Spintherobolus Eigenmann, 1911 was proposed over a century ago to include Spintherobolus papilliferus Eigenmann, 1911 from headwaters of the rio Tietê in Paranapiacaba Municipality, São Paulo State, Brazil (Eigenmann, 1911). Fourteen years later, the miniature Spintherobolus broccae Myers, 1925 was described from coastal streams in Rio de Janeiro State and Santos Municipality, São Paulo State (Myers, 1925). Weitzman, Malabarba, (1999) reviewed the genus and described two more small species: Spintherobolus ankoseion Weitzman & Malabarba, 1999 from coastal creeks in Santa Catarina State, and Spintherobolus leptoura Weitzman & Malabarba, 1999 from the lower rio Ribeira de Iguape, the largest coastal river basin in southern São Paulo State (Malabarba, 2003). Weitzman, Malabarba, (1999) and Bührnheim et al., (2008) also proposed the putative basal position of the larger species, S. papilliferus, sister to a clade with the remaining three species. All four species are endemic species of the Brazilian Atlantic Rainforest (Malabarba, 2003), listed as endangered at some level (Brasil, 2022) according to the International Union for Conservation of Nature (IUCN) categories and criteria (IUCN Standards and Petitions Subcommittee, 2019).

Spintherobolus papilliferus (Fig. 1) is currently classified as an Endangered (EN) species (Brasil, 2022) due to the small area of occurrence (approximately 8 km²), decrease in habitat quality related to expansion of the city and pollution, and low number of recorded (n = 4) localities (Castro et al., 2008; Oyakawa et al., 2009; Akama et al., 2018). Throughout most of the 110 years since its description, S. papilliferus was registered occasionally in the southern headwater streams of upper rio Tietê in Serra de Paranapiacaba, its type locality, and riacho Ipiranga (Weitzman, Malabarba, 1999), the former a mountain chain dividing rio Tietê drainage running inland towards the rio Paraná basin and the series of relatively short coastal systems draining into the Atlantic Ocean (e.g., Abell et al., 2008), and the latter almost entirely supressed near São Paulo city centre (e.g., Mattox, 2008). The species has not been found in the type locality since the 1980s (Castro et al., 2008), but has been sampled once in Parque Estadual da Cantareira in the northern headwater streams of rio Tietê and once in Estação Biológica de Boracéia, municipality of Salesópolis near the rio Tietê spring (Castro et al., 2008; Marceniuk et al., 2011). In 2015, S. papilliferus was found for the first time in coastal drainages of São Paulo State, specifically in rio Sertãozinho of the rio Itapanhaú basin (MZUSP 119089). In 2019 it was discovered in rio Juquiá, in the headwaters of rio Ribeira de Iguape basin (Cetra et al., 2020), a record that is explored in more detail herein.

The known distribution of Spintherobolus papilliferus is within the Atlantic Rainforest biome, one of the 33 biodiversity hotspots in the world, thus having priority in conservation efforts (Myers et al., 2000), but the second most threatened forest on Earth (Campanili, Prochnow, 2006). It is estimated that approximately 15% of the original 1.300.000 km² forested area remains today, with over 80% of the fragments with less than 50 ha, only 2–10% of which under protection (Menezes et al., 2007; Ribeiro et al., 2009; Fundação SOS Mata Atlântica, INPE, 2017; Wagner et al., 2020) and a rapid increase in deforestation in recent years (Fundação SOS Mata Atlântica, INPE, 2022). In light of the ongoing rapid loss of the Atlantic Rainforest, especially in Southeastern Brazil, and the fact that most endangered freshwater fishes in South America are small species restricted little streams (Castro, Polaz, 2020), the aim of this study is to provide new information regarding the geographic distribution of Spintherobolus papilliferus, the environmental conditions in which it was found in rio Juquiá, and discuss the expansion in the known distribution of the species for its biogeography and conservation.

FIGURE 1 |

Study area. The rio Ribeira de Iguape basin is located between coordinates 23º45’S 46º45’W and 25º30’S 50º10’W in the states of São Paulo and Paraná. Sampling of the new population of Spintherobolus papilliferus occurred in the municipality of Juquitiba, SP, in the coordinates 23º59’13.20”S 47º00’10.8”W, in a stream with no name draining into the headwaters of rio Juquiá, itself the largest left affluent of rio Ribeira de Iguape (Fig. 2). The stream belongs to a dense ombrophile Atlantic Rainforest with a native plant coverage relatively well preserved. According to the Köppen scale, its climate is classified as Cfb with subtropical humid summers and mild winters (CBH-RB, 2012). The average precipitation is 1400mm/y and reaches 2000 mm/y in the slopes of Serra do Mar (CBH-RB, 2012). The area lies over a pre-Cambrian crystalline bedrock that suffered influence of neotectonisms during the Mesozoic and Cenozoic. As a result, the upper reaches of the basin are characterized by steep terrains along the hillsides with fast flowing rivers that drain into the traphogenic basin of rio Ribeira de Iguape (Ross, 2002; Riccomini et al., 2004). The sampling site in rio Juquiá, municipality of Juquitiba, São Paulo State, does not belong to a protected area.

The characterization of abiotic parameters followed Cruz et al., (2013) and Teshima et al., (2015). The area where the specimens were sampled (Fig. 2) was shallow with large patches of leaflitter and transparent waters. The average depth was 26 ± 12 cm and the average width was 6.80 ± 2.25 m. The average water flow was 0.100 ± 0.036 m/s under the influence of a close waterfall upriver. The substrate was composed of leaflitter (21.53%), wood debris (18.46%), sand (20%), silt (2.3%), pebble (20.77%), cobble (0.8%), boulder (3.84%) and rock outcrops (12.3%). Water pH was 6.19, conductivity 0.037 mS/cm and temperature 17.66 ºC.

FIGURE 2 |

Sampling. Fish was caught in two fieldtrips (September 12 and November 22, 2019) using handnets and an electrofishing gear (LR-24 Electrofisher, Smith Root). Immediately after capture, some specimens were placed in a photograph tank to capture their live coloration with a Nikon D3200 and Mikro lenses of 60 mm. Specimens were then euthanized with eugenol and preserved in 4% formalin (n = 6) or kept in 100% alcohol for DNA analyses (n = 2). Formalin specimens were transferred to 70% alcohol and deposited in the collection of Departamento de Ciências Biológicas, UNESP, câmpus São José do Rio Preto (DZSJRP 23020), Laboratório de Ictiologia de Sorocaba (LISO 426) and Museu de Zoologia da Universidade de São Paulo (MZUSP 127727). Specimens for molecular analyses were deposited in Laboratório de Biologia de Peixes (LBP 31056).

Morphological analyses. Specimens were identified according to Weitzman, Malabarba, (1999). Seventeen measures and six meristic characters were collected from the specimens of rio Ribeira de Iguape following Fink, Weitzman, (1974). Measurements were taken point-to-point with a digital calliper on the left side of the body whenever possible. Morphometric data is presented as percentages of standard length (SL), except for subunits of the head which are given as percentages of head length (HL). Simple statistical descriptors (i.e., sampling number, mean, standard deviation and range) were calculated for each of the measurements. Counts were made under a Zeiss Discovery V20 stereomicroscope and presented as their range associated with the absolute frequency of each count between parentheses. Photography of the preserved specimen was made with the above-mentioned stereomicroscope connected to a digital Zeiss AxioCam.

For comparative purposes, ten specimens from the upper rio Tietê and seven specimens from rio Itapanhaú were also examined, measured and had their meristics counted following the methods described above. Data are presented in a table with each column representing a different drainage (Itapanhaú – IT, Ribeira de Iguape – RI, upper Tietê – UT). We used multivariate analysis of variance (MANOVA) to detect patterns between multiple dependent variables [(body depth (BD), predorsal distance (PD), preanal distance (PA), caudal-peduncle depth (CPD), and upper jaw length (UJL)] and statistically test the differences between three drainage means (RI, IT, UT). The null hypothesis is that the mean drainage vectors are the same for all drainages or don’t differ significantly. We selected the dependent variables that showed a statistically significant correlation with body depth (BD). When dependent variables are correlated, the MANOVA has greater statistical power and limits the joint error rate, and we can assess patterns between multiple dependent variables (Frost, 2019). We used Linear Discriminant Analysis (LDA) as a posthoc test. The LDA finds a linear combination of features that separates the drainages. We used a scatter plot with the two linear discriminants on the X and Y axes to match the independent variable, i.e., the drainage (Radečić, 2022). Percentage data received arcsine square root transformation. All the above analyses were carried in the R environment (R Development Core Team, 2020) and RStudio Team, (2020). We used “manova” and “lda” functions.

The list of comparative material examined follows museum acronym, number of specimens, locality, date of collection, and collectors. The material is deposited in MZUSP, all from São Paulo State, Brazil: Spintherobolus papilliferus: MZUSP 39080, 6, 18.5–61.1 mm SL, ribeirão Curiango, affluent of rio Claro, Estação Biológica de Boracéia, Salesópolis, 11 Sep 1988, Otávio Froehlich; MZUSP 63135, 1, 21.4 mm SL, riacho da Cachoeira do Engordador, Parque Estadual da Cantareira - Núcleo do Engordador, São Paulo, 20 Jun 2020, MZUSP/Uniban Team; MZUSP 72837, 3, 48.7–57.4 mm SL, last stream crossing the road before city of Paranapiacaba, 20 Mar 1980, Ricardo M. C. Castro; MZUSP 119089, 7, 20.9–46.5 mm SL, affluent of rio Sertãozinho, rio Itapanhaú drainage, Bertioga, 29 Aug 2015, LAPEC Team.

Molecular analyses. Fifteen specimens were included in molecular analyses. Fourteen sequences of Spintherobolus were generated herein (Tab. 1) and a single sequence of Triportheus guentheri (Garman, 1890) was retrieved from GenBank (HM404957.1) to root the tree. DNA extraction followed Ivanova et al., (2006) and partial sequences of the cytochrome c oxidase subunit I (COI) gene were amplified by polymerase chain reaction (PCR) with primers FishF1/FishR1 and FishF2/FishR2 (Ward et al., 2005), L6252-Asn/H7271-COXI (Melo et al., 2011) or FishF6 and FishR7 (Jennings et al., 2019). PCR amplifications were performed in a total volume of 12.5 µl, with 1.25 µl of 10X buffer, 0.25 μl of MgCl 2 (50 mM), 0.2 μl dNTPs (2 mM), 0.5 μl of each primer (5 mM), 0.1 μl of PHT Taq DNA polymerase (Phoneutria), 1.0 μl of genomic DNA (200 ng) and 8.7 μl ddH 2 O. PCR conditions consisted of an initial denaturation (5 min at 94 °C), followed by 30 cycles of chain denaturation (50s at 94 °C), primer hybridization (45s at 50–54°C) and nucleotide extension (1 min at 68 °C), and a final extension (10 min at 68 °C). All PCR products were checked on 1% agarose gels and then purified with ExoSap-IT (USB Corporation) following the manufacturer’s instructions. The purified PCR products were subjected to sequencing procedures with the BigDye Terminator v. 3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems) and purified by ethanol precipitation. The PCR products were loaded onto an ABI 3130 DNA Analyzer automatic sequencer (Applied Biosystems).

Raw sequences were assembled to consensus using Geneious v. 7.1.9 (Kearse et al., 2012) and posteriorly aligned with Muscle (Edgar, 2004) under default parameters. Substitution saturation was determined using Xia et al., (2003)’s approach in DAMBE v. 5.3.38 (Xia, 2013). Nucleotide variation, substitution patterns and best-fit model of nucleotide evolution were estimated in MEGA v. 10 (Kumar et al., 2018). The overall mean distance, the intraspecific distances and interspecific distances were estimated using the Kimura-2-parameter model (K2P; Kimura, 1980) and maximum likelihood (ML) analysis was conducted with the Hasegawa-Kishino-Yano (HKY)+G model and 1,000 bootstrap replicates in MEGA v10.

Identifying the new specimens from rio Ribeira de Iguape basin using the most comprehensive key (Weitzman, Malabarba, 1999) leads to Spintherobolus papilliferus (Fig. 1). Our samples have four of the five autapomorphies for this species: (i) number of branched anal-fin rays (9 or 10); (ii) eye-diameter smaller than in the remaining Spintherobolus species (23.2–32.7% of HL in our samples, 18.4–26.0% of HL in Weitzman, Malabarba, 1999); (iii) long caudal peduncle (26.1–30.5% of HL in our samples, 21.3–27.0% of SL in Weitzman, Malabarba, 1999); and (iv) adult body size larger than remaining Spintherobolus species (39.6 mm SL in our samples, up to 60.8 mm SL in Weitzman, Malabarba, 1999). The fifth autapomorphy, number of vertebrae, was not accessed in the present study.

Morphometric and meristic data are presented in Tabs. 2 and 3, respectively. While almost all data greatly overlap between drainages, five measures were different between Spintherobolus papilliferus from the rio Ribeira de Iguape and the other two populations. Values of ranges and means for these two characters are lower in the population from rio Ribeira de Iguape than those of the remaining drainages. The MANOVA detected a pattern between these five dependent variables (body depth, pre-dorsal distance, pre-anal distance, caudal-peduncle depth, and upper jaw length) which were significantly different between the three drainage means (rio Ribeira de Iguape, rio Itapanhaú, and upper rio Tietê basins) (Fig. 3A). As body depth increases, all other dependent variables tend to increase. Moreover, for any given body depth, specimens of S. papilliferus from the rio Ribeira de Iguape drainage tend to have lower values of these measures than specimens from Itapanhaú and upper Tietê basins, i.e., specimens with equal body depth tend to have smaller measures in rio Ribeira de Iguape. We can conclude that there is an association between drainage and the relationship among the morphometric variables. The group mean vector of rio Ribeira de Iguape is significantly different from the other drainage means (Fig. 3B). We assumed it was a crucial factor for rejecting the null hypothesis.

FIGURE 3 |

The molecular dataset resulted in a matrix of 531 bp, with 146 variables sites. The nucleotide frequencies were 25.4% adenine, 15.5% guanine, 30.8% thymine, and 28.3% cytosine. DAMBE indicated no saturation for either transitions or transversions in both asymmetrical (Iss.cAsym) and symmetrical (Iss.cSym) topologies. The best nucleotide model selected for the matrix was HKY+G. The maximum likelihood (ML) tree showed high bootstrap values supporting each of the analyzed species (Fig. 4). The overall mean of genetic distances among Spintherobolus species was 6.9%. The values of interspecific distances ranged from 1.5% between S. leptoura and S. ankoseion to 16.2% between S. papilliferus and S. ankoseion (Tab. 4). Interestingly, the genetic distances between the three smaller species (S. ankoseion, S. broccae, and S. leptura) range from 1.5–3.5%, while these distances are much higher (14.8–16.2%) when any of these three species are compared to S. papilliferus.

FIGURE 4 |

The populations of Spintherobolus papilliferus from the rio Ribeira de Iguape, rio Itapanhaú, and upper rio Tietê present all meristic data and almost all morphometric data greatly overlapped. The exceptions are body and caudal-peduncle depths, predorsal and preanal distances, and upper jaw length, which are relatively smaller in samples from rio Ribeira de Iguape basin (Tab. 2). The MANOVA and LDA confirmed these differences as statistically significant (Fig. 3). Despite these five morphometric differences, DNA barcoding corroborates samples from both coastal drainages (i.e., Ribeira de Iguape and Itapanhaú) as belonging to the same species, with the other three species apart in their own group (Fig. 4). It is unfortunate that we did not have access to molecular samples from the upper rio Tietê to verify if the coastal samples would indeed group with the crystalline shield, but our meristic data and the majority of our morphometric data confirm that our three samples belong to the same species. Moreover, several other fish species have a similar distribution pattern in both upper rio Tietê and coastal basins (e.g., Ribeiro, 2006; Oyakawa, Menezes, 2011), some of which will be further discussed below.

It is interesting that specimens of S. papilliferus from rio Ribeira de Iguape have a more streamlined body reflected by their shallower body and caudal-peduncle, and this could be interpreted as a populational variation from the river as our sample was caught in a rapids pool, therefore with high water flow, a typical environment of this region (Ross, 2002). Streamlined body in fishes is usually associated with faster and more agile forms (e.g., Webb, 1984; Astudillo-Clavijo et al., 2015), and a similar pattern of shallower body and caudal-peduncle associated with lotic environments was demonstrated by Dunn et al., (2020) with species of Galaxias Cuvier, 1816. Indeed, McLaughlin, Grant, (1994) demonstrated that juveniles of Salvelinus fontinalis (Mitchill, 1814) raised in faster flowing waters tend to develop shallower bodies and caudal peduncles, illustrating a clear case of phenotypic plasticity. A similar experiment with S. fontinalis in the lab also demonstrated phenotypic plasticity in juveniles raised in fast-flowing and slow-flowing conditions, however found out that fish raised in faster currents tend to have a deeper caudal peduncle probably due to swimming muscle increase in the tail (Imre et al., 2002). Our hypothesis is that S. papilliferus from the fast-flowing waters of rio Ribeira de Iguape basin are under a selective pressure which makes their body more streamlined. Their body is also shorter with lower values of preanal and predorsal distances, in addition to a shorter upper jaw (Fig. 3A).

Until recently, Spintherobolus papilliferus was restricted to the highlands of the upper rio Paraná basin, where rio Tietê empties, while all the other species of the genus are lowland coastal river species: S. broccae from coastal rivers in Rio de Janeiro and São Paulo states, S. ankoseion from coastal rivers of Santa Catarina and Paraná states, and S. leptoura from lower rio Ribeira de Iguape basin. This led Ribeiro, (2006) to classify Spintherobolus in his pattern B of distribution of species defined as “sister-group relationships between endemic ichthyofauna of the Brazilian coastal drainages and adjacent shield that includes intermediate degree of inclusiveness, in which, in some cases, both sister-clades underwent some subsequent radiation”. Also, according to this comprehensive work of Ribeiro, (2006), this cladogenetic events would have happened in the Tertiary. The presence of S. papilliferus in coastal drainages (Fig. 5) indicates that this interpretation needs to be updated. We propose that diversification of Spintherobolus species fits best Ribeiro’s (2006) pattern C, that “reflects the most recent vicariant events between upland crystalline shield rivers and the adjacent coastal drainages that lead to the recognition of shared species between them”.

Ribeiro (2006) provided examples of species that fit his pattern C, including Characiformes, Siluriformes, Gymnotiformes, Cichlidae and Poeciliidae fishes along the entire Brazilian coast. Specifically in Southeastern Brazil, Ribeiro, (2006) described the pattern in which the upper rio Tietê shares species with the headwaters of rio Ribeira de Iguape and other adjacent coastal rivers such as Hollandichthys multifasciatus (Eigenmann & Morris, 1900), Pseudocorynopoma heterandria Eigenmann, 1904, Hyphessobrycon bifasciatus Ellis, 1911, Hyphessobrycon reticulatus Ellis, 1911, and Gymnotus pantherinus (Steindachner, 1908). However, these are examples of species widespread in the coastal drainages that were also found in the upper rio Tietê system, suggesting a recent colonization of the latter basin by species originally from the coastal drainages (Ribeiro, 2006), which is not the case of Spintherobolus papilliferus with a restricted area in coastal drainages (i.e., so far, only found in two localities).

Alternatively, the distribution of Spintherobolus papilliferus in both upper rio Tietê and costal drainages such as rios Ribeira de Iguape and Itapanhaú could be explained by headwater capture from the crystalline shield towards the ocean as the Serra do Mar mountain chain that divides both systems suffered erosional retreat since its initial formation in the Paleogene (Almeida, Carneiro, 1998; Ribeiro, 2006). A similar model was proposed for the headwaters of rio Guaratuba, a coastal river whose upper reaches are close to rio Claro, a tributary of the upper rio Tietê (Ribeiro et al., 2006). As a matter of fact, †Megacheirodon unicus (Travassos & Santos, 1955) is a fossil in the clade including Spintherobolus from the Tremembé Formation and dated as Oligocene (35 to 23 million years ago), therefore within the boundaries of the Paleogene (Malabarba, 1998). †Megacheirodon unicus was considered as sister to Spintherobolus until the description of Amazonspinther dalmata Bührnheim, Carvalho, Malabarba & Weitzman, 2008 a miniature fish from the rios Madeira and Purus in the Amazon basin (Bührnheim et al., 2008). Therefore, the most recent relatives of Spintherobolus are from the Amazon and crystalline shield and the distribution of Spintherobolus papilliferus in coastal drainages could be explained by headwater captures due to retreat of the Serra do Mar (e.g., Camelier et al., (2018)), a hypothesis that must be tested with a phylogeographic approach when pertinent material is available.

Serra et al., (2007) studied the ichthyofauna of rio Itatinga, an affluent of rio Itapanhaú, and found four species that were shared between this system and upper rio Tietê with relatively restricted geographic distributions: Coptobrycon bilineatus (Ellis, 1911), Glandulocauda melanogenys Eigenmann, 1911 [= Glandulocauda melanopleura (Ellis, 1911)], Pseudotocinclus tietensis (Ihering, 1907), and Taunayia bifasciata (Eigenmann & Norris, 1900), three of which endangered species except for Glandulocauda melanopleura (Akama et al., 2018). Interestingly, Spintherobolus papilliferus, another endangered species, is added to this list as it also occurs in a headwater affluent of rio Itapanhaú, in addition to rio Ribeira de Iguape basin.

FIGURE 5 |

Recently, Camelier et al., (2018) presented an integrative approach to Glandulocauda melanopleura in which disjunct populations of the upper rio Tietê, and headwaters of coastal drainages such as Guaratuba, Itatinga, Itanháem, and Ribeira de Iguape have some morphological variation albeit monophyletic according to molecular data. In addition to confirming the monophyly of G. melanopleura, the genetic data showed two interesting results: (i) there are two distinct and well corroborated lineages, one from upper rio Tietê and rio Itanháem, and another exclusive of rio Guaratuba; (ii) there is a single haplotype endemic of rio Guaratuba, and two other haplotypes shared by upper rio Tietê and rio Itanháem indicating a recent event of river capture between these two latter drainages. Spintherobolus papilliferus seems to have a similar diversification history to that of G. melanopleura, with disjunct populations resulting from river capture from the uplands towards the coastal drainages that yields to intraspecific morphological variation.

Although the updates in the geographic range of the species, Spintherobolus papilliferus might not change to a less strict conservation status according to the criteria and standards of the IUCN (IUCN Standards and Petitions Subcommittee, 2019). Spintherobolus papilliferus was found in only four locations, one of them without records after the 1980s (Castro et al., 2008) and another almost entirely suppressed with the growth of the city of São Paulo (e.g., Mattox, 2008). The other two locations are protected by Conservation Units (Parque Estadual da Cantareira and Parque Estadual da Serra do Mar). The estimated area of occupancy was 8 km². We report herein two additional locations where the species can be found, one in the rio Sertãozinho, an affluent of rio Itapanhaú system whose preservation status is not known to us, and another in an affluent of rio Juquiá, rio Ribeira de Iguape basin, which is relatively well preserved although it does not belong to conservation units. This raises the number of locations and will likely increase the estimated area of occupancy of S. papilliferus, but will probably be not enough to change the classification of the species as Endangered.

Sampling efforts should be done in places where Spintherobolus papilliferus might still survive (e.g., Parque Estadual da Cantareira, Estação Biológica da Boracéia) to investigate if it is still present in these locations. It is of uttermost importance that the few sites where S. papilliferus occurs are kept under protection. This is especially relevant due to the apparent co-occurrence of S. papilliferus with other endangered species such as Coptobrycon bilineatus, Pseudotocinclus tietensis, Taunayia bifasciata, and Cambeva paolence (Eigenmann, 1917), all of which have restricted distributions in the upper rio Tietê and coastal rivers such as the headwaters of rio Ribeira de Iguape, rio Itapanhaú and rio Guaratuba (e.g., Ribeiro et al., 2006; Serra et al., 2007). Creation and monitoring of conservation units are important measures to mitigate our impact on freshwater ecosystems such as the Atlantic Rainforest streams, but these need to take into consideration the connectivity of aquatic habitats to fulfil their aims (Albert et al., 2020).