Reservoirs facilitate colonization of river catchments by a native invasive fish through provision of pelagic larval rearing habitat

Dams on rivers are known to facilitate the colonisation and spread of aquatic alien and native invasive species, but the actual mechanisms involved are poorly understood. Since the construction of the Solina Dam on the upper San River system in Poland, European perch (Perca fluviatilis) have expanded their distribution into the headwaters of this river system, becoming a native invader. In this study, we assessed the spread of perch in detail over time upstream of the Solina Reservoir, and used otolith trace element microchemistry to determine the spawning and larval rearing locations of perch in the catchment upstream of the dam. Extensive sampling over several years across the catchment upstream of the Solina Reservoir confirmed the widespread occurrence of perch into the headwaters of the tributary river systems, with smaller size classes dominating locations closer to the Solina Reservoir. Despite perch being widely distributed upstream of the Solina Reservoir, otolith microchemical analysis indicated the populations from various reservoir tributaries mostly shared the same spawning and larval rearing habitat, most likely the Solina Reservoir. Our results suggest that reservoirs can facilitate the colonisation of river systems by providing a critical habitat element that would be otherwise missing from riverine landscapes, i.e., an extensive and productive pelagic larval rearing environment. This research shows that the impacts of large dams can extend many kilometers upstream from the river reaches directly affected by the resulting impoundment.


Introduction
The construction of a dam on a river invariably has dramatic impacts on the ecology of fish communities, disrupting habitat for some species, but also creating opportunities for others (Johnson et al. 2008;Głowacki and Penczak 2013;Liew et al. 2016;Xie et al. 2018;Turgeon et al. 2019). Dams can form significant barriers to migration, preventing the natural movement of fish along rivers, often resulting in the extirpation of migratory species (Liermann et al. 2012;Turgeon et al. 2019;Sumizaki et al. 2019). Marked changes to the hydrodynamics of rivers occur, with altered patterns of discharge downstream, but also with the creation of new lentic environments upstream (Johnson et al. 2008;Jellyman and Harding 2012).
The impacts of impoundments may extend well upstream from the actual dam, including some effects that extend beyond the river reaches directly affected by impoundment (Martinez et al. 1994;Hladík et al. 2008;Jellyman and Harding 2012;Głowacki and Penczak 2013). The loss of longitudinal connectivity associated with dams may lead to the significant loss of migratory species into upstream river reaches (Penczak et al. 1998;Han et al. 2007;Sumizaki et al. 2019). Additionally, the lakes formed upstream of dams can act as focal points for range expansion and invasion by both native and exotic species (Han et al. 2007;Grabowska et al. 2011;Johnson et al. 2008;Jellyman and Harding 2012). This can potentially extend the distribution of species that are able to use the impounded habitat further upstream into river reaches they would otherwise be unable to colonise (Reid 2004;Han et al. 2007;Hladík et al. 2008;Song et al. 2016). Whilst the varied impacts of dams on riverine fish communities are well recognised (Johnson et al. 2008;Jellyman and Harding 2012;Liew et al. 2016;Xie et al. 2018;Turgeon et al. 2019), the mechanisms that actually drive all these changes (particularly those that extend upstream) are less understood, with changes variously attributed to alteration in the hydrology and temperature regimes, and the creation of suitable habitat for key life history stages (Johnson et al. 2008;Głowacki and Penczak 2013;Liew et al. 2016;Xie et al. 2018;Turgeon et al. 2019). Knowledge of the mechanisms that facilitate invasions by opportunistic species remains limited, particularly with respect to the early life histories of invasive species (Grabowska et al. 2011;Sajdlová et al. 2018).
Small pelagic larvae are a life history feature of many fish species (Winemillar and Rose 1993;Augspurger et al. 2017), enabling high levels of fecundity relative to maternal body size (Smith and Fretwell 1974). A critical landscape element that must be present to complete the life cycle of species with tiny pelagic larvae is a productive pelagic larval habitat within the wider catchment landscape, often a lake or pond (see Bylak et al. 2014;Augspurger et al. 2017). Pelagic habitat can also provide a refuge from visual predators for early life history stages (Wang and Eckmann 1994;Urho 1996;Vejřík et al. 2016). High fecundity, leading to high rates of recruitment into adult populations, is recognised as a key trait of many invasive fish species, potentially enabling them to achieve ecological dominance in the habitats they colonise (Liu et al. 2016;Augspurger et al. 2017).
The construction of an impoundment in an otherwise fluvial landscape can create the ideal pelagic habitat required for larval rearing and recruitment into adult populations (Martinez et al. 1994). Whereas adult fish often have relatively broad environmental and ecological tolerances, and can thus range widely across a catchment, small pelagic larvae are often fragile, with a limited capacity to tolerate the harsh physical conditions that may be experienced in a river (Schiemer et al. 2003;Closs et al. 2013;Jones and Closs 2016). With only limited endogenous energy resources, small larvae have limited tolerance to starvation, and hence are highly dependent on continuous access to the abundant, small and preferred planktonic prey typically found only in productive and stable pelagic environments (Bremigan and Stein 1994;Graeb et al. 2004;Tonkin et al. 2006). If an impoundment provides suitable larval rearing habitat, large numbers of pelagic larvae may survive, and the resulting post-larval fish can subsequently range more widely in both downstream and upstream directions (Martinez et al. 1994;Penczak et al. 1998;Hladík et al. 2008;Vašek et al. 2008).
Under certain circumstances, human alteration of an ecosystem may allow native species to become invasive (Carey et al. 2012). Although the term "invader" is typically paired with adjectives such as "non-native" and "alien", certain native species can also cause ecological impacts that rival those of wellknown invasive species (Valéry et al. 2008). Therefore, the term "native invader" (Simberloff 2011) or neonative species (Essl et al. 2015) has been proposed. In this paper we use the term native invader because it accurately reflects the nature of the ecological impact of such a species. Identifying when, where, and why species become invaders within their native ranges requires additional scientific inquiry (Carey et al. 2012).
The European perch (Perca fluviatilis L.; hereafter perch) is a widely distributed freshwater fish species of Europe and Northern Asia, often comprising a dominant and important element of the aquatic fauna, particularly in lowland rivers and lentic habitats, with adults also occurring in faster flowing streams (Kottelat and Freyhof 2007). Perch can also be invasive, having been successfully introduced to Australia, New Zealand and South Africa (Thorpe 1977). Adult perch can tolerate a broad range of physico-chemical conditions (Rask 1984;Christensen et al. 2019) and habitats (Mehner et al. 1995), however, the pelagic larvae of perch rear best in stable, relatively warm conditions (12-20 °C), and require access to high densities of small zooplankton soon after hatching (Treasurer 1988;Wang and Eckmann 1994). Perch are therefore more common in lakes and reservoirs than rivers (Irz et al. 2006), but are also often observed to increase in abundance and range across catchments following the construction of an impoundment within a river system (Penczak et al. 1998;Kukuła 2006;Hladík et al. 2008;Głowacki and Penczak 2013). Hence, they might be considered to be an invasive native species under certain circumstances (Valéry et al. 2009;Carey et al. 2012). The larvae of European perch may also be very abundant in the pelagic habitat of impoundments (Matena 1995;Matveev et al. 2002;Čech et al. 2007a, b), suggesting that such habitats can provide conditions that facilitate larval survival and enable juvenile recruitment into the adult populations located elsewhere in waterways downstream and upstream of the impoundment (Penczak et al. 1998;Kukuła 2006;Hladík et al. 2008).
In Poland, perch appear to have been a major beneficiary of the construction of the Solina Dam, and the extensive steep-sided Solina Reservoir that was created as a result of the dam (Kukuła 2006). Perch were first recorded from the upper San River catchment in small numbers following the construction of the Myczkowce Reservoir in 1960 just downstream of the present-day Solina Dam (Rolik 1971;Wajdowicz 1979). Following the construction of the Solina Dam in the late 1960s, perch increased in abundance and began expanding their distribution upstream (Kukuła 2006). The range of perch upstream from the reservoir is now limited by other natural and artificial barriers to migration, suggesting perch populations upstream of the reservoir are either wholly or partly sustained by individuals rearing and then migrating upstream from the Solina Reservoir (Kukuła 2006). Of specific concern is the expansion of perch into the streams of the Bieszczady National Park, threatening the ecological integrity of the aquatic ecosystems in this internationally significant reserve (Kukuła 2006;Taggart-Hodge and Schoon 2016), which is part of the Natura 2000 network (Council of the European Communities 1992).
In this study, we test the hypothesis that the creation of a potentially extensive pelagic larval rearing habitat (Solina Reservoir) has enhanced the survival and recruitment of perch larvae into post-larval populations, thus contributing to range expansion and increased abundance of perch across the wider catchment of the upper San River system (see Penczak et al. 1998;Kukuła 2003Kukuła , 2006Hladík et al. 2008). We analysed available data on the distribution and abundance of perch upstream of the Solina Dam, using records collected at various times from prior to the construction of the dam to 2018. We also analysed the otolith microchemistry of perch collected from various populations upstream of the Solina Dam to determine whether populations distributed across the upper catchment are sustained by recruits from a single (i.e., Solina Dam) or from multiple (tributary) larval rearing sources (see Fig. 1). Subtle differences in otolith microchemistry can provide a robust approach for assaying stock structure in widely dispersed fish meta-populations (see Elsdon and Gillanders 2003;Warburton et al. 2018). We compared larval and postlarval otolith microchemical signatures (see Warburton et al. 2018) from perch collected from multiple sites located both in and upstream of the Solina Reservoir. Otolith microchemistry reflects the ambient physiochemical environment in which fish were living at the time of formation and deposition of otolith material (Elsdon and Gillanders 2003). Hence, our specific tested predictions were: (i) The construction of the Solina Dam and associated reservoir created ideal conditions for perch spawning and larval rearing, leading to the widespread colonisation and persistence of perch through much of the upper San river catchment, a montane landscape that would otherwise be largely inhospitable for the early life history stages of perch. (ii) If perch larvae are rearing in the reservoir and migrating upstream as juveniles, then larval trace element signatures in otoliths will be similar (reflecting rearing in the Solina Dam), but the adult trace element otolith signatures will be site-specific (reflecting the ambient conditions of the location in which they were caught). Perch that remained in Solina Dam over their entire life history will have relatively similar larval and post-larval otolith trace element signatures. (iii) Alternately, if larval and adult otolith trace element signatures from across the catchment are relatively similar and site-specific, then postlarval migration is not occurring, with fish more or less remaining resident in the reach in which they were spawned.

Study area
The  (Hennig et al. 1991). The Solina Reservoir is a canyon-shaped submontane reservoir (mean water level is 420 m a.s.l.) The basin of the reservoir is 60% forested. The shores of the Solina Reservoir are very steep and the littoral zone is very narrow. Due to its steep shoreline more than 90% of the lake can be classified as   (Kukuła 2006).
Streams analysed in the present study flow in channels composed of Carpathian flysch. In the streams of the Flysch Carpathians, the water conductivity is ~ 200 μS cm −1 , due to the geology of the area. The Flysch Carpathians consist of Cretaceous flysch and Palaeocene deposits, i.e., sedimentary rocks (turbidite) (Săndulescu and Visarion 1988). These streams have well-oxygenated water (8.2-10.9 mg L −1 ) and mean summer temperatures typically range from 17 to 20 °C.
Perch were sampled from multiple river sites grouped within the 6 main hydromorphological regions upstream of the Solina Reservoir; i.e. groups based on sub-catchment, channel width, water depth and substrate (Rzonca and Siwek 2011, Fig. 1). Each abbreviation for regions consists of a letter and a number-'S' denotes regions in the San River subcatchment, 'W' denotes regions in the Wołosaty subcatchment area, and 'K' denotes the region in the Solinka River sub-catchment area. Numbers mean a distance from the Solina Reservoir, i.e., regions more distant have higher numbers. Region S3 comprised five sites in the upper San River, characterized by a flat valley with the character of a highland stream, 6-8 m wide. Substrates comprised stony sections with faster flowing water, alternating with gravelsandy sections. Region S2 comprised three sites in the middle course of upper San River, where the riverbed has a width of 20-40 m. Region S1 included three sites at the lower part of the upper San River, where the riverbed was 30-60 m wide. Both the middle and lower sections of the upper San have the character of a mountain river, with a diverse riverbed, and alternating sections of fast-flowing runs and slower flowing pools. The riverbed in these sections is wide and the water depth is relatively shallow, with many transverse rocky bars. Region W2 consisted of three sites in the upper part of the Wołosaty stream (tributary of the San River), where the riverbed is 5-8 m wide, and the stream has a mountain character. Region W1 comprised three sites along the lower reaches of the Wołosaty stream, where the stream is wider (10-20 m) than the upper section. Region K1 included three sites from the Solinka River. Here the riverbed has a width of 15-30 m, the bottom is mostly rocky, and comprised of varying water velocities. A single site was located in the Czarny Stream, which was characterized by relatively shallow water, a riverbed width of 6-8 m, and substrates dominated by pebbles-cobbles.

Field sampling
Assessments of the distribution of perch in the upper San River catchment were based on records from 1965 (Rolik 1971), electrofishing data from 1980 and 2000 (J.M. Włodek, K. Kukuła unpublished data), and regular electrofishing data from 2009 to 2018. Electrofishing surveys in streams of the Solina Reservoir basin were conducted to assess changes in perch abundance and population size structure at 21 sites variously grouped within the six hydromorphological regions (Fig. 1). Surveys were conducted once in late summer / early autumn (August-October), for seven years in the following periods: 2009-2012, 2014-2015 and 2018. Following initial habitat assessments, single-pass electrofishing was used to sample fish. Electrofishing was performed with a battery-powered backpack electroshocker (Hans Grassl GmbH IG600; Hans Grassl GmbH, Schonau am Konigssee, Germany) set at 3 A and 140-200 V. All captured fish were identified, counted and then released at the point of capture. All perch were measured for total length (Tl, accuracy of 1 mm), and counted to calculate their abundance per 100 m 2 of the site surface area (i.e., individuals per 100 m 2 ).

River discharge
Data available from the Institute of Meteorology and Water Management in Poland (IMGW 2020) was used to assess the influence of discharge on perch recruitment and abundance. The average San River discharge for each of the studied years was calculated using the available data for the measurement point located in the hydromorphological region S2. The calculations include daily measurement data from July to September, typically the time when perch appear in the tributaries of the Solina Reservoir (Kukuła 2006).

Perch otolith samples
In 2017 and 2018, from August to October, perch were collected in sufficient numbers for meaningful otolith extraction and analysis from the hydromorphological reaches S3, S1, K1, W1 and the Czarny Stream in 2017 (Fig. 1), and S2, S1, K1, W2 and Czarny Stream in 2018. Otoliths were also prepared from perch caught by angling in the Solina Reservoir from four sites (Sites KRI and KRII in the flooded Solinka River valley, and Site SRI and SRII in the flooded San River valley). At least fifteen perch from each site in the hydromorphological region, belonging to different size classes, were anesthetized and then euthanised with Propiscine. Propiscine is a lowtoxicity, safe and very effective product (Kazuń and Siwicki 2012), recommended by the Polish Institute of Inland Fisheries. We were aiming for a sample of ten analysed otoliths per sites to provide a reasonable level of statistical power to compare microchemical signatures across sites, and allowing for the possibility of some failed otolith preparations (i.e., loss or cracking of otoliths during otolith extraction or preparation), although this was not always achieved (Warburton et al. 2018 Otolith microchemistry Sagittal otoliths of P. fluviatilis were cleaned of any remaining tissue, washed with ultra-pure water, air dried, and stored in micro-centrifuge tubes. As final preparation, otolith surfaces were again cleaned in ultra-pure water, allowed to air dry, and then mounted sulcus-side up on a standard glass microscope slide with Crystalbond 509 thermoplastic. Each otolith was polished to expose their cores with sandpaper and lapping film of decreasing grain size. After polishing, otoliths were washed in ultra-pure water, air dried, and randomly mounted on a gridded microscope slide using double-sided adhesive tape. All utensils used during preparation and transfer were washed with ultra-pure water and dried with lint-free wipes between each otolith. Twelve isotopes commonly found in otoliths, including 43 Ca, were measured using LA-ICP-MS: 7 Li, 23 Na, 24&25 Mg, 31 P, 39 K, 55 Mn, 85 Rb, 86 Sr, 137&138 Ba. Data were collected on an Agilent 7900 ICP-MS coupled to an ASI RESOlution M-50 laser ablation system powered by a Coherent 193 nm ArF excimer laser located in the Centre for Trace Element Analysis at the University of Otago, Dunedin, New Zealand. Otolith mounts were placed in the sampling cell and visually located via a 400 × magnification video imaging system. The laser was fired on an ablation path drawn across the core to the outer edge of the otolith with a pre-ablation spot size of 75 μm, ensuring that uncontaminated otolith material was sampled, and a final spot size of 50 μm and scan speed of 10 μm sec −1 . The laser repetition rate was set at 10 Hz and the on-sample fluence was 2.5 J cm −2 . Ablation occurred in an atmosphere of pure helium to minimise re-condensation of ablated materials and potential elemental fractionation. Software-controlled gas flows of helium and nitrogen and ICP-MScontrolled argon carrier gas were tuned to maximise signal intensities on 138 Ba and 88 Sr while minimising oxides and mass fractionation by monitoring ablation of the NIST (National Institute of Standards and Technology) SRM (Standard Reference Material) 612 certified reference calibration glass. External standards (NIST 610, NIST 612 and MACS-3) were run every 6-8 samples to allow correction for any instrument drift and bias, and background signals were collected for 20 s between samples to establish a baseline and account for possible machine drift. Standard values for NIST were obtained from Jochum et al. (2011) and the standard value for MACS-3 was obtained from Chen et al. (2011).

Population analysis
Perch abundance was estimated from the data collected from multiple sites within each hydromorphological group and across multiple years. PER-MANOVA with pairwise comparisons was used to test if there were significant differences in perch abundance between hydromorphological regions. The fish abundance data were square-root transformed. Because of the large number of zeros in the data, analyses between samples were made using a zero-adjusted Bray-Curtis coefficient, including a virtual dummy species for all samples, prior to computing similarities (Anderson et al. 2008). A similarity matrix based on Bray-Curtis distances was built, and a permutational multivariate analysis of variance (PERMANOVA) with two factors: Region (fixed with 6 levels), and Year (fixed with 7 levels), was used to analyse the variation in perch abundance. We did not include the data from the Czarny Stream as electrofishing in this stream was carried out only three times during the study period.
Differences between the mean abundance of perch and mean total length of the perch across the hydromorphological regions (S1-S3, W1, W2, K1) were analysed using a non-parametric ANOVA (Kruskal-Wallis test), and post hoc tests for Kruskal-Wallis ANOVA (Dunn's test).
Additionally, for each region, separate analyses were performed (Kruskal-Wallis test) comparing the abundance of perch and the total length of perch in particular years (2009,2010,2011,2012,2014,2015,2018). Prior to statistical analysis, a Brown-Forsythe test was used to evaluate variance homogeneity, and a Shapiro-Wilk test to test the normality of the logtransformed data (Zar 2010). A Kruskal-Wallis test was also used to analyse the differences between the mean summer discharge in the San River in the years when perch studies were conducted. Statistical analyses of population data were completed using STA-TISTICA 13 (TIBCO Software Inc., Palo Alto, CA, USA), and multivariate analyses using PRIMER v7 (Anderson, et al. 2008).
Otolith microchemistry Data were processed using the Trace_Elements_IS data reduction scheme in IOLITE software version 3.71 (Paton et al. 2011) run on Igor Pro 7, with 43 Ca set as the internal standard, resulting in elemental data being expressed as a molar ratio of mols element/mols Ca. NIST 610 was used as the calibration standard, while NIST 612 (Jochum et al. 2011) andMACS-3 (Chen et al. 2011) were used for calibration verification and a matrix matched reference material. The results for the selected elements were within the published values of NIST 610, 612 and MACS-3, respectively, and all of the measurements were above the instrument's level of detection.
To interpret the otolith signatures, the otolith core was identified by using a spike in 55 Mn (Ruttenberg et al. 2005) as well as a peak in 138 Ba, which were cross validated by inspecting the transect distance at the core. In order for otoliths to be included in the analysis, each transect was assessed for a complete uninterrupted trace from the core to the edge; this resulted in final sample sizes of n = 76 in 2017 (S3, S1, K1, W1, Czarny Stream, and sites KRI, KRII, SRII in Solina Reservoir), and n = 123 in 2018 (S2, S1, K1, W2, Czarny Stream, and sites in the Solina Reservoir (KRI, KRII, SRI, SRII). The section of the otolith selected to represent the 'larval / early juvenile' was a 50 μm section, 50 μm from the core, which represents a period of early life history excluding the period of yolk-sac consumption, and thus reducing the likelihood of incorporating maternal effects into the analysis (see Kristensen et al. 2008). A 50 μm edge section, from the end of the otolith ablation at the edge of the otolith and in towards the core, was used from each transect to represent the otolith trace element signature at and close to the time of 'capture'. Element concentrations from each of these two periods were averaged to smooth out fine scale variation, with the average being used as a single value for statistical analysis.
Data analysis was performed in Paleontological Statistics (PAST) version 3.20 (Hammer et al. 2001) and R version 3.5.1 (R Core Team 2018). Data were log 10 -transformed to meet assumptions of normality prior to conducting any statistical tests. Boxplots for each element were used to assist in determining the appropriate suite of elements to use in a linear discriminant analysis (LDA). Five elements ( 7 Li, 25 Mg, 85 Rb, 86 Sr, 138 Ba) were selected for the LDA based upon the inspection of boxplots, previous water chemistry analysis and on the likelihood of the element's stability in the otolith (Campana 1999). An LDA was run on the above element suite to examine the clustering of otolith microchemical concentrations to test for patterns or structuring within the larval and capture segments. Classification success was used to assess how accurately the linear discriminant analysis could assign individuals back to their site of collection based on their otolith microchemical analyses.

Qualitative changes in perch distribution from 1960s to 2018
Prior to the construction of the impoundments on the San River, the larger watercourses of this catchment (San, Solinka) supported an ichthyofauna comprising species typical of Carpathian rivers, with numerous spotted barbel, common minnow, chub, stone loach and gudgeon (Gobio gobio (L. 1758)) (Rolik 1971). Following the construction of the small Myczkowce Reservoir, just downstream of the present-day site of the Solina Reservoir (Fig. 2), the first specimens of perch were recorded in the 1960s in the upper San River basin (Indicated by diamond in Fig. 2). The abundance and range of perch increased following the completion of the Solina Reservoir in 1969, and by 1980 the range of perch included San River to the mouth of the Wołosaty Stream, the lower and middle Solinka River, and the Wetlina (Kukuła 2003 ;  Fig. 2). Some 20 years later, the upper boundary of perch occurrence in the Solinka River moved a few kilometers upstream (Kukuła K., unpubl. data). In the Solinka tributary, the Wetlina Stream, perch initially (1980) extended their range approximately 10 km upstream of the confluence with the Solinka River ( Fig. 2 and cf. Figure 1). Subsequently, perch were lost from the Wetlina Stream, most likely due to a barrier to upstream migration caused by newly created natural waterfalls that were created following a rockslide into the stream that occurred in 1980 near the confluence with the Solinka River ( Fig. 2; Dzi uban 1980). In the San River, the upper boundary of perch occurrence shifted upstream by almost 40 km, with perch also appearing in the mountain tributary of the San River, the Wołosaty stream (13 km upstream from the mouth to San River). The maximum range of perch in recent years (2017-2018) covered the upper course of the San River, to an altitude of 785 m above sea level (~ 5 km from the sources of the San), in Wołosaty Stream up to the site located only 7 km downstream from the source of the Wołosaty Stream, as well as the lower and middle part of the Solinka River and the lower section of the Czarny Stream (Fig. 2). In the early 1990s, perch were recorded for the first time in the mountain streams of the Bieszczady National Park. In 2018, perch was recorded just ~ 5 km downstream of the headwaters of the San River.

Changes in perch distribution and abundance 2009-2018
The results of the PERMANOVA analysis of perch densities across the hydromorphological regions by time indicated that the Region and Year, as well as the interaction terms were significant (Table 1). The pairwise tests indicated that perch densities in S3 differed significantly from S2, S1, and K1 regions; S2 differed significantly from S1 and W2; S1 differed significantly from K1, W1, and W2; and K1 differed from W2 (Table 2). Based on data collected between 2009 and 2018, the lowest average perch abundance was found in regions S3 and S2 (upper reaches of the San River) and amounted to approximately 2.5 ind. 100 m −2 (Fig. 3). The highest average abundance of perch was found in the region S1 (lower San River)-5.5 ind. 100 m −2 (Fig. 3a). Perch abundance changed over time, with the highest abundances generally occurring at sites nearer the reservoir in 2015 and 2018 when very low discharge was recorded (Fig. S1), and greater variation in abundance further upstream (Fig. 4). Generally, perch abundance did not differ significantly across individual years in the S1 region, and also partly in S2 where only 2018 had significantly higher densities (Fig. 4).
The mean total lengths of perch across the hydrogeomorphic regions were significantly different (Fig. 3b). Over 60% of all perch caught had a total length (TL) between 10 and 15 cm, while more than 32% did not exceed 10 cm in total length. Only six individuals were longer than 20 cm, and all of them were caught in region S2. The lowest average lengths (Tl = 7.2 cm) were found at the sites from region S1 close to the Solina Reservoir, while the highest average total length of perch was in region S2 (Fig. 3b), in the middle reaches of the San River. In the S1 region in 2009, 2010, 2011 and 2014, individuals with a total length of ~ 6 cm dominated among the perch caught (Fig. 5a). Perch 10-12 cm in total length dominated the middle (S2) reach of the San River. In the  Rolik (1971) S3 region, most of the perch caught in were 10-14 cm in total length. In the W1 region, only a single perch was smaller than 9 cm, and fish 10.5-15 cm in length dominated. In the W2 region, the largest number of perch caught was in the range of 9-12 cm TL. In the Solinka River (region K1), the differences between the lengths of the perch caught in individual years were non-significant, and the most numerous were individuals with a length of ~ 9 cm TL. However, in 2015 and 2018, smaller individuals with a length of ~ 6 cm were also quite numerous (Fig. 5).  Table 2 Pair-wise tests results from PERMANOVA of the perch abundance across the six hydromorphological regions sampled in upper San River basin S3-W1 refers to hydromorphological regions as defined in Fig. 1; 999-PERMANOVA permutations; NS-differences were nonsignificant; *-P < 0.05; **-P < 0.01  Table S1A, 2017 = 33%; Table S2A, 2018 = 40%) and a high degree of overlap of trace element signatures by hydromorphological river reach in the LDA plot (Fig. 6). Of note, the larval trace element signatures (obtained from fish caught in the various river hydromorphological reaches) also exhibited a high degree of overlap with larval otolith trace element signatures of fish caught in the Solina Reservoir in both years (Fig. 6). In contrast, comparison of the trace element signatures of the 50 μm edge segments, relating to when and where the fish were actually captured, had relatively high jack-knifed classification success (Table S1B, 2017 = 67%; Table S2B, 2018 = 84%,), with the LDA plot of the otolith trace element signatures exhibiting clear clustering relative to capture site (Fig. 6). Most of the otolith edge segment misclassifications in both 2017 and 2018 were between either adjacent sites, or the sites within the Solina Reservoir, indicating relatively similar trace element signatures limited the ability of the LDA to discriminate between sites (Tables S1, S2, Fig. 6).

Discussion
Abundance and distribution of perch in the upper San River catchment The abundance and distribution of perch has increased steadily and significantly in the upper San and Solinka rivers following impoundment of the San River, first by the Myczkowce reservoir in 1960, and then by the larger Solina Dam and reservoir in the late 1960s (Rolik 1971;Wajdowicz 1979;Kukuła 2003Kukuła , 2006. Given that perch have expanded their range into the upper catchment relatively recently, their description as an invasive native species in this upper part of the catchment seems justified (Valéry et al. 2009;Carey et al. 2012). Our observations suggest that the presence of the reservoir has contributed to the increased abundance and distribution of perch across the upper catchment (see also Kukuła 2003Kukuła , 2006, although we acknowledge that other environmental changes that we have not assessed may have also contributed. Since the last published assessment of perch abundance in the catchment (Kukuła 2006), perch have remained abundant and are increasingly widespread (this study). They are now seasonally the fourth most abundant species in the catchment by percentage, and they are continuing to extend their range upstream, including the smaller streams of the Bieszczady National Park. Their distribution  Table S1 (2017) and Table S2 (2018) for number of samples per hydrogeomorphic river reach, and Linear Discriminant Analysis classification success of P. fluviatilis to site of spawning or capture based on otolith trace element signatures upstream from the Solina Reservoir now extends to most river reaches where access is not blocked by either natural waterfalls or constructed weirs. Generally, mean abundances are higher and more stable at sites closer to the Solina Reservoir, with populations in these areas dominated by smaller fish, although there is considerable annual variation in abundance, particularly at sites further upstream. Our finding that perch populations across the upper San River catchment are dominated by relatively small perch suggests sustained successful spawning and recruitment within the system (see also Kukuła 2006). Further, populations closest to the lake (S1) are strongly skewed towards smaller size classes, with larger fish generally upstream, suggesting that much of the spawning and rearing of the early larval stages is occurring in or close to the Solina Reservoir, and with juveniles or adults subsequently migrating upstream into riverine environments. In the Great Lakes in North America, the larvae of closely related yellow perch (Perca flavescens) rear in the highly productive plumes of tributary rivers (Reichert et al. 2010;Carreon-Martinez et al. 2015). Perch larvae are small, pelagic and relatively fragile, and hence unsuited for life in fast flowing montane rivers (see Wang and Eckmann 1994;Kristensen et al. 2008). Indeed, even in pelagic environments, perch are prone to recruitment failure due to starvation if adverse conditions are encountered soon after hatching (Wang and Eckmann 1994;Karås 1996).
This suggests that the range and abundance of perch in the upper San River catchment is at least partly determined by the fairly specific habitat requirements of the early life history stages of perch, i.e., eggs and larvae (Treasurer 1988;Wang and Eckmann 1994;Urho 1996). If large numbers of perch are to invade and persist, then suitable littoral habitat for spawning (Čech et al. 2012) and pelagic habitat for larval rearing needs to be present and accessible within the wider landscape (Wang and Eckmann 1994). Like many other reservoirs built in submontane landscapes, the Solina Reservoir is relatively deep (dam height 58 m, average reservoir depth of 22.4 m), with a steeply sloping underwater profile and limited littoral habitat or macrophyte beds. However, after the construction of the Solina Reservoir, large numbers of submerged trees and shrubs remained in the littoral habitat, providing an ideal habitat for perch reproduction (Kukuła 2006). Perch fry were very numerous within a few years of reservoir construction (Wajdowicz 1979), and perch are now one of the species most frequently caught by anglers in the Solina Reservoir (Kukuła 2006, Kukuła unpublished data). Similar patterns of perch recruitment have also been observed in other reservoirs in the region, including the impoundment on the Czarna Orawa River, in Slovakia, where perch are now very abundant (Augustyn and Nowak 2014). Currently, large fluctuations in the water level of the Solina Reservoir (up to 10 m), and the associated erosion of the shoreline limit the development of littoral macrophytes (Prus et al. 1999(Prus et al. , 2002a. However, in the last 20 years, some submerged macrophytes, mainly Potamogeton and Elodea, have developed along some of the reservoir shorelines (Kukuła K., unpubl. data).
The pattern of increasing abundance of perch upstream of the Solina Reservoir is consistent with increases in perch abundance and range seen elsewhere following the construction of impoundments along river systems (Penczak et al. 1998;Kukuła 2006;Hladík et al. 2008;Matveev et al. 2002;Głowacki and Penczak 2013). Although adult perch are capable of surviving in a relatively wide range of lentic and lotic environments, spawning and larval rearing is typically most successful over submerged macrophyte beds and in the relatively protected, productive pelagic environments typical of lentic systems (Treasurer 1988;Wang and Eckmann 1994;Čech et al. 2007a, b). Although some perch spawning upstream of the Solina reservoir cannot be ruled out, particularly in the tributary reaches closest to the reservoir, the Solina Reservoir likely provides a stable environment where larvae can survive and recruit into adult populations in relatively large numbers in most years (see Treasurer 1988;Wang and Eckmann 1994;Irz et al. 2006;Čech et al. 2007a, b). Greater variation in the abundance and mean length of perch further upstream suggests conditions for perch are not consistently favourable in the upstream submontane riverine environment where river discharge is highly variable and there is a general lack of the ox-bow lakes or slow flowing riverine pools (A. Bylek and K. Kukuła, personal observations). Given the lack of stable pelagic habitat upstream of the Solina reservoir, the early life history stages of perch would be particularly vulnerable to exposure to sustained periods of adverse conditions (see Treasurer 1988;Irz et al. 2006). The loss of perch from the Wetlina stream after 1980 following the formation of a barrier due to a rockslide near the confluence with the Solinka River (Dziuban 1980) supports our contention that the stability and persistence of upstream populations is partly dependent on the supply of recruits from the large and relatively stable perch population now present in the Solina Reservoir.

Otolith microchemistry and perch life history
The increase in perch abundance across the upper San River catchment has occurred subsequent to the construction of the Solina Dam and Reservoir. Our analysis of perch otoliths collected from across the upper San River catchment provides compelling evidence that a large proportion of the perch in the upper San River catchment are completing their larval and early juvenile life stages in the Solina Reservoir, then migrating widely upstream to support the extensive populations of perch that are clearly now present in the upper reaches of the San and Solinka rivers. This upstream migration likely provides a degree of propagule pressure (Colautti et al. 2006), sustaining and providing a degree of resilience for perch populations in upstream riverine habitats where some spawning also likely occurs, but where conditions for spawning and larval rearing are unlikely to be consistently suitable (Irz et al. 2006).
Large numbers of fish rearing in a single waterbody (i.e., Solina Reservoir), where relatively homogeneous environmental conditions are more likely to occur compared to riverine sites in separate catchments, is the most parsimonious explanation for the relatively consistent otolith microchemical signature close to the core of otoliths seen in many of the analysed otoliths (see Elsdon and Gillanders 2003;Elsdon et al. 2008;Izzo et al. 2018). Ambient environmental conditions (particularly water chemistry) are the primary driver of trace element uptake and incorporation into otoliths (Elsdon and Gillanders 2003;Elsdon et al. 2008;Izzo et al. 2018). The similarity of most of the otolith trace element signatures deposited during the early life history of the perch from across the upper San River catchment compared with the otolith trace element signatures collected from the perch actually caught in the Solina Reservoir also lends weight to the argument that a large proportion of perch in the upper San River catchment are spawning and completing the early part of their life history in the Solina Reservoir. Further, variation in the ambient environmental conditions across the catchment is also sufficient to generate distinctive and reach-specific variation in otolith trace element signatures. Clearly, the edges of otoliths collected from perch in different hydromorphic river reaches have distinctive trace element signatures, providing further evidence of a post-larval upstream migration (see Elsdon and Gillanders 2003;Elsdon et al. 2008;Izzo et al. 2018). Upstream migration of perch from the Solina Reservoir is plausible given extensive perch migration occurs in other systems (Lilja et al. 2003;Syväranta et al. 2008;Hall et al. 2022).

Reservoirs as a focal point for colonisation by aquatic invasive species
Our research provides evidence of a key mechanism by which dams and associated impoundments can facilitate the colonisation of freshwater systems by native and non-native invasive fish species, i.e., by creating extensive pelagic habitat, thus enabling species with a specific life history requirement for pelagic larval rearing habitat to recruit in abundance and then migrate upstream (or downstream). Previous work has identified impoundments as focal points for native and alien fish invasions (Johnson et al. 2008;Grabowska et al. 2011;Jellyman and Harding 2012), and their potential to act as "stepping stone" habitats for the spread of invasive species across landscapes (Johnson et al. 2008;Głowacki and Penczak 2013). The specific focus of our study has been on the habitats and fish communities upstream of the Solina Reservoirwhilst these systems are not directly impacted by the Solina Reservoir, the fish communities in these riverine systems have nonetheless been significantly altered since the construction of the dam, with native but expansive perch populations now abundant upstream of the Solina Reservoir (Kukuła 2006, this study). Our otolith microchemical results indicate that the creation of the Solina Reservoir has enabled perch to recruit in much greater numbers, by successfully spawning and completing their pelagic larval and early juvenile life history stages in the extensive pelagic environment of the reservoir. Upstream of the Solina Reservoir, perch have now expanded their range to include most areas downstream of natural waterfalls and artificial weirs. Such obstacles are difficult for fish to overcome, especially at low water levels (Kukuła 2006;Jones et al. 2021).
In the upper San River catchment, the Solina Reservoir is an extensive lacustrine habitat in an otherwise largely fluvial landscape. Whilst adult perch are found in a wide variety of habitats, their pelagic larvae require stable lentic habitats if they are to successfully recruit in large numbers into adult populations consistently (Treasurer 1988;Wang and Eckmann 1994). High fecundity is a trait associated with many invasive and nuisance fish species (Kolar and Lodge 2002;Liu et al. 2016), and is typically achieved by reductions in egg size and subsequent larvae, often at the expense of larval resilience and survival (Smith and Fretwell 1974;Winemillar and Rose 1993;Augspurger et al. 2017). Pelagic larval fish often have a quite specific need for rearing in a stable and productive pelagic habitat, and if that habitat is available, they can then recruit into adult populations in huge numbers (Augspurger et al. 2017). Many new impoundments are also relatively nutrient-enriched and have impoverished biotic communities that lack abundant predators and competitors (Čech et al. 2007a, b;Johnson et al. 2008), further enhancing their potential as productive pelagic larval rearing habitat. Clearly, the potential of large impoundments to provide critical lentic habitat for species with pelagic larval stages (including invasive fish, crustacea and molluscs- Johnson et al. 2008), enabling them to colonise and recruit into catchments they would otherwise be largely excluded from, should be included in the risk assessment for any planned impoundment.