How likely is Lepomis gibbosus to become invasive in Poland under conditions of climate warming?

Background. Despite increasing reports of non-native freshwater fish dispersal in Poland, a risk identification and risk assessment (RA) of their current or future impacts has not been undertaken. In this study, to advise policy and management decisions the Aquatic Species Invasiveness Screening Kit (AS-ISK) was applied for the first time in Poland (the RA area) to identify whether or not non-native pumpkinseed, Lepomis gibbosus (Linnaeus, 1758), a freshwater sunfish, posed a high risk of being invasive in the RA area. Material and methods. The AS-ISK was used to screen L. gibbosus for its potential invasiveness in the RA area under current climate conditions of the RA area (i.e. humid continental) and future predicted climate conditions (i.e. temperature increase by 1.5–3.0°C). The risk screening was based on available evidence of the species’ life-history traits (LHT) from its introduced European range, including both ambient and artificially-heated environments. Results. A LHT-based model for predicting L. gibbosus invasiveness revealed that the population in the Oder Canal, which receives heated-water discharge from the “Dolna Odra” electric power station, is amongst the most invasive in Europe. The basic AS-ISK score of 16.5 suggests the species already poses a risk of being invasive in Poland, and this risk is expected to increase under future, warmer climate conditions (AS-ISK Climate Change score = 28.5). Factors and traits affecting L. gibbosus’ invasiveness were: current rate and range of spread, high climatic match, parental care, relatively small size at maturity, opportunistic foraging behaviour, and elevated likelihood of being illegally stocked. Conclusion. Although L. gibbosus is known to cause adverse impacts in some circumstances, these are poorly understood for most of Europe, including the RA area, where the species is likely to disperse and establish new viable populations more widely, especially under future climate conditions. This first application of AS-ISK in Poland emphasises the need for national-level risk screening of non-native species in general, and freshwater fishes in particular, as part of Poland’s non-native species management strategy for the control and containment of invasive species.


INTRODUCTION
A major challenge facing government agencies responsible for the environment is the prediction of which species are likely to become invasive under future (warmer) climatic conditions (e.g. Rahel and Olden 2008). In Europe, the likely response of freshwater fishes to climate change has been modelled for some countries (e.g. the UK and France), where the introduced freshwater sunfish, the pumpkinseed, Lepomis gibbosus (Linnaeus, 1758), is predicted to benefit from the warmer conditions (Buisson et al. 2008, Britton et al. 2010, Fobert et al. 2011. However, no such projections nor risk analysis (i.e. risk identification, assessment, management and communication) have been undertaken for any freshwater fishes in Poland, where an increase in mean annual temperature by 1.5-3.0°C, similar to the UK (Buisson et al. 2008, Britton et al. 2010, is predicted to occur by the end of the 21 st century (Anonymous 2013).
A warmwater fish indigenous to eastern North America (Scott and Crossman 1973), L. gibbosus has a native range that extends southward from temperate New Brunswick (Canada) to sub-tropical peninsular Florida (USA). Widely introduced in Europe ( Copp and Fox 2007), L. gibbosus populations are also found in a few locations in Brazil (de Magalhães andRatton 2005, Santos et al. 2012). The first L. gibbosus introductions into Europe, beginning at the end of the 19 th century, originated from stocks in Canada and New York State (USA) exported to France (Arnold 1990), where the species' establishment was said to have been 'long and laborious' (Kunstler 1908). Imported to Germany from the USA in the early 1880s (Yavno et al. 2020) by Max von dem Born, who was a pioneer of non-native fish and crayfish importations for aquaculture (Copp et al. 2005, Kowarik andRabitsch 2010), and in the same year to France (Yavno et al. 2020). Lepomis gibbosus was subsequently introduced to the Třeboňsko region of the Czech Republic in 1929 (Baruš and Oliva 1995). Since then, the species has established populations in at least 28 countries of Europe (Copp and Fox 2007), where the northernmost (known) reproducing population is situated in southern Norway (Cucherousset et al. 2009)-this believed to have been an unauthorised release of fish imported from the Czech Republic for the aquarium trade (Sterud and Jørgensen 2006).
The motivations for L. gibbosus introductions are varied, such as for aquaculture use in France (Kunstler 1908), as a forage fish for introduced piscivorous fishes in Iberia (Elvira and Almodóvar 2001), as an ornamental pond fish in England (i.e. an 18 th century equivalent of 'koi carp': Copp et al. 2002), and as an aquarium fish in some countries (e.g. Tandon 1976, Sterud andJørgensen 2006). However, in some cases, there have been inadvertent transfers between water bodies as a contaminant of intentional stocking of aquatic species, such as those of young-of-year (YoY) common carp, Cyprinus carpio Linnaeus, 1758 (see Tandon 1976), and native aquatic plants . But, the reported, intentional (but illegal) introductions into lakes in Denmark, stocked along with rainbow trout Oncorhynchus mykiss (Walbaum, 1792), were based on the presumption that L. gibbosus would free O. mykiss from the fish louse Argulus foliaceus (J.K. Jensen, pers. comm.). Notably, this same parasiteremoval role of L. gibbosus has also been observed at a commercial angling venue (B. Brown, Tanyard Fishery: pers. comm.), in a region of England where the species has been present since at least as early as the 1910s (Wheeler and Maitland 1973).
The first records of L. gibbosus from the inland waters of Poland (i.e. including areas of former Germany and the Czech Republic) were in the 1920s-1930s. These refer to single individuals in the River Oder near Krosno Odrzańskie (Pappenheim 1927) and Słubice (Boettger 1934), as well as the River Nysa Łużycka near Gubin (Hoffman 1928). Rare reports refer also to the River Warta near Gorzów Wielkopolski (Thienemann 1950), Martwe Lake near Dziwnów (Wiktor 1959), the Pilchowice Reservoir on the River Bóbr near Jelenia Góra (Balon 1964), and to common carp aquaculture ponds 'Andrzej' and 'Grabownica' in Milicz on the River Barycz (Witkowski 1979). All of the above-mentioned records were of transient fish and all within the River Oder system. In the 1980s, a self-sustaining population was established in the lower River Oder, at a site influenced by warm water from a nearby "Dolna Odra" electric power station (Heese and Przybyszewski 1985). And in the following years, other self-sustaining populations appeared in adjacent water bodies of the city of Szczecin (Porębski and Małkiewicz 1995, Gruszka 1999, Zięba et al. 2016. Currently, L. gibbosus is one of 37 alien fish species recorded in Polish waters (Witkowski and Grabowska 2012). Although L. gibbosus is considered a warmwater species, its native and introduced ranges include countries with cold climates (e.g. Canada, Norway, and Switzerland). And although the species' life-history traits are known to vary in response to temperature (Copp andFox 2007, Masson et al. 2015), the role of local climate in the species' establishment success remains less well studied. Nonetheless, the populations reported to be the most invasive are mainly those located in the southern parts of Europe (Copp andFox 2007, Almeida et al. 2014), but as mentioned here above, mean annual temperatures in Poland are expected to rise, by the end of 21 st century.
The aim of the presently reported study was to assess the current and future invasiveness potential of L. gibbosus in Poland, and specifically to: 1) evaluate the existing published life history pertaining to L. gibbosus invasiveness potential within a wider European context; and 2) assess the species' invasiveness ranking using the Aquatic Species Invasiveness Screening Kit (AS-ISK) of Copp et al. (2016b). The outcomes of the presently reported study will serve to demonstrate to environmental managers and stakeholders in Poland the potential use of AS-ISK as a decision-support tool for informing legislation, policy and management (i.e. prevention, control, containment, eradication) of potential, existing and future undesired translocations of non-native fishes, such as L. gibbosus, in the country. 1994), the LHT-based model for predicting L. gibbosus invasiveness (Copp and Fox 2007) was employed. The most recent version of this model (Masson et al. 2015) was used because it is based on all European L. gibbosus populations for which both back-calculated total length (TL) at age and age at maturity have been published. This highly (statistically) significant model permits the mean age at maturity (in years) to be predicted from mean juvenile growth (i.e. TL at age 2). To predict the mean age at maturity for the RA area, the published TL at age 2 value for the core Polish population (data from Heese and Przybyszewski 1985, cited in table 1 in Copp and Fox 2007) was plotted on a re-drafted version of this model. Because heated-water populations of L. gibbosus are known to mature at age 1 (Dembski et al. 2006), the mean TL at age 1 of this core population was also plotted on the LHT model (for information purposes) because it inhabits a canal of the lower River Oder (53°12′50.34′′N, 014°28′8.31′′E) that receives heated-water effluent from the "Dolna Odra" electric power station. The water temperature therefore is by up to 8ºC higher than that of the river upstream and in the warmest months the water temperature reaches 26-30ºC Kondratowicz 2005, Domagała andPilecka-Rapacz 2007).
To identify the invasiveness potential of L. gibbosus, the AS-ISK decision-support toolkit * was used. Responses to these Qs provide a Basic Risk Assessment (BRA) score (Copp et al. 2016c), which is complemented by six additional 'climate change' questions that ask the assessor to foretell, based on their knowledge of the species, the likely effects of predicted future climate on the risk screening (specifically, the risks and magnitude of introduction, establishment and dispersal). Response scores to these six Climate Change Assessment (CCA) Qs are added to the BRA score, yielding a composite BRA + CCA score. To each question, the assessor must provide a response and a justification for their response (including bibliographic references) and then rank their confidence in that response. The confidence ranking categories are: 1 = low, 2 = medium, 3 = high, 4 = very high (Copp et al. 2016c). In all cases, an overall score < 1 assigns a status of 'low risk' (hence, not likely to be invasive), whereas values ≥ 1 identify alien species as potentially invasive and posing either a 'medium risk' or a 'high risk'. Importantly, it is advisable to identify a 'threshold' value for the RA area concerned by way of a 'calibration' process to distinguish between species of medium and high risk of invasiveness (Copp 2013, Hill et al. 2017. Because there has been no calibration of either FISK or AS-ISK for (freshwater fish in) Poland and the only FISK application to round goby Neogobius melanostomus (Pallas, 1814) in the River Oder Estuary (Czerniejewski and Kasowska 2017) relied on the 'reference' UK threshold of 19 ), which shares the same freshwater ecoregion as Poland (Abell et al. 2008), the choice of BRA and BRA + CCA thresholds to distinguish between medium vs high risk was based on the following three axioms: 1) Based on a global evaluation of 36 applications of FISK worldwide (Vilizzi et al. 2019), and noting that the AS-ISK BRA score is loosely equivalent to the FISK score being based on the original 49 LHT-related Qs (see Introduction), a FISK threshold of 8.2 has been identified for Köppen-Geiger climate class D, in which the country of Poland lies entirely. In light of the above: (i) the threshold of 8.0 (i.e. 8.2 -0.2) was used to distinguish between medium and high risk for the BRA score; and (ii) the threshold of 10.8 (i.e. 8.2 + 2.6) was used to distinguish between medium and high risk for the BRA + CCA score. Notably, the latter case, use of information from only one study to derive an estimate of the BRA + CCA threshold falls within the scope of Bayesian adaptive management practice (Hilborn andMangel 1997, Prato 2005).

RESULTS
The re-drafted and updated LHT-based model with points plotted from the Heese and Przybyszewski (1985) values for TL at age (Copp andFox 2007, Masson et al. 2015) revealed that, regardless of the mean age at maturity used (i.e. ages 1 and 2), the L. gibbosus population in the heated canal near the "Dolna Odra" electric power station (the most abundant and viable in Poland) is within the 'invasive' range for European populations (Fig. 1).
Based on the reference threshold score of 8.0, the BRA score for L. gibbosus in Poland (16.5) falls within the 'high risk' category (Table 1). When the potential effects of climate change on the risk screening responses are taken into consideration, L. gibbosus' BRA + CCA score increases to 28.5 (hence well above the 10.8 threshold) reflecting an even higher risk of the species being invasive in Poland in the future (Table 1). Factors and traits that increased L. gibbosus' AS-ISK score included a history of being invasive elsewhere, high climatic match, parental care, relatively small size at maturity, opportunistic foraging behaviour, and elevated likelihood of being illegally stocked. Traits that reduced the overall score included no likelihood of hybridisation with native species and low risks posed to native threatened or protected taxa (Table 1; Appendix 1). Overall, it is likely that L. gibbosus will continue to disperse and establish in the RA area under current climate conditions, and more likely under predicted future climatic conditions (Appendix 1). In the latter case, the risks of establishment and dispersal would increase the species' risk of invasiveness (Qs 50-52), and also the magnitude of future potential impacts (Qs 53-55).
The mean confidence levels for responses to Qs contributing to the BRA, CCA, and BRA + CCA scores for L. gibbosus in Poland were 2.45 (± 0.18 SE), 2.33 (± 0.61 SE) and 2.44 (± 0.17 SE), respectively, which suggests comparability among AS-ISK groups of Qs. Relative to L. gibbosus BRA and BRA+CCA scores published for other RA areas, those for Poland were consistently lower (Table 1) than for Lake Marmara (Turkey), River Neretva catchment (Bosnia and Herzegovina, and Croatia), and Thrace and Anatolia (Turkey).

DISCUSSION
The risk of invasiveness of L. gibbosus in Poland, such as revealed in this AS-ISK assessment, is a function of the species' probability to be introduced into, and to establish self-sustaining populations in novel environments (Copp and Fox 2007). This is evinced by the increasing number of populations reported in recent years (e.g. Graczyk et al. 2016, Zięba et al. 2016, including isolated water bodies where the releases were presumably the abandonment of unwanted pet fish (Zięba et al. 2016). Other accidental introductions have been as a contaminant of consignments of grass carp Ctenopharyngodon idella (Valenciennes, 1844), which were imported as 'fry' from Hungary (Graczyk et al. 2016). Similar accidental introductions as a contaminant have recently been reported in England, though the vector was water-filled trays of native plants stocked into a small angling water body for habitat enhancement ). In addition to natural spread, which is likely to be facilitated by the warmer temperatures and greater hydrological variability predicted for future climate conditions (Fobert et al. 2013, Zięba et al. 2016, L. gibbosus has a number of attributes that make it likely to be the target of intentional, future releases into new water bodies. Amongst these are the species' ease of capture by rod-and-line fishing (Evangelista et al. 2015), entrainment in landing and keep nets, its attractive coloration for ornamental purposes and aquarists (Copp et al. 2002), as well as its unusual (for Polish ichthyofauna) mating and nest-guarding behaviours (Scott andCrossman 1973, Almeida et al. 2012).

Fig. 1.
Life-history-trait-based model for predicting Lepomis gibbosus invasiveness in Europe (Copp and Fox 2007), redrawn from Masson et al. (2015), presenting mean age at maturity (in years) as a function of mean juvenile growth (TL at age 2, in mm). The proposed physiological transition phase (shaded zone) between non-invasive and potentially invasive pumpkinseed populations is hypothesised as extending from the minimum age at maturity (the 45° line that traces from the intercept, at 'i') and the end of juvenile growth (which for many pumpkinseed populations is age 2; the 45° line that traces through the age 2 intercept with the regression slope, at 'ii'). Data points are given for the mean TL at ages 1 and 2 (filled circles) for the L. gibbosus population in the heated River Oder canal (from Table 1 Regardless of the extent of the current non-native range of distribution, population density control for L. gibbosus should include both a reduction in the risk of escape of larvae and juveniles from semi-isolated water bodies (Fobert et al. 2013) and, where possible, eradication from isolated water bodies. Eradication may be possible by repeated depletion in small water bodies where only a few specimens are present; this was the case for the invasive, topmouth gudgeon, Pseudorasbora parva (Temminck et Schlegel, 1846), which was eradicated from a small pond by repeated electrofishing depletion ). However, an attempt to remove L. gibbosus by intensive angling failed, and this led only to a decrease in maximal individual fish mass, which resulted in a population shift towards smaller size at maturity and overall slower growth (Evangelista et al. 2015). Whereas, repeated removals of L. gibbosus from isolated pond in the River Oder drainage resulted in reductions in spawning stock and in subsequent juvenile abundance in catches (Zięba et al. 2016). Also, introductions of native piscivores (especially northern pike Esox lucius Linnaeus, 1758), a species known to prey on introduced L. gibbosus (see Guti et al. 1991, have proven effective to prevent L. gibbosus from becoming the dominant species thereby impacting on local biodiversity (van Kleef and Jongejans 2014). Other native piscivorous fishes known to prey on non-native L. gibbosus include brown trout, Salmo trutta Linnaeus, 1758, European eel, Anguilla anguilla (Linnaeus, 1758), and Eurasian perch, Perca fluviatilis Linnaeus, 1758 (see Stakėnas et al. 2013). Elsewhere, population expansion by L. gibbosus in the Mirgenbach Reservoir (France) coincided with the extinction of native population of E. lucius (see Masson et al. 2015), though this has been attributed to changes in water quality, such as increased temperature and copper concentrations (Masson et al. 2008).
For fish populations outside their native ranges, LHTs have proved to be particularly good predictors of nonnative fish establishment success (e.g. Fausch et al. 2001, Marchetti et al. 2004, Olden et al. 2006, Tarkan et al. 2016) and invasiveness (Copp and Fox 2007, Masson et al. 2015, Copp et al. 2016b). However, owing to the relation between latitude and temperature, intra-specific variation in LHTs at large spatial scales (cf. latitudinal clines) is also observed (e.g. Copp andFox 2007, Fox and, including phenotypic plasticity and evolutionary adaptation (Muñoz et al. 2016). In the case of L. gibbosus, and similar to other species (Blanck and Lamouroux 2007), a decrease in growth and increase in age at maturity with latitude has been observed in non-native coolwater and warmwater populations of Europe (Copp et al. 2002, Cucherousset et al. 2009, Copp and Fox 2007, and this pattern is observed in both native and non-native populations (Fox and Copp 2014). In particular, significant variation has been found in adult L. gibbosus growth (age 2-3 increment in TL) and mean age at maturity with increasing latitude (Cucherousset et al. 2009). Under natural temperature conditions, juvenile growth rates (TL at age 2) in L. gibbosus appear to decrease significantly with increasing latitude, and differences in growth may extend into the adult stage resulting in stunted adult body size and regardless of lifespan (e.g. Copp and Fox 2007). Consequently, in case of ambient-temperature water bodies, invasiveness may reflect the thermal water regime. Within the more northerly part of its introduced range, such as the UK (Fobert et al. 2013) and the Netherlands (van Kleef et al. 2008), L. gibbosus is still not considered as being invasive, whereas in its southern range of distribution the species already shows traits testifying to its high invasiveness (Fobert et al. 2013).
Results of studies on L. gibbosus LHTs in artificiallyheated water bodies provide further evidence that shifts in Table 1 AS-ISK scoring output for Lepomis gibbosus in some risk assessment (RA) areas within its non-native range of occurrence Section/Category Poland Lake Marmara River Neretva Thrace and Anatolia growth rate, age at maturity or YoY survival are a likely response to climate change (e.g. Dembski et al. 2006, Masson et al. 2015, Zięba et al. 2015. The exceptional water temperature regimes in the centre of L. gibbosus distribution in Poland are caused by the continuous inflow of post-cooling water from the Electric Plant "Dolna Odra" into an adjacent canal, which results in water temperature increases of ≈8°C in summer and up to 15°C in winter relative to the adjacent River Oder Kondratowicz 2005, Domagała andPilecka-Rapacz 2007). As a result, the L. gibbosus population from the heated Oder Canal can be considered to possess LHTs that place it amongst the most invasive of all known populations of L. gibbosus in Europe (Fig. 1). In warmer water bodies, this phenomenon could result in more protracted spawning in some ( , thereby increasing food intake and assimilation efficiency but also reducing energy expenditure (Cucherousset et al. 2009). As a result, contrary to ambient water sites, YoY individuals from warmwater populations are known to suffer lower mortality in their first winter of life (Zięba et al. 2015), and whilst facing a physiological trade-off between maturation and continued juvenile growth, they tend to mature earlier (Masson et al. 2015). Apart from a shift in the range of distribution, which is an expected response to climate change conditions (Parmesan 2006), LHTs adjustments allow the species to increase their overall survival (Zięba et al. 2015) and colonisation rate (Fobert et al. 2013). So, once relocated to warmer water bodies, L. gibbosus are expected to experience an increase in both recruitment and propagule pressure (Fobert et al. 2013).
In conclusion, this first application of AS-ISK in Poland emphasises the need for a more extensive risk screening of non-native aquatic species in general and non-native fishes in particular in the RA area. This will require reliable assessment and computation of AS-ISK BRA and BRA + CCA reference thresholds in order to provide environmental managers and stakeholders in Poland with the knowledge of potential adverse impacts of non-native aquatic species, both under current and predicted climatic conditions. In fact, this represents a crucial step for planning control and containment of invasive species as part of environmental management initiatives. With specific regard to L. gibbosus in Poland, the most viable and abundant population occupies an environment that is warmer than natural water bodies in its native range and relative to most non-native locations within Europe, including other infested waters in Poland. Being that L. gibbosus in both its native and introduced ranges occupies water bodies across broad climatic gradients (Fox and Copp 2014), this species' phenotypic plasticity renders it capable of expanding its range in Poland under both current and future climate conditions. However, despite the BRA and BRA + CCA scores, which classify L. gibbosus as likely to pose a high risk of invasiveness, the species' known adverse impacts are not yet fully understood for the RA area, and the evidence from other northerly parts of Europe is equivocal. For example, studies in southern England of the species' microhabitat and trophic interactions with native fishes found limited or no evidence for adverse impacts , Jackson et al. 2016, Copp et al. 2017a), whereas impacts have been recorded in managed ponds in the Netherlands (van Kleef et al. 2008) and in natural streams of the Iberian Peninsula (Almeida et al. 2014). In fact, the bulk of evidence for L. gibbosus impacts in Europe comes from the Iberian Peninsula, where the species is invasive mainly in human-degraded environments (e.g. Ferreira et al. 2007, Almeida et al. 2009), though impacts in at least one 'natural' water course have been documented (Almeida et al. 2014). As such, further research is needed in Poland to assess the potential impact of L. gibbosus on this country's native species and ecosystems. Yes L. gibbosus has established populations in at least 28 European countries, and has been the subject of both intentional and incidental introductions (Copp and Fox 2007). This species is known to be easily reared in captivity, and was introduced in 1927 into the RA area for the purpose of keeping ornamental fish in ponds (Witkowski and Grabowska 2012)

REFERENCES
Very high 2 1.02 Is the taxon harvested in the wild and likely to be sold or used in its live form?
Yes L. gibbosus was reported as by-catch from commercial fishers operating in the RA area and is also the target of rod-and-line fishing. This species is also known to be illegally stocked in isolated ponds nearby the RA area (  Higher To the best of the Assessor's knowledge, future climate change within RA area may influence L. gibbosus potential impact on ecosystem services/socio-economic factors by e.g. changing the attractiveness of angling venues, reducing nature reserve purity, consequently leading even to penalties from the EU Low