Summer Fish Migrations in the River Neretva (South-Eastern Adriatic Coast, Croatia) as a Consequence of Salinization Ljetne migracije riba u rijeci Neretvi (jugoistočna obala

River Neretva Estuary, located in the middle South-Eastern Adriatic coast, is a very important traditional fi shery ground and biodiversity-rich ichthyologic area. Historically, the estuary situated in Croatia and Bosnia and Herzegovina was composed of wetlands, lakes, and lagoons, but in the 20th century, it was changed into a large agriculture area, settlements and port facilities. The major fl ow was channelled with large banks to prevent fl oods. All these activities enabled signifi cant intrusion of seawater 20 km into the inland area. This led to the marinization of major fl ow freshwater ecosystems. This article presents data on summer fi sh migrations along with the major fl ow of River Neretva, during the May-September period. The temperature and salinity showed that seawater started to enter inland major fl ow from middle May up to the end of September. In total, the 1,429 fi sh individuals were caught. The numerically dominant species were Atherina boyeri (23.58%), Sardina pilchardus (21.62%), Sparus aurata (11.20%), Chelon ramada (8.19%) and Solea solea (7.63%). Dominant species in the total mass were: Chelon ramada (43.45%), Solea solea (14.27%), Sparus aurata (10.14%), Chelon aurata (7.66%) and Anguilla anguilla (3.75%). The major result of this study points to the fact that major Neretva fl ow of 20 km in length from the river mouth is under seasonal summer impact of seawater, which leads to salinization and complete marinization of this ecosystem during the warm period of the year. The marinization of the large Neretva fl ow area creates new nursery and feeding grounds for marine estuarine opportunist fi sh species, such as small pelagics European pilchard and anchovy, and also enlarge these grounds for the marine estuarine dependent species, such as grey mullets, sea breams, and fl atfi shes.


INTRODUCTION / Uvod
Estuaries are one of the areas with the highest biological productivity and important role in the functioning of both marine and inland aquatic systems. They serve as nursery grounds for many marine species and provide a migratory route for fi sh species [11]. Due to their position within the drainage basin, these areas are characterized by a high level of human activity that can compromise their ecological integrity, because of increasing urban and industrial development [10,27]. Although a very important ichthyologic and biodiversity area, the Neretva Delta (Croatia, Bosna and Herzegovina) has not been comprehensively researched with its marine and freshwater ecosystems. Most of the research partially addressed either the recruitment of the fi sh juveniles [5,13,19], individual marine species ecology [3,21], the ecological signifi cance of some freshwater fi sh endemic species [19,31], or the evaluation of signifi cant habitats like Parila lagoon [35]. The recent biodiversity changes and invasions of fi sh and crabs reported ongoing invasions of blue crab, Callinectes sapidus in the brackish ecosystems [14,30] and records of several invasive marine fi sh [23] and freshwater fi sh [15, 22, 34,]. A holistic approach is challenging because of the constant changes in the habitat of the Neretva Delta, which also resulted in changes of the composition of water fauna due to changes in signifi cant ecological factors, particularly salinity. The occurrence of salinization of major Neretva fl ow, with an increase in salinity during summer, due to seawater intrusion into the upper regions, has not been studied at the fauna change level. The developed models of sea intrusion to major Neretva fl ow showed that seawater should protrude 25 km from the river mouth, occupying complete riverbed from the depth of 1 m to the bottom at 8-10 meters [29]. Thus, major Neretva fl ow during the summer season represents the largest transitional ecosystem area in the whole Neretva estuary. However, this transitional ecosystem is of temporary nature and depends on the quantity of water fl ow, which is unpredictable as it depends on dam management in the middle Neretva reach. The fl ow quantity of 500 m 3 /sec is the freshwater amount needed to throw out seawater from the major fl ow to the river mouth [29]. This fl ow quantity started with higher electricity production which usually begins in October and fi nishes in April-May, but without strict dates. This also means that major Neretva fl ow is the largest transitional ecosystems during the summer season when the fl ow is under the process of salinization, but this also means that all marine fl ora and fauna vanished during the winter season, either dying or migrating to the sea. The similar is happening with freshwater organisms during the summer marinization period. A similar impact of lower river fl ow was described for the Upper Tagus Estuary in Portugal [37].
There is very little information on reproductive and food fi sh migrations in the Neretva River. In the last 10 years, there has been an increase in the number of thinlip grey mullet, Chelon ramada in the Neretva delta and its migration to upper parts of the delta was recorded [20,38]. However, most of the research to date has been carried out in the area of the Neretva mouth [3,4,5,6] and concerned the dynamics of sand smelt populations and three grey mullet species. Recently, the migrations of Petromyzon marinus from the sea to the freshwater parts of the Neretva Estuary was described [24].
The main objective of this study was to investigate the marine fi sh migrations through the Neretva River from May to the end of September when the sea deeply penetrates the Neretva River upstream causing salinization and marinization of the freshwater areas and ecosystems.

Study area / Područje istraživanja
The study was conducted in the major fl ow of the Neretva River (43.027773, 17.553131), 1 km downstream of the town of Opuzen and 10 km from the river mouth ( Figure 1). The river area of 500 meters length was sampled with gill nets and ell traps. The study area is part of the typical fl ow of the Neretva from the town of Metković to the river mouth. The right side of the river is of soil and mud mixture, and depths of 5-6 meters begin right along the coast, with a steep slope. As a result, this area is poorly overgrown with aquatic plants, and immediately after the reed zone, it plunges towards the bottom. The left side of the river has been protected by a 300 m long rock embankment to prevent erosion, which signifi cantly changed the ecosystem's features. This part of the coast has a slight slope towards the bottom of the river, which in some parts has areas overgrown with aquatic plants and algae, and rich in juvenile fi sh and small freshwater fi sh. Once a week, measurements of the temperature and salinity of the River Neretva were carried out, over the entire depth profi le from the surface to the bottom (about 10 meters), by YSI 850 multimeter probe.

Sampling and fi sh analysis / Uzorkovanje i analiza ribe
Monitoring of the fi sh migrations was carried out with two smaller gillnets (mesh sizes 40 and 28 mm and 50 m long) and one gillnet (mesh size 72 mm; length 100 m). Smaller gillnets were connected into one and were laid across the Neretva River throughout the study. Large gillnet was laid down the river from smaller depths in U-shape with curved ends. The 30-meter bigscale sand smelt gillnets (mesh size 20 mm) were laid along the right bank of the Neretva River. Also, fi ve heart-shaped steel traps were used, which were laid along both banks of Neretva. The diameter of the mesh was 50 mm. They were laid to the depths of 2-8 meters and inspected once a week with the addition of bait (old bread). Five traditional eel traps were used, which were inspected once a week. They were laid along the left and right banks of the river to a depth of 2-5 meters.
After the catch, all the fi sh were kept in iced water. Within three hours the following measurements were taken: total length with an ichthyometer (mm) and total mass with a precision scale (0.01 g). Weekly and monthly qualitative and quantitative analysis of the structure of the fi sh community and abundance and quantity of individuals and species were made. Total and monthly distribution of all dominant fi sh species (those that participated with >1% in the total sample) was calculated. The length-weight relationship was calculated using equation W=aLb; where W represents wet weight (g), a regression intercept, L total length (cm) and b regression slope [18]. Fultons condition factor was calculated following Froese (2006): K=100*W/L^3.
Age was determined from the number of scale annuli from 30 individuals. Scales from individuals of diff erent length classes were removed, cleaned, dried, mounted between microscope slides and photographed under a stereomicroscope with a camera. Each scale was read by three diff erent readers, taking it into the account that two reads need to be verifi ed as same to determine the age. The annuli were interpreted according to Bagenal and Tesch (1978), the edge of the scale was treated as the last annuli and the area between the previous annuli and edge of the scale was treated as the annual growth increment.
Statistical data analysis and charting were done using the statistical programs Statsoft Statistica 10.0 and Microsoft Excel. The data required to establish the age key was taken from www. fi shbase.org for the Adriatic Sea or from own age data. Figure 2 shows the temperature profile of the Neretva River during the monitoring period from May to the end of September 2012. It can be seen that the temperature is uniform throughout the profile only at the beginning of May and they indicate the "fresh" character of the Neretva waters. As early as mid-May, temperatures higher than the surface layers appear at five-meter depth, which indicates the beginning of the penetration of the sea in the bottom layers. Already in the middle of May, this temperature threshold is established at about three meters' depth, pointing to the clear character of the water profile that remains throughout the summer. During the summer months, the temperature of the fresh-brackish layer to a depth of about three meters ranges from 18 to a maximum of 21 °C during the afternoon due to warming from the sun. During the night the temperature drops for 1-2 °C. The temperature of the sea layer, which stabilizes during the summer at 2.8-3 meters' depth, ranges from 23 to 25 °C and has the temperature characteristics of the coastal sea.

Water salinity / Salinitet vode
The depth profi le of Neretva salinity during the monitoring period from May to the end of September is shown in Figure 3. The picture shows that the freshwater profi le of Neretva during monitoring was found only in early May when salinity through the whole column was 0.3 psu. Already at the end of May, a twolayer character of the water column was formed in the Neretva riverbed. Thus, at the end of May, the boundary of the freshbrackish layer fl uctuates at a depth of about 5 meters. Early at June, this limit moves to the depth of 3 m, at which salinity greater than 35 psu was recorded, which is a typical salinity of the Adriatic coastal water. More precise salinity measurements were performed only for depths of up to half a meter and at the boundary of two layers of water. Measurements have shown that during the summer months, the layer of typical freshwater is only a few centimetres deep. During midday, this layer is found to be only 2-3 cm. Already at 5 cm salinity increases to 1.5-2 psu, so it enters the category of water which is not recommended in intensive agriculture. The boundary layer between the brackish (salinity water up to 30 psu) and seawater (all salinity above 35 psu) is formed in Neretva major fl ow from early June at a depth of only about 2.8 meters. And below conditions of the marine ecosystem prevail (hence not the estuarine transitional brackish ecosystem). This part can be further stratifi ed into sections where salinity is from 5 to 10 psu, therefore areas that can be inhabited by some freshwater fi sh tolerating elevated salinity (carp, perch, etc.), and salinity areas of 10-25 psu inhabited by typical estuarine species (mullet, sea breams, fl atfi sh, eel, sand smelt, etc.).

Big-scale sand smelt, Atherina boyeri / Gavun oliga
The total of 332 individuals (1.6 kg in total) were caught. The species was dominant in number and well represented in mass. It was caught in all months of the survey except May, and most in August (35.2%) ( Figure 8A). Individuals were collected with a total length of 6.4 cm to 12.0 cm (9.2 ± 0.63 cm) and weighing 1.65 g to 7.34 g (4.97 ± 0.75 g). Most individuals belonged to the 9 cm (64.8%) length class ( Figure 8B). The exponent b from the equation of length-mass ratio has a value of 1.580, indicating a negative allometric growth ( Figure 8C). The age structure indicates that the majority of the population found in the study area is composed of 3-year-old big-scale sand smelt ( Figure 8D).

European pilchard, Sardina pilchardus / Srdela
In the study area, 306 individuals (1.2 kg in total) were caught. The European pilchard was dominant in number and well represented in mass. It was caught in all months of the survey, most notably in August (35.9%) ( Figure 9A). Individuals were caught with a total length of 5.5 cm to 12.2 cm (8.4 ± 0.88 cm) and weighing 0.91 g to 10.23 g (3.95 ± 1.31 g). Most individuals belonged to the 9 cm (41.50%) length class ( Figure 9B). The exponent b from the equation of length-mass ratio has a value of 3.147 (R2=0.815) and indicates positive allometric growth ( Figure 9C). The age structure indicates that the entire population found in the study area were composed of juvenile specimens, aged 0+ years.

Gilthead seabream, Sparus aurata / Komarča
Total of 158 individuals (6.7 kg in total) were caught during all months of the survey except May. Most were caught in September (57.0%) ( Figure 10A). Individuals were caught with a total length of 6.1 cm to 25.8 cm (14.33 ± 2.79 cm) and weighing from 1.91 to 263.82 g (42.16 ± 31.69 g). Most individuals belonged to the 13-17 cm (80.3%) length classes ( Figure 10B). The exponent b from the equation of the long-mass ratio has a value of 3.221 (R 2 =0.981) and indicates positive allometric growth ( Figure 10C). The age structure indicates that the majority of the aff ected population in the study area is composed of juvenile gilthead seabream aged 1 year (86.7%) ( Figure 10D).

Slika 10. Struktura komarče, ulov Sparus aurata: A) učestalost mjesečnog ulova; B) struktura duljine tijela; C) odnos duljine i težine i D) dobna struktura.
Thinlip grey mullet, Chelon ramada / Cipal balavac Thinlip grey mullet, with a catch of 117 individuals and 28.6 kg of mass, was one of the dominant species, both numerically and in biomass. It was caught in all months of the survey, most notably in July (30.8%) ( Figure 11A). Individuals were captured with a total length of 7.5 cm to 38.4 cm (32.09 ± 4.72 cm) and a mass of 3.59 g to 450, 32 g (244.04 ± 70.71 g). Most individuals belonged to the length class 31-33 cm (52.51%) ( Figure 11B). The exponent b from the equation of length-mass ratio has a value of 2.867 (R 2 =0.984) and indicates a negative allometric growth and increased growth in length ( Figure 11C). The age structure indicates that the majority of the thinlip grey mullet aged 3 and 4 years (91.3%) ( Figure 11D).

Common sole, Solea solea / List
With a catch of 109 individuals (9.4 kg in total), the common sole was one of the dominant species both in number and in biomass. It was caught in all months of the survey, most notably in August (42.2%) ( Figure 12A). Individuals were captured with a total length of 15.1 cm to 25.9 cm (21.01 ± 2.63 cm) and a mass of 3.59 g to 172.95 g (86.02 ± 34.89 g). Most individuals belonged to the 19-20 cm (31.20%) length class ( Figure 12B). The exponent b from the equation of length-mass ratio has a value of 3,209 (R 2 =0.958) and indicates positive allometric growth ( Figure 12C). The age structure indicates that the majority of the population in the study area was composed of the common sole of 1 and 2 years of age (99%) ( Figure 12D).

European anchovy, Engraulis encrasicolus / Inćun
Total of 57 individuals (total 0.3 kg) were caught and this species was relatively well numerically represented during the study. It was caught in all months of the survey except May, and most in July (56.1%) ( Figure 13A). Individuals were collected with a total length of 7.1 cm to 12.4 cm (9.9 ± 1.27 cm) and weighing from 1.48 g to 10.72 g (5.33 ± 2.00 g). Most individuals belonged to the 10 cm (43.9%) length class ( Figure 13B). The exponent b from the equation of length-mass ratio has a value of 3.004 (R 2 =0.806) and indicates positive allometric growth, that is, increased growth in mass ( Figure 13C). The age structure indicates that the entire population found in the study area is composed of juvenile European anchovies (0+ and 1+ years) ( Figure 13D).

Annular seabream, Diplodus annularis / Špar
Total of 40 individuals (1.6 kg in total) were caught during all months of the survey, most notably in August (40.0%) ( Figure  15A). Individuals were caught with a total length of 4.9 cm to Figure 13 The structure of European anchovy, Engraulis encrasicolus catch: A) frequency of monthly catch; B) body length structure; C) length-weight relationship and D) age structure Figure 14 The structure of slender goby, Gobius geniporus catch: A) frequency of monthly catch; B) body length structure and C) length-weight relationship Slika 14. Struktura glavoča bjelaša, ulov Gobius geniporus: A) učestalost mjesečnog ulova; B) struktura duljine tijela i C) odnos duljine i težine 19.1 cm (13.13 ± 2.48 cm) and weighing 1.8 g to 85.76 g (41.12 ± 20.86 g). Most individuals belonged to the 11-14 cm (62.25%) length classes ( Figure 15B). The exponent b from the equation of length-mass ratio has a value of 3.012 (R 2 =0.971) and indicates positive allometric growth, that is, increased growth in mass ( Figure 15C). The age structure indicates that the majority of the population found in the study area was at 3 years of age (37.5%) ( Figure 15D).

European eel, Anguilla Anguilla / Jegulja
In the study area, 28 individuals (2.5 kg in total) were caught and the eel was relatively well represented both numerically and in mass. It was caught in all months of the survey except May, and most in September (32.1%) ( Figure 16A). Individuals were captured with a total length of 9.5 cm to 64.2 cm (34.1 ± 10.22 cm) and weighing 1.78 g to 489.18 g (88.10 ± 86.12 g). Most individuals belonged to the 35 cm (46.4%) length class ( Figure 13B). The exponent b from the equation of length-tomass ratio has a value of 3.038 (R 2 =0.992) and indicates positive allometric growth, that is, increased mass growth ( Figure 16C). The age structure indicates that the majority of the population found in the study area was composed of juvenile eels 2 years of age ( Figure 16D).

Eastern mosquitofi sh, Gambusia holbrooki / Gambuzija
In the study area, eastern mosquitofi sh inhabited the area along both shores and is probably the most abundant fi sh species in the Neretva River. A total of 35 individuals (8.0 g total) were selected to determine the species accurately. Individuals were caught with a total length of 2.0 cm to 4.1 cm (2.8 ± 0.33 cm) and weight from 0.11 g to 0.86 g (0.23 ± 0.12 g). Most individuals belonged to the 3 cm (88.6%) length class. The exponent b from the equation of length-mass ratio has a value of 2.971 (R 2 =0.869) and indicates negative allometric growth. The majority of the population was 2 years old.

DISCUSSION / Rasprava
The major fl ow of the River Neretva was freshwater from the surface to the bottom at the beginning of May. The seawater started to intrude in the middle of May and stay until the end of September when this research was accomplished. According to hydrological models, the 500m 3 of fl ow is needed to kick out the seawater from the major riverbed and fl ows bellow 180 m 3 enable intrusion of the seawater 20 km inland, including an area in Bosnia and Herzegovina [39] while during the summer season the fl ows are bellow 80m 3 [29]. The measurements of salinity in this study showed that only a few centimetre of the river water column was below 0.3 psu, what is drinkable and irrigation water characteristics. The water column from 10 cm to 200 cm represents brackish water of diff erent increasing salinities, while water below 2,8 m to the bottom is typical seawater of 35 psu and higher. This means that major Neretva fl ow is not the only estuarine brackish ecosystem, but also a typical marine ecosystem with typical marine biota during the summer period. This is supported with catching of typical marine fi sh, like a brown ray, Raja miraletus) or octopus (Octopus vulgaris) 20-25 km from the river mouth (https://caportal.net/2019/11/07/ foto-raza-kroz-neretvu-stigla-do-visica/). As a result of human activities salinization is occurring all over the world and is expected to be worse due to climate changes and increased water demand [9]. Such salinized rivers have characteristics of degraded freshwater systems but also represents habitats for salt-tolerant faunas. Loss of biodiversity, changes in community composition and a decline in the species richness such as insects represents consequences that can occur in more salinized rivers [26]. On the opposite side, salinization creates new habitat for the marine species, serving as a large new nursery ground for the younger stages, as was documented by this study. The highest number of fi sh species and individuals were caught in August. These numbers were smallest during May, increasing during June and July, and started to decrease in September. The numerically dominant species in the investigated area were two small pelagic species: the big-scale sand smelt (23.58%) and the European pilchard (21.62%). While a big-scale sand smelt is a common inhabitant in the Neretva Estuary [3] and typical estuarine resident, the presence of the younger cohort of European pilchard point to the importance of the Neretva Estuary as a nursery ground for the small marine pelagic species. This is enhanced with a documented presence of two other small pelagic, European anchovy and sprat, in the study area. Similar fi ndings were published for other European estuaries, i.e. Gironde Estuary (France), where densities of all the studied species (i.e. European anchovy, sprat or European seabass) significantly increased in recent time, accompanied by changes in water temperature and seawater intrusion [33]. The eff ects of global changes in this estuarine system favour its nursery function for marine juveniles and some evidence of eff ects at a wider scale (community and ecosystem levels) may also alter the structure and functioning of this system [16]. The intrusion of the seawater inside the River Gironde, modify the potential habitat area for marine fi sh and positively aff ect the abundance of juvenile species using the system as a nursery ground [33], as described for the Neretva Estuary. This classifi ed all three small pelagic species as marine estuarineopportunists, and migration into and out estuaries is often seasonal [25]. The next group of highly abundant species belong to the marine estuarine-dependent guild; the species which rely on the protected waters of estuaries for providing a suitable nursery habitat [7]. These species in Neretva Estuary are gilthead seabream (11.20%), the thinlip grey mullet (8.19%) and the common sole (7.63%). Among the freshwater fi sh species, most abundant was Gambusia holbrooki (4.55%). It is documented that this species tolerates estuarine salinities [1,36] and this is a reason while is considered as highly invasive in Neretva Estuary [22].
The most interesting fi sh fi nding is a high number of small pelagic species like European pilchard and anchovy in the major fl ow. Subsequently, these small fi sh attracted large pelagic species, like bluefi sh and pompano, which were caught by sport fi shermen [22] in the area of Neretva mouth. These changes should have impacts on predator-prey interactions and competition. Recent studies have shown that Argyrosomus regius could colonize the entire estuarine area, which can cause a high niche overlap. His colonization can increase of predation pressure and cause food competition with other species (e.g. European sea bass) which at the end can aff ect the functioning of the whole estuary, including the nursery function [12]. However, during our survey, these large marine predators were not sampled in a higher number, but interviews with sport fi shermen indicated their presence in major Neretva fl ow.
The sampled big-scale sand smelt in the major Neretva fl ow was 3 years old. This is confi rmation of previous study hypothesis on sand smelt in the coastal waters (mouth of River Mala Neretva), where most sand smelts were in their fi rst and second year, while the number of older individuals was very low due to their potential migration upstream [3]. Recent fi ndings confi rm this hypothesis for the River Neretva waters. The bigscale sand smelt is typical estuarine resident species, as reported for Porto-Lagos Lagoon (Greece) [28] or Venice lagoon [17].
The reasons for salinity increase in the last 30 years in the Neretva Estuary are: exploitation of sand at the mouth of the Neretva River and reducing the summer fl ow due to the operation of hydropower plants. Although it is clear that these two activities coincide with an increase in salinity, due to the lack of more serious scientifi c and professional research, it is diffi cult to assess their impact on the salinity process of the Neretva waters [39]. Rough estimations of the major Neretva fl ow property for the 20 km river length (app. 100m width, app. average 10 m depth) during extended summer season show that 80% of water column (from 3 to 10 m depth) represent typical marine ecosystem, while 19% represent estuarine ecosystem (from 0.3 m to 2.5-3m depth) and only 1% remains as a freshwater ecosystem (surface to 0.3 m). This is documented with a dominance of typical small marine pelagic fi sh and several other typical marine fi sh species followed with typical estuarine resident species like sand smelt and estuarine dependent species, like younger gilthead sea bream and several grey mullet species. The freshwater ichthyofauna is tightened in the few several centimetres of the water column and dominated with mosquitofi sh.
The main result of this study points out that major Neretva fl ow of 25 km in length from the mouth is under strong seasonal summer impact of seawater, thus leading to its salinization and complete marinization of this ecosystem during the warm period (May-October) of the year. This is immediately followed by an abrupt change during higher river water fl ow in the October-May period, which forced marine fi sh to migrate back to sea, while sessile marine and brackish organisms vanished from the major fl ow. However, this event creates new transitional ecosystem and additional nursery ground, which favours migration of smaller marine estuarine opportunist pelagic species and estuarine dependent marine species juveniles and adults, contributing to potential better recruitment of marine fi sh species in the Neretva Estuary.