Spatial patterns and temporal trends in the fishery landings of the Messolonghi-Etoliko lagoon system ( western Greek coast ) *

The Messolonghi-Etoliko lagoon is one of the largest lagoon system in the northern Mediterranean coast. This area contains six clearly distinct lagoons with different topographic and hydrological features. The fishery landings of the lagoons are based on the ontogenic and seasonal migrations of the species. The total annual fishery landings of the barrier fish traps are estimated as 195 mt and are mainly composed of 16 species belonging to 8 families. Eel (Anguilla anguilla), the four species of Mugilidae (Liza saliens, L. aurata, L. ramada and Mugil cephalus), the two species of Sparidae (Sparus aurata and Diplodus annularis) and one species of Mullidae (Mullus barbatus) represent more than 92% of the total annual landings. The composition of the fishery landings varies between lagoons. The cluster analysis showed three groups of lagoons. The first group comprised lagoons in which the landings were dominated by eel, the second group was dominated by Mugilidae species and S. aurata, and the third by L. ramada and M. cephalus. The diversity index of the fishery landings is fairly constant in time, except for the Etoliko lagoon, where repeated anoxic crises decreased the diversity index value to almost zero in 1992. All the landing series showed a dominant annual cycle. Two seasonal patterns of the fishery landings were observed and linked to the fish spawning behaviour and/or their reaction to environmental forcings. The first one concerns species caught from summer to early autumn and the second pattern concerns species trapped during their autumn to winter offshore migration.


INTRODUCTION
Around the Mediterranean, an area of at least 6500 km 2 of coastal lagoons (Pearce and Crivelli, 1994) is exploited as fishing grounds (Ananiades, 1984;Kapetsky, 1984;Ardizzone et al., 1988;Peja et al., 1996;Anonymous, 2001).Coastal lagoons and estuaries are key ecosystems and an intensive exploitation of the increased natural productivity of these ecosystems is carried out by large local fishing communities (Kapetsky, 1984).
Different types of fishing gears and methods are used, such as barrier traps, fyke nets, trammel nets and longlines (Kapetsky and Lasserre, 1984;Ardizzone et al., 1988;Peja et al., 1996;Anonymous, 2001).At a regional scale, the lagoon fisheries play an important role in both socio-economic and cultural aspects.Marine, fresh-water or brackish fish populations inhabit the majority of the Mediterranean lagoons, which may be resident or migrant.Also, most of these ecosystems play an important role as nurseries for numerous fish species, offering food and shelter to the larval and juvenile stages (Weinstein, 1979;Blader and Blader, 1980;Tzeng and Wang, 1986).Many papers have been published on the influence of gradients of some abiotic parameters (i.e.salinity, temperature, dissolved oxygen and type of substrate) in the spatial and temporal distribution of fish species (Harmelin-Vivien et al., 1995;Rogers and Millner, 1996;Wang and Tzeng, 1997;Jenkins and Sutherland, 1997;Jenkins and Wheatley, 1998;Guidetti, 2000;Ishitobi et al., 2000;Blanc et al., 2001).
One of the most important types of exploitation is the use of barrier traps catching the fish during their seasonal or ontogenic offshore migration.The lagoon fisheries are based on the passage of fish from these important nursery and feeding grounds.These fishes are trapped at the lagoon-sea interface by passive and fixed gears during their movement from the lagoon ecosystem to the sea.
The aim of this paper is to analyse the barrier trap fishery landings in the different lagoons composing the Messolonghi-Etoliko lagoon system (Western Greek coast) during the period 1988-1998.In this area there are six clearly distinct lagoons with specific topographic and hydrological characteristics and different fish species compositions.The temporal changes, both seasonal and long term, are also analysed in order to define the migration pattern of the different species and to compare the migration timing in the different lagoons.15000 ha extending between the Acheloos and Evinos rivers.The six distinct lagoons (Fig. 1) can be defined by their topographic and hydrological characteristics (Table 1) (Anonymous, 2001).The bottom of the western and northwestern area of this system is muddy in contrast with the southern and south-eastern part of the R2 and R3, which is formed by sand and mud and is almost totally covered by Cymodocea nodosa (Bogdanos and Diapoulis, 1984).
Figure 2 presents the main seasonal changes of the air temperature and rainfall during the period 1992-1994, affecting the dynamics of both the abiotic and biotic parameters of the lagoons.Because of the limited depth of the lagoons, the meteorological changes rapidly affect the abiotic parameters of the water masses.The mean monthly air temperature ranged from 5 to 30 o C, with seasonal fluctuations typical of the Mediterranean climate.Two periods of high rainfall were observed, one in spring and one from October to December.

Fishing exploitation of the lagoons
The Messolonghi-Etoliko lagoon system is the property of the Greek State.The exploitation is a common extensive culture based on the seasonal entrance of young fish in the lagoons and the autumn to winter offshore fish migration.Eight fishing co-operative enterprises leasing the lagoons and exploiting the barrier traps installed at the interface between the lagoon and the sea supply the main fish catches.The so-called independent fishermen who fish mainly with lights, spears and longlines use the non-leased area of the lagoon system.Gill nets and dip nets are also used in the lagoon throughout the year.The total annual fish catches have decreased from 1500-2000 mt in the 1960s to 1300-1500 mt in recent years (Kotsonias, 1984;Dimitriou et al., 1994) and are provided by 200 fishermen working at the barrier traps and 700 fishing in the lagoon.The barrier traps provide the main fish production.These are passive gears and remained unchanged (both SPATIAL-TEMPORAL TRENDS OF LAGOONS FISHERIES LANDINGS 503  their nature and the effective fishing period) during the period concerned by the study.The number of fishermen fishing in the lagoons outside the barrier traps (the independent fishermen) and their fishing practices were also stable.Thus, the total fishing effort providing the recorded landings was fairly stable in time.

MATERIAL AND METHODS
The available monthly fishery landing data from the barrier fish traps in the six regions were used.The available time series for R3, R4, R5 and R6 lagoons refer to 1988-1998, while those for R1 and R2 refer to 1995-98 and 1994-98 respectively.In practice, the fishing period starts in April and continues until March of the next year, so the annual landings considered in the study correspond to this period and to the annual landings of the year covering the April to December period.Data were provided by the fish-wharf of Messolonghi.
The species landings used in this study were computed by the sum of the landings of all the commercial categories defined by the different size ranges of the precise species.
A hierarchical cluster analysis (Ward method, Hair et al., 1998) based on the arcsine-square-root transformation of the landing composition data was applied to determine the similarity/dissimilarity between regions.The squared Euclidean distance was used as a dissimilarity measure between regions for cluster analysis.The data transformation is necessary before analysis to normalize their distribution (Snedecor and Cochran, 1980).In order to observe annual changes in the composition of the fish production, the Shannon-Wiener diversity index where p is the relative importance, fraction, of biomass of the species I, was computed for the 6 lagoons and for the period 1988-1998.The analysis of variance (ANOVA) was applied to test for significant differences in mean H' between regions.Furthermore, Fisher's LSD test was applied to determine which lagoons differ from each other.
The average monthly species composition of landings from the different lagoons was used in a principal component analysis (PCA) in order to investigate the spatial and temporal patterns formed by the seasonal variability of the landing composition in the different lagoons.In particular, PCA was applied on the (species) X (arcsine-square-root of average monthly species landings proportion, period 1995-1998) data table.Only species representing a mean proportion higher than 5% of the total annual landings of each lagoon were retained in this PCA.
Fast Fourier Transform (FFT) analysis, based on log-transformed and detrended monthly landings, was used to identify the main frequencies (ω i ) describing the seasonal dynamics.A harmonic regression model (HREG) was adjusted to the time series of the dominant species landings: where X t is the logarithm of the species landings at time t, f(t) =at+bt 2 describe the temporal trends, c, a, b, b 1i , b 2i are coefficients estimated by least squares regression techniques, and ω i is the frequency of a cyclic component (Diggle, 1990;Swokowski et al., 1994).The frequencies, as well as time t and t 2 , used in the final model were selected through stepwise variable selection.Thus, only the significant components for both trends and cyclic fluctuations were retained in the final model.All models were developed using the SPSS version 8 statistical package.From the solution of the first derivative of HREG models, the time (month) of maximum landings for each species was estimated (Swokowski et al., 1994).In the case of presence of several peaks, their relative importance was expressed as a proportion of the landings in the month containing each peak to the total landings of all the months containing a defined peak in the same piscatorial year.All these calculations were performed for the year 1993 (t=61-72 months).

RESULTS
The recorded annual fishery landings of the six regions are presented in Table 2. Lagoon R3 had mean annual landings 3 to 6 times greater than the rest of the lagoon system.The year-to-year fluctuations of the total regional landings were limited except for those of 1988 and 1989.
The mean annual landings of the last four years was about 195 mt and composed of fourteen species belonging to six families (Mugilidae, Sparidae, Moronidae, Mullidae, Anguillidae, Goibiidae and Soleidae) and one species of Cephalopoda (Table 3).The landings of eel (Anguilla anguilla), four species of Mugilidae (Liza saliens, Liza aurata, Mugil cephalus and Liza ramada), the two species of Sparidae (Sparus aurata and Diplodus annularis) and one species of Mullidae (Mullus barbatus) represented about 92.1% of the total annual landings (Table 3).
Three groups of regions based on their species composition were identified by cluster analysis (Fig. 3).The regional differences in the species composition are presented in Table 3.There is a marked difference in the composition of fish landings of the six regions.The eel dominated the annual landings of group A (Lagoons R4, R5 and R6), while the family of Mugilidae and S. aurata dominated the landings of group B (Lagoons R2, R3).In group C (Lagoon R1) the largest part of the annual landings consisted of eels and two Mugilidae species (M.cephalus and L. ramada).In 1995-1998 Lagoon R3 had the greatest species diversity, with a high presence of species like Lithognathus mormyrus, Diplo-dus puntazzo, Sarpa salpa and Sepia officinalis, while Lagoon R6 produced almost exclusively eels.
Significant differences between lagoons of the Shannon-Wiener Diversity index were observed (F=24.3,df=5,47, P<0.05).Fisher's LSD test showed that H' 3 >H' 1 =H' 2 =H' 4 =H' 5 >H' 6 , where H' i the mean value of H' of the lagoon Ri.Fairly limited temporal fluctuations characterise the composition of the landings in lagoons R1 to R5, with lagoon R3 showing the highest diversity.The lowest diversity was recorded in lagoon R6, where the index decreased in 1991 and dropped to almost 0 in 1992 (Fig. 4).
The first and second factors of PC analysis, explaining 67.3% of the total variance (43.5 and 24.8% respectively) show that the monthly species compositions of the six lagoon landings were fairly similar.Furthermore, two major groups of monthly landing profiles were identified.The first (Group I) was defined by the seasonal fluctuations of the landings of L. saliens, M. cephalus, D. annularis, L.
SPATIAL-TEMPORAL TRENDS OF LAGOONS FISHERIES LANDINGS 505 mormyrus, D. puntazzo and S. officinalis.The second (Group II) was composed of two distinct subgroups, the first related to the temporal pattern of the landings of S. aurata, L. aurata, Chelon labrosus, L. ramada, Dicentrarchus labrax, M. barbatus, S. salpa and Solea vulgaris (Sub-Group A), and the second was composed of the landings of A. anguilla and Gobius sp.(Sub-Group B) (Fig. 5).
The frequencies in the variability of the logtransformed and detrended monthly landings of the 8 dominant species during the 1988-1998 period identified through FFT analysis are shown in Table 4. Seven frequencies were identified in the series of monthly landings (cycles with period 0.2, 0.25, 0.33, 0.5, 1, 2.75 and 3.7 years: Table 4).All catch series displayed a dominant annual cycle, whereas cycles with a period of less than one year represent harmonics generated by non-sinusoidal periodic variability.The arithmetic coefficients of the HREG models fitted to the monthly landings using the estimated frequencies (as well as time t and t 2 ) as independent variables and stepwise variable selection are shown in Table 5.The r 2 value of HREG models ranged from 0.64 for A. anguilla landings to 0.93 for D. annularis landings.In this table only the significant coefficient for both trends and cyclic fluctuations were presented after the stepwise component selection.In the same table the values of the coefficients of t and t 2 are significantly different from 0 (P<0.05).In the retained models for the different species only one of the t or t 2 components appears in the model and this indicates that the temporal trend, increasing or decreasing, is monotonous during the study period.From these values it is clear that the landings of L. saliens, A. anguilla and S. aurata are marked by a decreasing temporal trend, and an increasing trend appeared in the landings of D. annularis and M. barbatus.
In Figure 6 the recorded monthly landings of the four dominant species and the values estimated by the harmonic regression model (HREG) are presented.It is clear that the eel landings show an important decreasing trend.The landings of about 80 mt recorded in 1988-89 dropped to about 30-40 mt in the last few years of the period.
The solutions of the first derivative equations of the HREG models showed that the maximum quantities of landings occurred in two periods: (a) summer-autumn (August to October) maximum of landings defined for L. saliens, M. cephalus and D. annularis; and (b) the autumn to winter transition (November-December) for L. aurata, A. anguilla, S. aurata, M. barbatus and L. ramada (Table 5).However, lower landings of all species were recorded during the rest of the fishing period.These results, in combination with the results of PCA (Fig. 5), sug-SPATIAL-TEMPORAL TRENDS OF LAGOONS FISHERIES LANDINGS 507 gest that the majority of the annual landings in the Messolonghi-Etoliko lagoon system occur in two periods: (a) summer-autumn and (b) the autumn to winter transition.

DISCUSSION
The Messolonghi-Etoliko lagoon system is a productive coastal ecosystem exploited by an important fishing community using fairly traditional fishing methods.The fisherman density in the area is about 4.6 fishermen per km 2 , which is equal to the mean density of lagoon fisheries all over the world (Kapetsky, 1984).
The fish species composition of shallow coastal ecosystems depends greatly on the species-specific abiotic preferences and on their resistance to the important and frequent fluctuations in the abiotic parameters (Harmelin-Vivien et al., 1995;Rogers and Millner, 1996;Wang and Tzeng, 1997;Jenkins and Sutherland, 1997;Jenkins and Wheatley, 1998;Guidetti, 2000;Ishitobi et al., 2000;Blanc et al., 2001).The composition of the fishery landings also depends on the selectivity of the fishing gears.The fish barrier traps used traditionally in the lagoons are passive gears and their efficiency depends only on the size.All fishes greater than a given size are caught: for example in the traps of the area all the individuals of L. saliens larger than 10 cm total length are trapped during their offshore migrations (Katselis, 1996).Thus, the composition of the fish landings reflects the species community migrating towards the sea and can therefore be used to analyse the migration patterns, which are governed by both ontogenic changes and environmental forcings.It should be also noted that the total fishing effort providing the recorded landings was fairly stable in time because both the barrier traps and the number of fishermen showed little change during the study period.The general decreasing trend in landings in the last years can also be linked to the decrease of about 50% in the eel landings since the mid-1980s.This decrease was synchronous in coastal ecosystems of both the Ionian and the Aegean Sea and coincided with the decrease in the entire European eel fishery production (Zompola et al., 2001).Some of the dominant species (L.saliens, A. anguilla and S. aurata) showed a decreasing trend in the recorded landings, while some rare species (D. annularis and M. barbatus) showed an increase with time.Part of the observed fluctuations can be also linked to some technical aspects.For the first two years of the study some changes in the leasing regime of the lagoons may be responsible for a decreased accuracy of the declared landings (Dimitriou et al., 1994).In addition, any changes in the entrance probability of fishes in the lagoons may affect the fishery production.Thus, several technical modifications of the topography at the lagoon frontal openings of areas R1, R2, R4 and R5 during the early 1990s could be responsible for the decrease in fish landings recorded during this period (Klaoudatos and Konides, 1997).
Of the 56 species recorded in the Messolonghi-Etoliko lagoons, 6 are freshwater species, 29 are marine species entering the lagoons occasionally, 15 are euryhaline species spawning in the sea and migrating in and off the lagoons during their life cycle, five are species inhabiting the lagoons, and the last is the eel, which passed through these ecosystems during its inland and offshore migrations (Leonardos et al., 2000).In the present study 92.1% of the barrier traps fishery landings was composed of eight euryhaline species of Mugilidae (4 species), Sparidae (S. aurata and D. annularis), Mullidae (M.barbatus) and eel (Table 3).The remaining part was composed of species entering the lagoons occasionally (S. vulgaris, L. mormyrus, D. puntazzo and S. salpa) and one Cephalopoda species (S. officinalis).
The composition of the recorded fish landings of the barrier traps revealed some important differences between the lagoons.In most cases the species composition depends on the nature of the bottom, the lagoon-sea exchanges and the freshwater runoff.For several species (S. aurata, D. annularis, S. salpa, L. mormyrus and M. barbatus), representing a large part of the landings, their presence in Lagoons R3 and R2 is linked to the sandy bottoms covered by seagrass beds (Bauchot and Hureau, 1986;Harmelin-Vevien et al., 1995;Guidetti, 2000).In contrast, the increased landings of eel in Lagoons R4, R5 and R6 is linked to the muddy bottom (Deelder, 1984) of these regions (Bogdanos and Diapoulis, 1984).Furthermore, the increased landings of eel in the above lagoons is probably also linked to the increased freshwater inputs through natural and artificial channels and pumping systems draining the entire cultivated region east of the Acheloos river.The local fishermen also report the presence of eels in these channels.Thus, this system of channels is perhaps used by the eels as a feeding ground and/or pathway during their migrations from freshwater to the sea and vice-versa.These facts might explain the results of the cluster analysis (Fig. 3).
In the majority of the regions the inter-annual fluctuations of the species diversity index was low.The highest value of the index was observed in Lagoon R3 and was linked to the constant presence of several marine species entering the lagoon occasionally.This lagoon communicates with the sea through a long frontal area of 6 km enhancing the sea-lagoon exchanges.It is expected that the marked influence of the sea and the large surface of the lagoon lead to strong environmental gradients increasing the habitat heterogeneity which influences the diversity of the fish community.The species diversity in Lagoon R6 showed a particular pattern.A sharp decrease in the species diversity was observed in 1992, 1994 and 1997.Massive fish mortalities due to pronounced anoxic conditions were responsible for these fluctuations (Leonardos and Sinis, 1997).After these accidents, eel dominated the landings.Local fishermen reported that this species inhabits the inland ecosystems around the lagoon (natural and artificial channels and pumping systems draining the entire cultivated region), so it was not seriously affected by anoxic conditions.
The results of FFT analysis indicated that the monthly landings of species of Messolonghi-Etoliko lagoon system exhibited a strong seasonal cycle (Table 5).Indeed, the monthly landings of all species in barrier traps increased from a minimum in July to a maximum from August to December, depending on species, and declined thereafter.This marked seasonal cycle is most likely related to seasonal entrance of fish in the lagoons during the spring (inshore migrations) and their autumn to winter offshore migration.
Several species of Mugilidae, Sparidae and Moronidae just before and during their spawning season migrate massively offshore (Oren, 1981;Bauchot and Hureau, 1986;Smith, 1990).Massive displacements towards the deeper and more stable seawaters are also frequently observed during periods of unfavourable environmental conditions in the lagoons and shallow areas (Ishitobi et al., 2000).The observed offshore migration of L. saliens, L. aurata and M. cephalus seems to be linked to their spawning behaviour.The highest landings of L. saliens were recorded in July and August and the majority of the individuals (66%) were in advanced reproductive maturity stages (Katselis, 1996).Similar results were obtained for M. cephalus and L. aurata during August to November respectively (Vidalis et al., 1997;Hotos et al., 2000).In October-November, 96% of the S. aurata individuals caught in the barrier traps were 0 + age group (Dimitriou et al., 1994) and 81% of them were males reaching their reproductive maturity (Dimitriou and Katselis, 1999).
The offshore migration of some species may be linked to their spawning behaviour and/or their reaction to environmental forcings.Large landings of L. ramada were recorded in October-November but only 34% of the individuals were close to their reproductive maturity (Minos, 1996).The main offshore movement of D. annularis and M. barbatus (October and November respectively) seems to be linked to the rapid seasonal decrease in the water temperature since spawning take place in spring and eggs and larvae are recorded in sea waters in late spring and early summer (Karagitsou et al., 1997).
The eel landings seem to be concentrated in December.This is not due to a massive simultaneous migration but to the progressive stocking of trapped eels during the previous months and the global sale of the stocked production in December when the product price is higher (Messolonghi auction port, unpublished data).The main migration period coincides with the high rainfall period (Fig. 2).Deelder (1984) suggested that during the year two migrations can be observed, and both of them are linked to fishing activities.The first one concerns the eels migrating towards the sea during the period of heavy rainfall, usually a winter reproductive migration.In some other cases, massive offshore migrations can also be observed in early spring under the influence of important freshwater discharges (Westin, 1998).
In addition, the FFT analysis also indicated a 2-4 year periodicity in the landings of some species (Table 5).Cycles of the same period were also identified in the monthly fishery landings in the Greek seas (Stergiou et al., 1997) and in different biotic and abiotic variables in the Mediterranean, Black and Azov Seas (Stergiou, 1992).Similar cycles have also been observed in the physical environment and marine populations in other areas of the world and have generally been related to short-term oceanatmosphere interactions (Stergiou et al., 1997).
The landings of the barrier traps fisheries of the Messolonghi-Etoliko lagoon system are based on the migratory behaviour of several euryhaline species.These seasonal onshore-offshore movements are major events in the life cycle of the species, which take advantage of the increased productivity and the higher spring and summer temperatures of these particular ecosystems.Unfortunately, the barrier traps interrupt this life cycle and in some cases they do not allow the potential spawners to reach the spawning grounds and participate in the stock reproduction.In such a case, the lagoon exploitation seems completely independent from the regional coastal fisheries management.As the performance of the barrier traps increases, this separation will become more pronounced, with a negative impact on both coastal and lagoon fisheries.The results of the present study and the suggested models could be used for a more efficient management of the lagoon fishing activities, especially through the development of time limitations in the activity of the barrier traps.The main objective should be the integrated management of both the coastal and lagoon fishing activities.
FIG.1.-Map of the Messolonghi-Etoliko lagoon system.The six lagoons and the fish traps located at the lagoon openings communicating with the sea are presented.
FIG.2.-Mean monthly air temperature and total monthly rainfall for the period 1992 to 1994.The line that is limited by dots and arrows shows the entrance to the lagoons and the fishing periods respectively (data were collected from the meteorological station of the salt works of Messolonghi).
on the arcsine-square-root transformation of the average monthly proportion of the species landings by region(1995)(1996)(1997)(1998).The shadow areas indicate the eight dominant species.

TABLE 1 .
-Morphological and hydrological characters of the six distinct lagoons of the Messolonghi-Etoliko lagoon system.DPS, drainage pump station.

TABLE 2 .
-Annual fluctuations of the fishery landings (mt) of the barrier fish traps in the different Lagoons of the Messolonghi-Etoliko lagoons system during 1988-98.Ri, lagoons.
-Dendrogram of six regions in the Messolonghi-Etoliko lagoons obtained by cluster analysis (Ward's methods) of the fisheries landings species composition during the period1995-1998.

TABLE 4 .
-Frequencies identified through Fast Fourier Transform (FFT) analysis applied to the log-transformed monthly landings of the dominant species in the Messolonghi-Etoliko lagoon system, January 1988 -December 1998.

TABLE 5
. -Arithmetic coefficients of the harmonic multiple regression (HREG) models between the log-transformed monthly landings of the eight dominant species and the various frequencies identified by the Fast Fourier Transform (FFT) analysis of the Messolonghi-Etoliko lagoons.Study period: January 1988 -December 1998 (t=1 to 132 months).c is a constant, m 1 , m 2 are the months of maximum landings (B), Bm 1 , Bm 2 are landings (mt) observed in m 1 and m 2 respectively, and Bm 1 %, Bm 2 % are their relative importance (in percentage).