A total of 210 specimens of the genus Scomberomorus were collected in different seasons from 2019 to 2021 from different predesignated sampling sites along the coast of India, covering samples by type locality (topotypes) (Figure 1, Table 2). Pampus candidus, Brama sp., Lepidocybium flavobrunneum, and Kajikia audax were selected as outgroups for phylogenetic analysis based on the available literature (Miya et al., 2013) and digitally documented after collection (Supplementary Figure 1). All specimens were tentatively identified by morphological taxonomy (Collette and Nauen, 1983) and their fin or muscle tissues were preserved in absolute ethanol for future molecular studies. In addition, data mining of mitochondrial COI, Cyt b, NADH dehydrogenase subunit 2 (NADH2), ATPase subunits (ATPase 6 and ATPase 8), and nuclear [aldolase exon 5/intron 5 (ALD), rhodopsin (RHO), titin-like protein (Tmo-4c4)] genes were extracted from previous literature and nucleotide databases to represent the maximum species in Scomberomorus. The identity of the species underrepresented in GenBank (e.g., S. sinensis, Scomberomorus tritor, and S. multiradiatus) was deciphered by comparison, wherever possible, with either the morphological characters or the limited genetic data. Total genomic DNA (gDNA) was extracted from all tissue samples by the standard phenol–chloroform method (Sambrook et al., 1989). Isolated gDNA was visualized in 0.8% agarose gel, while NanoDrop One (Thermo Fisher Scientific, USA) determined the quality and quantity. High-quality gDNA was stored in duplicate at −80°C for further downstream processes.

The identified specimens and the outgroups were barcoded to confirm their genetic identity. After this, no fewer than five individuals were selected for multigene amplification, taking into account collection sites and topotypes. We targeted 11 regions of DNA, eight genes from the mitochondrial (mt) genome [coding: COI, Cyt b, ATPase subunits 6 and 8, NADH2; non-coding: D-loop/Control Region (CR), 12S rRNA (12S ribosomal RNA), and 16S rRNA (16S ribosomal RNA)] and three nuclear loci (ALD, RHO, and Tmo-4C4) which, depending on their availability in GenBank, were assigned to different analyses. Amplification was carried out in a 50-µl reaction using basic polymerase chain reaction (PCR) conditions and cycling parameters following the primers and annealing temperatures (Table 3). Our analysis included multiple mt genes to increase the phylogenetic signal and to inspect the presence of accidental pseudogene amplification, while nuclear genes in particular provide regions with slower substitution rates (Banford et al., 1999). Amplified DNA fragments were visualized in 1.2%–2% agarose gel depending on the size of the product, purified, and bidirectionally sequenced.

Amplified sequences were aligned using the ClustalW algorithm (Thompson et al., 1994), which was manually edited using BioEdit v7.1.9 (Hall, 1999) to remove gaps and severe base pair mismatches. Curated coding regions were translated into amino acids to check for the presence of pseudogenes. All generated sequences were compared to the sequences downloaded from nucleotide databases using the search tool BLAST (Altschul et al., 1990). All amplified sequences were deposited in GenBank and the accession numbers with details were given along with the sequences retrieved from GenBank and BOLD (Supplementary Tables 1, 2 and Table 4).

Initially, phylogenetic analysis using a single locus (COI and Cyt b with a truncated length of 632 and 763 bp, respectively) was implemented to incorporate maximal sequences representing the nominal species in Scomberomorus, while multigene analysis focused on increasing the phylogenetic signals, thereby verifying the validity of the primary investigation results. Three hundred and seventy-eight sequences belonging to 15 nominal species (Supplementary Table 1) and 122 sequences from available species (Supplementary Table 2) of the genus were included in the COI and Cyt b-based analyses, respectively. COI contained sequences from all species (except S. sinensis, S. multiradiatus, and S. tritor) along with those generated in this study (n = 148). We also evaluated the possible genealogical relationships between haplotypes in the putative species complex containing cryptics through a haplotype network with a maximal connectivity limit of 95%, mapped with their COI barcodes using the TCS algorithm implemented in POPART version 1.7 (Leigh et al., 2015). For the multilocus dataset, five sequences per species were selected from our samples. It included alignments of COI, Cyt b, ATP8, ATP6, NADH2, RHO, ALD, and TMO resulting in 632, 1,032, 168, 684, 1,047, 762, 410, and 499 base pairs, respectively. These gene fragments were selected to contain the maximum number of congeneric species, and only those with better data coverage/mitogenomes were included in the analysis. The saturation in each codon of each gene was tested using DAMBE v.7 (Xia, 2018) with no evidence of saturation detected at any of the codon positions. The concatenation resulted in a final length of 5,235 bp (Table 4).

Mitochondrial data were retrieved from the complete mitogenomes of the outgroups (except for P. candidus) after the species were confirmed with barcodes (accession nos. OM436012–OM436014). The individual and concatenated matrices were analyzed under the Bayesian inference (BI) and maximum likelihood (ML) method. The best-fit model (GTR) and partition for the datasets were identified using PartitionFinder v1.1.1 (Lanfear et al., 2017) under the parameters greedy algorithm, unlinked branch length, and Bayesian information criterion (BIC). BI analysis was performed through Markov chain Monte Carlo (MCMC) sampling implemented in MrBayes v3.2 (Ronquist et al., 2012) with 5 million iterations in two independent runs, sampling every 1,000 generations and 10% burn-in. Tracer v1.7 (Rambaut et al., 2018) evaluated the convergence of analyses and ensured that the effective sample size (ESS) values were above 200. Finally, the ML analysis was performed in IQ-TREE by 1,000 standard bootstrap with a given best-fit model and partition. The BI and ML phylogram topologies and their respective statistical supports—posterior probability (PP) and bootstrap support (BS) scores—were visualized in FigTree v.1.4.3.

The delineation analysis was performed at two levels with previously used data from COI and Cyt b fragments, followed by a final validation using a multiple gene data matrix, considered here as one locus. Three species delineation analyses [two tree-based methods: the Generalized Mixed Yule Coalescent model (GMYC; Pons et al., 2006; Fujisawa and Barraclough, 2013) and multirate Poisson tree processes (mPTP; Kapli et al., 2017); and a distance-based method: Automatic Barcode Gap Discovery (ABGD; Puillandre et al., 2012)] were used to delineate and discover diversity among genera. The ultrametric trees generated separately for single and multiple genes by MCMC sampling in BEAST v2.6.4 (Bouckaert et al., 2019) were used as input files for single (sGMYC) and multiple-threshold (mGMYC) models, respectively. In BEAST analyses, independent runs were performed on single and multilocus data incorporating a relaxed clock log-normal and a partition and best-fit model akin to phylogenetic analysis for 10 million generation sampling every 1,000 generations. After assessing convergence between runs and ESS scores, both sGMYC and mGMYC analyses were performed using SPLIT packages implemented in R 4.0.2 (R Core Team, 2020). ML phylogram for mPTP analysis executed in Python v.3 (Van Rossum and Drake, 2009) with PTP package (Zhang et al., 2013) was generated in IQ-TREE v1.6.12 (Nguyen et al., 2015), with provided partition and 1,000 standard bootstrap. A distance-based ABGD analysis was performed following the arguments Pmin = 0.001, Pmax = 0, steps = 20, relative gap width = 1, Nb bins (for distance distribution = 20), and genetic distance as simple distance.

An additional confirmatory species delineation for the cryptic species complex using concatenated gene data was performed using the Bayesian Phylogenetics and Phylogeography program (BP&P; Yang and Rannala, 2010). This is a robust method that models the evolution of multilocus data and is widely used for explicit taxon identification and in the analyses of taxon evolution and divergence (Luo et al., 2018). The analysis was carried out with a user-specified guide tree which estimates the probability that each node represents a speciation event (i.e., a split between two distinct species) in bpp v4.4.1 (Yang, 2015). A total of 27 sequences with the concatenated dataset (Table 4) were used for the construction of the guide tree [L, ((T,K),(E,(V,W)))], where L = S. lineolatus, T = S. guttatus, K = S. koreanus, and E, V, and W = S. aff. guttatus, in BEASTv 2.4. rjMCMC algorithm 1 with fine-tuning parameter epsilon = 2 and species model prior m = 1, mutation rate across locus = 1, and an independent rates clock model were chosen to run the analysis. An inverse gamma prior IG (alpha, beta) for ancestral population size (theta) and root age (tau) were estimated using assigned prior values. The assigned alpha and beta values for theta prior were 6 and 0.002, whereas the tau prior values were 4 and 0.06, respectively.

Intra- and interspecific sequence divergence (uncorrected p-distance) of nominal species and MOTUs were calculated after species delineation analysis using the above COI, Cyt b, and concatenated gene sequences under uniform rates in MEGA 11 (Tamura et al., 2021). In MEGA 11, the criterion for the complete deletion of missing data was adopted, resulting in 3,316 nucleotides (all mitochondrial and nuclear ALD fragments used for multilocus phylogenies) being required to calculate genetic distance.

The estimation of divergence time of 10 nominal species was performed using the sequences of eight mitochondrial genes (COI, Cyt b, ATP6 and 8, NADH2, 12S rRNA, 16S rRNA, and D-loop) with a total length of 5,554 bp. Of these, the sequences of six species were retrieved from GenBank (https://www.ncbi.nlm.nih.gov/genbank/), while the remaining four were gathered from this study. To include maximum molecular data and to ensure full coverage of sequences without any data loss from all the species, only mitochondrial genes were used for this analysis. The concluding data matrix for the analysis contained only one individual per species (Table 5). The estimation of evolutionary rates and approximate time to most recent common ancestors (TMRCA) of each species included in the analysis were assessed by BEAST v2.6.4 (Bouckaert et al., 2019). The analysis was conducted with an uncorrelated relaxed clock model using the GTR/GTR+I+G nucleotide substitution model recovered from PartitionFinder v1.1.1. Speciation for the tree prior was selected as the Yule model and alternative calibration points were incorporated with three fossils under the normal distribution. Firstly, the root age of the basal node of the phylogram was calibrated with a lower and upper boundary of 79.66–98 million years ago (mya) (Santini et al., 2013). The second calibration point was fixed to the Acanthocybium solandri–Gymnosarda unicolor clade with a minimum age of 52 mya and an upper boundary of 79.66 mya to calibrate its stem age (Purdy et al., 2001; Santini et al., 2013). Also, we constrain the crown node of the Scomberomorus genus with a Paleocene fossil between 56 and 59.2 mya (Sepkoski, 2002). Input file prepared by BEAUti in the BEAST package. The MCMC run for 100 million generations was executed twice, thinning every 1,000 trees to recover time-stamped phylogenies, and 25% of the initial samples were discarded as burn-in by TreeAnnotator v2.6.4. Tracer v1.7 evaluated the convergence of analysis and ensured that the ESS scores for all parameters were above 200. The chronogram after analysis was viewed in FigTree.

Since some of the aforementioned literature discussed the variation in the number of vertebrae in S. guttatus, this meristic trait was checked for 25 specimens each of S. guttatus, collected from the type locality, east coast (Odisha and West Bengal) and west coast (Gujarat) of India, along with five specimens from other Scomberomorus species. Vertebral counts were determined both from X-radiographs taken at the respective sites and by manual counting after preparing a dry vertebral column following James (2008). Specimens were boiled to 110°C to separate the meat from the bone and the bones were brushed under running water after boiling.

Identity of COI barcodes for Scomberomorus spp. generated in this study were verified with all available species (n = 13) in nucleotide databases. The reliable barcodes for S. lineolatus and S. koreanus were also generated here and their incorrectly labeled sequences in various nucleotide databases were thus recognized. The identity of S. koreanus could be confirmed using the published Tmo-4C4 sequence (DQ388095; Orrell et al., 2006). Some species like S. tritor and S. sinensis were exclusively represented by the Cyt b gene in NCBI. Search results employing Cyt b were correct identifications in most cases, apart from some errors (see Supplementary Tables 1, 2). Although there is no genetic information for S. multiradiatus, morphological evidence and a geographic distribution restricted to the Gulf of Papua (Collette and Russo, 1984) isolated it from the others.

The COI-based Bayesian phylogeny and IQ tree contained two major clades (Figure 2). Major clade I branched into two subclades with a Bayesian posterior probability (BPP) of 0.75. Subclade I was further bifurcated into two groups with high BPP (0.97). Species morphologically identified as S. guttatus have been resolved into distinct barcode clusters representing diverged species in phylogenetic analysis. In this study, sequences identical to those of S. guttatus collected and identified from the type locality (Tranquebar, India) and adjacent areas are considered to represent S. guttatus (Bloch and Schneider, 1801) in sensu stricto (see Discussion). The cryptic species with indistinguishable morphological resemblance to S. guttatus is referred to as S. aff. guttatus in the phylograms and in the manuscript. Thus, group I included S. guttatus, S. aff. guttatus, S. koreanus, and S. plurilineatus. Also, S. aff. guttatus contained two reciprocally monophyletic lineages tentatively designated as S. aff. guttatus 1 and S. aff. guttatus 2. The latter had two haplogroups, hereinafter referred to as S. aff. guttatus 2a and S. aff. guttatus 2b in the manuscript. Scomberomorus aff. guttatus 2a was represented by a larger number of sequences, while the other had only a limited number. Sequence analysis indicated a restricted distribution of S. guttatus in the Bay of Bengal, Bangladesh, and Myanmar. Scomberomorus aff. guttatus 1 included individuals from the Bay of Bengal and the western Pacific Ocean (South China Sea) and some from the Arabian Sea, while S. aff. guttatus 2 comes exclusively from the Arabian Sea. Scomberomorus plurilineatus clusters with this cryptic species, while S. koreanus and S. guttatus, clustering around S. plurilineatus, occupy an interchangeable position with respect to BI and IQ-based phylograms. In addition, S. munroi and S. niphonius form sister clade to the cryptic species + S. plurilineatus + S. koreanus + S. guttatus. Subclade II included Scomberomorus queenslandicus and S. commerson, the latter of which was further subdivided into two groups tentatively designated as S. cf. commerson 1 and S. cf. commerson 2 residing mostly in the Pacific and Indian Oceans, respectively.

Major clade II also had two subclades: one subclade included S. cavalla and S. semifasciatus, while the other included S. lineolatus and the rest of the species. Scomberomorus concolor and S. sierra formed a strongly supported group in this, while S. maculatus + S. regalis and S. brasiliensis formed another. The Cyt b gene generated a tree topology (Supplementary Figure 2) that was incongruent with the COI-based phylogram.

The TCS haplotype network based on mutational steps revealed a clear grouping pattern with three major haplogroups in the cryptic complex of S. guttatus (Supplementary Figure 3A) and two haplogroups in S. commerson (Supplementary Figure 4A). The geographical distribution of these haplogroups was also depicted (Supplementary Figures 3B–D, 4B). The analysis showed a clear genetic differentiation between the haplotypes of S. aff. guttatus 1 and S. aff. guttatus 2 from S. guttatus. The first two mentioned were 10 mutational steps apart, and when taken together, they were 61 mutational steps away from the third. Similarly, the two groups in S. commerson were 18 mutational steps away from each other.

The concatenated multilocus phylogeny depicted the relationships between the 10 nominal species (Figure 3). The Bayesian and IQ phylogenetic trees showed the same topology. There were two major clades (BPP-1) and the first one had two strongly supported subclades: one with (S. aff. guttatus 1 + S. aff. guttatus 2) + (S. guttatus + S. koreanus), while the second subclade contained S. lineolatus and S. cf. commerson 2. Major clade II clustered S. cavalla outside the group of S. concolor and S. sierra and another group of S. brasiliensis, S. maculatus, and S. regalis.

The ABGD analysis detected a barcode gap at 1.7% identifying 17 hypothetical species (MOTUs) in the initial partition. The GMYC and mPTP analyses detected 21 and 17 MOTUs, respectively. The results of the species delineation analyses are given in Figure 2. Of the consensus MOTUs (n = 17), 12 nominal species, viz., S. cavalla (Cuvier, 1829); S. commerson (Lacepède, 1800); S. concolor (Lockington, 1879); S. koreanus (Kishinouye, 1915); S. lineolatus (Cuvier, 1829); S. munroi Collette and Russo, 1980; S. plurilineatus Fourmanoir, 1966; S. queenslandicus Munro, 1943; S. brasiliensis Collette, Russo and Zavala-Camin, 1978; S. niphonius (Cuvier, 1832); S. semifasciatus (Macleay, 1883); and S. sierra Jordan and Starks, 1895 were recovered by all the analyses. The two nominal species S. regalis (Bloch, 1793) and S. maculatus (Mitchill, 1815) were identified as a single MOTU in all analyses except GMYC. Interestingly, three consensus MOTUs or hypothetical species could be identified in nominal S. guttatus with all delimitation methods. Apart from S. guttatus from the type locality, two MOTUs, i.e., S. aff. guttatus 1 and S. aff. guttatus 2, have been identified on the Indian coast including the island ecosystems (Andaman and Nicobar Islands). Scomberomorus aff. guttatus 2a had a smaller intercladal divergence (0.59%) with S. aff. guttatus 2b. However, the maximum genetic divergence was found between S. aff. guttatus 1 and S. aff. guttatus 2b. Sequence identity searches based on the BLAST results of various genes and species delimitation analyses indicated that certain COI barcodes of S. guttatus are classified as S. plurilineatus. Some others actually represented S. aff. guttatus 1 and S. aff. guttatus 2, a misidentification primarily due to the deceptive spotting pattern in the S. guttatus complex.

Scomberomorus commerson also branched into two allopatric putative species in all analyses, though GMYC analysis delimited two more entities (Figure 2). Scomberomorus niphonius (Cuvier, 1832) formed a single MOTU except in GMYC. Scomberomorus lineolatus and S. koreanus, previously not represented in nucleotide databases, represented two different MOTUs.

The COI-based genetic distance values between the recognized nominal species and consensus MOTUs containing novel hypothetical species are shown in Tables 6A, B, respectively. The nominal species on analysis showed a maximum intraspecific divergence of 10.60% (S. guttatus), followed by S. commerson (4.43%), while the minimum intraspecific distance was 0.00% in S. plurilineatus and S. lineolatus. The maximum inter-MOTU distance was 12.35% (between S. semifasciatus and S. cavalla), while the minimum distance was 0.44% between S. maculatus and S. regalis. The p-distance values between S. guttatus and S. aff. guttatus was 10.26%–10.41%, while it was 2.28% between S. aff. guttatus 1 and S. aff. guttatus 2. Between S. cf. commerson 1 and S. cf. commerson 2, it was 3.74%.

GMYC analysis of the Cyt b gene amid the limited number of proxies from each geographic region revealed 18 MOTUs representing the species records in GenBank and identified here (Supplementary Table 2; Supplementary Figure 2). Despite the absence of S. plurilineatus, S. queenslandicus, and S. regalis that were present in the COI-based analysis, the Cyt b data revealed two additional GMYC entities representing two other nominal species, viz., S. tritor and S. sinensis. Interestingly, another GMYC species of uncertain taxonomic status represented as S. guttatus in GenBank from Taiwan and as Scomberomorus sp. here was evident in Cyt b-based analyses, while their COI barcodes were not available. The observed p-distance between S. aff. guttatus 1 and S. aff. guttatus 2 was 1.82%, while Scomberomorus sp. showed a divergence of 10.42%–10.55% from above. These three showed a divergence of 10.38% to 12.57% from S. guttatus. The p-distance was 4.31% between S. cf. commerson 1 and S. cf. commerson 2. These values are comparable to the COI-based distances between the above MOTUs. A divergence of 6.61% was observed between Scomberomorus sp. and its nearest neighbor S. koreanus (Supplementary Table 3). Cyt b sequences from S. sierra of the eastern Pacific (JQ434072–JQ434074) were seen to represent S. concolor. Similarly, sequences such as DQ497884 (Taiwan) identified as S. koreanus and AY986968 identified as S. commerson (China) represented S. aff. guttatus 1.

GMYC analysis of the multilocus data revealed 10 nominal species or 11 MOTUs (Figure 3). The genetic distance values between the nominal species and consensus MOTUs are given in Tables 7A, B. Concordant results with the previous two single-locus analyses indicating three MOTUs in nominal S. guttatus were demonstrated here. The maximum intraspecific divergence (12.2%) was observed in the nominal species S. guttatus. The p-distance values between S. guttatus and S. aff. guttatus 1 and S. aff. guttatus 2 were 11.69% and 11.99%, respectively. The p-distance was 2.14% between S. aff. guttatus 1 and S. aff. guttatus 2. The maximum inter-MOTU distance was 13.11% (between S. cavalla and S. cf. commerson 2), while the minimum was 0.60% between S. maculatus + S. regalis and S. brasiliensis. Similar to the COI-based analysis, S. maculatus and S. regalis formed a single GMYC entity here.

According to rjMCMC algorithms 1 and 0, the BP&P analysis considered S. lineolatus, S. guttatus, S. koreanus, and three lineages in S. aff. guttatus as distinct species, with all nodes resolved with a posterior probability of 1.0, that corroborated with the multilocus tree topologies from phylogenetic analyses. The sister relationship of S. guttatus with S. koreanus was recovered here. The status of the species in S. aff. guttatus is restored similarly to all the tree topologies. In contrast to the phylogenetic and other species delineation analyses, BP&P recognized S. aff. guttatus 2a and S. aff. guttatus 2b into two separate species (Supplementary Figure 5).

The topology of the chronogram (Figure 4) mostly corroborated with both single and multilocus phylograms. The crown age of the Scomberomorus genus has been estimated at 57.3 mya in the late Paleocene [highest posterior density (HPD): 56.3–58.4 mya] and diversification in the Eocene. The regalis group species, S. concolor and S. sierra, split from S. cavalla by 46 mya (HPD: 41–49.5 mya), whereas S. concolor and S. sierra recently separated by 8.5 mya (HPD: 7.2–9.9 mya). The probability of diversification of Indo-West Pacific species was predicted in the early Eocene, i.e., 54.4 mya (HPD: 51.9–56.5 mya). The predicted recent common ancestral ages between S. commerson–S. lineolatus and S. munroi and S. niphonius have been estimated to be 45 mya (HPD: 37–45.5 mya) and 41.9 mya (HPD: 38.5–45.6 mya), respectively. The crown age of S. guttatus–S. koreanus species has been estimated at 24.9 mya (HPD: 22.5–27.5 mya), while S. koreanus split from S. aff. guttatus group by 20.2 mya (HPD: 17.8–22.8 mya). Split in S. aff. guttatus into three lineages was observed at 3.7 mya (HPD: 3–4.5 mya) and 0.91 mya (HPD: 0.6–1.2 mya).

Vertebral counts and X-radiographs of all available species from Indian waters are provided for comparison (Supplementary Table 4; Figure 5). Scomberomorus koreanus, S. commerson, and S. lineolatus showed a total vertebral count of 46, 44, and 46, respectively. The vertebral count of S. guttatus was 47–49 (usually 48), while it was 49–51 (usually 50) for S. aff. guttatus 1 and S. aff. guttatus 2 (Supplementary Figure 6).

COI and Cyt b fragments generated incongruent gene trees in this study, which is not unexpected in Scombridae, where the phylogenetic signal varies with genes (Miya et al., 2013). The low BPP support for the subclade containing S. queenslandicus and S. commerson to the major clade I in the COI-based phylogram indicates their rather fragile affiliation within this major clade. Similarly, different levels of resolution for the species–level relationships could be observed in numerous groups of fish in individual gene trees (Clements et al., 2003; Wright et al., 2012). The incongruence in gene trees arises due to differences in their evolutionary history and various other events like incomplete lineage sorting, gene duplications, extinctions, introgression or hybridization, etc. (Som, 2015). MtDNA introgression occurs between species in Scomberomorus (e.g., Morgan et al., 2013), and the possibility of incomplete lineage sorting in the youngest lineages in Scombridae has been suggested (Ollé et al., 2022).

Collette and Russo (1984), based on a cladistic analysis of 58 characters, identified six species groups in Scomberomorus, viz., the “sinensis group,” which contains only S. sinensis; the “commerson species group” with S. commerson, S. niphonius, S. queenslandicus, and S. cavalla; the monotypic “munroi species group”; the “semifasciatus species group” with S. semifasciatus, S. plurilineatus, and S. lineolatus; the “guttatus species group” with S. guttatus, S. multiradiatus, and S. koreanus; and the “regalis species group” with S. regalis, S. tritor, S. maculatus, S. concolor, S. sierra, and S. brasiliensis. They considered S. tritor to be the basal primitive species in the regalis species group and considered S. sierra, S. brasiliensis, and S. regalis to be the three most advanced species.

The “guttatus species complex” in the COI-based phylogenetic tree contained S. aff. guttatus 1, S. aff. guttatus 2 (2a and 2b), S. koreanus, S. guttatus, and S. plurilineatus, and the grouping of the last one to this complex deviated from the “guttatus species group” proposed by Collette and Russo (1984). The TCS network showed consistent genealogical relationships. A similar genetic structure consistent with the phylogenies has been observed in the TCS network topology of the cryptic S. guttatus complex and S. commerson. Overall, S. aff. guttatus has two distinct haplogroups with a star-like topology with the predominance of a single haplotype, suggesting recolonization events through founder effects or population expansions after a bottleneck event (Allcock and Strugnell, 2012). A similar situation was observed in S. commerson. A star pattern with radial distribution was also shared in S. guttatus. The formation of haplogroups in S. aff. guttatus 2 can be an outcome of incomplete lineage sorting due to recent divergence (Van Velzen et al., 2012).

In the coalescent species tree, the clustering of S. munroi with S. niphonius and the placement of the new world taxa (Banford et al., 1999) in a highly supported clade with S. cavalla outgroup agree with Santini et al. (2013). Species–level relationships deduced from molecular data were generally inconsistent with the previously published morphological groupings by Collette and Russo (1984).

Phylogenies illustrate the historical relationships within a group of species and help study the processes of cladogenesis, and the shape of a phylogenetic tree provides clues to patterns of speciation and extinction within the clade. Taxon sampling has a strong impact on the accuracy of phylogenetic reconstruction methods, and reduced taxon sampling leads to an increase in phylogenetic tree imbalance (Heath et al., 2008; Nabhan and Sarkar, 2012). Increased sampling of taxa allows more accuracy, as exemplified in Salmonidae (Crête-Lafrenière et al., 2012). It is expected that the strength of the phylogenetic signal inferred from the concatenated gene sequences will be useful as the number of taxa increases, and therefore, we recommend global sampling and inclusion of all available species to elucidate the systematic relationships in the genus Scomberomorus.

Through delimitation analyses of COI and Cyt b gene fragments, we confirmed the identity of all valid species except S. maculatus and S. regalis. The latter two formed a single MOTU, an observation which is not surprising in light of the previous reports pointing to mitochondrial introgression in Scomberomorus (Banford et al., 1999; Paine et al., 2007; Morgan et al., 2013) and other scombrids (Puncher et al., 2015). Banford et al. (1999) observed that these two readily distinguishable inhabitants of the Western Atlantic are seasonally sympatric, and hypothesized the probability of ephemeral hybridization and subsequent mitochondrial introgression and recommended a comprehensive geographic and molecular assessment. Although the single-locus species delimitation approach can sometimes lead to an over- and/or underestimation of species diversity (Hofmann et al., 2019), its authenticity was confirmed here with the concordant result of the GMYC analysis of multilocus data.

The incongruence in the number of MOTUs from various molecular species delimitation methods can be due to their difference in a number of aspects. ABGD, GMYC, and PTP were developed for the analysis of single-locus data, although the latter two are often applied to concatenations of multilocus data. ABGD is sensitive to singleton sequences, while GMYC showed tendencies for oversplitting the number of species when there are violations of the model (Pentinsaari et al., 2017). Moreover, PTP and GMYC are sensitive to gene flow (Luo et al., 2018). Here, we used simple rules to find consensus between discordant MOTU delimitations using the best fit between species and MOTUs.

The Indo-Pacific king mackerel, S. guttatus, is dominant next to S. commerson in FAO Fishing Area 51 (Siddeek, 1995; Devaraj and Mohamad Kasim, 1998; Secretariat, IOTC, 2015). It is noteworthy that the nominal S. guttatus showed high intraspecific divergence, an indication of cryptic speciation events (Ge et al., 2021), and contained three consensus MOTUs with morphological similitude in all analyses except in BP&P, which produced four Bayesian species. BP&P is believed to be more accurate than other multilocus coalescent methods, showing lower rates of species overestimation and underestimation, and is generally robust to various potential confounders (Luo et al., 2018). Nevertheless, this method tends to delimit only the population structure and not the number of species, and hence, validation of genome-based results with multiple data types such as phenotypic and ecological information is required (Sukumaran and Knowles, 2017).

The species distribution along the Indian coast derived from our extensive sampling is depicted in Figure 6. Analysis of specimens from the type locality of S. guttatus and regions up to Chennai (Cuddalore, Kovalam, Nagapattinam, Kovalam) indicated the occurrence of only one haplogroup from these areas on the Coromandel Coast. The same formed a fraction of samples from other sites of fishery (Chennai, Visakhapatnam, Odisha, and West Bengal) located in the Bay of Bengal. The species having a genetic identity with the above deserves the status to be adorned as the original S. guttatus identified by Bloch and Schneider in (1801), as it is the only available species in type locality. It differed from the other two MOTUs in the nominal species by high genetic divergence. The lean and slightly cylindrical body shape of this species resembles the only existing type specimen collected from Tranquebar, India (Figure 7A: ZMB 8759: lectotype, stuffed, 361 mm SL) although its meristic trait such as vertebrae is not visible (Figure 7B). The cryptic MOTUs, viz., S. aff. guttatus 1 and S. aff. guttatus 2, hitherto hidden among the nominal species, were widespread in the collections. The Scomberomorus sp. from Taiwan recovered in Cyt b-based analysis and clades with S. koreanus represent a species with an uncertain taxonomic status that cannot be verified with the current level of information.

The allopatric putative species in S. commerson, i.e., S. cf. commerson 1 and S. cf. commerson 2, were consistent with some previous studies using mtDNA CR (Fauvelot and Borsa, 2011) with extensive geographic coverage across oceans. Scomberomorus commerson has a typically pelagic lifestyle at all developmental stages (Thangaraja and Al-Aisry, 2001) and is considered to be more migratory than all other congeners (Collette, 2001). Incidental occurrence of eggs/larvae of the Pacific Ocean lineage in the Indian Ocean (Supplementary Table 1) supports this observation. Significant differences in morphometry and spawning season between the Indo-Pacific populations were observed by Collette and Russo (1984) and Kaymaram (2010), respectively. Fauvelot and Borsa (2011) stated that the samples from the northwestern Indian Ocean, consisting of the Persian Gulf and the Oman Sea (identical to S. cf. commerson 2) and those from the Western Pacific, i.e., Southeast Asia and Oceania (S. cf. commerson 1), represent two cryptic species with a deep phylogeographic split. The additional GMYC entities observed in this study also support previous observations of geographic differentiation in this species (e.g., Sulaiman and Ovenden, 2010; Vineesh et al., 2018).

However, S. lineolatus, another valid species which is least abundant in the seerfishes in India (CMFRI FRAD, 2020), showed a limited distribution in Pamban up to Tuticorin (personal observation of the authors), although the literature describes its occurrence from Mumbai, Malabar to Palk Bay and Chennai (Collette and Russo, 1984). The shift in the fishery for S. guttatus and S. lineolatus from their former areas of frequent occurrence (Southern Gulf of Mannar and Visakhapatnam, respectively) is reminiscent of a similar trend observed for S. concolor (Fitch and Flesching, 1949). It is interesting to note the restricted distribution pattern of the nominal S. guttatus and S. koreanus along the Indian coast with an apparent break in the marine regions adjacent to peninsular India, the reasons for which will be discussed in the upcoming session.

Although the phylogenetic investigation in the regalis species group has been performed, a comprehensive investigation on divergence time estimation for the Scomberomorus genus is yet to be undertaken. To procure the most reliable information from fossil-calibrated molecular data, molecular dating was carried out with species having at least 70% of the genes selected for the analysis. The results derived from our study differ from all previous investigations (Banford et al., 1999; Santini et al., 2013). The first split from the most recent common ancestor of Scomberomorus divided the Western Atlantic (S. cavalla)–Eastern Pacific species (S. concolor and S. sierra) with the Indo-West Pacific species group (S. munroi, S. niphonius, S. semifasciatus, S. commerson, S. lineolatus, S. guttatus, S. koreanus, and S. aff. guttatus). The recent split of the regalis species group suggested by previous nuclear sequence data (Banford et al., 1999) was also supported in this study, though the divergence time varies, which can be rectified if more species from the regalis group are added for divergence dating. The Indo-West Pacific species group has undergone multiple diversification events; of these events, the most interesting divergence was perceived in the crown group, comprising S. guttatus–S. koreanus and encircling cryptic species. Scomberomorus koreanus and S. aff. guttatus diverged from the common ancestor (S. guttatus) during the Miocene followed by population expansion of S. aff. guttatus during the Pliocene. Both divergence and expansion of the latter were probably materialized due to climatic or oceanographic changes that occurred during the Miocene–Pliocene periods that might have changed the population connectivity leading to range expansions (Renema et al., 2008). The early Miocene period is characterized by the redistribution of salt and heat resulting in significant changes in flow and climate patterns in the Indian Ocean (Bialik et al, 2019). Furthermore, the diversification of S. aff. guttatus into three lineages in the Pliocene could be accelerated by the increase in productivity associated with the biogenic bloom, and shifts in monsoon activity occurred in the Miocene period in the Indian Ocean (Karatsolis et al., 2022). The superimposed climatic changes with local adaptations followed by reproductive aggregation possibly led to sympatric speciation in this genus (Qvarnström et al., 2016; Beal et al., 2020).

The variations in vertebral counts, both in the precaudal and caudal regions, are observed in a number of species in Scomberomorus. The vertebral count observed for S. commerson, S. lineolatus, and S. koreanus agrees with previous observations on these species (Collette and Russo, 1984). The wide range (47–51) in nominal S. guttatus is discussed here. Jones and Silas (1961) observed 48 or 49 vertebrae in S. guttatus examined from the peninsular (Vizhinjam) and south-east coast (Tuticorin) of India. They indicated that Kishinouye (1923) observed the vertebral count of 51 for Cybium guttatum from Formosan waters and observed an insightful range of variation of this meristic character in S. guttatus. Subsequently, in 1964, Silas redescribed the species from specimens collected in the southern part of the Gulf of Mannar (Tuticorin, India). Later, Devaraj (1976) observed vertebral counts of 49–50 (mostly 50) in samples collected from the northern part of the Gulf of Mannar, which he considered to be the typical S. guttatus. He suggested that the S. guttatus from the southern Gulf of Mannar reported by Silas (1964), as well as the specimens of the contiguous Arabian Sea, characterized by shortening of the precaudal region and elongation of the caudal region, may belong to a different race suspected to be S. guttatus. It should be noted that the occurrence of S. guttatus in the southern Gulf of Mannar and Pamban is currently very rare and the authors were only able to collect very few specimens during the period of this study (2019–2021), contrary to previous observations (Jones and Silas, 1961; Devaraj, 1976; Devaraj, 1998). However, the species shares the fishery with S. commerson in the northern part of the Gulf (from Tranquebar to Nagapattinam), from where the authors were able to collect only one species with 47–49 (usually 48) vertebrae.

The cryptic species S. aff. guttatus in the Indian Ocean has 49–51 vertebrae (usually 50), while its vertebral count in the Western Pacific is usually 51 (Lee and Yang, 1983). Collette and Russo (1984) reported that the Indian Ocean population has a mode of 50 vertebrae, while populations in the East Indies, Gulf of Thailand, and China have 51 as mode. We examined the X-rays of lectotype (MNHN A.5771, Java, 1,075 mm FL, vertebrae 51) and paralectotype (MNHN A. 5715, Mumbai, 1,145 mm FL, vertebrae 50) of Cybium kuhlii (http://n2t.net/ark:/65665/3b5aba155-8948-4776-b613-ff753884c14d and http://n2t.net/ark:/65665/3cf8430ff-c20c-4d82-9424-cf17063a007b), currently synonymized with S. guttatus, to confirm our observations. The lectotype of C. kuhlii from Java and the paralectotype from Mumbai represent the Pacific and Indian Ocean lineages, respectively.

The number of vertebrae is often cited as a meristic character of diagnostic importance in fish taxonomy. The count sometimes differs between populations of a species along with its geographical range, or across the ranges of related species (e.g., Roul et al., 2021), similar to the observations here. Parallel differences in vertebral counts can also occur within the species (Hubbs, 1922). The vertebral number is influenced by a number of factors like ambient temperatures during the ontogeny of individual fish (Jordan, 1892), inheritance, etc., and the separation of environmental and genetic influences is a complex process (McDowall, 2008). Therefore, this character can be considered only in combination with other features for diagnostic purposes.

The historical provenance of Spanish mackerels from the Indian subcontinent begins with S. guttatus, originally described as Scomber guttatus by Bloch and Schneider (1801) from Tranquebar (Tharangambadi), Tamil Nadu, India. Two years later, Russell (1803) described the seerfishes vernacularly known as “Wingeram” and “Konam” from Vizagapatam (present-day Visakhapatnam) on the Coromandel coast of India as Scomber Wingeram and Scomber Konam. They added that the latter was rarer and of inferior quality compared with the English seerfish (Wingeram) of the same size. Later that year, Shaw (1803) from the same locality, based on Russell’s description and figures of Wingeram and Konam, described them as Scomber leopardus and Scomber maculosus, respectively. Cuvier then described in Cuvier and Valenciennes, 1831, Cybium lineolatum from Malabar, Cybium interruptum from Pondicherry (Puducherry), C. kuhlii from Bombay, and C. commerson from Pondicherry and Malabar in India. Day (1878) also reported five species of seerfishes from the Indian Seas under the genus Cybium, viz., C. kuhlii, C. interruptum, C. guttatum, C. commersonii, and C. lineolatum. Jones and Silas (1961), in examining their taxonomy from India, refined all to three valid species, S. guttatus, S. lineolatus, and S. commerson. They synonymized Day’s C. kuhlii and C. guttatum with S. guttatus, C. interruptum and C. lineolatum with S. lineolatus, and C. commersonii with S. commerson. Because the original description of Scomber guttatus in Bloch and Schneider (1801) is sparse, Silas (1964) redescribed S. guttatus. He considered “Wingeram” defined by Russell (1803), later described as Scomber leopardus by Shaw (1803), identical to S. guttatus. Considering the main characters, he also considered Cybium koreanum Kishinouye, 1915 as a subspecies of S. guttatus and divided the latter into two subspecies, viz., S. guttatus guttatus (Bloch and Schneider, 1801) and S. guttatus koreanus (Kishinouye, 1915). However, Devaraj (1976) proved that S. koreanus is distinct from S. guttatus and not a subspecies of the latter, unlike Silas (1964).

Scomberomorus aff. guttatus in the cryptic complex with two MOTUs showed a p-distance of 10.26%–10.4% with S. guttatus. There exists a divergence of more than 2% between the two MOTUs, which is widely accepted as the heuristic limit for species delineation using COI barcodes (Pereira et al., 2013). Although geographical populations of conspecifics can exhibit genetic distances in excess of 2% (Ward et al., 2005), the likelihood that these MOTUs represent two potentially cryptic species of recent origin cannot be ignored. Albeit, their divergence is not reflected in the aldolase exon 5/intron 5 regions, although the distance estimates of this nuclear gene are high among closely related taxa in the S. regalis complex that have been dated to a recent origin (Banford et al., 1999). This is due to insufficient time for the accumulation of mutations in nuclear markers (Allio et al., 2017). Microsatellite-based analysis of S. guttatus revealed a unit stock from the Persian Gulf (Abedi et al., 2011), and mtDNA sequences in GenBank from this region were identical to S. aff. guttatus 2.

Scomberomorus guttatus is characterized by the fine auxiliary branches radiating on either side of the anterior third of the lateral line, intestine with 2 folds and 3 limbs, and 47–52 vertebrae (Collette and Nauen, 1983). The first two features are common to S. guttatus and the cryptic species identified in this study, while differing in their morphometry and modes of vertebral counts. The occurrence of doppelganger in the S. guttatus species complex which is too cryptic to be identified by external morphology and its geographic populations also add to the difference in vertebral numbers noted by previous researchers. However, it can be observed that typical S. guttatus has a lesser body depth compared to the above. Their cryptic nature can be attributed to their morphological convergence in response to analogous selective regimes (Ingram and Mahler, 2013) which are discussed subsequently. Hence, based on the genetic results of specimens from type locality, vertebral counts of 47–49 (usually 48), and examination of lectotype (ZMB 8759; Figure 7), the identity of S. guttatus (Bloch and Schneider, 1801) is confirmed (Figure 8A) and its morphological redescription is strongly recommended.

It may be added that Shaw (1803) described Scomber leopardus, which was later synonymized with S. guttatus (Bloch and Schneider, 1801), based on the image and the original description of “Scomber Wingeram” provided by Russell (1803) from Visakhapatnam on the coast of Coromandel. Our study showed the occurrence of both species on the Coromandel coast, but since the type locality contains only S. guttatus which is less abundant in Visakhapatnam, the available senior synonym Scomber leopardus Shaw, 1803 (= Scomberomorus leopardus) merits the distinction to be accepted for the cryptic species. Based on the barcode criterion of mtDNA clustering and critical examination of Russell’s drawing (33.2 cm long), we here resurrect Scomberomorus leopardus (Shaw, 1803) (Figures 8B, C; Supplementary Figure 7) from the synonymy of S. guttatus as a valid species, which needs to be redescribed. The vertebral counts of this species have a mode of 50 for the Indian Ocean lineage, while it is reported to be 51 for the Western Pacific lineage.

It can be added that some junior synonyms, viz., Cybium interruptum and C. kuhlii, are also available in the literature (synonymized with S. guttatus) originally described by Cuvier and Valenciennes (1831) from Pondicherry and Bombay, respectively. Although C. interruptum was successively included in the same manuscript by Cuvier and Valenciennes (1831), we believe that C. interruptum cannot be applied to S. aff. guttatus 1 due to the availability of a senior synonym, Scomber leopardus Shaw, 1803 from the same geographic region. If the Arabian Sea counterpart of S. leopardus (i.e., S. aff. guttatus 2; Figure 8C) and the counterpart from the Bay of Bengal and Western Pacific (S. aff. guttatus 1; Figure 8B) turn out to be distinct on further analysis, the senior name Cybium kuhlii [= Scomberomorus kuhlii (Cuvier in Cuvier and Valenciennes, 1831)] will be applicable to the former. S. leopardus can be retained for the latter since its genesis lies in the Bay of Bengal region. In general, naming/reviving a species generally requires multiple lines of evidence along with the “Generalized Lineage Species Concept” (Coates et al., 2018), and therefore, a detailed study on the apparent phenotypic differentiation in the cryptic species and the redescription of S. guttatus are in progress.

Similarly, for the Indian Ocean complement of S. commerson (i.e., S. cf. commerson 2) which has been shown to be distinct on the basis of detailed molecular analysis, the name Scomber maculosus Shaw, 1803 (= Scomberomorus maculosus) described by Shaw (1803), based on the Konam of Russell (1803), will probably be applicable. However, in the absence of a type specimen of S. commerson, originally described as Scomber commerson after a figure from Commerson’s manuscripts, and the absence of any indication of type locality in the original description, the identity of the species can only be established by designation of a type based on the Rules of the International Code of Zoological Nomenclature.

The S. guttatus species group contains S. guttatus, S. koreanus, and S. multiradiatus (Collette and Russo, 1984). The first two are available in the northern Indian Ocean. Scomberomorus guttatus, a neritic species thought to be less migratory than S. commerson, inhabits the coastal waters at depths between 15 and 200 m and is generally encountered in turbid waters with lower salinity (Fischer and Whitehead, 1974; Collette, 2001) which corroborate with the collection data from this study (Table 2). The habitat of S. guttatus is captured from river mouths and estuaries in Indonesia (Hardenberg, 1936), West Bengal in India (Hora, 1953; Dutta et al., 2021), etc., indicating their tendency to ascend estuaries of large rivers. Similarly, S. koreanus exhibited a remarkable ability to ascend brackish parts of rivers (Kishinouye, 1923). Species range boundaries indicate a species’ ecological niche in space (Sexton et al., 2009), and species morphology is partially related to its ecological niche (Schluter, 2000).

Habitat use, diet, and environmental tolerance are usually measured niche subsets (Slatyer et al., 2013). Selection on habitat preferences can be a strong reproductive barrier that promotes divergence with gene flow (Berner and Thibert-Plante, 2015). The niche breadth and size of the geographical range of S. guttatus is of lesser magnitude in India limited to the east coast, with more aggregations in Tamil Nadu and West Bengal and also in Bangladesh, while the niche expanse is wider for S. leopardus. The range of all Scomberomorus spp. in India is shown in Figure 6. The west coast of India is a coast of submergence (except for the Malabar Coast; Figure 1) with a steep slope and narrow width with only a few west-flowing rivers, forming estuaries while draining into the Arabian Sea. In contrast, the east coast is an emergent coast (Swathi Lekshmi, 2020) with the Bay of Bengal receiving discharge from many east-flowing rivers, fostering larger estuaries and major deltas. The estuaries on the west coast lie mainly between Gujarat and Karnataka, while the extensive east coast estuarine system is in the stretch from West Bengal to Tamil Nadu (Faruque and Ramachandran, 2014a; Faruque et al., 2014b) (Figure 6). The intensity of the seasonal monsoons and the precipitation patterns show a great difference in the Indian subcontinent, and the latter along with other factors creates fluctuations in salinity by freshwater runoff and determines the hydrography of the aforementioned monsoonal estuaries. Tamil Nadu receives more annual precipitation during the winter monsoon (October to November) than most other parts of the country which get it during the southwest monsoon (June–September) (Venkataramana et al., 2021). The Ganges/Brahmaputra and Irrawaddy/Salween River systems transport huge amounts of water and sediment to the northeastern Indian Ocean during and after the rainy season (June to August), resulting in the formation of freshwater pools and seasonal salinity differences (Jöhnck et al., 2020). Deltas and estuarine systems formed by the river Ganges in the northern part of the Bay of Bengal (Gangetic delta in Bangladesh) and the Southern Bay of Bengal (Hoogly-Matlah estuary forming the Sunderbans delta in India) form a highly productive marine ecosystem (Mukhopadhyay et al., 2006) that maintains a salinity gradient due to seasonal freshwater discharge and regular tidal influxes. Scomberomorus guttatus, S. leopardus, and S. koreanus have a preference for coastal and estuarine habitats which results in their adherence to these areas.

As Jordan (1922) pointed out in one of his ecogeographical rules, the nearest relative of any given form often occurs in adjacent areas usually across some barrier to distribution. This holds true for the S. guttatus complex. This occurrence represents an early declaration of the concept of allopatric speciation or perhaps vicariance (McDowall, 2008). The peninsular region of India can act as an apparent geographical barrier due to its steep slope and non-availability of suitable habitats for larval settlement. Besides, Scomberomorus spp. have pelagic eggs and larvae (Vijayaraghavan, 1955; Jones, 1961; Collette, 1986; da Cunha et al., 2020). Although information on larval dispersal with monsoon currents is sparse for the S. guttatus complex, the range overlap between the MOTUs of the lineages in S. aff. guttatus (= S. leopardus) near Mangaluru (southern Karnataka) indicates that the larvae entering the currents and circulations in the southern peninsula of the Indian subcontinent (Rao and Sreenivas, 2020) settle only in their most appropriate niches.

Recruitment and population dynamics of fish are mainly driven by their feeding ecology and dietary plasticity (Sui et al., 2021). Vijayaraghavan (1955) suggested that the dietary component of S. guttatus changes with habitat and season. The gut content analysis of S. guttatus from the north-eastern Arabian Sea showed that their major food is Acetes sp. followed by clupeids (Chellappan et al., 2018). This area is rich in extremely euryhaline Acetes spp., which provide food for many predators (Deshmukh, 2007; Metillo et al., 2015). This contradicts the feeding pattern observed on the south and south-east coasts of India, where the prey is Sardinella spp. and whitebaits (Devaraj, 1998). Piscine prey abundance is expected to affect larval survival and recruitment of Scomberomorus, as illustrated by the larval survival strategy with an early switching of diet from planktivory to piscivory, which reduces the duration of the larval stage (Shoji et al., 2005). Chale-Matsau (1996) posited that the abundance of S. plurilineatus on the KwaZulu-Natal coastline was related to the appearance of the inshore marine prey species Sardinops sagax along the coast rather than to spawning. It can be observed that the highly productive areas along the Indian coast where S. guttatus spp. inhabit provide crustacean and piscine preys that can aid in their larval recruitment.

The spawning season observed for S. guttatus on the south-east coast of India is April to July with a spawning peak in July (Krishnamoorthi, 1958), while Devaraj et al. (1987) suggested an extended spawning season from January to August with a peak from April to May. Collette and Nauen (1983) indicated it from April to July around Rameswaram Island between India and Sri Lanka. On the east coast of India (Andhra Pradesh), the peak spawning period is between November and April (Uma Mahesh et al., 2017), while on the west coast along the northern Arabian Sea, it is May (Chellappan et al., 2018). Season is longer with a peak in January to August in Indonesian waters (Noegroho et al., 2018), but the peak recruitment is from May to July in Bangladesh (Rashid et al., 2010). The difference in spawning times in S. guttatus can be due to differences in biotic and abiotic parameters in the area (Roa-Ureta et al., 2019) coupled with life-history traits of the nominal S. guttatus which is now revealed to contain cryptic species.

Fidelity to spawning sites (probably upwelling areas) rather than adaptations to oceanographic regimes has been postulated as one of the reasons for the incidence of genetically discrete groups in S. sierra (López et al., 2010). In the current context of limited information on spawning grounds along the coastline, except for the reports by Devaraj (1987) and Jones (1961), which were limited to the south-east and south-west coasts of India respectively, the above postulate cannot be verified. Recent studies evidence that marine species with pelagic larval stages are subjected to high levels of local recruitment, which can lead to reduced gene flow, niche specialization through local adaptation, and sexual selection, leading to divergence and speciation (Milá et al., 2017). Fišer et al. (2018) pointed out that cryptic species show differences in their requirements along their Grinnellian niche. Speciation is a protracted process and a continuum of reproductive isolation (Stankowski and Ravinet, 2021), which are effects from the chance fixation of diverse, epistatic incompatibilities in discrete populations subjected to identical selective pressures (Rosser et al., 2019). Though many factors point to the probability that the distribution pattern of S. guttatus species complex in India is more connected with ecological niche and feeding regime, further studies are recommended to establish the relevant eco–evo interactions (Lowe et al., 2017).