First confirmed records of Prussian carp, Carassius gibelio (Bloch, 1782) in open waters of North America

Carassius gibelio (Prussian carp) is considered native from central Europe to Siberia and invasive in Asia, Europe, and the Middle East. Todate, there have been no published occurrences of C. gibelio in open waters of North America. Although there are reports from the late 1800’s that C. gibelio was imported into the USA, some doubt exists regarding the identity of those fish and it is unknown whether any were released into the wild. Here we report the first confirmed records of C. gibelio in the wild in North America and estimate its current distribution by compiling records from private and public sources. Morphology and DNA sequences were used to verify identification. Initial specimens of C. gibelio were found in a shallow lake in 2006 in Alberta, Canada. Routine provincial government and private fish surveys from 2008–2014 further identified C. gibelio throughout natural streams and manmade drainage ditches. These surveys document that C. gibelio is now widely distributed throughout southern Alberta. Further spread is likely due to connectivity of the waterways and the lifehistory attributes of this fish.


Introduction
Carassius gibelio (Bloch, 1782), generally referred to as Prussian carp, is a prolific invader of freshwater ecosystems, particularly in Europe and the Middle East (Kottelat and Freyhof 2007).Initial introductions outside its native range of central Europe and/or Asia likely originated from accidental aquaculture escapees in the 1600s (Lever 1996; Kalous et al. 2012).Since then, C. gibelio has exhibited secondary spread by natural dispersal through river and ditch systems (Grabowska 2010), escape from aquaculture (Toth et al. 2005), and intentional introductions by anglers (Witkowski 1996).C. gibelio's successful spread and establishment is attributed to its unusual gynogentic reproductive strategy and its ability to thrive under diverse environmental conditions (Perdikaris 2012).The impacts of introduced C. gibelio on native fisheries and ecosystem function are substantial (Paulovits 1998;Slavík and Bartoš 2004;Gaygusuz et al. 2007;Perdikaris 2012;Tarkan et al. 2012) and could cause economic damage; therefore, early detection programs that delineate its distribution and enable rapid management responses may be desirable.
Wild C. gibelio populations have never been documented outside Asia, Europe, or the Middle East.However, there is a record of importation from Germany to New York state in 1876 (Ferguson 1876) and an isolated, unconfirmed observation of hybrids (Carassius auratus  C. gibelio) in the Hudson River, New York around this date (Redding 1884), although the hybrids were later considered Carassius carassius (Cole, 1905).These records indicate that C. gibelio have been imported into North America, yet it is unclear if individuals were ever released into the wild.Also, inconsistent use and confusion over the C. gibelio nomenclature make it difficult to confirm if the fish found in the Hudson River were indeed C. gibelio or a hybrid (Fuller et al. 1999).Here, we present the first confirmed records of C. gibelio in open waters of North America.Our initial detection occurred during an annual sampling of shallow lakes in 2006.DNA sequences were used to verify identification.We subsequently compiled provincial and private fish surveys in Southern Alberta, Canada, which indicates that C. gibelio's current distribution has rapidly expanded to two additional river basins.We also discuss likely pathways of initial introduction and patterns of spread between major river basins in Southern Alberta.

Study region
We assembled data from the Red Deer, Bow, Oldman, and South Saskatchewan River basins in southern Alberta, which originate at the continental divide of the Canadian Rocky Mountains and flow east to Hudson's Bay.This landscape is part of North America's Prairie Pothole Region and use is now predominately grazing pastures and row crops.Due to its arid environment, open irrigation canals were constructed in the early 1900's that criss-cross the landscape throughout southern Alberta.These irrigation canals are organized into 13 geographically separate irrigation districts, but are inter-connected through natural waterways that include major rivers and small prairie streams.

Field surveys: shallow lakes
We collected C. gibelio using overnight sets of baited minnow traps during annual (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)2012) surveys of shallow prairie lakes in the Bow and Red Deer River basins east of Strathmore, Alberta.These lakes have varying degrees of connectivity to each other and nearby irrigation canals and/or major rivers (Jackson 2003).Our initial identification was based on the morphological characteristics outlined in Kottelat and Freyhof (2007).Our sampling efforts were not intended to target this species; however, we found incidental catches while conducting other studies.

Literature and database survey
To collect additional records and determine whether our catch was the first record of C. gibelio in Alberta, we searched government databases, scientific literature, technical reports, and contacted the Royal Alberta Museum for physical specimens.Due to the connected waterways of the agricultural irrigation canals, we also contacted each irrigation district in Alberta.We collected geographic location, catch numbers, and size for each available record.We focused our search on the Bow, Red Deer, Oldman, and South Saskatchewan River basins due to their proximity to the first C. gibelio record.To find any historical records associated with C. gibelio's importation into the United States in 1876, we searched the online collections of the American Museum of Natural History, Smithsonian National Museum of Natural History, and the Harvard Museum of Comparative Zoology for existing physical records of C. gibelio.

Specimen identification
Potential misidentification of C. gibelio and species level taxonomic uncertainty with other Carassius spp. is common (Japoshvili et al. 2013), particularly with C. auratus (L.), which is found in the Bow, Red Deer, and Old Man River basins in Alberta (Nelson and Paetz 1992; Government of Alberta 2014) and with C. carassius, which to our knowledge has not been identified in Alberta.We followed the dichotomous key of Kottelat and Freyhof (2007); C. carassius and C. gibleio are easily defined from one another based on the shape of the free edge of the dorsal fin and the peritoneum color.The only non-overlapping morphological character between C. gibelio and C. auratus is their color, which can be difficult to judge (Table 1).Due to morphological similarities, the definition of species within the Carassius genus is not always certain, and is particularly true of C. gibelio and C. auratus where earlier research defined C. gibelio as the wild form of C. auratus (Kottelat and Freyhof 2007).Recent genetic studies have since shown that C. gibelio and C. auratus are distinct species (Rylkova et al 2010) and that C. gibelio may have two separate clades (Kalous et al 2012).This potential misidentification and taxonomic confusion may create an under or overestimation of C. gibelio's distribution.We therefore compared DNA from 25 suspected specimens with published sequences from the Barcode of Life project to support our morphological identification (Ratnasingham and Hebert 2007).We also compared major morphological characters identified by Kottelat and Freyhof (2007) in the same 25 fish from Alberta (Table 1).

DNA sequencing
To aid and support our morphological identification, we isolated DNA preserved in 95% ethanol from a specimen collected in 2012 in Long Lake, seven specimens collected from West lake in 2014 (both  Sequences were aligned with Codon Code Aligner (v3.7.1) and sequences were compared with the BOLD database to identify the most likely species (Ratnasingham and Hebert 2007).

Occurrence in shallow lakes
Our analysis indicates that records of occurrence began in 2006 and increased through 2012 (  of lakes and Bland Lake, and found three specimens in Long Lake and none in Bland Lake.
We also set baited fish traps in Barnett Lake in 2012, which has an ephemeral input from the irrigation canals 2 km northwest of West Lake, but did not catch C. gibelio.However, at ice break-up in the spring of 2013, we discovered a fish kill where hundreds of C. gibelio lined the shoreline of Barnett Lake.In 2014, seven additional individuals were caught in West Lake.

Occurrence in streams and irrigation canals
The Government of Alberta and private data for southern Alberta suggest that C. gibelio are present in natural streams and irrigation canals in the Bow, Red Deer, and South Saskatchewan River basins (Table 2).In 2008, a contracted fish survey of one of the irrigation districts discovered ~5,000 individuals (total length range: 3-20 cm), and estimated 20,000-30,000

Historical museum records
No historical records of C. gibelio were found in the online collection databases in the American Museum of Natural History, Smithsonian National Museum of Natural History, or the Harvard Museum of Comparative Zoology.There were also no records of C. gibelio in the Royal Alberta Museum.

Confirmation of identification
The specimens from our shallow lake surveys and those preserved by the Government of Alberta from 2009 matched positively with the description of C. gibelio in Kottelat and Freyhof (2007) with the exception that some of our specimens had 32 gill rakers compared to the minimum of 37 gill rakers described in Kottelat and Freyhof (2007) (Table 1).Otherwise, all other morphological features fell within the range of C. gibelio characters described in Kottelat and Freyhof (2007).This also means there was considerable overlap in morphological characters between C. auratus and C. carassius.Our specimens all had a concave dorsal fin, a silver body color, a black peritoneum, and a highly serrated last dorsal ray.The concave dorsal fin edge, black peritoneum, and a highly serrated last dorsal ray distinguished our specimens from C. carassius and the silver color distinguished it from C. auratus.
A 650 bp COI fragment successfully amplified in 25 individuals.The molecular identity of 24 specimens most closely matched C. gibelio (98.57-100%).The COI fragment from one individual in 2014 from West Lake most closely matched C. auratus (99.02%), however, the next highest match was C. gibelio (98.81%).All individuals were compared to reference sequences from Russia, Germany, France and Greece (GenBank accession codes: NC_014177 and GU170401).

Discussion
There were no confirmed published records of C.
in open waters in North America until our discovery in 2006.An intensive fish population survey conducted by the Government of Alberta in 2006, which included the area in which C. gibelio is currently found, found no C. gibelio (personal communication with Dr. Mike Sullivan).Subsequent records from multiple sources reveal that C. gibelio is now widespread with occurrences in three major river basins in southern Alberta.However, our estimated current distribution and timing of introduction may be conservative due to the absence of comprehensive and extensive surveys throughout Alberta before and after 2006.It is also possible that this species has been overlooked in other surveys due to misidentification or population densities below the level of detection.
To clarify and confirm our identification, we sequenced DNA from 25 individuals caught in three locations.Although the DNA analysis confirmed our morphological identification, specimens from only three locations were sequenced, which leaves the rest of the C. gibelio distribution in southern Alberta unverified.Morphological identification is difficult and we believe DNA analysis would help to confirm identification, especially given C. gibelio's overlapping distribution with C. auratus in southern Alberta.Overall, molecular and morphological approaches support C. gibelio identification, despite that one individual matched C. auratus.Given the overlapping distribution of C. auratus in this region and the close phylogenetic relationship between C. gibelio and C. auratus, perhaps this result is not surprising.Taxonomic uncertainty also adds to the need for additional molecular work.Kalous et al. (2012) suggest that C. gibelio is indeed a distinct Carassius species, but also reports that C. gibelio is not monophyletic, suggesting it may actually be two species.A neotype was designed to help aid identification and to clarify C. gibelio species identity.Therefore, we recommend further genetic sequencing to confirm the distribution of C. gibelio in Alberta and that the specimens need to be compared to the neotype identified by Kalous et al. (2012).
The irrigation canals may have inadvertently facilitated C. gibelio's current widespread distribution throughout southern Alberta.Beginning in the late 1800's, artificial waterways were constructed to supply water to farms, which increased the connectivity of waterways and linked many of the once isolated shallow lakes, watersheds, and even large drainage basins together for the first time (Figure 2).As seen in Europe (Galil et al. 2007) and in the Laurentian Great Lakes (Rothlisberger and Lodge 2013), artificial waterways that link basins can become "invasion highways" that facilitate and accelerate the spread of aquatic invasions.Although it is intuitive to understand how C. gibelio spreads once established, we do not know how they were initially introduced to Alberta.
In Poland, C. gibelio was introduced into new regions intentionally by anglers and unintentionally by aquaculture, followed by subsequent spread from areas of initial introduction (Grabowska 2010).It is doubtful that anglers introduced this species to southern Alberta because C. gibelio has low sport value compared to salmonids already present and anglers would have had to bring individuals from overseas.Aquaculture may be a more likely source of introduction.Globally, aquaculture has become one of the leading vectors of aquatic species introductions (Naylor et al. 2001).In Alberta, aquaculture grows a variety of fish species including the closely related koi carp (Cyprinus carpio L.) and C. aurutus (Alberta Agriculture and Rural Development 2006), the latter being classified as invasive to Alberta (McClay et al. 2004).It is possible that C. gibelio was introduced into North America unintentionally in shipments of these or other commercially viable fish.Indeed this is a common vector for new introductions (Slavík and Bartoš 2004).It is also possible that C. gibelio was purposefully or accidentally imported by the aquarium industry as bait fish or as contaminated stock and accidentally released.In the Laurentian Great Lakes, C. auratus and Cyprinus carpio are widely distributed and are the top occurring fish recorded (100 and 85%, respectively) in aquarium stores near Lakes Erie and Ontario (Rixon et al. 2005).Although relevant data from aquarium stores in Alberta are not available, C. gibelio admixtures to shipments of ornamental cyprinids cannot be excluded.Moreover, the aquarium trade has been proposed as the most possible pathway of introduction of the con-generic Carassius langsdorfii Temminck & Schlegel, 1846 in Central Europe from Japan (Kalous et al. 2007;2013).
The impacts of C. gibelio on native fish communities and aquatic ecosystems in Southern Alberta have not been estimated.However, the negative impacts described in Europe, Asia, and the Middle East suggest impacts in Alberta could be substantial.The establishment of C. gibelio in Europe has caused multiple fishery declines, particularly with native cyprinids (Economidis et al. 2000;Gaygusuz et al. 2007).C. gibelio have a gynogenetic reproductive strategy that results in reproductive competition that may reduce native cyprinid population size (Tarkan et al. 2012).In Alberta, the May -August spawning period of many native cyprinids overlaps with C. gibelio's, and increase the likelihood of reproductive competition and subsequent population decline (Nelson and Paetz 1992;Kottlat and Freyhof 2007).Direct competition for pelagic zooplankton and benthic invertebrates may also reduce native cyprinid populations (Lusk 2010).Additionally, by removing pelagic zooplankton, C. gibelio can weaken the top-down control zooplankton excerpt on phytoplankton, resulting in turbid water.Furthermore, consumption of lake benthos stirs up bottom sediments, which can contribute to increased turbidity.The irrigation canals, small prairie streams, and shallow lakes also typically lack piscivorous fish, leading to a lack of predation on C. gibelio.
First detected in a single lake in 2006, C. gibelio has since been documented as occurring in parts of three adjacent major drainage basins in the Canadian province of Alberta.Because sampling effort has been uneven, additional sampling may reveal an even broader distribution and greater abundance.The potential consequence of this uncontrolled spread is especially high since the Oldman and South Saskatchewan River basins border the Milk River, which is part of the Mississippi River basin.A recent risk assessment and climate match model conducted by the U.S. Fish and Wildlife Service concluded that C. gibelio has a moderate to high climate match with the U.S.A. states that border Alberta and that the United States is at high risk of invasion (U.S. Fish and Wildlife Service 2012).This suggests that southern Canadian provinces with similar climates are equally vulnerable to invasion.In the absence of management efforts to control their populations and spread, C. gibelio is likely to continue spreading into new basins, causing unknown consequences to native biota, ecosystems, and economies.

Figure 2 .
Figure 2. Map of Canada (A) illustrating the area of Alberta where C. gibelio have been found; (B) confirmed occurrences (black dots) of C. gibelio in the Bow (BR), Red Deer (RD), and S. Saskatchewan River basins (SS).The adjoining Milk (MR) and Oldman River (OM) basins are also labeled; and (C) enlarged view of the square in 'B' identifying the site of first occurrence (Triangle; West Lake, n = 3).Arrows indicate artificial connections between the Red Deer River basin and the Bow River basin.
individuals in a pool within an irrigation canal in the Red Deer River basin(Haag and White 2008).In 2009, the same contractor found C. gibelio at the same site as 2008 and confirmed an additional finding in another pool within the canal system.Again, C. gibelio were found in high abundances (approximately 500 individuals; total length range: 3-20 cm)(Haag et al. 2010).Also in 2009, the Government of Alberta received 17 individuals from an unnamed tributary to the Medicine River.In a survey of small prairie streams in 2010, the Government of Alberta (2014) discovered and recorded C. gibelio in Spruce Creek in the Red

Table 1 .
Kottelat and Freyhof (2007)eristics for 3 Carassius species as given byKottelat and Freyhof (2007)compared to characteristics of 25 specimens collected from the wild in Alberta, Canada.

Table 2
observation for 2006.Subsequent surveys in 2007 and 2008 revealed C. gibelio in all three lakes in the West Lake chain (West, East, and Long Lake).In 2007, we also discovered C. gibelio in nearby Bland Lake, within the Red Deer River basin.In 2012, we returned to the West Lake chain

Table 2 .
Records of C. gibelio in the Bow, Oldman, Red Deer, and South Saskatchewan River basins in Alberta, Canada.* Indicates locations where DNA sequencing was done on C. gibelio individuals.** Indicates that exact coordinates are not known.