Stream ichthyofauna of the Tapajós National Forest, Pará, Brazil

Abstract The fish fauna of freshwater streams in the Tapajos National Forest was surveyed and a list of species is presented. The sampling was conducted from 2012 to 2013 during the dry season. Fish were collected with dip nets and seine nets in 22 streams of 1st to 3rd order. Sampling resulted in 3035 specimens belonging to 117 species, 27 families and six orders. The most abundant species were Bryconops aff. melanurus, Hemigrammus belottii, and Hemigrammus analis. Four undescribed species were recognized, one of which is known only from the area of this study. A significant dissimilarity was observed in fish species composition among drainage systems. This is the first survey of the stream ichthyofauna in the Tapajós National Forest, and it presents relevant information for future studies and decision-making in the management and conservation of fish fauna in this conservation unit.


Introduction
The Neotropical region has the richest and most diverse fauna of freshwater fishes in the world, reaching a number of more than 5400 valid species (Reis 2013) and estimates of the final number of more than 8000 species (Reis et al. 2016). Among its watersheds, the highest species richness is located in the Amazon River basin (Santos and Ferreira 1999;Reis et al. 2003), where the number of fish species remains undefined, particularly those inhabiting small streams. In these environments, despite having low primary production (Walker 1990), a rich fish fauna is supported, composed mainly of small-sized fish species (Henderson and Walker 1986;Castro 1999).
The main objective of the Tapajós National Forest (FLONA Tapajós), founded in 1974, has focused on the multiple use of forest resources and scientific research (SNUC 2000). However, studies of the fish fauna in streams are still needed. Collecting data on species composition in restricted geographical areas, such as conservation units, is an important initial step in decision-making related to the management of fish communities and conservation. Thus, the present study aimed to provide a list of fish species and to test difference in fish species composition among different drainage systems in the Tapajós National Forest.

Study area
The Tapajós National Forest (FLONA Tapajós), located in western Pará State, approximately 3°24'S, 55°03'W ( Fig. 1), holds an area over 527,000 hectares encompassing part of the Aveiro, Belterra, Placas, and Rurópolis municipalities (ICMBio 2014). The FLONA Tapajós is bordered in the west by the Tapajós River, in the east by the highway BR-163, connecting Cuiabá (Mato Grosso State) to Santarém (Pará State), in the south by the Cupari River, and in the north its border is perpendicular to intersection 65 km on BR 163 North. Streams in the FLONA Tapajós streams flow directly in the Tapajós River or drain into two distinct river systems − Curuá-Una and Cupari rivers.

Data collection
Twenty-two streams of 1 st to 3 rd order were sampled (Fig. 2) during the dry season from September 2012 to November 2013. Nine streams belong to the Curuá-Una river system, six drain into the Cupari River, and seven flow directly in the Tapajós River (Table  1). Fish sampling followed a part of the protocol proposed by Mendonça et al. (2005), in which a 50-m section of each sampled stream was blocked with fine-mesh nets (5 mm between opposite knots). After blocking a section, two collectors were actively sampling for about two hours using dip nets and seine nets.
Specimens were anesthetized in a solution containing eugenol (clove oil), fixed in 10% formalin solution, and subsequently transferred to 70% ethanol. They were counted and identified to the lowest possible taxonomic level. Species were identified with the use of dichotomous keys for different taxonomic groups (e.g. Géry 1977;Kullander 1986;Vari 1992;Buckup 1993;Mago-Leccia 1994;Netto-Ferreira et al. 2009;Oyakawa and Mattox 2009;Caires and Figueiredo 2011;Peixoto et al. 2013) and diagnoses of species (e.g. Zanata et al. 2009;Marinho and Langeani 2010) as well as with the assistance of fish taxonomy experts. The use of the terms "cf". "aff.", and "sp". follows Bengtson (1988). Taxonomic classification follows Reis et al. (2003). Voucher specimens are deposited in the Fish Collection of Universidade Federal do Oeste do Pará (UFOPA-I) (Appendix 1). Fish were collected under ICMBio license number 35649-2.

Data analysis
An overall estimate of the fish species richness was calculated by means of the Jackknife 1 method (Krebs 1999), utilizing estimatS 8.2 (Cowel 2009). Alpha diversity was estimated by the Shannon-Wiener index (H') (Shannon and Weaver 1963). To test difference in fish species composition among drainage systems, an analysis of similarities (ANOSIM) was applied with 999 permutations, using Bray-Curtis as a distance metric to measure the degree of dissimilarity between sites based on quantitative data (abundance) and Jaccard index for qualitative data (presence/absence of species). The analyses were done with the software PAST (Hammer et al. 2001).
The most abundant species were Bryconops aff. melanurus and Hemigrammus belottii (332 specimens each, 10.9% of the total species recorded), Hemigrammus analis  Table 3. The distribution of most species was related to drainage basins; from 117 species recorded, 38 were restricted to streams flow directly into the Tapajós River, 47 were collected only in streams draining into the Cupari River basin, and 11 were recorded only in streams draining into the Curuá-Una River basin. Six species were common to streams of the Curuá-Una and Cupari river drainages. One species was shared among streams flow directly into the Tapajós River and streams draining into the Curuá-Una River; thirteen species were shared among streams flow into the Curuá-Una and Cupari rivers, as well as streams draining directly into the Tapajós River (Fig. 4).

Discussion
The fish fauna of Tapajós National Forest, as well as the lower Tapajós River, is one of the most understudied and undersampled among aquatic systems in the Amazon drainage and so far all species found during the survey represent new records for the studied area. The number of species recorded (117) is one of the highest among known fish faunas in streams of the 1 st to 3 rd order in the Amazon drainage (e.g. Mendonça et al. 2005;Montag et al. 2008;De-Oliveira et al. 2009;Dias et al. 2009;Barros et al. 2011). However, the richness of species should be higher and reach up to 183 species. Therefore, more efforts should be employed in surveying the fish fauna of streams in the FLONA Tapajós. The Neotropical fish faunas are characterized by the predominance of species from the orders Characiformes and Siluriformes (e.g. Angermeier and Karr 1983;Arbeláez 2004;Baumgartner et al. 2006;Arbeláez 2008;Scarabotti et al. 2011;Pedroza et al. 2012;Raiol et al. 2012;Claro-García and Shibatta 2013;Volcan et al. 2013, Ramos et al. 2014. Characiformes is one of the largest orders of fishes with at least 2000 valid species (Eschmeyer 2015). In the Neotropical region, Characiformes, Siluriformes, and Gymnotiformes, or Ostariophysi, constitute about 77% of the freshwater fish fauna; however the order Perciformes has over 515 freshwater species, in some cases alternating with Gymnotiformes as the third richest order (Albert et al. 2011). In the present study, Perciformes presented three species more than Gymnotiformes.
If families are concerned, the largest number of species in the Neotropical region is contained in Characidae and Loricariidae (Schaefer 1998); however, similar to this study, other faunistic surveys in small streams of the Amazon drainage revealed an inversion in the number of species in the families Cichlidae and Loricariidae (e.g. Mendonça et al. 2005;Barros et al. 2011).
The highest values of richness were observed at sampling stations 8, 9 and 21 (Table 3). Stations 8 and 9 were at river sections characterized by the greatest depth and width. In streams, studies indicate that an increase in species richness is positively related to the habitat complexity and shelter availability as well as current velocity and stream size (Garutti 1988; Meffe and Sheldon 1988;Abes and Agostinho 2001;Súarez and Lima-Junior 2009). In the Neotropical region, substrate, depth and current speed are among the most important physical features, and a combination of such environmental features produces a mosaic of microhabitats, which can explain the downstream increase in species richness (Casatti 2005).
Station 21 is near to the mouth of a stream draining into a lake, and its high values of richness is resulted of the presence of species typically recorded near lakes such as Catoprion mento, Hemigrammus analis, H. levis, H. hyanuary, H. stictus and Dicrossus maculatus (Siqueira-Souza and Freitas 2004;Lima et al. 2013;Kullander 2011). Four new species were recorded, Curculionichthys sp. n., Aspidoras sp. n., Hyphessobrycon sp. n., and Moenkhausia sp. n, the Aspidoras being known only from the present study. Some specimens received provisional identification with the use of "cf. ", "aff.", or "sp.", which may be indicative of the recognition of other new species after more refined analysis, or may even indicate insufficient research for some taxonomic groups (e.g. Ancistrus and Apistogramma).
The existence of dissimilarity in fish species composition of different, however geographically close, drainage systems within the Tapajos National Forest indicates that geographic isolation coupled with environmental characteristics is responsible for the structuring of fish communities, in accordance with observed by Schleuter et al. (2012) in temperate regions and Barros et al. 2013, in tropical streams. Furthermore the smaller drainage basins can significantly influence the stream fish assemblages composition (Mendonça et al. 2005;Barros et al. 2013) since headwaters streams often support exclusive species that do not occur in the river system, allowing constitute single assemblages that are fundamental to compose the regional fish diversity (Paller 1994;Meyer et al. 2007) and must be priority included in units conservation planning for freshwater systems.

Authors' contribution statement
CSO, ACC and FRR collected the data, identified the species, filled the database and wrote the text.