The main channel and river confluences as spawning sites for migratory fishes in the middle Uruguay River

Evidence indicates that migratory fish reproduce in the middle Uruguay River, but the location of spawning sites remains unknown. To identify spawning sites in the basin, fish eggs and larvae were sampled monthly between October 2016 and January 2017. The sampling was carried out in three sections along the middle Uruguay River, covering two environments: the main channel and the mouth of tributaries (Comandaí, Ijuí, and Piratinim rivers). A total of 11,519 eggs and 3,211 larvae were captured, belonged to ten migratory species. Eggs and larvae of migratory fishes, were widely distributed, with predominance of segmented eggs, and larvae in yolk-sac and pre-flexion stages, with higher densities near the confluence with the Piratinim River. Larvae assemblages showed spatial variation, indicating that spawning sites differ among migratory species. This study provides new information about fish reproduction in the middle Uruguay River, revealing that migratory species spawn in different localities along the main channel and tributaries. This region may function as a critical site for fish reproduction, although it is currently threatened by the risk of hydropower expansion, emphasizing the need for measures that preserve their environmental conditions, hydrological connectivity and ecological functions.


INTRODUCTION
Freshwater fish display a range of reproductive tactics (Nakatani et al., 2001), among which migratory behavior is an important feature shared by many species. The purpose of reproductive migration is to release eggs in habitats that maximize survival, and spawning sites generally differ from nursery and feeding grounds (Suzuki et al., 2009;Ávila-Simas et al., 2014). The diversity of migratory behavior is remarkable and may involve anadromous, catadromous, or potamodromous life cycles. For example, the European eel Anguilla anguilla (Linnaeus, 1758) migrates toward the ocean to spawn, while the Atlantic salmon Salmo salar Linnaeus, 1758 migrates toward continental rivers, i.e., homing (Hendry et al., 2004). Some European cyprinids migrate between lentic and lotic environments (Skov et al., 2008), while some Asian cyprinids migrate toward upstream reach spawning (Lucas, Baras, 2008). In the Neotropical region, which holds extraordinary fish diversity (Albert et al., 2020), migratory dynamics and spawning sites remain poorly understood, but a diversity of behavior has been reported (Makrakis et al., 2012;Duponchelle et al., 2021). Studies conducted in the Paraná, Amazon, and São Francisco basins indicate that most fish migrate toward the upper sections and tributaries to spawn (Vazzoler, Menezes, 1992), which may involve hundreds (Agostinho, Zalewski, 1996;Nakatani et al., 1997;Lopes et al., 2019a) to thousands of kilometers (Godoy, 1975;Barthem, Goulding, 1997;Barthem et al., 2017).
Recent research conducted in the Uruguay River basin has shown different patterns, as some migratory species reproduce at various sites along the river channel (Reynalte- Tataje, Zaniboni-Filho, 2008), in addition to the confluence of tributary rivers (Zaniboni-Filho, Schultz, 2003). These studies have been conducted in the upper Uruguay River, a region characterized by waterfalls and anthropic impacts (i.e., dams), which pose significant constraints to fish migration (Reynalte-Tataje, 2008;Hermes-Silva et al., 2009;Silva et al., 2012). The middle Uruguay River represents a different scenario, with a long fluvial stretch of over 800 km, with no dam or significant geographic barrier that could restrict fish movements. At the upper limit of this segment, the Salto do Yucumã a large waterfall, constitutes a partial barrier for fish movement and restricts the dispersion to upper Uruguay; however, no other barrier exists downstream until the Santo Grande Dam. The absence of significant barriers may clarify whether reproductive dynamics are similar to those observed in the upper Uruguay River, or to those described in other basins (e.g., migration to upstream reaches). In this sense, it is essential to understand the reproductive dynamics and critical environments in river segments that have been subjected to limited impacts, especially if we consider that most ichthyoplankton studies in Brazil and the world have been conducted in altered environments (Sanches et al., 2006;Hermes-Silva et al., 2009;Mu et al., 2014).
In this study, we investigated the distribution of eggs and larvae of migratory fish to identify spawning sites in the middle Uruguay River basin. Based on patterns observed in the upper Uruguay River (Hermes- Silva et al., 2009;Reynalte-Tataje et al., 2012a), we investigated the hypothesis that fish spawn in different sites along the main channel and the confluence with tributaries, and predicted the predominance of eggs and larvae in early development stages across these environments. Particularly, this study examined (i) spatial variations in the abundance of eggs and larvae of migratory fish; (ii) the distribution of different embryonic and larvae stages; and (iii) the taxonomic composition of larvae assemblages in the main channel and tributaries.

MATERIAL AND METHODS
Study area. The Uruguay River is an important sub-system of the Río La Plata basin. The upper basin has its sources at 1,800 m elevation. The river is channelized in the valley, with steep slopes and considerable variation in water flow. The middle basin is a long unregulated segment of about 800 km. It is a lowland region that begins at 130 m of elevation, with an average slope of 0.16%. The lower basin is about 350 km long, with a difference in elevation of about 1 m (Zaniboni-Filho, Schulz, 2003). The wet period occurs between June and October, with relevant inter-annual variations (Zaniboni-Filho, Schulz, 2003). According to Reynalte-Tataje, Zaniboni-Filho (2008), the driest period occurs during the austral summer.
This study was carried out in the middle Uruguay River in a period characterized by regular climatic conditions, without the occurrence of extreme climatic events (INPE, 2021). In this stretch, the river channel is characterized by slow currents, pools, channels, backwaters, islands, and some rapids. Floodplains and wetlands are absent, but these environments are found approximately 150 km downstream. Three sections were selected along a 70 km river segment. We sampled two sites in each section, one site in the tributary river and the other in the Uruguay River, upstream from the confluence with the tributary, totaling six sites (Fig. 1). river segment, 1,300 m wide, with shallow areas, deep pools, rapids, and fluvial islands. The sampling site in the tributary (hereafter Trib-P) is located on the Piratinim River, 150 m upstream from its confluence with the Uruguay River. The tributary is 130 m wide with a drainage area of 7,500 km 2 . This tributary is not regulated by upstream dams.
Sampling. Monthly samples were taken between October 2016 and January 2017, during the reproductive period of migratory species in the basin (Reynalte-Tataje et al., 2008a). Sampling occurred during the first two hours after dusk, following the method described by Hermes- Silva et al. (2009). Fish eggs and larvae were collected with conical-cylindrical plankton nets, 500 µm mesh size, 1,5 m long, and equipped with a mechanical flow meter. Each site was sampled for three days at 24-hour intervals, by positioning two nets against the water flow near the surface. Nets remained set for 10 min, a procedure that was repeated, generating four samples/day/site. Daily abundance was then calculated as the sum of the four samples divided by the total filtered volume of the four samples, totaling three monthly replicates for each sampling site. We collected 72 samples (6 sites x 4 months x 3 replies). All material was stored in polyethylene bottles and fixed in 4% formalin.
In the laboratory, eggs and larvae were sorted under a stereomicroscope with the aid of a Bogorov plate. It is not possible to identify eggs of migratory species, but these eggs have a higher perivitelline space (> 30 %) (Nakatani et al., 2001); then, eggs with this characteristic were assigned as belonging to migratory fishes. Eggs were separated according to their degree of embryonic development: Segmentation (S), Head-Tail (HT), and Free-Tail (FT) (Nakatani et al., 2001). Larvae were identified as the lowest taxonomic level following Nakatani et al. (2001) and Reynalte-Tataje, Zaniboni-Filho (2008). Larvae were also separated according to the notochord-flexion development of caudal fins and supporting elements, as proposed by Ahlstrom, Moser (1976)  The larvae were classified as belonging to migratory or non-migratory species following specific literature (e.g., Carolsfeld et al., 2003;Reynalte-Tataje, Zaniboni-Filho, 2008;Massaro et al., 2019). Neotropical migratory fishes have complex life cycles and migrate over long distances (>100 km) between feeding and spawning sites (Carolsfeld et al., 2003). These fish are potamodromous and migrate seasonally between river habitats; they do not have parental care and release eggs in the water current. Otherwise, non-migratory fishes include a variety of behaviors (e.g., sedentary, short-distance displacements, rheophilic, limnophilic, and parental care), differing substantially from patterns observed for migratory species (Winemiller, 2003;Suzuki et al., 2004). The vouchers was deposited in the Coleção Ictiológica do Núcleo de Pesquisas em Limnologia, Ictiologia e Aquicultura, Universidade Estadual de Maringá, Maringá (NUP), Paraná, Brazil.
Data analysis. The number of eggs and larvae was converted to density (individuals/10 m 3 of filtered water), following Tanaka (1973) modified by Nakatani et al. (2001). In order to investigate differences in the abundance of eggs and larvae among sampling sites, we used the non-parametric Kruskal-Wallis test, followed by Dunn's comparison, 6/16 ni.bio.br | scielo.br/ni Spawning sites migratory fish in the middle Uruguay River because the data did not meet parametric assumptions. To test differences in taxonomic composition, stages of embryonic development, and stages of larval development between sampling sites, we used a permutational multivariate analysis of variance (oneway PERMANOVA) with 999 permutations, based on Bray-Curtis similarity applied to untransformed data (Anderson, 2005). The indicator value method (IndVal; Dufrêne, Legendre, 1997) was employed to determine species, embryonic stages, and larval stages that indicated of each sampling site. Statistical significance implied p < 0.05. All analyses were run in PCORD v. 5 (McCune, Mefford, 2011) or the R environment (R Development Core Team, 2018).
Egg density. We recorded migratory and non-migratory fish eggs at all sampling sites. Eggs of migratory fish were more abundant at Chan-P and Trib-P ( Fig. 2A; p < 0.05). Most eggs (> 75 %) were in the segmentation stage (with 2 to 16 blastomeres) (Fig.  3). Eggs in the head-tail stage were also widely distributed, but with low frequency, while the free-tail stage was less abundant (Fig. 3). Significant differences in embryonic development stages were detected among sampling sites (PERMANOVA; pseudo-F = 2875.2, p < 0.0001). The highest presence of eggs in the segmentation stage was found at the mouth of the three tributary rivers (Trib-C, Trib-I, and Trib-P) and also at Chan-P (IndVal, p < 0.05; Tab. 1; Fig. 3).

Sampling site
Embrionyc stage* IndVal p   Larvae density and composition. Larvae belonged to four orders, 20 families, 44 genera, and 35 species were captured. Siluriformes accounted for 49.5% of the catch and Characiformes summed 43.5%. We recorded larvae of migratory and non-migratory fish in all sampling localities, however, only migratory species larvae were analyzed (representing 16.1% of all larvae captured). The largest density of migratory species larvae was registered in the Piratinim River confluence area, mainly at Trib-P ( Fig. 2B; p < 0.05). The list of identified larvae and their distribution in the different sampling sites are shown in the Tab. S1.
We detected different development stages, but initial phases predominated (Yolk-sac larvae and preflexion). The distribution of larval stages differed among sampling sites (PERMANOVA; pseudo-F = 699.4, p < 0.0001). The highest proportion of the LV stage was found at Trib-P, and the PF stage at Trib-I, Chan-I, and Chan-P (IndVal, p < 0.05; Tab. 1; Fig. 4).

DISCUSSION
Our results evidence migratory fish spawns at different localities along the middle Uruguay River, with the main channel and tributaries working as spawning sites. Confirming our prediction, we found eggs and larvae in different localities, with a predominance of early development stages. It indicates that spawning is taking place close to the sampling area (approximately 10 km upstream) considering the average water velocity in the Uruguay River (<1.0 m.s -1 ) (Cardini et al., 2004); however the occurrence of other development stages suggest also to spawning activity in more distant sites. In fact, eggs and larvae in early development were also recorded in the Uruguay River approximately 210 km upstream from the study site (Ziober et al., 2015), indicating that the middle section works as a macro spawning area. These findings support our hypothesis that fish spawn in different sites along this river segment, including the main channel and the confluence with tributaries. This stretch is freeflowing and surrounded by two protected areas (Parque Estadual do Turvo in Brazil, and Reserva de la Biosfera Yabotí in Argentina), creating suitable conditions for the reproduction of migratory species. Therefore, migratory fishes in the Uruguay River basin apparently have multiple spawning sites, which include the main channel and tributaries in the upper (Reynalte-Tataje et al., 2012a;Silva et al., 2015) and middle reaches of the basin (Ziober et al., 2015;Reynalte-Tataje et al., 2017). However, middle Uruguay may have substantial importance considering that upper and lower reaches are disturbed by river regulation (Fig. 5); middle Uruguay, therefore, seems to work as a relevant reproduction site for migratory fishes, with the potential to supply stocks in other locations of the basin (Zaniboni-Filho, Schulz, 2003;Fuentes et al., 2016;Reynalte-Tataje et al., 2017;Serra et al., 2019).
Eggs from migratory species were collected in all sampling sites, and they accounted for most individuals in the samples compared to larvea. Eggs at early development stages (segmentation) were the most abundant, showing that most spawning activities took place in areas close to our sampling sites. Studies about the embryonic development of migratory fishes indicate that segmentation starts within the first two hours after hatching (Ninhaus-Silveira et al., 2006;Reynalte-Tataje et al., 2008b;Luz et al., 2018;Araújo et al., 2020). It means that organisms would disperse downstream over a few kilometers from the spawning site in a flowing river characterized by the presence of  Spawning sites migratory fish in the middle Uruguay River pools and rapids. This information is relevant as it indicates that spawning grounds are near the six sampling sites, in the main channel and tributaries. Eggs at more developed stages were captured near the confluence with the Ijuí and Piratinim rivers (Ch-I and Ch-P sites), and they may have been originated from spawning activities occurring near the confluence with the Comandaí River (Ch-C site) and/or other sites located farther upstream. In this river segment, the spawning area must be limited to the segment downstream from the Yucumã Falls, as this waterfall blocks upstream displacements. In fact, a study carried out in 24 sampling sites that included locations above and below this waterfall showed spawning activity (migratory and non-migratory fish) mainly downstream from the Falls, with almost no spawning in the upstream region (Ziober et al., 2015). Therefore, our study provided evidence that spawning occurs in different localities of the middle Uruguay River, especially in areas close to the confluence with the Comandaí, Ijuí, and Piratinim rivers. Larvae were also widely distributed in the study area, although in lower densities if compared with eggs. However, the percentage of larvae in the ichthyoplankton was above 20%, which is higher than the value observed (ca. 5%) in upper Uruguay (Silva et al., 2012;Reynalte-Tataje et al., 2013). The predominance of larvae in early development stages (yolk-sac larvae and preflexion) also indicates that spawning sites are close to the study area, but the incidence of advanced stages, although in low densities, points to reproductive activity in the upper stretch of the Uruguay River and tributaries. Studies carried out on upper Uruguay and other Neotropical basins indicate that larvae may  (Agostinho, Zalewski, 1996;Suzuki et al., 2009). Thus, although some larvae develop in the river channel, most probably drift towards lowland areas located 130 km downstream from the confluence with the Piratinim River, where they find suitable conditions for development.
Although eggs and larvae were registered in the main channel and mouth of tributaries, we recorded spatial variations in the structure of assemblages. Studies have shown that ichthyoplankton diversity is related to morphometric aspects of the river, where smaller rivers seem less attractive to reproductive fish (Hermes- Silva et al., 2009;Reynalte-Tataje et al., 2013;Silva et al., 2015). In fact, low densities of eggs and larvae occurred in the mouth of the Comandaí River, the smallest system in the area. Otherwise, higher densities of eggs and larvae, at the early stages of development, occurred in the Piratinim River. This tributary is well preserved and composed of different microhabitats such as rapids, channels, islands, and pools, reinforcing the relevance of instream habitat diversity for fish reproduction (Oliveira, Ferreira, 2008;Reynalte-Tataje et al., 2013;Silva et al., 2015). However, the Ijuí River, the largest subbasin in the study area, presented the lowest richness of migratory species. The presence of two large dams upstream may explain this pattern since flow regulation and limited connectivity negatively affect the reproduction of some migratory species (Agostinho et al., 1993;Agostinho et al., 2007). A similar pattern was observed in the upper São Francisco River, where migratory fish seem to avoid regulated tributaries (Lopes et al., 2019b).
One important aspect is the relevance of confluence areas for spawning and larvae drift. The importance of these environments for fish reproduction had also been verified in the upper Uruguay (Hermes- Silva et al., 2009;Reynalte-Tataje et al., 2012a). Some studies suggest that environmental variation in the confluence (temperature, electrical conductivity and flow) can act as triggers for fish spawning (Lucas, Baras, 2008;Rakowitz et al., 2008;Reynalte-Tataje et al., 2012c). These findings point to the importance of fluvial conditions and hydrological connectivity for reproduction dynamics, as this setting seems to create an adequate environment for spawning and egg/larvae transportation.
The larvae captured in this study, mainly in drifting stages, indicate that migratory species have different preferences for the spawning sites. Pseudoplatystoma corruscans seems to spawn mainly in the upper part of the middle Uruguay River, a pattern supported by other studies, such as Reynalte-Tataje et al. (2017) and Pachla et al. (2020). Differently, Prochilodus lineatus and Megaleporinus obtusidens spawn in regions further downstream and close to lowland regions. The dependence of larvae of this species on microcrustaceans (Paes et al., 2011) and juveniles on detritus (P. lineatus) and plant material (M. obtusidens) (Bayol, de Yuan, 1996;Reynalte-Tataje, Zaniboni-Filho, 2020) may partly explain the behavior of spawning in a section closer to lowland areas. Larvae of Salminus brasiliensis and Pimelodus maculatus were found mainly in the most preserved tributary river, which indicates that spawning occurred in this river. Greater activity of S. brasiliensis and P. maculatus spawning in tributary rivers was also observed in the upper Uruguay (Zaniboni-Filho, Schulz, 2003;Reynalte-Tataje et al., 2012a). The spatial segregation of migratory species shows that the middle Uruguay as a whole is important for this group of fish, not only because their home range is extensive, but also because each species demands specific spawning sites and nursery grounds. Nevertheless, more studies are needed to confirm these patterns, especially to follow multiple reproductive events.
Our results indicate that the middle Uruguay works as a critical site for fish reproduction. The river channel and tributaries seem to play a particular role, where migratory fishes spawn at multiple sites, with distinct preference among species. The greater flow in the Piratinim River, in particular, must attract reproductive fish and work as an important spawning ground. Habitat integrity in the middle Uruguay (e.g., natural flow regime, tributaries, and riparian forests) is probably a key factor allowing the reproduction of migratory fishes. It is worth noting that we recorded 10 migratory species in the area, i.e., out of 15 species existent in the whole basin (Reynalte-Tataje et al., 2012d). This scenario illustrates the importance of lotic stretches that preserve environmental heterogeneity for the maintenance of migratory fish populations (Lopes et al., 2019b;Massaro et al., 2019). Further studies are needed to confirm the patterns reported here, and to understand the dynamics of multiple reproductive cycles and the role of other river stretches. This information is very important considering that river regulation has disrupted the functioning of the Uruguay River basin in its upper reaches. Moreover, hydropower expansion is planned for the middle Uruguay River, such as the Garabí-Panambi complex, which would impact habitat integrity, flow and river connectivity. These dams would negatively affect populations of migratory fishes, as observed in other Neotropical basins (Agostinho et al., 2016;Pelicice et al., 2018), including endangered species such as Brycon orbignyanus (Valenciennes, 1850), P. corruscans and S. brasiliensis (Pachla et al., 2020). These results point to the need for conservation measures that preserve environmental conditions and ecological functions of rivers in the Uruguay basin. In particular, the maintenance of the hydrological connectivity between upper and lower sections of the middle Uruguay River is crucial to preserve migratory populations, as these fish need to migrate between river channels, different tributaries and the floodplain area located in the downstream section.