Mitochondrial lineages affiliated with B. variegata were identified for 1238 individuals sampled in 355 localities based on sequences of the cytochrome b (cyt b) gene, as gathered from previous studies (Szymura et al. 2000; Canestrelli et al. 2006; Hofman et al. 2007; Pabijan et al. 2008, 2013; Zheng et al. 2009; Fijarczyk et al. 2011; Cornetti et al. 2016; Talarico et al. 2020; Dufresnes et al. 2021a; Pröhl et al. 2021; Tzoras et al. 2023), and complemented by additional sequences for 17 individuals from 17 localities (methods in Dufresnes et al. 2021a). The new sequences were uploaded on GenBank under accessions PQ757961–PQ757977.

Sequences were manually aligned and trimmed (1096 bp) in Seaview 5 (Gouy et al. 2021), and unique haplotypes were identified with DnaSP 6 (Rozas et al. 2017). To visualize their relative divergence, a phylogenetic network was produced in SplitsTree 4 (Huson and Bryant 2006), based on uncorrected p distances. Haplotypes were assigned to the six lineages defined in the study of Fijarczyk et al. (2011) and lineage distributions were mapped in QGIS 3.24 (QGIS.org 2022). The list of localities is provided in Table S1.

For the purpose of a taxonomic description, which requires stating diagnostic characters, we further screened the cyt b haplotype alignment for substitutions unique to the Rhodope lineage using MOLD (Fedosov et al. 2022), as implemented in iTaxoTools (Vences et al. 2021); haplotypes with missing data were discarded for this analysis. The full gene sequence of B. variegata was used as reference (GenBank accession JX893177).

We gathered 19 full mitogenomes representative of the mitochondrial diversity of B. variegata (n = 8) and its closest relatives B. bombina (n = 9) and B. orientalis (n = 2) sequenced in previous work (San Mauro et al. 2004; Jung et al. 2006; Pabijan et al. 2008, 2013; De Cahsan et al. 2021; sequence list in Table S2). Sequences were handled in Seaview for manual alignment, trimming (17,240 bp), and gene/tRNA partitioning using the annotations of JX893177 as a reference.

The dataset was used to infer time-calibrated Bayesian phylogenies with BEAST 2.6 (Bouckaert et al. 2019), using the birth-death tree model with 1/X priors for the birth and death rates, GTR+G+I models of sequence evolution for each partition, and a strict clock calibrated by two most recent common ancestor (MRCA) priors, one on the root (prior 1) and one on the B. variegata/bombina divergence (prior 2). For those, we implemented the age estimates obtained by Pabijan et al. (2013) based on two fossil calibration schemes, which the authors considered equally plausible. The first calibration (calibration I) involved 21.3 Mya (95% Height Posterior Density [HPD]: 24.4–20.1 Mya) for prior 1 and 9.0 Mya (12.7–4.9 Mya) for prior 2, which was implemented with a log-normal prior of M = 0.35, S = 0.73 and offset = 19.7, and a normal prior of mean = 9.0 and σ = 2.3, respectively. The second calibration (calibration II) involved 14.1 Mya (13.1–17.4 Mya) for prior 1 and 6.5 Mya (8.9–4.2 Mya) for prior 2, which was implemented by a log-normal prior of M = 0.35, S = 0.73 and offset = 12.7, and a normal prior of mean = 6.5 and σ = 1.4, respectively. Short preliminary runs of 10 million iterations were conducted in order to optimize operator parameters as provided by the BEAST output. Final optimized runs were conducted for 100 million iterations with sampling every 50,000 and were monitored for stationarity and effective sample sizes of parameters in Tracer 1.7 (Rambaud et al. 2018). Consensus trees were produced with the TreeAnnotator module of BEAST, discarding the first 20% as burn-in.

In parallel, we also reconstructed a maximum-likelihood phylogeny in IQ-TREE 1.6 (Trifinopoulos et al. 2016), using the model finder option to find the best model of sequence evolution, and 1000 ultrafast bootstrap replicates.

Moreover, pairwise divergence between lineages was computed in MEGA 11 (Tamura et al. 2021) for the DNA barcoding genes cyt b, 16S ribosomal DNA (16S), and cytochrome c oxidase subunit I (cox1). These loci are frequently examined for amphibian species delimitation, notably to apply ad hoc speciation thresholds based on patterns of reproductive isolation (Dufresnes et al. 2021b; Dufresnes and Litvinchuk 2022).

We built a supermatrix spanning four polymorphic fragments of nuclear genes previously sequenced in Bombina, namely sodium-calcium exchanger gene intron 1 (ncx1, 702 bp), recombination activating gene intron 1 (rag-1, 1061 bp), intron 2 (rag-2, 578 bp) and rhodopsin gene (rho, 1374 bp), for a subset of nine samples representative of the mitochondrial diversity of B. variegata (n = 6), as well as B. bombina (n = 1) and B. orientalis (n = 2) as outgroups (sequence list in Table S3). The samples considered are composite, as they combine sequences from two independent studies (rag-2 and rho: Zheng et al. 2009; ncx1 and rag-1: Fijarczyk et al. 2011), noting that correspondence was reliably established based on cyt b barcoding and geographic origins. In total, the dataset consists of 15 partitioned sequences (6 out of the 9 samples were heterozygous) for a total length of 3715 bp.

The supermatrix was analyzed with the multigene coalescent-based species tree method implemented in *BEAST (Heled and Drummond 2010), using independent site models (GTR+G+I), clock models (strict clocks) and trees for each gene, and a birth-death model for the species tree (with 1/X priors for the birth and death rates). Sequences were set in eight groups that distinguished the main mitochondrial lineages of B. variegata and the two outgroup species. As above, the analysis was run for 100 million iterations, sampling every 50,000, and monitored in Tracer. To see alternative topologies among sampled trees, the nuclear phylogeny was displayed as a cloudogram produced in DensiTree 2 (Bouckaert and Heled 2014).

For exploratory purposes, the concatenated alignment was also analyzed with IQ-TREE as for the mitogenome dataset above, but noting that unlike mitogenomes (where all genes are fully linked and share the same genealogy), such analysis can provide spurious topologies when the combined sequences are unlinked and form different gene trees (Maddison 1997).

To infer the nuclear diversity of B. variegata, we further analyzed double-digest Restriction Associated DNA sequencing (ddRAD-seq) data obtained for 10 samples representative of the mitochondrial lineages of that species, and one B. bombina sample to be used as outgroup (Table S4). These samples were included in a genomic library prepared with a custom protocol (http://dx.doi.org/10.17504/protocols.io.kxygx3nzwg8j/v1), as fully described in Dufresnes et al. (2025). In brief, this protocol involves enzyme restriction with MseI and SbfI, ligation of barcoded adapters, purification, PCR amplification with Illumina primers, size selection of fragments between 400–500bp, purification and concentration for paired-end sequencing on an Illumina NextSeq550 with the 2×75 bp kit. Raw reads were demultiplexed with STACKS 2.63 (Catchen et al. 2013) and uploaded on the NCBI Sequence Read Archive under BioProject PRJNA949685 (accessions SRR31792557–SRR31792567). Assembly and cataloging was performed with the denovo_map.pl pipeline with the -rm-pcr-duplicates flag and default -m, -n, and -M parameters. The catalog contained 230,498 loci with an average coverage of 15.3 per sample (9.9–19.6). Among these we called the loci polymorphic and sequenced among all samples with the module popu­lations of STACKS and retaining invariable sites (-p 11 -phylip-var-all), which produced a supermatrix alignment of 4759 loci totaling 553,608 bp. From this alignment, a maximum-likelihood tree was built with IQ-TREE as above.

Our analyses confirmed the occurrence of an evolutionarily unique population in the Rhodope mountains, previously confounded with its sister lineage B. v. scabra (see Results). To document morphological variation, we measured 58 adult specimens curated at the Museum Koenig Bonn (ZFMK), the Museum für Naturkunde Berlin (ZMB), and the Muséum d’histoire naturelle de Genève (MHNG), representing both sexes and of known geographic origin – so they could be assigned to each lineage. Twelve characters were measured with a digital caliper (0.1 mm precision) by one of us (CD) to avoid measurer bias: L. body length, from tip of snout to center of cloacal opening; F. femur length, from center of cloacal opening to distal end of the femur bone; T. tibia length, from knee to heel; D.p. length of the first toe, from the distal point of the base of the internal metatarsal tubercle to the end of the toe; L.c. head length, from the tip of the snout until the posterior edge of the mandible; Lt.c. head width, between the posterior edges of jaw articulations; Sp.oc. snout width, between the front edges of orbits; D.r.o. snout length, from the front edge of snout to the interior bony orbital border; Lt.p. maximum width of the upper eyelid; Sp.p. inter-orbital distance, as the minimum distance between orbits; L.o. maximum length of the upper eyelid; Sp.n. inter-nostril distance. Measured specimens are listed in Table S5. A schematic showing how each character was measured is available as Fig. S1.

To examine body shape without the effect of body size, we applied allometric corrections with the R package GroupStruct (Chan and Grismer 2022) using the multispecies method to perform lineage- and sex-specific adjustments by body length (L.). We then performed a Principal Component Analysis (PCA) on the corrected dataset using the R package FactoMineR (Lê et al. 2008) and built convex hulls to distinguish sexes and lineages (chull function of the R package grDevices; R Core Team 2024). The effect of lineage and sex (and their interaction) was assessed by a Multivariate Analysis of Variance (MANOVA) with the R package stats (R Core Team 2024).

To find characters diagnosable in the field, we further computed simple body ratios (characters/L.) and visualized differences between lineages using boxplots (R package ggplot2; Wickham 2016), testing for significance using Kruskal-Wallis rank sum tests (R package stats), and applying a Bonferroni correction for multiple testing.

Finally, we explored differences in body length (L.) by considering body length measurements of an additional 120 B. v. scabra specimens of known sex and origin, namely the 4 specimens of the type series of the new Rhodope taxon (see Results), 96 live specimens captured in Bulgaria during monitoring surveys, and 20 specimens curated at the Zoological Institute of Russian Academy of Sciences, St. Petersburg (ZISP). The list is provided in Table S5. While these were measured by different persons, measurer bias is expected to be negligible for this standard character. The effects of sex and lineage were statistically tested on this extended dataset (n = 178) by a two-way Analysis of Variance (ANOVA), which statistical significance was assessed in a non-parametric way, namely by a permutation procedure in which the assignment of lineage and sex was reshuffled 1000 times. The variation was visualized using boxplots (R package ggplot2).

We quantified variation in the proportions of yellow/orange vs. dark coloration on the ventral side of 197 specimens of B. v. scabra of known geographic origin and attributable to lineages. To this end, we gathered photographs where the throat and belly are visible for (1) field-caught specimens, taken from our own libraries and the citizen-science platform iNaturalist (2024), and (2) the museum specimens measured above. On each photo, the forelimbs, hindlimbs, and any foreign object were manually cropped out in Adobe Photoshop 2019 to retain only the head, thorax and abdomen.

Each photo was processed using a modified version of a Python script published by Sanchez et al. (2018), which was originally developed to quantify black-to-yellow color proportions in fire salamanders (Salamandra salamandra) and similarly colored amphibians. Here, three key adjustments were made to the script in respect to our target species: (1) the color range was expanded to include shades of orange and red, in addition to yellow; (2) the threshold for black pixels was adjusted and the spectrum was broadened to accurately capture even the faded dark patterns in preserved specimens; and (3) the “-background” argument was removed since the photos were cropped, and any pixels not classified as black or color were counted as “other”. The modified script outputs the absolute number of black, color and “other” pixels, the total number of pixels, the proportion of black and color pixels relative to the total (excluding “other” pixels), and the color-to-black ratio. It is made available on Zenodo (https://doi.org/10.5281/zenodo.14359153).

Differences in color-to-black ratios between the two lineages were analyzed with a Kruskal-Wallis test. Additionally, to account for the potential effect of museum preservation on coloration contrast, a two-way ANOVA was conducted (significance tested by a permutation procedure as above), treating specimen type (field-caught vs. museum) as a secondary explanatory variable.

We examined the nomenclature with the goal of attributing available names to phylogeographic lineages. To this end, we first mapped these lineages according to our results by updating the distribution shapefile of B. variegata designed by Dufresnes et al. (2021a) based on a synthesis of available atlases and occurrence datasets. We then researched the scientific names that could be associated with these lineages using the lists provided by Lescure et al. (2011) and Frost (2024) as starting references. We evaluated (1) the type material and information associated with each taxon; (2) to which populations they correspond to; and (3) whether they represent senior or junior synonyms in respect to the principle of priority of the ICZN.

The nomenclature of the yellow-bellied toad B. variegata is intertwined with the nomenclature of the fire-bellied toad B. bombina. Both were considered the same species for more than a century of Linnean classification (Lescure et al. 2011) and consequently, the affiliation of some historical names prior to their distinction remains questionable. Therefore, our nomenclature overview encompassed taxa proposed for both or either B. variegata and B. bombina. When necessary, we took actions to affirm the affiliation of taxa of uncertain origin to promote nomenclatural stability in present and future revisions.

Analyses of the cyt b gene (1096 bp) in 1238 toads from 355 localities revealed 202 B. variegata haplotypes grouped into the four main mitochondrial lineages retrieved by previous studies (Fig. 1):

Figure 1. 

Relative divergence (network) and geographic distributions (map) of the mitochondrial lineages of B. variegata identified among 1096 bp sequences of the cyt b gene (n = 1238 individuals from 355 localities), defined as in Fijarczyk et al. (2011). Arrows indicate introduced populations.

(1) the Carpathian lineage (C) is found in the B. v. variegata populations of the Carpathian Mountains and can be divided into two closely related mitogroups inhabiting the northern and southwestern ranges (C-W) and southeastern ranges (C-E);

(2) the Apennine lineage (A) extends across all the populations of the Apennine B. v. pachypus;

(3) the Rhodope lineage (R) is restricted to the easternmost B. v. scabra populations of the Balkan Peninsula;

(4) the Balkano-Western lineage (BW) ranges from the southern to the northwestern edges of the distribution, and can be divided in two closely related mitogroups, one found in the southern populations of the Balkan Peninsula, attributed to B. v. scabra (BW-S), and one found across the populations stretching from the Dinarides to Western Europe, attributed to B. v. variegata (BW-N). The BW-S mitogroup is also found in Central Italy and the Aegean islands due to human-mediated introductions (Talarico et al. 2020; Tzoras et al. 2023).

The mitochondrial trees obtained with BEAST and IQ-TREE, based on full mitogenomes (17,240 bp), provided robustly supported topologies that recovered the four B. variegata lineages in two major clades, one including lineage C, and one regrouping lineages A, R, and BW (Figs 2, S2). Calibration I expectedly featured slightly older node ages than calibration II: the split between lineage C and A/R/BW was estimated to 7.1 Mya vs. 4.7 Mya; the split between lineages A and R/BW was estimated to 3.4 Mya vs. 2.3 Mya; the split between lineages R and BW was estimated to 2.1 vs. 1.4 Mya; the diversification of lineage C was estimated to 0.6 vs. 0.4 Mya, and the diversification of lineage BW was estimated to 0.5 vs. 0.4 Mya.

Figure 2. 

Bayesian phylogeny obtained with BEAST based on full mitogenomes (17,240 kb). Nodes are annotated with dating estimates and their 95% HPD (node bars) obtained from two calibrations (dark grey: calibration I; light grey: calibration II). Sequence labels indicate accession numbers, geographic origins, taxa and mitochondrial lineages. Asterisks denote full branch support (posterior probabilities of 1.0).

Pairwise sequence divergences at barcoding genes (Table 1) were on average 3.6% at 16S, 6.7% at cox1, and 10.4% at cyt b between B. variegata and B. bombina. The highest intraspecific divergence in B. variegata involved the Carpathian vs. other lineages, with on average 2.8% at 16S, 5.0% at cox1, and 8.7% at cyt b. The Apennine B. v. pachypus featured an average divergence of 1.8% at 16S, 2.5% at cox1, and 4.6% at cyt b from its most closely related lineages (BW and R). The two lineages affiliated to B. v. scabra (BW-S and R) featured divergences of 1.3% at 16S, 1.6% at cox1, and 2.4% at cyt b.

The nuclear species tree obtained with *BEAST based on the four gene fragments (3715 bp in total) recovered B. variegata as monophyletic and as the sister species of B. bombina (Fig. 3). Within B. variegata, this phylogeny differs from the mitochondrial phylogeny in two aspects. The first discordance is the position of Balkano-Western B. v. variegata, which branches with Carpathian B. v. variegata in the nuclear tree (vs. with Balkano-Western B. v. scabra as in the mitochondrial tree). This renders both B. v. variegata and B. v. scabra monophyletic: B. v. variegata populations carrying BW-N, C-W and C-E mtDNA form one nuclear clade, while B. v. scabra populations carrying BW-S and R mtDNA form a distinct nuclear clade. The second discordance is the position of B. v. pachypus, which is placed as the sister clade of B. v. variegata in the nuclear tree (vs. as the sister clade of B. v. scabra/Balkano-Western B. v. variegata as in the mitochondrial tree). Moreover, while the B. v. variegata samples feature shallow branches, the nuclear divergence within B. v. scabra rather compares to the divergence of B. v. pachypus.

Figure 3. 

Cloudogram of the nuclear species trees obtained with *BEAST based on four gene fragments (3715 bp) and distinguishing samples representative of the different B. variegata mitochondrial lineages. Node labels indicate Bayesian posterior probabilities; sequence labels indicate taxa and mitochondrial lineages; the thick and thin lines show the average tree root and all the sampled trees, respectively.

Support of the gene fragment tree is low to moderate, with posterior probabilities of nodes ranging from 0.62 to 0.92 within the B. variegata clade, a typical issue given the much lower level of polymorphism at nuclear than mitochondrial sequences.

The IQ-TREE analysis confirms the homogeneity among the B. v. variegata sequences, as well as the distinctiveness between the two lineages of B. v. scabra and compared to other taxa (Fig. S3). The topology should however be interpreted with caution, both due to the low polymorphism of the dataset and because such analysis inappropriately assumes a single genealogy shared among the nuclear genes. Accordingly, the IQ-TREE topology appears uninformative, as seen from the nested placement of B. bombina within B. variegata (Fig. S3).

The ddRAD-seq data (4759 loci totaling 553,608 bp) provided a robust phylogenomic tree that distinguished the different phylogeographic lineages of B. variegata (Fig. 4). In accordance with the mitochondrial phylogeny (Fig. 2), B. v. pachypus is more closely related to B. v. scabra than B. v. variegata in the phylogenomic tree, and B. v. scabra forms two distinct lineages (Fig. 4). As in the gene-based nuclear tree, the analysis also retrieves B. v. variegata as monophyletic i.e., samples from the Carpathian ranges, the northern Balkans and Western Europe are grouped together despite their distinctive mitochondrial ancestry (BW-N and C; Fig. 4).

Figure 4. 

Maximum-likelihood phylogeny obtained with IQ-TREE of 4759 concatenated RAD loci (553,608 bp). Colors distinguish the different B. variegata mitochondrial lineages. Sequence labels indicate sample names, geographic origins, taxa and mitochondrial lineages. Asterisks denote robust branch support (bootstrap >95). The tree is rooted by B. bombina.

Based on 11 morphological characters corrected by body length (L.), the two lineages of B. v. scabra differ in body shape (Fig. 5; variable loadings in Table S6). In the PCA, specimens attributed to the Balkano-Western vs. the Rhodope lineage show almost no overlap, which was particularly marked for females. Accordingly, lineage (F = 25.9, P < 0.001), sex (F = 2.2, P = 0.03) and their interaction (F = 3.1, P = 0.004) were all significant in the MANOVA.

Figure 5. 

Morphological comparison of specimens attributed to the two lineages of B. v. scabra. A PCA on 11 characters corrected by body length distinguishing the males (triangles) and females (circles) of each lineage; B boxplots showing variation in body length (L.) among sexes and lineages. For the list of characters, see Methods (section Morphometric analyses in Bombina variegata scabra).

Males and females from the Rhodope lineage were larger and smaller than their Balkano-Western counterparts, respectively, which resulted in a significant interaction between sex and lineage (F = 7.2, P = 0.007), but not for sex (F = 1.0, P = 0.29) and lineage (F = 0.15, P = 0.691) in the two-way ANOVA (significance tested by a permutation approach). Average raw measurements and their standard deviation are detailed per sex and lineage in Table 2.

Comparisons of body ratios (character/L.) between the two B. v. scabra lineages suggest overlapping distributions for all characters (males and females combined), but four of them show significant differences (Kruskal-Wallis tests, P < 0.05) (Fig. 6). Specifically, the femur (F.), tibia (T.) and first toe (D.p.) are proportionally shorter, and the eyelid is proportionally longer (L.o.) in the Rhodope lineage. The eyelid also tends to be larger in the latter (Lt.p.), but this was marginally not significant (P = 0.06). After Bonferroni correction, only the difference for the femur (F.) remains significant (P < 0.0045).

Figure 6. 

Body ratios (character/body length L.) between the Rhodope (purple) and Balkano-Western (light green) lineages of B. v. scabra (males and females combined). Brackets indicate significant differences (Kruskal-Wallis test) at P < 0.05 (dash) and at the adjusted P < 0.0045 for multiple testing (plain). For character abbreviations, see Methods (section Morphometric analyses in Bombina variegata scabra).

Color/dark ratios range from 0.26 (almost entirely dark) to 10.3 (almost entirely colored) (Fig. 7). The Rhodope lineage is overall more colored (2.9 ± 2.0) than the Balkano-Western lineage of B. v. scabra (2.1 ± 1.4) and the difference is significant (Kruskal-Wallis test, χ² = 7.3, P = 0.007). Whether specimens were field-caught or museum-preserved does not affect this pattern; in the two-way ANOVA, the lineage difference is significant (F = 8.6, P = 0.003) but the specimen type is not (F = 0.20, P = 0.67).

Figure 7. 

Proportion of color/dark coloration on the ventral side of specimens previously attributed to B. v. scabra. Bracket indicates significant difference (Kruskal-Wallis test) at P < 0.05. The body area analyzed is emphasized in the two colored (top) and dark (bottom) examples.

Our analyses suggest that the species B. variegata features four genuine evolutionary lineages eligible for a taxonomic classification. First, the nuclear similarity among the two divergent mitochondrial lineages attributed to nominal subspecies B. v. variegata (mtDNA C and BW-N) implies the existence of a single taxon among these populations, which extends from the Carpathians to Western Europe; the mitochondrial discordance is discussed below (see Discussion). Second, we confirm the independent evolution of the Apennine populations known as B. v. pachypus (mtDNA A). Third and fourth, we documented the mitochondrial, nuclear and morphological differentiation of two lineages among the populations attributed to B. v. scabra, one in the Western and Southern Balkans (mtDNA BW-S) and one in the Rhodope (mtDNA R). As argued below (see Discussion), we preliminarily consider these Northern, Apennine, Balkan and Rhodope lineages as four distinct subspecies of the yellow-bellied toad B. variegata. Their respective distributions are provided in Fig. 8, with the list of available names that apply to each of them according to the following nomenclature overview. Important illustrations are depicted in Fig. 9, along with a timeline summarizing the nomenclatural history of European Bombina.

Figure 8. 

Geographic distributions of the four subspecies of B. variegata and location of the type localities of available names for each (stars). Red: Northern subspecies; yellow: Apennine subspecies; light green: Balkan subspecies; purple: Rhodope subspecies. The dashed and question-marked areas correspond to potential contact zones/introgression and regions of unclear subspecies assignment, respectively. Senior synonyms are shown by big stars and are labelled with capital letters; junior synonyms are shown by small stars and are labelled with numbers.

Figure 9. 

Illustrations of the ventral side of European Bombina toads in nomenclaturally important specimens (top) and summary of the nomenclatural history (bottom). A Bottom right specimen in plate XXVII of Rösel (1758, pl. XXII) considered by Linnaeus (1766) to represent both his Rana variegata and Rana bombina, and (partly) used by Laurenti (1768) to describe his Bufo igneus. The color pattern unambiguously corresponds to the yellow-bellied toad (now B. variegata), further noting that Rösel depicted the fauna of Nuremberg (Germany) where only this species is present. B Illustration of Bufo igneus in Bonnaterre (1789, pl. 6–fig. 6), which is clearly a reproduction of Rösel’s plate. C Illustration of La Cepède’s invalid Rana sonans (La Cepède 1788, pl. XIII) and (D) reproduction by Bonnaterre (1789, pl. 2–fig. 3) for Rana sonans Bonnaterre, 1789; the depicted specimen is designated as the lectotype of this taxon. E Illustration of Bufo pluvialis Daudin, 1803 (pl. XXVI); the depicted specimen is designated as the lectotype of this taxon; F Bombinator igneus (Laurenti, 1768) and (G) Bombinator pachypus Bonaparte, 1838 (labeled “Bombinator pachidactylus” on the plate), as distinguished by Bonaparte (1838). H The fire-bellied toad and (I) the yellow-bellied toad, as distinguished by Boulenger (1886, pl. L). J NHMW 6698/1, the neotype of Bufo pluvialis Daudin, 1803 (credit: DJ). K MZL 11988, the neotype of Rana variegata Linnaeus, 1758 (credit: CD). The timeline is annotated with years of descriptions (diagonal labels), years of important revisions (vertical labels), and commonly used names (inside labels) for B. bombina (top) or B. variegata (bottom). The left illustration is taken from what appears to be the earliest natural history account for a fire/yellow-bellied toad (Gessner 1563).

Rana variegata Linnaeus, 1758 – First published in Systema Naturae 10^th edition without mention of a specimen and a locality other than “Exteris regionibus [foreign countries]” (Linnaeus 1758: 211). There is no trace of specimens labelled Rana variegata in Linnaeus’s inventoried collections (Holm 1957). In his 12^th edition of Systema Naturae, Linnaeus (1766) had already replaced this name by Rana bombina Linnaeus, 1761 (now Bombina bombina), assuming that both represent the same species: the one depicted by Rösel von Rosenhof (1758: pl. XXII, XXIII), namely a yellow-bellied toad from Nuremberg, Germany (Fig. 9A). The nomen variegata remained in synonymy for more than a century and a half. In the late 18^th and early 19^th centuries, most naturalists continued to assume that there was only one species of yellow/fire-bellied toad in Europe, and favored other names for it, notably Rana bombina and Bufo igneus Laurenti, 1768 (e.g., La Cepède 1788; Latreille 1800; Daudin 1803; Fitzinger 1843; Duméril and Bibron 1841). Due to its unclear origin, variegata was unused or overlooked even after the yellow- and fire-bellied toads were distinguished (Boulenger 1886). Accordingly, its synonymy was frequently accompanied by question marks (e.g., Bedriaga 1890: 581; Boulenger 1897: 142), but it does appear in priority in a few works (e.g., Bombinator variegatus in Bedriaga 1881: 291). It was eventually resurrected by Mertens and Müller (1928), who affiliated it to the yellow-bellied toad by restricting its type locality to “Schweiz [= Switzerland]”, yet without any justification or taxonomic act. According to Gollmann et al. (2012), this association may be linked to the account of the “Wasser Krott [water toad]” by the Swiss naturalist Gessner in his Thierbuch, which allegedly corresponds to the yellow-bellied toad based on the provided drawing (Gessner 1563: 170; perhaps the earliest naturalist illustration of the species, Fig. 9) and mentions “Demnach so ist auch ein geschlecht so in den wassere wohnet klein/mögen Wasserkrötlein genennet werden/in unseren landen Zügger von ihrer stimm her/so ihnen angeboren ist [Accordingly, there is also a species that lives in the water, small, and could be called water toads, in our Zürich lands their sound is specific to them]” (Gessner 1563: 169). Gessner’s Thierbuch is an abridgement of his Historia animalium, the first and one of the most exhaustive pre-Linnean natural history work. However, unlike for some of his other taxa, Linnaeus did not refer to Gessner’s opus in his description of Rana variegata, hence any connection of this taxon with Switzerland, and by extension, with the yellow-bellied toad, is baseless. In addition, according to the ICZN, so-called restrictions of type locality (e.g., “terra typica restricta”) without providing any evidence to support this claim are nomenclaturally invalid for not being associated with a lectotype or neotype designation (Dubois 2011). There is no potential lectotype of Rana variegata from the information provided by Linnaeus (1758). A neotype is therefore necessary to unambiguously affiliate Rana variegata to the yellow-bellied toad and restrict its type locality to Switzerland, as it has been assumed for nearly a century. We here designate MZL 11988 as the neotype of Rana variegata Linnaeus 1768 (Fig. 9K). MZL 11988 is a male yellow-bellied toad curated at the Muséum Cantonal de Zoology of Lausanne (MZL); it was collected in 1966 in Promenthoux, Prangins, Vaud, Switzerland, which becomes the type locality of Rana variegata. Switzerland is only inhabited by the yellow-bellied toad, namely the Northern subspecies (mtDNA BW-N) in our figures. Rana variegata Linnaeus, 1758 thus applies to this lineage and is the oldest available name for it, Bombina variegata variegata (N°A in Fig. 8).

Rana bombina Linnaeus, 1761 – Described from “Scaniae compestris fossis australibus [= southern Scania/Skåne]” in Sweden (Linnaeus 1761: 101), based on specimens curated in Uppsala (Boulenger 1888), at least two of which were featured in the Alströmer-Linné collection inventoried by Thunberg (Holm 1957: 43). For a century, it was widely used to designate all yellow-/fire-bellied toads (believed to be a single species), initially as Rana bombina (Linnaeus 1761) and Bufo bombinus (Latreille 1800). It was subsequently classified in a separate genus, as Bombina bombina (Oken 1816), Bombinator bombinus (Merrem 1820), or Bombitator bombinus (Wagler 1830; an emendation of Merrem’s Bombinator). The genus Bombinator was adopted for the rest of the 19^th century. When distinguishing the yellow and fire-bellied toads, Boulenger briefly considered Bombinator bombinus as the yellow-bellied toad (Boulenger 1886), then as the fire-bellied toad under the synonymy of Bufo igneus Laurenti, 1768 (Boulenger 1888). Applying the law of priority, Stejneger (1905, 1907) affirmed both the genus Bombina and the name Bombina bombina to the fire-bellied toad, an arrangement that persists until today. Given that only the fire-bellied toad inhabits Sweden, Rana bombina Linnaeus, 1761 accordingly applies to this species and is the oldest available name for it, as Bombina bombina.

Bufo igneus Laurenti, 1768 – Given as “Rœsel. Tab XXIII. XXII.” and “Habitat in paludibus nostris Danubialibus ingenti in copia; autumno omnis via, qua itur in Nusdorff, iisdem scatet” [= lives in great abundance in our Danubian marshes; in the autumn every road that leads to Nusdorff is covered with it]” (Laurenti 1768: 29). Laurenti’s detailed description (Laurenti 1768: 129–133) corresponds to the fire-bellied toad and accordingly, this species used to occur along the Danube from Vienna to Nussdorf, now a suburb of Vienna (Schweiger et al. 2021). He further documented four experiments aimed at evaluating the species’ toxicity, implying that he had specimens at his disposal (Laurenti 1768: 131–133). At the same time, he specifically referred to Rösel’s plates and gave his species Rösel’s adjective igneus [fiery], in reference to the belly coloration; but Rösel’s drawings unambiguously represent the yellow-bellied toad (Fig. 9A). This confusion is understandable, as again, yellow- and fire-bellied toads were not distinguished in Laurenti’s time. In the 19^th century, igneus was increasingly preferred to Linnaeus’s bombina (e.g., Wagler 1830; Fitzinger 1843), notably as Bombinator igneus, the name kept by Boulenger (1886, 1888) to differentiate the fire-bellied toad from the yellow-bellied toad. It was eventually replaced by Rana bombina Linnaeus, 1761 by application of the priority rule (Stejneger 1907). While Bufo igneus was unanimously attributed to the fire-bellied toad, its type series is heterogeneous: it includes both yellow-bellied toads B. variegata (the Nuremberg specimens depicted by Rösel) and fire-bellied toads B. bombina (the Danubian specimens examined by Laurenti). To affirm this taxon solely to the fire-bellied toad, we here designate as the lectotype of Bufo igneus Laurenti, 1768 the specimen Laurenti kept in captivity after his experiment XII, narrated as “unamque ex iis fanam nuncin longam hyemem adhuc confervo [= one of them, in fact, I still keep healthy even now through the long winter]” (Laurenti 1768: 312). If they were ever curated, Laurenti’s specimens are not present at the Naturhistorisches Museum Wien (NHMW; G. Gassner pers. comm.). The lectotype is probably lost but its origin (the Danubian marsches from Vienna to Nussdorf) is unambiguous. Bufo igneus Laurenti, 1768 thus belongs to Bombina bombina.

Bufo salsus Schrank in Schrank & Moll, 1785 – Documented from “Berchtesgaden” in the Bavarian Alps, with no mention of specimens (Schrank and Moll 1785: 308–309). The synonymy with Rana bombina Linnaeus, 1761 is questioned in the description, and was enforced by Latreille (1800). After the distinction of yellow and fire-bellied toads, early 20^th century authors such as Stejneger (1905, 1907) and Nikolsky (1918) used salsa to designate the yellow-bellied toad, assuming that it was the oldest available name for this species. Only the yellow-bellied toad is present in Bavaria, namely the Northern lineage (carrying mtDNA BW-N). Bufo salsus Schrank in Schrank & Moll, 1785 thus applies to this lineage, and is a junior subjective synonym of Bombina variegata variegata (N°1 in Fig. 8).

Rana sonans Bonnaterre, 1789 – Originally mentioned in La Cepède (1788) as a substitute name for Rana campanisona Laurenti, 1768, Rana bombina Linnaeus, 1761 and Rana variegata Linnaeus, 1758. The species was reported from Germany (“Allemagne”), with an illustration and a French name, “la grenouille sonnante” (La Cepède 1788: pl. XIII, 293–294; Fig. 9C). Although La Cepède lists the Latin name “Rana sonans” in his Synopsis Methodica (La Cepède 1788: 457), the name is not made available by his work, which has been considered invalid as non-binomial by the International Commission of Zoological Nomenclature (2005). Like many of La Cepède’s names, Rana sonans was made available by Bonnaterre (1789; see Dubois et al. 2021), who associated it to a reproduction of La Cepède’s illustration (Bonnaterre 1789: 7, pl. 2–fig. 3; Fig. 9D). However, Bonnaterre’s diagnosis of Rana sonans corresponds to the bull frog Lithobates catesbeianus (Shaw, 1802); this appears to be an unintentional mix-up, perhaps due to the presence of both species next to each other’s on the original plate of La Cepède (1788: pl. XIII). Instead, Bonnaterre referred to “la grenouille sonante” for Rana bombina and mentions the same illustration (Bonnaterre 1789: 4, pl. 2–fig. 3). To clarify this situation, we designate the specimen depicted in La Cepède (1788: pl. XIII) and Bonnaterre (1789: pl. 2–fig. 3) as the lectotype of Rana sonans (Fig. 9C–D). La Cepède (1788: 294) indicated that several specimens of this species are kept at the “Cabinet du Roi [= now Muséum National d’Histoire Naturelle Paris, MNHN]”, thus potentially including the lectotype, but his collections are not present at MNHN (A. Ohler pers. comm.). The issue remains whether this taxon belongs to the yellow or the fire-bellied toad, since both species occur in Germany. La Cepède (1788) and Bonnaterre (1789) clearly distinguished Rana sonans (and its intended synonyms) from Bufo igneus Laurenti, 1768 (= the fire-bellied toad), referring to the latter with Laurenti’s description and including a different drawing – a partial reproduction of Rösel’s plate (Bonnaterre 1789: pl.6–fig. 6; Fig. 9B). This distinction and classification (in different genera) tend to suggest that their Rana sonans was meant for a distinct species. Accordingly, La Cepède’s original diagnosis “ventre marbré de noir & de blanc [belly marbled with black and white coloration]” is consistent with the yellow-bellied toad, and the lectotype features a ventral coloration pattern also more typical of that species (Fig. 9C–D). Therefore, we conclude that the lectotype of Rana sonans is a yellow-bellied toad from Germany. Only the Northern lineage (mtDNA BW-N) inhabits this region, hence Rana sonans Bonnaterre, 1789 applies to this lineage and is a junior subjective synonym of Bombina variegata variegata (N°2 in Fig. 8).

Bufo ignicolor Fibig & Nau, 1789 – Originally mentioned in the Synopsis Methodica of La Cepède (1788: 460) as Buffo ignicolor, with the genus name “Buffo” intentionally misspelled to honor Buffon, La Cepède’s mentor, according to Dubois and Ohler (2009). In the account, La Cepède (1788: 362–364) used the French name “le couleur de feu” and explicitly referred to Bufo igneus Laurenti, 1768, citing the observations of Laurenti from the Danube. As mentioned above, La Cepède (1788) is considered an invalid work, and the name next appears in Fibig and Nau (1789: 297), where it is listed as Bufo ignicolor, with reference to “le couleur de feu” and to “Laur. n. 13” (= Bufo igneus Laurenti, 1768). Bufo ignicolor Fibig & Nau, 1789 thus appears as a junior objective synonym of Bufo igneus Laurenti, 1768, and by implication, it refers to Bombina bombina.

Bufo pluvialis Daudin, 1803 – Mentioned on a plate (Daudin 1803: pl. XXVI) and intended as a substitute name for Rana bombina Linnaeus, 1761 (and Bufo bombinus). The description does not state specific localities but mentions “assez frequent dans les eaux stagnantes et croupies des contrées méridionales de l’Europe [relatively frequent in the stagnant freshwaters of southern European countries]” (Daudin 1803: 75–76). It ends by referring to Bufo igneus and Rösel’s plates 22 and 23. The given geographic range could apply to either the yellow- or fire-bellied toad, and it is unclear which species is represented in Daudin’s drawing (Daudin 1803: Pl. XXVI; Fig. 9E). The diagnosis tends to suggest the fire-bellied toad, according to details on the coloration: “dessous du corps presque lisse et d’un jaune-orangé, avec des taches bleuâtres [nearly smooth ventral side with yellow-orange coloration and blueish blotches]”; and the webbing extent: “cinq doigts presqu’entièrement palmés aux postérieurs [five toes almost entirely webbed]”. In addition, Daudin specifically distinguished a separate variety within his species, which features characteristics closer to the yellow-bellied toad: “blanchâtre à taches noires en dessous; pieds à bandes brunes, jaunes en dessous, ayant leurs doigts séparés [whitish underbelly with black blotches; feet with dark bands, yellow underneath, and with separated toes]”; he accordingly linked this variety to Schrank’s Bufo salsus (= B. variegata variegata, see above). To fix the name, we first designate the specimen illustrated by Daudin (1803: pl. XXVI) as the lectotype of Bufo pluvialis (Fig. 9E). The fate and exact nature of this specimen being unknown, we then restrict its application to the fire-bellied toad by designating NHMW 6698/1 as the neotype of Bufo pluvialis Daudin, 1803 (Fig. 9J). NHMW 6698/1 is a fire-bellied toad specimen curated at NHMW; it was collected in 1889 by Steindachner in “Jedlesee b. Wien [= Jedlesee, Vienna, Austria]”, which becomes the new type locality of Bufo pluvialis. Specifically, Jedlesee is located on the left bank of the Danube just a few hundred meters of Nussdorf (the village mentioned by Laurenti in its description of Bufo igneus), so this act stabilizes Bufo pluvialis as a synonym of Bombina bombina but also of Bufo igneus, as Daudin seems to have intended. This dual restriction is relevant for future revisions, given that B. bombina diversified into several lineages in southeastern Europe (Fijarczyk et al. 2011). Bufo pluvialis Daudin, 1803 now unambiguously refers to Bombina bombina.

Rana cruenta Pallas, 1814 – Mentioned from “Rossia media [= central Russia]” with no reference to any specimen (Pallas’s collections are not known to exist, Kuzmin 1996). Some elements of diagnosis and the mention “Forma R. temporariae, cujus et magnitudo [= shape and size of Rana temporaria]” (Pallas 1814: 12) led many authors to interpret this taxon as a member of the true frogs and it was consequently associated with several Ranid species in the past (reviewed by Kuzmin 1996). However, Pallas questioned his taxon as a variety of “R. bombinae”, did not refer to any other species corresponding to the fire-bellied toad in his work, and some details on the coloration consistently match that species e.g., “subtus cinerea, lituris sparsis rubris, punctis que rubris circa clunes [gray underneath, with scattered red stripes and red dots around the hips]”. It is thus parsimonious to believe that Rana cruenta is a Bombina toad (Kuzmin 1996). The fire-bellied toad is the only Bombina species inhabiting Central Russia and these populations all consist of a single shallow lineage (Fijarczyk et al. 2011). The other Bombina species occurring in Russia, Bombina orientalis (Boulenger 1890), is restricted to the Far-East (notably Primorsky Krai on the Pacific coast), which was part of the Chinese empire (Manchuria) in Pallas’ time, and was not visited by him. Hence, Rana cruenta Pallas, 1814 quite likely refers to Bombina bombina.

Bombinator pachypus Bonaparte, 1838 – Identified in the Italian mountains and explicitly distinguished from Bombinator igneus (Laurenti, 1768) – the only accepted Bombina species at this time (see Fig. 9F–G). Bonaparte described this taxon following correspondences with Fitzinger from Vienna, who examined specimens from “Alpi Apuane [Apuan Alps]” sent to him by Salvi (Bonaparte 1838); these specimens are however absent from the herpetological collection of NHMW (G. Gassner pers. comm.). The description further mentions occurrence in “Monti Ascolani”. Bonaparte’s distinction was ignored until Boulenger (1886) separated the fire-bellied toad in Eastern Europe from the yellow-bellied toad in Western Europe (including Italy), based on detailed drawings (Boulenger 1886, pl. I; Fig. 9H–I). Initially, Boulenger attributed Bombinator igneus to the fire-bellied toad and Bombinator bombinus to the yellow-bellied toad, as he believed the latter to occur in Sweden given Linnaeus’ description and reference to Rösel’s drawings (see above). After being informed that Sweden was instead inhabited by the fire-bellied toad, Boulenger (1888) synonymized Bombinator bombinus with Bombinator igneus and used Bombinator pachypus for the yellow-bellied toad instead. In these arrangements, Boulenger disregarded the law of priority and voluntarily perpetuated igneus and pachypus rather than older names like bombina, variegata and salsa, attracting criticism and prompting further revisions (Stejneger 1907). Eventually, Bombina salsa and then Bombina variegata were used for the yellow bellied-toad, and the Apennine pachypus became one of its subspecies, as Bombina salsa pachypus (Stejneger 1907) and then Bombina variegata pachypus (Mertens and Müller 1928). More recent authors have considered it as a different species (e.g., Bombina pachypus; Lanza and Vanni 1991). The Apuan Alps are inhabited only by the Apennine lineage of B. variegata (mtDNA A), hence Bombinator pachypus Bonaparte, 1838 is the oldest available name for it, as Bombina variegata pachypus (N°B in Fig. 7).

Bombinator pachidactylus Quattrocchi & Battistelli in Bonaparte, 1838 – An alternative name for Bombinator pachypus that is mentioned only on the legend of the plate representing that species, signed by Quattrocchi, the illustrator, and by Battistelli, the lithographer. It is likely to be an error, which may have contributed to Boulenger’s open criticism of Bonaparte’s work (Boulenger 1886: 499). Bombinator pachidactylus Quattrocchi & Battistelli in Bonaparte, 1838 is an objective synonym of Bombinator pachypus and thus of Bombina variegata pachypus (N°6 in Fig. 8).

Bombinator brevipes Blasius, 1839 – Reported in Blantenburg [Blankenburg (Harz)] with locality given as “am Unterharze, namentlich bei Goslar, in einem Sumpfe am Fusse des Hartzensteins und bei Osterode vorkommend [occurring in the Lower Harz, especially near Goslar, in a swamp at the foot of the Hartzenstein, and near Osterode]”, and no reference to any specimen (Blasius 1839: 667; see also Mertens and Wermuth 1960). It was subsequently viewed as a mountain form and given as a variety of Bombinator igneus (Koch 1872). Boulenger (1886) synonymized it with Bombinator pachypus, and later considered it as a variety of that species from outside the Apennines, namely in northern Italy and eastern Europe (Boulenger 1896). The Harz mountains are only inhabited by B. variegata, namely from the Northern lineage (carrying mtDNA BW-N), which makes Bombinator brevipes Blasius, 1839 a junior subjective synonym of Bombina variegata variegata (N°3 in Fig. 8).

Bombinator scaber Küster, 1843 – Discovered in Montenegro with locality given as “kleine Lache bei Cettigne [= small puddle near Cetinje]” and explicitly distinguished from Bombinator igneus (Laurenti, 1768), but with no reference to any specimen (Küster 1843: 656). It was kept as a subspecies of B. variegata by Mertens and Müller (1940). Cetinje, close to the Bay of Kotor, is only inhabited by B. variegata, namely the Balkan lineage (mtDNA BW-S), which makes Bombinator scaber Küster, 1843 the oldest available name for it, as Bombina variegata scabra (N°C in Fig. 8).

Bombinator appeninicus Gistel in Gistel & Bromme, 1850 – Distinguished from Bombinator igneus with distribution given as “Italien auf den Apenninen [Italy in the Apennine Mountains]”, without mention of a specific locality or any specimen (Gistel and Bromme 1850: 333). Stejneger (1907) considered it a synonym of Bombina salsa pachypus (= Bombina variegata pachypus). The Apennine Mountains are inhabited only by the Apennine lineage of B. variegata (carrying mtDNA A), so Bombinator appeninicus Gistel in Gistel & Bromme, 1850 indeed appears to be a junior subjective synonym of Bombina variegata pachypus (N°7 in Fig. 8).

Bombina maculatus Gistel, 1868 – Described from “Regensburg” in Bavaria, Germany, with no reference to any specimen (Gistel 1868: 162). It is synonymized by Mertens (1936). Bavaria is only inhabited by B. variegata, namely the Northern lineage (mtDNA BW-N), which makes Bombina maculatus Gistel, 1868 a junior subjective synonym of Bombina variegata variegata (N°4 in Fig. 8).

Bombinator pachypus var. Kolombatovici Bedriaga, 1890 – Given as a variety of Bombinator pachypus (= the yellow-bellied toad, now B. variegata) of large size and documented from “Dalmatien [Dalmatia]”, based on specimens from “Spalato [Split, Croatia]” (Bedriaga 1890: 568). It is subsequently considered a subspecies of the yellow-bellied toad, as Bombina salsa kolombatovici (Stejneger 1907) then Bombina variegata kolombatovici (Mertens and Müller 1928). The Dalmatian coast (including near Split) seems to be only inhabited by the Balkan lineage of B. variegata (mtDNA BW-S), so Bombinator pachypus var. Kolombatovici Bedriaga, 1890 appears to be a junior synonym of Bombina variegata scabra (N°8 in Fig. 8).

Bombinator pachypus var. nigriventris Dürigen, 1897 – Discussed as a black-bellied variety characteristic of “bosnischen Exemplaren [Bosnian specimens]” (Dürigen 1897: 546), without information on whether these were collected and curated. The author mentioned similarities with an (unnamed) local variety from Montenegro documented by Schreiber (1875: 96). This taxon was subsequently considered as a subspecies of the yellow-bellied toad by Stejneger (1907), namely as Bombina salsa nigriventris, but is explicitly rejected by Fejérváry (1923) and Mertens (1928), who qualified it as a local melanistic variation. The taxon corresponds to either the Northern lineage of B. variegata (mtDNA BW-N), which reaches the northern half of Bosnia, or the Balkan lineage of B. variegata (mtDNA BW-S), which potentially extends along the Adriatic coast of Bosnia. Their respective distributions in the area remain incompletely delimited and are potentially complex, especially as these lineages may hybridize and admix. Bombinator pachypus var. nigriventris Dürigen, 1897 is thus either a junior synonym of Bombina variegata variegata, or a junior synonym of Bombina variegata scabra (N°10 in Fig. 8).

Bombina salsa var. csikii Fejérváry, 1923 – Described based on 4 specimens from “Ipek” (Mus. Hung. Amph. [MNH] 2540/5), 1 specimen from “Mount Korab collected, at 1800 m.” (MNH 2540/6) and 1 specimen from “Kula Lums (from the banks of the Luma)” (MNH 2561/3), all near the border of present Albania, Kosovo/Serbia and North Macedonia. The author refers to a form with extremely developed horny spines and a black ventral coloration, which Schreiber (1912: 176) already noted in Montenegro. Partly similar observations and reference to Schreiber’s earlier edition of his work (Schreiber 1875) had previously motivated the distinction of the form nigriventris by Dürigen (1897) in Bosnia, but Fejérváry rejected the latter based on the examination of the Bosnian specimens he had at his disposal. Mertens and Müller (1928) considered Fejérváry’s taxon as the subspecies Bombina variegata csikii, until they synonymized it with Bombina variegata scabra (Mertens and Müller 1940). The area covered by the type locality (northern Albania) is only inhabited by the Balkan lineage (mtDNA BW-S), hence Bombina salsa var. csikii Fejérváry, 1923 is a junior subjective synonym of Bombina variegata scabra (N°9 in Fig. 8).

Bombina variegata gracilis Bolkay, 1929 – Described from “Zabrgje (Northeastern Bosnia) [=Zabrđe]” based on 9 males and 5 females (and “some younger larvae”), curated at the Bosnia-Hercegovina State Museum in Sarajevo (= now National Museum of Bosnia and Herzegovina, ZMBH). The author also mentions that specimens collected in the “Save [= Sava]” river belong to this form. The diagnosis emphasizes weakly developed warts compared to western Balkan populations attributed to B. variegata kolombatovici (Bolkay 1929: 2, pl. I). Northeastern Bosnia is only inhabited by the Northern lineage (mtDNA BW-N), so Bombina variegata gracilis Bolkay, 1929 is a junior subjective synonym of Bombina variegata variegata (N°5 in Fig. 8).

Bombina bombina danubialis Călinescu, 1931 – Described from several sets of unspecified specimens collected in south and southeastern Romania, namely 16 from “Lunca Mofleni, lânga Caiova, judetul Dolj [Lunca Mofleni, near Craiova, Dolj county]”, 5 from “Lacul Herăstrău, lângă Bucureşti [Herastrau Lake, near Bucharest]”, 1 from “Băneasa”, 3 from “Olteniţa, judeţul Ilfov [Olteniţa, Ilfov county]”, 13 from “un izvor în pădurea Comana (Vlaşca) [a spring in the Comana forest (Vlasca)]”, 51 from “Craiova (canalul cald al moarei Mendel) şi împrejurimi (Lunca Mofleni, pârâul Izvorului din Luncă şi Balta Şerca) [Craiova (hot water canal of Mendel mill) and surroundings (Lunca Mofleni, Izvorului stream from Luncă and Balta Şerca]”, 39 from “Tulcea (bălţile şi smârcurile dinspre grădinării, sub dealul Monumentului) [Tulcea (the puddles and marshes from gardens, under the Monument hill)], 3 from “insula farului la Sf. Gheorghe, judeţul Tulcea [lighthouse island at Sf. Gheorghe, Tulcea county], 5 from “Periprava, judeţul Tulcea [Periprava, Tulcea county]” and 14 from “Vâlcov (canale părăsite), judeţul Ismail [Vâlcov (abandoned channels), Ismail county]” (now in Odesa Oblast, Ukraine). The author emphasized the distinctiveness of his taxon in terms of belly coloration (uneven yellow/yellow-orange blotches, never red), and the presence of dorsal spiny warts (Călinescu 1931: 135–140). From the geographic origin, description and drawings, Bombina bombina danubialis Călinescu, 1931 unambiguously refers to Bombina bombina.

Bombina bombina var. viridis Marián, 1959 – Given as a color variety characterized by a greenish dorsum and grey-pink flanks, sighted in “vielen Stellen der ungarischen Ebene [numerous places in the Hungarian plain]” and in the same biotopes as the nominal form. The author mentions the collection and terrarium breeding of specimens (perhaps subsequently curated in Móra Ferenc Múzeum, Szeged, Hungary). From the geographic origin and description, Bombina bombina var. viridis Marián, 1959 clearly refers to Bombina bombina.

Bombina bombina arifiyensi Özeti & Yilmaz, 1987 – Described as a subspecies inhabiting Anatolian Turkey, with holotype SZE 9/1983-11 from “Arifiye” [Sakarya Province, Turkey]. This taxon was noted for differences in color and morphology patterns compared to specimens collected from Thrace Province (European Turkey). From the geographic origin, description, and photographs, Bombina bombina arifiyensi Özeti & Yilmaz, 1987 refers to Bombina bombina.

Finally, a few non-Bombina taxa have occasionally been associated with fire/yellow-bellied toads in some historical work, e.g., Rana Rubeta Linnaeus, 1758 (a synonym of Bufo bufo (Linnaeus, 1758)) as the “Feuerkröte” [= the fire-bellied toad] in Lindaker (1791: 112), and listed with Bombinator igneus in Schreiber (1912: 178); Rana campanisona Laurenti, 1768 (a synonym of Rana temporaria Linnaeus, 1758) with Rana sonans in La Cepède (1788: 294).

There are, to the best of our knowledge, no other names that apply to European Bombina populations. While three of our B. variegata lineages delimited as subspecies have already been named (B. v. variegata, B. v. pachypus, B. v. scabra), the Rhodope populations seem to have never been the focus of any taxonomic work. We therefore describe it as a new subspecies in the next section.

Bombina variegata rhodopensis Lukanov, Denoël, Jablonski & Dufres­nes, subsp. nov.

https://zoobank.org/E61B7D1D-FF8D-48A4-93C9-DE43EAC3B2B4

Identity

Previously identified as a divergent mitochondrial (cyt b) lineage, attributed to the subspecies Bombina variegata scabra by Hofman et al. (2007) and Fijarczyk et al. (2011) and labeled as the “Rhodopean”, “R” or “RD”, in reference to its distribution in the Rhodope Mountains of southeastern Bulgaria and northeastern Greece. This new taxon corresponds to the Rhodope lineage (mtDNA R) that is purple-coded in our figures. Given its level of divergence (see Discussion), we describe this lineage as a new subspecies, Bombina variegata rhodopensis subsp. nov.

Holotype

BG-IBER-VER-000010561, adult male collected on 23 April 2024 by SL, MD, DJ and CD in a water fountain at the northeastern exit of Kostino, Kardzhali Municipality, Kardzhali Province, Bulgaria (41.7039°N, 25.3028°E; elevation: 563 m a. s. l.), and curated at the Institute of Biodiversity and Ecosystem Research of the Bulgarian Academy of Sciences (IBER-BAS), Sofia, Bulgaria. Measurements in mm: L. 42.7, F. 16.3, T. 17.3, Lt.c. 12.8, Sp.p. 4.8, L.o. 4.4, Sp.n. 2.4. The holotype and the type locality are depicted on Fig. 10.

Figure 10. 

The holotype and type locality of Bombina variegata rhodopensis subsp. nov. Credit: CD (top and middle) and DJ (bottom).

Paratypes

BG-NMNHS-HER-000000000552 (L. 41.6 mm), adult male collected on 23 April 2024 by SL, MD, DJ and CD at the type locality and curated at the National Museum of Natural History of the Bulgarian Academy of Sciences (NMNHS-BAS), Sofia, Bulgaria; MNHN-RA-2024.0001 (L. 46.0 mm), adult female collected on 24 April 2024 by SL, MD, DJ and CD in Lisitsite, Kardzhali Municipality, Kardzhali Province, Bulgaria (41.6108°N, 25.4543°E; elevation: 233 m a.s.l.), and curated at MNHN. NHMW 41962 (L. 39.7 mm), adult male collected on 24 April 2024 by SL, MD, DJ and CD in Panichkovo, Chernoochene Municipality, Kardzhali Province, Bulgaria (41.8565°N, 25.1517°E; elevation: 753 m a.s.l.) and curated at NHMW. These three specimens are depicted in Fig. 11.

Figure 11. 

The paratypes of Bombina variegata rhodopensis subsp. nov. Credit: CD.

Diagnosis

General characteristics similar to those of the yellow-bellied toad B. variegata. It is the sister taxon of the Balkan subspecies B. v. scabra, from which it is distinguished by substantial mitochondrial, nuclear (especially phylogenomic) and morphological divergence. Specifically, B. v. rhodopensis subsp. nov. differs from B. v. scabra by 1.3% of sequence divergence at 16S, 1.6% of sequence divergence at cox1, and 2.4% of sequence divergence at cyt b (Table 1). Based on the mitochondrial time tree, the two taxa initiated their divergence during the Early Pleistocene, either around 2.1 or 1.4 Mya, depending on the calibration (Fig. 2). On the cyt b gene, the following nucleotides distinguish B. v. rhodopensis subsp. nov. from any other B. variegata subspecies: a “G” in the site 469, a “G” in the site 726 and an “A” in the site 801 (positions relative to the full gene sequence). The new subspecies also features a unique combination of alleles at the nuclear genes ncx1, rag-1, rag-2 and rho (Figs 3, S3), as well as at 4759 RAD loci (Fig. 4). Externally, B. v. rhodopensis subsp. nov. differs from B. v. scabra by having on average a shorter femur, tibia, and first toe, as well as bigger eyes, noting that these rely on a small sample of B. v. rhodopensis subsp. nov. specimens (Table 2). Body sizes are on average similar (43.6 mm in B. v. rhodopensis subsp. nov. vs. 43.3 mm in B. v. scabra), with higher variation between the sexes in B. v. rhodopensis subsp. nov. (Fig. 5). From the examined specimens, the sexes appear morphologically dimorphic (Table 2, Fig. 5), as also seen in B. v. scabra (Radojičić et al. 2002; Fig. 5). Compared to B. v. scabra, B. v. rhodopensis subsp. nov. features a higher average proportion of yellow vs. dark coloration on the ventral side, although with wide overlap (Fig. 7).

Etymology

The name rhodopensis is a Latin toponymic adjective given in reference to the Rhodope Mountains in the southeastern part of the Balkan Peninsula (Bulgaria and Greece) where the new taxon is mostly distributed. It spotlights a rare case of Rhodope endemism in vertebrates – Rhodope endemics are so far known only from plants and invertebrates.

Vernacular names

Rhodope yellow-bellied toad (English), Родопска жълтокоремна бумка (Bulgarian), Κιτρινομπομπίνα της Ροδόπης (Greek), Rodop Sarılı Kurbağa (Turkish), Sonneur à ventre jaune des Rhodopes (French), Kunka žltobruchá rodopská (Slovak).

Distribution

Bombina v. rhodopensis subsp. nov. is essentially restricted to the Rhodope Mountains and their foothills (from sea level up to 1600 m a.s.l.) in southeastern Bulgaria, northeastern Greece, and the adjacent part of Turkish Thrace (Global Biodiversity Information Facility 2024). In Greece, it was documented eastward up to the National Forest Park of Dadia – Lefkimi – Soufli close to Evros (Maritsa) River which makes the border with Turkey (Petrov 2004; Valakos et al. 2008; Kret and Poirazidis 2015; Pafilis and Maragkou 2020; Strachinis 2024). In Bulgaria, population isolates exist east of the Rhodopes, namely in Sakar Mountain and perhaps Strandzha Mountain near the Black Sea (Boev et al. 2008; Stojanov et al. 2011). In Turkey, it is very rare and restricted to Karacahasan Mountain (Enez District), close to the Evros River (Kariş et al. 2017). The eastern and northern margins of B. v. rhodopensis subsp. nov. correspond to the shifts from forest hills towards open lowland habitats colonized by the fire-bellied toad (B. bombina), where they probably form hybrid zones. The southern boundary of the range follows the coastal foothills of the Rhodopes (Valakos et al. 2008). According to the mtDNA barcoding, the transition with B. v. scabra in the west might follow the Nestos (Mesta) River valley, which separates the Rhodopes from the Rila massif in the north, and from the Pirin massif in the west; the mtDNA of both subspecies were reported in the middle part of the Nestos River in Greece (Platanias-Pteleas). Nevertheless, many sampling gaps remain in the Pirin, western Rhodopes and southern Rila, so the exact subspecies boundaries shall be fine-tuned by multilocus genotyping.

Natural history

The new subspecies inhabits similar habitats as B. v. scabra, being found in various aquatic sites such as mountain brooks, rivers, ponds, natural and artificial lakes or water-filled ruts and puddles (Petrov 2004). The type series was found in drinking throughs, a valuable habitat for this subspecies as it is for B. v. scabra in northwestern Greece (Denoël 2004). The breeding season starts in March and lasts until late July. Reports of several thousand eggs laid at the bottom of a slow-flowing stream in Eastern Rhodopes suggested possibly up to 200 per female (Stojanov et al. 2011). Toads are most active during the day and at dusk, and hibernation occurs on land. Diet studies indicate that Bulgarian populations are mostly insectivorous, preferring water beetles and winged insects, with arachnids and snails having a minor share (Donev 1984). Bombina v. rhodopensis subsp. nov. displays the anti-predator defense posture (“Unkenreflex”), well-known in Bombina toads (Baijger 1980), and which we could observe in Bulgaria (see also Kariş et al. 2017 for Turkey).

Conservation

Bombina v. rhodopensis subsp. nov. was reported widespread over the eastern Rhodopes of Greece and Bulgaria (Petrov 2004), but due to its much smaller distribution than that of B. v. scabra, it may be more vulnerable. In Greece, it is abundant above 200m (Strachinis 2024). In Turkey, it subsists only in a few isolated localities with potentially small population sizes (e.g., Kariş et al. 2017), and any potential pressure (habitat change, collection) might threaten its persistence in the country. Informed conservation management would benefit from assessing the continuity of the distribution of B. v. rhodopensis subsp. nov. across its range, in respect to known occurrence data (e.g., Valakos et al. 2008; Global Biodiversity Information Facility 2024) and from quantifying habitat loss and threats. As B. variegata is listed in the European Union Habitats Directive Annex 2 (= “Natura 2000”), dedicated protected areas could be declared for the conservation of B. v. rhodopensis subsp. nov.