Trammel Net Selectivity of Common Carp ( Cyprinus carpio L . , 1758 ) in Manyas Lake , Turkey

The aim of this study was determine the selectivity properties of trammel net which have 100, 110, 120, 130 and 140 mm mesh size for Cyprinus carpio in Manyas Lake. Fieldwork was carried out in two different stations of Manyas Lake, in a monthly period from April 2007 to March 2008 and with a total of 24 catching trials. The SELECT method was used to determine the selectivity parameters. In the experiments, 208 C. carpio with length ranges between 28.4-65.4 cm were caught. According to the Lognormal model, the optimum total lengths were determined as 39.05, 42.95, 46.85 and 50.76 cm for 100, 110, 120, 130 and 140 mm mesh sizes respectively. As a result, in terms of the length at first maturity and legal status, the mesh size of nets used for C. carpio catching in Manyas Lake should not less than 120 mm.


Introduction
Cyprinus carpio is very important species for Turkey's inland fisheries.It is located many inland water in Turkey both natural and depend on introduce operation.C. carpio is main target species for inland fishermen due to the commercial value.Therefore catch pressure on that species has increased and this has caused decreasing the production depend on fisheries year to year.The annual production was 15631 tons in 1996, but that was 8276 tons in 2013 with the decrease of total production to 47% from 1996 to 2013 (TUIK 2014).The overfishing, invasive species (especially Carassius gibelio), pollution and habitat destruction have negative effect on production.The contribute to Turkey economy of C. carpio was ₺36996402.00($19461547.00) in 2013 (TUIK 2014).
In the catching of C. carpio which is bulkier than the other species of inland water, trammel net is used by the fishermen.According to fishermen the reason of this is that C. carpio often break away from gilnets by plucking the net rope.In support of this idea Çubuk (2000) reported that the trammel nets were more productive than gillnets for C. carpio cathing in Beyşehir Lake.
The trammel net design enables catching fish by two different processes; (a) gilling and entangling, like conventional gill nets, and (b) trapping large fish in the bags of the inner netting (FAO 2000;Karakulak and Erk 2008).It is very easy to put into practice the results of selectivity studies that on trammel nets.From this perspective the results of selectivity studies are very valuable that will use in legal regulations for fish stock sustainability.
The first reproduction length of C. carpio reported as (with a male and female ranking as a fork length) 26.2 cm, 31.4 cm in Çıldır Lake (Yerli and Zengin 1998); 29 cm, 39 cm in Nazik Lake (Şen 2001) and 18 cm and 31 cm in Karasu Stream (Şen and Elp 2009).
In this study, the selectivity property of trammel nets that have 100, 110, 120, 130 and 140 mm mesh sizes were determined for fishing of C. carpio.

Materials and Methods
The study was conducted at two different stations in Manyas Lake (Figure 1).Fish samples were caught monthly, between April 2007 and March 2008.In catching trials, inner panel of the trammel nets that are made of monofilament material with 100, 110, 120, 130 and 140 mm stretched mesh sizes which have 0.20 mm rope thickness with E=0.50 hanging ratio were used.Outer panel mesh size of the trammel nets with 500 mm (for 100, 110, 120, 130 mm inner panel) and 600 mm (for 140 mm inner panel).The depths of all nets were 50 meshes and which have 100 m long inner panel.All nets were set in the afternoon and lifted the following morning.Caught fish were classified according to the nets and total lengths determined by 1 mm precision measurement board and weights with 1g precision digital scale.
The SELECT (Share Each Length-class's Catch Total) method was used to determine the selectivity ( (Millar 1992;Millar and Holst 1997;Millar and Fryer 1999).This method assumes that the number of fish of length l caught with a mesh size with j size has a nlj Poisson distribution, and is defined by the following equation (Acarlı et al. 2013): where λl is the abundance of fish of size l caught in net; pj (l) is relative fishing intensity (the relative abundance of fish of size l that j mesh size can catch).The Poisson distribution of the number of fish of size l caught by fishing gear with J mesh size is defined as pj ( l )λ l .rj(l ) is the selectivity curve for j mesh size (Acarlı et al. 2013).


The data obtained from field studies were analyzed by R code developed by Millar (2011).The length selectivity of each mesh size was described by four different models (normal location, normal scale, gamma and lognormal) of the 'Share Each Length's Catch Total' (SELECT) method (Millar and Fryer 1999).
Model deviance of all models evaluated when selecting the most suitable model from calculations.The model has greater standard deviation shows that the model in question is not appropriate to the obtained data (Park et al. 2004;Akamca et al. 2010).The most suitable model was chosen taking into account the lowest deviance value.The SELECT method's equations as follows: Normal Location: Kolmogorov-Smirnov (K-S) test was used to determine differences between size frequency distributions of fish caught by nets that have varying mesh size (Siegel and Castellan 1989;Karakulak and Erk 2008;Acarlı et al. 2013).

Results
After the 24 catching operations, a total of 208 common carp that have lengths range of 28.4-65.4cm were caught.While the most productive net was 120 mm mesh size with 65.4% catching capacity of total catch, 110 mm mesh size net was unproductive with 2.9% catching of total catch.The distribution of caught fish according to the nets was showed in Table 1.Average length (±SD) for 100, 110, 120, 130 and 140 mm mesh size gill nets were determined as 37.2±5.0,40.1±4.3,50.3±7.2,47.9±2.2,44.6±6.0 respectively (Table 1).The total length-frequency distribution for fish caught using different mesh size are shown in Figure 2.
By comparing the deviances of four model of SELECT method, lognormal model with the lowest deviance 453.26 gave the best fit (Table 2).The selectivity curves that were drafted by lognormal parameters via R Code software that have been showed in Figure 3.The optimum length and spread values calculated in regard to the lognormal for each net panels that have different mesh size are given in Table 3.The modal lengths and spread value estimated for this model are presented in Table 3.According to the results of K-S test that applied for query difference of length frequency distributions of fish caught by nets, differences were founded in all nets (Table 4).Ho: There are no significant differences in the length frequency distributions (α=0.05;k=1.36).

Discussion
The determination of selectivity is needed to ensure the management of commercial gill net fishing, protection of small fish and prevention death and escape of them.Therefore, it is necessary to determine the optimum mesh size for obtaining maximum continuous product (Hamley 1975;Sümer et al. 2010).
The one of main principle for sustainable of fish population is that at least one maturating in their natural location.Hence minimum landing size of fish must be longer than first maturity length.The main purpose of this study is determination the trammel nets that catch the C. carpio that have above first maturity length.For this purpose many selectivity studies have been carried out by different researcher used different method on different localities.The comparisons of the selectivity studies of C. carpio between present and previous research results are showed in Table 5.When analyzing the Table 5, there is significant differences between optimum catch length of similar mesh size.For instance, Balık (1999) reported that optimum length as 39.33 cm for 130 mm mesh size and that result is very smaller than 50.76 cm for same mesh size that reported by us.The reason of this is differences originated from selectivity method, locality and sampling period and technical differences of gillnet-trammel net.Despite that, the optimum length that reported as 38.12 cm for 101.5 mm net by Carol and García-Berthou (2007) with SELECT method is quite close to 39.05 cm that reported by us for similar net (100 mm) and method.Balık (2008) reported that Holt (1963) method is one of the most commonly used methods for estimating gillnets selectivity, however this method is restrictive.Recently, the SELECT method has been used commonly this is a statistical "model that estimates gillnet selection curves from comparative gillnet catch and provides a cohesive approach to selectivity analyses.Balık and Çubuk (2001) reported that selectivity of gillnets can vary for each fish species and even for population of same species in different habitat therefore selectivity should be determine separately for each species which catching by this nets (Sümer et al. 2010).
Both consideration (Table 6) the first maturity that reported by Yerli and Zengin (1998), Şen (2001), Şen and Elp (2009) and legal status, the mesh sizes of trammel nets that used in catching of C. carpio in Manyas Lake shouldn't less than 120 mm.

Figure 1 .
Figure 1.Study area and sample station.

Figure 3 .
Figure 3. Selectivity curves (A) and deviance residual plots (B) of trammel net for the C. carpio.

Table 1 .
Catch composition data of trial nets.

Table 2 .
Selectivity parameter values of C. carpio.
Figure 2. Catch size frequency distributions for the C. carpio.

Table 3 .
Optimum length and spread values of C. carpio according to the Lognormal.

Table 4 .
Result of the K-S test used to compare length frequency distributions of catch.

Table 5 .
The comparisons of the selectivity studies of C. carpio between present and previous research results.
a mesh size (stretched); b mesh size (bar length).

Table 6 .
Determination criteria of fit mesh size for C. carpio by first maturity and legal status.