A new species of Amazonian snouted treefrog (Hylidae: Scinax) with description of a novel species-habitat association for an aquatic breeding frog

The genus Scinax is one of the most specious genera of treefrogs of the family Hylidae. Despite the high number of potential new species of Scinax revealed in recent studies, the rate of species descriptions for Amazonia has been low in the last decade. A potential cause of this low rate may be the existence of morphologically cryptic species. Describing new species may not only impact the taxonomy and systematics of a group of organisms but also benefit other fields of biology. Ecological studies conducted in megadiverse regions, such as Amazonia, often meet challenging questions concerning insufficient knowledge of organismal alpha taxonomy. Due to that, detecting species-habitat associations is dependent on our ability to properly identify species. In this study, we first provide a description of a new species (including its tadpoles) of the genus Scinax distributed along heterogeneous landscapes in southern Amazonia; and secondly assess the influence of environmental heterogeneity on the new species’ abundance and distribution. Scinax ruberoculatus sp. nov. differs from all nominal congeners by its small size (SVL 22.6–25.9 mm in males and 25.4–27.5 mm in females), by having a dark brown spot on the head and scapular region shaped mainly like the moth Copiopteryx semiramis (or a human molar in lateral view, or a triangle), bicolored reddish and grey iris, snout truncate in dorsal view, bilobate vocal sac in males, by its advertisement call consisting of a single pulsed note with duration of 0.134–0.331 s, 10–23 pulses per note, and dominant frequency 1,809–1,895 Hz. Both occurrence and abundance of the new species are significantly influenced by silt content in the soil. This finding brings the first evidence that edaphic factors influence species-habitat association in Amazonian aquatic breeding frogs.

Presently, twenty-nine valid species of Scinax occur in Amazonia sensu lato (Eva & Huber, 2005).However, the number of species occurring in this region (see Table S1) is currently underestimated (Fouquet et al., 2007;Ferrão et al., 2016).An integrative approach combining morphological, bioacoustics and DNA barcoding data revealed existence of seven confirmed candidate Scinax species distributed in different parts of the Purus-Madeira Rivers Interfluve (hereafter PMRI) (see Ferrão et al., 2016).Here we describe the confirmed candidate species Scinax sp.7 from Ferrão et al. (2016).
Species are widely recognized as fundamental units in ecology (Gotelli, 2004).Therefore, quantifying species-habitat associations, both at population and assemblage level, is dependent on accurate taxonomic classifications (Isaac, Mallet & Mace, 2004;Bortolus, 2008).Taxonomic inaccuracy may not affect the conclusions of ecological communities studies when a single target taxon is misidentified (e.g., Bortolus, Schwindt & Iribarne, 2002;Bortolus, 2006).However, misidentifying two or more morphologically similar species as a single species may considerably affect the degree of reliance indicated by patterns of species-habitat association (Dexter, Pennington & Cunningham, 2010) because frogs are globally threatened by habitat loss, habitat disconnection, disease and alien species (Kats & Ferrer, 2003;Stuart et al., 2004;Lips et al., 2006;Becker et al., 2007), and some attention have been focused on generating scientific knowledge to support its conservation and management.Species misidentifications should be more common in ecological studies from tropical regions because of the high levels of morphologically cryptic species that co-occur (Kreft & Jetz, 2007;Jenkins, Pimm & Joppa, 2013;Pimm et al., 2014).Therefore, only limited taxonomic data are available for many groups of organisms from tropical rainforests (Giam et al., 2012).
The Amazonia is the largest and most species-diverse tropical rainforest worldwide (Corlett & Primack, 2011).Nonetheless, molecular approaches applied in systematics have revealed that biodiversity in the Amazonia is greatly underestimated by cryptic biodiversity in many groups of organisms, such as frogs (Ron et al., 2012;Jungfer et al., 2013;Caminer et al., 2017).Different species of frogs react differently to the same anthropogenic effect, therefore the taxonomic knowledge is an important key to the conservation of frogs, especially in the Neotropics where the biodiversity is megadiverse.However, the taxonomy and systematics of many frog groups from the Amazonia and the role of environmental variables in shaping the distribution of species remain poorly understood.Furthermore, most taxonomic and ecological studies have been conducted on a local scale; thus, regional approaches are lacking (Allmon, 1991;Tsuji-Nishikido & Menin, 2011;Rojas-Ahumada, Landeiro & Menin, 2012;Dias-Terceiro et al., 2015;Jorge et al., 2016).
Studies that have used frogs to demonstrate patterns of species-habitat association in Amazonia were conducted mostly in the Manaus region (Brazil), on the north bank of the Amazon River (Allmon, 1991;Menin et al., 2007;Menin, Waldez & Lima, 2011;Tsuji-Nishikido & Menin, 2011;Rojas-Ahumada, Landeiro & Menin, 2012;Jorge et al., 2016).In this region, many terrestrial-breeding anurans are influenced by edaphic variables like slope, clay content and pH, while aquatic-breeding anurans are influenced by distance of streams and number of trees (Menin et al., 2007;Menin, Waldez & Lima, 2011;Landeiro, Waldez & Menin, 2014).However, anurans from other parts of Amazonia may present distinct patterns of species-habitat association, once the environmental factors that filter species occurrence and abundance may be expected as a response to regional sets of environmental conditions.As example, the Manaus region exhibits environmental elements that are not observed in the central and northern PMRI, which is the area sampled for this study.This portion of PMRI exhibits forests 17-27 m high, topography relatively flat with local variation from 1 to 3 m, silty soil and shallow water-table (Cintra et al., 2013;Martins et al., 2014;Schietti et al., 2016), whereas the Manaus region is characterized by forests with 30-37 m high closed canopy, undulating topography formed by valleys and plateaux ranging from 40 to 100 m a.s.l., low proportion of silt in the soil, and deep water-table (Ribeiro et al., 1999;Castilho et al., 2006;Schietti et al., 2014).
Reports have indicated that the treefrog genus Scinax is adequate for quantifying species-habitat associations because it is an extraordinarily species-rich genus that is widely distributed across different habitats in the Amazonia (Duellman & Wiens, 1993;Fouquet et al., 2007;Ferrão et al., 2016).However, the influence of environmental factors on the species distribution and abundance of Scinax is poorly understood, which is likely because of the lack of standardized sampling and the difficulty in identifying the species.Here, we also investigate the influence of environmental variables on the new species' spatial distribution based on sampling from standardized plots distributed along an approximately 600 km long transect in the PMRI.

Study area
The interfluve between the Purus and Madeira rivers covers approximately 15.4 million hectares of the southern Brazilian Amazonia in an area that is drained by a large and complex stream network (Maldonado et al., 2012).At a broad scale, the northern portion of the PMRI is covered by a tropical lowland rainforest with emergent canopy and the south is covered by open rainforest lowlands with palm trees (IBGE, 1997) (Fig. 1A).At a finer scale, the number of trees and palms in the plots range from approximately 2,000 to 11,500 individuals per hectare (diameter at breast height > 1 cm) and the biomass of plants is lower in plots located in the northern and southern extremes (Schietti et al., 2016).
The soil is generally shallow (Martins et al., 2014), with a predominance of silt, followed by sand and clay (Cintra et al., 2013).At a regional scale, the topography of PMRI is relatively flat and the altitude is between 27 and 80 m.Temporary ponds occur in lower areas during the rainy season (Rossetti, Toledo & Góes, 2005) and are formed by local ranges in elevation from 1-3 m.

Sampling design and collection effort
We collected data from 110 sampling plots (size 250 × 10 m) distributed along a 600 km transect in the PMRI, between the municipalities of Careiro da Várzea (03 • 11 32 S, 59 • 52 09 W) and Humaitá (07 • 13 06 S, 63 • 05 31 W) (Fig. 1B).The plots were distributed in 11 long-term ecological research sites (hereafter RAPELD) (Magnusson et al., 2013), with approximately 50 km of space between neighbouring modules.Each RAPELD module consists of two parallel 5 km trails and 10 sampling plots, with five plots per trail.The plots were spaced 1 km apart (Fig. 1).The sampling plots followed altitudinal contours to reduce the environmental heterogeneity within each sampling unit (Magnusson et al., 2013).
We sampled Scinax specimens using time-and space-constrained visual searches (modified from Campbell & Christman, 1982) and acoustic searches for males in breeding activity.All sampling plots were surveyed by two observers over 90 min.The plots were surveyed three times during the rainy season: in January/February 2013, November 2013, and January/February 2014 (990 sampling hours in total).Although the breeding seasons of most species of Scinax are seasonal, and some of them are explosive breeders, the new species was observed breeding along the entire rainy season.Due to that, we do not expect changes in abundance along RAPELD sampling modules caused by our sampling protocol or by sampling sessions.Adults were collected via occasional encounters on access trails and in areas surrounding sampling sites.Tadpoles were encountered in ponds unconnected to streams in the RAPELD module 9, where active breeding adults were encountered.
Adult specimens were killed with a 2% benzocaine solution, fixed in 10% formalin and conserved with 70% ethanol.Tadpoles were killed with a 5% lidocaine solution and conserved in 5% formalin.We used 5% formalin in tadpoles because the labial papillae and jaws are malleable after fixation at that concentration and it facilitates posterior description.Muscular tissue samples of all adults and one tadpole were collected before fixation in formalin, and preserved in absolute ethanol.Adults and tadpoles were deposited in the herpetological section of the Zoological Collections of the Instituto Nacional de Pesquisas da Amazônia (INPA-H), Manaus, Brazil.Tissue samples were deposited in the Albertina Pimentel Lima's Laboratory at Instituto Nacional de Pesquisas da Amazônia (INPA).See Appendix S1 for the specimens used in the comparisons.
Specimens were collected from RAPELD sampling modules under permit of Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) and Centro Nacional de Pesquisa e Conservação de Répteis e Anfíbios (RAN) Number 13777.ICMBio and RAN are institutes of Ministry of Environment, Government of Brazil.These permits were subject to approval of all procedures for collecting and euthanizing frogs.

Morphometric and acoustic data
Morphometric data were collected for adults using a digital calliper, and measurements were performed to the nearest 0.1 mm.Small measurements were taken with a digital calliper under a stereomicroscope.Specimens were sexed through the presence or absence of vocal sac, nuptial pads, vocal slits, and/or eggs.Males were defined as sexually active adults by the observation of expanded vocal sacs.Females were classified as adults when eggs were observed inside the body cavity, or when they were collected in amplexus with or near active males in temporary ponds.All males and females analysed in this study were classified as adults.Sixteen taxonomically important traits for identifying Scinax were measured as explained bellow (see Appendix S3 for measurement data).Ten morphometric characteristics were measured according to Duellman (1970): snout-vent length (SVL), head length (HL), head width (HW), horizontal eye diameter (ED), upper eyelid width (UEW), internarial distance (IND), interorbital distance (IOD), horizontal tympanum diameter (TD), tibia length (TL) and foot length (FL).Three characteristics were measured according to Napoli (2005): eye-nostril distance (END), fourth finger disk diameter (4FD), and fourth toe disk diameter (4TD).The length of the tarsus (TAL), hand (HAL) and thigh (THL) followed Heyer et al. (1990).The webbing formulae follow Savage & Heyer (1967) as modified by Myers & Duellman (1982).The enumeration of fingers followed Fabrezi & Alberch (1996) that demonstrated the loss of finger I in the anuran evolution.Description of the colour in preservative is based on all paratypes and given in percentage (between parentheses).Colour in life is described based on photographs of five specimens (except the colour of iris, which was denoted in all paratypes).
We used a Principal Components Analysis (PCA) to detect sexual dimorphism in the morphometric data by visually checking for overlapping individuals in the morphometric multivariate space.The PCA was performed using the SVL and 15 morphometric ratios (measurement/SVL) from 28 males and 6 females.Because the SVL and ratios have different scales, we used the command line ''scale = TRUE'' in the prcomp function of the R platform (R Core Team, 2016).We also used a Multivariate Analysis of Variance (MANOVA) to test for significant differences between the PCA scores by sex.The first two Principal Components (PCs) of the PCA were used as dependent variables, and sex was used as a factor in the MANOVA model implemented in the R platform (R Core Team, 2016).
Because of the wide distribution of the new species along the PMRI, we verified whether morphological variation occurred in males and females across the RAPELD modules.Given that no significant morphological dimorphism was found between males and females (see results), we performed a MANOVA where the first two components of the PCA quoted above were used as dependent variables and RAPELD sampling modules were used as a factor.
The advertisement call of one male was recorded using a PMD 660 digital recorder (Marantz, Kanagawa, Japan) and a ME 66 directional microphone (Sennheiser, Wedemark, Germany).The calls were analysed using oscillogram and spectrogram (Blackman window, 80 Hz of frequency resolution and 1,024-point Fast Fourier Transform) generated using Raven 1.5 software (Bioacoustics Research Program, 2014).Terminology of acoustic parameters following Köhler et al. (2017).As the advertisement call of the new species consists of a single note, we considered this unit a call (see Köhler et al., 2017).The following spectral and temporal parameters were obtained from 20 calls of the recorded male: call duration (s), number of pulses per call, pulse duration, inter-pulse interval, pulse repetition rate (pulse/s), and minimum, maximum and dominant frequency (Hz) of the call.

Species-habitat association
Given that the term ''habitat'' has different concepts and confusion over its use can be result of ambiguity, we follow the definition by Block & Brennan (1993): ''the subset of physical environmental factors that a species requires for its survival and reproduction''.
We selected three predictor variables to investigate the influence of the environment on the distribution and abundance of the new Scinax species from the PMRI: (1) Forest structure, which was represented by the number of trees (diameter at breast height > 1 cm).We selected tree density to represent forest structure because this variable has been identified as an important factor affecting the distribution and abundance of frogs with aquatic reproduction in Amazonia (Menin, Waldez & Lima, 2011;Landeiro, Waldez & Menin, 2014).(2) Soil structure, which was represented by the percentage of silt because the soil texture in the study area consists primarily of silt (Cintra et al., 2013).Additionally, silty soils are structurally unstable, which increases water retention (Juo & Franzluebbers, 2003).(3) Depth of underground water, which was selected because shallow underground water can overflow lower areas of the study transect.Both soil structure and depth of underground water are variables directly related to the water availability in the pond, and indirectly related to the breeding preferences of the new species, which deposits its eggs only in ponds not connected to streams.Although the distance of stream is a variable commonly used to explain the distribution of aquatic breeding frogs in Amazonia, we do not evaluate it because the reproduction of the new species occurs only in ponds not connected to lotic water bodies.See Appendix S2 for sampling methods related to predictor variables.
Our standardized sampling procedures registered a higher number of Scinax individuals compared with previous studies conducted in the Brazilian Amazonia (e.g., Ribeiro, Lima & Magnusson, 2012).However, the number of recorded specimens was low.Hence, we used the sum of recorded individuals (instead of the mean) for the three sampled periods to represent the abundance of Scinax in each plot (Bueno et al., 2012).Occasional encounters were not included in the ecological analyses described below.
The normality of the data was assessed using the Shapiro-Wilk test (P < 0.05 in all cases).The soil and forest structure data were normalized using log(x), and the underground water and frog abundance data were normalized with log(x +1).We tested for correlations among the environmental variables using Spearman's coefficient.The environmental variables were not correlated to each other.We investigated the effects of the predictor variables on the distribution (presence/absence) of the new species of Scinax using the inflated zero regression model (Zeileis, Kleiber & Jackman, 2008), which was run using the R-package pscl (Jackman, 2015).The effect of each predictor variable on the abundance of the new species was tested using simple linear regression.The significance level in regression models and zero-inflated model was α = 0.05

Taxonomic statement
The electronic version of this article in Portable Document Format (PDF) will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone.This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN.The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/.The LSID for this publication is: urn:lsid:zoobank.org:pub:D50A0044-0EA6-4619-B0FF-42A7443C4EAB.The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS.34604,34614,34615,34622,34598,34624,34627,34629) 34609,34611,34612,34617,34618,34621,34625,34626,34628) collected by Miquéias Ferrão and Rafael de Fraga in  Etymology.The specific epithet ruberoculatus is composed of two words in Latin, ''ruber'' (red) and ''oculatus'' (having eyes).The name is an adjective in concordance with the masculine gender of the genus Scinax and refers to the reddish colour of the upper part of the iris.Suggested English common name: 'Red-eyed snouted treefrog'.

Taxonomic account
Generic placement.We assign the new species to Scinax (sensu Duellman, Marion & Hedges, 2016) based on general morphological similarity to other members of the genus, cloacal tube of tadpoles positioned above the margin of the lower fin (a putative synapomorphy of the former S. ruber Clade according Faivovich et al., 2005), and our previous molecular data (Ferrão et al., 2016).
Diagnosis.A small species of the genus Scinax characterized by the following combination of characteristics: SVL 22.6-25.9mm in males and 25.4-27.5 mm in females; snout truncate in dorsal view and rounded in lateral view; tarsal tubercles indistinct; tubercles on the lower jaw, knee, and heel absent; diameter of disc on fourth finger represents 60% of tympanum diameter; skin on dorsum smooth; dentigerous processes of vomers triangular; bilobate vocal sac and nuptial pads in males; Finger III<V; in life, ground colour of dorsum light grey or light brown; a large brown or grey spot on the head and scapular region shaped like the moth of the species Copiopteryx semiramis (Cramer, 1775), or a human molar in lateral view, or a triangle; dorsal or dorsolateral stripes absent; whitish cream stripe in the lower portion of the flanks; anterior and posterior surfaces of thighs brown; webbing between toes light to dark grey; belly white to greyish-white with light brown to brown blotches laterally; males with vocal sac semi-translucent white; iris bicolored, upper half reddish, lower half grey; advertisement call consisting of a single pulsed note, with note duration of 0.134-0.331s, 10-23 pulses/note, dominant frequency 1809-1895 Hz; tadpoles with labial teeth formula 2 (2)/3, absence of labial arm, and presence of dark brown blotch on the distal part of the tail.
Comparison.Currently, the genus Scinax sensu Duellman, Marion & Hedges (2016) includes 71 species (Frost, 2017).Morphologically, Scinax ruberoculatus sp.nov.can be distinguished from all these species and from six confirmed candidate species identified by Ferrão et al. (2016) by following combination of characters (characters of other species in parentheses or brackets unless otherwise stated).
Holotype measurements (mm).SVL 25.9; HL 9.6; HW 8.8; ED 3.2; TD 1.5; UEW 2.4; IND 2.0; TAL 7.2; FL 9.7; HAL 7.6; 3FD 0.9; 4TD 0.9; END 2.9; TL 13.0; THL 12.1; NSD 0.8.Holotype colouration.In preservative, dorsum light grey dorsum (Fig. 2A); anterior portion of the head amber-yellow; a dark brown blotch similar in shape to the moth Copiopteryx semiramis extending from the interocular region to middle flanks; infraocular and infratympanic regions white; grey chevron-shaped blotch and dark grey dots on the sacral region; background of hand beige-coloured , arm and forearm with diffuse amber-yellow pigmentation; a narrow horizontal amber-yellow stripe between the hand and forearm and on the anterior portion of the arm; anterior and posterior area of thigh light brown, dorsal surface of thigh beige; beige blotch on the knee; tibia light grey with three diffuse brown stripes; beige-coloured blotch on the tibia-tarsus articulation; tarsus, feet and toes light cream; throat cream with inconspicuous light grey blotches (Fig. 2B); vocal sac and chest cream; belly yellowish cream; white band between the flanks and belly with few light grey blotches; perianal region white.Except by the colour of the bicolored iris (reddish upper and grey lower), the colour in life was not recorded.
Variation.The adult paratopotypes and paratypes are similar to the holotype.Variation occurs in the presence and number of small dorsal tubercles (completely absent in some individuals).Variation in foot webbing is as follows: Measurements are provided in Appendix S3 and summarized in Table 1.
The first axis (PC) of the PCA summarized 23% of the variation in the morphometric data, whereas PC2 summarized 16.3%.See Appendix S4 for values of other axes.Sexual dimorphism was not observed in the body shape (morphometric ratios) of S. ruberoculatus sp.nov.(Pillai trace = 0.100, df = 31, P = 0.19), and the body shape of both sexes overlapped in the morphometric multivariate space (Fig. 4A).Males and females of S. ruberoculatus sp.nov.did not show distinct morphological variations along the RAPELD sampling modules (Pillai trace = 0.38, df = 54, P = 0.41).The body shape of specimens from all modules overlapped in the morphometric multivariate space (Fig. 4B).
In preservative (Figs.5A-5I), the colour pattern of the paratypes shows the following variations (frequency in % of paratype specimens): Dorsal ground colour varies from light grey (Fig. 5B) to brown (Fig. 5E).A grey to dark brown blotche on the head and scapular region is shaped like the Neotropical moth Copiopteryx semiramis (60%), a human molar in lateral view (26%), or a triangle (14%).A grey to dark brown chevron in the sacral region is present in 42% of paratypes (Figs.5C-5H).Some individuals have tiny black spots (Figs.5A-5C) on the dorsal surface of the body and limbs (42%).Light brown or dark grey spots on the upper lip are more concentrated and conspicuous in some individuals (60%).Grey to brown stripe between the eye and the tympanum is present in some individuals (31%); a grey or dark brown supratympanic stripe (95%); a cream to yellowish cream stripe in the lower portion of the flanks bordered by brown spots dorsally (100%).One to three light
In life (Figs.6A-6F), dorsal ground colour varies from light grey (Fig. 6D) to brown (Fig. 6E).Blotches, chevrons, or stripes on the dorsal surfaces are more conspicuous than in preservative.Iris bicolored, upper half reddish, lower half grey; both parts separated by a narrow central red streak (Fig. 6).Dorsal surface of the arm cream (Fig. 6A) to yellowish Full-size DOI: 10.7717/peerj.4321/fig-6 cream (Fig. 6E).Anterior and posterior surfaces of thighs uniformly brown, dorsal surface of thigh yellowish cream to light brown.A whitish cream stripe in the lower portion of the flanks.Throat, vocal sac, and chest grey to semi-translucent white; belly white to greyish-white with light brown to brown blotches laterally (Fig. 6F).Ventral surface of the hand light grey, ventral surface of foot light grey to brown.Webbing light to dark grey.

Notes.
n, number of measured tadpoles.
spiracle opening; maximum height of the tail higher than body height (MTH/BH = 1.2).The intestinal mass is visible and positioned subparallel to the longitudinal body axis.Oral disc is in anteroventral position (Fig. 8).Marginal lip papillae bordering the entire posterior region of the lip up to second third of the anterior lip; one row of papillae on the anterior lip and three rows of papillae on the posterior lip.Labial teeth formula 2 (2)/3; A-2 hiatus 0.1 mm; teeth more developed in rows A-1, A-2 and P-1 than in P-2 and P-3.
Posterior border of the upper jaw bow with serrated cutting edge; and V-shaped lower jaw with serrated cutting edge.
Colouration of tadpoles (stages 34-39, n = 13).In live and preserved specimens, bronze body and tail muscle; tiny dark brown spots on body and tail that are denser posteriorly and may resemble small bars in the lower portion of the tail; dark brown stripe from snout to eye; dark brown postocular stripe; translucent tail; dark brown blotches conglomerated in the last third of the tail; and dark brown tip of tail (Fig. 8).
Colouration of newly metamorphosed specimens (INPA-H 35412, n = 1).Living specimens, greyish-brown dorsally with small dark grey spots in the sacral region; loreal region brown; iris red with black border; postocular region and anterior portion of flanks brown; inguinal region light grey; dorsal surfaces of arm and elbow, knee and heel cream; three dark grey stripes on dorsal surface of tibia; dorsal surface of discs dark grey; belly light grey, semi-translucent (Fig. 9).
Notes on the natural history.Individuals of Scinax ruberoculatus sp.nov.were observed mainly in primary and old-growth secondary lowland rainforests (39-68 m a.s.l.) where they occupied edge situations.Its breeding season was correlated with the rainy season in the northern PMRI (November-March).Active males vocalized while sitting on the vegetation in horizontal position 1-2 m above the ground around temporary ponds.The number of calling males was higher on rainy nights.In two large temporary ponds (>25 m 2 ) males of S. ruberoculatus sp.nov.shared calling sites with Dendropsophus minutus (Peters, 1872), D. rhodopeplus (Günther, 1858), D. sarayacuensis (Shreve, 1935), and Scinax sp. 1 (sensu Ferrão et al., 2016).Only males of S. ruberoculatus sp.nov.were found in small temporary ponds (<4 m 2 ).During the day, inactive individuals were observed between leaves of palm trees.
The zero-inflated model revealed a significative and positive effect of soil silt content on the occurrence of this species across the whole study area (occurrence = −5.836intercept −4.978 number of trees + 3.402 silt content + 0.242 underground water; = 3.299 silt P = 0.012).The soil silt content also explained 41% of the abundance of S. ruberoculatus sp.nov.(Fig. 11) in the plots where the species was found (abundance = −2.261+ 1.214 silt content; r 2 = 0.41, F1,9 = 6.48,P = 0.031).Forest structure (P = 0.567) and underground water (P = 0.260) did not have a significant effect on the abundance of the new species.

DISCUSSION
Scinax ruberoculatus sp.nov. is the 72nd described species of the genus Scinax (sensu Duellman, Marion & Hedges, 2016) and the 30th Scinax species known to occur in the Amazonia sensu lato (see Table S1).Recent studies have demonstrated that the species richness of the genus Scinax is greatly underestimated (Fouquet et al., 2007;Ferrão et al., 2016;Menezes et al., 2016).Despite that, the rate of species descriptions for Amazonian Scinax has been low in the last decade (2006-2016; 0.3 species per year) compared with that of other frog genera, e.g., Allobates Zimmermann and Zimmermann, 1988 (0.8 species per year) and Boana Gray, 1825 (0.7 species per year).However, the description rate of Amazonian Scinax may increase in the next years.At least other six potential unnamed species of Scinax pending formal description have recently been reported in the PMRI (Ferrão et al., 2016).Additionally, the geographic distribution of some species (e.g., S. blairi, S. cruentomma, S. iquitorum, and S. wandae) needs to be reviewed.Over the past decade, our research group has conducted standardized frog sampling along 23 permanent RAPELD sampling modules and/or grids (∼450 plots) distributed across a longitudinal gradient of approximately 1,500 km in the Brazilian Amazonia (Fig. 10).Despite this high level of sampling effort, S. ruberoculatus sp.nov.has only been observed in seven sampling modules in northern PMRI.However, sampling gaps occur in areas closer to the banks of the PMRI and the neighbouring interfluves in the southern Amazon; therefore, the range of S. ruberoculatus sp.nov.may be broader than shown here.
In this study, we addressed the influence of environmental heterogeneity on the distribution of a species of the genus Scinax in Amazonia for the first time.Our results indicated that the occurrence and abundance of S. ruberoculatus sp.nov. is positively affected by the silt content in the soil.An explanation for this phenomenon can be seen in the fact that increased silt content directly reduces the soil drainage capacity (Juo & Franzluebbers, 2003).In areas with slight altitude variations (1-3 m in the northern PMRI; Rossetti, Toledo & Góes, 2005), temporary water bodies in silty soils persist for longer periods of time compared with temporary ponds in well-drained, sandier substrates.Because temporary water bodies play a crucial role in the reproduction of S. ruberoculatus sp.nov., the soil structure represents an important ecological factor that affects the spatial distribution of the species in the study area.Alternatively, the soil texture may also affect the species composition of invertebrates in Amazonia, such as ants (Vasconcelos, Macedo & Vilhena, 2003;Oliveira et al., 2009;Souza et al., 2016), mites (Moraes et al., 2011) and termites (Dambros et al., 2013).Therefore, the distribution of S. ruberoculatus sp.nov.could be indirectly affected by the soil texture, which promotes the availability of certain invertebrate groups that represent prey species.However, the relationships between invertebrates and soil texture have not been investigated in the PMRI, and additional data are necessary to test this hypothesis.
Environmental heterogeneity may have different effects on frog distribution and abundance in the northern Amazon River depending on the reproductive mode (Menin et al., 2007;Menin, Waldez & Lima, 2011).Species that exhibit terrestrial reproduction are primarily affected by soil characteristics, such as the clay content and pH, whereas species that exhibit an aquatic reproductive mode are influenced by the tree density and distance from streams (Landeiro, Waldez & Menin, 2014).Since S. ruberoculatus sp.nov.uses temporary ponds to reproduction, our results are inconsistent with the above generalization made for aquatic breeding frogs but similar with generalization for terrestrial breeding frogs.Nevertheless, we conclude that environmental variables with the ability to filter the occurrence and abundance of species can also cause regional variations in species-habitat associations, independent of the species' reproductive modes.Future investigations should include additional frog species and a wider spatial scale to further elucidate this relationship.
By investigating a species of snouted treefrog, our study showed that alpha-taxonomy and ecology can be integrated into a single framework via the sampling of standardized units along a heterogeneous Amazonian landscape.Such an approach has the potential to reveal additional species and their ecological relationships not only in the study area but also in other megadiverse regions for which insufficient data on both the biota and environmental predictors are still available.
• Rafael de Fraga performed the experiments, analyzed the data, wrote the paper, reviewed drafts of the paper.
• Jiří Moravec analyzed the data, contributed reagents/materials/analysis tools, wrote the paper, reviewed drafts of the paper.

Figure 1
Figure 1 Sampling area in the Purus-Madeira Rivers Interfluve and schematic representation of RAPELD sampling modules and plots.(A) Distribution of RAPELD sampling modules along a 600 km transect.Legend: green colour (M1-M9) = tropical lowland rainforest with emergent canopy; gold colour (M10-M11) = open rainforest lowlands with palm trees.(B) General configuration of each module with ten sampling plots.Open squares represent plots where environmental variables used in this study were measured.BRA, Brazil.Full-size DOI: 10.7717/peerj.4321/fig-1

Figure 10
Figure 10 Geographic range of Scinax.ruberoculatus sp.nov.Numbers indicate the RAPELD sampling module.Yellow circles: RAPELD sampling modules where the new species was observed.White circles: RAPELD sampling modules where the new species was not observed.Triangles: RAPELD sampling modules outside the study area where the new species was not observed.The diameter of the yellow circles indicates the percentage of plots occupied by the new species within each sampling module (10% in 3-5; 20% in 7-8; 40% in 9; 50% in 2).Full-size DOI: 10.7717/peerj.4321/fig-10

Table 1 Measurements (in mm) of type series of Scinax ruberoculatus sp. nov.
Values are presented as the mean ± standard deviation, with the range in parentheses.