Leptodactylid frogs of the genus Adenomera Steindachner, 1867 comprise 30 described species distributed throughout South America east of the Andes (Frost 2024). The taxonomic history of this genus is very complex and, over the last 50 years, numerous systematic studies have reviewed its taxonomic validity, phylogenetic position and species diversity (Heyer 1973(Heyer , 1974;;Frost et al. 2006;Pyron and Wiens 2011;de Sá et al. 2014). The genus was originally described by Steindachner (1867) to accommodate a single species, A. marmorata. Later, Lutz (1930) synonymised the genus with Parvulus, a subgenus of Leptodactylus Fitzinger, 1826, but Parker (1932) soon gave priority to the name Adenomera and elevated it as a subgenus of Leptodactylus. Four decades later, Adenomera was resurrected by Heyer (1974) to accommodate taxa of the Leptodactylus marmoratus species group.
To avoid paraphyly of Leptodactylus rendered by Vanzolinius Heyer, 1974, Frost et al. (2006), supported by evidence from Heyer (1998) and Kokubum and Giaretta (2005), declared Adenomera a synonym of Lithodytes Fitzinger, 1843 and the latter taxon a synonym (subgenus) of Leptodactylus. Based on molecular data, Pyron and Wiens (2011) recovered Adenomera as sister to Lithodytes and this clade as sister to Leptodactylus. The authors also removed the two former taxa from the synonymy with Leptodactylus. The sister relationship between Adenomera and Lithodytes was corroborated by de Sá et al. (2014) through a total evidence analysis, which recovered the clade comprising the two genera as sister to the one grouping Hydrolaetare and Leptodactylus. Fouquet et al. (2014) performed a comprehensive phylogenetic analysis and recovered eight major clades within Adenomera: A. lutzi clade, A. heyeri clade, Adenomera sp. I clade, A. andreae clade, A. marmorata clade, A. thomei clade, A. martinezi clade and A. hylaedactyla clade. However, Carvalho et al. (2021) recovered Adenomera sp. I as A. juikitam and, based on acoustic, morphologic and genetic data, concluded that A. juikitam instead belongs to the A. heyeri clade.
The genus Adenomera displays a high prevalence of morphologically cryptic species (e.g. Angulo and Icochea (2010); Carvalho and Giaretta (2013a); Carvalho et al. (2020a); Zaracho et al. (2023)). Some species also show high levels of intraspecific polymorphism (e.g. Cassini et al. (2020)) and congeneric sympatry and syntopy are common; up to three species may occur in the same region (e.g. Carvalho et al. (2021)). These factors make species delimitation in Adenomera challenging. Nevertheless, 15 of the 30 currently recognised species were described in the last 10 years (Frost 2024) and several candidate species still await formal description (Fouquet et al. 2014). The massive advance in the taxonomy of Adenomera has been made possible by the use of integrative taxonomy (Carvalho et al. 2019a(Carvalho et al. , 2019c)). In particular, despite morphological crypsis, advertisement calls are markedly divergent amongst species and represent a powerful source of reliable diagnostic characters (Angulo and Icochea 2010;Carvalho and Giaretta 2013a, b;Carvalho et al. 2019cCarvalho et al. , 2021)).
The Adenomera andreae clade comprises four described species-A. andreae (Müller, 1923); A. chicomendesi Carvalho, Angulo, Kokubum, Barrera, Souza, Haddad & Giaretta, 2019;A. guarayo Carvalho, Angulo, Barrera, Aguilar-Puntriano & Haddad, 2020;and A. simonstuarti (Angulo & Icochea, 2010)-and two candidate species, Adenomera sp. D and Adenomera sp. T (Fouquet et al. 2014). While the A. andreae clade is restricted to Amazonia, none of the nominal species has a restricted geographic distribution. Adenomera andreae shows the widest range, being distributed throughout Amazonia (Carvalho et al. 2019c), while A. chicomendesi and A. guarayo are widely distributed in south-western Amazonia (Carvalho et al. 2019a(Carvalho et al. , 2020a)). Adenomera simonstuarti is distributed in western and south-western Amazonia (Carvalho et al. 2020b).
Adenomera simonstuarti was described from Peruvian Amazonia, based on morphological and acoustic data of four males and two females (Angulo and Icochea 2010). Subsequently, Fouquet et al. (2014) reported that the species was more widespread than previously thought, also occurring in Venezuela, Ecuador and Brazil (States of Acre and Amazonas). They also suggested the existence of more than one species hidden under the name A. simonstuarti (Fouquet et al. 2014, appendix S2a). Recently, Carvalho et al. (2020b) sequenced additional specimens from Brazil referred to as A. simonstuarti and their delimitation analysis recovered eight lineages within this name (hereafter, the A. simonstuarti complex). They also re-described the species' advertisement call, based on recordings from the type locality in Peru and an additional locality in the upper Juruá River Basin (Acre, Brazil). Based on molecular, morphological and bioacoustic data, Carvalho et al. (2020b) recognised their lineage 3 as A. simonstuarti sensu stricto. They also identified the other lineages as putative new species, pending confirmation with additional data (e.g. acoustic and morphologic data).
Poorly sampled environments in Amazonia usually harbour undocumented biodiversity of anurans (Ferrão et al. 2016;Vacher et al. 2020). Physiognomies comprising the white-sand ecosystems (hereafter, WSE) exemplify such environments (Adeney et al. 2016). The WSE occupies an area of 5% of the Amazonia and comprises two main physiognomies in Brazil: campina-open environments characterised as patches of grasslands or scrublands (canopy < 7 m) on a matrix of exposed sandy soil; and campinarana-closed-canopy, forested environments characterised by thin-trunked trees of low stature (canopy < 20 m) (Anderson 1981;Ferreira 2009;Adeney et al. 2016). Despite the increasing interest in WSE organisms (Capurucho et al. 2013;Fine and Baraloto 2016;Vicentini 2016;Lamarre et al. 2016;Borges et al. 2016;Fraga et al. 2018;Gonella et al. 2020), studies of anurans from WSE are rare. The few such studies recently published show that WSE represents a source of poorly known and new species of anurans, many of which appear to be specialists in or endemic to these environments (Carvalho et al. 2019a;Ferrão et al. 2019Ferrão et al. , 2022;;Mônico et al. 2023).
In the present study, we sampled an unreported lineage of the Adenomera simonstuarti complex from the whitesand forests of Central Amazonia and, by using integrative taxonomy, describe it as a new species.
Fieldwork was conducted between 2019 and 2023 in three long-term ecological research sites (RAPELD) in the Rio Negro Sustainable Development Reserve (hereafter, RDS Rio Negro), Municipality of Iranduba, State of Amazonas, Brazil (Fig. 1). Modules are located near km 18 (3°06'33.6"S, 60°40'29.0"W; 73 m above sea level [hereafter [a.s.l.]), km 26 (3°03'31.0"S, 60°45'42.0"W; 73 m a.s.l.) and km 50 (2°50'10.0"S, 60°50'20.0"W; 19 m a.s.l.) of the AM-352 highway. Adults were euthanised with 2% aqueous benzocaine topical solution, fixed in 10% neutral-buffered formalin and preserved in 70% ethanol. Before fixation, tissue samples of each specimen were collected and stored in 100% ethanol. Tadpoles were collected from two foam nests in the calling site of two uncollected males near the sampling module at km 18. They were euthanised as described above, fixed and preserved in 5% neutral-buffered formalin. Adults were deposited in the herpetological collections of the Instituto Nacional de Pesquisas da Amazônia (INPA-H; Manaus, Brazil), Museu Paraense Emílio Goeldi (MPEG; Belém, Brazil) and Museu de Zoologia da Universidade de Campinas (ZUEC-AMP; Campinas, Brazil); tadpoles were deposited at INPA-H.
Advertisement calls of six males of the new species (IN-PA-H 44867 [holotype], MPEG 44649, were recorded with a Sennheiser K6/ME66 unidirectional microphone (Sennheiser, Germany) coupled to a Marantz PMD660 digital recorder (Kanagawa, Japan) and with a Sony PCM-D50 digital recorder with built-in microphone. Recordings were stored in wav files with a sampling rate of 44.1 kHz and sample size of 16 bits. The microphone was positioned 50-100 cm from the calling male. Air temperature during all recordings was 25 °C. Recordings were deposited in the Neotropical Jacques Vielliard sound repository of the University of Campinas (FNJV; Campinas, Brazil) under accession numbers FNJV 59561-66.
To facilitate interspecific comparisons, 16 specimens and the advertisement calls of six males of Adenomera simonstuarti sensu stricto were collected and recorded, respectively, at Unidade de Gestão Ambiental Acurauá, Municipality of Tarauacá, State of Acre, Brazil. A specimen from this locality (INPA-H 40967) was included in the phylogenetic inference of Carvalho et al. (2020b) and nests with samples of A. simonstuarti sensu stricto from Peru and male advertisement calls of the Acre population match with those in the original description by Angulo and Icochea (2010). All males of the Acre population were found in the field by their vocalisation, ensuring that we collected the target species.
The description of external morphology of adults of the new species is based on 21 males and five females. Sex was determined through direct assessment of sexual characters: the presence of vocal slits, vocal sac and a fleshy ridge on the snout tip in males and absence in females. Maturity was determined, based on breeding behaviour in males (calling activity) and examination of secondary sexual characters in females (mature oocytes visible through the belly skin). The following 16 morphometric measurements (Watters et al. 2016) were taken to the nearest 0.1 mm using digital calipers and an ocular micrometer coupled to a stereomicroscope: snout-vent length (SVL), head length (HL), head width (HW), snout length (SL), eye-nostril distance (EN), eye diameter (ED), interorbital distance (IOD) internarial distance (IND), tympanum diameter (TD), upper arm length (UAL), hand length (HAL), forearm length (FLL), thigh length (THL), tibia length (TL), foot length (FL) and tarsus length (TSL). Toe tip development (character states) follows Heyer (1973). Snout shape follows Heyer et al. (1990). Terminology for other morphological characters follows Carvalho et al. (2020a). We follow the colour catalogue of Köhler (2012): colour names are italicised; cc, colour code. Repeated colours do not repeat codes. See Suppl. material 1: table S1 for morphometric raw data.
The larval developmental stage was determined according to Gosner (1960). The following morphometric measurements were taken with a micrometer coupled to a stereomicroscope from 10 tadpoles at stages 35 (n = 7) and 41 (n = 3): total length (TL), body length (BL), tail length (TAL), maximum tail height (MTH), tail muscle height (TMH), tail muscle width (TMW), inter-nostril distance (IND) and interorbital distance (IOD) (Altig and McDiarmid 1999); body height (BH), body width at spiracle level (BW), eye-nostril distance (END), eye diameter (ED) and oral-disc width (ODW) (Lavilla and Scrocchi 1986); body width at eye level (HW) (Lima et al. 2015); and vent-tube length (VTL) (Lins et al. 2018). Morphological description is based on seven tadpoles at stage 35. Terminology and diagnostic characters follow Altig and McDiarmid (1999) and Schulze et al. (2015).
Description of the advertisement call and the following acoustic parameters follow Carvalho et al. (2019a) and Köhler et al. (2017): call duration (CD), notes per call (NpC), note duration (ND), note repetition rate (NrR) note rise time (NrT), pulses per note (PpN), pulse duration (PD; measured for the first, central and last pulses of each note), pulse repetition rate (PrR), dominant frequency (DF), fundamental frequency (FF) and frequency modulation (FM). See Suppl. material 1: table S2 for bioacoustic raw data.
Calls were analysed with Raven 1.5.
1 (Bioacoustics Research Program 2014) configured as follows: Hamming window (size = 20 ms), filter bandwidth 65 Hz, overlap 90%, hop size 2 ms and Discrete Fourier Transform 1,024 samples. The dominant frequency and rise time were measured with the peak frequency and peak time relative functions. Figures were produced in R platform (R Core Team 2021) with the packages seewave 2.1.0 (Sueur et al. 2008) and tuneR 1.3.2 (Ligges et al. 2017). Seewave was set as follows: Hamming window, Fast Fourier Transform 256 points, overlap 90%.
Genomic DNA was extracted from tissues of four specimens of the new species using a Wizard genomic DNA Purification Kit (Promega Corp., Madison, WI, USA) according to the manufacturer's protocol. Fragments of cytochrome c oxidase subunit I (COI) were amplified through polymerase chain reaction (PCR) using the primers CHmL4 (5'-TYTCWACWAAYCAYAAAGAY-ATCGG-3') and CHmR4 (5'-ACYTCRGGRTGRC-CRAARAATCA-3') (Che et al. 2012). Reaction conditions were: 60 s at 94 °C followed by 35 cycles of 94 °C (20 s), 50 °C (50 s) and 72 °C (90 s) and final extension of 10 min at 72 °C. The final volume of the PCR reaction was 15 μl and contained 0.6 μl of 50 mM MgCl 2 , 1.2 μl of 10 mM dNTPs (2.5 mM each dNTP), 1.5 μl of tampon 10× (75 mM Tris HCl, 50 mM KCl, 20 mM (NH 4 ) 2 SO 4 ), 0.5 μl of each primer (10 μM), 9.55 μl of ddH 2 O, 0.15 μl of 1 U Taq DNA Polymerase and 1 μl of DNA (30-50 ng/μl).
The PCR products were purified using Exonuclease I and Thermosensitive Alkaline Phosphatase (Thermo Fisher Scientific, Waltham, MA, USA). Subsequent sequencing reactions were performed using standard protocols of the Big Dye TM Terminator Kit (Applied Biosystems, Waltham, USA). We used an automated sequencer ABI Prism 3130 (ThermoFisher Scientific, Waltham, USA) to sequence the amplicons. Sequences were edited with Geneious 5.3.4 (Kearse et al. 2012). Newly-generated sequences are deposited in the online repository GenBank under accession numbers OQ974333-36.
To infer phylogenetic relationships, we inserted the generated sequences into a dataset containing sequences retrieved from GenBank (Suppl. material 1: table S3). Our dataset contains the genes cytochrome b (Cytb), cytochrome c oxidase subunit I (COI), recombination activating gene 1 (RAG1) and pro-opiomelanocortin C. These sequences represent all species of the Adenomera andreae clade, including all known lineages of A. simonstuarti, as well as Adenomera sp. D and Adenomera sp. T (Fouquet et al. 2014;Carvalho et al. 2020b) and species belonging to the other six clades (Suppl. material 1: table S3). Lithodytes lineatus was used to root the tree. To align sequences of each gene, we used the MAFFT online server following default parameters under the G-INS-i strategy.
The final matrix was concatenated in Geneious 5.3.4 and comprises 53 terminals and 3,293 base pairs (bp) (667 for Cytb, 657 for COI, 1,422 for RAG1 and 547 for POMC).
We divided the dataset considering first, second and third codon positions for each protein-coding gene and we used PartitionFinder 2.1.1 (Lanfear et al. 2017) under the corrected Akaike Information Criterion (AICc) to infer partition schemes and evolutionary models. The best evolutionary models for partitions in the concatenated matrix were TIM+G for Cytb 1 st and COI 3 rd positions; SYM+I+G for Cytb 2 nd position; GTR+I+G for Cytb 3 rd position; TRNEF+I+G for COI 1 st position; F81+I+G for COI 2 nd position; TRN+I+G for RAG1 1 st and 2 nd positions; GTR+G for RAG1 3 rd and POCM 1 st positions; TVM+I+G for POMC 2 nd position; and GTR+I for POMC 3 rd position. Phylogenetic relationships were reconstructed through Maximum Likelihood (ML) using IQTREE (Nguyen et al. 2015) implemented in the online server http://iqtree.cibiv.univie.ac.at/ (Trifinopoulos et al. 2016). Clade support was estimated with 10,000 ultrafast bootstrap replicates (Hoang et al. 2018) using 5,000 maximum iterations, 3,000 replicates and a minimum correlation coefficient of 0.99. Lineage numbering within the A. simonstuarti species complex follows Carvalho et al. (2020b), except by A. simonstuarti 3, which is referred to as A. simonstuarti sensu stricto (SS) in the present study.
Based on COI alignment, we calculated pairwise genetic distances (uncorrected p-distance and Kimura two-parameter distance; Kimura (1980)) between the new species and closely related taxa of the A. simonstuarti species complex using MEGA 6 (Tamura et al. 2013). Mean distances are presented as in the main text, minimum-maximum values in the Suppl. material 1: table S4.
Due to the phenotypic similarity between A. simonstuarti sensu stricto and the new species, we performed a Principal Component Analysis (PCA) associated with a Multivariate Analysis of Variance (MANOVA) to test for a statistical difference between the morphometric multidimensional spaces of each species. Analysis was performed only for males due to the low number of females collected for A. simonstuarti sensu stricto. The same 16 morphometric measurements taken from the new species were also taken from 14 adult males of A. simonstuarti. To perform morphometric PCA, we transformed the raw data into 15 morphometric ratios: HL/SVL, HW/SVL, SL/SVL, END/SVL, IND/ SVL, ED/SVL, IOD/SVL, TD/SVL, FAL/SVL, UAL/ SVL, HAL/SVLL, TL/SVL, FL/SVL, THL/SVL and TAL/ SVL. The PCA and MANOVA were run using the functions prcomp and manova of the package stats 4.1 (R Core Team 2021); an ellipse representing the standard errors of points in the graphic representation of PCA was drawn using the function ordiellipse of the package vegan 2.5-7 (Oksanen et al. 2020) with parameter kind set as se. See Suppl. material 1: table S1 for morphometric measurements of Adenomera sp. nov. and A. simonstuarti sensu stricto.
Succinct morphological comparisons of adults were made with all 30 nominal congeners but detailed morphological
Individuals of Adenomera sp. nov. nest together as a new monophyletic lineage (bootstrap support = 100) within the A. simonstuarti species complex (sensu Carvalho et al. (2020b)), which nests within the A. andreae clade (Fig. 2). The new species is sister to the lineage A. simonstuarti 2 from the lower Juruá River in Brazil. Clades representing A. simonstuarti 2 and Adenomera sp. nov. are the shallowest within the species complex; the average p-distance for COI between them equals 2.9% (2.5-3.3%) (Table 1; Suppl. material 1: table S4). Peruvian and Brazilian individuals of A. simonstuarti sensu stricto are recovered as sister to the clade comprising A. simonstuarti 1, A. simonstuarti 2 and Adenomera sp. nov. Genetic p-distance between the new species and A. simonstuarti sensu stricto averages 5.2% (4.3-5.9%).
The first two principal components (PCs) of morphometric PCA explained ~ 53% of data variance. Spaces occupied by Adenomera sp. nov. and A. simonstuarti SS are significantly different (Pillai = 0.286, df = 32, p = 0.004) and do not overlap (Fig. 3). The three morphometric ratios that contribute most of the variation along PC1 are END/SVL, ED/SVL and DSL/SVL. See Table 2 for data regarding other PCA variables. Taxonomic account Order Anura Fischer von Waldhein, 1813 Family Leptodactylidae Werner, 1896 Subfamily Leptodactylinae Werner, 1896 Genus Adenomera Steindachner, 1867 Adenomera albarena sp. nov. https://zoobank.org/0FAB2CE9-3291-453D-8C58-E092BD288DCA Tables 3, 4, Figs 4-7, 9B-D Chresonymy. Adenomera gr. heyeri (Lima et al. 2021). Type material. Holotype. INPA-H 44867, an adult male collected at km 26 of the AM-352 highway, Rio Negro Sustainable Development Reserve (03°05'35"S, 60°40'36"W; 76 m a.s.l.), Municipality of Iranduba, State of Amazonas, Brazil, on 11 December 2020 by M. Ferrão, A. P. Lima and W. E. Magnusson. Paratypes. Twenty-four adults collected at the same locality as the holotype; eight males MPEG 44649, IN-PA-H 44868-73 and ZUEC-AMP 25694 collected on 11 December 2020 by M. Ferrão, A. P. Lima and W. E. Magnusson; four females INPA-H 44874-75, ZU-EC-AMP 25695 and MPEG 44650 collected on 10 December 2021 by M. Ferrão, A. P. Lima and B. Martins; four males INPA-H 44876-77 and MPEG 44651-52 collected on 11 December 2021 by M. Ferrão, A. P. Lima and B. Martins; four males INPA-H 44878-80 and ZUEC-AMP 25696 collected on 12 December 2021 by M. Ferrão, A. P. Lima and B. Martins; a male INPA-H 44881 collected on 19 January 2022 by B. Martins; a male INPA-H 44882 collected on 3 February 2022 by B. Martins; a male ZUEC-AMP 25697 and a female IN-PA-H 44883 collected on 14 May 2022 by B. Martins. One adult male INPA-H 44885, collected at km 50 of the AM-352 Highway, Rio Negro Sustainable Development Reserve (2°50'10.0"S, 60°50'20.0"W), Municipality of Iranduba, State of Amazonas, Brazil, on 12 January 2023 by B. Martins. Etymology. The specific epithet albarena is formed by the combination of two Latin words: "alba" (white) and "arena" (sand). This is a reference to the white-sand forests of central Amazonia, the distinctive environment inhabited by this species.
Vernacular names. White-sand terrestrial foam-nesting frog (English), rana terrestre de arena blanca (Spanish) and rãzinha da areia branca (Portuguese).
Diagnosis. The species Adenomera albarena is recognised by the following combination of characters. Interspecific comparisons. Adenomera albarena differs from all congeners, except A. simonstuarti by having a nearly solid dark-coloured stripe on the underside of the forearm (Heyer 1973(Heyer , 1975;;Kwet and Angulo 2002;Almeida and Angulo 2006;Kok et al. 2007;Kwet 2007;Angulo and Reichle 2008;Berneck et al. 2008;Kwet et al. 2009;Angulo and Icochea 2010;Carvalho andGiaretta 2013a, 2013b;Carvalho et al. 2019aCarvalho et al. , 2019bCarvalho et al. , 2019cCarvalho et al. , 2019dCarvalho et al. , 2020aCarvalho et al. , 2020cCarvalho et al. , 2021;;Cassini et al. 2020;Zaracho et al. 2023).
Amongst Amazonian congeners, adult male Adenomera albarena have SVL 21.2-23.0 mm, which is smaller than A. glauciae (SVL 27.6-30.4;Carvalho et al. (2020b)), A. gridipappi Carvalho et 6.1 ± 0.2 (5.9-6.5) 6.2 ± 0.4 (5.5-6.8) 6.2 ± 0.5 (5.8-6.5) (2019a, 2019b, 2019c)), it also differs acoustically from these species (see below).
The advertisement call of Adenomera albarena is composed of incomplete pulses, which differs from A. aurantiaca, A. guarayo, A. inopinata and A. phonotriccus (complete pulses; Carvalho et al. (2019bCarvalho et al. ( , 2020aCarvalho et al. ( , 2021))). Amongst those with incomplete pulses, calls of A. albarena sp. nov. differ from A. amicorum, A. glauciae, A. gridipappi and A. simonstuarti sensu stricto by having a single note (multi-note calls; Angulo and Icochea (2010) 2021)) and A. simonstuarti sensu stricto (1,224 Hz;this study). The dominant frequency of Adenomera albarena sp. nov. is placed in the second harmonic, which differs from A. simonstuarti sensu stricto (dominant frequency in the fundamental harmonic).
Lack of labial teeth distinguishes tadpoles of Adenomera albarena from exotrophic tadpoles of A. guarani, A. saci and A. thomei (present in all mentioned species; De la Riva (1995); Almeida and Angulo (2006); Carvalho and Giaretta (2013a)). Endotrophic tadpoles of A. albarena differ from those of A. hylaedactyla and A. andreae by the presence of a spiracle (absent in all mentioned species; Menin et al. (2009); Menin and Rodrigues (2013)); from A. marmorata by an internarial distance 44-52% of IOD (IND/IOD = 74%; Heyer et al. (1990)).
Description of the holotype. Adult male (Figs 4A, B, C, 5A, C). Dorsal skin glandular, warty on flank. Dorsolateral fold indistinct. Sacral region, dorsal surface of tibia and posterior surface of tarsus with white-tipped tubercles. Vertebral stripe in sacral region. Throat, belly and ventral surface of limbs smooth. Pair of lumbar glands. Posterior surface of thigh with a pair of paracloacal glands. Snout subovoid in dorsal view and acuminate in profile. Nostril closer to the snout tip than to the eye and orientated dorsolaterally; fleshy ridge on the snout tip. Eye nostril distance 83% of eye diameter, eye diameter equals internarial distance. Head wider than long. Internarial distance > 25% of head width. Canthus rostralis defined; loreal region slightly concave. Triangular interorbital blotch. Tympanum distinct, nearly 60% of eye diameter; black-coloured supratympanic fold well developed, extending from posterior corner of eye to base of arm. Postcommissural gland ovoid. Subgular vocal sac; vocal slits present. Vomerine teeth in two straight rows posterior to choanae and arranged in transverse series parallel to choanae. Tongue lanceolate (sensu Duellman (1970)) and free behind. Relative finger lengths IV < I ≈ II < III; fringes or webbing on fingers absent; finger tips rounded, slightly expanded, but without disc; inner metacarpal tubercle elliptical; outer metacarpal tubercle rounded; distinct rounded greyish subarticular tubercles on the underside of fingers; supernumerary tubercles rounded; antebrachial tubercle absent. Elliptical axillary gland. Tibia slightly longer than thigh (TL/THL = 1.01). Relative toe lengths I < II < V < III < IV; toe tips flattened or slightly flattened, with visible expansions (character states C and D, sensu Heyer (1973)); fringes or webbing absent; inner metatarsal tubercle elliptical; outer metatarsal tubercle rounded. Tarsal fold from the inner metatarsal tubercle extends 2/3 of tarsus length. Subarticular tubercles elliptical or rounded, supernumerary tubercles rounded. Colour of the holotype in preservative. See Figs 4A-C, 5A, C. Iris chrome orange (colour code [cc] 74). Snout tip Pale neutral grey (cc 296), as well as the fleshy ridge. Blotches on upper and lower lips Pale neutral grey. Postcommissural gland with melanophores. Tympanum Dark grey (cc 45). Supratympanic fold Vandyke brown (cc 281). Thoracic dorsal surface of body Hair brown (cc 277); lumbar region Cinnamon-drab (cc 50) with white-tipped tubercles. Interorbital region Sepia (cc 279). Flank Pale buff (cc 1). Triangular interorbital blotch Dark greyish-brown (cc 284). Dorsal surface of forelimbs Pale buff (cc 1) with Drab (cc 19) blotches. Dorsal surface of hind-limbs Tawny olive (cc 17); transverse bars Sepia with white-tipped tubercles. Vertebral stripe Pale buff in sacral region. Paracloacal region and lumbar glands Sepia. Throat Light buff (cc 2) with melanophores and a greater density around the jaw; belly Light buff and chest and underside of limbs pale Pinkish buff (cc 3). Underside of forearm with nearly solid Brownish-olive (cc 276) stripe. Palm of hand, sole of foot, digits and subarticular tubercles almost completely covered with melanophores. Stripe in metatarsal and proximal and medial phalanx Fuscous (cc 283) (Figs 5C, D). Intraspecific variation. Morphometric variation of the new species is summarised in Table 3. The type series shows three dorsal colour patterns: dark blotches few or absent (Figs 6A, 7A); many dark blotches (Figs 6B, 7D); and a dorsolateral stripe (Figs 6C, 7G). Sixty-eight percent of the type series (including the holotype) has the first pattern, 20% have many dark blotches and only 12% have a dorsolateral stripe. A sacral stripe is present in 64% of specimens. About 52% of paratypes have a toe tip shape in stage D (sensu Heyer (1973)), while 48% have an intermediate shape, stages C or D. All individuals have a triangular mark on the head (Fig. 6), which is less visible in individuals that lack dorsal blotches. Texture of the dorsum varies from rough to smooth, with few to many glandules. Throat and belly show slight variation in melanophore density (Fig. 7B, E, H).
Advertisement call. The advertisement call of Adenomera albarena consists of a single note with partially fused pulses. Pulse number varies from 11 to 21; pulse duration from 4 to 23 ms; and pulse repetition rate from 94 to 138 pulses per second. Note duration varies from 100 to 199 ms and note repetition rate from 0.6 to1.2 notes per minute. The fundamental frequency of the note coincides with the first harmonic and varies from 1,765 to 2,239 Hz; the dominant frequency varies from 3,746 to 4,349 Hz and corresponds to the second harmonic (Table 4; Fig. 8).
Tadpole. Body elliptical in dorsal and ventral views, globular in lateral view (Fig. 9A-C); BH 86-96% and 51-58% of BW and BL, respectively (Table 5). Body length 36-41% of TL, with a well-marked constriction in the postorbital region (Fig. 9A) that is more pronounced in dorsal than ventral view and less marked in some tadpoles. Snout rounded in dorsal and lateral views; eye-nostril distance 6-7% of BL. Internarial region convex; internarial distance 44-52% of IOD. Nostrils small, rounded, located and directed anterolaterally; visible in lateral view, poorly visible in dorsal view; closer to snout than to eyes. Border of external nares without fleshy marginal rim; border smooth, slightly below the level of marginal region. Eyes large, diameter 34-39% and 139-190% of IOD and END, respectively; located laterally, but directed anterolaterally. Interorbital region slightly concave; interorbital distance 46-50% of BW. Spiracle very small, single, sinistral; directed posterodorsally, located immediately above the edge of the lateral body surface at the level of the hind-limb insertion; poorly visible in dorsal and lateral views. Spiracular opening elliptical; aperture small, narrower than spiracle width, inner wall fused to the body wall. Widely coiled intestines positioned ventrally, perpendicular to main body axis, concealing other internal organs. Vent tube medial with sinistral displacement; comma-shaped, directed upwards; opening very small, rounded, directed dorsally; mostly free from the ventral fin, dorsal wall attached only near the body junction. Tail moderately long, length 142-178% and 59-64% of BL and TL, respectively; low, maximum tail height 83-94% of BH; tip rounded. Musculature moderately robust with strongly acuminate tip, which reaches the tail tip; higher than wide near the tail-body junction, tail muscle width 70-79% of TMH; tail muscle height 43-49% of MTH. Dorsal fin external margin slightly convex, originating from posterior third of body; higher than body, slightly higher than ventral fin, with maximum height at its central portion.
Ventral fin external margin poorly arched; maximum height at mid-tail; same height as or shorter than body. Oral disc small, ODW 29-33% of BW; unemarginated; located and directed anteroventrally (Fig. 9D). Upper and lower labium present; upper labium with 4-6 short, rounded papillae on lateral region, interleaved by a large medial gap, but arranged in a straight line; posterior labium projected posteriorly, with 3-6 short rounded papillae laterally, interleaved by a medial gap, but arranged in a straight line; anterior gap larger than posterior. Submarginal papillae absent. Jaw sheath keratinised only at the external borders; upper jaw sheath arch-shaped, lower jaw sheath V-shaped. Serrations on each sheath extend its entire length. Labial teeth absent; two labial ridges on the upper labium and three on the lower, formulae 2(2)/3 (1).
In preservative, dorsal surface of body brown; anterior half of body darker and with more melanophores than posterior half, with very fine translucent vermiculation on posterior half. Dorsal hind-limbs translucent with numerous melanophores. Tail mostly translucent grey, brown at the tail/body junction; caudal musculature whitish-grey; fins translucent grey; small melanophores on the caudal musculature. Vent tube translucent grey. Ventral surface of body translucent grey with numerous melanophores anteriorly; posterior portion translucent, except for lateral regions, which are light brown.
Distribution, habitat and natural history. Adenomera albarena is known only from the white-sand ecosystems between West Negro and the Solimões Rivers, specifically in the RDS Rio Negro and nearby localities, Municipalities of Iranduba and Manacapuru, Amazonas, Brazil, where these ecosystems are dominant (Figs 1, 10A). Two other species of Adenomera occur in sympatry: A. hylaedactyla and A. andreae. Although these species occur in the same region, their habitat is not the same; A. andreae mainly inhabits non-flooded forests, while A. hylaedactyla occurs in open areas. On the other hand, Adenomera albarena inhabits white-sand forests subject to flooding regimes close to streams. At the type locality, A. andreae and the new species occur in syntopy at the border between forests subject to flooding regimes and those that do not flood. Males call from the ground, above or hidden in the leaf litter (Fig. 10B). They start calling at ~ 16:00 h and continue calling until ~ 19:00 h. Isolated individuals sometimes call later, but it is unusual.
Adult males are found easily and juveniles are also not difficult to observe, but females are very secretive (Fig. 10C). Males excavate underground chambers in which foam nests are built and females deposit their eggs (Fig. 10D). The chambers are very difficult to find when they are located under leaf litter amongst the roots of palm trees and ferns.
Conservation. The known geographic distribution of Adenomera albarena comprises an area of approximately 150 km 2 within the RDS Rio Negro and nearby localities. Although the species is known only from a small area, it is very common there and likely occurs in other parts of the RDS Rio Negro and, potentially, in the nearby Jaú National Park. Despite its abundance, the new species occurs exclusively in white-sand forests subject to flooding regimes, which are highly vulnerable to anthropogenisation (e.g. from pollution, deforestation, mining, free-ranging livestock, irregular occupation and recreational use of water). Indeed, several riverine areas in this environment at RDS Rio Negro, including the type locality, have already been impacted by some of these anthropogenic drivers. Long-term monitoring that compares populations between pristine and anthropogenised areas is essential to evaluate whether and how these drivers impact the conservation status of A. albarena.
For many years, several undescribed species of Amazonian anurans have been erroneously assigned to nominal species that were believed to be geographically widespread (e.g. Allobates caldwellae Lima et al. 2020, Atelopus manauensis Jorge et al. 2020; Pristimantis guianensis Mônico et al. 2022). However, the integration of molecular, acoustic and morphological data has allowed contemporary taxonomists to more accurately delimit and describe such cryptic species (Fouquet et al. 2014;Carvalho et al. 2021;Moraes et al. 2022). Crypsis amongst species of the genus Adenomera, combined with their syntopy and sympatry, has been especially challenging, as exemplified by the A. simonstuarti species complex. Fouquet et al. (2014) recovered six lineages within this complex (including A. simonstuarti sensu stricto). Carvalho et al. (2020b) subsequently considered additional data and delimited eight candidate species. However, none of those candidate species corresponds to the one from RDS Rio Negro that we describe here as a new species. As it is the easternmost taxon within the A. simonstuarti complex, this species complex may be more widespread and diverse than previously thought.
Considering the current taxonomic uncertainty of lineages Adenomera simonstuarti 1-2 and 4-8, the description of A. albarena may introduce taxonomic instability by rendering A. simonstuarti as paraphyletic in a few scenarios of species delimitation (e.g. conspecificity of A. simonstuarti SS + lineage 1 + lineage 2). In addition to a previous molecular analysis that supports eight candidate species (Carvalho et al. 2020b) and the phenotypic divergence between A. albarena and A. simonstuarti SS (present study), Fouquet et al. (2014) document phenotypic divergence between A. simonstuarti SS and A. simonstuarti 4 (DF 2,830 Hz in A. simonstuarti 4 vs. 1,224 Hz in A. simonstuarti SS), which are also distinct from A. albarena (DF 3,349 Hz). Moreover, the slight genetic divergence between A. albarena and A. simonstuarti 1-2 and the large divergence between them and A. simonstuarti SS support the heterospecificity of the two nominal species. A comprehensive taxonomic revision is needed to resolve these taxonomic issues, but it may take decades to complete due to the difficulties in obtaining necessary field samples. On the other hand, solving lineage-by-lineage issues is much less time-consuming and can support conservation planning at the species level across threatened landscapes, such as the whitesand environment (WSE) inhabited by A. albarena.
Adenomera albarena is the first species of its genus to be described from a WSE. Although this species is likely a WSE specialist, its known distribution is limited to WSE in the Negro-Solimões interfluve. Other frogs may be specialists in or endemic to WSE in this interfluve, including Scinax albertinae (Ferrão et al. 2022), Pristimantis campinarana (Mônico et al. 2023) and Osteocephalus vilarsi (Ferrão et al. 2019). Moreover, the geographic distributions of two additional candidate species, Rhinella aff. proboscidea and Pristimantis aff. orcus, apparently are associated with WSE and limited to the West Negro-Solimões interfluve (unpublished data). The congruence of frog species sharing the same habitat specialisation and pattern of restricted distribution highlights the Jaú Region as an area of endemism within Amazonia (Borges and da Silva 2012).
Two reproductive modes are reported for Adenomera: endotrophic tadpoles that complete development entirely in a subterranean foam nest without an exotrophic feeding phase (mode 32, sensu Haddad and Prado (2005)); and exotrophic tadpoles with early developmental stages completed in a subterranean foam nest and later free-living aquatic stages (mode 30, sensu Haddad and Prado (2005)). A spiracle is present in all Adenomera with known exotrophic tadpoles (A. guarani, A. saci and A. thomei;Heyer (1973); Almeida and Angulo (2006); Carvalho and Giaretta (2013a); Zaracho and Kokubum (2017); Zaracho et al. (2023)). Amongst species with endotrophic tadpoles, the spiracle is absent in A. andreae and A. hylaedactyla (Heyer and Silverstone 1969;Kokubum and Sousa 2008;Menin et al. 2009;Menin and Rodrigues 2013) and present in A. marmorata (Heyer et al. 1990), A. aff. hylaedactyla from south-eastern Brazil (Kokubum and Giaretta 2005) and A. albarena. The condition of the spiracle differs interspecifically, ranging from a tube free from the body wall in some species to an attached tube in other species. The larval life history and morphological diversity make Adenomera an interesting model to investigate the evolution of reproductive features in anurans at small phylogenetic scale. Morphological and behavioural studies of Adenomera are urgently required for such evolutionary studies, since tadpoles are known for only a few species.
Amongst the three sympatric species of Adenomera living in white-sand environments at RDS Rio Negro, all of them with endotrophic tadpoles, only A. albarena inhabits forests subject to flooding and has subterranean foam nests that submerged for at least a few days; flooding habitats are not occupied by A. andreae and A. hylaedactyla. Hence, the spiracle in A. albarena may increase larval survivorship during short flooding periods, enabling the occupancy of flooding habitats by this species. Detailed ecophysiological studies evaluating survival rates in larvae of these three species under distinct flooding conditions are needed to test this hypothesis.
Martins, B. da C. et al.: New Adenomera from white-sand forests of Brazilian Amazonia