Revisiting bioluminescence and sucrose utilization in aquatic pathogens Vibrio harveyi and V. campbellii using genome-wide in silico mapping and phenotyping

Abstract

Vibrio harveyi is a major bacterial pathogen of shrimp and finfish aquaculture. Traditionally, bioluminescence and sucrose fermentation have served as key phenotypic marker for its identification. However, frequent misidentification with closely related species like V. campbellii necessitates a reassessment of these phenotypic traits. Therefore, these traits were evaluated for genomic distribution, targeted phenotypic validation and its potential role in evolution and speciation. We generated chromosome-level assemblies for seven strains, including V. harveyi SB1 reference genome, and performed genome-wide mapping of 282 strains (204 V. harveyi and 78 V. campbellii), followed by phenotypic validation of 49 isolates. In silico analysis revealed that only 2.9% of V. harveyi strains carry luminescence operon (luxCDABEGH), whereas 100% strains of V. campbellii carried either a functional luxCDABEGH (87.2%) or a defective luxBG operon (12.8%). The functional sucrose operon (scrRAKB) was present in 89.5% strains of V. harveyi (yellow colonies on TCBS agar) but was absent in all V. campbellii (green colony) except strain 170502. Mobilome and synteny analysis revealed horizontal gene transfer of scr operon in 1% strains, while no mobile genetic elements were associated with the luxCDABEGH operon in V. harveyi, despite rare occurrence. Core genome phylogeny indicated that V. harveyi represents an early-evolved lineage, whereas V. campbellii is a recently evolved species within the Harveyi clade. The evolutionary trajectory of V. campbellii further suggests that luminescence-defective strains (e.g., type strain CAIM519^T) evolved alongside a group of strains carrying luminescence operon flanked by mobile-genetic elements (e.g., BAA-1116). Phenotypic assays and PCR screening of the luciferase gene (luxA) and sucrose uptake gene (scrA) results were consistent with the genomic findings. Collectively, the present study demonstrates that V. harveyi is predominantly non-luminescent and sucrose-fermenting (yellow), while V. campbellii is primarily luminescent and sucrose non-fermenting (green colonies), providing refined phenotypic criteria for their differential diagnosis.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-37651-3.

Introduction

Vibrio harveyi and V. campbellii are major aquatic pathogens causing significant economic losses in aquaculture. They have been associated with luminescent vibriosis in shrimp hatcheries, where they can cause up to 100% mortality in mysis and early postlarval stages^1–3. They have also been linked to acute hepatopancreatic necrosis disease (AHPND) in shrimp^4–7. Beyond crustaceans, V. harveyi has also been implicated with vibriosis in several marine farmed fishes, including Asian seabass (Lates calcarifer)^8,9, European seabass (Dicentrarchus labrax)¹⁰, gilthead sea bream (Sparus aurata)^10,11, rainbow trout (Oncorhynchus mykiss)¹², Atlantic salmon (Salmo salar)¹², grey mullet (Mugil cephalus)¹³ and milk fish (Chanos chanos)¹⁴. Additionally, V. harveyi served as model organism for advancing the understanding of quorum sensing, a cell-density dependent regulatory mechanism controlling community behaviors such as bioluminescence¹⁵.

Despite great pathogenic significance in aquaculture and serving as a model organism in microbial research, V. harveyi and V. campbellii have been frequently misidentified. For example, comparative genomics study by Lin et al.¹⁶ revealed that 28% strains within the Harveyi clade were misidentified. Notably, V. harveyi BAA-1116, a model strain widely used in quorum sensing and luminescence studies, was reclassified as V. campbellii¹⁶. Studies from Indian shrimp farms and hatcheries revealed that 93 to 100% suspected V. harveyi strains were, in fact, V. campbellii^2,17. This taxonomic confusion largely arises from the poor discriminatory power of the 16S rRNA gene, which share > 99% sequence identity between these species, well above the 97% threshold of species boundary¹⁶ as well as from incomplete understanding of their differential phenotypic traits. Over the past two decades, multilocus sequence analysis^18,19 and genome-based phylogeny²⁰ have markedly improved species resolution within the Harveyi clade, providing far superior discriminatory power compared to 16S rRNA based approaches. Species-specific PCR assays^21,22have also been developed for rapid detection following culture on Vibrio-selective media. However, a gap remains between these molecular tools and traditional phenotypic markers such as bioluminescence and sucrose fermentation that continues to be widely used in field diagnostics due to their simplicity and ease of use.

Bioluminescence, encoded by luxCDABEGH operon, is one of the most widely cited phenotypic traits of V. harveyi^1,15,23–25. The enzyme luciferase (LuxAB) generates light by oxidizing FMNH₂ and a long-chain fatty aldehyde, provided by LuxGH and LuxCDE, respectively^26–28, and is regulated by quorum sensing networks^29,30. Another widely used diagnostic test is sucrose fermentation on thiosulfate citrate bile salts sucrose (TCBS) agar, where sucrose fermenters produce yellow colonies and non-fermenters form green colonies^31,32. In Gram-negative bacteria, sucrose utilization is encoded by the scrRAKB operon³³. This encodes for sucrose-specific phosphotransferase system (PTS) ScrA, sucrose-6-phosphate hydrolase, ScrB, fructokinase ScrK, and the regulatory protein ScrR, and has been functionally characterized in V. alginolyticus³⁴.

There are conflicting reports with respect to colony characteristics on TCBS agar (green/yellow) and bioluminescence in V. harveyi. Many studies describe V. harveyi as non-sucrose fermenter (green colony on TCBS) and predominantly luminescent^24,35,36. Conversely, other studies report it as a sucrose-fermenter (yellow colonies) and majority of strains being non-luminescent^37,38. Farm-level interpretations have further compounded this confusion, where the dominance of yellow colonies on TCBS agar is often taken as an indicator of “healthy flora,” while green colonies are regarded as problematic³⁹. These inconsistencies, against a backdrop of widespread species misidentification, necessitate a systematic re-evaluation of these phenotypic markers. The availability of large number of high-quality bacterial genome sequences allows robust phylogenomic classification using average nucleotide identity (ANI)^40,41 and core genome phylogeny⁴². Once species identity is reliably established, genotype–phenotype associations⁴³ such as bioluminescence and sucrose utilization can be reassessed with accuracy. Accordingly, this study first generated chromosome-level assemblies for six V. campbellii strains and one V. harveyi strain. We then delineated species boundaries and evolutionary pattern across 315 publicly available genomes of Harveyi clade, including 288 strains of V. harveyi and V. campbellii, using ANI and core genome phylogeny. Further, we analysed the distribution of the luxCDABEGH (bioluminescence) and scrRAKB (sucrose utilization) operons across 282 confidently identified genomes (204 V. harveyi and 78 V. campbellii), supported by phenotypic validation in 49 representative isolates. Finally, the study investigated the potential role of horizontal gene transfer in atypical operon patterns by analysing associated mobile genetic elements.

Results

Bacterial strain profile for insilico and phenotypic study

A total of 315 genome assemblies from NCBI database were selected for pangenome, core-genome based phylogeny and average nucleotide identity (Supplementary Table S1). These strains were collected during 1935 to 2024. V. harveyi was isolated from diverse range of sources, including 147 strains from 43 different aquatic animals, 30 strains from human, and 20 from environmental sources. In contrast, V. campbellii was mostly recovered from seawater (35 strains) and aquatic animals, primarily shrimp (36 strains). Overall, 75 strains had chromosome-level genome assemblies, eight strains were assembled up to 5 contigs/scaffolds, while the remaining strains had contigs/scaffold ranging from 13 to 921. A highly curated secondary dataset was used for evolutionary and speciation analysis, comprising 82 genomes; 70 complete genomes, six strains assembled up to 5 contigs and remaining 6 with contig numbers ranging from 52 to 93 (Supplementary Table S1).

Forty-nine bacterial isolates were subjected to phenotypic characterization and PCR screening and is presented in Table 1. Nucleotide BLAST analysis of rpoD gene identified 18 isolates as V. harveyi and 31 isolates as V. campbellii. The V. harveyi isolates showed the highest sequence identity with the type strain ATCC14126, while the V. campbellii isolates were most similar to V. campbellii CAIM 519ᵀ, and each group clustered phylogenetically with its respective reference strain (Supplementary Fig. S1). This was further confirmed by species-specific multiplex PCR using the hemolysin. Eighteen isolates of V. harveyi produced the desirable 454 bp amplicon while 31 V. campbellii isolates had 328 bp amplicon (Supplementary Fig. S2).

Table 1.

Bacterial isolates used for phenotypic characterization and validation of in silico results on luminescence and sucrose fermentation.

Sl.No	Species	Isolate name	Details of isolation	Disease	Year of isolation	Accession number for rpoD
1	V. campbellii	LB1	P. monodon brooder gills	Luminescent vibriosis	2006	MW428977
2	V. campbellii	LB3	P. monodon spawning tank water	Luminescent vibriosis	2006	MW428978
3	V. campbellii	LB10	P. monodon brooder surface	Luminescent vibriosis	2006	MW428979
4	V. campbellii	LB14	P. monodon maturation tank water	Luminescent vibriosis	2006	MW428980
5	V. campbellii	LB16	Brooder intestine	No	2006	MW428981
6	V. campbellii	LB25	Faecal matter	No	2006	MW428982
7	V. campbellii	LB27	Brooder surface	No	2006	MW428983
8	V. campbellii	LB33	Brooder surface	No	2006	MW428984
9	V. campbellii	LB37	P. monodon Maturation tank water	Luminescent vibriosis	2006	MW428985
10	V. campbellii	LB39	Brooder surface	No	2006	MW428986
11	V. campbellii	LB67	Faecal matter	No	2006	MW428987
12	V. campbellii	LB102	P. monodon Maturation tank water	Luminescent vibriosis	2006	MW428988
13	V. campbellii	LB131	Pond water	No	2007	MW428989
14	V. campbellii	LB135	Pond sediment	No	2007	MW428990
15	V. campbellii	LB157	Pond water	No	2007	MW428991
16	V. campbellii	LB164	Pond water	No	2007	MW428992
17	V. campbellii	LB171	Pond water	No	2007	MW428993
18	V. campbellii	LB178	Pond sediment	No	2007	MW428994
19	V. campbellii	LB186	Shrimp gills	No	2007	MW428995
20	V. campbellii	LB195	P. monodon Postlarvae	Luminescent vibriosis	2007	MW428996
21	V. campbellii	LB198	P. monodon Postlarvae	Luminescent vibriosis	2007	MW428997
22	V. campbellii	LB204	P. monodon Postlarvae	Luminescent vibriosis	2007	MW428998
23	V. campbellii	LB210	P. monodon Postlarvae	Luminescent vibriosis	2007	MW428999
24	V. campbellii	LB217	P. monodon Postlarvae	Luminescent vibriosis	2007	MW429000
25	V. campbellii	LB235	P. monodon Mysis	Luminescent vibriosis	2007	MW429001
26	V. campbellii	LB314	Brooder surface	Luminescent vibriosis	2008	MW429002
27	V. campbellii	LB503	P. monodon Postlarvae	Luminescent vibriosis	2009	MW429003
28	V. campbellii	LB515	P. vannamei postlarvae	Luminescent vibriosis	2018	MW429004
29	V. campbellii	LB516	P. vannamei postlarvae	Luminescent vibriosis	2018	MW429005
30	V. campbellii	BAA- 1116	Seawater	–	1993	JF930357.1 (Procured from ATCC)
31	V. campbellii	LMG 11216 T (CAIM519T)	Seawater	No	1971	A8140_02280 (Procured from LMG)
32	V. harveyi	SB1	Asian seabass	Vibriosis	2020	QMY49_12835
33	V. harveyi	LMG 4044 T (ATCC14126)	Dead Amphipod	Yes	1935	Procured from LMG
34	V. harveyi	LMG7890 (ATCC 35084)	–	–		JF930379.1 (Procured from LMG)
35	V. harveyi	LMG 19643	–	–		(Procured from LMG)
36	V. harveyi	2301	Penaeus vannamei gut	No	2023	PV550895
37	V. harveyi	2426	Hepatopancreas of P. japonicus	Black spot on exoskeleton	2024	PV550896
38	V. harveyi	2427	Hepatopancreas of P. japonicus	Black spot on exoskeleton	2024	PV550897
39	V. harveyi	2432	Gut of P. japonicus	Black spot on exoskeleton	2024	PV550898
40	V. harveyi	2437	Blood of P. japonicus	Loose shell	2024	PX116711
41	V. harveyi	2438	Muscle of P. japonicus	Loose shell	2024	PV550899
42	V. harveyi	2439	Haemolymph of P. japonicas	Loose shell	2024	PV550900
43	V. harveyi	2441	Hepatopancreas of P. japonicus male	Loose shell	2024	PV550901
44	V. harveyi	2452	Haemolymph of P. vannamei	Running mortality	2024	PV550902
45	V. harveyi	Y12-23	Lates calcarifer	-	2023	PV550903
46	V. harveyi	FCD 2025/2	Kidney of Asian seabass brooder affected with disease	Vibriosis affected	2025	PX116707
47	V. harveyi	FCD 2023/4	Asian seabass fry	Vibriosis affected	2023	PX116708
48	V. harveyi	FCD 2023/7	Asian seabass fry	Vibriosis affected	2023	PX116709
49	V. harveyi	Y1/12/23	Asian seabass	–	2023	PX116710

Whole genome sequencing

In the present study, one highly pathogenic strain of V. harveyi (SB1) and 6 representative strains of V. campbellii (LB3, LB10, LB102, LB135, LB198, LB314 and LB503) were sequenced on PacBio RS II and Illumina Novaseq platform. All the strains had chromosome level assembly. V. campbellii strains had two chromosomes while V. harveyi SB1 had two chromosomes and additional 153.4 kb plasmid (Table 2). V. harveyi SB1 has been designated as reference genome at NCBI database, providing key resource for genomic studies. It has the genome size of 5.87 Mb. Comparatively, the genome size of sequenced V. campbellii strains were smaller and ranged between 5.62 Mb (LB198) to 5.74 Mb (LB3).

Table 2.

Whole genome sequencing information and assembly statistics of V. harveyi and V. campbellii strains sequenced in the present study.

	V. harveyi	V. campbellii
	SB1	LB3	LB10	LB135	LB198	LB314	LB503
Sequencing information: PacBio raw reads
No. of Polymerase Reads	138,596	143,067	184,843	123,094	146,450	64,680	157,065
Number of polymerase read base	11,981,660,465	12,143,024,654	15,852,613,494	10,619,531,394	12,622,234,805	5,513,630,052	13,708,080,644
Sequencing information: PacBio HiFi reads
Total HiFi reads	89,477	91,529	1,23,472	73,554	95,007	39,546	97,614
Average length	8,791.4	8,455.8	8,006.2	11,162.9	9,076.5	10,089.2	10,694.9
Total Base	786,625,340	773,948,040	988,538,194	821,076,498	862,329,476	398,988,339	1,043,975,381
Sequencing information: Illumina reads
Illumina total reads	26,158,232	18,865,742	23,395,124	19,926,546	20,027,342	22,863,006	21,956,386
Illumina mean read length	151	151	151	151	151	151	151
Assembly method	Flye polished by NextPolish
Total length of assembly (Mb)	5.869	5.737	5.643	5.703	5.615	5.615	5.637
Total number of chromosomes	2 Choromosomes and 1 Plasmid	2 Choromosomes	2 Choromosomes	2 Choromosomes	2 Choromosomes	2 Choromosomes	2 Choromosomes
Average depth in PacBio	134	135	175	144	154	71	185
Average depth in Illumina	673	496	626	528	539	615	588
Busco score	100	99	100	100	100	100	100

Average nucleotide identity (ANI)

The ANI is a whole genome similarity matrix and has been presented as heatmap and dendrogram in Fig. 1. The ANIm score threshold of 95% is selected as the criteria to assign a species. The analysis revealed that four strains of V. harveyi (BSW5, BSW7, 1DA3 and KC13.17.5) were misidentified as their ANIm score ranged from 88.4% to 90% with reference to V. harveyi type strain ATCC 14126 T (Fig. 1; Supplementary Table S2). The ANIm score identified these strains as V. campbellii (KC13.17.5), V. jasicida (BSW5 and BSW7) and V. owensii (1DA3). Similarly, three strains of V. campbellii (Tx.T18, VB18PR-0194–1 and VB18PR-0122–1) were found to be misclassified, as they exhibited lower ANIm score (84.7 to 92.8%) relative to V. campbellii type strain CAIM519T. Of these, VB18PR-0194–1 and VB18PR-0122–1 had the closest similarity to V. owensii LMG25443T (ANIm = 94.1%) but did not satisfy the criteria for inclusion within this species. Notably, Tx.T18 appeared the most divergent, with its maximum ANI (94.2%) with V. chagasii LMG21353, a species of the Splendidus clade of Vibrio (Supplementary Fig. S3 and Supplementary Table S3). Overall, ANIm score identified 204 strains as V. harveyi and 78 strains as V. campbellii.

Fig. 1.

Phylogenetic tree of Harveyi clade species based on average nucleotide identity (ANI) generated using pyANI. Vh, V. harveyi; Vc, V. campbellii; Vo, V. owensii; Vj, V. jasicida; Vr, V. rotiferianus; Vp; V. parahaemolyticus

Pangenome analysis and core genome-based phylogeny

Pangenome analysis of 315 strains identified 2,379 core (genes present in 99% of strains), 1,173 soft-core (95—99% strains), 2,527 shell (15—95% strains) and 35,503 cloud genes (< 15% strains), yielding a pangenome of 41,582 genes (Fig. 2A). The curated 82-genome dataset showed an expanded core genome (2595 genes) and markedly reduced cloud (22,132 genes) and pangenome (28,146 genes) (Fig. 2B). The pangenome was found to be open, as inclusion of additional genomic sequences continued to increase the pangenome size (Fig. 2C,D).

Fig. 2.

Pangenome characteristics of Vibrio harveyi and V. campbellii (A) Distribution of pangenome statistics (core, soft-core, shell, cloud and total genes) in 315 genomes datasets. (B) Distribution of pangenome statistics in the curated dataset of 82 genomes. (C) Pangenome accumulation curve based on 315 genomes. (D) Pangenome accumulation curve based on the curated dataset of 82 genomes

Core genome-based maximum-likelihood phylogenies were constructed using the full dataset of 315 genomes and the curated dataset of 82 genomes (Fig. 3A,B). Each recognized species formed a monophyletic lineage supporting clear species boundary within Harveyi clade. Both the phylogenetic analysis consistently placed V. harveyi as an early diverging lineage, clustering closely with V. rotiferianus. In contrast, V. campbellii emerged as the most recently diverged species, showing closer proximity to V. owensii and V. jasicida, which formed a sister-species. V. harveyi appears as a genetically cohesive lineage with minimal internal subdivision, whereas, V. campbellii exhibited greater internal diversification, suggesting ongoing speciation process within the species. Overall, the inferred speciation trajectory within Harveyi clade followed the order of V. parahaemolyticus → V. rotiferianus → V. harveyi → V. jasicida/V. owensii → V. campbellii. The phylogenetic analysis further identified 204 strains as V. harveyi and 78 strains as V. campbellii (Fig. 3A). Similar to ANIm score four V. harveyi strains (BSW5, BSW7, 1DA3 and KC13.17.5) were misidentified. These strains clustered with V. campbellii (KC13.17.5), V. owensii (1DA3), and V. jasicida (BSW5, BSW7). Similarly, three V. campbellii strains were found misidentified. Strain VB18PR-0194–1 and VB18PR-0122–1 clustered adjacent to V. owensii, forming a distinct lineage separated from the V. campbellii groups. Although, these isolates do not form a monophyletic group with V. owensii, their phylogenetic placement indicates a greater genomic affinity to V. owensii than to the canonical V. campbellii cluster. Among the misidentified strains, V. campbellii Tx.T18 was highly divergent, forming a distinct lineage that branched outside even to V. rotiferianus and V. parahaemolyticus.

Fig. 3.

Core-genome based maximum likelihood phylogeny of the Harveyi clade (A) Phylogenetic tree constructed using the dataset of 315 genomes. The tree was generated from a concatenated alignment of 2,379 core genes produced by Panaroo pipeline. (B) Phylogenetic tree constructed using the curated dataset of 82 genomes. The tree was generated from a concatenated alignment of 2595 orthologous core genes produced by Panaroo. Different species are indicated by different color shading. Branches with bootstrap value greater than 75 are highlighted in blue. The strain sequenced in the present study is depicted in blue, whereas misidentified strain are shown in red. V. cha, V. chagasii; Vh, V. harveyi; Vc, V. campbellii; Vo, V. owensii; Vj, V. jasicida; Vr, V. rotiferianus; Vp; V. parahaemolyticus. V. campbellii strains are further classified into three groups: Vc_Gr-L (strains possessing luminescence operon), Vc_Gr-LM (strains carrying luminescence operon flanked with mobile genetic elements), and Vc_Gr-LD (strains harboring luminescence-defective operon). Trees were visualized and annotated using iTOL and InkScape (version 1.4.2)

The core genome-based phylogeny resolved V. campbellii into three major groups: Gr-L (luminescent), Gr-LM (luminescence with mobile genetic elements) and Gr-LD (luminescence-defective) (Fig. 3B). In both the phylogenetic tree, Gr-LD consistently clustered in close association with Gr-LM, suggesting a shared evolutionary history. However, the two phylogenies differed slightly in their clustering pattern of Gr-L group. The highly curated phylogeny revealed a clear evolutionary progression from Gr-L to Gr-LM and ultimately to Gr-LD (Fig. 3B). However, in the 315-genome phylogeny, luminescent Gr-L strains were distributed on either side of the Gr-LM/Gr-LD cluster, reflecting increased topological complexity associated with broader strain diversity. Overall, the Gr-L was the most dominant group, accounting for 83.3% of the strains (65 isolates), and it carried functional luminescence operon. All V. campbellii strains sequenced in the present study were the part of Gr-L group. The Gr-LD constituted the second most dominant group (12.8%; 10 isolates) and included type strain CAIM519T. The Gr-LM comprised three strains, including the model bioluminescent strain BAA-1116.

Comparative genomic metrics

A significant difference (p < 0.001) was observed in key genomic parameters between V. harveyi and V. campbellii. For example, the GC content was significantly higher (p < 0.001; R² = 0.74) in V. campbellii (45.43 ± 0.02%) compared to V. harveyi (44.94 ± 0.01%) (Fig. 4A). However, the genome size and number of protein coding genes were significantly higher in V. harveyi (5.94 ± 0.02 Mb; 5323 ± 15.63 genes) compared to V. campbellii (5.76 ± 0.03 Mb; 5089 ± 32.82 genes) (p < 0.001; R² = 0.20 and 0.27, respectively) (Fig. 4B,C). Further, the ANI analysis indicated that V. harveyi is a highly conserved species as the type strain ATCC 14126 showed 0.01–1.43% divergence. In contrast, V. campbellii exhibited higher genomic plasticity, as its type strain CAIM 519^T differed 1.00–3.52% from other strains.

Fig. 4.

Comparative analysis of genomic parameters between Vibrio harveyi and V. campbellii and among V. campbellii phylogenetic groups (A–C) Comparative analysis of (A) GC content, (B) genome size, and (C) number of protein-coding genes between V. harveyi and V. campbellii. (D–F) Corresponding comparisons of (D) GC content (E) genome size and (F) number of protein coding-genes among Gr-L, Gr-LM, and Gr-LD groups of V. campbellii. Difference in genomic parameters between V. harveyi and V. campbellii were analysed using a two-tailed independent t-test. Genomic indices among three groups of V. campbellii were compared using one-way ANOVA followed by Tukey’s multiple comparison. V. campbellii strains were classified into three phylogenetic groups; Gr-L (luminescent), Gr-LD (luminescence-defective), and Gr-LM (luminescent with mobile genetic elements) based on core genome phylogeny and synteny organization. Data represent mean ± SE. The level of significance is indicated as: P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***)

Within V. campbellii, notable genomic differences were observed among the three phylogenetic groups. The luminescent groups Gr-L and Gr-LM exhibited higher genomic indices such as elevated GC content (45.5%) (p < 0.001; R² = 0.43), larger genome sizes (~ 5.80 Mb) (p < 0.001 for Gr-L and p < 0.05 for Gr-LM; R² = 0.25), and greater number of protein-coding genes (5,137 to 5,394) (p < 0.01; R2 = 0.17) compared to the Gr-LD group (GC content 45.1%; genome size 5.43 Mb; protein-coding genes 4809) (Fig. 4D,F). However, no significant difference (p > 0.05) was observed for these genomic indices between Gr-L and Gr-LM groups.

Mapping of luminescence and sucrose operon

Luminescence (luxCDABEGH) and sucrose (scrRAKB) operons were in silico mapped in 204 strains of V. harveyi and 78 strains of V. campbellii (Fig. 5, Table 3, Supplementary Tables S4, S5, S6 and S7). BLASTP analysis showed 93.1–100% identity for the lux operon against the query sequence (Table 3). Only 2.94% strains of V. harveyi (6 isolates; ATCC 14126, ATCC 33843, Fish5, Fish10, XH2145, and VARA 4 1.1) carried a functional luxCDABEGH operon while one isolate (Hep-2a-10) possessed partial luxBEG operon (Fig. 5). The remaining 197 strains, including V. harveyi SB1 lacked all luminescence-related genes. In contrast, all V. campbellii strains carried either complete or partial lux operon; 83.4% had complete luxCDABEGH operon, 12.8% harboured defective luxBG and 3.8% (3 strains) had an operon with missing luxB or luxC. All V. campbellii strains sequenced in this study carried functional luxCDABEGH operon. Defective luxBG operon were observed in the type strain CAIM 519 ^T and widely studied strain DS40M4.

Fig. 5.

Distribution of luminescence and sucrose operon genes in Vibrio harveyi and V. campbellii Strains sequenced in the present study has been shown in blue color. Luminescence operon (luxCDABEGH) has been shown as blue color shaded circle while scrRAKB operon as red shaded square. The empty circle/square represents the absence of gene. The tree was visualized and annotated using iTOL. Vh, V. harveyi; Vc, V. campbellii

Table 3.

Protein BLAST of luminescence and sucrose operon.

Gene	Identity %	Coverage	E value	Bit score
Luminescence operon
V. harveyi
LuxC	96.9	100	0	927–929
luxD	97.0–97.4	100	0	616–620
luxA	98.9	100	0	742
luxB	98.1–98.5	100	0	668—669
luxE	97.9–98.4	100	0	778—781
luxG	93.1–96.1	100	2.91E-165–3.85E-169	458—468
LuxH	97.4–98.3	100	1.44E-164–1.71E-168	456–466
V. campbellii
LuxC	96.5–100	51–100	7.29E-167—0	471–953
LuxD	97.4–100	100	0	620–633
LuxA	98.9–100	97–100	0	726–748
LuxB	97.5–100	74–100	0	504—679
LuxE	98.4–100	100	0	782–793
LuxG	95.3–100	100	2.23E-168–6.68E-176	465–484
LuxH	98.7–100	100	2.49E-169–2.19E-172	468—475
Scr Operon
V. harveyi including operon1 of V423 and V424
scrA	98.1–98.7	52—100	0	481 −939
scrB	96.1–97.7	99	0	973—990
scrK	96.7–98.7	61—100	0	373—617
scrR	98.8–100	100	0	666—673
V. harveyi V423 and 424 operon2
scrA	91.65	100	0	865
scrB	74.32	99	0	769
scrK	81.31	99	0	512
scrR	86.89	100	0	586
V. campbellii 170,502
scrA	98.7	100	0	939
scrB	96.7	99	0	978
scrK	98.0	100	0	613
scrR	99.4	100	0	671

Mapping of the scr operon revealed that 89.2% of V. harveyi strains carried a complete and putatively functional scrRAKB operon, containing all genes required for sucrose utilization (Fig. 5). All V. harveyi strains harbored a single scr operon, except strains V423 and V424, which carried two copies. The scrRAKB operon in V. harveyi, including the first operon of V423 and V424, exhibited 96.1% (ScrB) to 100% (ScrR) protein identity against the V. alginolyticus reference sequence (Table 3). In contrast, the second scrRAKB operon of V423 and V424 showed lower identity, ranging from 74.3% (ScrB) to 91.6% (ScrA) with V. alginolyticus. BLASTP analysis indicated that these second operon shared the highest identity (95.8% for ScrB to 99.1% for ScrR) with V. crassostreae and V. toranzoniae, members of the Splendidus clade of Vibrio (Supplementary Table S8).

Despite high prevalence, 10.8% of V. harveyi strains lacked one or more scr genes. Notably, 6.4% (13 strains) lacked only the scrA gene. Nucleotide BLAST revealed that 12 of these strains carried scrA with ≥ 98.4% nucleotide identity, but with disruptive mutations. Nine strains had a guanine deletion at 628^th position, within a stretch of seven consecutive guanines, causing a frameshift (Supplementary Fig. S4). Additional mutations included a C → T substitution at position 178 introducing a premature stop codon, and an insertion at position 710 resulting in a frameshift. Among the examined 204 V. harveyi strains, only strain 345 lacked all four scrRAKB genes. In contrast, among 78 V. campbellii strains, all lacked the scr operon except strain 170502, which showed 96.7% (scrB) to 99.4% (scrR) identity with V. alginolyticus 170502.

Syntenic organization of luminescence and sucrose operon

In V. campbellii, the bioluminescence operon is located on chromosome 2 and organized syntenically as luxCDABEGH (Fig. 6A). Three distinct syntenic patterns for luminescence operon were noticed in V. campbellii. A) Functional luxCDABEGH operon (LB503 type): Present in the Gr-L group, this is the most dominant pattern and observed in 83.4% V. campbellii strains. The group is represented by highly luminescent strain LB503 (sequenced in this study). The operon is flanked upstream by an LruC domain-containing protein and downstream by molybdenum utilization genes (MoeB and MoeA). B) Functional operon with mobile genetic elements (BAA-1116 type): This is found in Gr-LM group (three strains; BAA-1116, BB120 and MP1). This pattern retains a functional luxCDABEGH operon with downstream MoeB and MoeA, but is flanked upstream by pseudo IS4 and IS3 type transposase, disrupting LruC, forming a potential hotspot for horizontal gene transfer. C) Defective luxBG (CAIM519^T type): Observed in the Gr-LD group (including type strain CAIM 519ᵀ and DS40M4), this pattern retains flanking lruC upstream and moeB/moeA downstream but lacks luxCDAEH, indicating partial operon loss without disruption to surrounding genes. In V. harveyi, the luxCDABEGH operon is similarly located on chromosome 2 with flanking regions analogous to the LB503 type and lacks nearby mobile genetic elements.

Fig. 6.

Syntenic organization of luminescence operon (A) and scrRAKB operon (B) in Vibrio harveyi and V. campbellii Vh, V. harveyi; Vc, V. campbellii; Va, V. alginolyticus. T3SS indicates start of T3SS operon by VopD. The luxCDABEGH operon and scrRAKB operon is represented by orange color. Blue color is representing the presence of transposable elements. The tree was generated using EasyFig and annotated using InkScape (version 1.4.2)

The syntenic organization of the scrRAKB operon is presented in Fig. 6B. In V. harveyi SB1, the operon is arranged as scrRAKB, similar to V. alginolyticus. However, few notable differences were also observed. In V. alginolyticus, scrRAKB is located on chromosome 2 with no genes of the type 3 secretion system (T3SS) in proximity. In contrast, in V. harveyi SB1 it is located on chromosome 1, with complete 29-gene T3SS operon at upstream. The upstream flanking genes include carbohydrate porin, carboxymuconolactone decarboxylase, a MerR transcriptional regulator, and T3SS. The downstream flanking genes include Yjdm, an ABC transporter, siderophore biosynthesis genes, and a class I SAM-dependent methyltransferase. The first scrRAKB operon of V. harveyi V423 and V424 displayed a syntenic pattern identical to SB1. However, the second operon in these strains showed evidence of horizontal gene transfer, being flanked upstream by a complete IS110 and a pseudogene of IS5, and downstream by a complete IS30, and pseudogenes of IS5, IS21, and IS982. In V. campbellii 170502, scrRAKB is also located on chromosome 1 adjacent to T3SS, with signs of horizontal gene transfer, as it lies in a mobile genetic hotspot flanked by IS4 and IS200/IS605 transposases.

Phylogenetic analysis of lux and scr operon

Phylogenetic reconstructions of the lux and scr operons are presented in Figs. 7A,B. The concatenated LuxCDABEG protein sequences from six V. harveyi strains formed a distinct monophyletic group, suggesting a shared evolutionary origin of their luminescence operons. Further, it clustered together with V. campbellii, and the closest sister group comprised three strains (V. campbellii LB102, LB198, and LB314) sequenced in this study.

Fig. 7.

Maximum likelihood Phylogenetic analysis of concatenated sequence of luminescence (LuxCDABEG) (A) and ScrRAKB (B) in V. harveyi and V. campbellii. The strains having full set of LuxCDABEG and ScrRAKB were selected for phyologenetic analysis. V. vulnificus ATCC 43382 served as outgroup for bioluminescence and V. alginolyticus ATCC 17749 for ScrRAKB clustering. Strains sequenced in the present study has been shown in blue color. Bootstrap value of more than 0.75 has been shown as star. Increasing size of star indicates higher bootstrap value. The tree was visualized and annotated using iTOL. Vh, V. harveyi; Vc, V. campbellii; Vv, V. vulnificus; Va, V. alginolyticus

In the present study, both V. harveyi V423 and V424 harbored two scr operons. The first operon (operon1) in both strains clustered with other V. harveyi, showing closest phylogenetic relation with V. harveyi ISF-200–6, consistent with their taxonomy. In contrast, the second operon (operon2) formed a distinct outgroup to all V. harveyi and V. alginolyticus strains, suggesting a possible acquisition via horizontal gene transfer (HGT) from distantly related donors. Evidence of HGT was also found in V. campbellii 170502, whose concatenated ScrRAKB protein sequence clustered within the V. harveyi group, showing highest similarity to V. harveyi strains such as Y6, 2020 V-1021, VB18PR-0039–2, and Vh427, implying possible acquisition from a V. harveyi-like ancestor.

Mobilome analysis

Mobilome analysis results are presented in Table 4 and supplementary Tables S9, S10, S11 and S12. Among six V. harveyi strains carrying luminescence operon, five had chromosome-level assemblies with operon consistently located on chromosome 2. The mobilome analysis revealed that IS element counts on chromosome 2 ranged from 0 to 5, with the closest IS located 117.1 kb (ATCC 14126) to 683.9 kb (ATCC 33843) from the lux operon. Further, among analyzed strains, an intact prophage (12B12) was detected on chromosome 2 in V. harveyi Fish10 which is positioned 98.9 kb from the lux operon. The other mobilome types like integrative/conjugative elements (ICE/IME/T4SS), integrons, and CALINs (clustered attC sites lacking integrase) were absent in these strains on chromosome 2. Overall, no mobile genetic elements were detected within 50 kb of the luminescence operon, suggesting its location in a genomically stable region.

Table 4.

Distribution of mobile genetic elements in Vibrio harveyi and V. campbellii strains suspected for horizontal gene transfer of the luminescence and scr operon.

Strains	Accession	Chr/contigs	Luminescence/scr operon coordinates (bp)	Complete integron	CALINs	IS	ICE/IME/T4SS	Prophage	MGE proximity to luminescence/scr operon
Luminescence operon (luxCDABEGH)
V. harveyi ATCC 14126/FDAARGOS_109	GCA_001471575.2	LOSK02000001.1(Chr II)	1,384,022 - 1,391,131	0	0	4 (1266958_1267701)	0	0	117.1 kb
V. harveyi ATCC 33843	GCA_000770115.2	CP009468.1 (Chr II)	318,403–325,512	0	0	5	0	0	683.9 kb
V. harveyi XH2145	GCA_021397735.1	CP090178.1 (Chr II)	373,388 - 380,497	0	0	4 (496818_497561)	0	0	123.2 kb
V. harveyi Fish5	GCF_028898825.1	NZ_CP118585.1 (Chr II)	1,410,033 - 1,415,017	0	0	1 (994039_995403	0	0	416.2 kb
V. harveyi Fish10	GCF_028898705.1	NZ_CP118597.1 (Chr II)	1,968,092 - 1,975,201	0	0	1	0	1 (Intact; 12B12) 2,074,095–2,100,556)	98.9 kb from prophage and 204.7 kb from IS
V. harveyi VAR A4 1.1	GCA_049311325.1	Contig18	11,063- 18,172	0	0	0	0	0	No
ScrRAKB operon
V. harveyi V424	GCA_050017355.1	JBNGUA010000001	980,982—986,207	0	0	7 (976,471–1,008,584)	0	0	Seven IS elements flank 2nd scr operon
V. harveyi V423	GCA_050017375.1	JBNGUB010000002	356,055—361,280	0	0	7 (351,544—383,657)	0	0	Seven IS elements flank 2nd scr operon
V. campbellii 170502	GCA_003691485.1	CP033134.1 (Chr 1)	1,214,367—1,219,603	1 (2,061,932–2,070,280)	4 ( 2,019,323–2,057,584)	30	0	0	Two IS elements flank scr operon

Potential HGT of the scr operon was assessed V. harveyi V423, V. harveyi V424, and V. campbellii 170502, where the operons were located on contigs JBNGUB010000002, JBNGUA010000001, and Chromosome 1, respectively. No prophage or ICEs were found near the operons. However, the second scr operon in both V. harveyi strains was flanked by seven IS elements and V. campbellii 170502 by two IS elements, suggesting possible HGT. Although V. campbellii 170502 carried integrons on chromosome 1, these were located 842 kb from the scr operon, indicating no direct association.

Phenotypic validation of luminescence and sucrose utilization in V. harveyi and V. campbellii

Eighteen V. harveyi and thirty-one V. campbellii strains were phenotypically characterized for luminescence (Fig. 8A and B). Among V. harveyi, only the type strain ATCC 14126 was luminescent, consistent with the presence of functional luminescence operon by in silico analysis. In contrast, all V. campbellii strains were luminescent, except the type strain CAIM519^T, which was sole non-luminescent strain. Luminescence intensity varied markedly, ranging from a 1,582-fold increase in V. harveyi ATCC 14126 to a 133,166-fold increase in V. campbellii LB164, relative to the negative control. The model strain V. campbellii BAA-1116 demonstrated an 80,870-fold increase, underscoring substantial inter-strain variability.

Fig. 8.

Luminescence behaviour and sucrose utilization in V. harveyi and V. campbellii (A) Luminescence on Zobell marine agar by representative strain of V. campbellii. (B). Luminescence by different strains of V. harveyi (n = 18) and V. campbellii (n = 31) at 12 h of incubation. (C). Yellow and green colouration on TCBS agar. Number 1 to 18 (yellow colony) represents eighteen strain of V. harveyi. All strains of V. campbellii produced green colonies. Therefore, number 19 to 30 is representative strains of V. campbellii (green colony) (D). Growth of V. harveyi (n = 18) and V. campbellii (n = 31) in M9-Sucrose medium

To validate in-silico findings on sucrose utilization, the same 49 isolates were screened on TCBS agar containing sucrose (Fig. 8C). All V. harveyi strains produced yellow colonies (scr⁺ phenotype), whereas all V. campbellii strains formed green colonies (scr- phenotype). Growth kinetics in well-defined M9 medium with 1% sucrose (Fig. 8D) further confirmed these results: V. campbellii strains showed no growth, while 100% V. harveyi isolates recorded growth in the medium. Among V. harveyi, Y1/12–23 exhibited the highest growth while the isolate LMG 19643 had the lowest growth.

PCR screening of luxA and scrA genes

The presence of luminescence operon in V. harveyi and V. campbellii was further verified in the 49 selected isolates by PCR screening of the luciferase gene luxA (Fig. 9A,B and supplementary Fig. S5 A, B, C, D). Consistent with phenotypic observations, among V. harveyi only the type strain ATCC 14126 produced the expected 828 bp amplicon, whereas all V. campbellii isolates, except the type strain CAIM519^T produced the amplicon. Sucrose utilization was further verified by screening scrA gene (Fig. 9C,D). All V. harveyi strains produced the desired 698 bp amplicon, confirming the presence of scrA gene. In contrast, no V. campbellii strain produced an amplicon, indicating the absence of scrA in this species.

Fig. 9 .

Gel picture showing presence of luxA (A and B) and scrA (C and D) gene. (A) and (B): M, 100 bp ladder; L1 to L18 represents 18 V. harveyi strains (L1, ATCC 14126; L2, LMG 7890; L3, LMG 19643; L4, SB1; L5, V2301; L6, V2426; L7, V2427; L8, V2432; L9, V2437; L10, V2438; L11, V2439; L12, V2441; L13, V2452; L14, Y/12–23; L15, Y2/12–23; L16, FCD2023/4; L17, FCD2023/7; L18, FCD2025/2) L19, Negative control for V. harveyi; L20 to L50 represents 31 V. campbellii strains (L20, LMG11216/CAIM 519 T; L21, BAA-1116; L22 to L50, 29 strains between LB1 to LB516; L51, Negative control for V. campbellii). (C)and (D): Distribution of scrA gene. M; 100 bp ladder; L1 to L18 represents 18 V. harveyi strains having desirable 698 bp amplicon. L19, Negative control for V. harveyi; L20 to L50 represents 31 V. campbellii strains with missing scrA gene. L51: Positive control for scrA using V. alginolyticus. L52: Negative control for V. campbellii

Discussion

V. harveyi is a major pathogen in aquaculture causing diseases in shrimp and fishes. There are several reports of misidentification of V. harveyi with closely related species like V. campbellii, V. owensii and V. jasicida^2,16,17,44. Average nucleotide identity (ANI) and core genome-based phylogeny identified seven misidentified strains; four labelled as V. harveyi (BSW5, BSW7, 1DA3 and KC13.17.5) and three as V. campbellii (Tx.T18, VB18PR-0194–1 and VB18PR-0122–1) (Fig. 1 and 3A). Consistent with earlier studies, KC13.17.5 was reclassified as V. campbellii, BSW5 and BSW7 as V. jasicida⁴⁵ and 1DA3 as V. owensii²⁰. The two V. campbellii (VB18PR-0194–1 and VB18PR-0122–1) strains exhibited ANIm values of 94.1% with V. owensii and 92.8% with V. campbellii, placing them below the 95% species delineation threshold⁴⁶, suggesting affiliation with V. owensii or a distinct lineage within Harveyi clade. Among all misidentified strain, V. campbellii Tx.T18 was most divergent and exhibited ANIm score of only 84.7% with V. campbellii and other strains of Harveyi clade. Further analysis revealed that it had maximum ANIm score of 94.2% with V. chagasii a member of Splendidus clade (Supplementary Table S3). Overall, ANI and core genome-based phylogeny confirmed 204 strains as V. harveyi and 78 as V. campbellii.

High-quality genome sequences enable clear reconstruction of evolutionary patterns. To investigate the speciation within the closely related Harveyi clade (V. harveyi, V. campbellii, V. owensii, V. jasicida, V. rotiferianus and V. parahaemolyticus), we constructed two core genome-based phylogenies. The first, based on 315 genomes, primarily aimed to identify misclassified strains and provide a framework for screening bioluminescence and sucrose utilization. For this dataset, we relaxed genome-quality thresholds (upto 1000 contigs) to include historically important strains (e.g., V. harveyi ATCC 35,804, CAIM 148, CAIM 464, CAIM 1075; V. campbellii CAIM 115), which were sequenced earlier and had highly fragmented genomes (Supplementary Table S1). To overcome this limitation, we constructed a curated dataset of 82 strains using mostly complete genomes (Supplementary Table S1). Phylogenetic analysis of both the datasets revealed a consistent evolutionary trajectory within the Harveyi clade; V. parahaemolyticus → V. rotiferianus → V. harveyi → V. owensii/V. jasicida → V. campbellii. In this topology, V. harveyi represents an earlier-diverging lineage, whereas V. campbellii emerges as the most recently evolved species⁴⁵. The concordance of this evolutionary pattern across datasets of different genome quality underscores the robustness of the inferred speciation framework and provides a reliable phylogenetic context for evolution of trait like bioluminescence.

V. harveyi has traditionally been described as luminescent bacterium and luminescence served as a distinctive and easily observed phenotypes for its identification^1,23,26,47. However, our earlier work demonstrated that all luminescent bacterial isolates initially suspected as V. harveyi were in fact V. campbellii². Similarly, in another similar study, 93% of strains presumed as V. harveyi were re-identified as V. campbellii¹⁷. This high rate of misidentification prompted us to re-examine luminescent properties in V. harveyi and V. campbellii. Contrary to the earlier reports describing V. harveyi as luminescent bacterium, our genomic analysis revealed that 97.1% of the 204 V. harveyi strains lacked luxCDABEGH operon and were therefore predicted to be non-luminescent (Fig. 5). Further, only six strains (2.9%) carried the operon, retaining the genetic capacity for bioluminescence. Of these lux-positive strains, only the type strain ATCC 14126 was available for phenotypic testing, and it exhibited luminescence consistent with its genomic profile (Fig. 8B). Similarly, Nakayama et al.⁴⁸ reported that only one out six V. harveyi strains tested exhibited luminescence. Defoirdt et al⁴⁹ further compared luminescence in seven V. campbellii and six V. harveyi strains. They reported luminescence in five V. campbellii strains and only one V. harveyi strain (BB120). Notably, BB120 was later reclassified as V. campbellii¹⁶. Thus, evidence from the present study and previous study strongly suggests that V. harveyi is predominantly a non-luminescent bacterium.

In the present study, only six out of 204 analyzed V. harveyi strains carried luxCDABEGH operon while the majority lacked all associated genes (Fig. 5). Phylogenetic analysis of the concatenated protein sequence of the operon revealed that these strains clustered closely with V. campbellii, a pattern incongruent with the core genome-based phylogeny. The HGT of luminescence operon has been previously documented in Vibrio species such as V. vulnificus VVL1 and V. chagasii 21N-12 and SB-52⁵⁰, supporting the possibility of inter-species transfer. However, in the present study, no mobile genetic elements were detected within 50 kb of the operon (closest located at 98.9 kb), indicating that, if HGT occurred, it was likely ancient followed by stablisation of the operon. Although speculative, these strains may represent a remnants of an ancestral luminescent V. harveyi population, now largely lost across the species due to niche-specific selective pressures that rendered bioluminescence dispensable. However, further study is needed to conclusively determine the evolutionary origin of luminescence operon in V. harveyi.

In the present study, each strain of V. campbellii carried either a functional luxCDABEGH (87.2% strains) or a defective luxBG variant (Fig. 5). Phenotypically, 30 out of tested thirty-one isolates were luminescent and carried luxA gene (Fig. 9A), confirming the high prevalence of bioluminescence within this species. These findings strongly suggests that V. campbellii is inherently a luminescent bacterium and corroborate previous reports implicating this species as a causative agent of luminescent vibriosis in shrimp hatcheries^2,51. Core genome-based phylogeny clustered V. campbellii into three major groups; Gr-L (luminescent), Gr-LM (luminescence operon with mobile genetic elements), and Gr-LD (luminescent defective) (Fig. 3B,C). Notably, in both the phylogenetic reconstructions, the Gr-LM and Gr-LD groups clustered together, suggesting that the presence of insertion sequences may have induced structural instability, ultimately leading to partial deletion of operon in Gr-LD. The strong phylogenetic concordance between synteny-based grouping and core genome-based phylogeny supports that bioluminescence in V. campbellii is not merely an accessory trait but an evolutionary conserved and functionally significant phenotype. This may reflect the adaptive advantage conferred by bioluminescence in natural marine environments, such as facilitating intercellular communication and niche colonization. As bioluminescence is tightly regulated by quorum sensing, a regulatory system that also governs virulence, colonization and other physiological processes, future research should investigate how distinct syntenic groups influence virulence potential and ecological adaptability of V. campbellii.

Despite the dominance of luminescent properties in V. campbellii, a sub-set (12.8%) of strains including type strain CAIM519^T, carried a dysfunctional luxBG operon and lacked luxCDAE genes (Fig. 6A). Core genome-based phylogeny clustered these strains into a distinct luminescence-defective (Gr-LD) group, suggesting a common evolutionary origin and conserved pattern of operon loss. Notably, the phylogeny also showed that the Gr-LD lineage clusters closely with the luminescence group flanked with mobile genetic elements (Gr-LM), representing the evolutionary transition (Fig. 3B). This highlights that the unstable luminescence operon within Gr-LM strains may have served as precursor for gene loss and emergence of the Gr-LD lineage. Consistent with this trajectory, Gr-LD strains exhibited significantly lower GC content, smaller genome size and reduced protein-coding genes compared to luminescent groups (Gr-L and Gr-LM) (Fig. 4E,F). Similar pattern has been reported for V. campbellii DS40M4, which lack luxCDAE and key quorum sensing genes such as luxM,—responsible for synthesizing autoinducer-1, a homoserine lactone signal molecule, along with markedly reduced regulon⁵². Several earlier studies have suggested that V. campbellii comprises two to three genomically distinct subgroups^45,53. Our findings further support these findings as phenotypically V. campbellii segregate into luminescent (Gr-L and Gr-LM) and luminescent defective (Gr-LD) groups. Notably, most Gr-LD strains originated from offshore, deep-sea environments ranging from the surface to hadal depths (> 6000 m), consistent with niche-driven reductive evolution⁵⁴. We hypothesize that as part of ecological adaptation in low-nutrient environments, these strains may have evolved to abandon quorum sensing-regulated community behaviours, such as bioluminescence, in favor of solitary survival strategies.

Colony morphology on TCBS agar serves as widely used test in the presumptive identification of Vibrio spp. In the past, several studies reported V. harveyi as non-sucrose fermenter (green colony on TCBS agar)^24,35,36. On the contrary, few studies found V. harveyi as sucrose fermenter (yellow colony on TCBS agar)^37,38. In the present study, all 18 V. harveyi isolates including the type strain ATCC 14126, produced yellow colonies (Fig. 8C), and grew luxuriantly in M9 minimal medium with sucrose as sole carbon source (Fig. 8D). Additionally, all isolates harbored the sucrose uptake gene (scrA) (Fig. 9C), consistent with their fermentation phenotype. Although phenotype testing was done on a limited number of strains, a complementary in silico analysis of 204 V. harveyi genomes revealed that 89.5% harbor a complete scrRAKB operon. This suggests that majority of V. harveyi strains have the metabolic capacity for sucrose utilization. The discrepancy in earlier reports may reflect strain-level variation, misidentification, or limitations in phenotypic characterization methods. This underscores the value of combining phenotypic and genomic data to define species-level metabolic traits within the Vibrio genus.

ScrA plays a critical role in sucrose metabolism by mediating the uptake via phosphoenolpyruvate-dependent carbohydrate phosphotransferase systems (PTS). It has been reported that mutation in scrA can abolish sucrose utilization⁵⁵. In the present study, BLASTP analysis indicated absence of ScrA in 6.4% (13 out of 204) V. harveyi strains. Subsequent BLASTN analysis showed high sequence identity (95 to 98.6%) against the scrA gene of V. harveyi SB1 suggesting the gene presence in these isolates. However, multiple sequence alignment revealed frameshift mutation (10 strains) or immature stop codon (1 strain). It is well-established that microbial populations often exhibit phenotypic heterogeneity, and up to 10% of bacterial strains can naturally lack a particular phenotype due to genetic variation, and other potential differences in transcription, translation and regulatory mechanisms^56,57. This raises the likelihood that a subset of V. harveyi strains harbor inactivating mutation in scrA, leading to a non-sucrose-fermenting phenotype. Notably, the mutation hotspot coincided with a missing guanine at position 628 within a homopolymeric G-rich region (seven consecutive guanines spanning positions 622–628). Many of these isolates were sequenced using Oxford Nanopore sequencing technology which is known to have elevated base-calling error rates, especially within homopolymeric stretches⁵⁸. Interestingly, one isolate (94/17) has been phenotypically characterized and reported to ferment sucrose⁵⁹, suggesting that scrA is likely intact and functional in few of these strains. Collectively, the presence of frameshift mutation in scrA provide genetic basis for the non-sucrose-fermenting phenotype and highlight the role of gene-level variation in shaping metabolic diversity.

In the present study, mapping of the scr operon in 78 isolates of V. campbellii revealed its absence in all the strains except V. campbellii 170,502 (Fig. 5). The lack of sucrose utilization in V. campbellii was further supported by phenotypic study; all 31 tested strains produced green colonies on TCBS agar, failed to grow in M9-sucrose medium and lacked scrA gene. These findings support earlier reports that V. campbellii is a sucrose non-fermenter^2,6. Based upon these studies it could be concluded that V. campbellii is a non-sucrose fermenter and form green colonies on TCBS agar. Among all examined strains, only V. campbellii 170,502 carried the scrRAKB operon. This was flanked by IS4 and IS200/IS605 family transposases and clustered phylogenetically with V. harveyi, strongly suggesting acquisition via horizontal gene transfer (HGT) from V. harveyi. Additional evidence for scr operon mobility was observed in V. harveyi strains V423 and V424, each carrying two distinct scr operons. While one operon clustered with other V. harveyi strains, the second formed an outgroup even to V. alginolyticus. BLAST analysis indicated highest identity with the scr operons of V. crassostreae and V. toranzoniae, both members of the Splendidus clade, suggesting interclade HGT. Mobilome analysis and synteny further revealed that the second scr operon in these strains was flanked by seven transposable elements, reinforcing the likelihood of acquisition via HGT. Similarly, HGT for scr operon has been reported in V. parahaemolyticus⁶⁰ and Photobacterium damselae susbsp damselae⁵⁵. Given that sucrose is the most abundant disaccharide in nature, acquisition of the scr operon likely confers a selective advantage by enabling utilization of diverse ecological niches, thereby favoring its horizontal transfer via transposable elements.

In the present study, V. harveyi appears to function as a niche specialist, with most strains isolated from marine animals, 43 different hosts including humans, indicating strong adaptation to host-associated environments. In contrast, V. campbellii behaves as an ecological generalist^61,62, being predominantly free-living in seawater, with host adaptation mostly limited to crustaceans. These ecological distinctions align with key phenotypic trait: V. harveyi as mostly non-luminescent consistent with host-associated requirements, whereas V. campbellii luminescent favorable for free-living or planktonic survival⁶². Collectively, the ecological, phenotypic, and genomic contrasts highlight that niche specialization in V. harveyi versus ecological generalism in V. campbellii has been a major driver of their evolutionary divergence and ongoing speciation within the Harveyi clade.

The present study combines the genome-wide in silico analysis of 282 strains (204 V. harveyi and 78 V. campbellii) with phenotypic validation of 49 representative isolates (18 V. harveyi and 31 V. campbellii) to assess bioluminescence and sucrose utilization. While this approach yielded robust comparative genomic insights, the phenotypic dataset represents only a subset of strains. As a result, our conclusions comparatively rely more on the in silico predictions, and it is possible that certain phenotypic variations were not captured within the tested isolates. Nonetheless, the integration of extensive genomic mapping with targeted phenotypic validation offers a strong framework for understanding these phenotypic traits in two major aquatic pathogens. Studies incorporating larger datasets across diverse geographic and ecological sources will further help to refine and validate these findings.

Conclusion

Bioluminescence and sucrose utilization in V. harveyi and V. campbellii were characterized by genome-wide operon mapping with targeted phenotypic validation. The distribution of bioluminescence (LuxCDABEGH) and sucrose (ScrRAKB) operons suggested V. harveyi as primarily non-luminescent, sucrose fermenter (yellow-colony on TCBS agar), whereas V. campbellii is predominantly luminescent and sucrose non-fermenter (green colonies). While the study provides a practical framework for presumptive identification of these two important aquatic pathogens, conclusions rely largely on in silico prediction as phenotypic validation was limited to a smaller subset of strains. Expanded phenotypic validation with broader strain diversity will further strengthen these findings.

Methods

Bacterial strains for genomic and phenotypic study

Bacterial strains for in silico analysis

A total of 374 genome assemblies (238 V. harveyi and 136 V. campbellii) were retrieved from NCBI GenBank collected during 1935 to 2024 (June 2025; Supplementary Table S1). A total of 59 accessions (13 V. harveyi and 46 V. campbellii) were excluded due to redundancy. This included 38 accessions for V. campbellii strain MP1 (sequenced for 39 times; MP1_replicate1 to MP1_replicate39)⁶³. Additional duplicity for V. campbellii comprised two accessions for type strain CAIM519T, one each for DS40M4, 151112 C, and HQB174 as well as three ENA-EUAN accessions (selected corresponding NCBI accession). For V. harveyi, duplicate included four accessions corresponding to type strain ATCC 14126 (which was sequenced for five times), five ENA-EUAN accessions, and one accession each for VH2, A2, 94/17 and HQB-19 (Supplementary Table S1). An additional 21 accessions (12 V. harveyi and 9 V. campbellii) were removed due to poor genome assembly (> 1000 contigs or < 4000 protein coding genes), and six accessions (five V. harveyi and one V. campbellii) for lacking annotated CDS at NCBI database. This filtering yielded 288 genomes (208 V. harveyi, 80 V. campbellii). To investigate evolutionary relationships and speciation patterns, we incorporated an additional 27 genome sequences from closely related species like V. parahaemolyticus (n = 5), V. rotiferianus (n = 4), V. jasicida (n = 9) and V. owensii (n = 9). This original dataset of 315 genomes were subjected for pangenome analysis, core genome-based phylogeny and average nucleotide identity (ANI) calculations. The analysis identified 282 strains comprising 204 V. harveyi and 78 V. campbellii and rest as misidentified strains belonging to other species. These 282 strains were subsequently used for genome wide in silico mapping of the bioluminescence and sucrose operon.

To better resolve phylogenetic relationship within the Harveyi clade, we generated a second set of dRep⁶⁴ filtered dataset (ANI threshold 99.9%; sa = 0.999; completeness ≥ 98%; contamination ≤ 5%). This refined dataset comprised 82 genomes, including V. harveyi (n = 34), V. campbellii (n = 24), V. owensii (n = 8), V. jasicida (n = 8), V. rotiferianus (n = 3), and V. parahaemolyticus (n = 5) (Supplementary Table S1). All species were represented by complete genome assemblies, except V. jasicida, which included one complete reference genome and seven high-quality draft assemblies (< 100 contigs).

Bacterial strains for phenotypic study and their molecular confirmation

In silico results were verified by phenotypic characterization and PCR screening of marker genes in 49 bacterial isolates; 18 V. harveyi and 31 V. campbellii (Table 1). V. harveyi isolates were recovered from disease affected Asian seabass (L. Calcarifer), Penaeus japonicus and P. vannamei while V. campbellii isolates were recovered from P. monodon and P. vannamei hatcheries and farms. Identity of these isolates was confirmed by multiplex PCR²¹ and partial sequencing of RNA polymerase sigma factor (rpoD) using published primers⁶⁵ (Table 5). Overall, 3 strains of V. harveyi (SB1, ATCC 14126 and ATCC 35084) and 9 strains of V. campbellii (CAIM519T, BAA-1116, LB3, LB10, LB102, LB135, LB198, LB314 and LB503) were the part of in silico as well as phenotypic analysis.

Table 5.

Primers sequence used for the characterization of bacterial isolates (rpoD) and confirmation of luciferase (luxA) and sucrose uptake (scrA) gene.

Sl. No	Gene	Primer sequence	Amplicon length	Reference
1	rpoD	70F: ACGACTGACCCGGTACGCATGTAYATG MGNGARATGGGNCANGT 70R: ATAGAAATAACCAGACGTAAGTTNGC Y TCNACCATYTCYTTYT	780	65
2	luxA	5F: AATTTGGAAACTTCCTTCTCACTTA 5R: TCACGCCATTGACCTTTATT	828	This study
3	scrA	F196: GGATCAGGCATCGTAAACCA R196: CAGTTCAGACCATCACCCAATA	698	This study

Whole genome sequencing

Whole genome sequencing was carried out for highly pathogenic V. harveyi SB1 and six representatives V. campbellii strains (LB3, LB10, LB135, LB198, LB314 and LB503). V. harveyi SB1 originated from vibriosis affected Asian seabass while V. campbellii strains were isolated from shrimp hatcheries and farm (Table 1). The genomic DNA was extracted using DNeasy Blood & Tissue Kits (Qiagen) according to the manufacturer’s instructions. The purity, concentration and integrity of the extracted DNA were determined using a NanoDrop 2000 (Thermofisher Scientific, Massachusetts, USA), Qubit 3.0 fluorometer (Thermofisher Scientific, Massachusetts, USA) and Agilent FEMTO pulse analyser (Agilent Technologies, California, USA), respectively. The whole genome sequencing was carried out at PacBio RS II and Illumina platform. For PacBio RSII sequencing, the DNA shearing was performed on Megaruptor 3 (Diagenode, Belgium) system. The SMRTbell library was constructed using the SMRTbell Express template Preparation Kit 2.0 (Pacific Biosciences, California, USA). The purified libraries were sequenced on SMRTcells containing 8M ZMW and sequenced in PacBio Sequel II system in CCS/HiFi mode at Nucleome Informatics, Hyderabad, India. For Illumina read, the library was constructed using the KAPA Hyper plus kit as per manufacturers’ protocol. The prepared libraries were quantified using Qubit 3.0 fluorometer (Thermofisher Scientific, Massachusetts, USA) and size of DNA was checked on Agilent 2100 bioanalyzer. The pooled libraries were sequenced using Illumina novoseq 6000 S4 flow cell.

Genome assembly was performed with the Flye 2.8.1⁶⁶ tool in the –pacbio-hifi mode. Polishing of these contigs for base errors and indels were carried out with Illumina short reads using NextPolish 1.4.0⁶⁷ brought the base consensus quality to 99.99%.

Average nucleotide identity

To determine the degree of genomic relatedness, pairwise average nucleotide identity (ANI) was calculated using the program Pyani v0.2.12 with MUMmer (ANIm)⁶⁸. Standard threshold of 95% identity and 70% sequence coverage was considered for species delimitation⁴⁶. The result was visualized using an heatmap generated by the R package gplots (https://cran.r-project.org/web/packages/gplots/index.html).

Pangenome analysis and core genome-based phylogeny

Pangenome analysis was performed using Panaroo pipeline⁶⁹, which employ graph-based filtering and alignment-correction strategies to retain conserved orthologs while removing spurious gene calls and collapsed paralogs. Panaroo was run with the options –clean-mode moderate using the.gff3 files generated by Prokka⁷⁰. The pangegnome accumulation curve were generated in R using custom script in Rstudio. Core genome-based phylogeny was constructed using 2,379 single-copy orthologus genes identified by Panaroo, aligned with MAFFT and trimmed by trimal under the automated1 setting⁷¹. Maximum-likelihood phylogenetic tree was inferred using IQ-tree 2 (v 2.2.6)⁷² under the best-fit substitution model GTR + F + I + R10. Node support was estimated using 1000 ultrafast bootstrap replicates, and the resulting phylogenetic tree was visualized using iTol⁷³. A second core genome phylogeny was constructed on the curated dataset of 82 genomes following the same workflow (Panaroo → MAFFT → trimAl → IQ-TREE2). In this dataset, 2,595 single-copy orthologus genes were used and phylogenetic inference was again performed under the automatically selected GTR + F + I + R10 model with 1,000 bootstrap replicates.

Genome wide mapping of bioluminescence and sucrose pathway in V. harveyi and V. campbellii

Mapping of bioluminescence (luxCDABEGH) and sucrose (scrRAKB) operons was carried out using query protein sequence described in Table 6. V. campbellii BAA 1116, a model strain for bioluminescence, was used as query sequence for luxCDABEGH²⁹ while well-characterized V. alginolyticus ATCC 17749 served as query sequence for scrRAKB operon^33,34. The protein BLAST with 50% minimum identity, query coverage and e value of 0.0001 considered as strain carrying the gene. The gene distribution pattern of luxCDABEGH and scrRAKB operons were visualized using iTOL with phylogenetic tree drawn from core genome-based phylogeny⁷³.

Table 6.

Query sequence used for mapping of luminescence (LuxCDABEGH) and sucrose (ScrRAKB) operon in Vibrio harveyi and V. campbellii.

Sl. No	Gene name	Gene accession	Gene description	Query organism	Reference
Bioluminescence (LuxCDABEGH operon)
1	LuxC	AGU98399.1	Fatty acid reductase	V. campbellii BAA-1116	29
2	LuxD	AGU98400.1	Myristoyl-ACP-specific thioesterase
3	LuxA	AGU98401.1	Luciferase alpha chain
4	LuxB	AGU98402.1	Luciferase beta chain
5	LuxE	AGU98403.1	Long-chain-fatty-acid luciferin-component ligase
6	LuxG	AGU97355.1	NAD(P)H-dependent FMN reductase
7	LuxH	AGU97356.1	3,4-dihydroxy-2-butanone 4-phosphate synthase
Sucrose (ScrRAKB operon)
1	ScrR	AGV19641.1	Sucrose operon repressor	V. alginolyticus ATCC 17749	34,33
2	ScrA	AGV19642.1	Type II phosphoenolpyruvate-dependent sucrose phosphotransferase system (PTS)
3	ScrK	AGV19643.1	Fructokinase
4	ScrB	AGV19644.1	Sucrose-6-phosphate dehydrogenase

Syntenic organization and phylogenetic analysis of bioluminescence and sucrose pathway

The syntenic organization of the bioluminescence and sucrose operon was visualized using genome visualization tool EasyFig⁷⁴ with manually curated Genbank file (gbff). Further, protein sequences of these two operons were extracted, aligned with MUSCLE and concatenated using the script catfasta2phyml.pl (https://github.com/nylander/catfasta2phyml). Phylogenetic reconstruction of the lux operon was performed with concatenated LuxCDABEG protein sequences among V. harveyi (6 strains) and V. campbellii (65 strains), carrying complete operon, using the luminescent V. vulnificus ATCC 43382 (VVL1) as the outgroup. For the scrRAKB operon, concatenated protein sequences from 182 strains of V. harveyi, and V. campbellii 170502 were analyzed with V. alginolyticus ATCC 17749 as the outgroup. Two V. harveyi strains (V423 and V424) carried two scr operons, and each operon was analyzed independently. Alignment was performed using MUSCLE, and maximum likelihood (ML) phylogenetic trees were constructed with RAxML-NG⁷⁵ using 1000 bootstrap replicates. The best-fit substitution model (JTT + I + G4) was identified with ModelTest-NG⁷⁶. The phylogenetic tree was visualized using iTOL⁷³.

Mobilome analysis

Mobile genetic elements were analysed in nine bacterial strains suspected for horizontal gene transfer (HGT). These included six V. harveyi strains suspected of acquiring the luminescence operon, two V. harveyi strains carrying two copies of the scr operon, and V. campbellii 170502 carrying scr operon. Prophage regions were predicted using PHASTEST⁷⁷, a high-throughput tool that identifies intact, questionable, and incomplete prophage. Insertion sequences (ISs) and transposable elements were annotated using ISEScan⁷⁸, which employs Hidden Markov Models. The identified transposases were further confirmed by blastP analysis. Integrons and CALINs (Clusters of attC sites Lacking Integron-integrase) were identified using IntegronFinder version 2.0⁷⁹. Integrated conjugative elements (ICEs), which mediate gene transfer via conjugation, were detected using ICEFinder 2.0⁸⁰.

Phenotypic characterization

In silico results were validated by bioluminescence and sucrose utilization assay. Bioluminescence assay was carried out as per our earlier described method². Briefly, V. harveyi and V. campbellii isolates were cultured in 10 ml LB broth (1.5% NaCl) at 30 °C, 120 rpm. A 200 µl bacterial culture (in triplicates) was transferred to a 96-well white opaque microplate (optiplate, PerkinElmer) and luminescence was recorded at 12 h post-incubation (Spark 10 M, Tecan, Grödig, Austria). Values were expressed as fold change of luminescence relative to the negative control.

Sucrose utilization was initially assessed on TCBS agar (30 °C, overnight), with yellow colonies scored as sucrose-fermenter and green colonies as non-fermenter. This was further confirmed by growth kinetics in M9 medium containing sucrose (10 g L⁻¹) as the sole carbon source⁸¹. In brief, culture was grown in LB broth to OD₆₀₀ ≈ 0.5, centrifuged (6000 g, 10 min), and resuspended in an equal volume of M9 broth. A 5 µl inoculum was transferred to 1 ml M9-sucrose in sterile 24-well microtitre plate (Tarson, India) and incubated statically at 30 °C in triplicates. The M9-sucrose broth without inoculum served as negative control and culture with V. alginolyticus served as positive control. The OD₆₀₀ was recorded at regular intervals using multimode reader (Spark 10 M, Tecan, Grödig, Austria).

Screening of luxA and scrA genes

In silico results were further verified by PCR screening of marker gene of bioluminescence (luciferase gene luxA) and sucrose uptake gene scrA. Primers were designed using IDT PrimerQuest tool (https://www.idtdna.com/pages/tools/primerquest) (Table 5). The luxA and scrA genes were screened using bacterial DNA extracted by DNeasy blood and tissue kit (Qiagen). The PCR reaction condition consisted of initial denaturation (95 °C, 5 min) followed by 35 cycles of denaturation (95 °C, 30 s), annealing (60 °C, 30 s), and extension (72 °C, 30 s) with final extension of 72 °C for 10 min.

Statistical analysis

Comparative genomic analysis was conducted to evaluate differences in GC content, genome size, and the number of protein-coding genes between V. harveyi and V. campbellii. A two-tailed independent sample t-test was performed to assess significant differences in these genomic features. Furthermore, V. campbellii strains were categorized into three distinct phylogenetic groups; Gr-L (luminescent), Gr-LD (luminescence-defective), and Gr-LM (luminescent strains harboring mobile genetic elements), based on core genome phylogeny and the distribution of the luminescence operon. A one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison was conducted to evaluate differences in GC content, genome size, and the number of protein-coding genes among the three groups. All statistical analysis was carried out in R4.0.2. The graphics were generated using GraphPad Prism version 7.0 (GraphPad Software, San Diego, CA, USA).

During the preparation of this manuscript, the authors used the web-based tool ChatGPT (version 5) to improve English language and rectify grammatical errors. Following its use, the authors carefully reviewed and edited the content, and take full responsibility for content of the publication.

Author contributions

Sujeet Kumar: Conceptualization, Whole genome sequencing, Genomic and in silico analysis, Writing—original draft B. Nishanthini: Bacterial isolation, Phenotypic characterization, PCR screening Abhaya Robinson: Bacterial isolation, Phenotypic characterization, PCR screening T. Sathish Kumar: Whole genome sequencing and editing of the manuscript Vidya Rajendran: Provided many bacterial isolates used for phenotypic characterization Vinay Kumar Katneni: Phylogenetic analysis and data interpretation P.S. Shyne Anand: Editing of the manuscript M. Makesh: Project administration, Funding support, Review & Editing M.S. Shekhar: Editing of the manuscript.

Funding

This work was carried out under the Consortia Research Platform on Vaccines and Diagnostics funded by the Indian Council of Agricultural Research, New Delhi, India under the grant number AS/2/3/2022-ASR-IV(PI.Xe-233289).

Data availability

The accession numbers can be found in the article in Table 1 (https:/www.nature.com/articles/s41598-025-09812-3) and supplementary Table 1. The whole genome sequencing carried out in the present study are available in GenBank, under the bioproject PRJNA377806, PRJNA1177612, PRJNA1178820, PRJNA972608. The rpoD gene was used for strain identification and its accession number is MW428977 to MW429005, PV550895 to PV550903, PX116707 to PX116711. The primary BLAST output for lux and scr operon, average nucleotide identity score, mobilome analysis, including insertion sequence, integrative conjugative elements, integron and prophage elements, are presented in supplementary Tables 2 to 12.

Declaration

Competing interests

The authors declare no competing interests.