The conquest (and avoidance?) of the brackish environment by Ponto-Caspian amphipods: A case study of the German Baltic Sea

Although an invasion of the brackish water biotopes was to be expected with the appearance of several Ponto-Caspian amphipods in German freshwaters two decades ago (and earlier), only recently (two years ago) the conquest of the mesohaline Baltic Sea could be observed. This discrepancy is a fortiori of interest as previous experimental studies showed that species like Dikerogammarus villosus and Obesogammarus crassus were able to tolerate both mesohaline and also polyhaline conditions. Two decades of invasion history in rivers and lakes have led to drastic faunal changes. If similar or analogue shifts will happen in brackish environments and if estuaries like the Stettin lagoon function not only as “a gate to the Baltic Sea” but also as a “catalyser” or acclimatisation area for invasive species remains to be seen. Simultaneously the question came up, why other also potentially brackish water species failed to colonise mesohaline waters although they partially arrived in the investigation area several decades ago (e.g. Chelicorophium curvispinum and Echinogammarus ischnus).


Introduction
The invasion of Ponto-Caspian species into the inland waters of Central Europe since the mid of the 19 th century involved a dramatic faunal change. Due to the construction of artificial waterways (and their function as corridors between river drainage areas formerly disconnected) and the increasing ship traffic as vector for non-indigenous species natural biogeographic borders (e.g. watersheds, mountains, and continents) were negotiated. A prominent "pioneer" was the zebra mussel (Dreissena polymorpha (Pallas, 1771)) arriving in Central Europe around 200 years ago (Minchin et al. 2002). At the beginning of the 20 th century the amphipod Chelicorophium curvispinum (Sars, 1895) was the second successful invader from the same origin (Grabowski et al. 2007b). Several other species of different taxonomical orders (six amphipods among them) have followed in the last decades. Bij de Vaate et al. (2002) describe three different introduction routes for aquatic invertebrates of Ponto-Caspian origin: the northern corridor (from the River Volga via River Neva to the Baltic Sea), the central corridor (from the River Dnieper via River Weichsel and Oder to the Baltic Sea and further via the rivers Elbe and Rhine to the North Sea) and the southern corridor as youngest connection (Main-Danube channel since 1992) from the River Danube into the tributaries of the North and Baltic Seas. Additionally, the active transport of species into lakes and rivers by human activities, e.g. tourism or fishery aspects is of importance (Gasiūnas 1963;Rewicz et al. 2014). The opening of the Main-Danubechannel in 1992 has accelerated the introduction and dispersal of allochthonous species significantly. The most prominent and successful species using this way is the amphipod Dikerogammarus villosus (Sowinsky, 1894) (see Rewicz et al. 2014). This euryoecious species  (Grabowski et al. 2007a;Rewicz et al. 2014), often occurring in very high abundances, managed to colonise the tributaries of Central and North-West Europe connected to the Main-Danube channel within one decade. Another decade later the species had already arrived in Great Britain (Rewicz et al. 2014). The dispersal processes are highly dynamic and diverse. Even isolated waters like some alpine lakes have been populated using different vectors (e.g. fishery and tourism) (Rewicz et al. 2014). The catchment area of the Baltic Sea is occupied by several Ponto-Caspian amphipods (Zettler and Zettler 2017) with different introduction histories (e.g. Jażdżewski et al. 2005). Here the salinity plays an important role as regulation factor for dispersion regarding the osmoregulatory capacities and the changing interspecific competition. Whereas the autochthonous amphipod fauna in surface freshwaters of the catchment area of the Baltic Sea is poor in species number (approximately 5 species) the diversity in the brackish environment, e.g. between 5 and 10 psu, is quite high (approximately 40 species) (Zettler and Zettler 2017).
Independently of their arrival, the Ponto-Caspian amphipod species have first established populations in freshwaters for a long time (several decades) and behaved as genuine freshwater species. Hitherto they have not colonised areas behind the "salty curtain" and have been found only in fresh and oligohaline waters (maximum 3 psu). They have stopped their successful invasion at these borders. In our investigation area the Stettin Lagoon ( Figure 1) is a "collecting basin" for all three invasion corridors or routes (see above) and was a terminal stop for further dispersal of seven Ponto-Caspian amphipod species (Zettler 2008;  Since the first occurrence of D. villosus in Northern Germany in 1999 (Müller et al. 2001) this species was restricted to inland waterways (freshwater) and to the oligohaline (0-3 psu) Stettin Lagoon. Additional sampling in 2016 showed an abundant population of D. villosus and Obesogammarus crassus (Sars, 1894) in combination with several autochthonous marine species (e.g. Gammarus salinus Spooner, 1947;G. zaddachi Sexton, 1912;Leptocheirus pilosus Zaddach, 1844;Microdeutopus gryllotalpa Costa, 1853) in these mesohaline waters. It is the aim of the present study to document the current distribution of Ponto-Caspian amphipod species in the estuaries, lagoons and mesohaline waters of the Western Pomeranian area, the co-occurrence of allochthonous and autochthonous species and their salinity preference in the field (spatially and chronologically).

Material and methods
After the first record of D. villosus and O. crassus at the SE coast of Rugia (Figure 1, Stn. 36), a systematic mapping was conducted in the same area (Stns. 28-43) to localise their current distribution. In the years 2016 and 2017, we sampled at 65 complementary stations. The selection of the stations was based on habitat preferences from former studies. Due to similar salinities and habitats, we also sampled a few stations at the outer coast (Stns. 27-30, 55). Additionally, we used the data of five long-term monitoring stations within the investigation area (Stns. 2,10,14,64,70). For the chronological appearance in different salinities, we used all the data from our database comprising the investigation area (including rivers and lakes) from 1897 to 2017. The data are translated into a database given in Supplementary material Table S1 (non-native species) and Table S2 (native species). References within the database refer to the study of Zettler and Röhner (2004).
With few exceptions (e.g. diving) almost all sampling effort was focused on shallow waters (0 to 0.5 m). We used a hand net (sieve size 1 mm) for scratching hard substrates, macrophytes and soft sediments. We dived at one station (Stn. 37) to verify the depth distribution. Macrophytes were sampled in a depth range between 1 and 2.5 m with a net bag. At all stations the salinity was measured using a refractometer. The sampled animals were fixed in 70% ethanol. They were identified based on the amphipod key by Zettler and Zettler (2017).

Investigation area
The Western Pomeranian lagoons (Figure 1) have developed after the last glaciation by abrasion of island cores and subsequent sand drift. Depending on the tributary size, the freshwater input differs significantly.
The Darss-Zingst-Lagoon (DZL) is a chain of lagoons ("Bodden") with a strong salinity gradient (0-10 psu) from the mouth of the River Recknitz (freshwater) to the small connection with the entire Baltic Sea. The salinity can vary significantly depending on wind speed and direction. The mean water depth is very shallow (1 m) and only in some parts deeper than 5 m.
Rugia Lagoons (RL): The island of Rugia has several bays and lagoons with a small freshwater catchment area. The salinity ranges between 5 and 10 psu. Some of these lagoons are very sheltered and resemble lakes with expanded reeds (Phragmites australis (Cav.) Steud.). On average, the water is between 2 and 4 m deep (usually shallower).
Greifswald Lagoon (GL): The large water body has a wide connection with the Baltic Sea and moderate salinity range between 4 and 8 psu. The mean depth range lies between 2 and 4 m (> 10 m in maximum). Some coasts are very wind exposed and have a boulder and gravel dominated shoreline (see Figure 3).
Stettin Lagoon (SL): This estuary of the River Oder is very shallow (< 2 m) and only the water ways can be deeper. The salinity is very low (0-3 psu) and only at the entrance to the Greifswald lagoon the salinity can be a slightly higher. Large and wide reed belts and freshwater macrophyte weeds (Potamogeton spp.) indicate the "lake-conditions".

Results
The checklist of all observed amphipods and their frequency within the distinct waterbodies is shown in Figure 4. Twenty-five amphipod species were detected during the campaigns in 2016 and 2017 (three species were recorded during previous years additionally -see Table S1 and Table S2). Due to methodological reasons, talitrids are underrepresented but also present. The omnipotence of G. tigrinus in all subareas is obvious. Currently this species has the widest distribution and occurs frequently in all water bodies along the whole salinity gradient ( Figure 5). From the autochthonous perspective G. zaddachi is still the characteristic species of the lagoons.
The Stettin Lagoon was dominated by allochthonous species (Figure 4)   exposed area (see Figure 3). However, this species was also very common in the deeper vegetated zone (here Fucus vesiculosus L.). In the adjacent stations (e.g. 40, 43 and 44), no records were made even under very similar environmental conditions. Additionally, no Ponto-Caspian species were observed in the inner parts of the lagoon (Stns. 33-35, 39, 41-42). Gammarus duebeni, G. zaddachi and G. oceanicus were the dominant amphipods there. Obesogammarus crassus shows a very similar distributional pattern as Dikerogammarus villosus ( Figure 6). However, it is restricted to the near shore area in water depths shallower than 20 cm.   by multifactorial analysis (Devin and Beisel 2007). We are following here the definition of salinity tolerance given by Grabowski et al. (2007a): "… defined as maximum observed salinity level at which a particular species formed a self-sustaining populations in nature". The extensive research on the salinity tolerance of Ponto-Caspian amphipod species occurring in their place of origin (Black Sea, Sea of Azov, Caspian Sea) provides inconsistent and imprecise data (compare Markovski 1954;Paavola et al. 2005;Grabowski et al. 2007a;Rewicz et al. 2014). Very often we only found the salinity range of the particular sea without specific measurements from locations where the species was actually found or lived. Merely data of some species and their salinity tolerance in the estuaries of the rivers Dnieper and Bug (tributary of the Black Sea) by Markovski (1954) were more relevant. According to him, D. villosus and O. crassus live there in a salinity range between 0.5 and 5 psu, with an optimum for D. villosus and O. crassus at 0.5 psu and 1.5 to 3 psu, respectively. Paiva et al. (2018) reported observations of O. crassus from the southern Caspian Sea at salinities between 9 and 10.5 psu. Our own observation at the Bulgarian Black Sea coast indicate the occurrence of some Ponto-Caspian species (e.g. Pontogammarus robustoides (G.O. Sars, 1894) und Echinogammarus ischnus (Stebbing, 1899)) in the river mouths or in oligohaline coastal lakes, but not in the open mesohaline areas. Only Pontogammarus maeoticus (Sowinsky, 1894), not yet introduced to Central European waters, lived in the Black Sea at salinities up to 17 psu (own observations, sea also Kocataş et al. 2003 andUzunova 1999). Very recently a study by Paiva et al. (2018) was published, in which the salinity range of O. crassus in its indigenous area was given between 12 and 34 psu and of P. maeoticus between 0.5 and 30 psu. It is the first time that we found such a high salinity tolerance of O. crassus in connection with field observations. However, we entertain some doubt about the correctness of the data. We assume that for most locations, salinity was determined based on the mean annual salinity from The World Ocean Atlas database (Black Sea, Caspian Sea), which is too imprecise for coastal waters, and not on real observations of amphipods within a distinct salinity. We, at least, were not able to reproduce this large range from the given references.

Discussion
In the Baltic Sea, observations of the Ponto-Caspian species in mesohaline waters with real salinity measurements are still rare. Dobrzycka-Krahel and Rzemykowska (2010) and Dobrzycka-Krahel et al. (2015) found D. villosus and O. crassus in the Bay of Danzig at salinities between 5.8 and 6.8 psu. Šidagytė et al. (2017) indicate a salinity range between 0 and 8 psu for the investigation area of the Lithuanian waters in which D. villosus and O. crassus were recorded, but not for the records in particular.
The salinity range for the Ponto-Caspian (and one North-American) species in our study area is given in Figure 5. The North-American species (G. tigrinus) colonise both fresh and brackish (up to 16 psu outside the present study area) waters from the beginning of its arrival. It was observed in the Stettin Lagoon for the first time in 1992 (Rudolph 1994), in the Darß-Zingst Lagoon in 1994 (Zettler 1995), in the western Baltic Sea already in 1975 (Bulnheim 1976). Up to now, the seven Ponto-Caspian species have behaved like freshwater species (in rivers and lakes) and have found their distribution boundary in the oligohaline Stettin Lagoon. Although they have occurred in high numbers and frequencies for more than two decades there, no dispersal tendencies into the Baltic Sea have been observed. Only recently and for two species the conquest of the mesohaline areas has become apparent. Why did it take so long and why now? These questions cannot be answered. Long-term measurements of the Federal Agency for Environment, Nature Conservation and Geology (LUNG) show clearly that the salinity of the lagoons has not decreased between 2015, 2016 or 2017. Additionally the salinity is not highly variable among seasons with low salted periods allowing the surviving of stable population. In contrast, the annual range of salinity in exemplarily the Greifswald Lagoon is moderately low (< 0.8 standard deviation) in the last 10 years (not shown in figure). Both Gruszka and Wozniczka (2008) and Zettler (2008) stressed the rapid colonisation of the Stettin Lagoon by D. villosus in contrast to the outlets of the lagoon to the open Baltic Sea. The gap between the population around the SE end of Rugia and the next occurrence in the Stettin Lagoon is also interesting. Maybe other vectors (fishery, artificial construction activities) can explain this jump from the oligohaline to the mesohaline areas. Although highly speculative separate introductions of two genetically diversified populations are also possible (see Hupało et al. 2018).
Further open questions are 1) do we now observe the starting shot for the conquest of other mesohaline areas at our coasts and 2) will other Ponto-Caspian amphipod species follow?
Dobrzycka-Krahel and Rzemykowska (2010) for instance also found Pontogammarus robustoides and Dikerogammarus haemobaphes at the outer coast in salinities between 5.8 and 6.1 psu. Markovski (1954) assigned D. haemobaphes to salinities between 0.2 and 5 psu with an optimum of 0.5 psu, P. robustoides even as stenohaline species up to 1 psu in the Dnieper-Bug estuary.
Particularly due to the rapid colonisation by Ponto-Caspian amphipods (especially by D. villosus), significant faunal changes were observed (e.g. Bij de Vaate et al. 2002;Jażdżewski et al. 2004Jażdżewski et al. , 2005Rewicz et al. 2014;Meßner and Zettler 2016). Dick and Platvoet (2000) described the disappearance of G. duebeni and G. tigrinus caused by the introduction of D. villosus into the Markermeer und IJsellmeer (freshwater to oligohaline). It is most likely that similar drastic faunal changes will also occur in the mesohaline areas of the Baltic Sea. Our own long-term observations in the Stettin Lagoon demonstrated significant faunal changes involving a high variability (Figure 7). Whereas Gammarus tigrinus, Obesogammarus crassus and Pontogammarus robustoides were abundantly and frequently found since the arrival of Dikerogammarus villosus, the formerly very abundant non-indigenous species Chelicorophium curvispinum is almost disappeared. Due to changing environmental conditions (e.g. salinity, river run off, eutrophication, and temperature seasonality), the colonisation success and development vary as well. Nevertheless, it is obvious that the species composition has changed and the species number has distinctly increased during the last decades. While the non-indigenous species already prevailed at the beginning of the monitoring (3 of 4 species), we have observed an additional impulse of introduction (7 of 9 species are allochthonous) in the last 10 years. The dispersal of non-indigenous species may be affected also due to the presence of stronger or weaker competitors (sensu Kobak et al. 2016).

Conclusions
Finally, although Ponto-Caspian NIS are currently not established in higher salinity areas of the North and Baltic Seas (Casties et al. 2016;Paavola et al. 2005), we assume a noteworthy and rapid spread of D. villosus and O. crassus along the southern Baltic Sea because the populations in the investigation area have already reached a sufficient and stable size. However, a lot of other abiotic and biotic parameters affect the distribution and dispersal of species. It is high speculative to predict the areas of future occurrence, nevertheless, the present observations are evidentially signs for the conquest of the brackish environment. Additionally, other Ponto-Caspian species will probably also have the potential to colonise these mesohaline waters. Based on the experience from freshwater habitats, we expect significant faunal displacements in coastal waters of the Baltic Sea in the near future.