Comparative Transcriptomic Analyses Identify Candidate Genes for Convergent Reproductive Shifts in a Bimodal Viviparous Amphibian

ABSTRACT

Shifts in reproductive mode represent key evolutionary innovations that shape species' life histories and evolutionary trajectories. Species showing bimodal reproductive strategies with multiple independent origins offer a rare opportunity to gain insights into the adaptive processes and mechanisms underlying convergent traits. The fire salamander, Salamandra salamandra, is the only amphibian exhibiting intraspecific variation in reproductive mode across multiple independent reproductive shifts, enabling investigation of the transition between larviparity (females give birth to aquatic larvae) and pueriparity (females give birth to fully developed terrestrial juveniles) within a single species and across different timescales. Pueriparity is an adaptive innovation that skips the aquatic larval stage, allowing individuals to exploit habitats with no available water bodies. The fire salamander is larviparous across most of its range, but pueriparity has evolved independently at least three times: once in the early Pleistocene within S. s. bernardezi in the mountains of northern Spain, and more recently on two land‐bridge islands (NW Spain) inhabited by S. s. gallaica. To identify candidate genes associated with these distinct reproductive modes, we compared gene expression profiles of the uterus and oviduct of pregnant females across two independent evolutionary transitions using RNA‐sequencing. We detected shared changes in maternal gene expression among pueriparous S. s. bernardezi and S. s. gallaica relative to their larviparous counterparts, in addition to differences unique to each independent evolutionary transition. Functional enrichment analyses indicated that differentially expressed genes were associated with reproductive timing, angiogenesis, and maternal signalling, consistent with the phenotypic differences observed in the uterine environment and embryonic development between the two reproductive modes. This study represents an important first step towards understanding the genomic basis of the evolution of pueriparity in a remarkable bimodal reproductive system, and provides transcriptomic resources and candidate genes for future research into the genomic architecture underlying this poorly understood adaptive trait.

1. Introduction

Reproductive mode is a fundamental life history trait, and shifts in reproductive strategy represent key adaptive innovations with profound effects on species' evolutionary trajectories. Identifying patterns of convergent evolution in reproductive traits offers insight into the processes and mechanisms underlying these adaptive transitions (Whittington et al. 2025). Amphibians exhibit remarkable diversity in their modes of reproduction, often driven by complex evolutionary adaptations to novel environments (Crump 2015; Gomez‐Mestre et al. 2012; Zamudio, Bell, Nali, et al. 2016). Despite this diversity, the genomic basis of amphibian reproductive evolution remains largely unexplored (Funk et al. 2018). Recent advances in sequencing technology have helped decipher the genomic architecture of adaptive traits, particularly in model organisms (Lehner 2013; Singh and Nüsslein‐Volhard 2015). However, uncovering the genetic basis of adaptive divergence remains challenging in natural populations of non‐model organisms (Barson et al. 2015; Ekblom and Galindo 2011; Steiner et al. 2009), especially for amphibians, where genomic resources remain limited (Kosch et al. 2024).

The ancestral reproductive mode in amphibians is oviparity, characterized by external egg‐laying with an aquatic larval life stage, but viviparity (live birth) has evolved multiple times independently across all three extant orders (Anura, Caudata and Gymnophiona; Liedtke et al. 2022). Among viviparous amphibians, some species are larviparous, giving birth to aquatic larvae, while others are pueriparous, bypassing the aquatic larval stage and delivering fully developed terrestrial metamorphs (Greven 2003). Pueriparity is a remarkable adaptation for a group that is typically characterized by a biphasic, aquatic‐terrestrial life cycle (Duellman and Trueb 1994). The shift to pueriparity confers semi‐independence from water, thus enabling survival in water‐limited habitats and dry environments (Dinis et al. 2025). Although pueriparity is an example of a homoplastic trait that is likely shaped by both genetic constraints (Wake et al. 2011) and environmental factors (Dinis et al. 2025; Dinis and Velo‐Antón 2017; Liedtke et al. 2017), the genetic basis of its evolution remains unknown.

In salamanders, all known cases of pueriparity occur in the subfamily Salamandrinae, specifically in the seven species of the genus Lyciasalamandra (Veith et al. 2020) and four in the sister genus Salamandra (Buckley 2012; Mulder et al. 2022). Salamandra contains six species, two of which are strictly pueriparous (the Alpine salamanders S. lanzai and S. atra ), whereas S. corsica , S. infraimmaculata , S. algira and S. salamandra are all referred to as larviparous. However, the latter two species display an exceptional intraspecific variability in their reproductive mode where larviparity and pueriparity co‐occur within a single species (Alarcón‐Ríos, Nicieza, Lourenço, and Velo‐Antón 2020; Alarcón‐Ríos and Velo‐Antón 2024; Dinis and Velo‐Antón 2017; García‐París et al. 2003; Velo‐Antón et al. 2007). Notably, S. salamandra exhibits mixed reproductive modes and transitional phenotypes across a contact zone where hybrids are viable and fertile (Figueiredo‐Vázquez et al. 2026). Ancestral state reconstruction across S. salamandra has shown that pueriparity evolved at least once during the early Pleistocene in the Cantabrian mountains (García‐París et al. 2003; Mulder et al. 2022) and twice independently during the late Pleistocene or early Holocene in insular populations along the northwest coast of the Iberian Peninsula (Lourenço, Sequeira, et al. 2018; Mulder et al. 2022; Velo‐Antón et al. 2007, 2012) (Figure 1). Applying genomic tools to study these independent reproductive transitions across Salamandra is a powerful approach that can disentangle the conflicting signals of neutral drift and genetic adaptation (Velo‐Antón et al. 2012; Whittington et al. 2022) and can uncover the genetic underpinnings of the shift to pueriparity in a phylogeographic comparative framework (Zamudio, Bell, and Mason 2016).

FIGURE 1.

(A) Study area in northwestern Spain. Larviparous range in blue and pueriparous range in red. The island transition in SW Galicia, Spain, includes two larviparous mainland populations (Coiro and Monteferro, blue circles) and the pueriparous population on the island of Ons (red circle). Across the mountain transition in the Cantabrian Mountains in northern Spain, we sampled the larviparous population of S. s. gallaica in Orallo (blue triangle) and the pueriparous population of S. s. bernardezi in Somiedo (red triangle). Symbols and colours are maintained across figures. (B) Summary of the ancestral state reconstruction across the sister species S. algira and S. salamandra from Mulder et al. (2022), highlighting four transitions to pueriparity, two of which were sampled for this study.

Within S. salamandra , the ancestral mode is larviparity (Mulder et al. 2022), in which females deliver ~20 to 80 larvae into water bodies, whereas pueriparity is characterized by the delivery of up to ~30 fully terrestrial metamorphs (Velo‐Antón et al. 2015). Despite lower fecundity, pueriparity offers reproductive independence from water (Dinis et al. 2025; Liedtke et al. 2017; Lourenço et al. 2017; Velo‐Antón et al. 2007), and has important ethological, ecological, physiological and morphological implications (Alarcón‐Ríos, Nicieza, Lourenço, and Velo‐Antón 2020; Buckley et al. 2007; Figueiredo‐Vázquez et al. 2021; Greven 2003; Lourenço, Antunes, et al. 2018; Lourenço et al. 2019). These differences in fecundity are caused by a series of heterochronic processes arising from the shift to pueriparity, such as incomplete fertilization of ovulated eggs, accelerated and asynchronous larval development in the reproductive tract, and developing larvae feeding on unfertilized eggs (oophagy) and siblings (adelphophagy; Buckley et al. 2007). Common garden experiments have shown that sexually mature females in controlled lab environments maintain their respective reproductive modes regardless of water availability (Alarcón‐Ríos, Nicieza, Lourenço, and Velo‐Antón 2020; Velo‐Antón et al. 2015). This suggests that the phenotype has an adaptive genetic basis likely encoded in the genome, rather than being plastically expressed depending on environmental conditions. This observation is further supported by > 20 years of fieldwork on a local pueriparous population, where females consistently deliver terrestrial juveniles despite the presence of suitable water bodies, as demonstrated by co‐occurring amphibians with a larval stage ( Discoglossus galganoi and Lissotriton boscai ) (Velo‐Antón et al. 2012, 2015).

The study of morphological and physiological changes associated with pregnancy in vertebrates has focused on the oviduct and uterus (Atkins et al. 2006; Biazik et al. 2012; Ramírez‐Pinilla et al. 2012; Shine and Guillette 1988; Wourms et al. 1988). For instance, patterns of gene expression in the uteri of viviparous amniotes are associated with eggshell and placenta formation, gas exchange, nutrient transportation, metabolism and immune system functioning (Brandley et al. 2012; Foster et al. 2020; Gao et al. 2019; Whittington et al. 2015). Viviparity in amphibians differs substantially from that of mammals and reptiles, and the genetic mechanisms underlying these associated morphological and physiological changes remain unknown. In both reproductive modes of S. salamandra , ovulated eggs are coated by a tough egg jelly produced by various glands along the oviduct, which is key for the subsequent egg fertilization process (Greven 2003). Contrary to other urodeles, where fertilization occurs in the uterus, fertilization in S. salamandra takes place in the glandular portion of the oviduct (Greven 1998; Joly and Boisseau 1973), with developmental differences between reproductive modes already evident during the initial stages of pregnancy (Buckley et al. 2007). In the pueriparous mode, nearly 50% of the eggs are not fertilized and instead provide additional nutrition for the developing embryos after larval hatching in the uterus (Buckley et al. 2007), representing a form of matrotrophy. The reduced clutch size in pueriparous females (Velo‐Antón et al. 2015) results from a combination of fewer coated eggs along the oviduct, reduced fertilization in the cranial part of the oviduct, and increased embryonic growth accompanied by oophagy and adelphophagy within the uterus (Buckley et al. 2007). The oviduct and uterus are therefore promising maternal tissues to investigate differential gene expression between larviparous and pueriparous salamanders, as they may explain the ontogenetic and physiological differences between these reproductive modes.

Here, we apply RNA‐sequencing (RNA‐Seq) to identify and quantify differences in uterine and oviductal gene expression in S. salamandra across larviparous and pueriparous populations representing two independent and convergent instances of reproductive mode evolution. We aim to: (a) describe general gene expression patterns and identify tissue‐specific transcription in the reproductive organs of female salamanders; (b) characterize convergent and lineage‐specific patterns of differential gene expression between larviparous and pueriparous females; and (c) identify candidate genes associated with the phenotypic differences between larviparous and pueriparous salamanders.

2. Material and Methods

2.1. Study Area

We focused on two regions where S. salamandra independently evolved pueriparity from the ancestral larviparous state (Mulder et al. 2022): (1) the early Pleistocene transition to pueriparity in S. s. bernardezi in the Cantabrian Mountains of northern Spain (henceforth the ‘mountain transition’) and (2) the more recent (late Pleistocene to early Holocene) transition within S. s. gallaica on two islands (Ons and San Martiño) in SW Galicia, Spain (hereafter the ‘island transition’). Reproductive modes are fixed within these lineages, precluding direct intra‐populational comparisons. Instead, we used a replicated design across both independent transitions to control for lineage‐specific genetic background. To isolate candidate genes associated with the reproductive signal we sampled geographically adjacent larviparous and pueriparous populations within each region to minimize confounding environmental variation. For the mountain transition, we sampled the pueriparous S. s. bernardezi population from Somiedo (Asturias province) and the larviparous S. s. gallaica from Orallo (León province; Figure 1), which are separated by only 17 km but divided by a high‐elevation ridge that restricts gene flow (Velo‐Antón et al. 2021). As pueriparity is fixed across the entire S. s. bernardezi range, we selected the geographically closest population of its sister lineage, S. s. gallaica, to represent the ancestral larviparous state. For the island transition, we sampled two larviparous mainland populations of S. s. gallaica (Coiro and Monteferro) and the pueriparous population on Ons Island, located ~7 km from the mainland. The pueriparous population of San Martiño island was not sampled due to its small and vulnerable population size (Velo‐Antón and Cordero‐Rivera 2017).

2.2. Field Sampling

Sampling for both transitions was completed within a 2‐week period in October 2016, when S. salamandra females are in the intermediate stages of pregnancy in this region (Table S1; Velo‐Antón and Buckley 2015). We searched for active, gravid females on two rainy evenings, sampling both reproductive modes for each transition zone on the same evening to control for reproductive stage and activity period. Salamanders were individually housed in a controlled indoor environment for 3 days prior to tissue sampling to minimize environmental influences on gene expression. This also allowed us to exclude individuals with full stomachs that might have been mistaken for being pregnant. We sampled six females for the mountain transition (three per locality; Figure 1; Table S1) and seven females for the island transition (three pueriparous females from Ons and four larviparous females from the two mainland localities). Each female was euthanized with an anaesthetic overdose (benzocaine; Ethyl 4‐aminobenzoate; Sigma‐Aldrich, Darmstadt, Germany), and the uterus and oviduct tissues were immediately dissected, embryos and eggs were removed, and the tissue was flash‐frozen in liquid nitrogen and stored at −80°C. We recorded the number and developmental stage of the larvae/juveniles/eggs found in each uterus (Table S1), and sampled right and left uteri separately as biological replicates because there were noticeable differences in the number and stage of development of the larvae/juveniles between the two uteri (Table S1). We only sampled one oviduct per female, always sampling the left side. We fully randomized tissue sampling and processing to minimize technical bias.

To generate a more comprehensive reference transcriptome, we included seven additional tissues from two previously collected, non‐pregnant females, one larviparous and one pueriparous (Table S1; heart, kidney, lung, liver, muscle, oviduct and uterus). These samples were used for transcriptome assembly and annotation and for tissue‐specific expression analysis, but excluded from the differential expression analyses.

2.3. RNA Extraction and Sequencing

We randomized tissue sample order prior to laboratory work. We extracted total RNA from approximately 25 mg of tissue using the RNeasy kit (Qiagen, Hilden, Germany) and assessed RNA integrity on an Agilent 2200 TapeStation (Agilent Technologies, Santa Clara, CA, USA). We re‐extracted samples with an RNA Integrity Number (RIN) below 7.5. We enriched for mRNA using the NEBNext Poly(A) mRNA beads, then double‐stranded cDNA was synthesized using the NEBNext first and second strand synthesis kits (all from NEB, Ipswich, Massachusetts, USA). We used an in‐house protocol to prepare DNA libraries using double‐indexed Nextera‐style adapters (Glenn et al. 2019). Libraries were quantified using KAPA library quantification kits and pooled equimolarly for sequencing. All uterus and oviduct samples for differential expression analyses were combined in one pool and sequenced across three lanes of a HiSeq 4000 with paired‐end 100 base pair (bp) reads at Macrogen (Seoul, South Korea). The two additional reference individuals were sequenced independently, the larviparous individual on a HiSeq 1500 with paired‐end 125 bp reads (CIBIO, Portugal) and the pueriparous individual on a HiSeq 4000 with paired‐end 150 bp reads (Berkeley, CA, USA).

2.4. Transcriptome Assembly and Annotation

All bioinformatic processing was performed on the Hydra High Performance Computing Cluster (Smithsonian Institution). Reads were filtered and trailing adapters were removed using trimmomatic v0.33 (Bolger et al. 2014) by setting the quality cut‐off at Q5, which is considered optimal for transcriptome assembly (MacManes 2014). All tissues were assembled together using trinity 2.4 (Haas et al. 2013) with default settings. We refined the transcriptome assembly using the EvidentialGene step of the transpi v1.3.0 pipeline to reduce redundancy and merge isoforms (Rivera‐Vicéns et al. 2022), and assessed completeness with busco v5.8.3 (Manni et al. 2021). We translated both nuclear and mitochondrial transcripts using the appropriate vertebrate code using transdecoder v5.7.1, and protein sequences were annotated with entap v.2.3.0, using the EggNOG 5.0 ortholog database (Huerta‐Cepas et al. 2019).

2.5. Confirming Genetic Relationships Between Larviparous and Pueriparous Population Pairs

To assess the genetic relationships of our populations and confirm that we had sampled two independent origins of pueriparity, we reconstructed phylogenetic relationships and visualized genetic distances between populations. We removed the poly‐A‐tail from the filtered reads using prinseq‐lite 0.20.4 (Schmieder and Edwards 2011) and mapped the reads using bowtie v2.2.9 (Langmead and Salzberg 2012) to a previously identified set of 3070 transcriptome‐derived genes that were considered single locus and phylogenetically informative for the genus Salamandra (Rodríguez et al. 2017). We retained concordantly mapped reads and removed duplicate reads with picard tools (https://broadinstitute.github.io/picard/). We called nuclear Single Nucleotide Polymorphisms (SNPs) using the Genome Analysis Toolkit (gatk) using the haplotype caller pipeline (McKenna et al. 2010), filtered SNPs by minimum depth of 5 reads, minimum SNP quality of 20, and removed sites that showed signs of excess heterozygosity (ExcessHet > 10.0, DP > 5, stand_call_conf > 20.0). We applied additional filtering using vcftools v0.1.16 to produce a strictly filtered dataset of nuclear SNPs (‐‐min‐alleles 2 ‐‐max‐alleles 2 ‐‐remove‐indels ‐‐max‐missing 0.6 ‐‐mac 2 ‐‐minQ 100 ‐‐minDP 15 ‐‐minGQ 30 ‐‐non‐ref‐ac 5). We performed a Principal Component Analysis (PCA) on all unlinked nuclear SNPs using custom R scripts to visualize genetic variation among the different populations and samples. To estimate evolutionary relationships among samples, we generated a maximum likelihood phylogeny from a concatenated alignment of all 3070 loci, using raxml 8.2.12 (Stamatakis 2014), applying the GTRCAT substitution model. Bootstrap support was computed on the best‐scoring tree using 100 iterations of rapid bootstrapping (Stamatakis et al. 2008).

2.6. Gene Expression Quantification

We quantified gene expression across the transcriptome by quasi‐mapping all uterus and oviduct samples against the filtered and annotated reference transcriptome using salmon v1.10.3 (Patro et al. 2017). Quasi‐mapping with salmon has been shown to be both faster and more accurate in estimating expression than traditional full‐mapping approaches (C. Zhang et al. 2017). Transcript counts were imported into R v4.1.1 using tximport 1.32 and combined to gene‐level counts using the entap annotations (Soneson et al. 2016). We normalized expression using the recommended edger offset to account for transcript length and library size (Love et al. 2018).

2.6.1. Expression Patterns

To assess gene expression variation across samples, we normalized count data and applied a variance‐stabilising transformation using the voom function from the limma v3.60.6 R package, which accounts for differences in library size and reduces heteroscedasticity across expression levels (Love et al. 2018). We performed principal component analyses (PCA) on the voom‐transformed expression matrix to visualize clustering patterns among samples based on tissue type and reproductive mode. To further assess global similarity among samples, we calculated Euclidean distances between expression profiles and visualized these relationships using sample‐to‐sample heatmaps generated with the pheatmap function (Pheatmap v1.0.12).

2.6.2. Tissue‐Specific Expression

To identify genes important for reproduction, we compared reproductive tissues (uterus and oviduct) to other tissues using Tau (Yanai et al. 2005), a tissue‐specific metric ranging from 0 (broadly expressed across tissues) to 1 (exclusive to a single tissue). Although this method is considered robust to sample size differences (Kryuchkova‐Mostacci and Robinson‐Rechavi 2017), we restricted the analysis to the reference samples with a more balanced representation across the seven tissues. We identified genes with high Tau values for uterus or oviduct, and we additionally highlighted genes specific to both reproductive tissues relative to non‐reproductive tissues.

2.6.3. Differential Expression Between Larviparity and Pueriparity

To identify convergent differentially expressed genes between both reproductive modes while accounting for the independent evolutionary origins of the transitions, we employed two different approaches. First, we applied quasi‐likelihood F‐tests from the R package edger 4.2.2 (Robinson et al. 2009) to compare expression between all pueriparous and larviparous individuals. In this model we included transition as a co‐factor to identify genes differentially expressed convergently across both transitions, while explicitly accounting for lineage‐specific variation. Additionally, we analysed each transition independently to identify the subset of genes that were independently significant across both transitions. This second analysis was performed only on uterine samples, which had sufficient sample sizes for split analyses. We applied the Benjamini & Hochberg false discovery rate (FDR) at a threshold of 0.05 (Benjamini and Hochberg 1995) to adjust p‐values for multiple comparisons and applied a log‐fold change (LogFC > 1) correction to only identify genes exhibiting at least a twofold change in expression. edger significance values and log fold change in expression between larviparous and pueriparous samples were also visualized using volcano plots. The most differentially expressed genes for each transition as ranked by their FDR scores were plotted individually to illustrate gene‐specific differences (Figures S4 and S5).

2.6.4. Functional Enrichment Analyses

To investigate biological processes associated with differential gene expression, we performed gene set enrichment analysis (GSEA) using the clusterprofiler R package. GO term annotations were derived from EggNOG functional annotation. We extracted and filtered gene‐to‐GO mappings, retaining only non‐obsolete biological process terms from the GO core ontology (go.obo). We ranked differential expression results from edgeR using a signed –log₁₀(p‐value), which incorporates both the significance and direction of expression change. We tested for significant enrichment with 10,000 permutations, adjusted for multiple comparisons by Benjamini–Hochberg correction (FDR < 0.05). We classified GO terms as enriched in either larviparous or pueriparous individuals based on the sign of the normalized enrichment score (NES). The top GO terms were visualized using dot plots and categorized by direction of gene expression to highlight functional divergence between reproductive modes.

3. Results

3.1. Reference Transcriptome

We analysed a total of 2.08 billion paired‐end Illumina reads, comprising 628 million from the 26 uterus samples, 308 million from the 13 oviduct samples and 1.14 billion across all 14 tissues for the two reference individuals. The initial trinity assembly yielded 1,402,216 contigs, which were reduced to 420,162 high‐quality transcripts following filtering with EvidentialGene. Of these, 132,645 were translated into putative protein sequences. The resulting reference transcriptome was highly complete, containing 88.3% complete BUSCO genes, 95% when including partial matches. A total of 26,567 transcripts were annotated with high confidence using the EggNOG 5.0 database, corresponding to 14,706 unique genes that were retained for downstream differential expression analyses. Although this represents only 20% of the putative transcript set, mapping statistics indicated that these annotated transcripts accounted for over 83% of mapped reads. This suggests that the annotated portion captured the vast majority of biologically relevant expression, and that most non‐annotated transcripts likely resulted from mis‐assemblies, spurious transcription, or genes with low expression.

3.2. Confirmation of Independence of Larviparous and Pueriparous Population Pairs

Our SNP calling pipeline yielded 1607 high‐quality, unlinked SNPs. Both the phylogenetic tree and PCA clearly separated the populations into four distinct groups with no evidence of hybridization (Figure S10). Importantly, the maximum likelihood tree confirmed that the two pueriparous populations are not closely related and that we sampled two independent origins of pueriparous reproduction, consistent with previous studies on this species (Burgon et al. 2021; Mulder et al. 2022).

3.3. Overall Expression Patterns

The PCA and heatmap plots of mRNA expression revealed a clear separation between uterus and oviduct samples (Figure 2, Figure S9). Within each tissue, samples showed moderate clustering by transition (Mountain vs. Island), with minimal clustering by collection locality within transitions. This suggests that local environmental variation between sites had limited influence on overall gene expression.

FIGURE 2.

Principal component analysis of mRNA transcript expression across both oviduct and uterus tissues after variance‐stabilising transformation. Plot dimensions are scaled proportionally to match the ratio of variance captured by PC1 and PC2. There is a split between the uterus samples at the top and the oviduct samples at the bottom. A version with individual sample names is provided in Figure S11.

3.4. Tissue‐Specific Expression

Tissue‐specificity was highest for the kidney and lowest for the uterus at a Tau score of 0.95 (547 and 40 specific genes, respectively) and intermediate for the oviduct (89 genes; Figure S1, File S1). When combining uterus and oviduct as a single reproductive tissue, 148 genes were identified as specific to the reproductive tract. Several of the specific genes in both uterus and oviduct tissue were associated with development and hormonal functions, with CUZD1 notable for being both highly expressed and oviduct‐specific. Genes specific to the Salamandra reproductive tract spanned diverse functions but were enriched for mucins and mucin‐modifying proteins (File S1).

3.5. Differential Expression Between Larviparity and Pueriparity

When comparing uterine expression between reproductive modes, 368 genes were significantly differentially expressed after FDR and log‐fold change (LogFC > 1) correction (Figure 3A,C, Table S2, Figure S2). Of these, 258 were upregulated and 110 downregulated in pueriparous individuals relative to larviparous ones. In the oviduct, which had a smaller sample size, only four genes were significantly differentially expressed after multiple testing correction—two upregulated and two downregulated (Figure 3B,D, Table S3, Figure S3).

FIGURE 3.

Volcano plots of (A) uterus transcripts and (B) oviduct transcripts. The x‐axis shows the change in expression (positive means higher expression in larviparous females), and the y‐axis indicates statistical significance as indicated by edgeR. Genes with FDR < 0.01 for the uterus and FDR < 0.05 for the oviduct are highlighted in yellow, and the top nine differentially expressed genes for each tissue are labelled by name. Log2‐scaled counts per million (CPM) expression levels for the top nine genes in uterus (C) and oviduct (D) samples split by reproductive mode. An annotated version of panels (C, D), featuring individual sample identifiers is provided with Figure S12.

In the uterus, 1736 genes were significant when analysing the island transition only (Figure S4, Table S4; 1133 upregulated and 603 downregulated), and 101 genes across the mountain transition (Figure S5, Table S5; 48 upregulated and 53 downregulated). Eleven genes were significantly differentially expressed and regulated in the same direction across both transitions, all of which were also identified as significant in the combined analyses (Table S6). We did not perform oviduct comparisons for the individual transitions due to the low number of samples.

3.6. Gene Set Enrichment Analyses

Gene set enrichment analyses (GSEA) of uterine expression profiles revealed that many enriched biological processes between reproductive modes were broad and related to general cellular functions such as transcription, translation and protein targeting. However, a consistent and biologically meaningful pattern was the upregulation of genes involved in vascular development, angiogenesis and vasculature regulation in pueriparous individuals (Tables S7–, S9, Figures S6–, S8), particularly evident in the island transition (Table S8, Figure S7). Additionally, multiple processes involved with cilium development and organization, and associated processes such as cell motility and cell migration, were also enriched in pueriparous populations.

4. Discussion

RNA‐Seq analyses across two independent transitions from larviparity to pueriparity reveal clear differences in the gene expression profiles of maternal reproductive tissue between both viviparous modes. Given the independent evolutionary origins and distinct environmental context of the two transitions, convergent expression patterns are likely associated with reproductive mode. Some differentially expressed genes unique to a single transition may also be linked to shifts in reproductive mode, although other environmental or genetic factors cannot be ruled out. Many of the genes significantly differentially expressed between reproductive modes are related to reproductive timing, maternal signalling and angiogenesis. We highlight and discuss several candidate genes linked to reproduction in S. salamandra and associated with the shift from larviparity to pueriparity.

4.1. Convergent Patterns of Differential Expression Across Mountain and Island Transitions to Pueriparity

Overall expression patterns in both uterus and oviduct samples clustered more strongly by geographic environment than by reproductive mode (Figure 2), likely reflecting the highly distinct environments of each transition. However, several genes in the uterus exhibited convergent differential expression across both mountain and island transitions to pueriparity, providing strong evidence that they are associated with the shift in reproductive mode. This set includes genes with putative roles in uterine function and embryonic development (Table S2). While these functional annotations are primarily derived from studies in mammals and the model amphibian Xenopus, which differ markedly in reproductive strategies, these findings offer valuable starting points for understanding the evolution of pueriparity in salamanders.

For instance, MCM6 was upregulated in both the uterus and oviduct of pueriparous individuals across both transitions. This gene encodes part of a helicase involved in initiating and continuing DNA replication, and is typically upregulated during the G0 phase of the cell cycle, and critical during periods of rapid cell division. In both Xenopus and Drosophila, MCM6 mRNA is maternally provided to the eggs (Ohno et al. 1998; Sible et al. 1998), and Drosophila larvae lacking a functional MCM6 copy do not show developmental problems until metamorphosis, when these maternal stores are depleted (Schwed et al. 2002). MCM6 also displays different expression profiles in brain tissue during larval development between paedomorphic (i.e., retaining larval characteristics when sexually mature) Ambystoma mexicanum and metamorphosing A. tigrinum (Boley 2009). In S. salamandra , increased maternal MCM6 supply from the uterus may support the acceleration of embryonic development and metamorphosis documented in pueriparous births (Buckley et al. 2007).

B9D1 was highly expressed in both the oviduct and uterus and consistently upregulated in the uterus of pueriparous individuals. This gene is required for ciliogenesis, a process essential for the formation and function of multiciliated epithelial cells (MCCs) in reproductive tissues. The oviduct epithelium in most vertebrates is lined with MCCs (Jantra et al. 2007; Spassky and Meunier 2017), but ciliated cells are also found in the uteri of viviparous amphibians (Greven 2024). MCCs are essential for the transport of eggs from the ovary to the uterus and for maintaining a suitable environment for fertilization, and are thickened in live‐bearing amphibians, sometimes even providing maternal nutrition (Gower et al. 2008; Greven and Guex 1994; M. H. Wake 2015). MCC differentiation is induced by oestrogens and associated with oviductal transport of embryos, and although B9D1 is not directly regulated by oestrogen receptors, it is downstream in the ciliogenesis pathway, and consequently its expression could be increased as a secondary effect of oestrogen‐induced MCC differentiation. Another candidate, TPPP3 (Tubulin polymerization promoting protein 3), is also associated with MCCs (Haider et al. 2019), and is additionally important for embryo implantation in the uterus in mammals and believed to mediate signalling between the uterus and the embryo (Shukla et al. 2018). In S. salamandra , TPPP3 was highly expressed across both reproductive modes, but upregulated in pueriparous females. Although there is no embryo implantation in either mode, TPPP3 in S. salamandra may be involved in the production of MCCs, and in enhanced signalling between the uterus and the embryo.

Conversely, several interesting genes—PDGFD, MTNR1A, NPR3 and STPG4—had low expression levels in pueriparous individuals but were highly expressed in the uterus of larviparous individuals. Platelet derived growth factor D (PDGFD) and other related growth factors are important for cell growth and embryonic development, including mesoderm patterning of the early embryo in Xenopus laevis (Ghil and Chung 1999). They might also influence the level of uterine vascularization, which is high in both reproductive modes (Greven 2003, 2011). MTNR1A (Melatonin Receptor 1A) is part of the melatonin pathway that regulates circadian rhythms and reproductive cycles in a wide array of vertebrates (Biase et al. 2019; Migaud et al. 2005; Wang et al. 2024; Wang et al. 2017). Higher uterine expression of MTNR1A slows embryonic development in fruit bats (Banerjee et al. 2009) and influences egg production in poultry (Feng et al. 2018; Li et al. 2013). The higher expression observed in larviparous females could thus relate to higher egg counts and slower embryonic development relative to pueriparous females (Buckley et al. 2007). Maternal factor gonad‐specific expression gene (STPG4, also called GSE) facilitates epigenetic modification in the embryo via altered methylation dynamics at several stages of development (Eckersley‐Maslin et al. 2018; Hatanaka et al. 2013), which in turn could facilitate changes in embryonic growth and development (Buckley et al. 2007). The NPR3 gene encodes a clearance receptor that binds and removes natriuretic peptides, which maintain meiotic arrest in Xenopus oocytes (Sandberg et al. 1993). Increased uterine expression of NPR3 could thus reduce natriuretic peptide levels, promoting oocyte maturation and potentially contributing to the greater fecundity in larviparous individuals.

Analysing uterine gene expression by transition and identifying overlapping candidates revealed 11 consistently differentially expressed genes. These largely overlapped with the combined analysis, with NPR3, PDGFD and MCM6 among the most prominent. Interestingly, LUM (lumican) was not among the top hits in the combined analyses but was significantly upregulated in pueriparous females across both transitions. This gene is involved in angiogenesis and has been found to be important for ovulation in humans (Kedem et al. 2022), and critical for placenta development in pigs (París‐Oller et al. 2021). Additionally, PRSS21, or testisin, upregulated in larviparous females, was initially identified for its role in sperm functioning (Hooper et al. 1999). It has also been found to be expressed and important in the angiogenesis of the mouse reproductive tract (Peroutka et al. 2020), and may aid fertilization when expressed in uterine fluid (Yamashita et al. 2008). Reduced PRSS21 expression in pueriparous uteri may therefore limit fertilization success.

We find evidence of convergent differential gene expression across independent transitions to pueriparity at the intraspecific level. Whether this pattern also extends to the interspecific level (i.e., within the genera Salamandra and Lyciasalamandra) remains unknown. Morphological and physiological contrasts between pueriparous S. salamandra and S. atra indicate that the evolution of pueriparity may proceed via different mechanisms. In S. salamandra , embryos obtain maternal resources through oophagy and embryonic cannibalism (Buckley et al. 2007). By contrast, in S. atra embryos feed on unfertilized eggs and on secretions from a specialized trophic epithelium located in the cranial uterus (Guex and Chen 1986). Therefore, intraspecific transitions may provide a more tractable system for detecting molecular convergence, as lineages share a more recent common ancestor and differ less in confounding genetic and environmental background factors. Indeed, studies investigating uterine gene expression in viviparous mammals, lizards, and sharks have failed to identify convergent genetic mechanisms underpinning independent origins of viviparity at these deeper evolutionary timescales (Foster et al. 2022). Similarly, no significant molecular convergence in protein‐coding genes has been detected among viviparous fishes from different families (Yusuf et al. 2023), and low levels of convergent amino acid replacements found across multiple origins of squamate viviparity could also be explained by chance (Eastment et al. 2024). Yet, similarities in gene expression in gestational tissues have been detected between viviparous lizards and mammals (Brandley et al. 2012; Recknagel et al. 2021; Smout et al. 2026). To better understand the potential for convergent evolution of pueriparity at broader evolutionary scales, comparative gene expression analyses of the uterus and oviduct across all independent transitions are needed.

4.2. Potential Independent Genetic Mechanisms in the Shift to Pueriparity

Our conservative approach of combining both transitions helped identify genes associated with reproductive mode rather than environmental factors or genetic drift. However, this may have overlooked genes specific to each transition. Given the independent evolutionary origins and disparate timings of these transitions (Lourenço, Sequeira, et al. 2018; Mulder et al. 2022), the shared pueriparous phenotype may also result from distinct underlying genetic architectures (Steiner et al. 2007; Wittkopp et al. 2003). Correspondingly, we found over 1700 differentially expressed genes within the island transition, and only 101 for the mountain transition (Tables S4 and S5, Figures S4 and S5). This disparity may reflect the larger sample size in the island group, greater environmental differences, or the small effective population size of the Ons population. Alternatively, lower within‐group variability in expression levels within the island transition may have reduced background noise, facilitating the identification of differentially expressed genes.

Among the island‐specific candidates, we highlight a few genes of particular interest due to their known roles in reproduction across vertebrates. NMB was the second most significant gene, upregulated in pueriparous females on Ons Island. In mice, NMB can initiate labour (Zhang et al. 2011) and thus may similarly regulate birth timing in Salamandra. NTRK3, upregulated in pueriparous females, is a neurotrophin receptor which regulates uterine growth in mammals (Wessels et al. 2014). GPER1, an oestrogen receptor, was consistently expressed in mainland larviparous individuals but absent in pueriparous females on Ons. GPER1 is important for oocyte development in alligators (Wen et al. 2023) and mediates maternal signalling in many vertebrates (Pang and Thomas 2010; Zhang et al. 2025).

These single‐transition candidate genes may reflect unique genetic architectures underlying convergent phenotypes. However, attributing these genes specifically to reproductive mode is challenging, as they may also reflect other environmental or evolutionary differences between populations. For example, S. s. bernardezi and S. s. gallaica differ in colour pattern (Alarcón‐Ríos et al. 2024) and morphology (Alarcón‐Ríos, Nicieza, Kaliontzopoulou, et al. 2020; Velo‐Antón and Buckley 2015), and both transitions occurred in distinct environments (Mulder et al. 2022). Similarly, the pueriparous insular and larviparous continental populations of S. s. gallaica exhibit morphological differences despite their recent evolutionary divergence and geographic proximity (Alarcón‐Ríos, Saabi and Velo‐Antón 2026; Velo‐Antón and Cordero‐Rivera 2017). Thus, some of these differentially expressed genes may also reflect broader physiological adaptations to local environmental conditions beyond reproductive mode (Dinis et al. 2025). Additionally, slight differences in gestational stage due to seasonal variation may obscure stage‐dependent gene expression (Velo‐Antón and Buckley 2015).

4.3. General Gene Expression Patterns of Reproductive Tissues in S. salamandra

In the reference transcriptomes, the kidney exhibited the highest tissue‐specificity, consistent with findings in mammals and fishes (Ramsköld et al. 2009; Salem et al. 2015). Tissue‐specificity was higher in the oviduct than in the uterus (Figure S1), suggesting that the oviduct is a more specialized organ in S. salamandra . Yet, we detected fewer significantly differentially expressed genes between the oviducts of different reproductive modes. This may be due to lower sample size (n = 13), the timing of our sampling (the embryos had already passed through the oviduct to the uterus), or the possibility that the oviduct plays a less significant role in embryonic development.

Although the oviduct and uterus of salamanders are relatively underdeveloped compared to mammals (Wake 1993), we identified numerous genes with tissue‐specific expression. The glycoprotein CUZD1 was highly specific to the oviduct and is known to be expressed in the reproductive tract of mice during late pregnancy (Liaskos et al. 2013). EVX1, specific to the uterus in S. salamandra , is important for embryo implantation in mice (Spyropoulos and Capecchi 1994). Several genes were specific to the combined reproductive tract tissue, including mucins (MUC6, MUC5AC, MUC3A) and enzymes likely involved in mucin modification (GAL3ST3, CHST1, CHST4, CHST9, GALNT13). Mucins are key components of reproductive tract secretions that create a protective, supportive environment for developing embryos. In mammals, uterine mucins form a barrier against pathogens and are carefully regulated to enable embryo implantation (Zhou et al. 2023). Similarly, some viviparous amphibians secrete mucoprotein‐rich ‘uterine milk’ that nourishes and protects embryos in utero (Sandberger‐Loua et al. 2017; Wake 2015). The discovery of multiple mucin genes with tissue‐specific expression in S. salamandra 's oviduct and uterus suggests that these mucus secretions play an important role in both reproductive modes.

4.4. Functional Enrichment of Genes Important for Vascular Development and Ciliary Processes in Pueriparous Individuals

Upregulation of genes involved in vascular development in pueriparous individuals suggests a functional shift in uterine physiology toward enhanced tissue remodelling and angiogenesis. This aligns with patterns observed in viviparous vertebrates, where increased uterine vascularization supports nutrient and gas exchange during internal development (Murphy and Thompson 2011; Van Dyke et al. 2014). In reptiles, histological work has documented increased vascular proliferation in the oviducts of species with varying degrees of viviparity (Parker et al. 2010; Stewart and Blackburn 2014). Such vascular adaptations are thought to facilitate not only embryonic support but also uterine secretory functions critical for sustaining developing offspring.

Furthermore, our Gene Set Enrichment Analysis revealed significant enrichment of GO terms related to cilium organization and cilium development in pueriparous individuals. This suggests an important role for multiciliated cells (MCCs) in the pueriparous uterus, where they might facilitate fluid movement and particle transport. Our findings thus suggest that the structural and cellular changes seen in vascular development and ciliated cells are mirrored at the transcriptional level, and that the transition to pueriparity may involve conserved molecular pathways that remodel the uterus to accommodate larval retention and intrauterine development.

4.5. Expanding the Comparative Framework Across Salamandra

Our genome‐wide transcriptomic approach enabled the identification of maternal candidate genes associated with pueriparity by comparing two independent transitions. This strategy is especially valuable in non‐model systems with limited prior knowledge of gene functions. The candidate genes identified here offer a valuable foundation for future functional studies across Salamandra and other lineages where pueriparity has evolved. Future work should test for signatures of selection in coding and regulatory regions of candidate genes across the different transitions. This approach could be expanded to larger sample sizes and the full Salamandra radiation (Burgon et al. 2021; Dinis et al. 2025; Mulder et al. 2022) using non‐lethal DNA sampling techniques such as tail‐tips. This includes the second island transition on San Martiño (S. s. gallaica; Velo‐Antón et al. 2012), the North African transition in S. a. tingitana (Dinis and Velo‐Antón 2017), and samples from hybrid zones between reproductive modes in both S. salamandra (Figueiredo‐Vázquez et al. 2026; Gippner et al. 2024; Uotila et al. 2013; Velo‐Antón et al. 2021) and S. algira (Dinis and Velo‐Antón 2017), which could be leveraged for admixture mapping.

Tissue choice and timing of sampling are critical in transcriptomic studies (Ferreira et al. 2020; Gao et al. 2019; Todd et al. 2016). By focusing on the oviduct and uterus, we quantified maternal differences in gene expression during pregnancy that may influence embryonic development. Interestingly, we observed substantial variation in larval number and developmental stage even between the two uteri of a single female (Table S1). This variation could be driven by external factors such as genetic differences among offspring or the presence of multiple paternal genotypes (Alarcón‐Ríos, Nicieza, Lourenço, and Velo‐Antón 2020; Caspers et al. 2014; Steinfartz et al. 2006). For example, a pueriparous female from Ons Island crossed with a larviparous male from the mainland can produce larvae, suggesting that paternal regulation may also play a role (Velo‐Antón et al. 2023). Future research should therefore examine both embryonic and paternal gene expression, as each may contribute to phenotypic differences between reproductive modes (Buckley et al. 2009). For instance, assessing gene expression in neural crest cells at different developmental stages (Buckley et al. 2007) could reveal time‐specific regulatory mechanisms that shape developmental trajectories (Gao et al. 2019).

5. Conclusion

The fire salamander, Salamandra salamandra , offers a unique opportunity to study convergent evolution of viviparity within a single species. To our knowledge, this is the first study to characterize gene expression of larviparous and pueriparous individuals and investigate the genetic basis of a remarkable shift in reproductive mode that enables amphibians to colonize and persist in water‐limited habitats. Shared gene expression changes across two independent transitions to pueriparity suggest that maternal gene regulation contributes to differences in embryonic development between viviparous modes. We highlight various candidate genes that may explain this key evolutionary transition, many of which are involved in hormonal regulation of reproductive timing, maternal signalling, angiogenesis and enrichment of multiciliated epithelial cells, factors distinguishing the two reproductive modes. These candidate genes establish a basis for future functional studies to resolve the putative mechanisms underlying this complex evolutionary shift. Together, these findings provide a foundation for future research integrating genetic and geographic data to explore the ecological and environmental conditions driving the evolution of pueriparity.

Author Contributions

K.P.M. and G.V.‐A. designed the study; K.P.M., G.V.‐A., A.L., I.G.‐M. contributed data or samples; G.V.‐A., I.M.‐S., D.B., M.C., I.G.‐M. and R.C.F. contributed reagents; K.P.M. analysed data; K.P.M., R.C.B. and G.V.‐A. wrote the first draft of the manuscript with contributions from all other authors.

Funding

This work was supported by Fundação para a Ciência e a Tecnologia, PTDC/BIA‐EVL/28475/2017, PTDC/BIA‐EVF/3036/2012, FCOMP‐01‐0124‐FEDER‐028325, UIDB/500027/2020, PD/BD/52604/2014, PD/BD/106060/2015, IF/01425/2014, CEECIND/00937/2018, CEECINST/00014/2018/CP1512/CT0002; Agencia Estatal de Investigación, CGL2017‐83131‐P, SEV‐2012‐0262, RYC‐2019‐026959‐I/AEI/10.13039/501100011033; Fonds Wetenschappelijk Onderzoek, 1224223N and 12AFD26N.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: A bar plot quantifying the number of tissue‐specific genes identified across all analysed tissues, based on the Tau score. Tau is the tissue‐specific score (0 = broadly expressed across tissues, 1 = completely specific to one tissue). Gene counts are presented for two stringency thresholds: highly specific (Tau ≥ 0.95) and extremely specific (Tau ≥ 0.99).

Figure S2: Detailed visualization of expression patterns for the top 20 differentially expressed genes in the Uterus. Each subplot depicts log₂‐normalized expression for a single gene, coloured by Reproduction status and shaped by Transition status.

Figure S3: Detailed visualization of expression patterns for the top 20 differentially expressed genes in the Oviduct. Each subplot depicts log₂‐normalized expression for a single gene, coloured by Reproduction status and shaped by Transition status.

Figure S4: Individual expression plots for the top 20 differentially expressed genes identified exclusively in Uterus samples from the Island transition. Each plot displays normalized and log₂‐transformed gene counts for Larviparous and Pueriparous samples.

Figure S5: Individual expression plots for the top 20 differentially expressed genes identified exclusively in Uterus samples from the Mountain transition. Each plot displays normalized and log₂‐transformed gene counts for Larviparous and Pueriparous samples.

Figure S6: A dot plot illustrating Gene Set Enrichment Analysis (GSEA) results for Gene Ontology (GO) Biological Processes in the complete Uterus dataset. It displays the top enriched gene sets, separated by their direction of enrichment (i.e., enriched in Larviparous vs. Pueriparous).

Figure S7: A dot plot illustrating Gene Set Enrichment Analysis (GSEA) results for Gene Ontology (GO) Biological Processes specifically in Uterus samples from the Island transition. It displays the top enriched gene sets, separated by their direction of enrichment.

Figure S8: A dot plot illustrating Gene Set Enrichment Analysis (GSEA) results for Gene Ontology (GO) Biological Processes specifically in Uterus samples from the Mountain transition. It displays the top enriched gene sets, separated by their direction of enrichment.

Figure S9: A heatmap illustrating sample‐to‐sample distances based on gene expression profiles as produced by the R package pheatmap. The clustering of samples reflects their similarity, providing a visual representation of the relationships between different experimental groups.

Figure S10: (a) Principal component analysis (PCA) of 1607 unlinked nuclear SNPs reveals genetic structure among the RNA‐Seq samples used in this study. The deepest split occurs between the S. s. bernardezi population and the three populations of S. s. gallaica. The mountain transition is more genetically divergent (PC1) than the island transition (PC2). (b) Maximum likelihood phylogeny based on a concatenated alignment of 3070 loci recovers the same relationships, with the pueriparous S. s. bernardezi forming the sister clade to all S. s. gallaica populations.

Figure S11: Principal component analysis (PCA) of gene expression profiles across uterus and oviduct samples from larviparous and pueriparous females. This version of the plot includes individual sample names for reference. See Figure 2 for the same analysis without labels for clarity.

Figure S12: Annotated visualization of expression patterns for the top nine differentially expressed genes in the Uterus and Oviduct, featuring individual sample labels. These plots correspond to the data shown in Figure 3C,D. Each subplot depicts log₂‐normalized expression for a single gene, coloured by Reproduction status and shaped by Transition status.

Table S1: List of samples included in this study. The first 14 samples (GVA6688 and GVA6082) were only used to generate the reference transcriptome.

Table S2: Comprehensive differential expression results for Uterus samples. It includes gene‐level metrics such as LogFC (Log of the Fold‐Change), LogCPM (Log of the Counts Per Million), p‐value (exact p‐value for differential expression) and FDR (significance value after correcting for multiple testing by means of the Benjamini & Hochberg false discovery rate) comparing Larviparous and Pueriparous reproduction types.

Table S3: Comprehensive differential expression results for Oviduct samples. It includes gene‐level metrics such as LogFC (Log of the Fold‐Change), LogCPM (Log of the Counts Per Million), p‐value (exact p‐value for differential expression) and FDR (significance value after correcting for multiple testing by means of the Benjamini & Hochberg false discovery rate) comparing Larviparous and Pueriparous reproduction types.

Table S4: Complete differential expression results for Uterus samples specifically from the Island transition. It includes gene‐level information such as LogFC (Log of the Fold‐Change), LogCPM (Log of the Counts Per Million), p‐value (exact p‐value for differential expression) and FDR (significance value after correcting for multiple testing by means of the Benjamini & Hochberg false discovery rate) detailing the gene expression differences between Larviparous and Pueriparous reproduction types within this transition.

Table S5: Complete differential expression results for Uterus samples specifically from the Mountain transition. It includes gene‐level information such as LogFC (Log of the Fold‐Change), LogCPM (Log of the Counts Per Million), p‐value (exact p‐value for differential expression) and FDR (significance value after correcting for multiple testing by means of the Benjamini & Hochberg false discovery rate) detailing the gene expression differences between Larviparous and Pueriparous reproduction types within this transition.

Table S6: Enumeration of genes that are significantly differentially expressed (FDR < 0.05) in both Island and Mountain Uterus samples, with the additional criterion of consistent log fold‐change direction across both transitions.

Table S7: Detailed results of the GSEA for GO Biological Processes performed on all Uterus samples. It lists each enriched GO term, its Normalized Enrichment Score (NES), p‐value, adjusted p‐value (FDR) and leading edge genes.

Table S8: Detailed results of the GSEA for GO Biological Processes applied to Uterus samples specifically from the Island transition. It lists each enriched GO term, its Normalized Enrichment Score (NES), p‐value, adjusted p‐value (FDR) and leading edge genes.

Table S9: Detailed results of the GSEA for GO Biological Processes applied to Uterus samples specifically from the Mountain transition. It lists each enriched GO term, its Normalized Enrichment Score (NES), p‐value, adjusted p‐value (FDR) and leading edge genes.

File S1: A supplementary Excel file comprising three distinct sheets, each enumerating genes identified as highly tissue‐specific (Tau ≥ 0.99) for ‘Reproductive Tract’, ‘Uterus’ and ‘Oviduct’, respectively. Tau is the tissue‐specific score (0 = broadly expressed across tissues, 1 = completely specific to one tissue) and Quant is a relative quantification of expression.

Data Availability Statement

All raw sequencing data are available on the NCBI Sequence Read Archive (SRA) under BioProject number PRJNA1450543. The bioinformatic pipelines and custom scripts used for this study are publicly available in the following GitHub repository: https://github.com/kvpmulder/Salamandra_RNAseq.