We obtained two genetic samples of “Afrixalus cf. laevis” during our fieldwork in the Congolian lowland rainforests in DRC in 2014 and 2023. Several individuals in a very early larval developmental stage were collected from a gelatinous mass surrounding the egg clutch on a leaf and stored in 96% ethanol (IVB-H-CD14-034; IVB-H: herpetological collection in Studenec, Institute of Vertebrate Biology of the Czech Academy of Sciences, Brno, Czech Republic). The larvae were collected in the Dalangba Forest near Lindi River, near Bafwabianga village, Tshopo Province, northeastern DRC (1.1543°N, 26.8082°E, 510 m a.s.l.) on June 16, 2014 (Fig. 1A, B). No other sightings of “Afrixalus cf. laevis” were recorded during the extensive fieldwork of our team in the lowland rainforests of DRC, until a single adult male (IVB-H-CD23-0847) was recently found calling atop a Ficus tree above a shallow, partially dried up swamp near Isandja-Bomongili village, Salonga National Park, Tshuapa Province, central DRC (2.0526°S, 21.3832°E, 455 m a.s.l.) on October 23, 2023 (Fig. 1C, D). The male, after recording its advertisement call, was collected, euthanized, and its muscle tissue sample was stored in 96% ethanol.

Figure 1. 

Afrixalus lacustris A, B clutch with developing larvae, DNA barcoded (IVB-H-CD14-034), found in a folded leaf near Bafwabianga village, Tshopo Province, DRC. The leaf was picked, opened and photographed on the ground C, D adult male (IVB-H-CD23-0847) from near Isandja-Bomongili village, Salonga National Park, Tshuapa Province, DRC, in day-time coloration from dorsolateral and ventral view, the black bar corresponds to 10 mm.

Species identities of the two samples were verified via DNA barcoding (Vences et al. 2012). Total genomic DNA was extracted using GeneJet Genomic Purification Kit (Thermo Fisher Scientific, USA). Fragments of the 16S rRNA gene (hereinafter 16S) of mitochondrial DNA were amplified using polymerase chain reaction and the 16SL1 (forward) and 16SH1 (reverse) primers for the IVB-H-CD14-034 sample (Palumbi et al. 1991; 539 bp), and 16Sc (forward) and 16Sd (reverse) primers for the IVB-H-CD23-0847 sample (Evans et al. 2003; 871 bp) [note: 16SH1 and 16Sd bind to the same region]. Obtained 16S sequences were compared with publicly available data (GenBank) using the BLAST search tool (Altschul et al. 1990). The most similar data were downloaded, aligned with our new data, and uncorrected p-distances were calculated using MEGA v. 11.0.13 (Tamura et al. 2021) after the complete deletion of gaps and missing data (Dolinay et al. 2021), resulting in 446 homologous sites. The newly generated sequences were deposited in the GenBank online database (IVB-H-CD14-034: PQ351303; IVB-H-CD23-0847: PQ351304.

To construct the phylogenetic tree, we used the same methodological approach as Greenbaum et al. (2022), reflecting their population-level divergence dating analysis, supplemented with our new data (Table 1). The analysis was performed on a 902 bp-long alignment using BEAST 1.10.4 (Suchard et al. 2018), GTR substitution model, coalescent tree prior, constant size growth prior, and a strict molecular clock set to 0.02 substitution/site per million years (Greenbaum et al. 2022). The analysis was run in triplicates for 10 million generations each, with sampling every 1000^th generation. The first 10% were discarded as burn-in, after convergence and effective sample size values were inspected using Tracer v. 1.7.2 (Rambaut et al. 2018), and a maximum clade credibility tree with median heights was created from the remaining post-burn-in 27,000 combined trees using LogCombiner 1.10.4 and TreeAnnotator 1.10.4 (Suchard et al. 2018).

The advertisement call of the male (IVB-H-CD23-0847; snout-vent length, SVL = 21 mm) was recorded on a hand-held recorder Zoom H5 using a shotgun microphone Zoom SGH-6. The recording was obtained at 20:15 at 23.7 °C ambient temperature from a distance of 2–3 meters. The analysis of the recording was performed in SoundRuler v. 0.9.6 (Gridi-Papp 2007) and Raven Lite v. 2.0.5 (The Cornell Lab of Ornithology, Ithaca). The terminology of acoustic parameters is following Köhler et al. (2017). The recording was deposited in the FonoZoo online database with the reference number 14858 (https://www.fonozoo.com).

Due to the existence of the isolated record from Sankuru, some distribution maps of “A. laevis” sensu lato have shown the range as continuous from Cameroon, across the Congo, to southwestern Uganda (IUCN SSC Amphibian Specialist Group 2013; Channing and Rödel 2019). However, in reality, the range was always known as composed of two disjunct areas in the west (Cameroon, Gabon; A. laevis sensu stricto) and east (DRC, Rwanda, Uganda; A. lacustris, A. phantasma), with the Sankuru record located in between but closer to the eastern area (Schiøtz 1999). Greenbaum et al. (2022) assigned the Sankuru specimens to A. lacustris, but with a note that this population requires additional scrutiny with molecular data.

In this study, we present two new records of A. lacustris substantially extending its distribution in the Congo Basin westward into lowland rainforests, as the BLAST comparisons of 16S retrieved A. lacustris as the most similar species for both our samples of “Afrixalus cf. laevis” (Fig. 2A, B). The IVB-H-CD14-034 sample was found to differ from the previously published A. lacustris 16S data by 1.5% uncorrected p-distance, the IVB-H-CD23-0847 sample differed by 1.9%, and the two samples differ from each other by 2.0%, which do not reach the 3% threshold suggested for identifying a potential candidate species in anurans (Vieites et al. 2009). The dated phylogenetic reconstruction performed on 16S (Fig. 2B) shows a very similar topology to that of Greenbaum et al. (2022) although with slightly younger age estimates. The split between A. lacustris and its sister species A. phantasma is estimated to have occurred 1.11 million years ago, Mya (0.74–1.51 Mya, 95% highest posterior densities, HPD) during the Early Pleistocene (Calabrian). Six distinct mitochondrial lineages (haplogroups) with uncertain relationships are identified within A. lacustris, which originated during the Middle Pleistocene (Chibanian; Fig. 2B, C, Table 1). The first divergent lineage is represented by the single sample IVB-H-CD23-0847 from the Central Congolian Lowland Forests (L-C haplogroup). Its divergence is estimated to have occurred ~650 thousand years ago (hereinafter as kya; 400–970 kya) at the beginning of the Middle Pleistocene. The second diverging lineage consists of a population inhabiting the Albertine Rift in the south of the A. lacustris range (Itombwe Plateau, DRC, and its vicinity; AR-S haplogroup), from where the holotype originates. Diversification within the AR-S haplogroup is estimated to have occurred ~150 kya (50–280 kya), roughly corresponding with the beginning of the Late Pleistocene. The third lineage is a single sample from Kahuzi-Biega, DRC (UTEP 22423; AR-C1 haplogroup). The fourth and fifth lineages form a common clade (although with low support) and are represented by a sample from Irangi, DRC (near Kahuzi-Biega; UTEP 20810, AR-C2 haplogroup), and the sample IVB-H-CD14-034 from the lowland Dalangba Forest in northeastern Tshopo Province, DRC (L-NE haplogroup), respectively. The sixth mitochondrial lineage consists of samples from southwestern Uganda and adjacent DRC (haplogroup AR-N). Diversification within the AR-N haplogroup is estimated to have occurred ~210 kya (110–360 kya), at the end of the Middle Pleistocene. Two subgroups (a, b), with their diversifications corresponding with the beginning of the Late Pleistocene, are detectable in the AR-N haplogroup, but they have only weak supports (Fig. 2B, Table 1). The two subgroups have partially sympatric distribution in Uganda, and one subgroup (AR-N (b)) has been found only in Uganda, suggesting that a forest refugium was probably located in this area during the Late Pleistocene. The diversification into the six main mitochondrial lineages (haplogroups) of A. lacustris can probably be attributed to Middle Pleistocene climatic fluctuations and their impact on the suitable forest environment, which has undergone repeated fragmentations (Maley 1996; Zachos et al. 2001). However, the lack of statistical support for the relationships among the six identified mitochondrial lineages prevents further discussion on the historical biogeography of this species.

Figure 2. 

A map of all known distribution sites of Afrixalus lacustris. Red symbols mark new localities presented in this study; the red question mark denotes Boteka – a site in need of verification, where “A. laevis” was collected. Green polygons mark Salonga National Park. White symbols denote previously known localities of A. lacustris summarized by Greenbaum et al. (2022); the white star corresponds to the type locality. Orange polygons show the known distribution range of A. laevis, questionable areas are marked with orange question marks. EG = Equatorial Guinea, Rwa = Rwanda, Bur = Burundi B phylogenetic tree and C phylogeographic map of A. lacustris based on Greenbaum et al. (2022) with the addition of our new data (in bold). Black dots indicate highly supported nodes, numbers at nodes denote estimated divergence time (Mya), and blue bars denote 95% HPD intervals. A single representative of A. phantasma is shown as an outgroup. Haplogroups are distinguished by different colors and abbreviations placed on the branches (see Distribution and phylogeography, and Table 1). The maps were created in ArcGIS v. 10.8.1 (Esri Inc., https://www.esri.com), with land cover visualized by implementing results of the GlobCover project (Arino et al. 2012), and country and provincial boundaries and shaded relief background downloaded from https://naturalearthdata.com.

The range extensions lie in two areas (Fig. 2A). The first site is located in the north of the known A. lacustris range in Tshopo Province, approximately 200 km westward from the nearest locality (Epulu, Ituri Province; Greenbaum et al. 2022). The second site is situated in the south of the A. lacustris range, near the northern bloc of Salonga National Park in Tshuapa Province, and lies approximately 260 km to the northwest of the isolated record from Omaniundu in Sankuru Province (Laurent 1982). The discussed record from Omaniundu (as “A. laevis” in Laurent 1982), with specimens exhibiting partly distinct morphology (Greenbaum et al. 2022), thus probably indeed represents A. lacustris. Our two new records suggest that A. lacustris is probably rather widespread in the Northeastern and Central Congolian Lowland Forests. However, the paucity of its findings in the field—we had failed to find this species during every year of fieldwork between 2015 to 2022—confirms that A. lacustris hides excellently in the foliage of shrubs in dense forests, which explains why this species has not been detected in many areas despite its wide geographic distribution (Laurent 1982; Greenbaum et al. 2022). Afrixalus laevis sensu stricto is similarly difficult to find (Greenbaum et al. 2022) and its southeastern extent is not well known. Besides Bioko Island, it is with certainty confirmed from Cameroon and Gabon (Greenbaum et al. 2022 and citations therein), but possibly extending deeper into the Congolian rainforests (Frost 2024). Unpublished data from the “Museum” database of the Royal Museum for Central Africa, which remains to be properly examined, indicate that “A. laevis” was also collected in 1985 in Boteka, Équateur Province, DRC (approx. 0.40°S, 19.11°E, 320 m a.s.l.). Whether this record represents an even more westward range extension of A. lacustris, an eastward extension of A. laevis, or potentially a new species, must yet be investigated.

The earlier larval sample (Fig. 1A, B) provides an important view into the reproductive biology of A. lacustris, as it uncovers that this species folds leaves to make a nest for its egg clutch. This is in line with its phylogenetic position, which is deeply divergent from its morphological convergent, A. laevis, which deposits eggs on leaf surfaces without folding the leaf (e.g., Amiet 2012). However, it is not known with certainty, besides the sister species A. phantasma, which other species are the most closely related to A. lacustris, whether the East African montane species A. dorsimaculatus, A. morerei and A. uluguruensis, or West African to northern Central African A. weidholzi, or A. orophilus from the Albertine Rift (Portik et al. 2019; Greenbaum et al. 2022; Nečas et al. 2022). Afrixalus dorsimaculatus and A. uluguruensis do not enfold egg masses in leaves, similar to A. laevis; A. weidholzi glues leaf edges into a nest during oviposition, while it is probably not known in A. morerei and A. orophilus (Harper and Vonesh 2002; Harper et al. 2010; Channing and Rödel 2019; Dehling et al. 2023).

The second sample provides DNA-based identification of the calling male (SVL = 21 mm), which was initially found on a Ficus tree around 1.5–2 m above the ground, then disturbed, escaped and re-found on the top of the tree about 3 m above the swampy ground (Figs 1C, D, 3D). The advertisement call consists of four to five pulsed notes (4.8 mean for 37 measured calls of the single found male; Fig. 3A–C). Each note consists of 10 or 11 pulses, but the pulsation towards the end of the note was often obscured in the waveforms. Call duration in A. lacustris averages at 210 ms (195–220 ms, only five-note calls measured, N = 31; 167 ms average for four-note calls, N = 6; 23.7 °C). In the sister species A. phantasma, Greenbaum et al. (2022) documented longer durations of five-note calls (they reported five to six, rarely four notes per call), however their recordings were made in montane habitats at substantially lower temperatures with a trend of shortening call duration with increasing temperature (from 572–620 ms at 10.9 °C to 388–397 ms at 16.2 °C, five-note calls only). If we consider this trend in A. phantasma, our recording of A. lacustris made at 23.7 °C approximately fits with the mean call duration of 210 ms to a theoretically expected value in A. phantasma at the same temperature. The mean dominant frequency 3720 kHz (3627–3795 kHz) of A. lacustris is similar to the dominant frequency of A. phantasma, as measured at the higher temperature (3660–3810 kHz, 16.2 °C). Greenbaum et al. (2022) also observed a similarly obscured pulsation in the rear part of notes in A. phantasma, discussed as probably caused by echo effects. However, as we found a similar veiling in A. lacustris, and a similar characteristic was found in most of the notes in A. orophilus (Dehling et al. 2023), it is possibly a normal characteristic of the advertisement call in these species. In general, the advertisement calls of A. lacustris and A. phantasma are very similar, which is not surprising in sister species of frogs with parapatric distribution (see e.g. Gvoždík et al. 2015).

Figure 3. 

Afrixalus lacustris and its advertisement call A part of a call series (oscillogram, 9 s), with ambient frog chorus in the background B oscillogram of a five-note advertisement call C spectrogram of the respective call D the recorded male A. lacustris (IVB-H-CD23-0847) in nocturnal coloration in situ, perching on a leaf from where it was calling.

When we compare the advertisement call of A. lacustris with A. laevis (Amiet and Goutte 2017), with which it was previously confused, the general characteristics of the calls are quite different. Thus, the morphological convergence is not mirrored in the phonetic parameters, which might be related to partially different habitats. For example, reproduction of A. lacustris may occur more frequently in stagnant waters, whereas that of A. laevis in streams. If the advertisement call of A. lacustris is compared with other potentially closely related species (A. dorsimaculatus, A. morerei, A. orophilus, A. uluguruensis, A. weidholzi; Portik et al. 2019; Greenbaum et al. 2022; Nečas et al. 2022), none of them displays a substantial similarity of their advertisement calls; perhaps only A. weidholzi has partially similar call characteristics (Schiøtz 1999; Amiet and Goutte 2017; Dehling et al. 2023).

Despite our relatively intense fieldwork in the Congolian lowland rainforests during last 10 years (especially GB), we have found A. lacustris only twice, in two distant areas in Tshopo and Tshuapa provinces, in both cases based on single findings – one clutch with larvae and a single calling male. In the first case (Tshopo), the habitat was dense vegetation overhanging a drying muddy place near a sandy-bottomed stream in primary forest, where also small rainwater pools and puddles were present nearby, as well as a large river (Lindi; Fig. 4A). In the second case (Tshuapa), the habitat was a small drying swamp in a forest opening, located near a forest edge (local guides informed us that the swamp is much larger, potentially connected with flooded forest, after heavy rains; Fig. 4B). In both cases, habitats were in accordance with the A. lacustris habitat description of Greenbaum et al. (2022). Among sympatric amphibians, we found in Tshopo: Arthroleptidae: Arthroleptis sp. aff. variabilis, Arthroleptis tuberosus Andersson, 1905; Hyperoliidae: Hyperolius bolifambae Mertens, 1938, H. langi Noble, 1924, H. ocellatus Günther, 1858; Pipidae: Xenopus pygmaeus Loumont, 1986, X. ruwenzoriensis Tymowska & Fischberg, 1973; Ptychadenidae: Ptychadena christyi (Boulenger, 1919); Ranidae: Amnirana cf. albolabris (Hallowell, 1856); and in Tshuapa: Arthroleptidae: Arthroleptis sp. aff. variabilis, Leptopelis christyi (Boulenger, 1912); Hyperoliidae: Hyperolius cf. kuligae Mertens, 1940 (eggs only), H. cf. veithi Schick, Kielgast, Rödder, Muchai, Burger & Lötters, 2010; Phrynobatrachidae: Phrynobatrachus sp. aff. auritus; Pipidae: Hymenochirus cf. boettgeri (Tornier, 1896), Xenopus pygmaeus; Ptychadenidae: Ptychadena aequiplicata (Werner, 1898); Ranidae: Amnirana cf. albolabris; Rhacophoridae: Chiromantis cf. rufescens (Günther, 1869) (taxonomy and nomenclature follow Badjedjea et al. 2022). Hyperolius spp. at both sites and Chiromantis cf. rufescens were found with A. lacustris in syntopy on the same or nearby shrubs.

Figure 4. 

Habitats of Afrixalus lacustris in the Congolian lowland rainforests A Dalangba Forest, near Bafwabianga village, Tshopo Province, northeastern DRC B Isandja-Bomongili, Salonga National Park, Tshuapa Province, central DRC; junior authors after the finding of A. lacustris; note foam nests of Chiromantis cf. rufescens (white double arrow).

The authors have declared that no competing interests exist.

No ethical statement was reported.

This study was funded by the Czech Science Foundation (23-07331S) and the Ministry of Culture of the Czech Republic (DKRVO 2024–2028/6.l.b, National Museum of the Czech Republic, 00023272), GB was supported by the International Foundation for Science (IFS # D-6074-2), and JC by the Masaryk University (MUNI/A/1489/2023).

Conceptualization: VG, TN. Data curation: TN, VG. Formal analysis: TN. Funding acquisition: VG, GB, JC. Investigation: TN, GB, JC, VG. Methodology: TN, VG. Project administration: VG. Resources: VG. Supervision: VG. Validation: VG. Visualization: TN, VG. Writing - original draft: VG, TN. Writing - review and editing: TN, GB, JC, VG.

Tadeáš Nečas https://orcid.org/0000-0002-5060-8394

Gabriel Badjedjea https://orcid.org/0000-0003-4284-4327

Janis Czurda https://orcid.org/0009-0000-9151-3482

Václav Gvoždík https://orcid.org/0000-0002-4398-4076

All of the data that support the findings of this study are available in the main text and public databases.