Ichthyofauna of the Mamanguape river basin, Northeastern, Brazil

The Mamanguape River Basin is located in a peripheral semi-arid area of South America, with its headwaters and middle reaches running through the Caatinga (from wetter to drier) and its lower reaches through the Atlantic Forest. The objective of this study was to inventory the fish fauna of the Mamanguape river basin through a comprehensive sampling and to discuss its ichthyofaunal dominance pattern. Sampling was conducted between 2015 and 2016 at 38 points throughout. The main river course was sampled at 18 fixed points during two expeditions (dry and wet seasons) using traw nets, cast nets, and dip nets, with a standardized effort. The tributaries were sampled during the dry season at 20 points using an adaptaptation of the AquaRAP methodology. A total of 32 freshwater fish species belonging to 26 genera, 16 families and six orders were recorded, predominantly from the order Characiformes and the family Characidae. Astyanax fasciatus had widest distribution and greatest abundance in the basin. Siluriformes were the second most prevalent order, with five species recorded. Cichla cf. monoculus, Poecilia reticulata and Oreochromis niloticus were non-native records for the basin. Apareiodon davisi, an Endangered species, was recorded.


Introduction
The greatest diversity of freshwater fish in the world is found in the Neotropics , and Brazil contains approximately 43% of this diversity (Buckup et al. 2007). Reis et al. (2016) report 5,617 described neotropical freshwater fish species. Given that Reis et al. (2003) listed 4,475 described species, there has been a significant advance in the taxonomic knowledge of neotropical ichthyofauna, with approximately 28% of the ichthyofauna described in the last eleven years. Approximately 100 species have been described per year, and estimates suggest a total of roughly 9,000 neotropical freshwater fish species (Reis et al. 2016). However, there is still a lack of information on the composition and distribution of this fauna in certain regions of Brazil such as the many coastal basins in the Northeast region which is part of the Northeastern Caatinga and Coastal Drainages ecoregion (NCCD, sensu Abell et al. 2008). Sampling efforts in these regions are needed to increase the level of knowledge about the composition of these fish fauna and to support an understanding of their historical patterns, as suggested by Langeani et al. (2009).
The Mamanguape River Basin in the NCCD ecoregion is an example of an area with a poorly known freshwater fish taxocene. A single previous study was conducted in the basin (Rosa & Groth 2004), aiming to evaluate the number of species occurring in Brejos de Altitude (high altitude rainforest enclaves occurring in semiarid zone of Northeastern Brazil, Prado 2003) of the States of Paraíba and Pernambuco. However, no list of fish species to the that river basin was provided. The Mangaguape River basin is one of many small systems that drain to the Atlantic Ocean, and as most northeastern Brazilian coastal drainages, it is influenced by different morphoclimatic zones such as the Caatinga and the Atlantic Forest (Andrade 1997;Langeani et al. 2009). The basin is located in the peripheral region of South America, which is highlighted by  as an area with low species richness but with high degree of endemism. The objective of the present study was to produce a comprehensive freshwater fish survey of the Mamanguape river basin and to provide an understanding on the dominance of fish groups along the basin.

Study area
The Mamanguape River Basin has its headwaters located on the eastern slope of the Borborema mountain range, at a mesoregion named Agreste (characterized by a more humid Caatinga, as defined by Prado 2003), and extends towards the northern coast of Paraíba State, where it flows to the Atlantic Ocean. It drains an area of 3,525.00 km², and is bordered by the Curimataú and Camaratuba River Basins on the North, by the Paraíba do Norte and Miriri River Basins on the West and South, and by the Atlantic Ocean on the East (CERH/PB 2004). Two-thirds of the basin are under the influence of a semi-arid climate, which causes intermittent flow regime in most of the basin. Pools of varying sizes remains along the river bed when the flow is interrupted. The river is perennial near the coast, where the basin is influenced by tributaries from wetter areas (Andrade 1997). In the present study, the basin was classified into reaches (Upper, Middle and Lower) according with the variation of the relief and vegetation as suggested by Andrade (1997).
The Upper Reaches comprise areas with elevations between 600 and 200 meters, and the vegetation is predominantly humid Caatinga, which is more arborescent and typical of Brejos; the Middle Reaches comprise areas with elevations between 200 and 70 meters and the vegetation is dry Caatinga, which is more shrubby and is typical of the Agreste; the Lower Reaches comprise areas with elevations below 70 meters associated to remnants of the Atlantic Forest (excluding estuarine domain).

Data collection
Sampling was performed at 38 collection points; 18 of which were distributed along the main river (named Fixed Points), and 20 were distributed along the northern and southern tributaries of the basin (named AquaRAP points) ( Figure 1, Table 1). The18 points along the main river were sampled during the wet season (September to October 2015) and the dry season (February to August 2016) using a trawl net (4 m long, 5 mm mesh), a cast net (15 mm mesh) and a dip net (5 mm mesh) with a standardized number of throws for each method of collection. The sampling of the tributaries was conducted during the months of February and March of 2016 using the same tools but following the adaptation of Willink et al. (2000)'s AquaRAP methodology: the three collection tool were used when the site's characteristic allowed its use the most frequent circumstance; when the collection site was restricted, only dip net was used. During the AquaRAP survey, some tributaries or their reaches closer to the headwaters were not sampled due to lack of access or pools. Sampling was permitted by ICMBio/SISBIOthroughlicense #48004-1/2015. The specimens were anesthetized with eugenol (10 ml of eugenol in 90 ml of ethyl alcohol, Lucena et al. 2013), fixed in 10% formalin for 2 to 8 days, and then transferred to a 70% ethanol solution (Malabarba & Reis 1987). Sorting and taxonomic identification of specimens and subsequent labeling was performed at the Laboratório de Sistemática e Morfologia de Peixes of the Federal University of Paraíba (LASEP/UFPB). Available material from the study area deposited at UFPB Ichthyological Collection but not collected during the field work was included in the list of species (SR, secondary records of table 2). Specimens were photographed in the field and in the laboratory. After identification, the lots were registered and deposited in the UFPB Ichthyological Collection (number of voucher specimens in Table 2).

Data analysis
Sampling effectiveness was evaluated using a rarefaction curve (Colwell et al. 2012). Richness was assessed using bootstrap (Efron 1979) estimator. A heatmap (Singh et al. 2011) was generated using a species presence/absence matrix to provide a visual profile of occurrence throughout the basin.

Results
A total of 26,805 specimens of 29 species, 24 genera, 14 families and six orders were collected (Table 2, figures 2 and 3). Callichthys callichthys, Gymnotus cf. carapo, and Hemigrammus unilineatus, were not sampled in the present study, but had been previously recorded in the UFPB Ichthyological Collection. Thus, the total number of taxa recorded from the basin is 32 species, 26 genera, 16 families and six orders.
Visual inspection of the rarefaction curve shows that the asymptote was reached (Figure 4), indicating an effective sampling of the studied area. The richness estimator yielded numbers of species approximately equal to that recorded in the field (29): boot, 28.8. Figure 5 shows that the different sampling approaches used in different areas of the basin produced similar results. The pattern of species dominance observed in the main river, which was sampled at fixed points during the wet and dry seasons, was similar to that of the tributaries, which were sampled by AquaRAP during the dry season.
Seven families were recorded from the order Characiformes, comprising 15 species and 52% of the species collected (including introduced species); one family was recorded from the order Cichliformes, comprising five species (two introduced) and 17% of the registered species; and two families were recorded from the order Siluriformes, comprising four species (12.5%) ( Table 2). Characidae was the family with the highest species richness (nine species or 31%), followed by Cichlidae (five species, 17%), Loricariidae (three species, 9%) and Poeciliidae (two species, 6%). Only one species was recorded from each of the other families, each representing 3% of the total species richness ( Table 2).
Eight individuals of Apareiodon davisi, a threatened species (Endangered, according to Portaria nº 445, December 17, 2014) (Brasil 2014), were recorded at five sampling points in areas of perennial flow or high to moderate flow (FP18, AP4, AP11, AP12 and AP14) and were recorded in all three reaches of the basin. Three exotic species were recorded to the Mamanguape River Basin: two from the family Cichlidae, the tucunaré Cichla cf. monoculos and the tilápia (or Xilapo, Xilapa) Oreochromis niloticus, and the guarú (or Barrigudinho) Poecilia reticulata (Poeciliidae). Two undescribed species of the genus Parotocinclus were recorded. Parotocinclus sp. 1, sampled at Upper and Middle Reaches (FP3, FP6 and FP7) of the main river, and Parotocinclus sp. 2, sampled at a single point in the Middle Reaches, the Jacaré River (AP11), one of the major tributaries of the basin. Apareiodon davisi, Cheirodon jaguaribensis and Hypostomus pusarum are endemic to the NCCD (Rosa et al. 2003  vivipara, Serrapinnus heterodon, Serrapinnus piaba and Steindachnerina notonota are distributed throughout the drainage network and form the main cluster in the heatmap ( Figure 5, cluster 1). In contrast, the heat map shows that some species were restricted to individual collection points such as Parotocinclus sp. 2, recorded at point AP11, and Eleotris pisonis, collected at point FP18 (clusters 6 and 7, respectively). Rhamdia quelen and Synbranchus cf. marmoratus, although collected at only four and six points, respectively, were recorded in all three Reaches of the basin (clusters 2 and 6, respectively). Awaous tajasica occurred only in the lower reaches (cluster 4), whereas Parotocinclus sp. 1 occurred in the upper and lower reaches (cluster 2). The introduced species Cichla cf. monoculus was recorded at only two points in the lower reaches (cluster 4), and Oreochromis niloticus was recorded at six points in the middle and lower reaches (cluster 5). The primary group produced by the heatmap (cluster 1) consisted of fifteen species belonging to eight families: Characidae (6 species), Cichlidae (2 species), Poeciliidae (2 species), Anostomidae (1 species), Crenuchidae (1 species), Curimatidae (1 species), Erythrinidae (1 species) and Loricariidae (1 species). The family Characidae had the broadest distribution and was the most common family in the entire basin, with one or more species present at 35 of 38 collection points. At the Fixed Points, which had more regular and extensive sampling, Characidae had an average of 30% of the species occurring in the three reaches. The species Astyanax bimaculatus, Astyanax fasciatus, Compsura heterura, Serrapinnus heterodon and Serrapinnus piaba, which occurred in all three reaches, co-occurred at 14 points (36.8% of the points). The species Cheirodon jaguaribensis, Hyphessobrycon parvellus and Hemigrammus rodwayi co-occurred with one or more species of Characidae at seventeen, ten and six points, respectively, whereas Hemigrammus marginatus co-occurred with other Characidae at five points, but only in the lower reaches ( Figure 5).

Discussion
Although two sampling methods were employed (Fixed Points in both wet and dry seasons vs. AquaRAP in the dry season) and performed in different areas of the basin (main river vs. tributaries), the results showed a similar pattern of species occurrence for both methods in both areas of the basin. This reinforces the existence of such patterns of distribution, as indicated by the heatmap. The pattern of species occurrence in the main river (higher frequency by collection point) may be due to the greater sampling effort: all gear types were used at all collection points, and a higher number of throw per gear type vs. selective gear used depending on the environmental structure. Moreover, the total mainstream sampling área was greater (greater volume of water sampled) than that of the tributaries, given the greater channel width and the formation of larger pools in themainstream.
The species grouped in the main cluster (cluster 1) were generalists, as is common for small fishes (Pelicice & Agostinho 2006, Fiori et al. 2016, or piscivorous such as Hoplias malabaricus (Rodrigues et al. 2017). Prochilodus brevis, a specialist detritivore (mud eaters - Silva et al. 2010), occurred only in the lower and middle reaches of the basin. Ramos (2012), citing the River Continuum Concept (Vannote et al. 1980), suggested the higher occurrence of detritivores, such as those of the genus Prochilodus, should be expected in the lower reaches of the basin, which have a larger amount of accumulated material in the river bottoms. Consistent with this prediction, the heatmap shows that Prochilodus brevis is more abundant in the lower reaches. Awaous tajasica and Eleotris pisonis were among the less common species and were collected only in the Lower Reaches of the main river. These two species are widely distributed in the coastal region of South America and are of marine-estuarine origin (Pezold & Cage 2002, Sarmento-Soares 2007. Poecilia reticulata and P. vivipara are widely distributed, the first was introduced into neotropical freshwater environments by release from aquariums or from use as controller of biological vectors (Ramos 2012). Rosa et al. (2003) emphasized that Siluriformes dominate over Characiformes in the ecoregions under the influence of the Caatinga (101 vs. 89 species), a statement also made by Lima et al. 2017 (143 vs. 132). However, it differs from the results of the present study, with 16 species distributed in seven families of Characiformes, whereas the Siluriformes are represented by five species belonging to three families (Callichthyidae, Loricariidae and Heptapteridae). In all the previously surveyed basins in NCCD, it was recorded a similar pattern of species composition in which the number of Siluriformes is smaller than those of Characiformes: in the Curimataú basin (12 characiform vs. one siluriform species, Ramos et al. 2005), in the Seridó basin (21 vs. 6, Silva et al. 2014), in a general survey of the ichthyofauna of the Rio Grande do Norte State basins (21 vs. 6, Nascimento et al. 2004) and in a survey of the ichthyofauna of the Mundaú River (16 vs. 6, Teixeira et al. 2017); the same was recorded in Parnaíba ecoregion (59 vs. 48, Ramos et al. 2014) and in Northeastern Mata Atlântica ecoregion (91 vs. 75, Camelier & Zanata, 2014). Only in the    São Francisco ecoregion Siluriformes dominate over Characiformes (85 vs. 77 species) . The above observations suggest that instead of the pattern of species dominance recorded in the whole Caatinga (Rosa et al. 2003, when comparing the dominance of Siluriformes and Characiformes in each particular basin of this domain, no siluriform dominance is recorded. In truth, the particular basins seem to include a relatively larger number of endemic siluriform than endemic characiform species to each of them, explaining the apparent dominance of Siluriformes when the numbers of the Caatinga basins are put together. However, it should be only a differential effect of the peripheral high level of endemism of South America ) which is larger in Siluriformes than in Characiformes in each particular basin of the Caatinga domain, although the dominance pattern of the basins running in a large part of the Caatinga reveals a Characiform dominance. Among the recorded species, only Apareiodon davisi is listed as endangered (Brasil 2014), with distribution restricted to points with perennial flow. This species, and the entire family Parodontidae, prefers environments with more turbulent waters, where it lives in proximity to the substrate (Graça & Pavanelli 2007). Ramos et al. (2005), Medeiros et al. (2006;2010) and Costa et al. (2017) recorded this species in lentic environments at Seridó and Paraíba do Norte rivers, which are rocky residual environments within intermittent rivers and are probably lotic during the flood period, which is consistent with the observations reported by Graça & Pavanelli (2007). This species was considered endemic to the coastal drainages of the mid-northeastern Caatinga (Rosa et al. 2003), which corresponds to the Caatinga region of the NCCD. The occurrence of A. davisi is currently recorded for the Jaguaribe, Piranhas-Açu, Paraíba do Norte, Curimataú, Ipojuca (Reis et al. 2003, Ramos et al. 2005, ICMBio 2017 and Mamanguape (present study) rivers. Characidium bimaculatum, previously considered to be endemic to NCCD (Rosa et al. 2003), is known from records in the São Francisco ecoregion (Melo & Espindola 2016) and possibly from Parnaíba basin (Ramos et al. 2014, listed as Characidium bimaculatum).
Some surveys of the freshwater fish of the NCCD ecoregion have been published, usually with limited sampling efforts or partial evaluating of the fish fauna composition. Rosa & Groth (2004), as mentioned above, surveyed a limited area of the Mamanguape River Basin but did not report the presence or abundance of species. Ramos et al. (2005) recorded 22 species in the Curimataú River Basin. The same richness (22 species) was observed by Torelli et al. (1997) and Gomes-Filho & Rosa (2001) in the Gramame River drainage (Paraíba State). Silva et al. (2014) recorded 35 species in the middle reaches of the Piranhas-Açu River Basin (Paraíba and Rio Grande do Norte States); Paiva et al. (2014) recorded 13 species in the Pratagi River Basin (Rio Grande do Norte State). The richness recorded in the present study, (32 species) apparently reflects a good estimate of the local fish fauna composition, as indicated by the estimators used. Moreover, the sampling showed a consistent pattern of ichthyofaunal composition for both the main course and its tributaries ( Figure 5) despite different sampling methods.  state that the periphery of South America is characterized by low species richness and a high level of endemism (in contrast to central South America's core of high diversity and low endemism). Rosa et al. (2003) mentioned 30 endemic species occurring in the named Nordeste Médio-Oriental ecoregion (sensu Rosa et al. 2003). When characterizing this ecoregion, the authors describe it as a hydrographic region of the Caatinga domain (Rosa et al. 2003: 140), although they acknowledge that species can occur outside of this domain, in upper and low stretches of the rivers, and highlights that endemism is stated for the ecoregion, not for the domain (Rosa et al. 2003: 137).  listed 38 endemic species in the NCCD. Among the species considered endemic to this ecoregion Apareiodon davisi, Cheirodon jaguaribensis and Hypostomus pusarum (according to Rosa et al. 2003, Reis et al. 2003, and Buckup et al. 2007 were recorded in the Mamanguape River basin. Reis et al. (2016) treated the NCCD and Parnaíba ecoregions as a unit termed Northeast Atlantic basin complex (ecoregions 325 and 326 of Abell et al. 2008, respectively) based on a low species diversity (<200 species). The authors documented 65 endemic species in their proposed basin complex. However, the authors did not cite Ramos et al. (2014), which recorded 146 species in the Parnaíba basin, 54 of which endemic. Addionally, Silva et al. (2015) surveyed the Gurgueia river, the main tributary of the Parnaíba basin. Adding their list to the survey carried out by Ramos et al. (2014) of Parnaíba basin, established its ichthyofauna as composed of 152 species. Rosa et al. (2003) cite a diversity of 82 species, 30 endemic to the NCCD, whereas  proposed to the same region a list of 88 species, 38 of which endemic. Recent descriptions have added six more endemic species to the ecoregion: Anablepsoides cearensis (Costa & Vono 2009) bringing the total to 94 species, 44 of which are endemic. In addition to the number of occurrence and endemic species of the Parnaíba (152/54; Ramos et al. 2014;Silva et al. 2015) and NCCD ecoregions (94/44, present compilation), the number of species is 246 with 98 endemic species. Thus, the two ecoregions should be treated separately, based on the criterion used by Reis et al. (2016). In this case, the number of endemic species is considerable, which corroborates the results of .
In the field, it was possible to observe the direct result of human action on the environment, primarily at the margins of rivers and streams. The suppression of riparian vegetation, the development of agricultural practices, the dredging of sand from the river bed and the breeding of animals were the most evident human activities. Elimination of riparian vegetation was observed throughout the Mamanguape basin, primarily in the middle and lower reaches ( Figure 6). This habitat modification, besides of other harms imposed to the ecossistem, opens the area for human activities, which further compromises the stability of the system as previously mentioned by Andrade (1997). The damage to the river's health and to the organisms inhabiting it caused by the removal of riparian vegetation is already well explored in the literature (Sweeney et al. 2004;Casatti et al. 2009;Teresa & Casatti 2010;Cruz et al. 2013).

Author Contributions
Katherine Morais Porto Viana, Raizze da Costa Avellar: contributed with data collection.
Leonardo Oliveira Silva: contributed with data collection, manuscript preparation, data analysis and interpretation.
Robson da Costa Tamar Ramos: contributed with the manuscript preparation, data analysis and interpretation.
Telton Pedro Anselmo Ramos: contributed with data collection, and data analysis and interpretation.
Yuri Gomes Ponce Carvalho-Rocha: contributed with data collection, and data analysis and interpretation.