According to the Living Planet Index by the World Wide Fund for Nature, freshwater ecosystems and biodiversity are considered among the most threatened on the planet, with an average decline of approximately 83% in freshwater organism populations since 1970 (Cooke et al., 2024). Indeed, freshwater biodiversity is disproportionately threatened and underprioritized compared to marine and terrestrial organisms, despite their intrinsic value and the provision of essential ecosystem services (Rees et al., 2021). Harrison et al.(2018) labeled this global phenomenon as “the freshwater biodiversity crisis”. Many of these estimates are based on well-known groups of aquatic vertebrates that have been monitored over decades. However, this situation cannot be applied to Amazonian stream fishes, whose biodiversity is still underestimated, and numerous areas represent spatial gaps that align with the seven shortfalls outlined by Hortal et al. (2015). Consequently, accurately predicting the magnitude of the freshwater biodiversity crisis for Amazonian stream fishes remains challenging.

The Amazon River basin boasts the most diverse regional assemblage of freshwater fishes globally, with 2,716 valid species (1,696 of which are endemic) representing 529 genera, 60 families, and 18 orders (Dagosta, de Pinna, 2019). Since streams harbor approximately 50% of the Amazonian fish fauna (Junk et al., 2007), these water bodies harbor approximately 1,400 species, which stands the Amazon basin as the world’s largest and most biodiverse basin for stream fishes as well (Frederico et al., 2021). As with any other freshwater system, Amazonian stream fishes face multiple threats, including deforestation for agriculture and livestock raising, dam construction, navigable waterway development, urban and mining pollution, and overharvesting for the aquarium trade (Vieira et al., 2008; Castello et al., 2013; Pelicice et al., 2021; Santana et al., 2021).

The shortfalls in knowledge about biodiversity need to be carefully recognized and quantified to make accurate predictions about how it might change in the future (Hortal et al., 2015). This need becomes even more urgent when it comes to megadiverse regions such as the Amazon (Carvalho et al., 2023). To complement the classical descriptors of biodiversity such as composition of species, abundance, species richness, and evenness (Magurran, 2011), there are the taxonomic distinctness indexes, that measure the average degree to which individuals in an assemblage are related to each other and how is the evenness of this relation (Warwick, Clarke, 1995; Clarke, Warwick, 2001; Zintzen et al., 2011).

In the Amazon region, particularly in the Pará State, essential drainages such as Tapajós, Xingu, and Trombetas contribute to the Amazon River and comprise 982, 821, and 494 fish species, respectively (Jézéquel et al., 2020). These drainages encompass the Tapajós-Juruena, Xingu, and Amazonas Guiana Shield ecoregions, respectively (FEOW, 2024). Notably, a large part of these territories contains officially protected areas, in the categories of full protection or sustainable use (MMA, 2024). However, it is important to note that protected networks designed with a focus on terrestrial systems and species are not good enough surrogates to adequately protect the remarkably rich diversity of Amazon stream fishes and these areas are ineffective in safeguarding Amazonian stream fishes (Dagosta et al., 2020; Frederico et al., 2021).

Therefore, this research aims to address knowledge gaps in the taxonomy and distribution of stream fish in the Eastern Amazon, with a particular focus on protected areas in the Pará State. The objectives encompass: (i) describing the ichthyofauna of streams associated with four conservation units in Pará State, (ii) detailing patterns of abundance, richness, diversity, dominance, occurrence, and taxonomic distinctness in each area, and (iii) evaluating the contribution of the present inventory compared to existing records in scientific collections encompassing the ichthyofauna of streams in Pará State.

Study areas. To conduct this study, a dataset was generated from four field expeditions to the state of Pará within the Amazon biome. The rainy season in this region occurs from December to May, and the dry season from June to November (Moraes, 2005). Because of this, sampling procedures were carried out during the dry periods (of 2021, 2022, and 2023) to avoid the excessive rainfall during the rainy season that may cause changes in the physical and chemical properties of the Amazonian streams and affect the logistics of the sampling procedures (Oliveira-Junior, Juen, 2019). During field expeditions data were systematically collected from 99 streams situated both inside and in the surroundings of the Parque Nacional da Amazônia, Parque Nacional do Jamanxim, Floresta Nacional Saracá-Taquera, and Reserva Biológica Nascentes da Serra do Cachimbo (Tabs. 1, S1; Figs. 1, S2). The selected streams represent diverse types of protected areas from two ecoregions (Tapajós-Xingu and Amazonas Guiana Shield; FEOW, 2024), encompassing a gradient of environmental conditions within and around these designated protected areas or zones.

Fish sampling and identification. In each stream, a 150-meter-long stretch was designated for fish collection. Following the methodology outlined by Prudente et al. (2017), a sampling effort of six hours was allocated for each stream, divided among three to four collectors, who used metal hand nets (55 cm in diameter, 2 mm mesh in opposite nodes) to catch fishes on the bottom, closer to the banks, and in the water column. Collected specimens underwent anesthesia with eugenol, were fixed in 10% formalin, and subsequently preserved in 70% ethanol. Species identification was carried out using specialized taxonomic keys (Géry, 1977; Van der Sleen, Albert, 2018), with identities confirmed by specialists.

Data analysis. Based on the species list compiled, the threatened categories were verified according to the Brazilian Red List (ICMBio, 2022). To complement the first objective and assess the similarity in assemblage structure among areas, a Similarity Analysis (ANOSIM) was conducted with all species of the 99 streams using the Bray-Curtis coefficient (Anderson et al., 2008) and 9,999 permutations, with the areas (PARNA AM, PARNA JX, FLONA ST, REBIO SC) as the grouping factors. ANOSIM is a non-parametric test facilitating the comparison of sample sets (in this case, streams from each area) based on similarity within and between groups (Anderson et al., 2008). The Bray-Curtis coefficient was selected for its appropriateness in handling abundance data (Clarke, Gorley, 2006). Among collected specimens, 35 were juveniles (belonging to the Characidae, Iguanodectidae, Loricariidae, and Rivulidae families), lacking evident diagnostic characters for effective taxonomic determination. Consequently, these juveniles were excluded from quantitative analyses. To visually represent the similarity between streams in a biplot, a non-Metric Multidimensional Scaling Analysis (nMDS) was performed. Both ANOSIM and nMDS analyses were carried out using the PRIMER 6.0 software (Clarke, Gorley, 2006).

For the second objective, the abundance and frequency of occurrence were determined for each species within the set of streams associated with each protected area. Shannon-Wiener diversity in each stream was calculated using the classical formula H'=−∑pi*logpi , for log base 2, where Pi represents the proportion of species i in each stream (Clarke, Gorley, 2006). Dominance in each stream was assessed using the Berger-Parker index, calculated with the formula d=n/nmax , where n represents the abundance of the most abundant species, and nmax is the total abundance in each stream (Magurran, 2011). The effective number of species, known as Hill numbers or the true diversity for each stream, was also calculated. Hill numbers quantify the number of species in a sample, depending on the common and rare species and incorporate relative abundance and species richness based on the value of the “q” exponent (Chao et al., 2014). Two Hill’s numbers were calculated: q = 0, which is insensitive to the differences in the relative abundances of species, and indicates the effective species richness; q = 1, which includes all species with a weight exactly proportional to their abundance in the community and can be interpreted as the effective diversity (Chao et al., 2014). Diversity indexes were calculated in the PRIMER 6.0 software (Clarke, Gorley, 2006). The species richness estimation for each area was determined by using two nonparametric richness estimators based on sampling coverage (Lee, Chao, 1994), namely ACE (Abundance Coverage Estimator) and ICE (Incidence Coverage Estimator), through the software EstimateS 9.10 (Colwell, 2013). The completeness at each protected area was calculated based on the ratio between observed richness and the total estimated richness obtained by ACE and ICE (Morales-Martínez et al., 2021).

Fish diversity at various taxonomic levels was assessed using two diversity measures: average taxonomic distinctness and variation in taxonomic distinctness (Clarke, Gorley, 2006). Average taxonomic distinctness (also known as Delta+ or AvTD) represents the average taxonomic path length between all pairs of species, indicating whether a species list from a specific locality shares the same taxonomic distinctness structure as the regional pool of species (Clarke, Gorley, 2006). Variation in taxonomic distinctness (also known as Lambda+ or VarTD) is the variance of these pairwise path lengths, reflecting the unevenness of the taxonomic structure in such an assemblage (Clarke, Warwick, 2001). Both Delta+ and Lambda+ are based on an established taxonomic hierarchy into species, genera, families, and orders, known as a master list (Marchant, 2007). For this study, the master list was produced with all species, genera, families, and orders from the 99 streams, while the species list comprises a matrix of species abundance from the four protected area. Both lists were combined in the TAXDTEST routine of the PRIMER 6.0 software (Clarke, Gorley, 2006) to obtain Delta+ and Lambda+ values per stream. Subsequently, these values were plotted against the number of species recorded at each stream, with 95% confidence limits for different numbers of species, resulting in funnel plots. To test whether Delta+ and Lambda+ varied significantly among the four protected areas, a Permutational Multivariate Analysis of Variance (PERMANOVA, Anderson et al., 2008) was performed. This method is suitable for testing data that do not meet the normality assumptions and can utilize any measure of similarity. Among the analysis options, Euclidean distance with 9,999 unrestricted permutations was selected, using the different areas as a fixed factor. These calculations were performed in the PRIMER 6.0 software (Clarke, Gorley, 2006).

To accomplish the third objective, records of freshwater fish in Pará from 11 scientific collections (Tab. S3) available in the SpeciesLink network (CRIA, 2024; accessed on 10/Jan/2024, specieslink.net) and in the ‘Sistema da Informação sobre a Biodiversidade Brasileira’ (SiBBr, 2024; accessed on 12/Jan/2024, https://www.sibbr.gov.br/) were downloaded. The initial data cleaning was based on the following criteria: (i) records from Ostariophysi families that are not typical from streams, according to to Buckup (2021) were removed; (ii) records without geographic coordinates, records from dams or lagoons were removed; (iii) records unidentified at the genus level were removed. After this process, 18,230 records remained. To keep only records from watersheds of similar size to our fish samples, a second round of data cleaning was conducted. To this, watersheds shapefiles for the state of Pará from the HydroSHEDS spatial database (Lehner et al., 2008), at level 10, were obtained. Watersheds larger than 4,300 hectares (the approximate area of the largest basin sampled in the current inventory, see Tab. S1) were excluded from the analysis. This resulted in 2,187 records. Because 38% of this total had undetermined specific epithet (cf., aff., sp.), this comparison was restricted to the generic level, adhering to the validity of generic epithets as outlined in Eschmeyer’s Catalog of Fishes (Fricke et al., 2024a).

Data availability. Fish collection is available in the Species Link system (http://splink.cria.org.br/) and in the TAOCA database (www.taoca.net).

In the 99 surveyed streams, a total of 13,602 specimens were collected, belonging to 176 species from 29 families and seven orders (Tab. 2). The small tetras from the Characidae family constituted 33% of all species and 39% of all specimens (Figs. 2, S4). Non-native specimens were not recorded. One threatened species according to the Brazilian Red List was registered: the Vulnerable (VU) Harttia dissidens Rapp Py-Daniel & Oliveira, 2001. This species was represented by three specimens collected in the surroundings of the PARNA JX (site PJF19).

For occurrence data (Tab. 2), small tetras were present in more than half of the PARNA AM streams, including Hemigrammus cf. bellottii, Moenkhausia cf. collettii, Hemigrammus cf. ocellifer, and the cichlids Apistogramma sp. and Aequidens cf. tetramerus. In the PARNA JX, the group of species occurring in more than half of the streams was more diverse, encompassing Characidium aff. zebra, Hoplias cf. malabaricus, Hemigrammus cf. ocellifer, Moenkhausia aff. melogramma, Moenkhausia oligolepis (Günther, 1864), Ancistrus sp. 3, Apistogramma sp., and Saxatilia aff. lepidota. In the FLONA ST, Copella callolepis (Regan, 1912), Pyrrhulina aff. brevis, Gymnotuscoropinae Hoedeman, 1962, Helogenes marmoratus Günther, 1863, and Laimosemion dibaphus (Myers, 1927) were present in over half of the streams. Finally, in REBIO SC, only one undetermined species of Knodus was present in more than half of the sampled streams.

The abundance pattern across the different sets of streams was distinct among the four areas (Tab. 2), but it emphasized the significant presence of small tetras in most cases. The dominant species in the PARNA AM streams were Hemigrammus cf. bellottii and Apistogramma sp.; in the PARNA JX, Moenkhausia aff. melogramma and an undetermined species of Microcharacidium were dominant; in the FLONA ST, the dominant ones were Pyrrhulina aff. brevis and Laimosemiondibaphus; and in REBIO SC, an undetermined species of Knodus and Aspidoras poecilus Nijssen & Isbrücker, 1976 were dominant. ANOSIM identified the structure of fish assemblages in the four areas as different (Global R = 0.715, significance level = 0.01%), as well as the pairwise comparisons (pairwise tests: PARNA AM, PARNA JX, R = 0.417; PARNA AM, FLONA ST, R = 0.825; PARNA AM, REBIO SC, R = 0.6; PARNA JX, FLONA ST: R = 0.938; PARNA JX, REBIO SC, R = 0.547; FLONA ST, REBIO SC: R = 0.871; significance level = 0.01% for all pairwise tests). Indeed, the most dissimilar area corresponds to FLONA ST, which virtually shows no overlap with the other areas (Fig. 3).

The patterns of richness and diversity revealed that, despite sampling a smaller number of streams in REBIO SC — which undoubtedly influenced the species richness in this area — the average richness per stream and indices that describe the diversity were low (Tab. 3). The diversity in REBIO SC streams was impacted by the high dominance of Knodus sp. In contrast, the presence of a large percentage of singletons and unique in this area suggests that there is still a significant portion of local diversity to be documented, as indicated by ACE and ICE (Tab. 3). In contrast, PARNA AM streams had the highest average effective number of species (5.7 for q = 0, Tab. 3) and PARNA JX streams had the highest effective species diversity (4.5 for q = 1, Tab. 3).

The average taxonomic distinctness (Delta+) was notably high in most assemblages, averaging 87.3 and ranging from 0 to 100 (Fig. 4A). This suggests that species are relatively dissimilar from each other in terms of their taxonomic classification, regardless of the protected area. Delta+ values for the studied streams generally fell within the 95% probability funnel, indicating that the fish assemblages exhibited the anticipated level of species diversity for a specific area, with only a few exceptions. Even in assemblages with low richness, Delta+ remained high. Only in the FLONA ST streams, Delta+ was significantly higher than in other areas (Tabs. 3–4). In terms of variation in taxonomic distinctness (Lambda+), the average expected value is 312.5, ranging from 0 to 661.2, and nearly all regions fell within the anticipated limits (Fig. 4B). Interestingly, almost half of the assemblages exhibit a high degree of unevenness concerning their taxonomic structure (Fig. 4B). Among areas, PARNA AM and PARNA JX had the highest values for Lambda+ (Tabs. 3–4).

Overall, 1,333 records (set of species from each stream) were obtained with this inventory, which added 61% to the filtered records from Pará streams (n = 2,187). It is important to mention that these records refer to streams reaches located in watershed up to 4,300 hectares, where 204 genera were registered. The present study added the record of other eight genera, which were not included in the stream records of other collections consulted: Ammoglanis, Apteronotus, Cetopsis, Cetopsorhamdia, Curculionichthys, Hemibrycon, Nannoxyropsis, and Platyurosternarchus. In the collections, the genera Moenkhausia, Hemigrammus, Bryconops, and Hyphessobrycon account for 18% of the records, which is equivalent to that recorded in the present study (20%), reinforcing the relevance of small tetras (Characidae and Iguanodectidae) for Amazonian stream fish diversity (Fig. S5). On the other hand, there are still important spatial gaps in the extreme Northwest and South of the state (Fig. 5), which deserve the direction of new sampling efforts.

The fish fauna recorded in this study predominantly comprises small-sized species, defined as those with a standard length of 15 cm or less at adult size (sensu Castro, Polaz, 2020). This group of fish represents the most threatened category, according to the Brazilian Red Book (Castro, Polaz, 2020; ICMBio, 2022). The sole threatened species in this investigation conforms to the small body-size criterion. The rheophilic species Harttia dissidens has been classified as Vulnerable (VU) under criterion A3c due to potential impacts on its natural habitat from hydroelectric power plants and road expansion, which could lead to a population decline of at least 30% over the next decade (Akama et al., 2018). The threatened species was collected outside the boundaries of PARNA JX, underscoring the necessity for surveys within the park to assess viable populations. Additionally, implementing measures such as the protection of riparian buffers is important for minimizing impacts on streams that extend beyond protected areas (Dala-Corte et al., 2020).

Effective species richness and diversity calculated by Hill numbers were the highest in the two largest protected areas (PARNA AM and PARNA JX). Larger protected areas will be more likely to have native forests throughout the longitudinal and lateral dimensions of the stream network, therefore influencing the richness and diversity of species in stream assemblages (see Montag et al., 2018a). When considering the entire set of streams, 24 species were singletons, with most of them being captured inside and in the surroundings of PARNA AM and PARNA JX (Tab. 2). Among these, some species are associated with woody debris and logs (e.g., Tatia intermedia, Trachelyopterusgaleatus, Cetopsis sandrae, and Batrochoglanis villosus), while others (e.g., an unidentified Characidium, Harttia rondoni, and Rineloricaria lanceolata) inhabit consolidated bottoms of fast-flowing habitats. As previously noted by other authors, rare species hold disproportionate importance in various aspects of assemblages, including functional structure and co-occurrence patterns (Leitão et al., 2016; Lobato et al., 2022). However, this importance may be severely threatened by the numerous activities and threats occurring in Brazilian Amazon protected areas, notably deforestation (Qin et al., 2023).

The loss of forests both along stream banks and in the watershed leads to well-known and devastating consequences for stream fish fauna (Montag et al., 2018b; Budnick et al., 2019; Dala-Corte et al., 2020), particularly for those species with limited mobility, such as those mentioned herein. This is attributed to a reduction in the presence of allochthonous structures like woody debris and logs, accompanied by an increase in siltation in fast-flowing water areas such as riffles. As a result, habitats become less diverse (Budnick et al., 2019; Zeni et al., 2019), providing fewer opportunities for shelter, feeding, and reproduction of species with limited mobility. Consequently, the survival of these species could be greatly compromised in streams affected by deforestation placing them as strong candidates for being in threat categories of red lists.

In terms of species richness, the importance of the Characidae members in the Amazon basin is well-established, as emphasized by several authors (Lowe-McConnell, 1987; Val, Almeida-Val, 1995; Montag et al., 2018b; Dagosta, de Pinna, 2019). Generally, following Characidae, the families Loricariidae and Cichlidae rank second, with their prevalence alternating based on the availability of suitable habitats, levels of physical degradation, water quality, and biogeographic history. For instance, in different reaches (vegetated habitats and beaches) of the same stream, Montaña et al. (2008) observed that species typically associated with vegetated habitats included small invertivorous cichlids, a small invertivorous doradid catfish, and small characids. On beaches, Montaña et al. (2008) observed that the most common species were small-bodied pelagic characids, indicating the role of instream habitat configuration on the assemblages’ structure. There are cases where biogeographic history, combined with habitat configuration, acts as determining factors, as evidenced by the species recorded in FLONA ST, where the abundant cichlids, lebiasinids, and rivulids were associated with marginal ponds (present study and Silva et al., 2022).

Characidae is the taxonomically most diverse characiform family, representing more than half of the diversity of the order, with 1,236 valid species (Toledo-Piza et al., 2024). Another aspect worth mentioning regarding the Characidae is that its taxonomy is highly dynamic, with 200 species described in just the last decade (2015–2024) (Fricke et al., 2024b) and a significant number of cryptic species that, when studied in greater detail, indicate a highly underestimated species richness of the Amazonian ichthyofauna (Escobar-Camacho et al., 2015; Guimarães et al., 2020). The most rich-species genera of Characidae recorded here, i.e., Hemigrammus, Hyphessobrycon, and Moenkhausia, comprise species with alike ecomorphological traits (Melo et al., 2022), in environments with high rate of diversification in the Neotropical Region, as Eastern of Amazon (Cassemiro et al., 2023). Their small size (cf. Cerezer et al., 2023), low vagility, and similar body shape, combined with the abundance of habitats and resources in the areas, partially account for the significant diversity (and abundance) observed in the studied regions.

The greater proximity between PARNA AM and PARNA JX to REBIO SC has biogeographic implications since these areas are included in the Tapajós-Xingu ecoregions (FEOW, 2024). By contrast, the stream fish fauna of the FLONA ST is not only the most distinct but also the most even regarding the taxonomic arrangement. The FLONA ST is in the Amazonas Guiana Shield ecoregion, specifically in the Trombetas River basin (FEOW, 2024), with an evolutionary history distinct from the other areas. Even though the sampling was quite complete, the total species richness and the richness per stream were low in this basin. However, it is important to highlight that there is a set of species from different families with very conspicuous occurrence, generally associated with stream margins, such as Crenuchusspilurus, Copella callolepis, Pyrrhulina aff. brevis, and Laimosemiondibaphus. Individuals of C. spilurus, e.g., spend most of the day sheltered under shaded areas among structures such as dead leaves and branches, roots, and plants, where they feed on particulate organic matters that sink slowly through the water column during daylight (Pires et al., 2016). For P. aff. brevis, in turn, the individuals move between flooded streamside areas and choose suitable pools while still permitting a swift return to the stream channel if conditions in the flooded area become unfavorable (Espírito-Santo et al., 2017). Therefore, not only can a different evolutionary history explain the results of taxonomic distinctness, but the phylogenetic inheritance needs to be combined with biotic interactions, habitat diversity, and disturbance events that lead to unique taxonomic composition and diversity patterns in these areas that are at the same time cradle and museum of diversity (Melo et al., 2022).

For the variation in taxonomic distinctness, which was particularly high in PARNA AM and in PARNA JX, it is necessary to consider that it represents the success of certain families. This is the case of Characidae, with 31 spp. in PARNA AM and 35 spp. in PARNA JX, contrasting to the four species in FLONA ST and 15 in the REBIO SC. In these two last protected areas, species were more evenly distributed among a wider range of higher taxonomic levels. It is important to note that the number of sampled streams in FLONA ST and in REBIO SC was lower than in other areas, which may affect the uniformity of family representativeness. However, stream fish assemblages in FLONA ST were well characterized, since sampling completeness approached 100%. If the low variation in taxonomic distinctness is influenced by environmental attributes, interactions with other species, or biogeographic and evolutionary history, this should be investigated in future studies. On the other hand, by increasing the number of sites in REBIO SC, there may also be an increase in the variation in taxonomic distinctness, as seen in PARNA AM and PARNA JX, both geographically closer to REBIO SC.

Numerous recent efforts have been made to enhance the understanding of stream fishes in Pará State within protected areas and their surroundings. Freitas et al.(2018), e.g., compiled two decades of studies on fishes in the Caxiuanã National Forest. Rosa et al. (2016), Silva et al.(2016, 2022), Soares et al.(2017), and Barros et al. (2019) investigated ecological aspects of the fishes of the Sacará-Taquera National Forest. Dutra et al. (2020) present a list of 286 fish species documented during a rapid survey of the rivers and streams in the Northern Pará drainage system. This survey encompassed five state protected areas in Pará, namely Faro State Forest, Trombetas State Forest, Paru State Forest, Grão-Pará Ecological Station, and Maicuru Biological Reserve. In these various studies, there is always a recurring theme emphasizing the need for more targeted efforts to address thematic gaps concerning Amazonian fishes, as well as the threats directed towards protected areas. Notably, the present study provides a snapshot of the ichthyofauna, as each locality was sampled only once. Recurrent sampling, coupled with the resolution of many uncertainties in species determination, could enhance richness in the study areas, as Freitas et al.(2018) pointed out in FLONA Caxiuanã. This temporal gap — Prestonian shortfall (Hortal et al., 2015) — in sampling has not been, but needs to be addressed, especially because more robust long-term data about species richness and population dynamics are required to understand how Amazonian biodiversity will respond to climate change (see Borba et al., 2021).

Despite conducting a comprehensive sampling effort across 99 streams, significant spatial gaps persist in the extreme Northwest and South of the state of Pará. These regions are characterized by the presence of indigenous lands, such as Nhamundá-Mapoera, Kaxuyana-Tunayana, and Zo’é in the North, as well as Kapôt Nhinore, Menkragnoti, Kayapó, and Panará in the South (ISA, 2024). Nevertheless, access to and research authorizations within these areas are significantly more restricted. It is important to note that indigenous lands have experienced limited research efforts when compared to strictly protected and sustainable use areas (Carvalho et al., 2023). Only through integrating indigenous lands into a more inclusive scientific approach (Carvalho et al., 2023) can we hope to minimize Wallacean shortfalls, as defined by Hortal et al. (2015).

In summary, the present study highlights the importance of small tetras for the fish fauna of Amazonian streams and emphasizes that, despite substantial efforts in documenting the fish fauna, further spatial and temporal sampling is needed, especially in indigenous territories. Although only one threatened species (the Vulnerable (VU) Harttia dissidens) was recorded, there is a significant portion of this fauna that could be affected in the short term, given its rarity in the areas where it occurs.