Introduction

The Atlantic salmon, Salmo salar L. 1758, is a native European anadromous fish species. It used to be one of the most important fish species in recreational and commercial fisheries in Europe (Frič 1893). Salmon used to be abundant across Northern, Western, and Central Europe. During the 20^th century, salmon populations have declined in the whole Europe. The population in the Elbe River basin perished completely. The main reason was river fragmentation but other factors such as presence of diseases and parasites, predation, climate change, water pollution, overfishing, and losses of spawning habitats were also important (Frič 1893, Parrish et al. 1998, Jonsson & Jonsson 2004, Wolter 2015). Recently, salmon reintroduction has become one of the main goals in environmental protection and fisheries management in Europe (see European Habitats Directive). Salmon reintroduction has already been somewhat successful in several European countries where new salmon populations have been established (Wolter 2015). Czech populations perished about 60 years ago and salmon is now listed as critically endangered species. In year 1998, a new salmon reintroduction programme named “Salmon 2000” was founded (Kortan et al. 2010). The goal of this programme is to establish a thriving salmon population in the upper Elbe River basin (Benda & Šmíd 2002, Wolter 2015).

The great cormorant, Phalacrocorax carbo L. 1758, is one of the most important piscivorous predators in European freshwater ecosystems (Keller 1995, Suter 1997, Čech et al. 2008). Cormorants are opportunistic predators and are able to quickly adapt to new sources of prey (Keller 1995, Suter 1997, Leopold et al. 1998, Emmrich & Duttmann 2011). It has been stated that bird predation can be a significant source of salmon mortality (Jepsen et al. 1998, Mather 1998, Koed et al. 2002, 2006, Ibbotson et al. 2011). Previous research suggested that stocked fish are especially vulnerable to bird predation (Jonsson et al. 1991, Maynard et al. 1994, Christensen 1996, Eklov & Greenberg 1998, Dieperink et al. 2001). Stocked salmon could therefore serve as easy prey for cormorants (Jackson & Brown 2011, Salvanes 2017). The effect of cormorant predation on newly established salmon population in the area of the upper River Elbe has not been studied yet. It is important to assess any obstacles that could prevent the reintroduction programme from being successful.

This study had three aims: firstly, to assess lengths and density of salmon parr in nursery streams; secondly, to assess numbers and lengths of adult salmon in nursery streams; thirdly, to discover the effect of cormorant predation on salmon parr in nursery streams. We expected that stocked salmon juveniles survive and grow in nursery streams. We also expected that a significant number of adult salmon would be observed in nursery streams. Lastly, we expected to find remains of a few salmon in cormorant pellets.

Material and Methods

Study area

Cormorant pellets were collected during winters 2014/2015 and 2015/2016 from cormorant roosting places at the upper River Elbe (Velké Březno, North Bohemia, Czech Republic, 100 km north of Prague, 50°40′34.2″ N, 14°07′28.5″ E) (Fig. 1). Cormorants roosted in this area from October to April. About 100 cormorants roosted in the area in October. The numbers increased to approximately 500 birds in November and remained constant till February. Then the numbers dropped to approximately 100 birds in March and April. All birds were gone by May (Agency of Nature and Landscape Protection, unpublished data).

Salmon stocking was conducted on three nursery streams: the River Kamenice (angling ground no. 443 015,50°50′15.1″N, 14°21′16.9″E), the Ještědský stream (angling ground no. 443 501, 50°42′30.5″ N, 14°47′58.0″ E), and the Liboc stream (angling ground no. 443 062, 50°17′02.8″ N, 13°15′47.6″ E). All three nursery streams are located in the Elbe River basin (Fig. 1).

Groups of 10–30 cormorants were observed hunting on the River Kamenice in both winters 2014/2015 and 2015/2016 (Czech Fishing Union, unpublished data). The River Kamenice enters the River Elbe 24 km downstream from the roosting colony (air distance). In contrast, no cormorants were observed on the Ještědský stream or on the Liboc stream (Czech Fishing Union, unpublished data). The Ještědský stream is situated 47 km from the colony where it enters the River Ploučnice (in Stráž pod Ralskem). The River Ploučnice then enters the River Elbe 13 km downstream from the colony (in Děčín). The Liboc stream is situated 58 km from the colony where it enters the River Ohře (in Žatec). The River Ohře then enters the River Elbe 16 km upstream from the colony (in Litoměřice). According to the work of Platteeuw & van Eerden (1995), Grémillet & Argentin (1998), and Carss & Ekins (2002) most of the River Kamenice, lower River Ploučnice, and lower River Ohře are well in the reach of the roosting colony of cormorants in Velké Březno (Fig. 1). The studied colony was the largest cormorant colony in the North Bohemia. No other permanent colonies were identified in the study area (Agency of Nature and Landscape Protection, unpublished data).

Fig. 1.

Map of the study area with location of the cormorant colony (full black circle), estimated reach of cormorants roosting in the study area (wide black circle), the River Kamenice, the Ještědský stream, and the Liboc stream (black rectangles).

Table 1.

The numbers of Atlantic salmon stocked into nursery streams (the River Kamenice, the Ještědský stream, and the Liboc stream) in the Czech Republic during years 2014 and 2015. Note: fry (n), the number of stocked salmon fry (standard length 20–30 mm); parr (n), the number of stocked salmon parr (standard length 80–100 mm). Numbers (n) are in thousands of fish.

Salmon stocking

Salmon stocking was conducted exclusively on three nursery streams — the River Kamenice, the Ještědský stream, and the Liboc stream (Czech Fishing Union, unpublished data). Salmon stocking is a part of a salmon reintroduction programme named “Salmon 2000”. The goal of this programme is to establish a thriving salmon population in the upper Elbe River basin. For this purpose, about 40000–80000 fish have been stocked annually since year 1998. About 400000 fish with total estimated weight of 145 kg were stocked in years 2014 and 2015 (Table 1). Salmon fry (standard length 20–30 mm) and salmon parr (standard length 80–100 mm) were stocked. Salmon spawn originated from fish in the River Götaälv and the River Ätran (western Sweden). Fish from those rivers are genetically related to the extinct salmon population in the upper Elbe River basin (Zahn et al. 2009). Higher survival rates were expected because the populations are genetically close (McCormick et al. 1998). Stocked fish were reared in a hatchery near Langburkersdorf (East Germany) and transported to the Czech Republic in polyethylene bags. Each bag had a volume of 80 1 and contained 20 1 of water and 60 1 of oxygen-enriched air. About 5000 fish were transported in one bag. Salmon fry were released into all three nursery streams during spring. Salmon parr were released into the River Kamenice during autumn. The stocking was conducted by fisheries experts from the Czech Fishing Union and the National Park Bohemian Switzerland. Following the methodology previously published by Crisp (1995) and McMenemy (1995), fish were released in widely dispersed small groups on spots where the flow was slow and the stream was shallow.

Table 2.

Results of electrofishing surveys conducted on nursery streams where Atlantic salmon was stocked in years 2014 and 2015. Note: n, number of fish individuals; %n, percentage share on fish community; SL mean min-max (mm), mean min-max standard length (mm); density, density of fish per 100 m².

Electrofishing surveys

All three nursery streams were surveyed by electrofishing. A 100 m section was surveyed each time. The nursery streams were surveyed in spring 2014 to assess fish survival and then in autumns 2014 and 2015 to assess fish abundances, densities, and sizes. On the River Kamenice, two sections were surveyed using a portable motorized EFG electrofishing device. On the Ještědský stream, three sections were surveyed using a battery-powered device type Lena 1. On the Liboc stream, one section was surveyed using a battery-powered electrofishing device type Lena 2. Captured fish were determined to species level, measured, and released.

Table 3.

The numbers of adult Atlantic salmon observed in the River Kamenice in years 2014 and 2015. Note: n, number of fish; TL (cm), total length (cm); N/A, data not available.

Results

Electrofishing surveys revealed that salmon parr were present in all three nursery streams. Salmon parr density was 3–81 fish per 100 m². Nursery streams were dominated by brown trout, Salmo trutta. Salmon was the second most abundant fish species in the nursery streams (Table 2). Other fish and fishlike species discovered in the nursery streams were the following: bullhead, Cottus gobio, stone loach, Barbatula barbatula, brook lamprey, Lampetra planeri, grayling, Thymallus thymallus, European chub, Squalius cephalus, European eel, Anguilla anguilla, and gudgeon, Gobio gobio.

Salmon lengths were not normally distributed (Shapiro-Wilk: n = 738, p < 0.01). Surveyed salmon were significantly larger than stocked salmon; this was true for the River Kamenice in autumn 2014 (Wilcoxon: n = 130109, W = 142, p < 0.01) and 2015 (Wilcoxon: n = 145075, W = 8804, p < 0.01), for the Ještědský stream in autumn 2014 (Wilcoxon: n = 60139, W = 0, p < 0.01) and 2015 (Wilcoxon: n = 20103, W = 0, p < 0.01), and for the Liboc stream in autumn 2014 (Wilcoxon: n = 40148, W = 0, p < 0.01) and 2015 (Wilcoxon: n = 20164, W = 0, p < 0.01). Stocked salmon were 20–30 mm and 80–100 mm long (standard length); recorded salmon were 50–162 mm long (standard length) (Table 2).

Only thirteen adult salmon were reported in the River Kamenice during years 2014 and 2015 (Table 3), while no such records were registered in any of both Ještědský and Liboc streams.

Altogether 912 cormorant pellets were collected during winters 2014/2015 and 2015/2016. Together 5482 diagnostic bones were found in the pellets. From those diagnostic bones, together 3915 fish were identified. The overall cormorant diet was composed of 24 fish species from six families. No salmon were identified in cormorant diet.

Discussion

We discovered that the salmon stocking programme produces a reasonable amount of smolts in individual streams/rivers involved in “Salmon 2000” programme but return rates of adult fish are very low. Ecological conditions in nursery streams were comparable to conditions in similar streams where functional salmon populations exist (Prevost et al. 1992, Jutila et al. 2006, Descroix et al. 2009, Johansen et al. 2010). Presence of other pollution-intolerant fish species (e.g. brown trout, bullhead) was a sign of good ecological conditions in the nursery streams (Weatherley et al. 1995, Geist et al. 2006, Horká et al. 2017). Salmon density was a bit lower than what is considered average but the lower density was somewhat helpful since juvenile salmon display territorial behaviour (Gibson 1966, 1993, McMenemy 1995, Rosengren et al. 2017). Historically, numbers of returned salmon adults were significantly higher in the upper River Elbe and in other rivers in Central and Northern Europe (Frič 1893, Aarestrup et al. 1999, Lajus et al. 2005, Breve et al. 2014, Wolter 2015). Recently, low numbers of salmon are most likely caused by relatively low number of juveniles stocked into a low number of nursery streams (only three streams/rivers in case of the whole upper River Elbe; North Atlantic Salmon Conservation Organisation 2017).

Previous research suggested that migrating smolts are subjected to heavy predation from piscivores (Mather 1998, McCormick et al. 1998, Breve et al. 2014). During smolt run, migrating smolts get killed, delayed, and disoriented by hydropower plants, dams, and weirs (Larinier 1998, McCormick et al. 1998, Aarestrup & Koed 2003, Larinier 2008, Thorstad et al. 2008, Breve et al. 2014). There are two large weirs situated on the River Elbe: the Geesthacht weir (Germany) and the Střekov weir (Czech Republic; causing a potential problem to only the Ohře River basin salmon population). Both weirs have functional fish passes that should allow small and large fish to pass through (Prchalová et al. 2011, Adam et al. 2012, Menzel & Schwevers 2012). Unfortunately, previous research suggested that fish passes can be ineffective (Larinier 1998, Chanseau et al. 1999). Furthermore, migrating salmon smolts suffer from high mortality in estuaries (McCormick et al. 1998, Koed et al. 2006). Anglers and poachers usually catch some adult salmon as well (North Atlantic Salmon Conservation Organisation 2017).

We discovered that the cause of low salmon return rates into the upper River Elbe was definitely not cormorant predation on nursery streams. Cormorants were absent in two out of three of these nursery streams and salmon remains were not identified in regurgitated pellets at nearby cormorant roosting colony. The absence of cormorants in the nursery streams greatly limited predatory impact of cormorants on the salmon population. It is possible that some cormorants caught a small amount of salmon but pellets of those specific birds were not found; Jepsen et al. (2010) discovered that a single cormorant can consume high amount of salmon individuals when the bird locates a salmon school.

Previous research suggested that overwintering cormorants display different diurnal behaviour than salmon parr. In colder temperatures, salmonid juveniles are usually active during night in order to avoid endothermic predators (Fraser et al. 1993, Heggenes et al. 1993). Inversely, cormorants are diurnal predators that mainly feed on diurnally active prey (McCormick et al. 1998).

Our results also suggest that migrating cormorants mostly miss the main smolt run in this area. Other authors claim that smolt run usually occurs from April to June (Blackwell et al. 1997, Rosengren et al. 2017). We observed that flocks of overwintering cormorants leave the upper River Elbe area in April (most birds prior the end of March; M. Čech, R. Lyach, pers. observ.).

We suggest that cormorants in our study area did not prey upon salmon because biomass of other fish in the environment was much higher than biomass of stocked salmon. Total biomass of stocked salmon was 145 kg while biomass of other fish in most streams and rivers in the area usually equals to 250–300 kg per hectare, in eutrophic the River Elbe even exceeds this boundary (Czech Fishing Union, unpublished data). Biomass of stocked salmon was therefore almost negligible when compared to biomass of other fish. Previous research suggested that frequency of salmon in cormorant diet is usually positively correlated to salmon abundance and availability in the environment (Warke & Day 1995, Blackwell et al. 1997). Many authors discovered that cormorants usually prey upon the most abundant and available shoaling fish species (Keller 1995, Suter 1997, Čech et al. 2008, Čech & Vejřík 2011, Emmrich & Duttmann 2011). In case of this study, the upper Elbe River basin is dominated by shoaling cyprinids (Prchalová et al. 2011, Horký et al. 2013, Valová et al. 2014). On the other hand, salmon are definitely an attractive prey for cormorants — salmon parr and smolts are usually 3–25 cm long (Ibbotson et al. 2011) and cormorants often prey upon fish of this size (Keller 1995, Suter 1997, Čech et al. 2008, Emmrich & Duttmann 2011).

Several previous studies showed similar results as our study (Harris et al. 2008, Bostrom et al. 2009). On the other hand, different studies reported heavy cormorant predation on salmon (Warke & Day 1995, Blackwell et al. 1997, Jepsen et al. 1998, Koed et al. 2006, Jepsen et al. 2010). Researchers suggested that heavy cormorant predation on stocked salmon is mainly caused by high vulnerability of stocked fish to bird predation (Jonsson et al. 1991, Maynard et al. 1994, Christensen 1996, Eklov & Greenberg 1998, Dieperink et al. 2001, Jackson & Brown 2011, Salvanes 2017). In those scenarios, stocked fish frequently served as easy prey.

Previous studies confirmed that diagnostic bones of salmonids can be retrieved from cormorant pellets (Suter 1995, Carss & Marquiss 1999, Čech & Vejřík 2011). Therefore, analysis of content of cormorant pellets can be used to estimate effects of cormorant predation on salmonid populations.

In conclusion, the salmon stocking programme is producing a reasonable amount of salmon smolts and the nursery streams are suitable for salmon populations. The main reason for poor salmon return rates is not cormorant predation on nursery streams but, most likely, a low salmon survival rate on the passage downstream. Therefore, we suggest that more studies should focus on monitoring of survival and return rates of salmon in the upper River Elbe to ensure that, in the future, the salmon reintroduction programme will be really successful.