NOTES Stergiou Forecasting the anchovy fishery in the eastern Mediterranean 



413 



ographic conditions. Catches exhibit a marked seasonal 

 pattern (Figs. 1, 2) which is hkely related to these 

 seasonal offshore and inshore migrations. Purse-seine 

 fishing activity in Greek waters does not operate in the 

 open sea but is mainly restricted to coastal areas where 

 schools of anchovy migrate on a seasonal basis. The an- 

 chovy starts its inshore migration in early spring, but 

 the peak occurs in coastal waters in May- August and 

 schools disperse again during late summer-fall (Fig. 

 2; Tsimenidis and Caragitsou 1984). 



In Greek waters trawling and coastal seining are pro- 

 hibited from the 1st of June to the 30th of September. 

 As a result, the landings of demersal species (e.g., 

 Merluccius merluccius, Micromesistius poutassou, 

 Lophius sp., MuUus sp., Pagellus sp.) are low. This com- 

 bined with the increased demand for fish in summer, 

 drives up fish prices in Greece. In this context, accurate 

 forecasts of the catches of anchovy (and of pelagic fish 

 in general) during June-September are essential for 

 market and industrial planning. APE in June-Septem- 

 ber was <11.2% in 1985 (mean 7.7%) and < 10.6% in 

 1986 (mean 7.8%). 



Strict monitoring and accurate forecasts are essen- 

 tial for pelagic fisheries that are heavily dependent on 

 a single year-class, inasmuch as they are prone to col- 

 lapse when conditions for recruitment in a particular 

 year are not favorable and fishing is intense (as for 

 anchovy in Greek waters). In this context, accurate 

 forecasts of the annual anchovy catch together with 

 information related to optimal management of an- 

 chovy estimated by deterministic fishery-management 

 models, can be used by resource managers for the 

 preseasonal adjustment of anchovy fishing mortality. 

 The model produced reasonable forecasts of the annual 

 1985 and 1986 catches. Total observed annual catches 

 in 1985 and 1986 were 17 544 t and 18 339 t, respec- 

 tively, while the model predicted 17 369 t and 20 210 t 

 (APE 1% and 10%, respectively). It must be pointed 

 out, however, that although the model will most likely 

 produce accurate forecasts if the anchovy stock is 

 under equillibrium, it may fail to produce reliable 

 forecasts for years characterized by weak year-classes 

 of anchovy. (In fact there may be a lag of some months 

 between the occurrence of a turning point and its 

 recognition by the model; for a general discussion on 

 turning points and out-of-sample forecasting, see 

 Schlegel 1985.) 



The model presented here has also an interesting 

 biological interpretation. The seasonal-difference term 

 of the model indicates the seasonal migratory nature 

 of anchovy in Greek waters. Moreover, climate-plank- 

 ton-anchovy interactions in the Eastern Mediterranean 

 have been found to involve time lags of 2 to 3 years 

 (Pucher-Petkovic et al. 1971). Yet, a 2-3 year cycle has 

 been identified in the variability of different biotic 



(zooplankton, phytoplankton, fish) and abiotic (air and 

 sea temperature, salinity and air pressure) components 

 of the Eastern Mediterranean/Black Sea ecosystem 

 (Polli 1955; Regner and Gacic 1974; S. Regner 1982, 

 1985; D. Regner 1985; Dement'Eva 1987; Petrova- 

 Karadjova and Apostolov 1988). The autoregressive 

 terms (X,_24, -Y,_2.5, Xt_2(,) of the model seem to be 

 consistent with these biological/oceanographical 

 observations. 



Citations 



Boudreault, F.R., J.N. DuPont, and C. Sylvain 



1977 Modeles lineaires de prediction des debarquement de 

 homard aux Iles-de-la-Madeleine (Golfe du Saint-Laurent). J. 

 Fish. Res. Board Can. 34:379-383 [in French. Engl, abstr.]. 

 Box, G.E.P.. and G.M. Jenkins 



1976 Time series analysis, forecasting and control. Holden- 

 Day, San Francisco, 575 p. 

 Dement'Eva, T.F. 



1987 A method for correlation of environmental factors and 

 year-class strength of fishes. J. Icthyol. 27:55-59. 

 Hilborn, R. 



1987 Living with uncertainty in resource management. N. 

 Am. J. Fish. Manage. 7:1-5. 

 Koslow, A.J., K.R. Thompson, and W. Silvert 



1987 Recruitment to Northwest Atlantic cod (Gadus morhua) 

 and haddock {Melanogramrrms aeglefiniis) stocks: Influence of 

 stock size and climate. Can. J. Fish. Aquat. Sci. 44:26-39. 



Makridakis, S., S. Wheelwright, and V. McGee 



1983 Forecasting: Methods and applications. John Wiley. NY. 

 926 p. 

 McLeod, A.I., and P.R.H. Sales 



1983 An algorithm for approximate likelihood calculation of 

 ARMA and seasonal ARM A models. Algorithm AS 191. Appl. 

 Stat. 1983:211-223. 

 Mendelssohn, R. 



1981 Using Box-Jenkins models to forecast fishery dynamics: 

 Identification, estimation and checking. Fish. Bull., U.S. 78: 

 887-896. 



National Statistical Service of Greece 



1968-88 Results of the sea fishery survey by motor vessels. 

 Section G, 20 issues (for the years 1964-86), Athens, Greece. 

 Petrova-Karadjova, V.J., and E.M. Apostolov 



1988 Influence of solar activity upon the diatoms of Black Sea 

 plankton. Rapp. Comni. Int. Mer Medit. 31, p. 224. 



Polli. S. 



1955 I cicli cHmatici di 5.6 e 8.0 anni e la loro realta' fisca. Riv. 

 Meteor. Aeronautica 2. 12 p. 

 Pucher-Petkovic, T., M. Zore-Armanda, and I. Kacic 



1971 Primary and secondary production of the Middle Adriatic 

 in relation to climatic factors. Thalassia Jugosl. 7:301-311. 

 Regner, D. 



1985 Seasonal and multiannual dynamics of copepods in the 

 middle Adriatic Sea. Acta Adriat. 26:11-99. 

 Regner, S. 



1982 Investigations of qualitative and quantitative composi- 

 tion of the larval fish stages in the plankton at the high sea 

 of the middle Adriatic. Stud. Mar. Fauna 11/12:45-60. 



1985 Ecology of planktonic stages of anchovy, Engraulis en- 

 crasicolus (Linnaus, 1758), in the central Adriatic. Acta 

 Adriat. 26:5-113. 



