Age, growth and natural mortality of coney (Cephalopholis fulva) from the southeastern United States

Age determination

Coney were sampled from fisheries landings along the SEUS coast (North Carolina through the Florida Keys) from 1998 to 2013 by port agents employed by the SRHS, which samples headboats, and the SEFSC Trip Interview Program (TIP), which samples commercial fisheries landings. Both surveys operate using a random sample design methodology. All specimens used in this study were killed as part of legal fishing operations and were already dead when sampled by the port agents, thus all research was conducted in accordance with the Animal Welfare Act (AWA) and with the U.S. Government Principles for the Utilization and Care of Vertebrate Animals Used in Testing, Research, and Training (USGP) OSTP CFR May 20, 1985, Vol. 50, No. 97. The majority of specimens from both fisheries were captured by conventional, vertical hook-and-line gear. Total length (TL) of specimens was recorded in millimeters. Whole weight (W, kg) was recorded for fish landed in the headboat fishery. Whole weights were unavailable for fish landed commercially because they were eviscerated at sea. Because of predetermined sampling protocols and time constraints imposed by their workload, as well as the protogynous nature of coney and their small size, samplers were not required to determine the sex of sampled coney due to uncertainty in macroscopic staging.

Sagittal otoliths were removed from 367 coney and stored dry in coin envelopes. Otoliths were sectioned in the transverse dorso-ventral plane on a low-speed saw, following the methods of Potts & Manooch (1995). The sections were mounted on microscope slides with thermal cement and covered with mounting medium before analysis. The sections were viewed under a dissecting microscope at 12.5 × with transmitted light. Readings were taken from the dorsal lobe of the otolith. Initial ring counts were made along the sulcal groove (vertically from the core) and then confirmed by following the rings out to the lateral plane (horizontal from the core). Each sample was assigned a ring count equal to the number of opaque zones. Two readers interpreted otolith sections. To ensure consistency between readers in the interpretation of growth structures, each individual read all 367 slides, then we calculated between-reader indices of average percent error (APE) following the methodology of Campana (2001).

Edge analysis was used to validate the annual deposition of the opaque zone in coney otoliths. The edge type of the otolith section was noted: 1, opaque zone forming on the edge of the otolith section; 2, narrow translucent zone on the edge, generally <30% of the width of the previous translucent zone; 3, moderate translucent zone on the edge, generally 30–60% of the width of the previous translucent zone; 4, wide translucent zone on the edge, generally >60% of the width of the previous translucent zone (Harris et al., 2007). Frequency of all edge types by month was then plotted to determine the period of opaque zone deposition. Edge analysis is based on the assumption that there is a yearly sinusoidal cycle in the plot of relative frequency of edge types over time (Campana, 2001). On the basis of this frequency analysis of edge type, all samples were assigned a chronological, or calendar, age obtained by increasing the opaque zone count by one if the fish was caught before that increment was formed and had an edge with a translucent zone that was moderate to wide (type 3 or 4). All fish caught after opaque zone formation would have had a chronological age equivalent to the opaque zone count.

Growth

Von Bertalanffy (1938) growth parameters were estimated from the observed length-at-age data, using the chronological age. Parameters were derived using PROC NLIN, a non-linear regression procedure using least squares estimation and the Marquardt iterative algorithm option, in SAS statistical software (vers. 9.3; SAS Institute , 1987). Appropriate statistical tests were employed to examine differences in mean size and age by fishery sector (Student’s t-test [P < 0.05] for normally distributed data; non-parametric Kolmogorov–Smirnov test [P < 0.05] for skewed distributions). We also examined mean length-at-age data by sector (recreational versus commercial) to determine if pooling of data was appropriate (i.e., there were no significant differences in length- at-age by sector).

Body–size relationships

We regressed fish whole weight (kg) on fish TL (mm) using data for all coney measured by the SRHS from 1976–2014 (n = 487). We evaluated a nonlinear fit, using PROC NLIN in SAS (SAS Institute , 1987), and a linearized fit of the log-transformed data, examining the residuals to determine which regression was appropriate.

Natural mortality

We estimated the instantaneous rate of natural mortality (M) using 2 methods.

1. Hewitt & Hoenig’s (2005) longevity mortality relationship: ${\displaystyle M\approx 4.22/t_{max},}$ where t_max is the maximum age of the fish in the sample.

2. Charnov, Gislason & Pope (2013) method, which uses the von Bertalanffy growth parameters: ${\displaystyle M={\left(L/L_{\infty }\right)}^{-1.5}\times K,}$ where L_∞ and K are the von Bertalanffy growth equation parameters (asymptotic length and growth coefficient) and L is fish length at age. The Hewitt & Hoenig (2005) method uses life span or longevity to generate a single point estimate, and it is an improvement to the original equation of Hoenig (1983). The newer Charnov method, which incorporates growth parameters, is an improvement to the empirical equation of Gislason et al. (2010) and is based on evidence that M decreases as a power function of body size. The Charnov method generates age-specific rates of M and is currently in use in Southeast Data Assessment and Review (SEDAR) stock assessments (E Williams, pers. comm., 2013).

Age determination

A total of 367 sagittal otoliths of coney were sectioned. The distribution of samples by state and fishery sector is shown in Table 1. The majority of samples (75%) came from the commercial sector in the Carolinas. Only 9% of aging samples were from Florida, with the majority of those samples from headboats. Opaque zones were counted on 353 (96.2%) of the 367 sectioned otoliths. Sections from the other 14 otoliths (3.8%) were judged illegible and were excluded from this study.

For our analysis of increment periodicity, we assigned an edge type to all 353 samples. Opaque zones on the otolith marginal edge occurred in samples collected from January to June (Fig. 1); occurrence of opaque zones was highest from March to June with peak formation in April (Fig. 1). We concluded that opaque zones on coney otoliths were annuli. Chronological ages resulting from edge analysis were assigned as follows: for fish that were caught from January to June and had an edge type of 3 or 4, the chronological age was the annuli count plus one; for fish that were caught during the same period and had an edge type of 1 or 2 and for fish that were caught from July to December, the chronological age was equivalent to the annuli count.

Growth increments of coney were moderately easy to interpret. Based on Campana’s (2001) acceptable value of APE (5%), agreement was good between the two readers (MLB–JCP APE = 6.4%, n = 353). Percent agreement values between the two readers were moderate (49%) but increased for estimates within (±) 1 year (87%) and (±) 2 years (92%). These results indicate acceptable between-reader agreement.

Growth

Coney in this study (n = 353) ranged from 217 to 430 mm TL and from age 1 to 19. There were only 11 (3%) fish older than age 12 (Table 2). Length and age distributions by sector are shown in Fig. 2. Visual examination of these size and age frequency distributions identified apparent differences by sector. Modal lengths were 325 mm TL for the commercial sector and 275 mm TL for the recreational sector (Fig. 2A). Mean lengths by sector were significantly different: 325 mm TL (±standard error [SE] 2.1) for the commercial sector versus 280 mm TL (SE 3.6) for the recreational fishery (Student’s t test: t = 11.00; P = 0.0001) (Fig. 2A). The modal age frequencies were age-5 years and age-3 years for the commercial and recreational sectors, respectively (Fig. 2B). Due to the visually skewed appearance of age frequency distributions, we tested for normality using PROC UNIVARIATE (SAS Institute , 1987) and found age frequency data to be non-normally distributed (Wilcoxon signed rank test, P < 0.0001). Mean ages were significantly different between fisheries: 6.6 years (SE 0.17) versus 4.4 years (SE 0.32) for commercial and recreational sectors, respectively (non-parametric Kolmogorov–Smirnov test; D = 0.51, P < 0.0001). Visual examination of mean size-at-age of coney by fishery sector (Fig. 3) revealed no significant difference in growth by sector for ages 2–11, as indicated by overlapping error bars. Additionally, we performed an analysis of covariance (ANCOVA) of length-at-age by sector, using age as the covariate, and found no significant differences in size at age between sectors for six out of nine ages for which we had adequate sample size for comparison. On the basis of these results, we pooled data across sectors. The resulting von Bertalanffy growth equation was: ${\displaystyle L_t=377\left(1-e^{-0.20\left(t+3.53\right)}\right)}$ for all fishery sectors combined (n = 353) (Table 3 and Fig. 4).

Body–size relationships

Statistical analyses revealed a multiplicative error term (variance increasing with size) in the residuals of the W–TL relationship, indicating that a direct nonlinear fit was not appropriate. We used a linearized ln-transform fit of the data, resulting in the following regression to describe the relationship: ${\displaystyle \ln \left(W\right)=3.03\times \ln \left(\mathrm{TL}\right)-11.15\left(n=487,r^2=0.91\right)}$ where r² is the coefficient of determination. This equation was transformed back to the form ${\displaystyle W=a\times {\mathrm{TL}}^b}$ after adjustment of the intercept for log-transformation bias with the addition of one-half of the mean square error (MSE) (Beauchamp & Olson, 1973), resulting in this relationship:

W = 1.457 × 10⁻⁸TL^3.03(n = 487; MSE = 0.022) (Fig. 5).

Natural mortality

The method of Hewitt & Hoenig (2005), which uses maximum age or life span (age-19 years in this study), estimated that M was 0.22. The method of Charnov, Gislason & Pope (2013), which produces age-specific estimates of M using von Bertalanffy growth parameters, estimated M of 0.49 for age-1, 0.28 for age-5, 0.22 for age-10, and 0.20 for age-19 (Table 2).