blstituto di lstologia ed Embriologia, Universith di Palermo, Via Archirafi 20, 90123 ..... 25th European Marine Biology Symp., Ed. Ferrara 1990, Olsen & Olsen.
ELSEV I ER
Fisheries Research 20 (1994) 9%107
Otolith reading as a tool for stock identification D. Levi a'*, M.G. Andreoli b, E. Arneri c, G. Giannetti c, P. Rizzo a alstituto di Tecnologia Pesca e Pescato, Consiglio Nazionale delle Ricerche, Via L. Vaccara n. 61, 91026 Mazara del Vallo (TP), Italy blstituto di lstologia ed Embriologia, Universith di Palermo, Via Archirafi 20, 90123 Palermo, Italy CLR.P.E.M., C.N.R., Molo Mandracchio, 60100 Ancona, Italy
(Accepted 29 March 1994)
Abstract A method is presented for using otolith readings to identify stocks, and a refined statistical analysis of age/length keys is proposed based on 95% probability ellipses of log-log plots. Three case studies are given: ( 1 ) an application for the identification of the redmullet stock in the central Mediterranean; (2) an approach for the identification of stock units of anchovies in the Adriatic Sea; (3) an "ab contrario" proof of the validity of the approach on hake in the central Mediterranean. The final discussion deals mainly with the justification of methods used. Evidence of independent circulation of water masses as gathered from oceanographic mesoscale studies seems to parallel preliminary evidence of unit fish stocks from statistical treatment of age-length relations.
1. Introduction A problem which is often neglected in population modelling is the identification o f the object to be modelled (Levi et al., 1990); so, one o f the major causes o f failure o f population dynamics when applied to the m a n a g e m e n t o f marine fisheries can be the p o o r delimitation o f unit stocks (Lassen, 1990). The distribution o f a fish stock is determined and delimited by its migration pattern. H a r d e n Jones ( 1968 ) notes that " t h e migration pattern is believed to be related to that o f the currents: the young stages drift with the current to the nursery ground, and *Correspondingauthor. 0165-7836/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0165-7836 ( 94 )00295-8
98
D. Levi et al. / Fisheries Research 20 (1994) 97-107
the spents return to the nursery ground; the spawning migration is against the current, and the spents return to the feeding ground with the current". The general model proposed by Harden Jones encompasses a range of situations from large migratory pelagic species in which the migration pattern is related to ocean gyres, to coastal rock-fish, in which the pattern is related to local tidal and coastal currents. Between these two extremes is the migration pattern of small, shoaling pelagics and of most demersal species trawled on the shelf. Their migration patterns are possibly related to the mesoscale currents. These can thus be considered as clues to the identification of stocks inhabiting the area therein delimited. The perception of unit stocks has grown through research on commercial fisheries for some decades now, and is mostly based on the summation of the outcomes of many different approaches in time, from tagging to immunogenetics. None of these approaches alone is ultimately conclusive and all require long and committed research. The method we propose probably has some major drawbacks and weaknesses, but has the undoubted advantage of being simple, as the data on which it is based are usually the first and easiest to be collected. It could be particularly useful for sea areas where biological and statistical data are sparse, e.g. the Mediterranean and most tropical seas. Also, it is more easily applicable to meaningful numbers of fish, allowing considerations to be formulated at the population level.
2. Materials and methods
Two different areas of the Mediterranean Sea were considered, the Sicily Channel and the northern-central Adriatic Sea (Fig. 1 ). In both areas, independent water masses and current systems are identifiable on the basis of oceanographic literature. In the Sicily Channel, stocks of the red mullet (Mullus barbatus, L. 1758 ) were studied. This species spawns in coastal water and has pelagic eggs and larvae. The adults swimming capacity is inferior to that of Mediterranean hake (Merluccius merluccius), but possibly superior to that of most coastal rock-fish, such as sparids. In the northern-central Adriatic Sea, stocks of anchovy, Engraulis encrasicholus (L. 1758 ) were studied. This is a small pelagic shoaling species and hence, more strictly linked to water currents throughout its life cycle. The test we developed as a working hypothesis is based on a statistical treatment of the growth characteristics of the adult fraction of the population. Hard body structures such as otoliths and scales are usually collected from the various length groups of fish populations to determine age (Holden and Raitt, 1974). Ideally age-readings should be proportional to age variance within a given length group and to the numbers in such groups of the population (Levi and Mortera, 1981; Mortera and Levi, 1981 ), but this is seldom the case as these variables are not known a priori, when sampling occurs. Initially, therefore, a higher than necessary number of otoliths is usually collected per length group. A fixed number of
99
D. Levi et al. / Fisheries Research 20 (1994) 97-107
:2J
fh
I
4/
?
S. Benedet to
L/
\
J-
~
8 o
9"
I0"
11 °
1~'
?r"
13°
l,f
If;'
16 ~
I ,'
i~'"
10 ~
Fig.1.Studyareas. them is analysed and where uncertainty a n d / o r variation is high, a further, variable number is analysed in a second phase. This additional analysis is usually necessary at the tails of the length distributions. Age/length keys are then constructed, from which average lengths at age and average ages at length with variance estimates can be drawn; from age/length keys it is then possible to obtain either the age structure of the population sampled for length, or part of the growth curve of the sub-sampled population. We tried yet another utilization of the same data, assuming they can also help in discriminating unit stocks (this is more widely discussed in the final section ). The bivariate set of data of average lengths at age and average ages at length was used to draw their confidence ellipses. Confocal ellipses of these data represent various areas under the bell-shaped surface of the bivariate normal distributions. For the ellipse we chose Jacobi's method (Shoup, 1985 ) to find the eigenvalues and vectors of a real symmetric matrix of the covariance matrix. This was in turn computed on the log transformed bivariate data set of average lengths and corresponding ages. The programme PRINCOMP by Sayla et al. ( 1988 ), was utilized for this pur-
lOO
D. Levi et al. / Fisheries Research 20 (1994) 97-107
pose. The eigenvalues here are analogous to variance values, with the eigenvector being the major and minor axes of the ellipses. In data as highly correlated as ours, represented by rather narrow ellipses, most of the variance was accounted for by the major axis, which was always much higher than the minor one (Sokal and Rohlf, 1981 ). Thus, the basic assumption of our method is that overlapping of the ellipses can be used as an inverse index of stock identity, being a geometric image of how age varies with length within a given population. In turn, a further assumption underlies this basic one: that a stock can be identified by its age-length relationship. This is discussed in the final section.
3. Results
3.1. Mullus barbatus - - central Mediterranean Region A of Fig. 1 is currently being investigated by seasonal trawl-surveys by the Institute of Fisheries and Fish Technology of Mazara del Vallo (Levi, 1988 ). Otoliths collected from 1200 fish, spanning the whole range of fish lengths (93322 m m ) from some 600 haul stations and from Stratified Random Sampling designed surveys lasting 2 years ( 1985/1986 - 1986/1987), were looked at to determine age, and 320 'safe' readings were eventually retained for constructing a single age-length key for the period considered. Age-length data from a study conducted in 1979 in the Gulf of Gabes, (Fig. 1 ), using comparable techniques for age determination, were available (Gharbi and Ktari, 1981 ). Plotted on the same graph, the 95% probability equal frequency ellipses for the two sets ofbivariate data give rise to only a very minor overlapping of the ellipses. (Fig. 2) Published data allowed further comparison with analogous data from the same species elsewhere in the Mediterranean, and other evidence was gathered suggesting that the Mullus barbatus in area A is probably part of a unique stock belonging to what was named an 'Eastern Aggregate', sharing similar probability distributions with populations sampled in the Aegean Sea, Cyprus and Israel (Levi et al., 1990). 3.2. Merluccius merluccius - - central Mediterranean
Otoliths were collected from the trawl surveys mentioned above and the age ,:f 629 hake in the 14-74 cm length range was determined. From Tunisian wate=, similar data were available for 1971 on 1455 hake ranging from 8 to 56 c~J~ (Bouhlal and Ktari, 1975 ). A similar calculation and plotting of 95% probability equal frequency ellipses for the two sets of data (i.e. estimated age vs. average corresponding lengths) resulted in ellipses with a high degree of overlap. (Fig. 3) Here, the picture is different from that seen previously for red mulle" i n that the Tunisian 95% prub-
D. Levi et al. / Fisheries Research 20 (1994) 97-107
101
1.60
1 150
2
I
~)
1.40
c
1.30
I
'~ 11 ~x
1.20
l
~
1,10
(¢~'~)
1.00
0.90 0.80 -0.30
I 0.00
I 0.30
iog
I 0.60
o£
I 0.90
a~-e
! 1.20
(y.)
Fig. 2. Ninety-five percent equal frequency ellipse plot. 1, Tunisia, 1979 (Gharbi and Ktari, 1981 ); 2, Sicily, 1985-1987 (Levi et al., 1990). 2.202.101
o
2.00"
g
1.90
o f
1.80
1 e n g t h s (c m)
1.70 1.60 1.50 1.40 1.30 1.20 1.10 1.00 0.90 -0.40
I - 0.10
I 0.30
log
I 0.70
of
I 1.10
ag*-
I 1.50
(y . )
Fig. 3. Ninety-five percent equal frequency ellipse plot. 1, Tunisia, 1971 (Bouhlal a n d Ktari, 1975 ). 2, Sicily, 1 9 8 5 - 1 9 8 7 (Levi et al., 1990).
ability equal frequency ellipse is almost entirely inside the ellipse constructed on data from Sicilian waters. The evidence seems here to be for a unique stock inhabiting the vast area of the Central Mediterranean Sea between Sicily and Tunisia. This calls for pooling of data collected in both parts of the Sicilian Channel before proceeding with any
A
-0,SN0
@.8@@@
-O2N0
1,20N
i,50N
] .......
0,91100
1,300
I
o •o~
log Of lengths (cm)
log of lengths (cm)
E
/ 2
l,O0O
11.O@
1,200
1.300
l
F
2
13@@
10g of lengths (cm)
................ -0.5000 -0.ZOO0 O.2UO I.SO00 8,9000 1,3U
...... -~'----
f
log Of lengths (cm)
6.8606 . . . . . . . . . . . . . . -O,SOOO -11,21166 0,2660 O,~OOO 0.9060
0.B0@0
1,3N
11,@@
1,300
1'3~
@,Sm -0.Sm
0,21100 O,SIIN I.gH0
1
(cm)
o~@9.@0@ []eL
1,2@~
l,31d9
6,9m
-O,ZM
(cm)
log 0£ lengths
1
@,gM
l,m
I.IU
1,2N.
log of lengths
c
0,8666 . . . . . . . . . . . . . . . -6,5M -O,~m 0,2900 0,5006 @,~OOO
°~ 0,~066 I
1.200
1,300
Fig. 4. Ninety-five percent equal frequency ellipse plot for Adriatic anchovies (1986). A: ( 1 ) San Benedetto, ( 2 ) Chioggia; B: ( 1 ) San Benedetto, ( 2 ) Ancona; C: ( 1 ) Vieste, (2) Chioggia; D: ( 1 ) Vieste, ( 2 ) Cattolica; E: ( 1 ) Vieste, ( 2 ) Ancona; F: ( 1 ) Vieste, ( 2 ) San Benedetto.
g
,f
I
-0,5@@@ -@.~@@@ 62@@@ O,~O6@ 0,9@@@ 13@@
@,80~
0,9000
L,066
1,100
L,20O
IA@@
eb
,,,..
t~
0 t-,J
D. Levi et al. / Fisheries Research 20 (1994) 97-107 1.3@0
7
I -
103
I
1
2
..~_
0~
j t~'-
o
H @.SM -g.SM
-11.2m
8,2888
0.50~
O.911W
log Of lengths
0.9~9 j
,.3~ (cm)
6.8~
t~
L
. l o g Of Le~cJLhs (C7)
Fig. 5. Ninety-five percent equal frequency ellipse plot for Adriatic anchovies ( 1989 ). G: ( 1 ) Vieste, (2) Chioggia; H: ( 1 ) Vieste, (2) Cattolica; h ( 1 ) Vieste, (2) Ancona; L: ( 1 ) Vieste, (2) San Benedetto.
further analysis of the hake population, which we were not justified in doing for red mullet in part A of Fig. 1. 3.3. Engraulis encrasicholus - - Adriatic Sea. A routine programme of biological sampling in ports and collection of catcheffort data from commercial fisheries on small pelagic fish has been conducted since 1975 in the Adriatic Sea by the CNR (NRC) Institute of Ancona (Levi, 1978 ) (Fig. I ). Semi-pelagic trawlers fish mostly in areas adjacent to or not far from their base ports. The present method was used to assess whether different unit stocks of anchovy inhabit the study area which is extended over four degrees of latitude. In this case, a reasonable working hypothesis could be established in the light of the different history of fishing in the northern and southern Adriatic, showing the shift in crisis years (Bombace, 1990), and based on the empirical evidence of what fishermen call 'big black' anchovies in the south compared with 'small silvery' ones in the north. Two years were selected for reading trials because they were immediately be-
104
D. Levi et al. / Fisheries Research 20 (1994) 97-107
fore and immediately after a collapse of the fishery: 1783 otoliths were analysed for 1986 and 1584 for 1989. If, as previously, we compare the 95% probability equal frequency ellipses for each year of fish sampled in ports from north to south (Fig. 1 ), in 1986 the 95% probability ellipse of the southern port of S. Benedetto is markedly distinct from all the others in a higher position, indicating bigger animals at the same age (i.e. the black, big anchovies). (Figs. 4 (a) and (b) ). The line for the southernmost port of Vieste is also higher (Figs. 4 (c), (d), (e)). Its ellipse is completely separable from that of the northernmost port of Chioggia and only partly overlaps with Cattolica and Ancona. It does, however, overlap more with its neighbour S. Benedetto (Fig. 4 ( f ) ) . For 1989, Vieste's ellipse is still the most separate and distinct from that of Chioggia (Fig. 5 (g)). It does not overlap with Cattolica, the port south of Chioggia (Fig. 5 (h)), nor with Ancona (Fig. 5 (i)), though it fully overlaps with its neighbour S. Benedetto (Fig. 5 (1)). Plotting the ellipses on the same graph would blur the picture, so we present separate ellipses for each port. On the whole, both in 1986 and 1989 it seems possible to distinguish between a northerly stock, fished from the ports of Chioggia, Cattolica and Ancona and a southerly one, fished from the ports of Vieste and S. Benedetto.
4. Discussion
The middle area of the Sicily Channel, below a depth of 250-300 m, is characterized by a water mass moving east-west, the Levantine Intermediate Water (LIW), with high salinity (38.8) and stable temperature of around 14 °C (De Maio et al., 1990). The surface layers are less dense and colder, and arrive from the Atlantic Ocean through the Straits of Gibraltar (Manzella et al., 1988; Hopkins et al., 1990). The southern limit of the water flow from Gibraltar which touches the Sicilian coastline can be identified in an ideal frontal isohaline between Sm> 37.50 to the north and Sm< 37.50 to the south. (Grancini et al., 1972). An overall picture on two different circulation systems in part A and B of Fig. 1 can thus be drawn (Mosetti, 1989). This complex environmental situation seems to hinder full genetic mixing of red mullets but not of hakes. It may be suggested that strong swimmers like hake are less sensitive to oceanographic barriers than apparently less active swimmers like red mullet. Other evidence supporting the existence of two separate stocks of red mullet, comes from parasitological observations. A large infestation by a trematode of the genus Stephanostomum seriously affected the red mullet fishery in the Tunisian waters for several months in 1990 (Laamouri and M'Hetli, personal communication, 1991 ). No such occurrence has so far been noted in the fish landed at the Sicilian base-ports of the present study area.
D. Levi et al. / Fisheries Research 20 (1994) 97-107
105
Also the northern and central Adriatic Sea are characterized by different physical and oceanographical properties. Average depth, salinity, temperature, input of nutrients and principal current patterns indicate a degree of separation between the two basins (Zorr, 1956; Artegiani et al., 1993 ). The circulation in the northern Adriatic Sea is strongly influenced by the variable inflow of the Po river, while the central Adriatic cyclonic gyres are strongly influenced by the inflow from the Jonian Sea through the Strait of Otranto (Franco et al., 1982). Two somewhat independent circulation patterns are therefore evident in the two parts of the Adriatic Sea (Mosetti, 1989 ). Piccinetti (1970) postulated that anchovies are capable of extensive migrations throughout the central and northern Adriatic and that there are no separate spawning grounds but did not reject the hypothesis of two populations, one located in the northern and the other in the central Adriatic. Stirn (1974), using echo-survey data, suggested that anchovies of the northern Adriatic migrate only inside this area. This migration pattern was also suggested by Varagnolo ( 1967 ) on the basis of the movements of the commercial fishing fleet during the year. Our findings seem to suggest two independent stocks of anchovies in the northern and central parts of the Adriatic Sea. Can our method, in the absence of better information, be considered as an acceptable 'test' of stock independence, when only age-length data are available for comparison, as in our case? In recent years, the amount of information on the growth process of fish has increased, mainly as a result of its statistical correlation with natural mortality (Pauly, 1980). Growth of fish is one of the first phenomena to be quantified and modelled. It can be viewed as a result of genetic adaptation (Ginzburg, 1981 ). As such, its variability should be higher at the boundaries of the distribution area (MacCall, 1990). The particular environmental parameter likely to influence growth is the temperature of water masses in the distribution area of the stock. Krueger (1964), quoted by Laevastu (1984) made a comparative study of the effects of temperature on the growth rates offish and of the formulae representing these effects. He found that the well-known formula of Arrhenius presenting the effect of temperature on the speed of chemical reactions reproduces best the effect of temperature on growth rate. This conclusion was experimentally determined a few decades earlier by Krogh (see Krueger, op. cit. ). If body growth is proportional to otolith growth, and if a detectable difference in otolith growth can be used to discriminate between stocks (e.g. Dawson, 1986; Umezawa and Tsukamoto, 1990; Moskness and Fossum, 1991 ) then fish growth alone could just as well be used to discriminate stocks, provided its variability is taken into account, and this is what our method tries to do. References Artegiani, A., Gacic, M., Michelato, A., Scarazzato, P., Kovacevic, V. and Russo, N., 1993. The Adriatic Sea hydrography and circulation in spring and autumn (during 1985-1987 ). Deep Sea Res. 2, 40 (6): in press.
106
D. Levi et al. / Fisheries Research 20 (1994) 97-107
Bombace, G., 1990. Fisheries of the Adriatic Sea. In: Marine Eutrophication and Population Dynamics. In: Proc. 25th European Marine Biology Symp., Ed. Ferrara 1990, Olsen & Olsen. Bouhlal, M. and Ktari, M.H., 1975. Croissance du merlu de la region du golfe de Tunis. Bull. Inst. Nat. Sci. Tech. Oceanogr. Peche. Salammbo, 4( 1): pp. 5-47. Dawson, W.A., 1986. Mackerel (Scomber scombrus L. ) otolith L analysis as a method of stock separation. ICES-CM-1986/H: 24, 13 pp. De Maio, A., Moretti, M., Sansone, E. and Spezie, G., 1990. Dynamical and hydrological features of the water masses in the Sicily Channel. In: Fourth POEM Scientific Workshop-Venezia, 2 pp., mimeo. Franco, P., Jeftic, L., Malanotte Rizzoli, P., Michelato, A. and Orlie, M., 1982. Descriptive model of the northern Adriatic. Oceanol. Acta, 5 (3): 379-389. Gharbi, H. and Ktari, M.H., 1981. Croissance des rougets en Tunisie. Bull. Inst. Nat. Sci. Tech. Oceanogr. Peche. Salammbo, 8: 5-40. Ginzburg, L.R., 1981. Genetic adaptation and models of population dynamics. In: D.G. Chapman and V.R. Gallucci (Editors), Quantitative population dynamics. Int. Coop. Publ. House, Fairland, MD. Grancini, G., Lavenia, A. and Mosetti, F., 1972. A contribution to the hydrology of the strait of Sicily. In: Oceanography of the Strait of Sicily. Sanclantcen Conference Proceedings, 7:68-81. Harden Jones, F.R., 1968. Fish migration. Edward Arnold, London, 325 pp. Holden, M.J. and Raitt, D.F.S. (Editors), 1974. Manual of fisheries science. Part. 2. Methods of resource investigations and their application. FAO Fish. Tech. Pap., ( 115 ) Rev. 1:24 pp. Hopkins, T.S., Minnett, P, J. and Nacini, E., 1990. Atlantic water in the Strait of Sicily. J. Geophys. Res., 95 (C2): 1569-1575. Krueger, F., 1964. Neuere matematische Formulierungdes biologischen temperatur-funktionund des wachstums. Helgolander Wiss. Neeresunters., 9:108-124. Laevastu, T., 1984. The magnitudes of the effects of temperature anomalies on the fish stocks as compared to the effect of fishing. ICES-C.M. 1984/C:3, 14 pp. Lassen, H., 1990. Biological uncertainties in Fish Stock Management: A Discussion, NAFO Scientific Council Studies Number 16. Special Session on Management under uncertainties; 5-7 September 1990, pp. 165-169. Levi, D., 1978. A summary report on the routine sampling programme for sardine and anchovy in the Adriatic. FAO, Rapport de la Septieme session du Group de Travail sur l'evaluation des resources et les statistiques de peche du Conseil General des Peches pour la Mediterran6e (CGPM). FAO, Rapports sur les peches, 204, pp. 115-117. Levi, D. and Mortera, J., 1981. Sampling problems, methods of extraction, storage, preparation, mountingof bony parts and reading techniques in fish age determination. General Fisheries Council for the Mediterranean (GFCM). Annex C of the Report of the technical consultation on methodologies used for fish age--reading. 5-9 October 198 l, Montpellier, FAO Rapport sur les peches, 257, pp. 47-62. Levi, D., 1988. Canale di Sicilia. Relazione sull'attivith svolta dall'Unith Operativa lstituto di Tecnologia della Pesca e del Pescato - - Mazara del Vallo. In: Atti Semin. Unit~ Oper. Resp. Progr. Ric. prom. nell'ambito dello Schema preliminare di piano per la pesca e l'acquacoltura. Ministero Marina Mercantile, Consiglio Nazinale delle Ricerche, Roma, 3:156 l - 1767. Levi, D., Andreoli, M. G. and Rizzo, P., 1990. Growth curves from representative samples of fish populations as possible hints for the identification of unit stocks. The case study of red mullet from the Sicilian Channel. In: Marine Eutrophication and Population Dynamics, Proceedings of the 25th European Marine Biology Symposium, Ed. Ferrara 1990, Olsen & Olsen, pp. 299-306. MacCall, A.D., 1990. Dynamics Geography of Marine Fish Populations. Washington Sea Grant Program, Books in Recruitment Fishery Oceanography, 153 pp. Manzella, G.M.R., Gasparini, G.P. and Astraldi, W., 1988. Water exchange between the eastern and western Mediterranean through the Strait of Sicily. Deep Sea Res., 35 (6): 1021-1035. Mortera, J. and Levi, D., 1981. Influence du biais dans la lecture de l'age sur la cl6 ages/longueurs, la courbe de croissance et l'analyse de populations virtuelles. Conseil General des Peches pour la Mediterranre (CGPM). Annex D du Rapport de la Consultation Technique sur les mdthodologies
D. Levi et al. / Fisheries Research 20 (1994) 97-107
107
utilizres dans la lecture de l'age des poissons. 5-9 October 1981, Montpellier, FAO Rapport sur les peches, 257: 63-72. Mosetti, F., 1989. Mari. In: Atlante Tematico d'Italia. Touring Club Italiano, Consiglio Nazionale delle Ricerche, 15. Moskness, E and Fossum, R., 1991. Distinguishingspring and autumn-spawned herring larvae (Clupea harengus L. ) by otolith microstructure. ICES J. Mar. Sci., 48:61-66. Pauly, D., 1980. On the interrelationship between natural mortality, growth parameters and mean environmentaltemperature in 175 fish stocks. J. Cons. Int. Explor. Sea, 39 (2): 175- 192. Piccinetti, C., 1970. Considerazioni sugli spostamenti delle alici (Engraulis encrasicholus L. ) nell'Alto e Medio Adriatico. Boll. Pesca Pisc. Idrobiol., 25 ( 1): 145-158 Saila, S.B., Recksiek, C. and Prager, M.H., 1988. Basic fishery science programs. A Compendium of Microcomputer Programs and Manual of Operation. Elsevier, Amsterdam, 230 pp. Shoup, T.E., 1985. Numerical Methods for the Personal Computer. Prentice-Hall, NJ, 236 pp. Sokal, R.R. and Rohlf, F. J., 1981. Biometry 2nd. Edn., W.H. Freeman, San Francisco, 859 pp. Stirn, J., 1974. Contributions to the knowledge of migrations and the volume of the pilchard and anchovy populations in the Northern Adriatic. Acta Adriatica, 16 (24): 401-422. Umezawa, A. and Tsukamoto, K., 1990. Determination of stock origin of the ayu based on the otolith microstructure. Jiseki bisai kozo ni yoru kosan-ayu no hanbetsuho. Nippon Suisan Gakkaish/Bull. Ja. Soc. Sci. Fish., 56 (12): 1919-1926. Varagnolo, S., 1967. Analisi della produzione ittica dei mercati di Chioggia e di Venezia. Arch. Oceanogr. Limnol. 15 (suppl.) :201-235. Zore', M., 1956. On gradient current in the Adriatic Sea. Acta Adriatica, VIII, (6)36.
