Five adult male specimens (GEP a168–170; CIB 121751; SYS a009518) of the genus Boulenophrys were collected on 11 August 2023 from Guangdong Yangchun Ehuangzhang Provincial Nature Reserve (21.851912°N, 111.421553°E; ca. 1,060 m a.s.l.) in Yangchun City, Guangdong Province, China. All examined specimens were fixed in 10% buffered formalin and later transferred to 70% ethanol. All studied specimens have been deposited at the Guangdong Polytechnic of Environmental Protection Engineering (GEP) facility in Foshan, Guangdong Province; the Museum of Biology, Sun Yat-sen University in Guangzhou, Guangdong Province, China (SYS); and the Herpetological Museum at the Chengdu Institute of Biology, Chinese Academy of Sciences (CIB) in Chengdu, Sichuan Province, China.

External measurements were recorded with a digital caliper (Neiko 01407A stainless steel 6-inch digital caliper) to the nearest 0.1 mm. These measurements are as follows: SVL (snout–vent length, from tip of snout to posterior margin of vent); HDL (head length, from tip of snout to the articulation of the jaw); HDW (head width, head width at the commissure of the jaws); SNT (snout length, from tip of snout to the anterior corner of the eye); IND (internasal distance, distance between nares); IOD (interorbital distance, minimum distance between upper eyelids); ED (eye diameter, from the anterior corner of the eye to the posterior corner of the eye); TD (tympanum diameter, horizontal diameter of tympanum); TED (tympanum–eye distance, from anterior edge of tympanum to posterior corner of the eye); HND (hand length, from the proximal border of the outer palmar tubercle to the tip of digit III); RAD (forearm or radioulna length, from the flexed elbow to the proximal border of the outer palmar tubercle); FTL (foot length, from distal end of shank to the tip of digit IV); TIB (crus or tibiofibula length, from the outer surface of the flexed knee to the heel). Sex was determined by external secondary sexual characters, such as the presence of vocal sacs, nuptial pads, or spines in males and their absence in females (Fei et al. 2009). Morphological characters of all recognized species of the genus Boulenophrys used for comparisons were based on an examination of museum specimens listed in Appendix I and information available in the literature (Table 1).

Five tissue samples were derived from the five adult specimens (GEP a168–170; CIB 121751; SYS a009518). Another six tissue samples (GEP asp061–066) were derived from the tail fins of different tadpoles from Guangdong Yangchun Ehuangzhang Provincial Nature Reserve (21.851912°N, 111.421553°E; ca. 1,060 m a.s.l.), and the tadpoles were not collected. In total, 94 samples were used for the phylogenetic analyses, encompassing 11 newly sequenced individuals and others downloaded from GenBank (https://www.ncbi.nlm.nih.gov/genbank/). Two samples of the genus Xenophrys Günther, 1864, were used as outgroups (Suppl. material 1). Two partial mitochondrial genes, the 16S ribosomal RNA (16S) and the cytochrome c oxidase 1 (COI), were used for phylogenetic analysis. DNA extraction, PCR amplification, and sequencing protocols followed Liu et al. (2018).

We used Clustal X 2.0 (Thompson et al. 1997) to align the sequences using default parameters. We then used PartitionFinder (Lanfear et al. 2012) to determine the optimal partitioning schemes, and we determined that partitioning by gene was optimal for 16S, while partitioning by codon position was optimal for COI, with GTR+I+G identified as the best-fit nucleotide substitution model for all partitions. We constructed the phylogenetic trees using maximum likelihood (ML) implemented in RAxMLGUI v.1.3 (Silvestro and Michalak 2012) and Bayesian inference (BI) using MrBayes v.3.2.4 (Ronquist et al. 2012). For the ML analysis, an optimal tree was obtained, and branch supports were evaluated with 1000 rapid bootstrapping replicates. For the BI analysis, two independent runs were conducted, each running for 10,000,000 generations and sampled every 1000 generations, with the first 25% of samples discarded as burn-in, resulting in a potential scale reduction factor (PSRF) of < 0.01. Then, the remaining trees were used to create a consensus tree. Nodes having ML bootstrap values (BS) ≥ 70 and BI posterior probabilities (BPP) ≥ 0.90 were considered well supported. Genetic distances among all Boulenophrys samples were calculated in MEGA 11 using the uncorrected p-distance model based on the 16S.

Among the focal species in this work, B. acuta has been assessed as Critically Endangered (ASG 2021), but B. yunkaiensis and B. gaolanensis have not been assessed. Here, the threat assessment of B. gaolanensis is relevant, as it is closely related to the taxa described in this manuscript. The geographic area of this study is characterized by low-elevation hills and mountains interrupted by low-elevation agricultural plains and large urban areas. As a result, the populations are disconnected both because of the landscape and human activities. To suggest an extinction risk assessment for these four species, we followed the IUCN Red List categories and criteria (www.iucnredlist.org/resources/redlistguidelines) due to their robustness, especially for species with restricted ranges (Maes et al. 2015), and because evaluations are conducted against quantitative thresholds for five criteria (IUCN 2012). However, we could not use criteria A (population size reduction), C (small and declining population size), and E (quantitative analysis of extinction risk) due to the absence of data on population sizes. Instead, we relied on criteria B (geographic range size) and D (very small population and/or restricted distribution) for the assessments. To determine the extent of occurrence (hereafter EOO) and the area of occupancy (hereafter AOO) for each species, we used the online platform GeoCAT (www.geocat.iucnredlist.org).

The BI and ML phylogenetic trees resulted in essentially identical topologies (ML phylogenetic topology in Fig. 2). In the phylogenetic result, the Boulenophrys samples from Ehuangzhang NR, Yangchun, formed three independent clades. Among them, samples (GEP asp062 and GEP asp064) from two uncaptured tadpoles clustered with the typical series of B. acuta from Heishiding NR, Fengkai, and that from Mt. Yunkai (Song et al. 2024), Xinyi, with good support (BS = 74; BPP = 0.95). In addition, samples (GEP asp061, GEP asp063, and GEP asp065–066) from four uncaptured tadpoles clustered with the typical series of B. yunkaiensis from Mt. Yunkai, Xinyi, with strong support (BS = 97; BPP = 0.99), and showed very shallow divergence. Thus, B. acuta and B. yunkaiensis were verified to be distributed in Ehuangzhang Nature Reserve based on molecular data, which also confirms the expansion of the distribution range of the two species. As the distribution range of the critically endangered B. acuta has been exceptionally underestimated, we suggest an update for the extinction risk assessment below.

Additionally, samples taken from five captured adult specimens (GEP a168–170, SYS a009518, and CIB 121751) formed a monophyletic lineage with strong support (BS = 100; BPP = 1.00). This new lineage is a sister taxon to B. acuta with strong support (BS = 97; BPP = 1.00) and significant divergence with a mean p-distance of 6.5% (Suppl. material 2). Further morphological examination also showed that the corresponding specimens of the new lineage can be distinguished from all recognized congeners. Thus, we describe the corresponding specimens as a new species, which is sympatric with B. acuta and B. yunkaiensis.

Recommendation for extinction risk assessment for Boulenophrys acuta

Distribution: Heishiding Nature Reserve, Fengkai County, Zhaoqing, Guangdong Province, China (Li et al. 2014; Lyu et al. 2023); Mt. Yunkai, Xinyi City, Guangdong Province, China (Song et al. 2024); Guangdong Yangchun Ehuangzhang Provincial Nature Reserve, Yangchun City, Guangdong Province, China (this study).

Extinction risk assessment: based on the analysis on GeoCAT, we determined an AOO of 3,170 km² and an AOO of 12 km². As the species is found at three independent locations (to assess against criteria B2a), and there is a continued observed decline in the (iii) quality of the habitat, we recommend the species be listed as Endangered following the criteria B1ab(iii) and B2ab(iii). If relying on the criteria D (very small population and/or restricted distribution) for the recommendation of assessment, the species would be listed as Vulnerable D2 because the AOO is smaller than 20 km² and the species occurs at fewer than five locations. This recommendation for assessment is a non-genuine downlisting (ASG 2021), from Critically Endangered (CR) to Endangered (EN), because of the discovery of two new populations.

Recommendation for extinction risk assessment for Boulenophrys gaolanensis

Distribution: Gaolan Island, Jinwan District, Zhuhai City, Guangdong Province, China (Song et al. 2024; Jian Wang, pers. comm.).

Extinction risk assessment: based on the analysis on GeoCAT, we determined an AOO of 0 km² and an AOO of 8 km². As the species is found at two independent locations (to assess against criteria B2a), and there is a continued observed decline in the (iii) quality of the habitat, we recommend the species be listed as Critically Endangered following the criteria B1ab(iii) and B2ab(iii). If relying on the criteria D (very small population and/or restricted distribution) for the recommendation of assessment, the species would be listed as Vulnerable D2 because the AOO is smaller than 20 km² and the species occurs at fewer than five locations. This recommendation for assessment as Critically Endangered (CR) has the potential to be updated through a non-genuine downlisting, as there are still a large number of potential distribution areas in the adjacent coastal mountainous regions.

Recommendation for extinction risk assessment for Boulenophrys yunkaiensis

Distribution: Mt. Yunkai, Xinyi City, Guangdong Province, China (Qi et al. 2021); Guangdong Yangchun Ehuangzhang Provincial Nature Reserve, Yangchun City, Guangdong Province, China (this study).

Extinction risk assessment: based on the analysis on GeoCAT, we determined an AOO of 29.8 km² and an AOO of 8 km². As the species is found at two independent locations (to assess against criteria B2a), and there is a continued observed decline in the (iii) quality of the habitat, we recommend the species be listed as Critically Endangered following the criteria B2ab(iii). If relying on the criteria D (very small population and/or restricted distribution) for the recommendation of assessment, the species would be listed as Vulnerable D2 because the AOO is smaller than 20 km² and the species occurs at fewer than five locations. This recommendation for assessment as Critically Endangered (CR) has the potential to be updated through a non-genuine downlisting, as the species is so far known to occur on two disconnected mountain ranges adjacent to likely suitable habitat.

Recommendation for extinction risk assessment for Boulenophrys yangchunensis sp. nov.

Distribution: Guangdong Yangchun Ehuangzhang Provincial Nature Reserve, Yangchun City, Guangdong Province, China (this study).

Extinction risk assessment: based on the analysis on GeoCAT, we determined an AOO of 0 km² and an AOO of 4 km². As the species is found at a single location (to assess against criteria B2a), and there is a continued observed decline in the (iii) quality of the habitat, we recommend the species be listed as Critically Endangered following the criteria B1ab(iii) and B2ab(iii). If relying on the criteria D (very small population and restricted distribution) for the recommendation of assessment, the species would be listed as Vulnerable D2 because the AOO is smaller than 20 km² and the species occurs at fewer than five locations. This recommendation for assessment as Critically Endangered (CR) has the potential to be updated through a non-genuine downlisting if the species was present in the continuous habitat likely to be suitable to support a larger population. However, the continuous and likely suitable area is less than 400 km², placing the species in the Endangered (EN) category under the current threat conditions even if it was present throughout the whole area, which surveys have already demonstrated not to be accurate.