Deep Pools: Ecological Sanctuaries for Steindachneridion melanodermatum , a Large Endemic and Endangered Pimelodid of the Iguaçu River

: Steindachneridion melanodermatum , an endemic and endangered species, is the largest catﬁsh in the Lower Iguaçu River basin. Currently, the wild population of this species no longer exists in most of this basin, probably due to the loss of habitat by successive hydroelectric dams. This study evaluated the spatial distribution and abundance of S. melanodermatum in the last free-ﬂowing river stretch of approximately 190 km downstream from the Salto Caxias Dam, upstream of the Iguaçu Falls, as well as some tributaries. Hydroacoustic and ﬁshing campaigns were performed between September 2010 to December 2016 to characterize the preferred habitat. A total of 180 specimens was sampled with higher abundances in a protected area near Iguaçu Falls in the Iguaçu National Park. Twenty-four deep pools were identiﬁed with maximum depths ranging from 5 to 25 m: 21 pools were along the main channel of the Iguaçu River, and three pools were in tributaries. Deep pools are preferential habitats for this species, especially the deep pool called Poço Preto (P23) and its surroundings. Conservation strategies are required to protect these habitats and prevent S. melanodermatum from becoming extinct, such as establishing deep pools as ecological sanctuaries, intensifying the inspection of illegal ﬁshing, and maintaining dam-free tributaries.


Introduction
Steindachneridion melanodermatum is a large catfish (>150 cm in total length) of the family Pimelodidae and is one of the six endemic and endangered species of the genus [1,2]. Its geographic distribution is restricted to the Lower Iguaçu River basin [3][4][5][6], where it is known as surubim do Iguaçu, bocudo, or monjolo. This species is the largest fish in the Iguaçu river basin and is possibly migratory [4,7,8], inhabiting free-flowing river sections with fast waters, rocky bottoms, and high depth [3].

Study Area
The Iguaçu River basin is 72,000 km 2 and borders Brazil (79% in the State of Paraná, 19% in the State of Santa Catarina) and Argentina (2%) [39,40]. The Iguaçu River is one of the leading and largest tributaries of the Paraná River basin, running approximately 1200 km from its source in the Serra do Mar to the confluence with the Paraná River [41]. It is subdivided according to the geomorphological characteristics in Upper Iguaçu (1st plateau), Middle Iguaçu (2nd plateau), and Lower Iguaçu (3rd plateau) [6].
The Lower Iguaçu River is characterized by a high drop with rapids and waterfalls, such as Salto Santiago (40 m), Salto Osório (30 m), and Salto Caxias (67 m) [41], which enabled the formation of five hydroelectric reservoirs in sequence in the main channel of the Iguaçu River [4,40,42]. In the Lower Iguaçu River, there are also the great falls, namely, the Iguaçu Falls (72 m), in a protected area found in the Iguaçu National Park bordering with Argentina, which promotes the geographical isolation of the distinct upstream ichthyofauna from the downstream ichthyofauna.
The study area comprised the last free-flowing river stretch of the Lower Iguaçu basin, which is approximately 190 km long from downstream of the Salto Caxias Dam (1240 MW) to upstream of the Iguaçu Falls ( Figure 1) and exhibits diverse landscape characteristics [37]. In this stretch, the Iguaçu River is embedded with a rocky bed consisting of magnetic basaltic rock that originated in the Mesozoic and presents the remaining fragments of the Atlantic Forest biome. Between the downstream of the Salto Caxias Dam to the upper limit of the Iguaçu National Park, the region is subject to anthropic actions (livestock, agriculture, dam) where a new hydroelectric power plant is located. The new Baixo Iguaçu Hydroelectric Power Plant (350 MW) is located 30 km downstream from the Salto Caxias Dam, whose dam axis is approximately 0.7 km from the upper limit of Iguaçu National Park [36]. However, the samplings for this study were conducted before this dam's construction.

Sampling and Data Analysis 2.2.1. Spatial Distribution and Abundance
Twenty-seven sampling sites were established, covering the main channel of the Iguaçu River and tributaries ( Figure 1 and Table 1), to collect S. melanodermatum samples (Figure 2a). Samplings were undertaken monthly (70 samplings) from January 2010 to January 2011 (13 samplings, 25 locations) and from September 2012 to December 2016 (57 samplings, 27 locations). Some of these sampling sites were in the Iguaçu National Park Conservation Unit (16, 17, 18, 19, 20, 21, 22, 26, and 27) (Figure 2b and Table 1). Between sites 13 and 18, the new Baixo Iguaçu Dam was located. The sampling sites downstream of the Salto Caxias Hydroelectric Plant (1-15 and 22-24) exhibited anthropized areas (Figure 2c) due to livestock, agriculture, and degradation of riparian vegetation on both banks [43] ( Figure 1 and Table 1). On the other hand, the sites that were within the boundary of the INP (16 to 20, 22, and 25) exhibited riparian vegetation that was well preserved in the right riverbank, while the left one impacted, except the sites 21, 26, and 27, which were fully within the INP with both margins being well preserved with dense forest (Table 1).  The fishing gear used to catch the S. melanodermatum were gillnets (total length: 10 m and meshes: 2.5 to 18 cm between opposite nodes), trammel nets (total length: 10 m and meshes: 6, 7, and 8 cm between opposing nodes), and longlines (20 hooks of 9/0 and 20 hooks of 5/0 baited with pieces of fish and bovine heart). Nets and longlines were installed at 3 p.m. on the right bank of the river and inspected every six hours, that is, 9 p.m. that night and 3 a.m. and 9 a.m. of the next day, with a standardized fishing effort at all sampling sites.
The caught fish were measured (total and standard length) and weighed. Of the 180 specimens used in this study, 63 individuals were released and the others (117) were sacrificed for studies of their reproductive biology (already published [36]), diet, and genetic diversity (in preparation). Voucher specimens (eight specimens) were deposited at the Zoology Museum of the Universidade Estadual de Londrina (UEL) (MZUEL 15702, 15703, 15704, 17091, 17092, 17093, 17504, and 17531).
The fish were classified into juvenile and adult using the first maturation size of the species (juvenile: 21.0-43.9 cm) according to Assumpção et al. [36].
Fish abundance was expressed as catch per unit of effort (CPUE) for each piece of fishing equipment. The multigear mean standardization method (MGMS) [44] was used to standardize the efforts of the equipment.
Spatial abundance was assessed using non-parametric variance analysis (Kruskal-Wallis) to test for possible differences in the sampled sites (1 to 27), considering the areas (inside and outside of the Iguaçu National Park) and environments (main channel and tributaries). The CPUEs were transformed into a square root. The program STATISTICA v. 7.0 was used for this analysis.

Deep Pools Mapping
A hydroacoustic campaign was conducted in the study area to identify deep pools (notably, potential habitats of S. melanodermatum). Here we assumed that deep pools in all areas were those with depths lower than −5 m. Depths and geographical coordinates were obtained using sonar (Garmin Echo 100, Garmin International Inc., Olathe, KS, USA) and GPS (Garmin GPSMAP 60CX color map navigator, Garmin International Inc., Olathe, KS, USA), respectively ( Figure 2d). In addition, the location of other deep pools in the Iguaçu River and tributaries was also provided by the Chico Mendes Institute for Biodiversity Conservation (ICMBio), Iguaçu National Park, regarding those of apprehension of predatory fishing of the species.
Bathymetric profiles were performed in deep pools (and around them) where the S. melanodermatum were caught. The transects were performed transversally to the main river channel (Figure 2e), with a spacing interval close to ten meters between the tracks [45]. To obtain the bathymetry data (river depths) and coordinates (longitudes and latitudes), we used a GPS (Garmin GPSMAP 60CX color map navigator) with sonar (Garmin Echo 100). The XYZ dataset that was collected in situ during the campaigns was processed using a geographic information system (GIS) to obtain the bathymetric digital terrain models (MDT); we implemented the triangular irregular networks (TIN) algorithm to represent the bathymetry surface. This interpolation approach is relevant for providing the spatial analysis of each deep pool in the study area.
The deep pools were classified according to their position in the river following Halls et al. [46]. Descriptions of the area and length of the pools, riparian vegetation, and habitat type (riffles, runs/rapids, pools) were also recorded. The area and length data were obtained based on Halls et al. [46]. The longitudinal extension of each pool along the thalweg was used to delineate the area, and the depth was digitized with 50 cm equidistance contours. A polygon was drawn manually for each pool in a geographic information system (GIS) by tracing the shallower depth contour crossed by the outlined pool line. Pool lengths, which were defined as the places where the riverbed profile crossed the zero-crossing line, were calculated as the distance along the thalweg between the start and end points.
The habitat type at each measurement point was subjectively assessed by the survey team [47]. The criteria for this assessment were: riffle-swiftly flowing with a high proportion of its water surface broken; pool-slow flowing with a smooth water surface; and rapids/runs-intermediate between pool and riffle with a wavy water surface.

Deep Pools Characterization
Twenty-four high-depth sites were identified in the study area, with maximum values ranging from 5 to 25 m ( Figure 4). These sites were called deep pools, especially P1 and P23 with the greatest depths (25 and 22 m, respectively). Most of the deep pools (21) were along the main channel of the Iguaçu River and 3 were in tributaries: Capanema

Deep Pools Characterization
Twenty-four high-depth sites were identified in the study area, with maxim values ranging from 5 to 25 m (   In addition to variations in depths of the deep pools P1 to P24, data on the area, length, riparian forest, and habitat type are shown in Table 2. The width of the transverse transect of the bathymetric profile is exhibited for deep pools and the surrounding area where the species were found (P1, P2, P3, P4, P5, P6, and P23) ( Figure 6 and Table 2). The riparian zone in the main channel of Iguaçu River, pools P1 to P5 and P7 to P8, and Capanema River, P6, which were all located outside of the protected area (INP), exhibited riparian forest that had been reduced to tiny fragments on one or both banks occupied by agricultural activities and pastures (Figure 7 and Table 2). In the protected area, the riparian zone exhibited dense preserved native forest on both riverbanks, as in pools P23 (Figure 7), P20 to P22, and P24 (Table 2), or just on one of the riverbanks (P9 to P17 and P19) ( Table 2).
The total area of the sampling transects for mapping the pools ranged from 1.253 to 258.39 ha (Table 2). Deep pools had a low initial depth, varying from 0.40 to 2.0 m, then increasing abruptly, with maximum depths of 5.0 to 25 m (Figures 6 and 7 and Table 2). The bathymetric transects conducted (P1, P2, P3, P4, P5, P6, and P23) had areas ranging from 2.06 (P6) to 258.39 (P23) ha that housed deep pools with areas ranging from 0.451 to 47.255 ha, with an average depth of 6.98 m, a maximum depth of 25.0 m, and different bathymetric profiles (Figures 6 and 7 and Table 2).  Rapids with a riparian zone that was poorly preserved with forest fragments and predominance of pasture on both banks (Figure 7).   Riffle stretches in some shallow areas, bedrock substrate forming extensive areas with large rocks, backwater areas that included foams on the water surface, and an island upstream of the pool (Taquaras' Island). The deep pool was in a meander. Both banks had dense riparian forests (Figure 7).

Discussion
Protected areas are one of the main tools used for the biological conservation of aquatic habitats and species worldwide [48][49][50][51], positively affecting fish productivity in freshwater environments [48]. The higher abundance of S. melanodermatum in the protected area was different from the unprotected area downstream of the Salto Caxias Dam to the upper limit of the Iguaçu National Park. Habitat changes and degradations caused by the hydroelectric plant (downstream of Salto Caxias Dam), high anthropic pressure (agriculture and pasture), and illegal and predatory fishing [9,22,36] along the main channel and tributaries may have contributed to the low abundance in this area. Studies carried out on Lake Tanganyika in Tanzania to assess and compare differences in fishing resources within protected areas and outside (unprotected areas) have also shown that species abundance and diversity were more significant in the protected area [52]. The abundance and diversity of vulnerable and endangered species were greater in the protected areas than the unprotected ones [9,53], showing the benefit of the conservation of the large catfish that are endangered in the Lower Iguaçu River basin.
Our study revealed that S. melanodermatum inhabited deep pools (high-depth sites) in the Lower Iguaçu River and tributaries. The deep pool called Poço Preto (P23) and its surroundings in the Iguaçu National Park can be considered a sanctuary for the species given the highest abundances. Thus, preserving this area is critical for the conservation of this large catfish that is threatened by human actions. Identifying sites that are essential to the conservation of freshwater species, together with information about critical species, improves the representation of freshwater biodiversity in protected areas [28,54].
Deep pools are of high ecological importance for conserving tropical fish species [55]. These locations serve as refuge habitats in the dry season for sedentary and migratory species in the Mekong River [56][57][58][59][60][61], as well as permanent habitats for other species, such as non-native species [61], and as spawning and feeding sites [38,56]. Deep pools are fundamental for sustaining the Lower Mekong River basin fisheries, providing critical spawning and refuge habitats for nearly 200 species of fish, including the Mekong giant catfish Pangasianodon gigas and other critically endangered fish species [46]. In addition, deep pools are an essential link between habitats for migratory fish species migrating over long distances, which use the pools for rest [56,62]. In dry seasons, pools and rapids can become an obstacle to fish trying to swim through them [63].
Considering that habitats with deeper areas attract larger fish [64], S. melanodermatum, the unique large species in the Iguaçu River basin, may depend on deep pools as a refuge habitat during periods when the river level is low during dry periods. Congenerous species also use deep sites, such as Steindachneridion parahybae from Paraíba River [65] and S. scriptum from Uruguai River [66][67][68][69]. In the Paraguay River, the existing pools are used as a refuge by several fish species, such as large pimelodids, especially the jaú (Zungaro zungaro, cited as Paulicea luetkeni) when the river level goes down [70]. During the water level elevation, the free movement of fish is possible, allowing them to select habitats, while at low levels, there is a limitation of access to pools or the main channel in large rivers [71]. On the Iguaçu River, drastic water level fluctuations occur downstream of the Salto Caxias Dam, which are caused by the hydroelectric operations during the week in the early hours of the morning and on the weekends throughout the day. Therefore, the identification of pools by fishers makes it easier to capture the S. melanodermatum, as in P1, which was a deep pool located close to the Salto Caxias Dam. In this sense, some deep pools in the Iguaçu River that are used as refuge sites become traps that are the targets of illegal fishing of S. melanodermatum.
This study is the first to describe deep pools in the Lower Iguaçu River basin, downstream of the Salto Caxias Dam to Iguaçu Falls. The Uruguay River also has stretches with waterfalls, followed by deep pools with depths from 20 to 30 m and bedrock with depths of 1 to 2 m [63]. In addition to their depth, deep pools differ in environmental characteristics, such as water velocity and type of substrate [72], and are classified according to their position on the riverbed, presence of islands, and substrate [46]. In our study, most deep pools were positioned in the meander and the middle of the main channel, with bedrock dominating and occupying small to large areas. Halls et al. [46] highlighted that pool length, area, and volume tend to increase with distance downstream, suggesting that discharge (which increases with growing catchment area) plays a key role in determining overall pool size. In large rivers worldwide, hydroelectric dams have altered the natural flow regime [73][74][75] and impacted sediment transport [73,74]. Reservoirs in the Upper Volga River have decreasing bottom irregularity and the resulting loss of efficiency of its use by fish for feeding and protection against predators, leading to the decline of rheophilic fish species [76]. In addition, the Volga cascade of dams caused the disappearance of iconic species, such as the Caspian lamprey (Caspiomyzon wagneri), Caspian inconnu (Stenodus leucichthys), Russian sturgeon (Acipenser gueldenstaedtii), and the Beluga (Huso huso) in the Upper Volga [75]. Hydroelectric dams change the water flow and transport sediments, threatening the deep pools [46]. The upcoming loss of some deep pools downstream from the Salto Caxias Dam reveals the severe impact that the new dam may have on the preferential habitat of S. melanodermatum. These pools could become silted up through high sedimentation rates over the years, which seriously harms many fish, particularly species that are dependent on deep-water areas for their survival [57,77]. Sedimentation promotes the fragmentation of aquatic habitats, which affects the health and production of fish and reduces primary production [18]. Given this context, we emphasize that S. melanodermatum and other fish species inhabiting deep pools in the main channel of the Iguaçu River could be imperiled.
The Mekong River and its tributaries are considered the last refuges for the megafauna of large fish at high risk of extirpation because of anthropogenic pressures, requiring political actions that include inspection for the conservation of these species in their natural habitat [26]. The occurrence of the S. melanodermatum in the main channel and tributaries of the Iguaçu River reveals their importance in the conservation of species that use deep pools as preferential habitats. Some studies report that tributaries are critical for spawning and recruiting many fish species, especially those that migrate [78][79][80]. The tributaries hold endangered and threatened species and supply various environmental conditions, including access to spawning habitats that provide refugia for early life stages [80]. The protection area needs to enable young individuals to reach sexual maturity within that area to efficiently sustain a fish population while maintaining the diversity of fish species regardless of fishing pressure [81]. The prevalence of adults of S. melanodermatum in the deep pools located in the main channel may indicate that tributaries are an environment for development and growth, and spawning areas of this species still need to be elucidated by identifying the presence of eggs and early larvae [36]. As there is no knowledge regarding movement behavior, spawning sites, and initial development areas of S. melanodermatum, it is essential to protect and preserve not only the Iguaçu River, but also its tributaries, such as the Capanema, Floriano, Gonçalves Dias, and Santo Antônio Rivers, where the species was recorded.
Deep pools were extensively studied and mapped to protect and conserve the fish species of the Mekong River, with the establishment of Fish Conservation Zones (Fisheries Conservation Zones (FCZs)) or sanctuaries, where fishing is not allowed [38,72], with direct effects on the increase in the abundance of several species, both sedentary and migratory [56]. In this way, protected areas benefit the conservation of freshwater fish [48,53,82]. The Poço Preto (P23) deep pool and its surroundings, located in the protected area of Iguaçu National Park, were found to be fundamental for the maintenance of S. melanodermatum since downstream of the Salto Caxias Dam, several anthropic actions further expose the species to threats. This survey and delimitation of priority areas for the spatial distribution of the endemic species highlight the need to create new conservation units in areas of their occurrence to minimize population losses and ensure the protection of aquatic biodiversity. Based on this information, our results indicate where S. melanodermatum are present, particularly showing which are the locations with the highest abundances, and it is also essential to consider that the low abundance of S. melanodermatum in the downstream stretch of the Salto Caxias reservoir are the results of the high local anthropic action. Thus, the definition and creation of environmental protection areas become essential for maintaining and conserving this species.
Establishing areas for habitat protection, such as sanctuaries, protected areas for fish, or prohibiting fishing in certain regions, are measures that have been used recently for habitat conservation for some endemic fish species [26,83]. Therefore, we recommend that Poço Preto (deep pool P23, location 26) and the deep pools located downstream of the Salto Caxias Dam and in the Capanema River, a tributary of the new reservoir and a target of new hydroelectric projects, are established as conservation areas and areas where fishing is prohibited (with intense inspection). The studies carried out by Assumpção et al. [9,36] emphasized the importance of the protected area, namely, the Iguaçu National Park, as well as Poço Preto for the maintenance and conservation of S. melanodermatum and other endemic species in the Iguaçu River basin, highlighting this region as a sanctuary for the Iguaçu River's ichthyofauna. Mitigation and environmental compensation programs should not only restrict lentic sites but the management carried out should address fishery production and aim to conserve biological diversity [8], especially endemic and endangered species.
The greatest abundance of the S. melanodermatum occurred in sites that bordered Argentina in the Iguaçu National Park, which is also a protected area. Studies carried out in the INP in the Argentina border also registered the species in Porto Três Marias, in the Garganta do Diabo walkway, and in the mouth of the Ñandú River, which are all locations near the Iguaçu Falls [84]. Considering that our study area borders Argentina, it is of extreme importance for the environmental agencies of both countries to elaborate specific strategies for the conservation of S. melanodermatum and come to a joint agreement for the inspection of the illegal fishing that occurs in the pools in the INP. The management of transboundary waters is necessary, and the fisheries face challenges that need to be managed [25].
Humans have altered natural river connectivity in multiple ways by placing structures in the longitudinal or lateral flow paths, such as dams and levees [85]. Dams and reservoirs, along with their upstream and downstream propagation of fragmentation and flow regula-tion, are the leading contributors to the loss of river connectivity [85,86]. The construction of new hydroelectric power plants, such as the Baixo Iguaçu HPP hydroelectric power plant that is in operation, will result in the loss of habitats and restrict the area of occurrence of the S. melanodermatum, which may cause a decrease in genetic diversity via fragmentation. Endemic species are particularly vulnerable to environmental changes, increasing the risk of extinction [87]. Steindachneridion melanodermatum is becoming rare and difficult to capture, especially in non-protected areas downstream of the Salto Caxias Dam, where our captures were low. Although it is distributed in these deep pools, unfortunately, it was impossible to make correlations due to the scarcity of fish sampled. Thus, the new Baixo Iguaçu HPP caused the reduction of the dam-free stretch in the Iguaçu River and prevented S. melanodermatum from accessing the deep pools upstream (8) and downstream (16) of the new dam in the main channel and tributaries. This species is seriously threatened with a risk of disappearing if conservation measures are not established [36] in its last refuge. Because of developing new monitoring technologies and the current push for more robust national and international mechanisms for biodiversity management, threatened fish should no longer be neglected in conservation and sustainability commitments.

Conclusions
Given watersheds' current status and prospects, actions are needed to protect these threatened river systems, which provide habitats for endemic, rare, and endangered species, such as S. melanodermatum. In the Lower Iguaçu River, special attention should be given to the main channel and tributaries downstream of Salto Caxias Dam, upstream of Iguaçu Falls, and in protected areas of the last free-flowing river stretch. Steindachneridion melanodermatum prefers deep pools, but its greater abundance occurred in entirely preserved INP regions, evidencing the importance of these sites in conserving the species. Therefore, the leading conservation strategies that are required to safeguard S. melanodermatum are: (i) establishing deep pools as ecological sanctuaries, (ii) the intensification of illegal fishing inspection by environmental agencies, and (iii) the maintenance of dam-free tributaries. In addition, evaluating and monitoring the transport of sediments, silting, and the S. melanodermatum population in the deep pools using hydroacoustic cameras are mandatory to support the management and conservation of this species.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request. Noronha Chiaretto, Dhonatan Oliveira dos Santos, and Fabio Luiz Paetzhodt. Coordination for the Improvement of Higher Education Personnel (CAPES) provided a doctoral scholarship to Lucileine de Assumpção, and the National Council for Scientific and Technological Development (CNPq) provided a Productivity Grant in Technological Development and Innovative Extension (DT) to Sergio Makrakis.

Conflicts of Interest:
The authors declare no conflict of interest.