One, 3: e3881.

Coll, M., Palomera, I., Tudela, S., and Sarda, F. 2006. Trophic flows, ecosystem structure and fishing impacts in the South Catalan Sea, Northwestern Mediterranean. Journal of Marine Systems, 59:

63–96.

Coll, M., Palomera, I., and Tudela, S. 2009. Decadal changes in a NW Mediterranean Sea food web in relation to fishing exploitation.

Ecological Modelling, 220: 2088–2102.

Corrales, X., Coll, M., Ofir, E., Heymans, J. J., Steenbeek, J., Goren, M., Edelist, D.et al.2018. Future scenarios of marine resources and ecosystem conditions of the Eastern Mediterranean under impacts of fishing, alien species and sea warming. Scientific Reports, 8: 14284.

Corrales, X., Coll, M., Ofir, E., Piroddi, C., Goren, M., Edelist, D., Heymans, J. J. et al. 2017b. Hindcasting the dynamics of an Eastern Mediterranean marine ecosystem under the impacts of multiple stressors. Marine Ecology Progress Series, 580: 17–36.

Corrales, X., Ofir, E., Coll, M., Goren, M., Edelist, D., Heymans, J. J., and Gal, G. 2017a. Modeling the role and impact of alien species and fisheries on the Israeli marine continental shelf ecosystem.

Journal of Marine Systems, 170: 88–102.

Criales-Hernandez, M. I., Duarte, L. O., Garcı´a, C. B., and Manjarre´s, L. 2006. Ecosystem impacts of the introduction of bycatch reduc- tion devices in a tropical shrimp trawl fishery: insights through simulation. Fisheries Research, 77: 333–342.

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(14)

Cury, P. M., Shannon, L. J., Roux, J., Daskalov, G. M., Jarre, A., Moloney, C. L., and Pauly, D. 2005. Trophodynamic indicators for an ecosystem approach to fisheries. ICES Journal of Marine Science, 62: 430–442.

Darwall, W. R. T., Allison, E. H., Turner, G. F., and Irvine, K. 2010.

Lake of flies, or lake of fish? A trophic model of Lake Malawi.

Ecological Modelling, 221: 713–727.

Du¨zbastılar, F. O., O¨ zbilgin, H., Aydın, C., Metin, G., Ulas¸, A., Lo¨k, A., and Metin, C. 2010. Mortalities of fish escaping from square and diamond mesh codends in the Aegean Sea. Fisheries Research, 106: 386–392.

Du¨zbastılar, F. O., Breen, M., Aydın, C., O¨ zbilgin, H., O¨zgu¨l, A., Ulas¸, A., Metin, G.et al.2017. Seasonal variation in mortality of red mullet (Mullus barbatus) escaping from codends of three dif- ferent sizes in the Aegean Sea. Scientia Marina,81: 339–349.

E.C. 2006. Council regulation (EC 1967/2006) concerning manage- ment measures for the sustainable exploitation of fishery resour- ces in the Mediterranean Sea, amending regulation (EEC) no.

2847/93 and repealing regulation (EC) No. 1626/94. 409, 75 pp.

Erguden, D., Turan, C., Gurlek, M., Yaglioglu, D., and Gungor, M.

2010. Age and growth of the Randall’s threadfin bream Nemipterus randalli (Russell, 1986), a recent Lessepsian migrant in Iskenderun Bay, northeastern Mediterranean. Journal of Applied Ichthyology, 26: 441–444.

Eryas¸ar, A. R., O¨ zbilgin, H., Go¨kc¸e, G., O¨zbilgin, Y. D., Saygu, I., Bozaoglu, A. S., and Kalecik, E. 2014. The effect of codend cir- cumference on selectivity of hand-woven slack knotted codend in the North Eastern Mediterranean demersal trawl fishery. Turkish Journal of Fisheries and Aquatic Science, 14: 463–470.

Eryas¸ar, A. R. 2014. Selectivity of commercial and alternative codends and fish behaviour in the selectivity grids in Mersin Bay demersal trawls. PhD thesis, Mersin University. 143 pp.

Eryas¸ar, A. R., and O¨ zbilgin, H. 2015. Implications for catch compo- sition and revenue in changing from diamond to square mesh codends in the northeastern Mediterranean. Journal of Applied Ichthyology, 31: 282–289.

EU. 2013. European Regulation (EU) No 1380/2013, 2013 of the European Parliament and of the Council of 11 December 2013 on the Common Fisheries Policy, amending Council Regulations (EC) No 1954/2003 and (EC) No 1224/2009 and repealing Council Regulations (EC) No 237.

FAO. 2011. Review of the State of World Marine Fishery Resources.

Rome. 235 pp.

GFCM. 2007. Report of the Tenth Session of the Scientific Advisory Committee. Nicosia, Cyprus. 144 pp.

Go¨kc¸e, G., Saygu,_I., and Eryas¸ar, A. R. 2016. Catch composition of trawl fisheries in Mersin Bay with emphasis on catch biodiversity.

Turkish Journal of Zoology, 40: 522–533.

Gu¨cu¨, A. C., and Bingel, F. 1995. State of the fisheries along the Turkish Mediterranean coast. Oceanographic Literature Review, 42: 679

Guijarro, B., and Massutı´, E. 2006. Selectivity of diamond- and square-mesh codends in the deepwater crustacean trawl fishery off the Balearic Islands (western Mediterranean). ICES Journal of Marine Science, 63: 52–67.

Hall, M. A., Alverson, D. L., and Metuzals, K. I. 2000. By-catch: prob- lems and solutions. Marine Pollution Bulletin, 41: 204–219.

Heath, M. R., Cook, R. M., Cameron, A. I., Morris, D. J., and Speirs, D. C. 2014. Cascading ecological effects of eliminating fishery dis- cards. Nature Communications, 5: 3893.

Heymans, J. J., Coll, M., Link, J. S., Mackinson, S., Steenbeek, J., Walters, C., and Christensen, V. 2016. Best practice in Ecopath with Ecosim food-web models for ecosystem-based management.

Ecological Modelling, 331: 173–184.

Hines, D. E., Ray, S., and Borrett, S. R. 2018. Uncertainty analyses for Ecological Network Analysis enable stronger inferences.

Environmental Modelling and Software, 101: 117–127.

Horbowy, J., and Luzenczyk, A. 2017. Effects of multispecies and density-dependent factors on MSY reference points: example of the Baltic Sea sprat. Canadian Journal of Fisheries and Aquatic Sciences, 74: 864–870.

Hoshino, E., Pascoe, S., Hutton, T., Kompas, T., and Yamazaki, S.

2018. Estimating maximum economic yield in multispecies fisher- ies: a review. Reviews in Fish Biology and Fisheries, 28: 261–276.

ICES. 2007. Report of the ICES-FAO Working Group on Fish Technology and Fish Behaviour (WGFTFB), ICES Document CM 2007/FTC. Dublin, Ireland. 197 pp.

Kempf, A., Mumford, J., Levontin, P., Leach, A., Hoff, A., Hamon, K.

G., Bartelings, H. et al. 2016. The MSY concept in a multi-objective fisheries environment—lessons from the North Sea. Marine Policy, 69: 146–158.

Kennedy, M. C., and O’Hagan, A. 2001. Bayesian calibration of com- puter models. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 63: 425–464.

Libralato, S., Christensen, V., and Pauly, D. 2006. A method for iden- tifying keystone species in food web models. Ecological Modelling, 195: 153–171.

Libralato, S., Coll, M., Tudela, S., Palomera, I., and Pranovi, F. 2008.

Novel index for quantification of ecosystem effects of fishing as removal of secondary production. Marine Ecology Progress Series, 355: 107–129.

Link, J. S. 2010. Adding rigor to ecological network models by evalu- ating a set of pre-balance diagnostics: a plea for PREBAL.

Ecological Modelling, 221: 1580–1591.

Lucchetti, A. 2008. Comparison of diamond- and square-mesh codends in the hake (Merluccius merlucciusL. 1758) trawl fishery of the Adriatic Sea (central Mediterranean). Scientia Marina, 72:

451–460.

Official Gazette. 2016. 4/1 Commercial Fishery Regulation Rescripts of Republic of Turkey (No: 2016/35).

Ok, M. 2012. Evaluation of the demersal fish assemblages of the northeastern Levant sea. PhD thesis, Middle East Technical University. 227 pp.

Ordines, F., Massutı´, E., Guijarro, B., and Mas, R. 2006. Diamond square mesh codend in a multi-species trawl fishery of the western Mediterranean: effects on catch composition, yield, size selectivity and discards. Aquatic Living Resources, 19: 329–338.

O¨ zbilgin, H., Eryas¸ar, A. R., Go¨kc¸e, G., O¨zbilgin, Y. D., Bozaoglu, A.

S., Kalecik, E., and Herrmann, B. 2015. Size selectivity of hand and machine woven codends and short term commercial loss in the Northeastern Mediterranean. Fisheries Research, 164: 73–85.

O¨ zbilgin, H., Tokac¸, A., and Kaykac¸, H. 2012. Selectivity of commer- cial compared to larger mesh and square mesh trawl codends for four fish species in the Aegean Sea. Journal of Applied Ichthyology, 28: 51–59.

O¨ zyurt, C. E., Yes¸ilc¸imen, H. O¨., Mavruk, S., Kiyaga, V. B., and Perker, M. 2017. Assessment of some of the feeding aspects and reproduction of S. undosquamis distributed in the Iskenderun_ Bay. Turkish Journal of Fisheries and Aquatic Science, 17: 51–60.

Papaconstantinou, C., and Farrugio, H. 2000. Fisheries in the Mediterranean. Mediterranean Marine Science, 1: 5–18.

Pauly, D., and Christensen, V. 1995. Primary production required to sustain global fisheries. Nature, 374: 255–257.

Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., and Torres, J. F.

1998. Fishing down marine food webs. Science, 279: 860–863.

Pauly, D., and Palomares, M. 2005. Fishing down marine food web: it is far more persuasive than we thought. Bulletin of Marine Science, 76: 197–211.

Piroddi, C., Bearzi, G., Gonzalvo, J., and Christensen, V. 2011. From common to rare: the case of the Mediterranean common dolphin.

Biological Conservation, 144: 2490–2498.

Piroddi, C., Giovanni, B., and Villy, C. 2010. Effects of local fisheries and ocean productivity on the northeastern Ionian Sea ecosystem.

Ecological Modelling, 221: 1526–1544.

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

(15)

Power, M. E., Tilman, D., Estes, J. A., Menge, B. A., Bond, W. J., Mills, L. S., Daily, G. et al. 1996. Challenges in the Quest for Keystones: identifying keystone species is difficult—but essential to understanding how loss of species will affect ecosystems.

Bioscience, 46: 609–620.

Rindorf, A., Dichmont, C. M., Levin, P. S., Mace, P., Pascoe, S., Prellezo, R., Punt, A. E.et al.2017. Food for thought: pretty good multispecies yield. ICES Journal of Marine Science, 74: 475–486.

Sala, A., Lucchetti, A., Piccinetti, C., and Ferretti, M. 2008. Size selec- tion by diamond- and square-mesh codends in multi-species Mediterranean demersal trawl fisheries. Fisheries Research, 93:

8–21.

Sala, A., and Lucchetti, A. 2010. The effect of mesh configuration and codend circumference on selectivity in the Mediterranean trawl Nephrops fishery. Fisheries Research, 103: 63–72.

Sala, A., Lucchetti, A., Perdichizzi, A., Herrmann, B., and Rinelli, P.

2015. Is square-mesh better selective than larger mesh? A perspec- tive on the management for Mediterranean trawl fisheries.

Fisheries Research, 161: 182–190.

Sa´nchez, P., Demestre, M., and Martin, P. 2004. Characterisation of the discards generated by bottom trawling in the northwestern Mediterranean. Fisheries Research, 67: 71–80.

Sarda`, F., Coll, M., Heymans, J. J., and Stergiou, K. I. 2015.

Overlooked impacts and challenges of the new European discard ban. Fish and Fisheries, 16: 175–180.

Saygu,_I. 2018. Ecosystem based assessment of the effects of shifting trawl codends in Gulf of Mersin, Eastern Mediterranean. PhD the- sis, Cukurova University. 186 pp.

Saygu,_I., Heymans, J. J., Fox, C. J., O¨ zbilgin, H., Eryas¸ar, A. R., and Go¨kc¸e, G. 2020. The importance of alien species to the food web

and bottom trawl fisheries of the Northeastern Mediterranean, a modelling approach. Journal of Marine Systems, 202: 103253.

Soykan, O., Tu¨rker, A. T., and Tuncay, K. H. 2015. Growth and re- production ofBoops boops,Dentex macrophthalmus, Diplodus vul- garis, andPagellus acarne(Actinopterygii: Perciformes: Sparidae) from east-central Aegean Sea, Turkey. Acta Ichthyologica et Piscatoria, 45: 39–55.

Steenbeek, J., Corrales, X., Platts, M., and Coll, M. 2018. Ecosampler:

a new approach to assessing parameter uncertainty in Ecopath with Ecosim. SoftwareX, 7: 198–204.

Stewart, P. A. M. 2002. A Review of Studies of Fishing Gear Selectivity in the Mediterranean. COPEMED Report, Aberdeen. 57 pp.

Thorpe, R. B. 2019. What is multispecies MSY? A worked example from the North Sea. Journal of Fish Biology, 94: 1011–1018.

Tokac¸, A., Herrmann, B., Go¨kc¸e, G., Krag, L. a., Nezhad, D. S., Lo¨k, A., Kaykac¸, H.et al.2016. Understanding the size selectivity of red mullet (Mullus barbatus) in Mediterranean trawl codends: a study based on fish morphology. Fisheries Research, 174: 81–93.

Ulanowicz, R. E., and Puccia, C. J. 1990. Mixed trophic impacts in ecosystems. Coenoses, 5: 7–16.

Ulrich, C., Vermard, Y., Dolder, P. J., Brunel, T., Jardim, E., Holmes, S. J., Kempf, A.et al.2017. Achieving maximum sus- tainable yield in mixed fisheries: a management approach for the North Sea demersal fisheries. ICES Journal of Marine Science, 74: 566–575.

Worm, B., and Branch, T. A. 2012. The future of fish. Trends in Ecology and Evolution, 27: 594–599.

Worm, B., Hilborn, R., Baum, J. K., Branch, T. A., Collie, J. S., Costello, C., Fogarty, M. J.et al.2009. Rebuilding global fisheries.

Science, 325: 578–585.

Handling editor: Coll Marta

Downloaded from https://academic.oup.com/icesjms/article/77/7-8/2918/5983850 by guest on 05 April 2024

Referensi

Dokumen terkait