Evidence for the unexpected dispersal of Leporinus tigrinus Borodin, 1929 through hydroelectric dams in the most regulated Neotropical river basin

The Upper Paraná River Basin (UPRB) is harmed by several human activities, among which species introduction is particularly intense, making it the Neotropical ecoregion with the highest number of non-native fish species. Leporinus tigrinus Borodin, 1929 is a native species of the Tocantins-Araguaia River Basin and was recorded for the first time in the UPRB in 1999. Since then, the species was progressively recorded in lower portions of the UPRB (downstream Corumbá River). Therefore, the aim of this study was to provide evidence of possible introduction, establishment and dispersion of the species in that region. For this purpose, an inventory of its occurrence was performed by sampling ichthyoplankton, juveniles and adults of fishes in tributaries of the Porto Primavera Reservoir, as well as by compiling data from scientific collections considering all the UPRB. Diet and reproductive aspects were also analyzed. Overall, 23 specimens of L. tigrinus (10 adults and 13 juveniles) and five eggs (identified by DNA Barcoding) were sampled, evidencing the occurrence, reproduction, and recruitment of this species in the basin. The results of this study support the conclusion that L. tigrinus is an introduced species in the lower portion of the UPRB, capable of dispersing over long distances by passing through several reservoirs and dams (from upstream to downstream), possibly via turbines, spillways, locks, and fishways. Its dispersal pathway differs from the patterns observed for most invasive fish species in this basin, and suggests a substantial downstream dispersion through dams, even in the absence of fishways.


Introduction
Non-native species are among the main drivers of recent extinctions (Bellard et al. 2016a) and have been leading to the impoverishment and homogenization of ecosystems worldwide (Olden et al. 2004;Daga et al. 2015;Toussaint et al. 2016). Most of the immigrant organisms perish en route or soon after their arrival in new locations (Mack et al. 2000). Nevertheless, some of them can become invaders, especially when they find suitable environments and dispersion corridors which favor their persistence, proliferation and spreading in the recipient ecosystem (Hulme 2009;Bellard et al. 2016b;Braasch et al. 2019).
Fluvial ecosystems are particularly susceptible to invasions as a result of being interconnected in their multidimensionality (Ward 1989). However, the current boom in global dam construction is limiting this connectivity (Nilsson et al. 2005;Zarfl et al. 2015;Makrakis et al. 2019) and restricting long free-flowing rivers to remote regions of the world (Grill et al. 2019). Dams can enable invasions by causing disturbances that favor the establishment of non-native species (Havel et al. 2005;Johnson et al. 2008). Also, a highly-regulated river basin may inhibit or even prevent the movement and dispersion of fish (Pelicice et al. 2015;Reis et al. 2016), potentially limiting the range expansion of introduced species.
In South America, the Upper Paraná River Basin (UPRB) is the most regulated river basin (Agostinho et al. 2008) -its main river and largest tributaries were converted into a reservoir cascade . Initially, it encompassed the stretches of the Paraná River Basin located upstream the Sete Quedas Falls, a natural barrier capable of preventing the dispersal of most fish species between the Upper and the Lower Paraná River basins (Vitule et al. 2012), two distinct ichthyofaunistic provinces (Abell et al. 2008). Following the construction of the Itaipu Dam and the subsequent flooding of the Sete Quedas Falls in 1982, there was massive colonization of the Upper Paraná River Basin by endemic species from the Lower Paraná River Basin (Júlio Júnior et al. 2009;Vitule et al. 2012). The dispersal barrier was relocated roughly 150 km downstream, being then represented by the dam. In 2002, a fishway known as "Canal da Piracema" was implemented in this dam, enabling some species of the Lower Paraná River to have access to the Upper Paraná River (Makrakis et al. 2007(Makrakis et al. , 2011, which has resulted in new biological invasions over time Ota et al. 2018).
Currently, the UPRB presents the highest number of non-native fish species out of all Neotropical ecoregions (105 species; Gubiani et al. 2018). It is possible to identify at least two distinct patterns of biological invasion in this basin. The first pattern was observed for the highest number of invasive fish (around 60 species), which migrated from the Lower Paraná River (Gubiani et al. 2018;Ota et al. 2018). These species have limited upstream dispersion due to dams, which restrict their range to the lowest reaches of the basin (Júlio Júnior et al. 2009;Casimiro et al. 2017). The other pattern is presented by species of human interest, such as those originating from fish farming, fish stocking, and aquarium trade, as well as species sought for sport fishing and live bait (Gubiani et al. 2018). These species have had multiple introduction events and foci, and as a result are more widely diffused in the basin (Daga et al. 2015;Garcia et al. 2018).
A third pattern, however, may be occurring in the UPRB, involving the downstream dispersal of invasive species, despite the fragmentation caused by dams. This phenomenon was potentially observed for Leporinus tigrinus Borodin, 1929, a fish species native to the Tocantins-Araguaia River Basin (Britski and Garavello 2007) with recent records for the UPRB. The first capture of this species in the UPRB dates from 1999, from a broad collection of fishes in the Corumbá River, an uppermost drainage of this basin ). On that occasion, the authors considered L. tigrinus as native to that drainage, and, by extension, the species was considered native to the UPRB (Langeani et al. 2007). L. tigrinus had never been recorded until that date in the lower portions of the UPRB (downstream Corumbá River), which have a well-documented ichthyofauna. For this reason, the species was considered introduced (Froehlich et al. 2017), or possibly introduced (Santos et al. 2013;Ota et al. 2018) in the region. Even when considered to be an alien species, the reasons behind its introduction continue to be ignored (Froehlich et al. 2017).
This study aimed to clarify the origin of L. tigrinus in the lower portion of the Upper Paraná River Basin (i.e., native vs. non-native), by providing evidence of its possible introduction, establishment and unusual dispersal across an intensely fragmented environment, and to test the hypothesis that dams of hydroelectric plants are not efficient to prevent its dispersal.

Study area
With a drainage area of 2,966,900 km² and extending through five countries (Brazil, Argentina, Bolivia, Paraguay, and Uruguay), the Paraná River Basin (also known as La Plata River Basin) is the fifth-largest in the world, and the second-largest in South America (Wolf et al. 1999). Its upper stretch, the Upper Paraná River Basin (UPRB), currently extends to the Itaipu Dam ( Figure 1) and is considered an ecoregion of high ichthyofaunal diversity (Langeani et al. 2007).
This basin drains great urban centers as well as important agricultural and industrial areas of Brazil. It is subject to various anthropogenic impacts, such as deforestation, pollution, habitat loss, the construction of hydroelectric dams (which affect all of its main watercourses), and the introduction of alien species (Agostinho et al. 2008;Pelicice et al. 2017).
The largest reservoir in the UPRB is formed by the Engenheiro Sergio Motta Hydropower Plant (also known as Porto Primavera). Some tributaries of this reservoir still preserve natural characteristics of the hydrological regime, such as the Aguapeí River (a dam-free watercourse), and the Pardo and Verde rivers (lightly regulated only in upper stretches). Along with the Paraná River floodplain, located downstream from the Porto Primavera Dam, these three tributaries are acknowledged as critical reproductive sites for many fish species (Da Silva et al. 2015, 2019. They contribute to the fish diversity maintenance in the Porto Primavera Reservoir, by acting as source environments (Marques et al. 2018).

Occurrence data
To determine the current range of L. tigrinus in the UPRB, we (i) compiled occurrence records from scientific collections available in an online database (SpeciesLink 2019), and (ii) performed field sampling.
Fish sampling was carried out from 2011 to 2018, in both breeding and non-breeding seasons, in three tributaries of the Porto Primavera Reservoir (Aguapeí, Pardo, and Verde rivers), where L. tigrinus was historically absent (Marques et al. 2018). In all, 141 samples (47 per tributary) were collected using two sets of 10-m-long gillnets of different mesh sizes (2.5; 3; 4; 5; and 6 cm), installed at 5 p.m. in the river channels and marginal lagoons. In the river channels, gillnets were installed close to the banks, in locations of low water flow, while in the lagoons they were distributed along the banks, as well as in limnetic zones. The time of exposure was approximately four hours, and the sampled fish were preserved in ice until they arrived in the laboratory, where specimens of L. tigrinus were identified according to Santos and Jégu (1989) and Ota et al. (2018).
All collections were conducted following the Ethics Committee on the Use of Animals of the Universidade Estadual do Oeste do Paraná (Protocol Number 59/09), and the System of Authorization and Information on Biodiversity (SISBIO) issued the authorization for the collections (no. 31137-5).
A spatio-temporal distribution map was made to check if L. tigrinus is spreading in the UPRB. This map considered the year and spatial position of the first record, and the following records which expanded the distribution of the species in the basin (records from both scientific collections and field samplings were considered). Finally, a dispersion graph, followed by a Spearman correlation between the spatial position and year of record, was constructed, using the map data. This analysis was performed through software SAS University Edition 9.4, and the significance level was 5%.

Reproduction and recruitment
To detect the reproduction and recruitment of L. tigrinus in the basin, the sex and gonadal maturation stage of collected fish were identified by the macroscopic analysis, according to Vazzoler (1996). Juvenile fish (with immature gonads) were considered to be an indication of recruitment.
Ichthyoplankton were also studied to detect the spawning of L. tigrinus. Sampling was carried out monthly from November 2017 to April 2018, at the same three tributaries of the Porto Primavera Reservoir. The collections were made through horizontal tows lasting 10 minutes along the surface of the water (at 20 cm depth), between 9 and 11 p.m., using a conical-cylindrical plankton net (0.5 mm mesh). The samples were preserved in 90% ethanol and were refrigerated until stored in a freezer at −20 °C. In the laboratory, fish eggs and larvae were separated from other organisms and detritus.

Molecular identification of ichthyoplankton (DNA barcoding)
Since the morphological identification of eggs and larvae is challenging and the early-development stages of L. tigrinus have not yet been described (Nakatani et al. 2001), the presence of the species in the ichthyoplankton was identified by DNA barcoding. For that, total genomic DNA of the sampled eggs and larvae was extracted in a buffer containing Chelex 100 and proteinase K. The cytochrome c oxidase subunit I gene (COI, 648 bp) was amplified using the primers described by Ward et al. (2005). The amplicons were sequenced and analyzed in ABI-PRISM 3500 XL automated sequencer (Applied Biosystems). The detailed protocol of the extraction, amplification, and sequencing of the COI gene is described in Frantine- .
The nucleotide sequences were edited and aligned using Clustal W (Thompson et al. 1994), a program incorporated into MEGA7 software (Kumar et al. 2016). Then, these sequences were compared to the sequences in GenBank database through BLAST (Basic Local Alignment Search Tool), which searches for regions of similarity between sequences and is mostly used for species identification (Kelly et al. 2011;Stoeckle et al. 2011;Abdullah and Rehbein 2016). Ichthyoplankton sequences with similarity higher than 98% to L. tigrinus sequences in Genbank were assigned to this species. This threshold was adopted because, in the UPRB, the maximum intraspecific genetic divergence observed for species of the genus Leporinus was 0.9%, and the minimum interspecific genetic divergence was 3.3% (Pereira et al. 2013).

Diet composition
Specimens of L. tigrinus collected in the field samplings also had their diet analyzed, after being preserved in 10% formaldehyde. Stomachs were removed through an abdominal incision, and only those which were full, or partially full, were considered.
The stomach contents were identified in the following categories using a stereomicroscope: organic detritus (particulate organic matter in different stages of decomposition), sediment (mineral particles), terrestrial plants (fruit, seeds and leaves of vascular plants), aquatic plants (bryophytes), filamentous algae, insects and macroinvertebrates (Arachnida, Bivalvia, Gastropoda, Annelida, and Diplopoda). Food items were quantified according to the volumetric method (Hyslop 1980), estimated using graduated test tubes and a glass counting plate (Hellawell and Abel 1971).

Occurrence and range expansion
The first record of L. tigrinus in the Upper Paraná River Basin (UPRB) occurred in 1999, in the Corumbá Reservoir, state of Goiás, Brazil , which still corresponds with the uppermost limit of the species distribution in the basin. Over time, the species was recorded sequentially in lower reaches of the basin, in the Paranaíba and Paraná rivers. Leporinus tigrinus was also recorded in some tributaries of these rivers, in stretches without dispersal barriers between them and the main streams (Table 1, Figure 1, Figure 2A). The cascade of eight reservoirs where the species currently occurs is formed by the Corumbá, Itumbiara, Cachoeira Dourada, São Simão, Ilha Solteira, Jupiá, Porto Primavera, and Itaipu dams (Table 2). Of these dams, only Porto Primavera and Itaipu have fishways (a fish ladder in the former and a lateral system in the latter).
The current southernmost limit of the species' range is the Canal da Piracema inlet, in the Itaipu Dam ( Figure 2A). The range expansion of L. tigrinus in the UPRB occurred in a descending direction (Spearman correlation: ρ = 0.97; p < 0.01) ( Figure 2B), covering approximately 1,300 km of river over 18 years (72.2 km year -1 ).
The field sampling carried out in the present study broadened the known occurrence area of the species as a result of new records in the three tributaries of the Porto Primavera Reservoir since 2013 ( Figure 1; Table 1). Through the sampling period, 23 specimens of L. tigrinus were collected, of which six were deposited in the ichthyological collection of the Museu de Zoologia da Universidade Estadual de Londrina (MZUEL19890, 19912, 19913, 19915, 19916). Since the first capture of L. tigrinus in the tributaries, its abundance corresponded to 0.27% of all collected fish (n = 8,613).

Reproduction and recruitment
The reproductive activity of L. tigrinus was detected in the three tributaries of the Porto Primavera Reservoir. Of the ten caught adults (mean standard length: 160.7 mm ± 23.4), six were evaluated. Two female specimens captured in December 2013, in the Pardo and Aguapeí rivers, were in reproduction (mature ovaries). The other four specimens were captured during the nonbreeding season (April and June) and presented resting ovaries.
Among the 793 eggs and 689 fish larvae sampled in the spawning season of 2017-2018, 132 eggs and 215 larvae were identified through DNA Barcoding   techniques. Five eggs collected in the Verde River were assigned to L. tigrinus since they presented 100% similarity with the sequence of this species in the GenBank database (accession number: KF568997.1, from a specimen collected in the Araguaia River; Ramirez et al. 2016). New sequences generated in this study were submitted to GenBank (MN186394 to MN186398). The sampled L. tigrinus eggs were pelagic, non-adherent, and spherical. One egg in embryonic development of cleavage-stage, with 2.54 mm in diameter and large perivitellinic space (0.89 mm), was photographically recorded ( Figure 3A).
Juvenile specimens of L. tigrinus were also recorded, which comprised 56.5% (n = 13) of the 23 collected specimens of this species, and 7.8% of all juveniles collected (n = 167) in the tributaries since the first record of L. tigrinus for this stretch of the basin ( Figure 3B). Two juveniles were collected in January 2013, three in January 2014, two in January 2015, and six in November 2018.

Discussion
The present study gathered new evidence that supports the hypothesis that L. tigrinus is an introduced species to lower portions of the UPRB. Pavanelli et al. (2007) and Langeani et al. (2007) considered L. tigrinus as native to the UPRB, based on a single record of the species in the Corumbá Reservoir in 1999 (NUP 1093; Table 1). Before that year, L. tigrinus has never been recorded in the basin, despite the high number of inventories made downstream of Corumbá River (e.g. Agostinho et al. 2007;Graça and Pavanelli 2007).
Data compiled in scientific collections along with those obtained in the present study revealed that the first record of L. tigrinus in the UPRB was followed by new records sequentially located in lower sections of the basin (Figure 2A). In the sampled section of our study, located approximately 750 km downstream from the area of the first record, the species had only begun to be captured in 2013, despite intensive collection efforts dating back to 1999 (Marques et al. 2018). The correlation between the chronology and the spatial position of subsequent records ( Figure 2B) suggests the dispersal of L. tigrinus to new reaches of the basin, in an upstreamdownstream direction.
Currently, the range of L. tigrinus in the UPRB spans the longitudinal stretch from the Corumbá Reservoir to the Canal da Piracema inlet (approximately 1,300 km), encompassing segments of the Corumbá, Paranaíba and Paraná rivers -a cascade of eight reservoirs. The species was also recorded in stretches of some tributaries without dispersion barriers with the main stream. Conversely, L. tigrinus was not recorded in the largest tributaries of the Upper Paraná River (Grande, Tietê and Paranapanema rivers), possibly because the upstream dispersal to these watercourses was prevented by dams. Thus, L. tigrinus seems to have expanded its distribution in a descending direction, by crossing dams and reservoirs, and laterally to dam-free stretches of tributaries. Such results suggest that dams are permeable to downstream but insurmountable to upstream movements of this species, in the absence of fishways. Until now, it was possible to identify at least two distinct patterns of biological invasion in the UPRB, involving different vectors and dispersal pathways. The first one, described by Júlio Júnior et al. (2009), refers to the species that accessed the Upper Paraná River after the flooding of the Sete Quedas Falls and the implementation of the Itaipu Dam fishway (Canal da Piracema), which constitutes a continuous source of propagules (Makrakis et al. 2007(Makrakis et al. , 2011. These species originating from the Lower Paraná River have their upstream dispersal limited by dams, although some species have reached upper stretches of the basin by utilizing fishways and ship locks (Júlio Júnior et al. 2009;Casimiro et al. 2017). Thus, in this pattern of invasion, stretches without barriers (natural or human-made) constitute an open route for dispersal of invasive species, whereas dams act as filters. Although dams are usually insurmountable, the passage of some species is possible when fishways are available (ladders, locks, elevators).
The other pattern encompasses multiple vectors (fish farming, fish stocking, aquarium trade, sport fishing, and live bait) with several and widespread introduction foci. The result is the presence of invaders in disseminated but not necessarily connected reaches of the basin (Daga et al. 2015;Garcia et al. 2018;Gubiani et al. 2018). Many of these species (such as cichlids and small characids) are non-migratory, generalists in terms of resource use, tolerant of environmental variations, and able to complete their entire life cycle in reservoirs, tending to remain in these locations (Agostinho et al. 2008). Others, such as migratory species originating from stocking (Garcia et al. 2018) or those escaping from fish farms (Forneck et al. 2016), display rheophilic behavior and present vast home range. However, the high human interest in these species makes it challenging to predict whether its presence in a given stretch of the basin is due to the existence of a local introduction focus or the dispersal from previously introduced sites.
The dispersal of L. tigrinus in the UPRB differs from these patterns, and the fact that this is the most-studied river basin in Brazil Langeani et al. 2007;Marques et al. 2018;Ota et al. 2018) allowed the detection of what may be a pathway for future biological invasions. Since L. tigrinus is not a species of particular human interest, a scenario involving multiple widespread introductions would not be realistic. As a result, the most plausible L. tigrinus invasion dynamics would require its introduction in an upper region of the basin (near or at Corumbá Reservoir), followed by its dispersion to lower reaches, despite intense fragmentation. Even though the vector of this possible introduction is unknown (Froehlich et al. 2017), species of the genus Leporinus are occasionally used as live bait, and L. tigrinus has potential for use in the aquarium trade. It is essential to highlight that the Upper Paraná and Tocantins-Araguaia river basins are contiguous, which facilitates the transport of species between them by human interaction.
The question of whether L. tigrinus is native or introduced to the Corumbá River Basin remains controversial. It is possible to speculate that the species is native to that river since that region drained north until the Miocene, with subsequent capture of rivers by orogenetic events and draining to the south (Beurlen 1970). Although there is a lack of dating of the origin of L. tigrinus, the Miocene was a period of significant diversification for Neotropical fishes (Albert and Reis 2011). In contrast, the lack of records of this species downstream of the Corumbá River Basin is strongly supported by scientific collections and long-term surveys that failed to collect the species in the region (e.g. from Britski 1972to Agostinho et al. 1997). This absence is highly unlikely to be due to species misidentification, since Leporinus octofasciatus Steindachner, 1915, another barred species that occurs in this basin, has a color pattern and pelvic-fin ray counts that readily distinguish it from L. tigrinus (see Ota et al. 2018). The Corumbá River was considered an area of fish endemism isolated by rapids and canyons (Pavanelli and Britski 1999). The formation of the Corumbá Reservoir may have created conditions for L. tigrinus to disperse downstream, which allowed it to achieve the current distribution in the UPRB gradually.
Dams and their reservoirs constitute barriers to upstream and downstream movements of fish, trapping populations between dams (Agostinho et al. 2008). Otherwise, the results of our study provide evidence that L. tigrinus was able to cross seven hydroelectric dams and their reservoirs, possibly through turbines and spillways (only the Porto Primavera Dam provides a fishway; Table 2). The use of these routes in the passage through dams had already been suggested by Makrakis et al. (2012) and Antonio et al. (2007), when studying migratory fish patterns in the Paraná River Basin. In North America, mark-recapture data in the Missouri River indicated that turbines and spillways could establish substantial downstream connectivity, even for species of large body size, despite the potential mortality (Pracheil et al. 2015).
Studies related to the spatial distribution of ichthyofauna in the Peixe Angical Reservoir, situated in the Upper Tocantins River, demonstrated that L. tigrinus preferably occupies bathypelagic (deep) waters in the lacustrine zone (Freitas et al. 2008). The strong preference towards deeper and inner regions of the reservoir could facilitate the entry of L. tigrinus specimens into turbines and spillways. It may provide indication as to how this species was able to cross seven dams in the UPRB, of which only one possess a fishway. Furthermore, an evaluation of the reproductive activity of fish species before and after the formation of a reservoir in the Tocantins River demonstrated that L. tigrinus is a species that is positively affected by reservoirs, showing a considerable increase in the percentage of spawningcapable females (from 30.4% to 83.3%) after the reservoir formation (Angulo-Valencia et al. 2016). This characteristic may have contributed to the species' success in the reservoir cascade of the UPRB.
An alien species is considered established when it constitutes persistent and self-sustainable populations in the new environment through reproduction and recruitment (Vermeij 1996). Although preliminarily, the present study indicates that L. tigrinus is possibly established in the sampled area, since we detected two females in reproduction, five fertilized eggs, and 13 juveniles of this species, which may be an indication of recruitment. Regardless of its presumed preference for lentic habitats (Freitas et al. 2008), our data suggest that L. tigrinus relies on lotic environments for its reproduction, since its pelagic eggs were collected drifting passively in a free-flowing stretch of the Verde River. The necessity to move in search of biotopes suitable for each life phase may also explain such a wide dispersal within the new area.
Considerable intraspecific divergence in the COI gene is observed in fish species with distributions ranging over multiple hydrographic basins (Ramirez et al. 2017;Silva-Santos et al. 2018;Bignotto et al. 2020). However, in the present study, the COI gene sequences from eggs collected in the UPRB (Verde River) showed 100% similarity to the sequence of L. tigrinus from the Tocantins-Araguaia River Basin. The high similarity would be due to a short period of isolation, since some level of genetic differentiation is expected between geographically isolated populations (Hebert et al. 2003). Thus, we can suggest that populations of L. tigrinus were separated in two different basins recently, reinforcing the hypothesis of introduction in the UPRB. However, research with additional genetic markers and a larger sample size is desirable to support this hypothesis.
The study of ecological and life-history traits provides essential information in understanding how species thrive in novel environments Liu et al. 2017), and plasticity is undoubtedly of central importance (Sol and Weis 2019). Regarding L. tigrinus, Monteiro et al. (2008) detected some versatility in its feeding habits, since it changed from piscivore to omnivore after the construction of a dam in the Tocantins River. This apparent trophic plasticity is reinforced by the present study, which revealed a third feeding habit for the species, predominantly detritivore. Organic detritus is also dominant in the diet of Hemiodus orthonops Eigenmann & Kennedy, 1903, a non-native fish of high abundance in the Upper Paraná . Bezerra et al. (2019) observed a tendency toward the dominance of bottom-feeding fish species in ecosystems altered by damming and species introduction, such as UPRB.
Our results support the view that L. tigrinus is non-native to the lower portion of Upper Paraná River Basin. Also, it is a candidate invasive species to this region, if considering the broader concept whereby a species is considered invasive when it spreads in a new area (Richardson et al. 2000;Heger et al. 2013). Furthermore, the knowledge of its unique dispersal pathway denotes substantial upstream to downstream connectivity in dams, even in the absence of fishways, and may explain the success of other invasive species in the occupation of reservoir cascades. Lastly, our findings indicate that L. tigrinus might continue its dispersal to lower reaches of the Paraná (La Plata) River Basin, since dams seem to be a permeable barrier to downstream movements of this species.