Size-age population structure of an endangered and anthropogenically introgressed northern Adriatic population of marble trout (Salmo marmoratus Cuv.): insights for its conservation and sustainable exploitation

Introduction

In the inland waters of industrialized countries, sport fishing is an extremely popular leisure activity that attracts social, cultural, and economic interests; salmonids make up the backbone of these fisheries (Unfer & Pinter, 2018; Brown et al., 2019; Arlinghaus et al., 2019). Unsustainable management of salmonid stocks is a widespread concern. Overharvesting and poorly designed hatchery and stocking programs fuel the spreading of invasive and interfertile non-native taxa that compete and hybridize with native populations (e.g., Naish et al., 2007). These anthropogenic impacts interact synergistically with habitat destruction, habitat degradation, and climate change, thus affecting the population size-age structure of native salmonids through density-dependent regulating mechanisms (e.g., Marschall & Crowder, 1996; Vincenzi et al., 2007a; Ayllón et al., 2019).

The marble trout Salmo marmoratus Cuvier, 1829 is a large salmonid exceeding one meter in total length (TL). It is characterized by a marbled coloration and is phylogenetically distinct from other Salmo species (e.g., Pustovrh, Sušnik Bajec & Snoj, 2011; Pustovrh, Snoj & Sušnik Bajec, 2014; Segherloo et al., 2021). Among salmonids, marbled coloration traits are also present in (i) hybrids between S. marmoratus and domesticated stocks of North-European brown trout (Delling et al., 2000; Djurdjevič et al., 2019; S. trutta Linnaeus, 1758 sensu Segherloo et al., 2021), (ii) at least one population of S. trutta (Otra River, Norway; Delling, 2002), and (iii) hybrids between salmonid species of different genera (Miyazawa, Okamoto & Kondo, 2010). S. marmoratus includes two geographically distinct populations, in the Northern and Southern Adriatic regions (Delling et al., 2000; Polgar et al., 2022 and references therein). In the North Adriatic region, this species is distributed in the middle and lower tracts of the orographic left tributaries of the Po River, a few of the small orographic right tributaries of the Po River in the Southwestern Alps, and several large rivers flowing into the Adriatic Sea (Fig. 1; Sommani, 1960; Meraner & Gandolfi, 2018; Lobón-Cerviá et al., 2019; Splendiani et al., 2020; Merati, Pascale & Perosino, 2021). In the upper drainage of the Slovenian Soča River, in rhithral conditions, several small, genetically pure (Berrebi et al., 2000; Snoj et al., 2000; Sušnik Bajec et al., 2015) and extremely differentiated (Fumagalli et al., 2002) populations live in small karstic streams isolated by insurmountable waterfalls. These populations are periodically exposed to drastic demographic and genetic bottlenecks induced by stochastic and catastrophic flood events (e.g., Pujolar et al., 2016). Although earlier studies (e.g., Boniforti, 1870; Pavesi, 1873; Delpino, 1935), including its original description (Cuvier, 1829), recorded this species also in subalpine lakes, there is presently no scientifically sound documented record of lacustrine or lacustrine-adfluvial populations (Meraner & Gandolfi, 2018). It is not possible to know whether these records are native marble trout populations, straying stocked individuals, lacustrine morphs of non-native Salmo species, hybrids between native and non-native taxa with divergent behavioral patterns, or native and extinct Salmo species (e.g., D’Ancona & Merlo, 1959; Behnke, 1972; Giuffra, Guyomard & Forneris, 1996). Confounding factors include the pervasive and frequent translocations that have occurred in this territory in historical times (e.g., Delpino, 1935), the lack of genetic data, and the strong convergent traits of lacustrine and lacustrine-adfluvial Salmo species (e.g., Snoj et al., 2010).

Scientific investigations of marble trout populations in the North Adriatic region have been historically affected by the presence of anthropogenic introgressive hybridization between the marble trout and the non-native and interfertile Atlantic brown trout S. trutta, introduced by stocking activities (Splendiani et al., 2016, 2019). Phenotypic hybrids between S. trutta and S. marmoratus were recorded since the early 20^th century (e.g., Delpino, 1935; Pomini, 1940; Sommani, 1948). Subsequent molecular studies confirmed the presence of hybridization between these two species in the wild (Giuffra, Guyomard & Forneris, 1996; Lucarda et al., 2000; Zerunian, 2003; Jug, Berrebi & Snoj, 2005; Turin, Zanetti & Bilò, 2006; Lucarda, 2007; Meldgaard et al., 2007; Meraner et al., 2010; Meraner & Gandolfi, 2018). It is still unclear whether these hybrids are adaptively advantaged or disadvantaged (positive or negative heterosis, respectively) either in sympatry or allopatry with ‘genetically pure’ marble trout, though possible ecological differences and interactions between genetically discriminated hybrids and ‘pure’ individuals have been explored (Meldgaard et al., 2007; Meraner et al., 2010; Simčič, Jesenšek & Brancelj, 2017). Several ecological studies, including some early investigations on population density, size-age structure, and fishing pressure, phenotypically discriminated between pure and hybrid individuals within populations, analyzing them separately (Alessio et al., 1990; Jelli & Duchi, 1990; Jelli, Alessio & Duchi, 1996; Turin, 2000). However, such analyses are scarcely informative, due to the substantial overlap between the coloration pattern of S. trutta × S. marmoratus hybrids and that of both parental species, which prevents from phenotypically discriminating hybrids from ‘genetically pure’ marble trout individuals, as evidenced by morphological, genetic, and transcriptomic studies (Delling et al., 2000; Meraner et al., 2010; Chiesa et al., 2016; Djurdjevič et al., 2019; Gandolfi et al., 2020). Studies that phenotypically determined S. marmoratus without discriminating between pure and hybrid phenotypes offer a synthetic scenario of introgressed marble trout populations, though not discerning between possibly different ecological contributions of hybrid and non-hybrid individuals. However, none of these studies examined the size-age structure of a population (e.g., Lorenzoni et al., 2012; Marchi et al., 2017). When genetically pure wild populations were found (Soča River headwaters), some ecological studies examined them in situ, investigating density and age structure (Šumer, Leiner & Povž, 2001) or their trophic ecology, including interactions with non-native trout species (Vincenzi et al., 2011; Musseau et al., 2015, 2017, 2019). Other studies used hatchery-bred progeny from broodstock obtained from these populations to conduct field experiments and investigate plastic or adaptive density-dependent responses to catastrophic events and climate change (e.g., Crivelli et al., 2000; Vincenzi et al., 2007a, 2007b, 2010, 2012a, 2012b, 2014; Vincenzi, Jesensek & Crivelli, 2020), or to measure metabolic potential, respiratory rate, and ecological tolerance in the laboratory (Simčič, Jesenšec & Brancelj, 2015a, 2015b, 2017).

Within the North Adriatic region, in northern Italy, the marble trout is a subendemic species with substantial cultural, conservation and economic importance (Meraner & Gandolfi, 2018). This species was first included in the Annex II of the EU Habitats Directive (European Union, 1992), and then in an action plan for freshwater fish conservation, which classified the Italian populations as ‘IUCN Endangered’ (Zerunian, 2003). However, in spite of past efforts to facilitate a reduction of the angling pressure through fishing regulations and supporting breeding programs, which resulted in moderate numerical increases of the populations since the 1980s (Turin, Zanetti & Bilò, 2006; Lucarda, 2007), Italian populations have been subsequently classified as ‘Critically Endangered’ by the Italian IUCN Red List, predicting an 80% decline, due to introgressive hybridization with the Atlantic S. trutta and habitat degradation (Bianco et al., 2013). The massive introduction of S. trutta dramatically impacted marble trout populations in northern Italy, and no ‘genetically pure’ populations are known. However, panmixia has been observed in very few cases and genetically pure individuals are present in most populations (Meraner & Gandolfi, 2018), possibly due to a partial overlap of the spawning period of the two parental species and to local differences in the relative size of introduced non-native stocks and native populations (Meraner et al., 2010; Meraner & Gandolfi, 2018). Like many other salmonid species, Italian marble trout populations also suffer habitat destruction and degradation from damming, channelization, water withdrawal, hydropeaking events, extraction of riverbed materials, artificial embankments, degradation or destruction of the riparian vegetation, and climate change (Zerunian, 2003; Becciu & Dresti, 2015; Dresti et al., 2016; Saidi et al., 2014, 2018), with possible but yet uninvestigated synergistic effects on population structure.

In the VCO Province of Piedmont, within the subalpine catchment of Lake Maggiore (Fig. 1), the Toce River hosts an ecologically and economically important trout fishery, where Salmo marmoratus co-occurs with other stocked non-native salmonids, including non-native S. trutta. Recent investigations confirmed the presence of hybridization between S. marmoratus and S. trutta in the Toce River (Gibertoni et al., 2014). No other hybridization events between native and non-native species have yet been documented in this system. Furthermore, no stocks of S. trutta with marbled coloration traits and no hybrids between salmonid species of different genera with marbled coloration were found in this system. Therefore, it is reasonable to assume that trout with marbled coloration patterns here only include ‘genetically pure’ S. marmoratus and S. trutta × S. marmoratus hybrids or variably introgressed individuals. On the other hand, the absence of marbled coloration traits in introgressed individuals cannot be ruled out.

Here, we analyze trout specimens with marbled coloration elements of the anthropogenically introgressed marble trout population of the Toce River to provide novel baseline data on its size-age structure. To this aim, we: (1) estimate biomass density and numerical density; (2) estimate length-mass relationships, size-age structure and body growth trajectories, length at first maturity, length at maximum yield per recruit, mortality; and (3) illustrate the possible effects of present minimum-length harvest limits (MLL) regulations on the population’s length structure, relative to an alternative fishing strategy. We then discuss future research developments to address possible management options, with each option appealing to different conservation, societal and political priorities.

Materials and Methods

Fish density, length-mass relationship, length-age analysis, mean length at first maturity, optimum length, and mortality

A length-weight regression for the Toce River marble trout population was estimated using subsets of sample A (n = 338) and sample B (n = 113) with measured TL and W data (Tables 1, S1), using the following allometric model:

(1) $${\log }_{10}\left(W\right)=b\cdot {\log }_{10}\left(TL\right)+{\log }_{10}\left(a\right)$$

A subset of sample A was used to measure numerical density (ind. m⁻³) and biomass density (g m⁻³) per river tract, when georeferenced linear transects were available (31 surveys, n = 458; Table 1). Within this subset, since wet-mass data are not available for 124 individuals, Eq. (1) was used to impute missing data (Table S1).

Other subsets of samples A (n = 186) and B (n = 109) with available length measurements and determined age (Table S1) were used for length-age analyses. Age classes were defined using time intervals of 1 year, excluding the lower limit value of each interval (i.e., 0 < class 0 ≤ 1, 1 < class 1 ≤ 2, etc.). Age-specific growth trajectories were estimated from this dataset by fitting von Bertalanffy, Gompertz, and logistic growth models to the data (Nelson, 2019) (Eqs. (2)–(4), respectively):

(2) $$T{L}_t=T{L}_{inf}\left[1-e^{-K\left(t-t_0\right)}\right]$$

(3) $$T{L}_t=T{L}_{inf}\cdot e^{-e^{\left[-G\left(t-t_0\right)\right]}}$$

(4) $$T{L}_t=T{L}_{inf}/\left[1+e^{-G^{\mathrm{\prime }}\left(t-t_0\right)}\right]$$where TL_t is TL at age t; TL_inf is the fish asymptotic maximum length, i.e., “the mean length that the fish would reach if they were to grow indefinitely” (Froese & Binohlan, 2000); in Eq. (2), K is a growth coefficient and t₀ is the theoretical age when TL = 0. In Eqs. (3) and (4), G and G′ are growth-rate coefficients, and t₀ is time at the curve’s inflection point (Gompertz and logistic models are sigmoid-shaped), when instantaneous growth rate is maximum (Quist, Pegg & Devries, 2012; Tjørve & Tjørve, 2017). In the Gompertz model, the absolute maximum growth rate at the curve’s inflection point (Tjørve & Tjørve, 2017) was estimated as AMGR = (G·TL_inf)/e, where G is the Gompertz growth-rate coefficient and e = 2.71828 is Euler’s number. The best model was selected based on the Akaike information criterion corrected for small samples (AICc) (“Statistical analyses”). Model assumptions of homoskedasticity and normality of residuals of the selected model were evaluated using a residual plot and a QQ plot, respectively. Because of the absence of individuals in age classes 7 and 8 in the dataset, and the presence of only three individuals in age classes 9 and 10, the effect of the heterogeneous distribution of available length-age data on the growth-trajectory model was evaluated by performing a sensitivity test, i.e., by fitting the selected growth-trajectory function without the three larger individuals.

The mean length at first maturity (TL_m) was estimated using the TL_inf estimated by the selected growth model, according to the empirical relationship (Froese & Binohlan, 2000):

(5) $${\log }_{10}\left(T{L}_m\right)=0.8979\cdot {\log }_{10}\left(T{L}_{inf}\right)-0.0782$$

Optimum length, or the length at maximum yield per recruit, was defined as the length at which “the product of the number of surviving individuals multiplied by their average weight results in the highest biomass” (TL_opt; Froese & Binohlan, 2000). TL_opt is typically larger than TL_m, and often corresponds to the maximum female fecundity (Beverton, 1992; Froese, 2004). Froese (2004) identified three key risks of fishery management: (i) recruitment-overfishing, i.e., decreasing recruitment by catching fish before their first reproduction, (ii) growth-overfishing, i.e., catching fish before they can realize their growth potential, and (iii) reducing the number of ‘mega-spawners’, i.e., fish that disproportionally contribute to recruitment due to their large size, thus increasing the resilience of the population to random recruitment fluctuations. In order to mitigate these risks, Froese (2004) argued that the whole catch should only contain fish with TL = TL_opt ± 0.1 TL_opt. Optimum length was estimated using the estimate of TL_inf obtained from the selected growth model, according to the empirical relationship (Froese & Binohlan, 2000):

(6) $${\log }_{10}\left(T{L}_{opt}\right)=1.0421\cdot {\log }_{10}\left(T{L}_{inf}\right)-0.2742$$

The same dataset used to estimate age-specific growth trajectories (n = 295; Table S1; Fig. 2A) was used to estimate the population’s instantaneous total mortality rate (Z), using an unbiased Chapman–Robson estimator (Z_CR; Chapman & Robson, 1960; Hoenig, Lawing & Hoenig, 1983; Smith et al., 2012; Ogle, Wheeler & Dinno, 2021; “Statistical analyses”). Assuming that the fish catch is proportional to the size of the population, the total annual mortality rate was estimated from a catch curve obtained from cross-sectional data, plotting catch numbers of 2016–2020 against fish age class (catch-at-age). Vectors of cross-sectional data are equal to vectors of longitudinal data, when assuming that (i) the mortality rate is constant across time and among ages (constant mortality), (ii) recruitment is constant for each cohort or fluctuates randomly (constant or randomly fluctuating recruitment), and (iii) the probability of capture is constant across time and among ages (constant vulnerability). Catch-curve methods further assume that (i) the population is closed (no outputs except mortality and no inputs except recruitment), (ii) Z is constant across ages and years on the descending limb of the curve, and (iii) the sample is unbiased for any age class (Ogle, 2019; Ogle, Wheeler & Dinno, 2021). Within the limits of these assumptions, the Chapman–Robson method estimates the total annual survival rate S along the descending limb of the catch curve, not including the first fully-recruited age (peak of the curve, in our sample: age 2; Fig. 2A), i.e., including age classes 3–6 (n = 80; Fig. 2A). S is then used to obtain an unbiased estimate of instantaneous total mortality rate, i.e., Z_CR.

Z_CR values were then substituted for Z to calculate annual total mortality rates (A_Z) as (Ogle, 2019):

(7) $$A_Z=1-e^{-Z}$$

The instantaneous natural mortality rate (M) of the population was estimated by using the maximum recorded age of S. marmoratus (t_max = 11+ years, northern Italy; Jelli, Alessio & Duchi, 1996), using the empirical one-parameter t_max-based estimator (Then et al., 2015; “Statistical analyses”), as:

(8) $$M=5.109/t_{max}$$

The proportion of the population that suffers natural mortality in 1 year (A_M) was obtained from the M estimate as in Eq. (7).

The instantaneous fishing mortality rate (F) of the population was calculated as (e.g., Ogle, 2019):

(9) $$F=Z-M$$

The proportion of the population that suffers fishing mortality in 1 year (A_F) was obtained from F as in Eq. (7). The exploitation ratio E, i.e., the proportion of recruits that is fished during all the years of their life (Ricker, 1975) was computed as:

(10) $$E=F/Z$$

Assuming that fishing and natural mortality rates follow a constant or parallel yearly variability, E roughly assesses whether a stock is overexploited, based on the assumption that the optimal value of E (E_opt) is approximately 0.5, i.e., that sustainable yield is optimized when F = M (Gulland, 1971).

Results

The total sample includes 579 trout specimens with marbled coloration elements, likely including ‘genetically pure’ S. marmoratus, S. trutta × S. marmoratus hybrids, and variably introgressed individuals (samples A, n = 466; sample B, n = 113; Tables 1, S1; Fig. S1). Fish are significantly larger and older in the middle tract of the Toce River (TOR2; Fig. 3), however, large fish (≥80 cm TL) were found in all river sections (Table 1), and the local length record was caught in the upper Toce River, near the confluence with the Diveria Torrent (TOR3; Fig. S3). The measured numerical and biomass densities are higher in TOR2 than in TOR1 and TOR3, though not significantly so (Kruskal–Wallis test, p = 0.299 and p = 0.081, respectively; Fig. 4). Six individuals collected in the middle Toce river and the individual collected in Lake Maggiore (21.5–85 cm TL, 1.5–9.8 years) exhibited one or more traits compatible with convergent traits found in trout living in deep waters and pelagic habitats (e.g., Snoj et al., 2010), i.e., contracted marbled coloration patterns, silvery background color, pointed caudal-fin lobes (‘swallow-tail’), and intact ventral caudal lobes (e.g., Fig. S4).

The estimated log₁₀ length (TL) to log₁₀ body mass (W) relationship is (Table 2, Fig. S5):

(12) $${\log }_{10}\left(W\right)=3.055\cdot {\log }_{10}\left(TL\right)-2.100$$

The Gompertz growth function has the lowest AICc value among the models tested (Fig. S6; Tables S3, S4). A biologically unrealistic TL_inf was estimated using the Von Bertalanffy model (Table S3; Fig. S6). The TL_inf ~112 cm estimated without the three larger individuals (sensitivity test; Table S5; Fig. S7) is not larger than 10% of that one obtained including them (TL_inf ~ 105 cm, G ~ 0.2, and t₀ ~ 4 years; Table 2; Figs. 2B; S8), indicating that the presence of a gap in the dataset of age classes 7 and 8 and the presence of only three individuals in classes 9 and 10 did not significantly bias the analysis. The growth trajectory fits a sigmoid-shaped model, suggesting a gradually accelerating growth rate at earlier life-history stages, and a slight deceleration after an average age (t₀) of ~4 years (Quist, Pegg & Devries, 2012). The estimated AMGR is 8.33 cm year⁻¹.

The mean length at sexual maturity is TL_m ~ 55 cm and the optimum length is TL_opt ~ 68 cm (Table 2). The unbiased estimate of instantaneous total mortality rate is Z_CR ~ 0.9 and the total annual total mortality rate is A_Z ~ 0.6; i.e., an average of 60% of the population die annually (Table 2). The empirically estimated natural instantaneous mortality rate is M = 0.464. The total annual natural mortality rate is A_M = 0.372, i.e., ~40% of the population die naturally each year. The fishing instantaneous mortality rate is F = 0.460 and the total annual fishing mortality rate is A_F = 0.369, that is ~37% of the population are fished annually. The exploitation ratio is E = 0.498, indicating an optimized sustainable yield.

However, following the rationale by Froese (2004; TL_opt ± 0.1 TL_opt), the total catch should only include fishes ranging ~61–75 cm TL, in order to prevent overfishing and increase recruitment resilience. The frequency distribution of 1-cm TL classes of our total angling catch (sample B, n = 113; Table 1; Fig. 5) shows that 16–23% of the total catch would have been retained, according to the past and current regulations (minimum length limit MML for S. marmoratus in the VCO Province before and after 2019). Only one fish (4–6% of the retained catch) has a size (68 cm TL) within the optimum length interval proposed by Froese (2004). Outside this interval, only two retained fish (8–11%) are mature individuals that could have reproduced before capture, while 14–22 retained fish (78–85%) are immature individuals that likely never reproduced. The only captured megaspawner (85 cm TL) would have been retained (Fig. 5).

Discussion

The marble trout is a species of high conservation interest (Bianco et al., 2013). Despite being widely distributed in the southern Alpine region, where it is typically found in large subalpine rivers and tributaries, the published information on the structure of these populations is fragmentary.

Some comparisons with published data on population density can be made. The Toce River marble trout population is 4–7 times denser in terms of number of individuals and more variable in terms of biomass (from five times less dense to 50 times denser in different sites) than that of the middle tract (15 km) of the Brenta River (Southeastern Alps; Turin, 2000). The Slovenian populations (Southeastern Alps) live in small and fast-flowing karstic headwaters, and are characterized by individuals with smaller average size, shorter lifespan, and earlier sexual maturity. Relative to them, the Toce River population is on average less dense in terms of number of individuals, but denser in terms of biomass (Šumer, Leiner & Povž, 2001). The higher average density, significantly larger size, and older age of the trout collected in the middle tract of the Toce River might indicate that this system offers more optimal conditions for growth, or might suggest the presence of ontogenetic migration patterns among the different river tracts, e.g., a tendency of moving to the middle river tract during growth. However, it should be noted that the relatively low density in the lower tract of the Toce River might be an effect of lower electrofishing and rod-and-line sampling efficiency in deep waters. In order to avoid similar sampling biases potentially caused by the low sampling efficiency in deeper waters, some previous studies explicitly limited populations studies to selected smaller and shallower tributaries of large rivers (e.g., Crivelli et al., 2000).

Log-log regressions between length and body mass are largely consistent with those previously estimated for the Brenta River population (Turin, 2000), and for both Slovenian and northern Italian populations (Lorenzoni et al., 2012). The growth trajectory model estimates a maximum growth rate of 8.2 cm decy⁻¹ at 3–5 years (inflection point), consistent with their dietary shift to a piscivory diet, at ~40 cm TL (Turin, 2000). At 3–5 years of age, the estimated fish sizes are consistent with those of the High-Po (Southwestern Alps), Adige, Piave, Avisio (Jelli, Alessio & Duchi, 1996) and Brenta Rivers’ populations (Central-eastern and Southeastern Alps; Turin, 2000). The asymptotic length estimated by the selected growth model is biologically compatible with maximum size records (139 cm TL~ 120 cm standard length (SL), Rienz River (Delling, 2002); SL/TL = 0.864 ± 0.015, n = 14). However, comparisons with growth models based on different populations should be avoided, since estimated parameters depend on local ecological conditions, such as temperature, food availability, current patterns, and local phenotypic adaptation (e.g., Vincenzi et al., 2007a).

Although the exploitation ratio suggests that the Toce River population is sustainably exploited, F = 0.91M is from 5% to 36% larger than optimal fishing mortalities estimated for teleosts (F_opt = 0.67–0.87; Zhou et al., 2012; Lester et al., 2014). Following the rationale by Froese (2004), the length-frequency distribution of our angling sample and estimated TL_opt and TL_m suggest that the present regulations (MLL) exacerbate risks of overfishing and loss in resilience to recruitment fluctuations. Most of the captured fish were also immature individuals, which may be either wild-born offspring or stocked fish. However, the length frequency distribution also shows a conspicuous lack of mature individuals larger than MLL, suggesting a significant effect of fishing pressure on the population structure. Resilience to recruitment fluctuations might be improved by implementing new regulations allowing to retain only fish of ~60–70 cm TL, thus both preventing overfishing risks and preserving ‘mega-spawners’. Harvesting regulations based on length interval limits are referred to as ‘harvest-slot length limits’ (HS). In the observed scenario, HS regulations would likely promote more balanced population size structures, higher biomass, density, and fecundity (Ayllón et al., 2019), and are expected to gradually increase the frequency of larger reproductive individuals in the population. However, changing the present MLL regulations to the proposed HS (60–70 cm TL) would have the immediate effect of drastically decreasing the harvestable catch. In our angling sample, this would correspond to a decrease from ~19% (n = 21) to ~1% (n = 1) of the total catch. Therefore, these actions could be only implemented with the empathetic support of the anglers and sporting fishing associations.

No genetically ‘pure’ populations have been found in northern Italy, contrary to those living in the drastically different Slovenian karstic ecosystems (e.g., Šumer, Leiner & Povž, 2001). Studies that phenotypically discriminated between ‘pure’ and hybrid individuals (e.g., Turin, 2000) are unreliable, as previously explained. On the other hand, sampling an introgressed marble trout population by selecting individuals with marbled elements in their coloration patterns may leave out hybrids or variably introgressed individuals without marbled coloration elements. No study in this region described the size-age structure of a marble trout population by analyzing the individual level of genetic introgression. These limitations of this and previous studies should be addressed by future investigations.

The combined effects of climate change and anthropogenic genetic introgression might challenge the sustainable management of marble trout populations, in order to face the contrasting goals of (i) favoring predicted upriver migrations induced by global warming, e.g., by creating ecological corridors, and (ii) preventing genetic mixing between subpopulations with different levels of genetic introgression presently separated by artificial hydrological barriers (e.g., Splendiani et al., 2019). Furthermore, although there is evidence that pure and hybrid individuals might have distinct ecological and physiological traits (e.g., Meraner et al., 2010; Simčič, Jesenšec & Brancelj, 2015a, 2015b, 2017), such studies were either conducted on artificial populations, or on extremely inbred and small populations living in karstic headwaters, and are also not comparable with our results.

Typically admixed and sympatric ecophenotypes with different life histories, not uncommon in salmonids (e.g., Charles et al., 2005; Snoj et al., 2010), could have divergent ecology and behavior, with distinct population size-age structures. The presence of some individuals exhibiting traits compatible with pelagic morphs might suggest the presence of different ecophenotypes in the Toce River, such as lacustrine, fluvial-adfluvial, or lacustrine-adfluvial morphs. However, further genetic and ecomorphological studies are needed to confirm this hypothesis, since the only available study lacks experimental evidence of migratory patterns (Gibertoni et al., 2014).

Conclusions

The description of density distribution, length-weight relationship, and length-age structure, and the estimates of growth trajectory, mean length at first maturity, optimum length, and mortality provide a first overview of the structure and dynamics of the Toce River marble trout introgressed population. The overlap of these estimates on the length-frequency distribution of our angling sample highlights the presence of overfishing risks posed by the current fishing regulations.

Future research may focus on (i) conducting genetic analyses on all sampled individuals, thus evaluating the effects of genetic introgression on population size-age structure; (ii) improving the sampling design to obtain more spatiotemporally balanced datasets, e.g., sampling periodically in a smaller number of selected sites; (iii) using river sections that allow a higher and more homogeneous fishing efficiency in shallow waters, as a genetic and ecological proxy of the larger system; (iv) tracking fish movements in the river and lake systems, investigating possible ontogenetic shifts, individual patterns, and presence of ecophenotypes with different behaviors and ecology; and (v) quantifying the relative effects of harvesting larger fish and stocking smaller fish on the population structure. In this latter case, measurements of the genetic distance between smaller fish in the wild population and in hatchery stocks might preliminarily demonstrate the actual contribution of stocking to the wild river population.

The long-term sustainability of inland recreational fisheries is a priority, if shared economic, research, and conservation-oriented goals are to be achieved (Cooke et al., 2015). Management strategies of such socio-ecological systems must merge the priorities and needs of all the stakeholders, from conservation scientists to anglers, if the common goal of bequeathing the structure and function of these valuable ecosystems to future generations is to be achieved.

Supplemental Information