### Study area and sampling

This study was conducted on the Bandırma Bay coast, south of the Marmara Sea (40°24′25″N–27°55′33″E; Fig. 1) in intertidal and shallow sub-tidal areas with sandy bottoms. *R. philippinarum* samples were collected on a monthly basis between September 2012 and August 2013. Samples were collected by towing parallel to the shoreline during low tide for 10 min (length of dredge mouth and height: 55 and 30 cm, respectively; number of teeth and length: 25 and 16 cm, respectively; mesh size: 5 mm) at a depth of 3–8 m using a mechanical dredge. Shell length (SL) and total weight (TW) of individual bivalves were measured for a period of 1 year. Size measurements were used to estimate growth parameters. The sea surface temperature varied between 8.70 °C in winter (February) and 24.10 °C in summer (July), with a mean of 16.20 ± 1.55 °C (Fig. 2). Seawater temperature in the sampling area was measured using a mercury bulb thermometer.

### Growth

In total, 10.626 *R. philippinarum* were sampled. Anterior–posterior length (SL) of individual specimens was measured using digital callipers (0.01-mm accuracy). Length–frequency distributions were constructed with 1-mm intervals for each month. Total, shell and wet meat weight of each bivalve were measured using an electronic balance (0.01-mg accuracy).

The length–weight relationship was determined according to the allometric equation defined by Ricker (1973): *Y* = *aX*
^{b}, where *Y* is TW, *X* is SL, *a* is the intercept and *b* is the slope. Parameters *a* and *b* were estimated by least squares linear regression using log–log transformed data:

$$\log \text{TW} = \log a + b\log \text{SL}$$

(1)

The coefficient of determination (*r*
^{2}) was used as an indicator of the linear regression quality. In addition, the 95 % confidence limit of *b* and the significance level of *r*
^{2} were also estimated. To confirm whether the value of *b* obtained by linear regression was significantly different from the isometric value (*b* = 3) and if they had negative (*b* < 3) or positive (*b* > 3) allometric relationships, a *t* test was applied with a confidence level of ±95 % (*α* = 0.05; Sokal and Rohlf 1987).

On the basis of monthly sampling frequency in the study area, 12 time-series datasets (1-mm SL size classes) were estimated using the electronic length–frequency analysis (ELEFAN) procedure in the length–frequency distribution analysis (LFDA) program (Kirkwood et al. 2001). Length was predicted as a function of age according to the von Bertalanffy growth equation (VBG, Eq. 2). This equation is used when a non-seasonal growth pattern is observed

$$L_{t} = L_{\infty } \left( {1 - e^{{ - K(t - t_{0)} }} } \right)$$

(2)

A study conducted by Hoenig and Hanumara (1990) found the Hoenig and Hanumara (1982) model used in fisheries better fit seasonal growth data; this model represents a combination of features from other models. Therefore, seasonal growth was described using the Hoenig and Hanumara (1982) version of the VBG equation:

$$L_{\text{t}} = L_{{\infty }} \left[ {1 - e^{{\left[ { - K\left( {t - t_{0} } \right) + \left( {C\frac{K}{2 \ne }} \right)\sin 2 \ne \left( {t - t_{\text{s}} } \right) - \left( {C\frac{K}{2 \ne }} \right)\sin 2 \ne \left( {t_{0} - t_{\text{s}} } \right)} \right]}} } \right]$$

(3)

where *L*
_{t} is the maximum anterior–posterior shell length (apSL; mm) at time *t*, *L*
_{∞} is the asymptotic apSL (mm), *K* (year^{−1}) is the growth curvature parameter, *C* is the relative amplitude (0 ≤ *C* ≤ 1) of seasonal oscillations, *t*
_{0} is the theoretical age when the SL is zero (years) and *t*
_{s} is the phase of the seasonal oscillations (−0.5 ≤ *t*
_{s} ≤ 0.5), which denotes the time of year that corresponds to the start of the convex segment of sinusoidal oscillation.

The time of the year when growth is slowest, known as the winter point (WP), was calculated as:

$$\text{WP} = t_{\text{s}} + 0.5$$

(4)

Seasonal and non-seasonal VBG curves were fitted to length–frequency distributions after first specifying a range of values for *L*
_{∞} and *K* to maximize the goodness of fit (Rn) for each curve, thereby optimizing data. Rn was calculated as:

$${\text{Rn}} = \frac{{10^{{\text{ESP/ASP}}} }}{10}$$

(5)

where ASP is the available sum of peaks, computed by adding the best values of the available peaks, and ESP is the explained sum of peaks, computed by summing all the peaks and troughs hit by the VBG curve. In the area on the score grid that the best maximum is found, maximization has been done on the small area (0.1 < *K*<0.5 year^{−1} and 60 < *L*
_{∞} < 70 mm), in order to obtain the highest score function possible. Through the value of this score function, growth parameters were determined to be stable.

The growth performance index (*Ø*′, Eq. 6) was compared using different growth values reported in the literature, according to the following formula (Eq. 6; Pauly and Munro 1984). In addition, we constructed a 95 % confidence interval for *Ø*′ from the different combination estimates and from those in this study (*α* = 0.05)

$$\emptyset^{\prime } = 2\log_{10} \left( {L_{{\infty }} } \right) - \log_{10} K$$

(6)

The maximum lifespan (*A*
_{95}, Eq. 7) was calculated using the inverse of the VBG equation, where we considered the maximum SL as 95 % of the *L*
_{∞} (Taylor 1958):

$$A_{95} = t_{0} + \frac{2.996}{K}$$

(7)

### Mortality

The instantaneous total mortality rate (*Z*, Eq. 8) was estimated using different methods. The Beverton and Holt (1956) equation for estimating *Z* was calculated as:

$$Z = K\left[ {\frac{{L_{{\infty }} - \overline{L} }}{{\overline{L} - L^{\prime } }}} \right]$$

(8)

where *L’* is the length when *R. philippinarum* were first fully recruited and \(\overline{L}\) is the mean length of all clams longer than *L’*.

The length-converted catch curve (LCCC; Pauly 1983, 1984a, b) was also used to estimate *Z* as follows:

$$\ln \left( {\frac{{N_{\text{i}} }}{{\Delta t_{\text{i}} }}} \right) = a + b t_{\text{i}}^{\prime }$$

(9)

where *N*
_{i} is the frequency in length class i, Δ*t*
_{i} is the time required for a clam to grow and reach length class i, a is the intercept, \(t_{\text{i}}^{\prime }\) is the relative age of individual clams that correspond to length class i and *b* is the slope that corresponds to *Z* with a sign change.

The natural instantaneous mortality rate (*M*, Eq. 10) was estimated using the empirical relationship defined by Pauly (1980):

$$\log M = - 0.0066 + 0.279\log \text{TL}_{\infty } + 0.6543\log K + 0.4634\log T$$

(10)

where *T* is the mean annual seawater temperature and *TL*
_{∞} is the asymptotic total length (cm) that *R. philippinarum* can reach. This empirical equation assumes that the length is measured as TL in cm (Gayanilo et al. 2005). Therefore, length–frequency analyses were reapplied to length composition data to obtain *TL*
_{∞} (cm), TL and *K* for use in Pauly’s empirical equation.

The fishing mortality rate (*F*) was calculated as:

The exploitation rate (*E*; Sparre and Venema 1992) was calculated as:

$$E = \frac{F}{F + M}$$

(12)

Moreover, instantaneous mortality rates were then converted to annual mortality rates (*A*) as:

The Beverton–Holt and LCCC *Z* were calculated using length–frequency distribution analysis version 5.0 (Kirkwood et al. 2001). *M* was estimated using the FISAT II program (Gayanilo et al. 2005). Significant differences between the Beverton–Holt and LCCC mortality rates were analysed by one-way analysis of variance (ANOVA; *F* test), using Microsoft Excel 2010 (Zar 1984).

### Reproduction

The reproductive activity of *R. philippinarum* was determined on the basis of the ash-free dry weight (AFDW)/dry shell weight (DSW) ratio. Each month, sub-samples of 35 clams were used to extract all their soft parts. The sub-sample used for condition index (CI, Eq. 14) analysis had an SL ranging from 20 to 50 mm. To determine the body mass cycle, all soft parts were removed and dried to a constant mass at 100 °C for 24 h to obtain DSW (g). AFDW (mg) was obtained by drying soft tissues in an oven at 550 °C for 7 h (Laudien et al. 2003). CI was calculated according to the following formula (Walne and Mann 1975):

$${\text{CI}} = \left( {\text{AFDW/DSW}} \right) \times 100$$

(14)

The monthly gonado-somatic index (GSI, Eq. 15), which is defined as the volume of gonadal tissue (*V*
_{gon}) relative to the total body volume (*V*
_{body}), was estimated using a method based on linear measurements of the gonad region, which forms a sheath around the digestive gland (Urban and Riascos 2002; Riascos et al. 2007).

$${\text{GSI}} = V_{\text{gon}} \text{/}V_{\text{body}} \times 100$$

(15)

A sub-sample of 35 specimens (SL = 40–50 mm) was used to study the reproductive cycle. The body mass cycle of individual bivalves was determined in the gonad stage on the basis of microscopic observations of fresh gonadal material. We used a semi-quantitative scale proposed by Guillou et al. (1990), which allowed us to classify males and females into four gonad stages: indifferent, ripe I, ripe II and spent.