Introduction of Mysis relicta (Mysida) reduces niche segregation between deep-water Arctic charr morphs

Niche diversification of polymorphic Arctic charr can be altered by multiple anthropogenic stressors. The opossum-shrimp (Mysis relicta) was introduced to compensate for reduced food resources for fish following hydropower operations in Lake Limingen, central Norway. Based on habitat use, stomach contents, stable isotopes (δ13C, δ15N) and trophically transmitted parasites, the zooplanktivorous upper water-column dwelling ‘normal’ morph was clearly trophically separated from two sympatric deep-water morphs (the ‘dwarf’ and the ‘grey’) that became more abundant with depth (> 30 m). Mysis dominated (50–60%) charr diets in deeper waters (> 30 m), irrespective of morph. Mysis and/or zooplankton prey groups caused high dietary overlap (> 54%) between the ‘dwarf’ morph and the two other ‘normal’ and ‘grey’ morphs. After excluding Mysis, the dietary overlap dropped to 34% between the two profundal morphs, as the ‘dwarf’ fed largely on deep-water zoobenthos (39%), while the ‘grey’ morph fed on fish (59%). The time-integrated trophic niche tracers (trophically transmitted parasites and stable isotopes) demonstrated only partial dietary segregation between the two deep-water morphs. The high importance of Mysis in Arctic charr diets may have reduced the ancestral niche segregation between the deep-water morphs and thereby increased their resource competition and potential risk of hybridization.


Introduction
Ecologically induced speciation may lead to a continuum of evolutionary differences within and among populations, with some groups being in the process of diversifying and others being reproductively isolated (Schluter, 2000;Hendry et al., 2009). Postglacial lakes are useful systems for studying the impacts of biodiversity changes caused by environmental, ecological and human-induced factors as they are semiclosed ecosystems with relatively well-defined habitats that can host polymorphic populations at different stages of evolutionary divergence (Schluter, 2000;Klemetsen, 2010;Hendry et al., 2017). Human activities may have large impacts on natural environments by rapidly changing the direction of evolutionary development, and in some instances, reverse the evolutionary processes that promote increasing biodiversity (Hendry et al., 2017). Multiple anthropogenic stressors in lake ecosystems, including pollution, commercial fishing and non-native species introductions, have reversed speciation processes (Seehausen et al., 2008;Alexander et al., 2017;Kuparinen & Festa-Bianchet, 2017). For example, reproductive breakdown has been observed in newly differentiated native morph-pairs of whitefish (Coregonus lavaretus) and stickleback (Gasterosteus aculeatus) following the introduction of competitive fish species or invasive crayfish (Taylor et al., 2006;Velema et al., 2012;Bhat et al., 2014).
In postglacial lakes, polymorphic fish populations often diverge along the benthic-pelagic resource axis (e.g. Schluter, 1996Schluter, , 2000. Charr (Salvelinus spp.) is one of the few genera that is also found to diversify along the shallow vs. deep-water benthic resource axis Klemetsen, 2010;Muir et al., 2016;Markevich et al., 2018). Knowledge about deepwater (profundal) morphs of Arctic charr (S. alpinus) is still relatively limited, although they seem to occur across the entire Holarctic region (Klemetsen, 2010). The deep-water morphs are typically reproductively isolated from co-occurring littoral and pelagic morphs (Hindar et al., 1986;Westgaard et al., 2004;Simonsen et al., 2017). Moreover, they express heritable specialised physiological, behavioural, and morphological adaptations (Klemetsen et al., 2002Knudsen et al., 2015) to effectively exploit (i.e. for foraging and mating) the deep-water niches (Knudsen et al. 2016a). Small-sized deep-water morphs feed mainly on benthic invertebrates, whereas the few known, large-growing deep-water morphs are specialised piscivores (Knudsen et al., , 2016bPower et al., 2009;Klemetsen, 2010;Moccetti et al., 2019). Although the ecology of the deep-water morph has been studied, little is known about the potential impacts of multiple anthropogenic stressors (e.g. hydropower operations and the introduction of nonnative species) on these deep-water morphs.
Lake Limingen, central Norway, has a polymorphic population of Arctic charr, consisting of three morphs: the upper-water 'normal' morph, the deep-water 'dwarf' morph, and the piscivorous 'grey' morph (Nyman et al., 1981;Aass et al., 2004). Following the damming of the lake in 1953 for hydropower production, brown trout (Salmo trutta) and Arctic charr population densities drastically declined (Aass et al., 2004;Gregersen et al., 2006). In 1969, Mysis relicta (hereafter Mysis) was introduced to compensate for reduced fish food resources and to mitigate the negative impacts of hydropower operations on fish and overall ecosystem productivity (cf. Hirsch et al., 2017). Mysis is an opossum shrimp native to Scandinavia, though previously absent from this Limingen region of Norway (Spikkeland et al., 2016). Mysis show a pronounced diel vertical migration pattern, with nocturnal foraging on zooplankton in the upper water column potentially resulting in food resource competition with zooplanktivorous Arctic charr (Moen & Langeland, 1989;Naesje et al., 1991;Koksvik et al., 2009). In contrast, deep-water fish (e.g. burbot Lota lota and profundal Arctic charr) may benefit from Mysis introductions through increased food availability (Langeland et al., 1991;Naesje, 1995). Whilst introduced Mysis populations have become an important prey resource for Arctic charr in Limingen and elsewhere (Garnås, 1986;Gregersen et al., 2006), detailed studies of their impacts on trophic differentiation among sympatric Arctic charr morphs have been lacking.
In this study, we investigated the habitat use, diet, parasite infections and stable isotope ratios (d 13 C, d 15 N) of the three sympatric Arctic charr morphs in Limingen. The aim of the study was to explore the degree of niche overlap between the three sympatric Arctic charr morphs about 50 years after the introduction of Mysis. We quantified niche overlap by using data on habitat use and stomach contents (recent niche-use) and by analysing the occurrence of trophically transmitted parasites and stable isotope values that reflect the temporally integrated trophic niches of individual fish (Knudsen et al., 2011). We hypothesised that existing depth-habitat preferences for the sympatric morphs would be maintained. However, due to damming and the introduction of Mysis, we also hypothesised that prey resource use would overlap, particularly between the upper-water 'normal' and the deep-water morphs.

Study lake
Lake Limingen (64°50 0 N, 13°13 0 E) is a large (surface area = 95.7 km 2 ), deep (Z max = 192 m, Z mean-= 87 m), dimictic, oligotrophic and relatively clear (Secchi depth = 9-12.7 m) lake situated at 418 m a.s.l. in the north boreal vegetation zone of central Norway. Originally, the lake drained to the Å ngermanälven watercourse in northern Sweden, but after hydropower development in 1953 most of the water was diverted to the Namsen watershed in Trøndelag County, Norway . Today, the lake is regulated with a maximum annual water level amplitude of 8.7 m. Spruce forests with some birch dominate the riparian vegetation, and there are only a few low-intensity farms around the lake. In addition to the polymorphic Arctic charr, the lake has a small population of brown trout, a littoral population of minnow (Phoxinus phoxinus; introduced in 1980s) and a very sparse population of three-spined stickleback (introduction date unknown) (Aass et al., 2004;Gregersen et al., 2006).

Fish material
Arctic charr were sampled in August 2016 with Nordic multi-mesh gill-nets consisting of 2.5 m panels with 12 different knot-to-knot mesh sizes from 5 to 55 mm (Appelberg et al., 1995). The nets were set in the littoral (1.5 m high benthic nets; 0-15 m depth), pelagic (6 m high offshore gill-nets set from the surface; above 30 m depth), and profundal (1.5 m high benthic nets; at 20-50 m depth) zones. Additional sampling with a pelagic pair trawl caught 63 'normal' morph Arctic charr (see details in Sandlund et al., 2017). Fish were weighed (closest 0.1 g) and measured (closest 1 mm, fork length, L F ). Otoliths were removed for age determination.
Individual Arctic charr were classified to one of three possible morphs based on head and body morphology, maturation, and colouration following guidelines produced from earlier studies of similar polymorphic populations (Skoglund et al., 2015;Simonsen et al., 2017). In total, we sampled 178 Arctic charr from Limingen, with stomachs analysed from 171 individuals. The catch per unit of effort (CPUE) was estimated as the number of fish caught per 100 m 2 gillnet per night. A subsample of mature individuals from all three morphs was assessed for parasite assemblages and sampled for stable isotope ratios (d 13 C and d 15 N). The number of each morph included in the parasite and stable isotope sampling were: 'normal' morph (n = 39; mean ± SD: L F-= 310.6 ± 83.7 mm; age = 6.7 ± 2.5 years), 'dwarf' morph (n = 27; L F = 173.9 ± 31.7; age = 6.4 ± 2.6), and 'grey' morph (n = 14; L F-= 297.1 ± 59.3; age = 9.5 ± 3.4).
Growth differences among morphs were described by mean length-at-age using a modified von Bertalanffy growth model (Roff, 1984): where L T is fish body length at time T, L ? is the asymptotic fish length, k is the growth coefficient, and A T is the age at time T. This simplified model has been shown to work well with inland polymorphic salmonids (Jonsson et al., 1988).

Diet
Prey items from fish stomachs were preserved in ethanol and later identified to the lowest feasible taxonomic level (23 different prey taxa in total) and subsequently sorted into five main categories: (i) zooplankton (e.g. Daphnia, Bosmina, Holopedium, Bythotrephes, copepods), (ii) surface insects (adult insects), (iii) benthos (e.g. snails, clams, insect larvae, benthic crustaceans), (iv) Mysis, and (v) fish. The contribution of each prey category to the diet was estimated by visual determination of the stomach fullness using a percentage scale ranging from empty (0%) to full (100%) (prey abundance; Amundsen et al., 1996). Among morph dietary overlap was quantified for all prey categories using Schoener's (1970) similarity index, which is commonly considered high when the overlap exceeds 60% (Wallace, 1981).

Parasites
All parasites from the body cavity, stomach, intestine, kidney, swim bladder, gills and eyes were enumerated from sub-sampled fish (see Table 3 for more details). Most of the parasite taxa are transmitted to Arctic charr via different prey items such as copepods (cestodes Dibothriocephalus spp., Proteocephalus sp. and Eubothrium salvelini), insect larvae (trematodes Crepidostomum spp.), and the benthic amphipod Gammarus lacustris (cestode Cyathocephalus truncatus, nematode Cystidicola farionis, and Acanthocephalan sp.). Mysis may also transmit the swim bladder nematode C. farionis (Black & Lankester, 1980); however, the intermediate host for this parasite is currently unknown. All taxa, except Dibothriocephalus spp., utilize Arctic charr as the final host (see Table 3 for further details). Larval Dibothriocephalus spp. (former Diphyllobothrium spp., see Waeschenbach et al., 2017) are able to use fish as parathenic hosts and re-establish in piscivorous individuals (Curtis, 1984), which also may be the case for Eubothrium sp. (Williams & Jones, 1994). Additionally, three parasite taxa are non-trophically transmitted to the fish, either from other fish, i.e. the parasitic gill crustacean (Salmincola edwardsii), or via trematode larvae released from intermediate snail hosts, i.e. Diplostomum sp. and Apatemon sp. We quantified parasite prevalence (percentage of hosts infected by the parasite) and abundance (number of parasites per host) following methods outlined in Bush et al. (1997). The exceptions were Diplostomum sp. and Apatemon sp., for which the prevalence and abundance were estimated from a single eye (at random).

Stable isotopes
Stable isotopes of carbon (d 13 C) and nitrogen (d 15 N) are commonly used to estimate the dietary sources (littoral vs. pelagic carbon) and trophic position of organisms in lake food webs, as well as the intra-and inter-specific niche segregation of fish populations (e.g. Boecklen et al., 2011;Layman et al., 2012). Here, a small piece of dorsal muscle tissue, obtained posterior to the dorsal fin, was dissected from a subsample of fish and frozen at -20°C. Tissue samples were dried at 60°C for 48 h and homogenised using a pestle and mortar. Approximately 0.3 ± 0.05 mg of dried tissue was weighed and placed in tin capsules for analyses completed at the University of Waterloo, Canada, on a Delta plus continuous flow stable isotope ratio mass spectrometer (Thermo Finnigan, Bremen, Germany) coupled to a Carlo Erba elemental analyser (CHNS-O EA1108, Carlo Erba, Milan, Italy). The machine analytical precision of ± 0.2% (d 13 C) and ± 0.3% (d 15 N) was determined through the repeat analysis of internal laboratory standards calibrated against International Atomic Energy Agency standards CH6 for carbon and N1 and N2 for nitrogen.

Statistical analyses
All statistical analyses were computed using R (version 3.4.2, R Core Team, 2017). Differences in the number of parasite taxa among morphs were compared using a general linear model fitted with a Gaussian distribution family. Differences in total parasite abundance and the abundance of each parasite taxon among charr morphs were examined using a series of generalized linear models, with the exception of five rare taxa (C. truncatus, E. salvelini, S. edwardsii, Acanthocephalan sp., unidentified nematode), which were excluded from further analysis. Generalized linear models were fitted with a quasipoisson distribution and log-link function due to the over-dispersion of abundance data. Fish age (years) was included as a continuous fixed factor in all general and generalized linear models to account for the influence of varying host age on parasite abundance and richness. Potential outliers were identified by graphically examining the raw data and by running models with and without outliers to assess their influence on model outcomes. Analysis of variance was used to assess whether the interaction term between morph and age provided additional explanatory power over the simpler additive model. Contrast analyses were constructed for each final model set by varying the base morph (intercept) to assess the significance of differences between morph pairs. Multivariate analyses were conducted using the package vegan (version 2.5-2, Oksanen et al., 2017). Parasite community composition differences among morphs were visualized by using individual Arctic charr in a non-metric multidimensional scaling analysis (NMDS) based on Bray-Curtis dissimilarities of log-transformed parasite abundances, including both trophically and directly transmitted parasite species. To visualize and explore the correlation between individual diet and parasite community composition, we used canonical correlation (vegan: CCorA, Oksanen et al., 2017) of logit-transformed prey volumes and log-transformed abundances of trophically transmitted parasites.
Non-parametric Kruskal-Wallis tests, followed by pairwise comparisons with Mann-Whitney U tests, were used to evaluate the significance of differences in d 13 C (reflecting littoral versus pelagic resource use) and d 15 N (reflecting trophic position) values among the three charr morphs. Isotopic niche overlaps were calculated between all pairs of morphs using the probabilistic method developed by Swanson et al. (2015), available in the R-package nicheROVER (Lysy et al., 2014). In this method, a Bayesian approach is employed to produce 95% probability niche regions and directional estimates of pairwise niche overlap. Niche overlap is defined as posterior probabilities that an individual of one morph falls within the niche region (95%) of the other morph. Potential outliers were identified graphically from the raw data, and two outliers were removed to ensure better fit to multivariate normal distribution of the data.

Results
Fish community, habitat preference and Arctic charr growth Arctic charr was the dominant species (n = 168) in the benthic habitats (i.e. littoral and profundal), whereas only nine 'normal' Arctic charr were caught in the pelagic zone (0.6 charr per 100 m 2 gillnet area).
Length-at-age differed among the morphs, with significant differences in mean length (t tests, P \ 0.05) observed between 'normal' and profundal 'dwarf' morphs for each age-class between 4 and 9 years (Fig. 1b). Estimated von Bertalanffy growth models indicated greater asymptotic lengths for the 'normal' and 'grey' morphs as compared with the 'dwarf' morph, with non-overlapping confidence intervals indicating significantly different maximal sizes for all morphs. Growth rate (k) similarly differed between the morphs as indicated by non-overlapping confidence intervals, being lower in the 'normal' and 'grey' morphs and highest in 'dwarf' morph (Table 1). It should be noted that the precision of parameter estimates for the 'grey' morph was possibly affected by the smaller number of fish available for estimating model parameters.

Dietary niches
Generally, the abundance (%) of Mysis in charr stomach contents increased with depth, independent of the morph considered (Fig. 2a), being about 10% in the upper water column (0-10 m) and [ 60% in deep waters ([ 50 m). In contrast, the diet of all Arctic charr captured in the uppermost water column (\ 30 m depth) was dominated by zooplankton ([ 53%) and surface insects ([ 20%). Mysis constituted 18% of the diet of the 'normal' morph, 35% of the 'grey' morph diet and 39% of the 'dwarf' morph diet (Fig. 2b). The 'dwarf' morph consumed zooplankton (29%) and benthos (23%), in addition to Mysis. The 'dwarf' morph ate much less Daphnia and Bythotrephes, but approximately equal amounts of Bosmina and Holopedium when compared with the 'normal' morph. The 'grey' morph relied more heavily on fish (38%) than the other morphs, which had less than 1.5% fish in their stomachs. Common consumption of Mysis and/or zooplankton prey groups caused a relatively high dietary overlap (54-56%) between the 'dwarf' morph and the two other morphs when considering all prey groups (23 taxa). Dietary overlap was lower (41%) between the 'normal' and the 'grey' morph. After removing Mysis as a prey group, the dietary overlap dropped to 34% between the two deepwater morphs, as the 'dwarf' morph fed mainly on zooplankton (49%) and deep-water zoobenthos (39%), while the 'grey' morph fed mainly on fish (59%).
Parasite community composition in individual Arctic charr appeared to be more similar in the two profundal morphs compared to the 'normal' morph ( Fig. 3). Of the three parasite taxa non-trophically transmitted to charr, Diplostomum sp. and Apatemon sp. were the most prevalent among morphs (* 40-65%), whereas S. edwardsii tended to occur in the 'normal' charr morph (20%). The abundance of non-trophically transmitted parasites was consistently low (\ 3 individual parasites per fish) and did not differ among morphs, although there was a positive relationship between Diplostomum sp. abundance and charr age (Tables 3, S2). Of the trophically transmitted parasites, the upper water-column 'normal' morph had the highest prevalence for five of eight parasite taxa, and the remaining three parasite taxa were most prevalent in the piscivorous 'grey' morph. Two Gammarus transmitted taxa, C. truncatus and Acanthocephala sp., were restricted to 'normal' charr morphs only. Dibothriocephalus spp. cestode larvae were the most prevalent trophically transmitted parasite and occurred in similar abundances in all morphs (Tables 3, S2). The copepod-transmitted taxa, i.e. Proteocephalus sp. and Eubothrium sp., were more prevalent in the 'normal' morph than in the deep-water charr morphs. Proteocephalus sp. abundance was greater in the 'normal' morph than in the 'dwarf' morph, with the abundance of this parasite declining with charr age (Tables 3, S2). The swim bladder nematode C. farionis was found most often in the piscivorous 'grey' morph, although it was in consistently low abundance in all charr morphs (Table 3).
The community composition of trophically transmitted parasites in individual charr was significantly explained, albeit moderately, by the diet composition of the individual (Canonical Correlation R adj 2 = 0.30, P \ 0.001; Fig. 4). Thus, when visualizing both the most recent diet (stomach contents) and the temporally integrated characterization of resource use as measured by trophically transmitted parasites, all morphs appeared to have different trophic niches (Fig. 4). The Fig. 3 Parasite community composition for the three morphs of Arctic charr, the 'normal' (red), the 'dwarf' (green) and 'grey' (grey) found in Limingen, visualized using Non-metric multidimensional scaling based on Bray-Curtis dissimilarities of parasite infra-communities (n = 59; stress: 0.23). Letters denote the mean for each morph two deep-water morphs were located closest to each other, indicating they have more similar trophic niches. The 'normal' morph was more separated (Fig. 4) and associated with higher infections of Proteocephalus sp. and Crepidostomum spp. as a result of feeding on a different assemblage of zooplankton species and insect larvae than the 'dwarf' or 'grey' morphs. The 'grey' and 'dwarf' morphs were mainly associated with infections of Dibothriocephalus spp. and C. farionis, linked to feeding on Mysis, mussels, chironomid larvae, and fish.

Discussion
We observed a partial niche segregation between the three sympatric Arctic charr morphs in Limingen, with the clearest segregation being between the upper water-column zooplanktivorous 'normal' morph and the two profundal morphs, the 'dwarf' and the piscivorous 'grey' morphs. Although we do not have directly comparable data from each of the morphs before the Mysis introduction (Gregersen et al., 2006), our results suggest that the Mysis introduction has reduced niche segregation between the three sympatric Arctic charr morphs as a result of common exploitation of this resource, with the strongest impacts being on the two deep-water morphs. The temporally integrated trophic tracers (parasite fauna and stable isotope values) pointed to a further partial dietary segregation between the two Arctic charr morphs with identical deep-water preferences, with the 'dwarf' and 'grey' morphs supplementing a Mysisbased diet with benthic prey and fish, respectively. The reduced trophic segregation has increased the apparent ecological similarity between the morphs and has the potential to enhance the probability for increased competitive interactions and hybridization. A corresponding trophic segregation between the 'normal' morph and the sympatric profundal morphs occurs in some other polymorphic lakes (Knudsen et al., , 2016aAmundsen et al., 2008;Moccetti et al., 2019). In Arctic charr, a zooplanktivorous diet is generally found in southern Scandinavian lakes (e.g. L' Abée-Lund et al., 1993;Fig. 4 The relation between the most recent trophic niche (stomach content, blue text) and community composition of trophically transmitted parasites representing a temporally integrated trophic niche (red text) in 58 individuals of the three Arctic charr morphs found in Limingen: 'normal' (red), 'dwarf' (green) and 'grey' (grey) morph. Letters denote the mean for each morph (Canonical correlation: R adj 2 = 0.30, P \ 0.001) Jensen et al., 2017;Paterson et al., 2019), in lakes regulated for hydropower production (e.g. Hirsch et al., 2017), and in northern lakes with benthivorous competitors (e.g. Skoglund et al., 2013;Eloranta et al., 2013). In northern lakes with deep-water morphs, the upper water-column 'normal' Arctic charr morph may also include littoral resources in the diet (Knudsen et al., 2010(Knudsen et al., , 2016aEloranta et al., 2013;Moccetti et al., 2019). In Limingen, however, the benthic resources in shallow littoral areas are restricted due to water level fluctuations that reduce littoral zone productivity (e.g. Hirsch et al., 2017) and the occupancy of available shallow areas by abundant minnows and a few brown trout (Aass et al., 2004;Gregersen et al., 2006). Low presence of littoral benthos in the diet of the 'normal' morph is supported by low infection by the few parasite species transmitted from benthic prey (i.e. Crepidostomum sp.), as also observed in other studies in this geographic region (Paterson et al., 2018(Paterson et al., , 2019. The small-sized deep-water 'dwarf' morph included both zooplankton and Mysis in the diet, resulting in a relatively high dietary overlap with the 'normal' and 'grey' morphs. Small-sized deep-water Arctic charr morphs typically specialize on softbottom benthos (Hindar & Jonsson, 1982;Knudsen et al., 2006Knudsen et al., , 2016aHooker et al., 2016;Moccetti et al., 2019), as do profundal whitefish morphs (Harrod et al., 2010;Praebel et al., 2013;Siwertsson et al., 2013) and brown trout (Piggott et al., 2018). In Limingen, excluding Mysis consumption reduced the apparent dietary overlap between the two profundal  Mean [range 95% credibility interval] probability (%) of finding an individual of the morph in the row within the niche region of the morph in the column morphs, with consumption of prey resources other than Mysis pointing to a more distinct benthivorous dietary niche for the 'dwarf' morph. Although separation into morph groupings was not reported, zoobenthos were noticeably more common in the diet of Arctic charr prior to the Mysis introduction (Gregersen et al., 2006). The introduction of Mysis may have induced a dietary shift by the 'dwarf' morph towards a more pelagic diet as a result of the diel vertical migration of Mysis within the water column. The 'dwarf' morph also had significantly higher d 15 N values and less diverse parasite fauna when compared to the 'normal' morph in Limingen and nearby lakes (Paterson et al., 2018(Paterson et al., , 2019, as has been reported for other polymorphic Arctic charr lakes (Knudsen et al., , 2016aSiwertsson et al., 2016). Despite the apparently large dietary overlap, the above suggests that the 'dwarf' morph has a less unique benthivorous diet in Limingen than in other lakes (see also Moccetti et al., 2019).
The relative importance of Mysis in the diet of the 'grey' morph is not typical for large-growing Arctic charr (but see Eloranta et al., 2015), although lake charr (S. namaycush) predate substantially on Mysis (e.g. Chavarie et al., 2016) particularly when introduced to oligotrophic lakes (e.g. Ellis et al., 2011). In Limingen, fish was an important prey for the 'grey' morph, but not for the sympatric 'normal' and 'dwarf' morphs, indicating the position of 'grey' Arctic charr as specialized piscivores (Adams et al., 1998;Power et al., 2005;Knudsen et al., 2016b;Moccetti et al., 2019). The inclusion of a specialized piscivore among lake-resident morphs is also found in other polymorphic Salvelinus spp. populations Markevich et al., 2018). Although the relatively high d 15 N values of the 'grey' morph partly reflected their piscivorous diet, the morph was less clearly separated from the 'dwarf' morph than has been evident in studies of other profundal morph-pairs (Knudsen et al., 2016a;Moccetti et al., 2019). The 'grey' morph had a higher diversity of trophically transmitted parasites than 'dwarf' morph, likely passed on via prey fish as has been noted elsewhere (Siwertsson et al., 2016;Moccetti et al., 2019). The 'grey' morph also had aggregated high Dibothriocephalus spp. infections (a cestode able to re-establish in predatory fish; e.g. Curtis, 1984), as is often seen in other piscivorous Salvelinus spp. morphs (Frandsen et al., 1989;Butorina et al., 2008;Siwertsson et al., 2016;Moccetti et al., 2019). However, the parasite data also suggest abundant ingestion of Mysis by the 'grey' morph, as C. farionis (a swim-bladder nematode potentially transmitted by mysids; Black & Lankester, 1980) were most frequent in the 'grey' morph. Overall, the parasite results (i.e. community structure and/or abundance) described here support previous conclusions that piscivorous predators are exposed to a portfolio of parasite species that differ from those found in sympatric invertebrate feeding morphs (Siwertsson et al., 2016;Moccetti et al., 2019).
The profundal 'grey' and 'dwarf' morphs in Limingen also showed greater similarity in diets and growth rates than the sympatric deep-water benthivorous and piscivorous charr morphs found elsewhere (Smalås et al., 2013;Knudsen et al., 2016a, b;Moccetti et al., 2019), likely as a result of Mysis consumption as has been noted for lake charr feeding on introduced Mysis in Flathead Lake, Montana (Ellis et al., 2011). The introduced Mysis is also one of the main reasons for the high dietary overlap, which was similarly reflected in the overlap in isotopic niches and parasite fauna between the deep-water Arctic charr morphs. When present, Mysis may dominate the diet of benthic and pelagic Arctic charr in Scandinavian lakes in all seasons, but especially during winter when zooplankton are scarce (Garnås, 1986;Naesje, 1995;Hammar, 2014). Introduction of Mysis in polymorphic Arctic charr lakes may therefore diminish the ecological segregation between sympatric morph pairs and alter the local selection regimes. Whilst there is no information regarding reproductive isolation (e.g. time and place of spawning) for Limingen Arctic charr, the morphs are thought to be genetically different (Nyman et al., 1981). In several other postglacial lakes, upper water-column morphs of Arctic charr and whitefish are genetically different from their sympatric benthivorous deep-water morphs, as well as from resident piscivorous morphs (Verspoor et al., 2010;Praebel et al., 2013Praebel et al., , 2016Siwertsson et al., 2013;Simonsen et al., 2017;Moccetti et al., 2019). The 'normal' morph in Limingen differs from the other two sympatric morphs in terms of habitat depth, whereas the two deep-water morphs segregate in terms of piscivory; yet, all three morphs prey on the introduced Mysis. Thus, reliance on Mysis clearly reduces the niche segregation between the morphs and has also likely affected energy flow pathways through the lake food web (e.g. Ellis et al., 2011). Ecological convergence (e.g. similarity in diet) as observed in the present 'dwarf' and 'grey' morphs, may even promote hybridization. The 'reverse speciation' process (increased hybridization) among native fish morphs has been reported from other lakes where the introduction of non-native competitors or potential prey has impaired ecological segregation (e.g. Taylor et al., 2006;Vonlanthen et al., 2012;Bhat et al., 2014).
Multiple human-induced stressors are evident in Limingen and common in many Scandinavian freshwater systems (Hirsch et al., 2017). Hydropowerinduced water level fluctuations provided the initial environmental stressor that reduced littoral benthic food resources for fish (Gregersen et al., 2006). Another human-induced ecosystem stressor was the introduction of Eurasian minnow, a typical shallowwater benthivorous resource competitor for salmonids Museth et al., 2010). Finally, the introduction of Mysis may have further altered the niche use of the 'normal' morph through increased competition for zooplankton resources (Langeland et al., 1991). There is generally little understanding about how multiple human-induced stressors may affect relatively simple postglacial lake ecosystems such as Limingen, and no information exists on how cumulative stressors can affect the evolutionary processes structuring polymorphic Arctic charr populations (Sandlund & Hesthagen, 2011). By introducing Mysis into a lake ecosystem with a littoral zone impaired by hydropower operations, the evolutionary selection regimes appear to have been modified, which may in turn induce a breakdown of the reproductive isolation between established morphs as a result of increasing the functional ecological similarity among the morphs.
For management of the scattered and unique deepwater morphs of Arctic charr, it is important to obtain an overview of the occurrence of intra-lake divergence within populations, describe their biological characteristics and determine the environmental prerequisites for their occurrence. Based on recent ecological and genetic studies, deep-water morphs of Arctic charr and whitefish are replicated in several locations and appear to originate locally Østbye et al., 2006;Klemetsen, 2010;Praebel et al., 2013Praebel et al., , 2016. Profundal morphs of Arctic charr have evidently inherited traits selected for surviving in cold, dark and nutrient-poor deep-water environments, including specific adaptations in trophic morphology, behaviour and growth (Klemetsen et al., 2002Knudsen et al., 2015). Other traits seem to a lesser degree to be under strong natural selection, such as temperature preference and vision capabilities (Siikavuopio et al., 2014;Kahilainen et al., 2016). Without appropriate knowledge of the occurrences of traits within and among populations of Arctic charr, and of northern lake resident fish in general, a full understanding of the functional biodiversity within these lakes will remain unknown. Functional diversity is an important component of biodiversity in northern lakes (Sandlund & Hesthagen, 2011) and its categorization is particularly important given the rapid anthropogenic induced environmental changes that are altering ecosystems and biodiversity faster than the diversity can be inventoried (Reist et al., 2013).
A second concern and challenge for management is to identify potential threats to these deep-water morphs that may reduce their abundance or even cause local extinction. There seems to be no population-specific, cold-water adaptations in deep-water Arctic charr morphs as they have the same estimated temperature preferences as those from Svalbard and most of Scandinavia (Larsson et al., 2005;Siikavuopio et al., 2014). Profundal morphs, however, tend to spawn later than sympatric upper water-column morphs, during the winter when lakes are normally ice-covered (Klemetsen et al., 1997;Smalås et al., 2017). The profundal zones in deep oligotrophic postglacial lakes are relatively stable environments, experiencing less variability in food supply and temperature regimes (e.g. Mousavi & Amundsen, 2012). Thus, populations inhabiting these lakes may be less affected by moderate global warming (Poesch et al., 2016), as they can thermally buffer in cold deep waters isolated from summer temperature stratification. Arctic charr populations that spawn in shallow areas may actively avoid the warm upper watercolumn layers during summer stratification (Murdoch & Power, 2012) but may alter spawning timing or habitat (e.g. Winfield et al., 2010;Jeppesen et al., 2012). Thus, upper water-column morphs may be more severely affected by an accumulation of anthropogenic-induced stressors, e.g. climate change and hydropower-induced water level fluctuations. Furthermore, if whole lake ecosystems are significantly modified, there may be cascading ecological consequences even for deep-water morphs, as has been suggested by the data from Limingen. This may include an increased risk of hybridization between morph-pairs that will eventually reduce the intraspecific biodiversity apparent in many Scandinavian lakes.