Vital population statistics based on length frequency analysis of the exploited Japanese eel (Anguilla japonica) stock in the Kao-Ping River, southern Taiwan

(1)

Vital population statistics based on length frequency analysis of the exploited

Japanese eel (Anguilla japonica) stock in the Kao-Ping River, southern Taiwan

By Y.-J. Lin1and W.-N. Tzeng1,2

1_{Institute of Fishery Science, College of Life Science, National Taiwan University, Taipei, Taiwan, ROC;}2_{Department of Life}

Science, College of Life Science, National Taiwan University, Taipei, Taiwan, ROC

Summary

Vital statistics such as growth, mortality, and maturity parameters are crucial in understanding the population dynamics of a species. A total of 7 074 Japanese eels (Anguilla japonica) in the lower reach of the Kao-Ping River, southern Taiwan, were collected with eel tubes in 1998 2004 and shrimp nets in 2004 2007. Data from 2004 were excluded due to mixed gear information and escapement of cultured eels; in subsequent years escaped cultured eels were identified and excluded from analyses. The estimated asymptotic length in the von Bertalanffy growth function (84.5–110 cm) was smaller, while the Brody growth parameter (0.30–0.44 year)1) was higher using electronic length frequency analysis (ELE-FAN) than when using ShepherdÕs length composition analysis (SLCA). The total instantaneous mortality rate (Z) was around 1 for periods 1998–2003 and 2 year)1 for 2005–2007 using length-converted catch curves. The 95% confidence intervals of Z did not overlap for two of the periods, suggesting that the mortality rates were significantly higher during 2005–2007, possibly due to the introduction of shrimp nets. The maturity function differed significantly between sexes, indicating that females become silver eels at a larger size. The Japanese eels in the lower reach of the Kao-Ping River were likely heavily exploited, thus management and conserva-tion acconserva-tions are strongly recommended.

Introduction

The Japanese eel, Anguilla japonica, is a catadromous fish widely distributed in Taiwan, China, Japan, and Korea (Tesch, 2003). The eels spawn in the tropical Pacific Ocean west of the Mariana Islands (Tsukamoto, 1992, 2006). The leaf-like larva, leptocephali, are transported passively via the North Equator-ial and Kuroshio currents and metamorphose to the glass eel stage over the continental shelf and further develop into pigmented elvers in the estuary (Cheng and Tzeng, 1996). The elvers then either migrate upstream to freshwater habitats or remain in brackish estuaries as yellow eels. After 4–10 years the yellow eels begin sexual maturation and become silver eels with enlarged eyes and darkened body color (Han et al., 2001, 2003; Tzeng et al., 2002). These silver eels then migrate to the spawning grounds in the ocean to spawn and die (Tesch, 2003). Vital population statistics such as growth parameters, total, natural, and fishing mortality and length at maturity are crucial in understanding population dynamics and making fisheries assessments (Quinn and Deriso, 1999; Jennings et al., 2001). Japanese eel populations, along with those of other temperate eel species, have been declining to historically low

levels (Haro et al., 2000; Feunteun, 2002; Tatsukawa, 2003; Tseng et al., 2003; Dekker, 2004; Han and Tzeng, 2006). However, the availability of these statistics for Japanese eel stocks is limited relative to species such as the European eel A. anguilla(De Leo and Gatto, 1995; Sveda¨ng, 1999; Dekker, 2004), American eel A. rostrata (Robitaille et al., 2003; Weeder and Uphoff, 2003), and Australian short-finned A. reinhardtiiand long-finned A. dieffenbachii eels, and A. aus-tralis(Francis and Jellyman, 1999; Hoyle and Jellyman, 2002; Doole, 2005). Without appropriate vital population statistics, efficient and active management of A. japonica stocks may be difficult.

Traditional methods of studying the dynamics of exploited fish stocks were mostly based on age-structured data. However, use of computers and theoretical advantages of analytical methods based on length-frequency data led to their development during the 1980s (Jones, 1984; Gulland, 1987; Pauly, 1987; Pauly and Morgan, 1987) and widespread use in recent years (e.g. O¨zbilgin et al., 2004; García and Duarte, 2006; Velasco et al., 2007). Length-frequency data is easier to obtain than age-structured data, and correctly ageing fishes in tropical waters may be difficult due to reduced seasonal growth contrast (Campana, 2001) or problems in interpreting otolith growth increments (Morales-Nin and Panfili, 2005). Thus, length-frequency methods were used to overcome such difficulties.

The lower reach of the Kao-Ping River in Taiwan (Fig. 1) is one of the most important fishing grounds for elvers, juveniles, and adult Japanese eels. Fishermen use bamboo eel tubes to harvest the juvenile and adult eels (Chang and Tzeng, 1990; Tzeng and Chang, 2001). In Taiwan, although great numbers are reared in aquaculture ponds, eels harvested in the wild attain prices 3–4 times higher than for cultured eels (Chang and Tzeng, 1990). Considerable quantities of both juveniles and adults are also caught as by-catch in shrimp nets in the lower reach, possibly because the eels follow the shrimp into the nets. Moreover, shrimp nets are widely used by eel fishermen because they are highly efficient in catching juvenile and adult eels (Lin and Tzeng, 2008), which further increases fishing pressure on the local eel stocks. However, information on growth, mortality, and maturity is limited, especially for yellow and silver eels. Thus, the path to meaningful manage-ment and restoration of A. japonica in Taiwan remains uncertain.

The study objective was to estimate the vital population statistics and evaluate the state of exploitation of A. japonica stock in the lower reach of the Kao-Ping River based on the length frequency data from the eel catch, 1998–2007.

J. Appl. Ichthyol. 26 (2010), 424–431 2010 Blackwell Verlag, Berlin ISSN 0175–8659

Received: July 27, 2009 Accepted: November 22, 2009 doi: 10.1111/j.1439-0426.2010.01453.x

U.S. Copyright Clearance Centre Code Statement:0175–8659/2010/2603–0424$15.00/0

(2)

Materials and methods Study area

The Kao-Ping River is the second largest river in Taiwan with a length of 171 km and a drainage area of 3 256 km2. Mean annual rainfall is 3 046 mm and characterized by a dry season in winter and spring (October May) and a rainy season in summer and autumn (June Sept.). The rainfall is mainly contributed by the monsoon in early summer and by irregular typhoons during late summer and autumn. As a result, the river water level ﬂuctuates in the mid- and upper reaches but is relatively stable in the lower reach. In the region where ﬁshermen harvest the eels and sand shrimp, salinity is 10 30 ppt and mean (±SD) bottom temperature is 25.9 ± 2.7C (Chen, 2005).

Sample collection and measurement

Monthly eel catch in numbers was recorded from one cooperative fisherman using eel tubes during 1998–2004 and from another two fishermen using shrimp traps in 2004–2007 (Table 1). Gear information in 2004 was mixed and unable to be distinguished clearly. Moreover, the local eel stock was greatly influenced in 2004 by escaped cultured eels due to Typhoon Mindulle in July 2004 (Chu et al., 2006), and which might have influenced the estimation of vital parameters. This appeared to be most significant in 2004, decreasing substan-tially in 2005 and nearly disappearing in early 2006 (Lin and Tzeng, 2008). Moreover, the escaped cultured eels could be distinguished from wild eels by their different body coloration (Chu et al., 2006). Thus, data were separated into two periods: 1998–2003 and 2005–2007; data from 2004 was excluded from the length-frequency analysis. Escaped cultured eels were also excluded from analyses in 2005 and 2006.

Captured eels were anesthetized immediately with ice, measured for total length (to nearest 1 mm) and total weight (g), then transferred to the laboratory and frozen ()20C). After defrosting, sex was determined by gross inspection of the gonads, and the development stage (yellow or silver eels) was determined by the body color, enlarged eyes, and blackened pectoral ﬁns (Han et al., 2003; Tesch, 2003).

Growth

Anguilla japonicusmales are reported to have a mean growth rate in length of >90% vs females in the Kao-Ping River (Tzeng et al., 2003), and 102% in the Pearl River (Tzeng et al., 2000). Thus, the difference between sexes is not as significant as for other temperate eel species, and it was felt to be adequate to pool length data for both sexes in order to increase the sample size for analysis. Eel growth was described by the von Bertalanffy growth function (VBGF):

Lt¼ L1½1 eKðtt0Þ

whereLtis the length at time t, L¥is the asymptotic length, K is

the Brody growth coeﬃcient, and t0 is the initial condition

parameter when the hypothetical length is zero (Quinn and Deriso, 1999).

The population maximum length, estimated following Formacion et al. (1992), was set as the preliminary value of the asymptotic length. In addition, the estimated population extreme length with 95% conﬁdence intervals was also used to estimate L_¥according to Froese and Binohlan (2000):

log₁₀ðL1; cmÞ ¼ 0:044 þ 0:9841 log10ðLmax; cmÞ Given the estimated maximum length as initial estimate of L¥, the preliminary value of K was obtained using monthly

catches-at-length data with 5-cm intervals by length frequency analysis (FiSAT-ELEFAN) (Pauly, 1987) and ShepherdÕs length composition analysis (FiSAT-SLCA) (Shepherd, 1987), respectively. The response surface analyses of ELEFAN and SLCA were then used to search for the optimal combination of L¥ and K (Gulland, 1987; Shepherd, 1987),

in which the preliminary estimates of L¥and K were used as

the inference points. These estimating procedures were completed using FiSAT software (Gayanilo et al., 1996). Growth performance index (ØÕ) (Pauly and Munro, 1984) was calculated as:

[0¼ logeðKÞ þ 2logeðL1Þ

Mortality

Natural instantaneous mortality (M) was estimated using PaulyÕs (1980) empirical formula (Gayanilo et al., 1996):

log_eðMÞ ¼ 0:0066 0:279 log eðL1Þ þ 0:6543 logeðKÞ þ 0:463 log_eðT Þ; SDðlogMÞ¼ 0:245

where L¥ and K are VBGF parameters and T is the mean

environment temperature, i.e. 25.9C in this study.

Total instantaneous mortality (Z) was estimated using the length-converted catch curves developed from length fre-quency distributions (Pauly, 1990; Gayanilo et al., 1996), given the estimated value of VBGF parameters. The instanta-neous ﬁshing mortality (F): F = Z–M, the exploitation rate (E): E = F· Z)1 and the mean annual survival rate (S): S = exp(–Z).

Sexual maturity

The silver eels were sexually maturing in preparation for spawning in the ocean. The maturation process, represented by the silvering process, was assumed to depend on size rather than age (Vøllestad, 1992; Oliveira, 1999). It was further assumed that the corresponding parameters for the maturity equations were the same during the period from 1998 to 2006 Fig. 1. Japanese eel Anguilla japonica sampling site, lower reaches of

(3)

and between wild and escaped cultured eels in 2004 (Chu et al., 2006). They were also assumed to be the same between different types of gear. However, the silvering of A. japonicus was significantly different between sexes in that the silver females were generally larger and older (Han et al., 2003), thus the silvering process was fitted separately in females and males. Let c(Li)be the proportion of the silver eels in the length

class Li,then the silvering pattern was ﬁtted by two models:

(1) Logistic model: cðLiÞ ¼

expða þ bLiÞ ½1 þ expða þ bLiÞ

; where a, b are parameters

(2) De Leo and GattoÕs (1995) model: cðLiÞ ¼

c_max ½1 þ expðkLi

g Þ

where cmax is the asymptotic maturation rate, k is the

semi-saturation constant which corresponds to the length at which the maturation rate = 0.5· cmax. g is a shape parameter

which is inversely proportional to the slope of the curve at L= k (De Leo and Gatto, 1995). Note that the two models are identical when cmax= 1. The optimal values of the

parameters of the two equations were found by minimizing the sum of squared error using the Microsoft EXCEL SOLVER routine.

After the parameters of above models were estimated, the length at 50% maturity (Lm50) was calculated as:

^ Lm50¼

^a ^

b ;for logistic model ^

k ^g log_eð2^c_max 1Þ ;for De Leo & Gatto0s model (

The null hypothesis that females and males have the same silvering process was tested by the analysis of the residual sum of squares (ARSS, Chen et al., 1992; Kimura, 2000) using the Fc: Fc¼_SSRðHSSH =PðQ1Þ

aÞ=ðnQP Þwhere SSH = SSR(H0)–SSR (Ha) = the sum of squared residuals for the null hypothesis – that of the alternative hypothesis. P = the number of parameters in the model (P equals 2 for the logistic model and equals 3 for De Leo & GattoÕs model), Q = the number of groups

compared (i.e. = 2 for comparisons between sexes), and n = total sample size. Fcfor the logistic and De Leo & Gatto

models was calculated and compared with a critical value Fcrit= FP(Q-1), n-QP,1-0.5a (Kimura, 2000). Statistical

signiﬁ-cance level (a) was set at 0.05.

Results Sample collection

A total of 7 074 eels were caught from 1998 to 2007, in which 4 601 eels were sexed and their developmental stages (yellow or silver) determined (Table 1). Annual catch increased from around 130 eels in 1998–2003 and to more than 700 individuals after 2004, which coincided with introduction of the shrimp net to this area. Prior to 2007 most eels were sexed, with females dominating the catch from 1998 to 2003. However, in 2004, males increased due to cultured eel escapement (Chu et al., 2006). Also, the catch record from shrimp nets included more months than those from eel tubes (Table 1), implying that the ﬁshing season for eel tubes was shorter than for the shrimp nets.

Growth

A total of 908 eels caught in 1998–2003 and 4 894 eels in 2005– 2007 were used for length-frequency analyses using FiSAT-ELEFAN and FiSAT-SCLA. Observed and predicted max-imum length, optimal combination of VBGF parameters, L¥

and K and the growth performance index ØÕ are shown in Table 2. Observed and predicted maximum length was 78.50 and 87.95 cm from 1998 to 2003, and 87.50 and 89.94 cm from 2005 to 2007, respectively. Overlapping 95% CIs for predicted maximum lengths indicated a statistically insigniﬁcant diﬀer-ence between the two periods. The predicted L¥from Lmaxwas

89.39 cm (95% CI: 84.11–97.17 cm) and 94.43 cm (95% CI: 84.92–100.39 cm) for 1998–2003 and 2005–2007, respectively. VBGF parameters (L¥, K and Ø) were estimated by ELEFAN

as 101.00 cm, 0.33 year)1, and 8.12, and by SLCA as 84.50 cm, 0.38 year)1, and 7.91 for eels from 1998 to 2003. For eels caught from 2005 to 2007, the corresponding values were 110.00 cm, 0.30 year)1, and 8.20, and 94.00 cm, Table 1

Monthly eel catch by cooperative ﬁshermen 1998–2007 in lower reach of Kao-Ping River

1998T 1999T 2000T 2001T 2002T 2003T 2004TS 2005S* 2006S* 2007S January 18 51 205 545 February 40 55 16 155 640 March 95 15 154 84 411 April 54 61 May 16 3 63 30 66 314 June 11 6 156 76 309 July 9 40 264 134 113 148 August 38 62 152 124 235 50 September 71 16 30 47 52 99 October 22 18 34 214 76 159 November 241 269 222 33 December 16 35 310 92 Sum 38 (38) 190 (190) 112 (112) 283 (283) 168 (168) 117 (117) 1272 (1272) 1254 (1092) 1813 (588) 1827 (841) F 23 110 81 196 117 94 519 682 401 560 M 4 24 11 28 28 9 689 206 39 9 U 11 56 20 59 23 14 74 204 148 272

*Escaped cultured eels were identiﬁed and excluded from the study.

Numbers in parentheses, number of eels sexed; Superscript T, eels caught by eel tubes; S, eels caught by shrimp nets; F, number of females; M, males; U, sexually undiﬀerentiated eels. Individuals in 2004 were excluded due to mixed gear information and escaped cultured eels.

(4)

0.44 year)1, and 8.26 (Table 2). L¥ estimate from ELEFAN

was higher compared to SCLA and outside the 95% CI of L¥

from Lmax, while the L¥ estimate from SCLA fell into this

range. Thus, the K estimate from ELEFAN I was smaller than that from SCLA due to the negative correlation between L¥

and K. The estimates of Ø were similar between periods and methods, except for the 1998–2003 period when the SLCA method gave the lowest value.

Mortality

Instantaneous total mortality rates Z, estimated from the length-converted catch curves based on growth parameters from ELEFAN and SLCA for eels caught from 1998 to 2003 were 1.13 year)1(95% CI: 0.99–1.36) and 0.90 (95% CI: 0.78– 1.03), respectively. The conﬁdence intervals include uncertain-ties in the catch curves only, assuming correct growth parameter values (Fig. 2a,b). For 2005–2007, Z values were estimated as 1.99 year)1(95% CI: 1.76–2.28) and 2.08 (95% CI: 1.92–2.26), respectively (Fig. 2c,d).

The annual natural mortality rates (M) from PaulyÕs empirical formula based on ELEFAN and SLCA growth parameters were 0.59 and 0.68 year)1, respectively, in 1998– 2003, and amounted to 0.54 and 0.74 year)1 in 2005–2007. According to the 95% conﬁdence interval of individual data in PaulyÕs regression these estimates of M are statistically uncertain by a factor of 3, when assuming correct growth parameters.

Instantaneous ﬁshing mortality rates (F) were derived as 0.54 and 0.22 year)1in 1998–2003, and 1.45 and 1.34 year)1in 2005–2007. The exploitation rates (E) during 1998 2003 for Japanese eels were estimated as 47.8 and 24.4%, with annual survival rates (S) of 32.3 and 40.7%. After 2005, exploitation rates increased to 72.6 and 65.0% with S of 13.7 and 12.5%, suggesting that ﬁshing pressure increased on the eels after the use of shrimp nets began (Table 2).

Maturity

The 4 601 eels sexed from 1998 to 2007 (Table 1) were used to calculate maturation rates (represented by the proportion of

silver eels) at a given size class. Females at about 40 cm TL started to become silver eels, this proportion increasing greatly from 50 to 70 cm TL (Fig. 3). All females larger than 80 cm were silver eels (Fig. 3a). Males began to silver at 35 cm TL, the proportions thereof increasing considerably with length so that all males larger than 70 cm were silver eels (Fig. 3b). Both the logistic and De Leo & Gatto models ﬁtted the original data well and were indistinguishable (Fig. 3). The slope (b) of the logistic model was greater and the intercept (a) smaller in females (0.18 cm)1and)11.55) than for males (0.15 cm)1and )8.47). For the De Leo & Gatto model, cmaxwas similar for

both males and females (1.00), while k was larger and g was smaller for females (64.54 and 5.59 cm) than for males (57.41 and 6.78 cm). The length at 50% maturity for females (64.54 cm) was larger than for males (57.41 cm) (Table 3). Moreover, according to the ARSS, the silvering process diﬀered signiﬁcantly between the sexes for both the logistic and De Leo & Gatto models (Fc= 33.56 for the logistic and

20.34 for the De Leo & Gatto model, both P < 0.0001).

Discussion

Length-frequency analysis methods

For adequate results on growth from length frequency analysis, a total of 1 500 or more individuals collected over at least 6 months with an apparent shift in modal length over time is advisable (Pauly, 1987), as is a sample size of at least 10 times the number of length classes (Gerritsen and McGrath, 2007). In 2005–2007 the sample sizes were for most months suﬃciently large enough to satisfy the Gerritsen and McGrath (2007) rule of thumb (Table 2). The change in ﬁshing gear used in 2004 was believed to substantially alter the catchability of the eel, thus the data were separated into two periods for estimation of growth parameters. However, the estimation of Zduring 2005–2007 is likely biased due to apparent changes in mortality after 2004.

Due to the lower eel catch rate of the eel tubes, the sample size was low for 1998–2003, which may have inﬂuenced the estimation of growth parameters. But the temporal scale was still large, as eels were collected over a 24-month period, which Table 2

Observed and predicted maximum length (Lmax), corresponding estimate of asymptotic length (L¥) from Lmax, estimates of VBGF growth

parameters (L¥and K), growth performance index(Ø) by length frequency analysis (ELEFAN) and ShepherdÕs length composition analysis

(SLCA) and corresponding estimates of instantaneous total mortality rate (Z), natural mortality rate (M),ﬁshing mortality rate (F), exploitation rate (E), and annual survival rate (S) of Kao-Ping River Japanese eels, 1998–2007

Period 1998–2003 2005–2007 N 908 4894 Lmax(cm) Observed (cm) 78.50 87.50 Predicted (cm) 87.95 (81.52–94.39) 89.94 (82.31–97.57) L¥from Lmax(cm) 89.39 (84.11–97.17) 91.43 (84.92–100.39)

Method ELEFAN I SLCA ELEFAN I SLCA

Growth L¥(cm) 101.00 84.50 110.00 94.00 K(year)1) 0.33 0.38 0.30 0.44 Ø 8.12 7.91 8.20 8.26 Mortality Z(year)1) 1.13 (0.89–1.36) 0.90 (0.78–1.03) 1.99 (1.76–2.28) 2.08 (1.92–2.26) M(year)1) 0.59 0.68 0.54 0.74 F(year)1) 0.54 0.22 1.45 1.34 E(%) 47.8 24.4 72.6 65.0 S(%) 32.3 (25.7–41.1) 40.7 (35.7–45.8) 13.7 (10.3–17.4) 12.5 (10.4–14.7)

(5)

might reduce the bias resulting from the small sample size in this period. Length class width (5 cm) was determined from a prior search of several possible widths (1–10 cm) to ensure distinct peaks in length distribution and clear changes in model length. However, the eel tube is more selective than the shrimp nets (Lin and Tzeng, 2008), perhaps making results from the

1998 to 2003 period inconclusive and inconsistent between the diﬀerent methods of analysis. Conversely, shrimp nets were less selective and the number of eels caught per month was much larger (mostly above 160 individuals, except for 2 months). Consequently, the use of length-frequency data was believed appropriate to estimate the vital population statistics of A. japonicus for the 2005–2008 period.

Growth

Anguilla japonicus growth was assumed to follow a von Bertalanﬀy growth function, which has been veriﬁed by Lin and Tzeng (2009a). The optimal combination of L¥ and K

were estimated by ELEFAN and SLCA using response surface analysis because the estimates of L¥ and K are highly

correlated (Gulland, 1987; Shepherd, 1987). Both methods may be aﬀected by diﬀerences in growth strategies, individual variability in growth, seasonal oscillations of growth, variation

(a) (b)

(d) (c)

0 50 100 Probablity (%) Total length (cm) Obs Logistic De Leo & Gato

0 50 100 0 20 40 60 80 100 0 20 40 60 80 100 Probability (%) Total length (cm) Obs Logistic De Leo & Gato

(a)

(b)

Fig. 3. Proportion of silver eels as a function of body length for (a) females and (b) males. Obs, observed in a given length class (squares on solid line); Logistic, predicted using logistic model (open circles on grey line); De Leo = predicted using the De Leo and Gatto (1995) model (asterisks on broken line)

Table 3

Estimates of maturation parameters for the logistic and De Leo & Gato models Model Parameters Logistica a b(cm)1) L50(cm) F )11.55 0.18 64.54 M )8.47 0.15 57.41 Pooled )9.77 0.16 63.11

De Leo & Gatoa cmax k (cm) g (cm)1) L50(cm)

F 1.00 64.54 5.59 64.54

M 1.00 57.41 6.78 57.41

Pooled 1.00 63.10 6.46 63.11

L50= 50% length at maturity. F, females; M, males; and pooled

sexes. a_{sex-separate models diﬀered signiﬁcantly from pooled the}

(6)

in recruitment pattern, size-dependent selection, and the width of length classes (Isaac, 1990). The lower reach of the Kao-Ping River is tropical (Lin and Tzeng, 2009b), thus any seasonal oscillation in eel growth was assumed to negligible. Elvers recruit to the estuary regularly in winter from November to January and are not found after April (Tzeng, 1997). Thus, the recruitment season of the Japanese eel in Taiwan is considered to be stable, making the use the length-frequency data possible (Pauly, 1984).

L¥estimated by ELEFAN was not included by the 95% CI

of L¥from Lnax, implying that L¥ from ELEFAN might be

overestimated in this study. However, the uncertainty of growth parameters cannot be fully examined in this study. The variances of estimation of growth parameters were not given by FiSAT, thus statistical comparisons could not be con-ducted. It is still necessary to estimate the growth of A. japonicususing other data, e.g. lengths-at-age derived from annulus in the calciﬁed structures that would enable compar-isons of growth between sexes, time period and diﬀerent data sources (Morales-Nin, 1989).

The values of L¥and K for Japanese eel vary considerably

among diﬀerent regions. In this study, L_¥ was estimated between 84.5 and 112 cm when stages (i.e. yellow and silver eels) and sexes were pooled, while it was 98.2 cm for yellow eels from the Quilu River in southern China (Guan et al., 1994), and 55.7 cm for male and 77.5 cm for female silver eels from the Pearl River in southern China (Tzeng et al., 2000). Estimated L¥ values fell within the upper range of published

values, while K was higher (0.29 0.45 year)1) than in other studies (0.07 year)1in Guan et al., 1994 and 0.21 year)1 for males and 0.14 year)1for females in Tzeng et al., 2000). For Kao-Ping River eels the fishing pressure may be very high, perhaps reducing the intra-specific competition for food and suitable habitats and thus possibly enhancing individual growth rates. However, the relevance of this or other factors requires confirmation.

Mortality

Japanese eels in the Kao-Ping River probably faced a heavier ﬁshing pressure after the introduction of shrimp nets. The 95% CI of Z for 1998–2003 did not overlap with that for 2005–2007

with growth parameters derived from both ELEFAN and SLCA, suggesting a significant increase in Z in 2005–2007. This may be contributed by the introduction of shrimp nets, which are less-selective than eel tubes and have been widely used by eel fishermen (Lin and Tzeng, 2008). In addition, F and E for A. japonicus were generally higher with a lower S than for other fished eel species (Table 4). Thus, the Japanese eel stock in the lower reach of the Kao-Ping River may suffer higher fishing pressure than do the eels in other regions.

Hoyle and Jellyman (2002) predicted a reduction in spawner per recruit of 48% for female A. australis and 96.5% for female A. dieﬀenbachii under an annual E of 10%. Conse-quently, the impact of an annual exploitation rate larger than 60% in 2007 on the spawning biomass of the local Japanese eel stock might be severe. To enhance elver production, not only the harvest of elvers but also the biomass of eels in the river and escapement of silver eels to the spawning stock should be managed. Moreover, because of the time lag between a decrease in stock size and decrease in recruitment (7–14 years for A. anguilla; Feunteun, 2002) due to the catadromous nature of eels, where they spend about 1 year as migrating leptocephali and several years in freshwater or brackish waters, a precautionary approach (Russell and Potter, 2003) should be applied.

However, the estimates of Z in this study still have a large uncertainty. Because Z was estimated from a length-converted catch curve, the input of diﬀerent growth parameters would obviously alter the shape of the converted catch curve and thus the estimates of Z (Isaac, 1990). Also, M was derived from PaulyÕs empirical formula rather than estimated directly from the population (e.g. mark-recapture experiment), and is correspondingly sensitive to the uncertainties in the growth parameters. Thus, the obtained high M compared to other eel species (Table 4) is at least in part due to high values of the growth parameters. Moreover, the estimation of Z was still under the inﬂuence of eel tubes for some years after the introduction of shrimp nets. The catch curves in 2005–2007 (Fig. 2c,d) appear to have a sudden drop after the relative age of 2.5 years, implying a sudden increase in Z for older individuals. Studies to estimate mortality rates independent of these growth parameters are required to further evaluate the current exploitation rate of eels in the Kao-Ping River.

Table 4

Comparison of mean annual natural mortality (M), ﬁshing mortality (F), total mortality (Z), exploitation (E), and annual survival rate (S) among other eel species

Species Locality M(year)1) F(year)1) Z(year)1) E(%) S(%) Note References

A. anguilla Valli di Conacchio Lagoon, Italy

30 –90 30% for age 1 to age 2, 90% for other ages

De Leo and Gatto, 1995

A. anguilla West coast, Sweden 0.23 0.31 0.54 57 58 Sveda¨ng, 1999

A. anguilla Ijsselmeer Lake, Netherlands

0.14 to >2 F increases

with length

Dekker, 2000

A. australis New Zealand 0.038 10 E increases

with length

Hoyle and Jellyman, 2002

A. dieﬀenbachii New Zealand 0.036 10 E increases

with length

Hoyle and Jellyman, 2002

A. rostrata Chesapeake Bay, US 0.25 0.27 0.52 52 59 Weeder and Uphoﬀ,

2003 A. rostrata St Lawrence River,

Canada

19–24 Caron et al., 2003

A. Japonica Kao-Ping River, Taiwan

(7)

Maturity

The maturity of Japanese eels, as represented by the silvering process, was assumed to be a function of size alone. This assumption seemed reasonable because the silvering of the eels was found to be dependent on size, not on age, for eel species such as A. anguilla (Vøllestad, 1992; De Leo and Gatto, 1995; Bevacqua et al., 2006), A. rostrata (Oliveira, 1999), A. austra-lis, A. dieffenbachii (Jellyman, 2001; Hoyle and Jellyman, 2002), and A. japonica (Lin and Tzeng, unpubl. data). The silvering process was significantly different between female and male Japanese eels and the observed larger size of females at 50% maturity (Table 3) is consistent with the published data for this species (Tzeng et al., 2002, 2003; Han et al., 2003). Sex-dimorphism in silver eels is common in A. anguilla, A. ros-trata, A. australis and A. dieffenbachia (Poole and Reynold, 1996; Oliveira, 1999; Jellyman, 2001; Tzeng et al., 2003), perhaps related to sex-dependent life history strategies (Helfman et al., 1987).

In the logistic model, it is assumed that all individuals (i.e. 100%) become silver eels at some older age or larger length class. The De Leo & Gatto model is more plastic because of addition of cmax, the asymptotic probability of being silver

eels, which can adjust the upper level of the silvering curve. Given that this study is the ﬁrst to describe the maturation of A. japonica, no comparative published data are available for this species. For A. Anguilla, cmax appeared to be more

variable for females compared to males, being 0.12 ± 0.03 for females in the Valli di Comacchio Lagoon in Italy (De Leo and Gatto, 1995) and 0.60–1 in southern France (Bevacqua et al., 2006). For males in the two regions, cmax

generally approached 1. For A. japonica in Kao Ping River in southern Taiwan, cmax was 1 for both sexes in this study,

suggesting that for A. anguilla and A. japonica, nearly all males tend to silver upon reaching some critical size, while for females the tendencies to become silver eels vary among regions and species. In other words, the conditions for females to become silver eels seem to be more plastic than for males.

Conclusion

Due to restrictions in the data set of the first time period and to the change in fishing method, the data on growth and mortality remain uncertain and require confirmation. In the absence of large environmental fluctuations, natural mortality rates were likely similar throughout the study period. The estimated decrease in survival rate, though uncertain in detail, was probably realistic due to the use of new fishing gear types that increased the fishing mortality and exploitation rate considerably.

Acknowledgements

This study was ﬁnancially supported by the National Science Council, Taiwan (NSC94-2313-B-002-070). We would like to thank the cooperative ﬁshermen and Dr S. L. Chang, Division of Biotechnology, Fisheries Research Institute, Taiwan for collecting the eel samples and providing the hydrological and gear information, as well as Prof. S. C. Fong, Institute of Marine Biology, National Sun Yat-sen University, Taiwan for giving valuable suggestions for analyzing length-frequency data with FiSAT; and our colleagues in the Fisheries Science, National Taiwan University, Taiwan for preparing the

samples. Finally, we express our special thanks to B.M. Jessop for reviewing and giving helpful comments on the manuscript.

References

Bevacqua, D.; Mellia`, P.; Crivelli, A. J.; De Leo, G. A.; Gatto, M., 2006: Timing and rate of sexual maturation of European eel in brackish and freshwater environments. J. Fish Biol. 69, (Suppl. C), 200–208.

Campana, S. E., 2001: Accuracy, precision and quality control in age determination, including a review of the use and abuse of age validation methods. J. Fish Biol. 59, 197–242.

Caron, F.; Verreault, G.; Rochard, E., 2003: Estimation of the population size, exploitation rate and escapement of silver-phase American eels in the St. Lawrence Watershed. In: Biology, management, and protection of catadromous eels. A. D. Dixon (Ed.). Am. Fish. Soc. Symp. 33, 235–242.

Chen, Y. D.; Jackson, D. A.; Harvey, H. H., 1992: A comparison of von Bertalanﬀy and polynomial functions in modeling ﬁsh growth data. Can. J. Fish. Aquat. Sci. 49, 1228–1235.

Chang, C. W.; Tzeng, W. N., 1990: Traditional ﬁshing methods for eels in the river (in Chinese). Fish. Exten. NTU 12, 27–36.

Chen, P. H., 2005: Distribution, growth and reproduction of splendid ponyﬁsh, Leiognathus splendens (Cuvier 1829) in the coastal waters oﬀ southern Taiwan. Master Thesis. National Sun Yat-sen University, Kaohsiung, Taiwan.

Cheng, P. W.; Tzeng, W. N., 1996: Timing of metamorphosis and estuarine arrival across the dispersal range of the Japanese eel Anguilla japonica. Mar. Ecol. Prog. Ser. 113, 89–96.

Chu, Y. W.; Han, Y. S.; Wang, C. H.; You, C. F.; Tzeng, W. N., 2006: The sex-ratio reversal of the Japanese eel Anguilla japonica in the Kaoping River of Taiwan: the eﬀect of cultured eels and its implication. Aquaculture 261, 1230–1238.

De Leo, G. A.; Gatto, M., 1995: A size and age-structured model of the European eel (Anguilla anguilla L.). Can. J. Fish. Aquat. Sci. 52, 1351–1367.

Dekker, W., 2000: Impact of yellow eel exploitation on spawner production in Lake Ijsselmeer, the Netherlands. Dana 12, 25–40. Dekker, W., 2004: Epilogue: worldwide decline of eel resources

necessitates immediate action. Fisheries 28, 28–30.

Doole, J. G., 2005: Optimal management of the New Zealand longﬁn eel (Anguilla dieﬀenbachii). Aust J Agric Resour Econ 49, 395–411. Feunteun, E., 2002: Management and restoration of European eel population (Anguilla anguilla): an impossible bargain. Ecol. Eng. 18, 575–591.

Formacion, S. P.; Rongo, J. M.; Sambilay, V. C., 1992: Extreme value theory applied to the statistical distribution of the largest lengths of ﬁsh. Asian Fish. Sci. 4, 123–135.

Francis, R. I. C. C.; Jellyman, D. J., 1999: Are mean size data adequate to monitor freshwater eel ﬁsheries? Mar. Freshw. Res. 50, 355– 366.

Froese, R.; Binohlan, C., 2000: Empirical relationships to estimate asymptotic length, length at ﬁrst maturity and length at maximum yield per recruit in ﬁshes, with a simple method to evaluate length frequency data. J. Fish Biol. 56, 758–773.

Garc´ia, C. B.; Duarte, L. O., 2006: Length-based estimates of growth parameters and mortality rates of ﬁsh populations of the Caribbean Sea. J. Appl. Ichthyol. 22, 193–200.

Gayanilo, F. C., Jr; Sparre, P.; Pauly, D., 1996: FAO-ICLARM stock assessment tools (FISAT) userÕs manual. FAO Comput. Inf. Ser. Fish. 8, 126.

Gerritsen, H. D.; McGrath, D., 2007: Precision estimates and suggested sample sizes for length-frequency data. Fish. Bull. 106, 116–120.

Guan, R.; Wang, X.; Ke, G., 1994: Age and growth of eels Anguilla japonicain a Chinese river. J. Fish Biol. 45, 653–660.

Gulland, J. A., 1987: Length-based methods in ﬁsheries research: from theory to application. In: Length-based methods in ﬁsheries research. D. Pauly, G. R. Morgan (Eds). ICLARM Conf. Proc 13, 335–342.

Han, Y. S.; Tzeng, W. N., 2006: Use of the sex ratio as a means of resource assessment for the Japanese eel Anguilla japonica: a case study in the Kaoping River, Taiwan. Zool. Stud. 45, 255–263. Han, Y. S.; Tzeng, W. N.; Huang, Y. S.; Liao, I. C., 2001: Silvering in

the eel: changes in morphology, body fat content and gonadal development. J. Taiwan Fish. Res. 9, 119–127.

(8)

Han, Y. S.; Liao, I. C.; Huang, Y. S.; He, J. T.; Chang, C. W.; Tzeng, W. N., 2003: Synchronous changes of morphology and gonadal development of silvering Japanese eel Anguilla japonica. Aqua-culture 219, 783–796.

Haro, A.; Richkus, W.; Whalen, K.; Hoar, A.; Busch, W. D.; Lary, S.; Brush, T.; Dixon, D., 2000: Population decline of the American eel: implication for research and management. Fisheries 25, 7–16. Helfman, G. S.; Facey, D. E.; Hales, L. S., Jr; Bozeman, E. L., Jr, 1987: Reproductive ecology of American eel. In: Common Strategies of Anadromous and Catadromous Fishes, M. D. Bethesda, M. J. Dadswell, R. J. Klauda, C. M. Moﬃt, R. A. Saunders, R. L. Rulifson, J. E. Cooper (Eds). Am. Fish. Soc. Symp. 1, 42–56.

Hoyle, S. D.; Jellyman, D. J., 2002: Longﬁn eels need reserves: modeling the eﬀects of commercial harvest on the stocks of New Zealand eels. Mar. Freshw. Res. 53, 887–895.

Isaac, V. J., 1990: The accuracy of some length-based methods for ﬁsh population studies. ICLARM Technical Report 27.

Jellyman, D. J., 2001: The inﬂuence of growth rates on the size of migrating female eels in Lake Ellesmere, New Zealand. J. Fish Biol. 58, 725–736.

Jennings, S.; Kaiser, M. J.; Reynold, J. D., 2001: Marine ﬁsheries ecology. 1st edn. Blackwell Science, Maiden, MA.

Jones, R., 1984: Assessing the eﬀects of changes in exploitation pattern using length composition data (with notes on VPA and cohort analysis). FAO Fish. Tech. Pap. 256.

Kimura, D. K., 2000: Using nonlinear functional relationship regression to ﬁt ﬁsheries models. Can. J. Fish. Aquat. Sci. 57, 160–170.

Lin, Y. J.; Tzeng, W. N., 2008: Eﬀects of shrimp net and cultured eels on the wild population of Japanese eel Anguilla japonica in Kao-Ping River, Taiwan. J. Fish. Soc. Taiwan 35, 61–73.

Lin, Y. J.; Tzeng, W. N., 2009a: Modelling the growth of Japanese eel Anguilla japonica in the lower reach of the Kao-Ping River, southern Taiwan: an information theory approach. J. Fish Biol. 75, 100–112.

Lin, Y. J.; Tzeng, W. N., 2009b: Validation of annulus in otolith and estimation of growth rate for Japanese eel Anguilla japonica in tropical southern Taiwan. Environ. Biol. Fish 84, 79–87. Morales-Nin, B., 1989: Growth determination of tropical marine ﬁshes

by means of otolith interpretation and length frequency analysis. Aquat. Living Resour. 2, 241–253.

Morales-Nin, B.; Panﬁli, J., 2005: Seasonality in the deep sea and tropics revisited: what can otoliths tell us? Mar. Freshw. Res. 56, 585–598. Oliveira, K., 1999: Life history characteristics and strategies of American eel, Anguilla rostrata. Can. J. Fish. Aquat. Sci. 56, 795–802.

O¨zbilgin, H.; Tosunoglu, Z.; Bilecenoglu, M.; Tokac¸, A., 2004: Population parameters of Mullus barbatus in Izmir Bay (Aegean Sea), using length frequency analysis. J. Appl. Ichthyol. 20, 231– 233.

Pauly, D., 1980: On the interrelationships between natural mortality, growth parameters and mean environmental temperatures in 175 ﬁsh stocks. J. Cons. Int. Explor. Mer 39, 175–192.

Pauly, D., 1984: Fish dynamics in tropical waters, a manual for use with programmable calculators. ICLARM Reviews 8, Manila. Pauly, D., 1987: A review of the ELEFAN system for analysis

length-frequency data in ﬁsh and aquatic invertebrates. In: Length-based methods in ﬁsheries research, D. Pauly, G. R. Morgan (Eds). ICLARM Conf. Proc. 13. 7–34.

Pauly, D., 1990: Length-converted catch curves and the seasonal growth of ﬁshes. Fishbyte 8, 33–38.

Pauly, D.; G. R. Morgan. (Eds), 1987: Length-based methods in ﬁsheries research. ICLARM Conf. Proc. 13, 468.

Pauly, D.; Munro, J. L., 1984: Once more on growth comparison in ﬁsh and invertebrates. Fishbyte 2, 21.

Poole, W. R.; Reynold, J. D., 1996: Growth rate and age at migration of Anguilla anguilla. J. Fish Biol. 48, 633–642.

Quinn, J. T., II; Deriso, R. B., 1999: Quantitative ﬁsh dynamics. Oxford University Press, New York.

Robitaille, J. A.; Be´rube´, P.; Tremblay, S.; Verreault, G., 2003: Eel ﬁshing in the Great Lakes⁄ St. Lawrence River system during the 20th century: Signs of overﬁshing. In: Biology, management, and protection of catadromous eels,A. D. Dixon, (Ed.). Amer. Fish. Soc. Symp. 33, 253–262.

Russell, I. C.; Potter, E. C. E., 2003: Implications of the precautionary approach for the management of the European eel, Anguilla anguilla. Fish. Manage. Ecol. 10, 395–401.

Shepherd, J. G., 1987: A weakly parametric method for estimating growth parameters from length composition data. In: Length-based methods in ﬁsheries research, D. Pauly, G. R. Morgan, (Eds). ICLARM Conf. Proc. 13. pp. 113–119.

Sveda¨ng, H., 1999: Vital population statistics of the exploited eel stock on the Swedish west coast. Fish. Res. 40, 251–265.

Tatsukawa, K., 2003: Eel resources in East Asia. In: Eel biology, K. Aida, K. Tsukamoto, K. Yamauchi (Eds). Springer Verlag, Tokyo, Japan, pp. 293–298.

Tesch, F. W., 2003: The Eel. 3rd edn. Blackwell Science, Oxford. Tseng, M. C.; Tzeng, W. N.; Lee, S. C., 2003: Historical decline in the

Japanese eel Anguilla japonica in northern Taiwan inferred from temporal genetic variations. Zool. Stud. 42, 556–563.

Tsukamoto, K., 1992: Discovery of the spawning area for Japanese eel. Nature 356, 789–791.

Tsukamoto, K., 2006: Spawning of eels near a seamount: tiny transparent larvae of the Japanese eel collected in the open ocean reveal a strategic spawning site. Nature 493, 929.

Tzeng, W. N., 1997: Short- and long-term ﬂuctuations in catches of elvers of the Japanese eel, Anguilla japonica. Developing and Sustaining World Fisheries Resources: the State of Science and Management. 2nd World Fisheries Congr. Proc., Victoria, Australia, pp. 85–89.

Tzeng, W. N.; Chang, C. W., 2001: Stock status and management prospect of freshwater eel Anguilla spp. in Taiwan. J. Taiwan Fish. Res. 9, 251–258.

Tzeng, W. N.; Lin, H. R.; Wang, C. H.; Xu, S. N., 2000: Diﬀerences in size and growth rates of male and female migrating Japanese eels in Pearl River, China. J. Fish Biol. 57, 1245–1253.

Tzeng, W. N.; Shiao, J. C.; Iizuka, Y., 2002: Use of otolith Sr:Ca ratios to study the riverine migratory behaviours of Japanese eel Anguilla japonica. Mar. Ecol. Prog. Ser. 245, 213–221.

Tzeng, W. N.; Iizuka, Y.; Shiao, J. C.; Yamada, Y.; Oka, H. P., 2003: Identiﬁcation and growth rates comparison of divergent migra-tory contingents of Japanese eel (Anguilla japonica). Aquaculture 216, 77–86.

Velasco, G.; Reis, E. R.; Vieira, J. P., 2007: Calculating growth parameters of Genidens barbus (Silurifomes, Ariidae) using length composition and age data. J. Appl. Ichthyol. 23, 64–69. Vøllestad, L. A., 1992: Geographic variation in age and length at

metamorphosis of maturing European eel: environment eﬀects and phenotypic plasticity. J. Anim. Ecol. 61, 41–48.

Weeder, J. A.; Uphoﬀ, J. H. J., 2003: Eﬀect of changes in growth and eel pot mesh size on American eel yield per recruit estimates in upper Chesapeake Bay. In: Biology, management, and protection of catadromous eels, A. D. Dixon (Ed.). Amer. Fish. Soc. Symp. 33, 169–176.

AuthorÕs address: Wann-Nian Tzeng, Institute of Fisheries Science, College of Life Science, National Taiwan Univer-sity, Taipei, Taiwan 106, ROC.