A Pan-Amazonian species delimitation: high species diversity within the genus Amazophrynella (Anura: Bufonidae)

Amphibians are probably the most vulnerable group to climate change and climate-change associate diseases. This ongoing biodiversity crisis makes it thus imperative to improve the taxonomy of anurans in biodiverse but understudied areas such as Amazonia. In this study, we applied robust integrative taxonomic methods combining genetic (mitochondrial 16S, 12S and COI genes), morphological and environmental data to delimit species of the genus Amazophrynella (Anura: Bufonidae) sampled from throughout their pan-Amazonian distribution. Our study confirms the hypothesis that the species diversity of the genus is grossly underestimated. Our analyses suggest the existence of eighteen linages of which seven are nominal species, three Deep Conspecific Lineages, one Unconfirmed Candidate Species, three Uncategorized Lineages, and four Confirmed Candidate Species and described herein. We also propose a phylogenetic hypothesis for the genus and discuss its implications for historical biogeography of this Amazonian group.


INTRODUCTION
Amphibians are undergoing a drastic global decline (Beebee & Griffiths, 2005). This decline is primarily attributable to habitat destruction, diseases (chytrid fungus) and global climate change (Collins, 2010). In Amazonia the primary threat is habitat destruction, although the chytrid fungus has reached the Amazon basin (Valencia-Aguilar et al., 2015;Becker et al., 2016), and is starting to have an impact on Amazonian and Andean anurans (Lötters et al., 2005;Lötters et al., 2009;Catenazzi & Von May, 2014). Most Amazonian amphibians are thought to have broad, often basin wide distributions, although their geographic distributions are generally poorly known. More detailed analyses generally reveal the existence of multiple deeply divergent lineages, suggesting cryptic diversity. Fouquet et al. (2007a) and Fouquet et al. (2007b) estimated that amphibian diversity of Amazonia is underestimated by 115%, while Funk, Caminer & Ron (2012) suggest this underestimate is closer to 150-350%. But even without taking into account the high levels of crypsis or pseudocrypsis (morphological differences apparent but overlooked) in widespread Amazonian anurans, Amazonia has the highest diversity of amphibians on this planet (Jenkins, Pimm & Joppa, 2013).
Delimiting species and their geographic distributions is therefore crucial for the understanding of impacts on the biodiversity of Amazonian anurans, and for the assessment of their conservation status (Angulo & Reichle, 2008). Previous studies suggest a prevalent conservatism in the morphological evolution of anurans (e.g., Elmer, Dávila & Lougheed, 2007;Robertson & Zamudio, 2009;Vences et al., 2010;Kaefer et al., 2012;Rowley et al., 2015), thus, species delimitation based solely on morphological characters may fail to differentiate among species. Conversely, delimiting species solely based on molecular characters or genetic distances harbors potential pitfalls that have been well documented (e.g., Carstens et al., 2013;Sukumaran & Knowles, 2017). Environmental data also have the potential to provide important information to taxonomy since species have distinct ecological requirements that determine their occurrence in time and space (Soberón & Peterson, 2005). Therefore, species delimitation relying on a pluralistic approach seeking to unite several lines of evidence (Dayrat, 2005;Padial et al., 2010) generally provides robust and consensual taxonomic hypotheses (e.g., Padial & De La Riva, 2009) especially in morphologically conserved groups, i.e., taxonomic groups harboring cryptic or pseudocryptic taxa (Cornils & Held, 2014).
Until 2012, only two species were recognized: Amazophrynella minuta from western Amazon and Amazophrynella bokermanni from eastern Amazon (Fouquet et al., 2012b). Since 2012 five additional species have been described from western Amazon (Amazophrynella vote, Amazophrynella manaos, Amazophrynella amazonicola, Amazophrynella matses and Amazophrynella javierbustamantei). The taxonomy of the genus remains, however, far from being resolved (Rojas et al., 2016). Although molecular phylogenetic analyses in Fouquet et al. (2012b), Rojas et al. (2015) and Rojas et al. (2016) provided evidence for the existence of multiple lineages, the scarcity of material suitable for morphological and bioacoustic analyses prevented the description of these lineages as new species.
In this study, we revisit the genus Amazophrynella, include specimens from new localities, and reconstruct intra-and inter-specific phylogenetic relationships. We delimit candidate species based on molecular data and subsequently seek support for these lineages combining qualitative and quantitative morphological data and environmental evidence. As a result of these analyses, we formally describe four new species of Amazophrynella from Brazil, Ecuador, French Guiana and Peru, and identify additional seven candidate species. Additionally, we provide new insights into the overall phylogenetic relationships for the genus, and discuss biogeographic history of this Amazonian group.

Protocol for species delimitation
We evaluated the status of populations of Amazophrynella, adhering to the unified species concept proposed by Queiroz De (2007), that conceptualizes species as lineages of ancestordescendent populations which maintain their distinctness from other such lineages and which have their own evolutionary tendencies and historical fates. We followed the consensus protocol of integrative taxonomy proposed by Padial et al. (2010). The concept of candidate species adopted in this study follows the subcategories proposed by Vieites et al. (2009) in using: Confirmed Candidate Species (CCS) for lineages that present high genetic distance and can be differentiated by other traits (i.e., morphological data), Deep Conspecific Lineages (DCL) for lineages that are genetically divergent but not supported by any other character (these characters being available), Unconfirmed Candidate Species (UCS) for lineages that are genetically divergent but no additional characters are available to support this divergence (these characters not available) and Uncategorized Lineages (UL) for lineages that do not corresponds to any of the above categories.

Focal species and morphological examination
Field work and visits to museum collections were carried out between 2011 and 2017. Field collection of specimens followed the technique of visual encounter surveys and pitfallbarrier traps (Crump & Scott Jr, 1994). Museum acronyms are found in Sabaj (2016) except for Museo de Biodiversidad del Peru (MUBI; this collection is part of Museo de Historia Natural, Universidad Nacional de San Antonio Abad, Cusco, Peru). Collecting permits in Peru were granted by Dirección General Forestal y de Fauna Silvestre del Ministerio del Medio Ambiente (MINAN; No. AUT-IFS-2017-055), in Ecuador by Ministerio del Ambiente (MA; 001-1-IC-FAU-DNB/MA) and in Brazil by the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio;No. 39792-1 and No. 32401). The material of Amazophrynella teko from Mitaraka (French Guiana) was collected during the ''Our Planet Reviewed'' expedition, organized by the MNHN and Pro-Natura International.
We examined topotypical material of A. minuta deposited at the collection of Amphibians and Reptiles of the Instituto Nacional de Pesquisas da Amazônia-INPA (INPA-H) and three syntypes (NHMG 462,NHMG 463,NHMG 464) deposited at the Göteborgs Naturhistoriska Museum, Sweden; five specimens of A. bokermanni (Izecksohn, 1993) from near the type locality (c. 30 Km) deposited at the INPA collection; the type series of A. vote (Ávila et al., 2012) deposited at the Coleção Zoológica de Vertebrados of the Universidade Federal de Mato Grosso-UFMT, Cuiabá, Mato Grosso, Brazil (UFMT-A) and INPA; A. manaos (Rojas et al., 2014) deposited at the INPA; A. amazonicola and A. matses (Rojas et al., 2015) deposited at the Museo de Zoología-Universidad Nacional de la Amazonia Peruana-UNAP and A. javierbustamantei (Rojas et al., 2016) deposited at the Museo de Biodiversidad del Peru (MUBI), Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (MHNSM). A list of the examined specimens is found in Appendix S1.
Qualitative morphological terminology was according to Kok & Kalamandeen (2008). Morphological comparison between specimens were made through visual inspection of diagnostic characters that include: dorsal skin texture, ventral skin texture, head shape, shape of palmar tubercle, relative length of fingers and venter coloration (Fouquet et al., 2012b;Rojas et al., 2014;Rojas et al., 2015;Rojas et al., 2016). We used ventral incision to perform gonadal analyses. Developmental stages of tadpoles were determined using Gosner's protocol (1960). Descriptive terminology, morphometric variables and developmental stages of tadpoles follow Altig & McDiarmid (1999). Spectral and temporal parameters of advertisement calls (when available) were analyzed in the software Praat for Windows (Boersma & Weenick, 2006). Bioacoustic terminology followed Köhler et al. (2017).

Morphological quantitative analyses
Quantitative measurements of body were obtained with a digital caliper (0.1 mm precision) following Kok & Kalamandeen (2008) with the aid of an ocular micrometer in a Leica stereomicroscope. Measurements were taken from the right side of specimens, and, if this was not feasible, from the left side. Measurements were: SVL (snout-vent length) from the tip of the snout to the posterior margin of the vent; HL (head length) from the posterior edge of the jaw to the tip of the snout; HW (head width), the greatest width of the head, usually at the level of the posterior edges of the tympanum; ED (eye diameter); IND (internarinal distance), the distance between the edges of the nares; SL (snout length) from the anterior edge of the eye to the tip of the snout; HAL (hand length) from the proximal edge of the palmar tubercle to the tip of finger III; UAL (upper arm length) from the edge of the body insertion to the tip of the elbow; THL (thigh length) from the vent to the posterior edge of the knee; TL (tibia length) from the outer edge of the knee to the tip of the heel; TAL (tarsal length) from the heel to the proximal edge of the inner metatarsal tubercle; FL (foot length) from the proximal edge of the inner metatarsal tubercle to the tip of toe IV. We rounded all measurements to first decimal place to avoid pseudoprecision (Hayek, Heyer & Gascon, 2001).
Principal Component Analyses (PCA) were performed on residuals obtained by linear-regressing each variable on SVL, thus removing the effects of size. We used only males specimens because of absence of females in some lineages. The PCA was used to detect groups representing putative species. We also performed a discriminant Function Analysis (DFA) to identify morphometric variables that contribute the most to species separation and to test the classification of specimens into mtDNA lineages. For DFA we used morphometric size-free data set. To determine the number of correct and incorrect assignments of specimens to each of the mtDNA lineages, we jackknifed our data matrix. The significance of differences of morphological variables among mtDNA lineages was tested using the Kruskal-Wallis (KW) non-parametric test. All the statistical analyses (PCA, DFA and KW) were performed in R v3.4.3 (R Development Core Team, 2008) using the stats package and setting the significance cut-off at 5%.

DNA amplification
DNA extraction, gene amplification and sequencing was carried out using standard protocols (Appendix S2). Sequence data were deposited in GenBank under the accession numbers MH269714-MH270330 (Table S2A).

Phylogenetic analyses and species delimitation
We collected molecular data for 230 individuals of Amazophrynella from 35 localities, including topotypical material for all nominal species and encompassing the entire distribution of the genus. We obtained a total of 1,430 bp from three mitochondrial loci (16S rRNA (16S), 480 bp; 12S rRNA (12S), 350 bp; and Cytochrome oxidase subunit I (COI), 600 pb (see Appendix S4, Table S4A)). The edition and alignment of the sequences was performed using Geneious v.6.1.8. (Kearse et al., 2012) and the Clustal W algorithm (Thompson, Gibson & Higgins, 2002). We used only unique haplotypes for phylogenetic reconstruction. We concatenated all loci, treating them as a single partition evolving under the same model of molecular evolution. The best model of molecular evolution (GTR+G+I) was estimated in JModelTest (Posada, 2008) and selected using the Akaike Information Criterion-AIC. Phylogenetic analyses were performed using Bayesian Inference (BI) using MrBayes 3.2.1. (Huelsenbeck & Ronquist, 2001). We generated 10 7 topologies, sampling every 1,000 th topology and discarding the first 10% topologies as burn-in. The stationarity of the posterior distributions for all model parameters was verified in Tracer v1.5 (Rambaut & Drummond, 2009). From the MCMC output, we generated the final consensus treemaximum clade credibility tree using Tree Annotator v1.6.2 (part of the Beast software package). For visualization and edition of the consensus maximum clade credibility tree, we used the program Figtree v.1.3. (Rambaut, 2009).
We used a Poisson tree processes (PTP) model (Zhang et al., 2013) to infer the most likely number of species in our dataset, as implemented in the bPTP server (http://species.h-its.org/ptp/). The PTP model is a simple, fast and robust algorithm to delimit species using non-ultrametric phylogenies, ultrametricity is not required because the algorithm models speciation rates by directly using the number of substitutions. The fundamental assumption is that the number of substitutions between species is significantly higher than the number of substitutions within species. In a sense, this is analogous to the GMYC (General Mixed Yule Coalescent) approach that seeks to identify significant changes in the rate of branching events on the tree. However, GMYC uses time to identify branching rate transition points, whereas, in contrast, PTP directly uses the number of substitutions (Zhang et al., 2013). For input, we used a BI tree estimated by MrBayes. We ran the PTP analyses using 10 5 MCMC generations, thinning value of 100, a burn-in of 10%, and opted for remove the outgroup to improve species delimitation. Convergence of MCMC chain was confirmed visually. To ensure that the lineages detected using PTP presented high genetic distance (>3.0%, sensu (Fouquet et al., 2007a;Fouquet et al., 2007b) we calculated uncorrected p-distance using the 16S mtDNA (Vences et al., 2005) in the program MEGA 7.0 (Kumar, Stecher & Tamura, 2016).

Environmental analyses
The environmental analyses were undertaken in order to test if delimited species occur in distinct climatic environments (Soberón & Peterson, 2005). We retrieved high resolution bioclimatic layers (30 arc-seconds ∼1 km, present environmental conditions) using the Community Climate System Model (CCSM4) from the WorldClim project (http://www.worldclim.org/) (Hijmans et al., 2005). To avoid geographic pseudoocurrence of points, localities were filtered using the program Geographic Distance Matrix Generator 1.2.3. (Ersts, 2014) considering a threshold of 1 km between localities. The localities of each lineage used for analyses are in Appendix S3, Table S3A.
To identify environmental variables that were most informative and test the classification of specimens into mtDNA lineages using ecological variables, we performed Principal Component Analysis (PCA) and Discriminant Function Analysis (DFA) separately for each lineages/species of the eastern and western clades. The analyses were performed using the 19 BioClim environmental variables in WordClim. Probability of correct assignment of individuals to lineages was tested using jackknife.

Electronic publication of new zoological taxonomic names
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:1C6046BE-CFC4-4060-A1CA-0C9C9C1C7A0A. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS.

Phylogenetic and species diversity
The concatenated data resulted in a strongly supported phylogeny ( Fig. 1), with high degree of divergence among putative and nominal species of Amazophrynella. The PTP model of species delimitation detected a total of eighteen lineages (posterior probability = 0.48-0.91) (Appendix S4, Fig. S4A) of which seven are nominal species and 11 are candidate species.  The phylogeny of Amazophrynella recovered the presence of two clades diverging basally, both strongly supported: one distributed in eastern and other in western Amazonia (see Fig. 1A). The eastern clade was formed by two strongly supported subclades, herein called northeastern (NE) and southeastern (SE) clades. The northeastern clade included three lineages and the southeastern clade seven lineages. The western clade was formed by two well supported subclades, herein called northwestern (NW) and southwestern (SW) clades. Both subclades were composed of four lineages (see Fig. 1A). Uncorrected p-distances for 16S mtDNA between pairs of sister lineages are presented in Table 1. Each lineage presented high genetic divergence (>3.0%) compared to its sister taxon and ranged between 3.0-3.2% (3.0 ± 0.1) to 4.0-6.0% (5.0 ± 0.1).
Our timetree recovered Dendrophryniscus as sister taxon of Amazophrynella (see Appendix S5, Fig

Morphological analyses
A total of 468 specimens (adult males and females) were examined for comparative morphological analyses (Table 2); these analyses did not include Amazophrynella aff. matses sp, A.sp2 and A sp3 (see Fig. 1). Measurements of males and females are presented in Tables 3 and 4. For morphometric analyses (Principal Components Analyses-PCA and Discriminant Function Analyses-DFA) we used 237 adult male specimens (87 from the eastern clade and 148 from the western clade). The specimens used in morphometric analyses are listed in Appendix S6.  The PCA of the eastern and western clades revealed a grouping of specimens based on morphometric traits and allowed us to distinguish all the mtDNA lineages in multivariate space (Figs. 3A and 3B). Character loadings, eigenvalues and percentage of variance explained for PCA (PC I-II) for morphometric variables for the eastern and western clades are provided in Appendix S7 and Table S7A-S7B.
In the eastern clade specimens of each lineage can be successfully separated based on morphometric traits using PCA (Fig. 3A). The first two principal components extracted by the PCA account for 57.7% of the variation found in the dataset. The first component (PC1) explained 37.48% of the total variation and the second component (PC2) explained 20.29% of the variation. Using DFA a total of 80% of specimens were correctly classified to phylogenetic groups. The number of individuals correctly assigned to each clade by DFA are presented in Table 5. The DFA showed that the variables that contributed the most to the morphometric separation were snout length, tarsal length, and head width. Head measurement traits (head width, head length, snout length, and intranasal distance) explained 93% of the classification by the first two discriminant axes (Appendix S8, Figs. S8A-S8B). Loadings and percentage of variance explained for discriminant axes (F1-2) of morphometric variables in eastern clade are provided in Appendix S8 and Table S8A).
In the western clade specimens of each lineage can be successfully separated based on morphometric traits using PCA (Fig. 3B). The first two principal components extracted

Environmental analyses
We obtained a total of 90 unique localities for final analysis, 43 localities of the eastern and 47 localities of the western clade, representing the occurrences of all species but Amazophrynella aff. matses sp, A.sp2 and A sp3 (see Fig. 1). The list of localities used Table 5 Male classification in morphological space. Successful classification in morphological space (males) recovered phylogenetic mt DNA lineages (Eastern and Western clades). In parentheses, the percentage of successfully classification. The numbers in the cells represent the numbers of individuals assigned to each clade by discriminant analyses. UCS and UL were not included.

Lineages (Eastern clade)
A. manaos (90%) for environmental analyses and discriminant function analyses are in Appendix S3 and Table S3A. The PCA of the eastern and western clades revealed a grouping of specimens based on environmental traits and allowed us to distinguish all the mtDNA lineages in the multivariate space (Figs. 3C and 3D). Character loadings, eigenvalues and percentage of variance explained for PCA (PC 1-2) analyses for environmental variables for the eastern and western clades are provided in Appendix S7 and Tables S7C-S7D.
In the eastern clade specimens of each lineages can be successfully separated based on environmental traits using PCA (Fig. 3C). The first two principal components extracted by the PCA account for 87.71% of the variation found in the dataset. The first component (PC1) explained 73.28% of the total variation and the second component (PC2) explained 14.43% of the variation. A total of 65% of specimens were correctly classified to their lineage. The numbers of individuals correctly assigned to each clade by DFA are presented in Table 6. The environmental variables that most contributed to separating lineages were mean temperature of the coldest quarter (bio11), maximum temperature of the warmest month (bio5), mean diurnal temperature range (bio2) and isothermality (bio3) (Appendix S8, Figs. S8A-S8C). Loadings and percentage of variance explained per discriminant axes (F1-2) of environmental variables in the eastern clade are provided in Appendix S8 and Table S8B. Table 6 Male classification in environmental space. Successful classification in environmental space recovered phylogenetic mt DNA lineages (Eastern and Western clades). In parentheses, the percentage of successful classifications. The numbers in the cells represent the numbers of individuals assigned to each clade by discriminant analyses. UCS and UL were not included.

Lineages (Eastern clade)
A. manaos (77%) A. teko sp. nov. In the western clade specimens of each lineages can be successfully separated based on environmental traits using PCA (Fig. 3D). The first two principal components extracted by the PCA account for 95.55% of the variation found in the dataset. The first component (PC1) explained 95.37% of the total variation and the second component (PC2) explained 0.18% of the variation. A total of 81% of specimens were correctly assigned to their candidate species. The numbers of individuals correctly assigned to each clade by DFA are presented in Table 6. The environmental variables that most contributed to group separation were annual mean temperature (bio1), mean diurnal temperature range (bio2), mean temperature of the warmest quarter (bio10) and mean temperature of the wettest quarter (bio8) (Appendix S8 and Figs. S8A-S8D). Loadings and percentage of variance explained for discriminant axes (F1-2) of environmental variables in the western clade are provided in Appendix S8 and Table S8B.

Taxonomic decisions
Our data analysis of Amazophrynella suggest the existence of 18 linages of which seven are nominal species, three Deep Conspecific Lineages, one Unconfirmed Candidate Species, three Uncategorized Lineages and four Confirmed Candidate Species ( Table 2). The four CCSs presented at least one diagnostic morphological character, monophyly with a strong phylogenetic support using the standard DNA barcode 16S fragment (Vences et al., 2005) and divergence from its sister taxon at environmental and morphometric data. Based on these results, herein we described A. teko sp. nov., A. siona sp. nov., A. xinguensis sp. nov., and A. moisesii sp. nov.
Comparison with other species (characteristics of compared species in parentheses). Amazophrynella teko sp. nov. is morphologically most similar to A. manaos from which it can be distinguished by: large SVL of males 12.9-15.8 mm, n = 13 (vs. 12.3-15.0 mm,      Variation (Fig. 9). There is little variation among the examined specimens. Sexual dimorphism was observed in SVL, with 12.9-15.8 mm (14.7 ± 0.8 mm, n = 13) in males and 17.9-21.5 mm (19.2 ±1.8 mm, n = 17) in females. Specimens (MNHN 2015.137, MNHN 2015.138, MNHN 2015.139, MNHN 2015) present lesser abundance of granules on arm insertion. In some individuals (MNHN 2015.143) the ventral and the dorsolateral region present one to three large tubercles. Subarticular tubercles more protruding and swollen in females. Blotches on belly display different sizes (larger vs. small, see Fig. 10). In life, venter coloration between cream to off-white . Palm and sole between light red and orange. In preserved specimens, the palmar tubercle is more flattened than in life.
Coloration of the holotype (in life). Head black brown, in dorsal view. Dorsum brown. Flanks brown. Scattered tubercles on flanks white. Dorsal surfaces of upper arm, arm and hand black. Dorsal surfaces of thighs, tibia, tarsus and foot black. Ventral surfaces of upper arm, arm and palm cream. Ventral surfaces of thighs cream, mottled with black blotches. In dorsal view, tarsus and tibia creamy, sole light red. Gular region brown. Belly cream with black tiny blotches. Posterior region of the thigh and cloaca with black blotches. Longitudinal white stripe on upper jaw extending from nostril to tympanum. Iris golden and pupil black. Color in preservative (∼70% ethanol, Fig. 10). Almost the same as color in life. We noted the progressive loss of dorsal coloration which eventually becomes black. The chest lost its coloration and became less intense. The dark blotches on venter became less evident. The coloration of the fingers and toes became pale red.
Bioacoustics (Fig. 11  French Guiana. All call parameters described by Lescure & Marty (2000) show an overlap with our recorded calls. Call trill emitted at regular intervals. pulses/call, n = 29). Time to peak amplitude between 0.06-0.13 s (0.08 ± 0.02 s, n = 29). The call has a downward modulation, reaching its maximum frequency near its beginning. Distribution and natural history (Fig. 1B). Amazophrynella teko sp. nov. have been recorded from the district of Saint Laurent du Marioni, Saint Georges and Camopi, French Guiana, the state of Amapá, Brazil and in the southern region of Suriname (A Fouquet, pers. obs., 2017). It occurs at elevations ranging from 70 m a.s.l. to 350 m a.s.l. The species is diurnal and crepuscular but is also active at night during peak breeding period, which normally occurs at the beginning of the rainy season (January-February). This species shows a conspicuous sexual dimorphism, with males being much smaller than females. The conservation status of this species remains unknown. The habitat destruction and pollution must affect their populations; however, due to its abundance we believe that this species probably needs not be classified above Least Concern category.
Etymology. The specific epithet is a noun in apposition and refers to the name of the Teko Amerindians who occupy the southern half of French Guiana; the area occupied by the Teko tribe also encompasses the type locality.

Coloration of the holotype (in life)
. Head brown, in dorsal view. Dorsum mostly brown. Flanks reddish brown. Dorsal surfaces of upper arm, arm and hand light brown. Dorsal surfaces of the thighs, tibia, tarsus and foot light brown. Ventral surfaces of upper arm light red, arm light brown, palm reddish brown. Gular region reddish brown. Belly bright red with yellow blotches. Axillar region with yellow granules. Ventral surfaces of thighs, tarsus and tibia reddish brown, sole reddish brown. Iris golden and pupil black.
Color in preservative (∼70% ethanol, Fig. 14). Almost the same as color in life. Dorsum became brown. We detected a gradual fading of the red and yellow coloration of the chest and venter. The blotches on venter became less evident. Fingers and toes became pale red. Tadpoles (Fig. 15). Duellman & Lynch (1969) described the tadpole of Amazophrynella siona sp. nov. as Atelopus minutus based on ten individuals at stage 31 and three at stage 40, from Sarayacu, Province of Pastaza, 400 m a.s.l. The morphological characteristics described by Duellman & Lynch (1969) are similar to those observed by us. We analyzed ten tadpoles at stage 30. Body ovoid in dorsal view. Total length 11.0-13.2 mm (11.5 ± 0.84 mm). Body length 3.6-4.8 mm (4.2 ± 0.3 mm); depressed in lateral view. Body height 1.2-1.9 mm (1.5 ± 0.2 mm), body widest posteriorly. Snout rounded in dorsal and lateral view. Eye diameter 0.3-0.5 mm (0.3 ± 0.1 mm). Eye snout distance 0.9-1.4 mm (1.2 ± 0.14 mm). Nostrils small, closer to eyes than to tip of snout. Inter nasal distance 0.5-0.75 mm (0.6 ± 0.1 mm). Inter orbital distance 0.5-0.75 mm (0.6 ± 0.09 mm). Spiracle opening single, sinistral and conical. Spiracle opening on the posterior third of the body. Centripetal wall fused with the body wall and longer than the external wall. Upper and lower lips bare, single row of small blunt teeth, sectorial disc absent. Jaw sheaths finely serrated. Two upper and three lower rows of teeth. Oral disc weight 0.8-1.1 mm (0.9 ± 0.1 mm). Dorsal fin originating on the tail-body junction, increasing in height throughout the first third of the tail and decreasing gradually in the posterior two thirds of the tail to a pointed tip, in lateral view. Ventral fin originating at the posteroventral end of the body, higher at the first third of the tail, decreasing gradually in height toward tail tip. Tail length 5.4-8.1 mm (6.8 ± 0.9 mm). Tail height 0.9-1.1 mm (0.9 ± 0.1 mm). Body and tail rosaceous with small dark pointed flecks on body in fixed specimens. In life, Duellman & Lynch (1969) reported brown body and spotted tail with black and small brown flecks on caudal musculature, the entire dorsal fin and posterior third of ventral fin.
Bioacoustics (Fig. 16). The advertisement call of Amazophrynella siona sp. nov. was described by Duellman (1978) as the advertisement call of Dendrophryniscus minutus from Santa Cecilia, Ecuador. We analyzed one call from the Reserva de . The call has a downward modulation, reaching its maximum frequency almost at the middle. Distribution and natural history (Fig. 1B). Amazophrynella siona sp. nov. have been recorded from Ecuador, in Provinces of Orellana, Sucumbíos and Pastaza and Peru in the Province Andoas, northern Loreto Department. It occurs at elevations ranging from 200-900 m a.s.l. The species is found in the leaf litter of primary and secondary forest, terra firme or flooded forest, and swamps. It is active during the day; at night individuals rest on leaves, usually less than 50 cm above ground. It breeds throughout the year (Duellman, 1978). This species shows conspicuous sexual dimorphism, with males being much smaller than females. The amplexus is axillar. Eggs are pigmented; males call from amidst leaf litter. Duellman & Lynch (1969) reported that this species deposited its eggs in gelatinous strands 245-285 mm long, with 245-291 eggs. It can be abundant at some sites (e.g., Cuyabeno reserve; SR Ron, pers. obs., 2018) Given its large distribution range (>20,000 km 2 ) which also includes vast protected areas and locally abundant populations, we suggest assignment this species to the Least Concern category.
Etymology. The specific epithet is a noun in apposition and refers to the Siona, a western Tucanoan indigenous group that inhabits the Colombian and Ecuadorian Amazon. The Siona inhabit the Cuyabeno Lakes region, an area where Amazophrynella siona sp. nov. is be abundant. While working in his undergraduate thesis in the early 1990s, SRR lived with the Siona at Cuyabeno. The Siona chief, Victoriano Criollo, had an encyclopedic knowledge of the natural history of the Amazonian forest, superior in extent and detail to that of experienced biologists. His death, a few years ago, represents one of many instances of irreplaceable loss of traditional knowledge triggered by cultural change among Amazonian Amerindians. A . teko,A. manaos,A. vote,A. minuta,A. bokermannni,A. javierbustamantei,A. matses,A. Amazonicola,A. siona sp. nov. A. teko sp. nov.,A. moisesii sp. nov. see Fig. 6 Measurement of the holotype (in mm). SVL 18.5, HW 5.0, HL 6.0, SL 3.1, ED 2.1, IND 1.6; UAL 6.6; HAL 4.1, FI 1.9, FII 1.9, THL 9.7, TAL 9.3, TL 5.7, FL 6.4. Variation (Fig. 18). Color in preservative (∼70% ethanol, Fig. 19). In preservative, the coloration is almost the same than life. The coloration of the dorsum became dark brown. Gular region and venter became white. The iris loses its coloration. The fingers and toes became cream.
Distribution and natural history (Fig. 1B). Amazophrynella xinguensis sp. nov. have been recorded from State of Pará, Brazil, at three localities: PDS Virola Jatoba, municipality of Anapú, Fazenda Paraiso, municipality of Senador José Porfirio (right bank of Xingu River) and Ramal dos Cocos, municipality of Altamira (left bank of Xingu River), all of them in area of influence of the Belo Monte dam. It occurs in elevations of 86-106 m a.s.l. This species is found amidst leaf litter. The amplexus is axillar (Fig. 18C). Reproduction occurs in the rainy season in tiny puddles. Males were found hidden in the leaf litter. Tadpoles and Holotype (Fig. 20) Variation (Fig. 23). Phenotypically, the new species present some variation among specimens. Sexual dimorphism was observed in SVL, with 12.2-15.8 mm (14.3 ± 1.5 mm, n = 15) in males and 16.4-20.9 mm (18.5 ± 1.6 mm, n = 15) in females. Some specimens present greater abundance of granules on dorsum (e.g., UFAC-RB 2690). Some individuals present greater abundance of small tubercles on dorsolateral region (e.g., UFAC-RB 2611, UFAC-RB 2603, UFAC-RB 2689, UFAC-RB 2692). Another specimen (UFAC-RB 2610) presents brown chevrons extending from the head to the vent, in dorsal view. Some individuals (e.g., UFAC-RB 829) present a line on dorsum, extending from the tip of the snout to cloaca. The pale yellow coloration of ventral surfaces may extend from thighs to the chest or just to the middle of the venter. In some specimens, the irregular black dots on venter vary in abundance and size (e.g., Figs. 24B vs. 24E). In life and preserved specimens, venter coloration between pale yellow and yellow. In some
Our results suggest that the genus harbors more than twice as many species as current estimates. In the last several years the systematics and taxonomy of the genus Amazophrynella has begun to be elucidated (Ávila et al., 2012;Rojas et al., 2014;Rojas et al., 2015;Rojas et al., 2016). Resulting from these studies, five new species (A. vote, A. manaos, A. amazonicola, A. matses and A. javierbustamantei-previously mistaken for A. minuta) were described. With the description of the four new species in this study, the total number of nominal species reaches 11 (Fig. 25), representing an important increase in species diversity of the genus. The number of undescribed species as a percentage of total is concordant with estimates from previous studies aiming to elucidate the species diversity of Amazonian frogs (e.g., Elmer, Dávila & Lougheed, 2007;Fouquet et al., 2007a;Fouquet et al., 2007b;Padial et al., 2012;Ron et al., 2012;Caminer & Ron, 2014;Gehara et al., 2014;Ferrão et al., 2016). Therefore, our study adds to this growing body of studies, and confirms the hypothesis that the species diversity within Amazophrynella is much higher than currently stability, but as a consequence continues, albeit to a lesser degree, to underestimate the true species diversity of Amazonian anurofauna.
A limiting factor of our study was the use of a single molecular marker (16S, 12S and COI mtDNA loci). The potential limitations for species delimitation using mtDNA have been discussed in literature (e.g., Rannala & Zang, 2003;Yang & Rannala, 2010;Dupuis, Roe & Sperling, 2012;Fujita et al., 2012). The use of additional nuclear markers is generally recommended as the use of these unliked markers has the potential to improve the accuracy of phylogenetic reconstructions and species delimitation. In spite of having used only mtDNA loci, our study also provides an extensive new morphological dataset, bioacoustic data and accurate collecting locality information which allowed us to associate environmental data with each specimen. All these additional data support and reinforce the inference based on the mitochondrial genes.
Our phylogenetic analysis also reveals a striking biogeographic pattern with a basal eastern and western divergence followed by a northern and southern split within both eastern/western clades (Fig. 2). Our basal east-west pattern dated to the Miocene and match similar patterns and divergence times detected in other groups of frogs (Symula, Schulte & Summers, 2003;Noonan & Wray, 2006;Funk et al., 2007;Garda & Cannatella, 2007;Fouquet et al., 2014). Paleoenvironmental reconstructions of Amazonian history suggest that there was a large lacustrine region in western Amazon which began to form at the beginning of the Miocene (∼24 Ma) (Hoorn et al., 2010). This lake and marshland system, known as Lake Pebas, existed in southwestern Amazonia, and was drained first to the north and then to the east (Hoorn et al., 2010). Paleoenviromental data suggest marine incursions into western Amazon during the Miocene, and Noonan & Wray (2006), for example, suggest the importance of these incursions for the diversification of Amazonian anurofauna. In general, however, marine incursions remain largely untested as a diversifying force (Noonan & Wray, 2006;Garda & Cannatella, 2007;Antonelli et al., 2010). In addition, it is reported that in early Miocene, the Purus arch was still active, and was a prominent landscape feature in central Amazon (Wesselingh & Salo, 2006;Figueiredo et al., 2009;Caputo & Soares, 2016) thus this geological formation also could explain the east-west pattern as well. While other hypotheses, such as Pleistocene refugia have also been proposed to explain this east-west pattern of diversity (Pellegrino et al., 2011), the Miocene marine incursions have the best temporal concordance with the basal east-west divergence pattern observed in Amazophrynella and other Amazonian anuran groups.
The northern and southern split within both the eastern and western clades occurred in early Miocene (∼20.1 Ma) in the eastern Amazonia clades, while the diversification of the western Amazonian clade commenced in the Middle Miocene (∼16.5 Ma). The beginning of the diversification of these clades appears to be asynchronous and therefore is unlikely attributable to a single event. The more recent date of diversification of the western clade is likely to have followed the last marine incursion, i.e., a colonization of newly available habitat in western Amazon from eastern Amazon. Independent of the absolute timing these divergence events, the four subclades are restricted to north and south of the Amazon River, a common pattern in many vertebrates species groups analyzed at the Amazonia-wide scale (e.g., Kaefer et al., 2012;Ribas et al., 2012;Fouquet et al., 2015;Oliveira, Carvalho & Hrbek, 2016). In the case of Amazophrynella species, ecological characteristics such as small body side, being a terra firme species and being restricted to reproducing in puddles (Rojas et al., 2016), clearly evidences these species' inability to disperse across rivers. This in turn implies that major Amazonian rivers should limit the distributions of lineages of Amazophrynella, a pattern observed in our phylogeny. However, the role of rivers in driving diversification of Neotropical frogs remains controversial (see Vences & Wake, 2007vs. Lougheed et al., 1999. But it is clear that geological and climatic changes in the Miocene and Pliocene played an important role in the diversification of Amazonian vertebrates (Bush, 1994;Glor, Vitt & Larson, 2001;Da Silva & Patton, 1998;Symula, Schulte & Summers, 2003;Santos et al., 2009;Kaefer et al., 2012;Fouquet et al., 2014;Gehara et al., 2014). However, only future process-based studies and biogeographic hypotheses testing will allowed us to reveal the mechanisms (e.g., dispersion, vicariance, founder event) by which Amazophrynella diversified. and Rommel R. Rojas (in Brazil), MINAN 0320 of Rommel R. Rojas (in Peru), and 001-1-IC-FAU-DNB/MA of Santiago R. Ron. In French Guiana, species of Amazophrynella are not protected and sampling does not require a license.

DNA Deposition
The following information was supplied regarding the deposition of DNA sequences: The sequences are available as a Supplemental File.

Supplemental Information
Supplemental information for this article can be found online at http://dx.doi.org/10.7717/ peerj.4941#supplemental-information.