Integrative Genomics and Precision Breeding for Stress-Resilient Cotton: Recent Advances and Prospects

Abstract

Developing climate-resilient and high-quality cotton cultivars remains an urgent challenge, as the key target traits yield, fibre properties, and stress tolerance are highly polygenic and strongly influenced by genotype–environment interactions. Recent advances in chromosome-scale genome assemblies, pan-genomics, and haplotype-resolved resequencing have greatly enhanced the capacity to identify causal variants and recover non-reference alleles linked to fibre development and environmental adaptation. Parallel progress in functional genomics and precision genome editing, particularly CRISPR/Cas, base editing, and prime editing, now enables rapid, heritable modification of candidate loci across the complex tetraploid cotton genome. When integrated with high-throughput phenotyping, genomic selection, and machine learning, these approaches support predictive ideotype design rather than empirical, trial-and-error breeding. Emerging digital agriculture tools, such as digital twins that combine genomic, phenomic, and environmental data layers, allow simulation of ideotype performance and optimisation of trait combinations in silico before field validation. Speed breeding and phenomic selection further shorten generation time and increase selection intensity, bridging the gap between laboratory discovery and field deployment. However, the large-scale implementation of these technologies faces several practical constraints, including high infrastructural costs, limited accessibility for resource-constrained breeding programmes in developing regions, and uneven regulatory acceptance of genome-edited crops. However, reliance on highly targeted genome editing may inadvertently narrow allelic diversity, underscoring the need to integrate these tools with broad germplasm resources and pangenomic insights to sustain long-term adaptability. To realise these opportunities at scale, standardised data frameworks, interoperable phenotyping systems, robust multi-omic integration, and globally harmonised, science-based regulatory pathways are essential. This review synthesises recent progress, highlights case studies in fibre, oil, and stress-resilience engineering, and outlines a roadmap for translating integrative genomics into climate-smart, high-yield cotton breeding programmes.

1. Introduction

Cotton (Gossypium spp.), the world’s most important natural fibre crop, underpins the global textile industry while supporting the livelihoods of millions of smallholder households across more than 90 countries, predominantly in subtropical and semi-arid agroecologies [1,2]. Upland cotton (Gossypium hirsutum L.) dominates cultivation, contributing ~90–95% of global fibre production [3]. In 2024–25, cotton world production was estimated at 120.96 million bales, with forecasts for 2025–26 projecting a decline to ~116.7 million bales, while mill consumption is expected to rise to a five-year high of ~118.1 million bales [4,5]. Cottonseed, accounting for ~15% of crop value, contains 15–20% oil and ~23% protein, contributing to both farm profitability and the bioeconomy through applications in edible oil, animal feed, and biofuel industries [6].

Despite its global significance, cotton production is increasingly constrained by multiple biotic and abiotic stresses. Biotic stressors, including Verticillium and Fusarium wilt, bollworms, whiteflies, and cotton leafroll dwarf virus, severely affect crop health and productivity. Abiotic factors such as drought, salinity, and heat further reduce yield and fibre quality, particularly during the flowering and boll maturation stages [7]. Yield gains derived from monogenic Bt traits have plateaued due to pest adaptation, while repeated selection within narrow germplasm pools has exacerbated genetic bottlenecks and reduced resilience [8]. Conventional breeding approaches including hybridisation, phenotypic selection, and backcrossing have delivered notable improvements but are hindered by long development timelines (10–15 years), high labour intensity, and erosion of genetic diversity during the creation of homozygous lines and multi-site testing [9]. Compounding these challenges, the allotetraploid cotton genome (~2.5 Gb), characterised by extensive duplication and structural complexity, has long impeded high-resolution mapping and trait introgression [10].

Beyond yield and fibre quality, cotton improvement must increasingly align with sustainability goals and farmer livelihoods. Cultivars resilient to heat, drought, and pests are vital for sustaining productivity in smallholder systems especially across Africa, Asia, and Latin America, where resource constraints and climate risks are acute [11]. Genomics-assisted breeding and genome editing expedite the deployment of stress-resilience alleles, help reduce reliance on agrochemicals and irrigation, and accelerate genetic gain via predictive breeding [12,13]. By anchoring these innovations in frameworks of ecological intensification and equitable access, genomic cotton improvement can contribute to sustainable development, improved income stability, and resilience under climate uncertainty [14].

Genomics-assisted breeding (GAB) has emerged as a transformative solution to overcome these constraints. Advances in high-throughput genotyping, linkage mapping, genome-wide association study (GWAS), and dense single-nucleotide polymorphism (SNP) arrays now allow early selection for complex traits such as yield, fibre quality, abiotic tolerance, and seed composition [15]. The establishment of reference genomes, coupled with resequencing of genomes and transcriptomes over 700 accessions, has enabled fine mapping of trait loci and discovery of elite alleles from diverse germplasm [16]. In addition, advances in pangenomics and graph-based genome representations have broadened our understanding of structural variation and allelic diversity across cotton germplasm. The development of high-quality chromosomal reference genomes has further revealed introgressed segments and structural variations associated with key fibre traits. These genomic resources now underpin genomic selection and haplotype-based breeding, which enable genome-wide prediction of breeding values and accelerate the improvement of complex quantitative traits such as fibre elongation, micronaire, and yield stability while maintaining genetic diversity [17].

Alongside GAB, development of precision genome editing approaches such as CRISPR/Cas has rapidly transformed cotton biotechnology. Cas9-mediated editing has been applied to key genes such as GhMYB25-like, while newer Cas12a orthologues (e.g., Mb2Cas12a) deliver >90% editing efficiency under variable conditions, with relaxed PAM requirements expanding the accessible target space [18,19]. Multiplex editing of loci such as GhPGF has generated glandless cotton suitable for edible seeds, while base and prime editing now allow precise, transgene-free single-nucleotide substitutions [20]. Nevertheless, the size and redundancy of the cotton genome necessitate refined vector design, rigorous off-target prediction and minimisation strategies, and genotype-specific transformation approaches.

Speed breeding has further accelerated breeding timelines. By exploiting controlled environments, extended photoperiods, and embryo rescue, cycle duration can be reduced to 71–85 days, enabling up to 3–5 generations per year [21]. When integrated with GAB and genome editing, speed breeding facilitates the rapid pyramiding of resilience and quality traits, as demonstrated in elite lines such as JND24-i3 [22]. However, adoption is constrained by high energy costs, genotype-specific responses, and scalability challenges for smallholder systems [23].

Looking forward, initiatives such as Cotton 2035 seek to integrate multi-omics (genome, transcriptome, epigenome), AI-driven phenomic modelling, and functional genomics, supported by robust public–private partnerships, to deliver climate-resilient, fibre- and oil-efficient cultivars (Figure 1) [24]. In this review, we summarised recent breakthroughs in genomics-assisted breeding, genome editing, and speed breeding, and critically assessed their integration into modern cotton pipelines. We also highlight case studies such as glandless seed production, drought tolerance, and fibre quality improvement and propose a roadmap towards the next generation of ideotypes that unite productivity, fibre excellence, stress resilience, and sustainability in smallholder systems exposed to climate uncertainty.

2. Genomic Characterisation and Utilisation of Cotton Germplasm for Pre-Breeding and Genetic Improvement

Harnessing and characterising the extensive genetic diversity conserved within global cotton germplasm repositories are fundamental to broadening the genetic base and accelerating crop improvement under climate stress. However, the potential of these resources remains largely underexploited due to limited multi-environmental phenotyping, incomplete genomic annotation, and poor integration of omics data for complex traits such as fibre yield, quality, and abiotic stress tolerance [25]. These polygenic traits are strongly influenced by genotype–environment interactions, requiring integrated genomic, phenotypic, and environmental datasets to uncover functionally relevant allelic variation [26]. Comprehensive molecular characterisation of large germplasm collections is often labour-, time-, and cost-intensive, constraining their routine use in breeding pipelines [27]. Recent advances in high-throughput sequencing, molecular markers, and automated phenotyping have revolutionised germplasm utilisation by enabling genome-wide diversity characterisation, haplotype discovery, and allele mining. These technologies have transformed conventional germplasm banks from static repositories into dynamic, data-driven platforms for pre-breeding and translational genomics. The integration of genomic data with phenotypic and environmental information is now allowing breeders to prioritise accessions containing favourable alleles for stress resilience, fibre quality, and yield stability, paving the way for genomics-assisted germplasm deployment in precision breeding.

A widely adopted strategy to improve germplasm utilisation efficiency is the development of core and minicore collections, which are curated and genetically representative subsets capturing the essential diversity of larger repositories [28,29]. These reduced panels enable intensive phenotypic and molecular evaluation of a manageable number of accessions while maintaining genetic breadth, thereby facilitating the efficient dissection of complex traits [30]. Major repositories including the USDA National Cotton Germplasm Collection and extensive holdings in China, India, Australia, Uzbekistan, and Europe preserve over 15,000 accessions of cultivated and wild Gossypium species [27,31,32]. From these resources, the USDA developed the Gossypium Diversity Reference Set (GDRS), encompassing ~20% of G. hirsutum accessions selected to maximise geographic and taxonomic diversity [33]. This reference panel has been comprehensively genotyped using SSR and SNP markers and phenotyped for fibre and seed traits, providing a widely used resource for allele discovery, genomic prediction, and comparative studies [34].

The functional value of these germplasm sets has been significantly enhanced by the advent of whole-genome resequencing, pangenomics, and genome-wide association studies (GWAS) [35]. For example, resequencing of 419 upland cotton cultivars (~6.5× coverage) revealed ~3.7 million SNPs, enabling the identification of 7383 trait-associated loci distributed across 4820 genes, including novel candidates related to flowering time, fibre length, and fibre strength [36]. Such large-scale SNP datasets now underpin marker-assisted selection (MAS) and genomic selection (GS) pipelines, accelerating the identification and pyramiding of favourable alleles for yield, quality, and stress resilience. Likewise, resequencing of 500 G. hirsutum accessions uncovered key alleles linked to fibre-quality enhancement, demonstrating the utility of diversity panels for allele mining and genomic prediction [37]. Complementary graph-based pangenome analyses have revealed extensive structural variants associated with fibre traits, offering a more complete representation of genomic diversity and structural polymorphism in cotton [38]. Together, these efforts enhance our understanding of cotton’s genomic architecture and provide actionable targets for designing climate-resilient ideotypes.

To broaden the genetic base and alleviate domestication bottlenecks, researchers have created synthetic amphiploids and chromosome segment substitution lines (CSSLs) carrying alleles from wild Gossypium species. For instance, introgression of approximately seventy-two percent of the G. anomalum genome into G. hirsutum generated seventy-four CSSLs, revealing twenty-four QTLs linked to plant height, drought tolerance, and fibre strength, and enabling the cloning of functionally significant alleles such as peroxiredoxin and callose synthase 8 (GhCalS8) [39]. Similar CSSL populations derived from wild donors have identified novel QTLs for stress tolerance and fibre quality, confirming the value of exotic germplasm for broadening the cotton gene pool [40]. Likewise, synthetic polyploids integrating wild genome segments have improved stress tolerance and fibre performance in elite upland cotton varieties [41]. Global breeding programmes are now routinely exploiting these resources through marker-assisted introgression and backcrossing, successfully integrating Verticillium wilt resistance, drought tolerance, and superior fibre traits into elite cultivars [42].

Taken together, these curated germplasm resources, synthetic polyploids, and CSSL populations constitute powerful platforms for allele mining, QTL discovery, and genomics-assisted breeding. Their strategic integration with speed breeding, GWAS, MAS, and genomic selection has markedly expanded the usable genetic diversity in modern cotton improvement pipelines [43,44]. Overall, genomically characterised germplasm encompassing core collections, diversity panels, synthetic amphiploids, and wild-derived CSSLs forms the foundational resource base for next-generation precision and climate-resilient cotton breeding.

3. Genome Sequencing Paving the Way for Molecular Marker Identification for Precision Breeding

Genome sequencing has become the cornerstone of cotton research, providing the resolution required to unravel the complexity of its polyploid genome, trace domestication events, and link structural diversity with fibre properties, stress resilience, and breeding potential. The availability of high-quality reference genomes, complemented by diverse molecular marker platforms, now underpins evolutionary studies, QTL mapping, GWAS, and genomic prediction frameworks. These integrated resources have laid the foundation for precision breeding, enabling the systematic incorporation of genomic information into modern improvement pipelines (Figure 2). The following sections provide an overview of recent breakthroughs in genome sequencing technologies and molecular marker systems that have transformed cotton genetics and accelerated the development of elite, climate-resilient cultivars.

4. Cotton Genome Sequencing Projects

Plant species prioritised for sequencing are typically selected based on economic significance, genome size, the availability of genetic and physical maps, transcriptomic datasets, and strong research-community support [45]. Sequence-based assemblies provide high-resolution representations of genomes, revealing chromosomal positions of genes, regulatory elements, full nucleotide sequences, and predicted functions [46]. Accordingly, high-quality reference genomes have become essential foundations for evolutionary biology, functional genomics, and crop improvement [47]. This rationale has driven numerous sequencing projects, beginning with rice (Oryza sativa) in 2002 [48].

Cotton (Gossypium spp.) has been a central focus of genomics due to its polyploid complexity and economic importance. The genus comprises eight diploid genome groups (A–G and K) and one allopolyploid group [49]. Genome size estimates are available for 37 of the ~50 Gossypium species, spanning both diploid and polyploid groups, and revealing striking interspecific variation [50]. Whole-genome sequencing (WGS) has become pivotal to evolutionary studies, QTL discovery, functional gene identification, and population genomics [51]. Cotton also serves as a model for polyploidisation, genome-size evolution, and unicellular fibre development [52].

A landmark contribution came Fang et al. (2017) [51], who re-sequenced 147 accessions, including wild relatives, landraces, and cultivars, generating a comprehensive genomic variation map and demonstrating dual domestication processes in G. hirsutum and G. barbadense. The first cotton genome sequenced was the diploid D-genome species G. raimondii (2012), estimated at ~740 Mb, though the draft assembly contained thousands of gaps and was ~60% repetitive [53,54]. The A-genome diploid G. arboreum followed in 2016, assembled into a fragmented draft with limited contiguity but showing comparable gene content to its D-genome counterpart [55].

With both diploid progenitors sequenced, the next milestone was the assembly of tetraploid genomes. Draft assemblies of G. hirsutum (TM-1) and G. barbadense were published in 2015, constructed primarily from Illumina reads [56,57]. Although fragmented, they confirmed strong collinearity between the At and Dt subgenomes and their diploid progenitors. Subsequent advances in long-read sequencing (PacBio), optical mapping (BioNano), and Hi-C scaffolding enabled near-chromosome-scale assemblies, resolving repeat-rich regions and markedly improving annotation. For example, Wang et al. (2019) and Chang et al. (2024) produced chromosome-scale assemblies of G. hirsutum and G. barbadense, identifying ~70,000 annotated genes, including lineage-specific genes associated with phenotypic diversity [58,59]. Updated assemblies of G. arboreum, G. herbaceum, and G. raimondii now provide robust frameworks for comparative genomics.

In parallel, chromosome-scale assemblies for elite cultivars such as ‘UGA230’, ‘UA48’, and ‘CSX8308’, together with an updated TM-1 reference genome, have revealed introgressed segments and structural diversity underpinning fibre-quality traits [60]. The most recent sequencing projects have expanded coverage across multiple tetraploid cottons, delivering chromosome-level assemblies ranging from ~1.5 Gb (A-genome diploids) to ~2.3 Gb (tetraploids), each containing tens of thousands of annotated genes [61]. The establishment of centralised databases such as CottonGen [62] and CottonFGD [63] has further facilitated access to genomic data and analysis pipelines. Comparative analyses continue to uncover major differences in gene content, repeat organisation, and transposable element activity, offering insights into evolutionary trajectories and functional complexity.

Table 1. Major milestones in cotton genome sequencing, grouped by diploid references, tetraploid references, and pangenome resources.

Group	Species/ Accession	Genome Type	Year	Assembly Size (Gb)	N50 (Contig/Scaffold)	Technology Used	Key Contributions
Diploid references	G. raimondii (D5)	Diploid	2012	~0.74	44.9–135.5 kb	Illumina short reads	First cotton genome; basis for comparative genomics
	G. arboreum (A2)	Diploid	2016	~1.7	72 kb	Illumina short reads	Insights into A-genome evolution and fibre traits
Tetraploid references	G. hirsutum (TM-1)	Tetraploid	2015	~2.3	<1 Mb (fragmented)	Illumina short reads	First tetraploid draft; confirmed At–Dt subgenome collinearity
	G. barbadense	Tetraploid	2015	~2.3	<1 Mb (fragmented)	Illumina short reads	Genomic basis for superior fibre quality
	G. hirsutum (updated)	Tetraploid	2019	~2.3	>10 Mb	PacBio, BioNano, Hi-C	Chromosome-scale assembly; improved annotation
	G. barbadense (updated)	Tetraploid	2019	~2.3	>10 Mb	PacBio, BioNano, Hi-C	Repeat resolution; functional gene discovery
Pangenome resources	Multi-accession panels	Mixed	2022–2024	1.5–2.3	Near-chromosome scale	Hybrid assembly + Hi-C + ONT	Structural variation, PAVs, CNVs, haplotype diversity

summarises major milestones in cotton genome sequencing, from diploid and tetraploid references to recent pangenome resources.

Together, cotton genomics has advanced from fragmented drafts to chromosome-scale reference assemblies, which now serve as anchors for understanding genome structure, evolution, and gene function. However, single-reference genomes capture only a fraction of the genetic diversity present across cultivated and wild germplasm. To address this limitation, pangenome approaches have emerged, integrating multiple high-quality assemblies to represent the full spectrum of genomic variation within a species. In cotton, these developments have provided unprecedented insights into genomic diversity, structural variation, and breeding potential. Early efforts built multi-accession reference panels for G. hirsutum and G. barbadense, revealing extensive presence–absence variations (PAVs), copy-number variations (CNVs), and large structural rearrangements invisible to single-genome approaches [38]. More recent graph-based pangenome assemblies, encompassing over ten accessions, have captured haplotype diversity across geographically distinct germplasm pools, enabling fine mapping of fibre-quality loci and adaptive alleles [64,65]. These resources have also revealed lineage-specific genes contributing to stress resilience and fibre elongation, while supporting the construction of haplotype-aware genomic prediction models. Advances in cotton reference genomes and pangenomes have clarified the relationships between genomic diversity and complex trait variation, establishing a robust foundation for precision breeding strategies that integrate genomics with genome editing, predictive modelling, and systems biology. Building on these genomic resources, molecular marker technologies now provide the tools to translate sequence diversity into practical breeding applications.

5. Emerging Pangenomic Frameworks for Structural Variation Analysis

Following the generation of high-quality reference genomes, cotton genomics is advancing from single-reference assemblies to comprehensive pangenomic frameworks that capture population-level variation across Gossypium species and cultivars [66]. Unlike linear references, which can mask structural polymorphisms, pangenomes constructed from multiple de novo assemblies and extensive resequencing datasets provide an inclusive view of genomic diversity. In cotton, these efforts have revealed abundant structural variants (SVs) including PAVs, CNVs, inversions, and translocations many of which are associated with fibre quality, stress tolerance, and disease resistance. Population-scale SV discovery has been enabled by long-read sequencing, Hi-C scaffolding, and optical mapping, which together produce chromosome-scale assemblies and robust SV detection [37,60]. These integrated platforms substantially improve the resolution of repeat-rich genomic regions and reveal variant types that remain inaccessible to short-read and single-reference analyses.

Building on these advances, recent population-level analyses have uncovered the genomic consequences of domestication and modern crop improvement. Early pangenome studies demonstrated that domestication and breeding left substantial portions of sequence and gene content missing from single-reference genomes. For instance, large-scale population pangenomic analyses have revealed thousands of previously uncharacterised, non-reference genes and extensive presence–absence variations (PAVs) linked to fibre quality, yield, and stress adaptation [58,67]. These discoveries highlight how gene presence–absence dynamics influence stress-response pathways and cell-wall biosynthesis, contributing to phenotypic variation and selection signatures. Subsequent multi-reference assemblies and deep resequencing have refined catalogues of structural variants (SVs) and clarified their agronomic relevance. High-quality assemblies for modern G. hirsutum lines that integrate long-read and Hi-C data have resolved complex repeats, enabled comprehensive SV discovery, and facilitated candidate-gene identification for fibre traits [60]. Comparative analyses across elite cultivars and landraces further revealed that SVs, more than SNPs, often underpin domestication-related phenotypes [68]. Looking ahead, graph-based pangenomes that preserve alternative haplotypes and local rearrangements are expected to transform variant discovery and association analyses. These frameworks will allow GWAS and genomic prediction models to incorporate non-reference alleles [69]. Although graph-enabled association analysis has been demonstrated in other major crops, its broader application in cotton remains at an early stage. Immediate priorities include constructing graph pangenomes from newly available chromosome-scale assemblies and benchmarking variation-aware GWAS for fibre and adaptation traits using cotton-specific datasets [70]. As cotton genomics enters an era of data-intensive precision breeding, graph-based pangenomes will serve as pivotal resources linking structural variation to predictive analytics. They will provide the foundation for connecting genomic diversity with phenomic, environmental, and physiological datasets, ultimately enabling in silico ideotype optimisation and trait simulation through AI-driven digital twins and multi-omic integration pipelines.

Crucially, integrating pangenomic variation with transcriptomic, epigenomic, proteomic, and metabolomic datasets will enable systems-level dissection of SV-linked gene regulation and metabolic fluxes relevant to fibre development and stress responses. Practically, these resources will support: (i) genomic selection models incorporating SV haplotypes for polygenic traits; (ii) CRISPR/Cas genome editing guided by pangenomic targets; and (iii) ideotype design for climate-resilient, high-quality fibre. By anchoring structural variants and non-reference genes within a population-scale framework, cotton pangenomes not only enhance trait dissection but also create a foundation for genome editing and translational breeding strategies that can accelerate the development of climate-resilient, high-quality cultivars.

6. Progress in Molecular Marker Development for Genetic Linkage Mapping

Genetic linkage mapping has been pivotal in dissecting the genetic architecture of agronomic traits in cotton, providing insights into both polygenic and single-gene loci [71]. The principle relies on tracking recombination events during meiosis to establish marker order and genetic distances [72]. While early morphological markers were constrained by low abundance and strong environmental influence, the advent of molecular markers enabled genome-wide, high-resolution mapping. These advances facilitated the identification of loci underlying fibre quality, yield, and stress resilience, while also accelerating marker-assisted breeding within cotton’s genetically narrow background [73]. The trajectory of marker development illustrates a steady overcoming of prior limitations, progressing from low-throughput, low-polymorphism systems to high-density next-generation sequencing (NGS)-enabled platforms.

The first molecular markers applied to cotton were restriction fragment length polymorphisms (RFLPs), which anchored the earliest linkage maps in tetraploid cotton [74]. RFLPs provided codominant inheritance and broad genome coverage, enabling QTL mapping for example, Mei, Syed [75] identified 14 fibre-quality QTLs in G. hirsutum × G. barbadense populations. However, their reliance on large DNA quantities, labour-intensive protocols, and low polymorphism limited their long-term utility in breeding.

In the 1990s, random amplified polymorphic DNA (RAPD) markers emerged as cost-effective alternatives that required no prior sequence information. They were widely applied for diversity analyses and preliminary QTL detection [76,77]. Nevertheless, their dominant nature and poor reproducibility undermined their effectiveness in polyploid genomes such as cotton, where allelic dosage complicates interpretation.

Amplified fragment length polymorphisms (AFLPs) combined restriction digestion with PCR amplification, generating highly polymorphic, genome-wide markers [78]. They were particularly effective for diversity analyses and for saturating maps in low-polymorphism populations [79,80]. However, like RAPDs, AFLPs were dominant markers with limited cross-laboratory reproducibility, which restricted their broader application in breeding.

A major breakthrough came with simple sequence repeats (SSRs), or microsatellites, which combined codominance, high reproducibility, and multi-allelic variation. These features were particularly valuable given cotton’s limited interspecific polymorphism [81,82]. Large SSR libraries were developed from expressed sequence tags (ESTs), bacterial artificial chromosomes (BACs), and enriched genomic libraries [83,84], while databases such as CottonGen and the Cotton Marker Database (CMD) standardised their application. SSR-based maps facilitated QTL discovery for fibre properties [85,86]; and early maturity [87], and other complex traits. Consensus SSR maps containing thousands of loci expanded genome-wide coverage [88,89]. Despite their success, SSRs lacked the scalability and resolution needed for modern genomic selection compared with SNP-based approaches.

The emergence of SNPs marked a transformative phase. SNPs are the most abundant and stable form of genetic variation, ideal for anchoring polyploid genomes. High-throughput arrays such as the Illumina CottonSNP63K [90] and Cotton80K [91] produced ultra-dense linkage maps, with marker spacing as low as 0.23 cM in interspecific populations, the highest resolution reported to date [92]. These platforms enabled the identification of stable QTLs for fibre, yield, disease resistance, and drought tolerance, as well as allele introgression from wild relatives. Nonetheless, SNP arrays are still constrained by ascertainment bias (variants limited to discovery panels) and cost barriers for resource-limited breeding systems.

Most recently, NGS-based marker systems including genotyping-by-sequencing (GBS), RAD-seq, and whole-genome resequencing have provided unprecedented resolution and scalability. These approaches allow de novo SNP discovery across diverse germplasm, overcoming the bias of fixed arrays. For instance, GBS identified 30 drought-tolerance QTLs and 89 candidate genes introgressed from wild donors [93], while whole-genome resequencing enabled GWAS of cold- and salt-tolerance alleles [94]. Their cost-effectiveness and flexibility now position NGS-derived markers as central to future cotton breeding pipelines.

Parallel to these technological advances, innovative mapping populations such as multi-parent advanced generation inter-cross (MAGIC) and nested association mapping (NAM), coupled with bulk segregant analysis sequencing (BSA-seq), have enhanced mapping resolution and gene discovery [95,96]. MAGIC populations have disentangled polygenic and epistatic effects, exemplified by recent studies resolving complex Verticillium wilt resistance loci [97].

Taken together, molecular marker development in cotton has progressed from labour-intensive, low-resolution tools to NGS-enabled, high-throughput systems that integrate seamlessly with genomics-assisted breeding. These platforms underpin marker-assisted selection (MAS), marker-assisted recurrent selection (MARS), and genomic selection, which are already accelerating cultivar development [81,98,99]. Looking forward, the convergence of high-resolution mapping with predictive genomics and machine learning will ensure efficient translation of genomic insights into resilient, high-yielding cotton cultivars. A consolidated overview of high-density linkage maps, marker systems, and associated traits is provided in Table 2, serving as a bridge to the next section on genomics-assisted breeding strategies.

Table 2. High-density genetic linkage maps in cotton: Populations, marker systems, and associated traits.

Population/Panel	Genotyping Assay	N (Mapping)	Mapped Markers	Map Distance (cm)	QTLs/Loci	Trait(s)	Reference
Multiple populations (meta-QTL synthesis)	Mixed (SSR/AFLP/SNP; published QTLs)	—	—	—	300+ QTLs integrated	Abiotic & biotic stress (VW, FW, RKN, drought, salt)	[6,100]
GWAS panel (diverse Upland cotton)	CottonSNP63K array (Illumina)	120 accessions	63,000 SNPs (genotyped)	—	5 QTLs/18 lead SNPs	Verticillium wilt resistance	[101,102]
RIL (G. hirsutum × G. barbadense)	High-density SNP (GBS + array)	not stated	high-density map	2865	119 QTLs (DI & incidence)	Verticillium wilt resistance	[103]
BC1/BIL population	SSR (~175–400)	94–200	175–400	2895–4403	10–17 QTLs	Verticillium wilt/heat tolerance	[97]
RIL (AS2 × MCU-13)	GBS + SSRs	48	1898	2808.3	19 QTLs	Drought tolerance (chlorophyll stability index, proline, etc.)	[93]
GWAS panel (US Upland cotton)	CottonSNP63K	376	26,301 polymorphic SNPs	—	53 (drought), 78 (salt), 8 (thrips)	Seedling-stage drought, salt, thrips	[104]
F2:3 interspecific (G. tomentosum × G. hirsutum)	GBS (~93k SNPs)	277	93,384 SNPs; 26 LGs	—	QTL clusters	Salt tolerance (seedling stage)	[105]
Introgression lines from wild donor (BC-derived)	GBS/resequencing	—	dense SNPs	—	30 stable QTLs + 89 candidates	Drought-tolerance physiological traits	[106]
Multiple populations (NemX, SJ-2, Pima S-7, etc.)	SSRs linked to rkn1	multiple	—	—	major-effect QTL(s)	Root-knot nematode resistance	[107,108]
Natural panel (200 accessions)	SLAF-seq + GWAS	200	~10,000 SNPs	—	significant SNPs; GhSAD1	Cold tolerance (ABA-mediated)	[94]
BSA-seq (extreme pools)	BSA-seq	pools	—	6.27 Mb (physical)	3 QTL intervals (A13, A10, A07)	Drought resistance (PH, SBW)	[109,110]

Table 2. High-density genetic linkage maps in cotton: Populations, marker systems, and associated traits.

Population/Panel	Genotyping Assay	N (Mapping)	Mapped Markers	Map Distance (cm)	QTLs/Loci	Trait(s)	Reference
Multiple populations (meta-QTL synthesis)	Mixed (SSR/AFLP/SNP; published QTLs)	—	—	—	300+ QTLs integrated	Abiotic & biotic stress (VW, FW, RKN, drought, salt)	[6,100]
GWAS panel (diverse Upland cotton)	CottonSNP63K array (Illumina)	120 accessions	63,000 SNPs (genotyped)	—	5 QTLs/18 lead SNPs	Verticillium wilt resistance	[101,102]
RIL (G. hirsutum × G. barbadense)	High-density SNP (GBS + array)	not stated	high-density map	2865	119 QTLs (DI & incidence)	Verticillium wilt resistance	[103]
BC1/BIL population	SSR (~175–400)	94–200	175–400	2895–4403	10–17 QTLs	Verticillium wilt/heat tolerance	[97]
RIL (AS2 × MCU-13)	GBS + SSRs	48	1898	2808.3	19 QTLs	Drought tolerance (chlorophyll stability index, proline, etc.)	[93]
GWAS panel (US Upland cotton)	CottonSNP63K	376	26,301 polymorphic SNPs	—	53 (drought), 78 (salt), 8 (thrips)	Seedling-stage drought, salt, thrips	[104]
F2:3 interspecific (G. tomentosum × G. hirsutum)	GBS (~93k SNPs)	277	93,384 SNPs; 26 LGs	—	QTL clusters	Salt tolerance (seedling stage)	[105]
Introgression lines from wild donor (BC-derived)	GBS/resequencing	—	dense SNPs	—	30 stable QTLs + 89 candidates	Drought-tolerance physiological traits	[106]
Multiple populations (NemX, SJ-2, Pima S-7, etc.)	SSRs linked to rkn1	multiple	—	—	major-effect QTL(s)	Root-knot nematode resistance	[107,108]
Natural panel (200 accessions)	SLAF-seq + GWAS	200	~10,000 SNPs	—	significant SNPs; GhSAD1	Cold tolerance (ABA-mediated)	[94]
BSA-seq (extreme pools)	BSA-seq	pools	—	6.27 Mb (physical)	3 QTL intervals (A13, A10, A07)	Drought resistance (PH, SBW)	[109,110]

7. Advances of Genomics Approach in Marker-Assisted Breeding

The maturation of molecular marker technologies has transformed genetic discoveries into practical breeding tools through marker-assisted selection (MAS). In cotton, where complex polyploidy, narrow genetic diversity, and strong genotype × environment interactions have historically constrained genetic gain, MAS provides a precise complement to conventional phenotype-based selection [111]. By enabling breeders to track desirable alleles directly at the DNA level, MAS bridges the gap between fundamental gene discovery and applied cultivar development.

The early phases of MAS in cotton highlighted both its promise and limitations. Morphological and biochemical markers, once the mainstay of selection, were rapidly superseded due to low resolution, environmental sensitivity, and inability to capture polygenic variation [112,113]. The advent of SSRs and SNP arrays provided reproducible, codominant, and genome-wide markers, facilitating the mapping and introgression of major resistance loci. For example, SSR-based MAS enabled the pyramiding of QTLs for Verticillium wilt and Fusarium wilt resistance into elite G. hirsutum cultivars, delivering durable multi-pathogen defence [114,115]. Similarly, high-density SNP arrays such as CottonSNP63K and Cotton80K facilitated the discovery of fibre-quality QTLs and their incorporation into breeding pipelines, directly improving fibre length, strength, and fineness [90,116].

Beyond disease and fibre traits, MAS has increasingly targeted abiotic stress resilience. Drought- and salinity-tolerance QTLs have been mapped through SNP- and GBS-based platforms and validated for early-generation selection, enabling the rapid accumulation of alleles for osmotic adjustment, chlorophyll stability, and root-system architecture [117,118,119]. These successes underscore how MAS can shorten breeding cycles, reduce reliance on costly phenotyping, and deliver genetic gains even in stress-prone environments. In Uzbekistan, MAS was successfully applied to fibre-quality breeding, where SSR markers tightly linked to a multi-trait QTL cluster on chromosome 7/16 enabled the development of the ‘Ravnaq’ cultivar series with enhanced fibre uniformity, strength, length, and micronaire [120]. Likewise, in Pakistan, MAS was deployed against cotton leaf curl virus (CLCuV), resulting in the release of resistant cultivars such as CIM-443 and CIM-240, which continue to support farmer resilience in virus-prone regions [121]. These examples demonstrate how MAS accelerates the translation of molecular discoveries into farmer-ready varieties.

Recent innovations are reshaping MAS into a more flexible and scalable framework. The adoption of multi-parent populations such as MAGIC and NAM has enhanced mapping resolution, enabling the capture of minor-effect QTLs and epistatic interactions that underpin yield and stress-complex traits [122]. Furthermore, marker-assisted recurrent selection (MARS) has proven valuable for pyramiding favourable alleles for fibre quality and drought resilience across successive cycles [89,123]. Together, these approaches expand MAS beyond major gene introgression toward the systematic improvement of quantitative, polygenic traits.

Nevertheless, MAS faces inherent constraints that limit its stand-alone application. Most agronomically important traits in cotton such as yield, lint percentage, and fibre uniformity are controlled by multiple small-effect loci with strong environmental dependence [124]. In such contexts, MAS targeting a limited number of QTLs cannot fully capture trait architecture. In addition, high-throughput genotyping costs remain prohibitive for many breeding programmes, particularly in developing countries, and marker–trait associations often fail to replicate across diverse environments or genetic backgrounds [125]. Thus, MAS is best positioned as a transitional platform rather than a terminal solution. Its validated markers and trait-linked loci provide the backbone for advanced frameworks including genomic selection (GS), genome editing, and AI/ML-driven predictive breeding. By integrating MAS with GS and functional genomics, breeders can simultaneously exploit large-effect loci through targeted introgression while capturing polygenic variation via genome-wide prediction. Compared with conventional breeding, genomic approaches substantially increase selection accuracy and genetic gain per unit time by enabling early-generation selection based on genome-wide marker effects rather than late-stage phenotyping. Conventional breeding cycles in cotton typically require 8–10 years to achieve varietal release, whereas genomic selection and integrated multi-omics pipelines can reduce this to 4–6 years by predicting breeding values before field evaluation [126]. Furthermore, genomic prediction models maintain higher efficiency under complex genotype × environment interactions, where conventional selection often loses accuracy due to environmental noise. Nonetheless, conventional breeding remains indispensable for large-scale field validation and capturing epigenetic and microenvironmental effects that current genomic models cannot fully explain [127]. Thus, genomic approaches complement rather than replace conventional methods, creating a hybrid framework where empirical and predictive strategies jointly drive genetic improvement and climate resilience in cotton [128,129].

Despite these advances, it is crucial to acknowledge methodological biases that may confound genomics-assisted breeding in cotton. Many high-density SNP arrays (e.g., CottonSNP63K, CottonSNP80K) are developed from a restricted discovery panel of elite cultivars and therefore are subject to ascertainment bias, resulting in under-representation of rare alleles and variants from landraces or wild relatives [130]; You et al. (2018) [131] likewise discuss how array design inherently skews allele-frequency distributions in polyploid species.

The process of SNP-array design and filtering further intensifies these distortions (e.g. as shown by Geibel et al., 2021 [132]). Moreover, associations discovered via GWAS or QTL mapping under specific environments or genetic backgrounds often fail to replicate or transfer in different populations or under stress conditions, especially due to strong genotype × environment (G × E) interactions and population structure. As mitigation strategies, it is increasingly common to complement array-based genotyping with whole-genome resequencing or genotyping-by-sequencing of diverse panels, perform cross-population and multi-environment validation, and apply meta-GWAS or cross-population genomic prediction frameworks to enhance robustness and generalisability [13].

In addition, the success of moving from MAS to genomic selection and AI/ML-driven predictive breeding ultimately depends on rigorous field validation and gradual operational adoption. Predictive models trained on limited populations or environments can lose accuracy when deployed broadly, underscoring the need for multi-environment trials and cross-population testing before routine use. In cotton and other crops, public and private breeding programmes are beginning to pilot genomic selection under real field conditions—for example, the CGIAR Excellence in Breeding Initiative’s Genomic Prediction Toolbox and multi-location pilots in sorghum, maize and wheat provide a template for embedding prediction models into breeding pipelines [128,133]. Investment in breeder training, interoperable data platforms, and extension partnerships will be essential to move genomic prediction and digital tools from proof-of-concept to routine practice in cultivar development.

8. Biotic Stress Resistance

Biotic stresses caused by fungal, bacterial, viral pathogens, and insect–nematode complexes remain a major constraint on global cotton productivity [134,135]. Of which, Verticillium wilt (VW), Fusarium wilt (FW), bacterial blight, cotton leaf curl disease (CLCuD), root-knot nematodes, and bollworm infestations are particularly destructive, with yield losses ranging from 10% to more than 50% depending on severity and environment. Conventional breeding has contributed to moderate tolerance in elite cultivars, yet the narrow genetic base of G. hirsutum limits durable resistance deployment [136,137]. Consequently, MAS, QTL mapping, and GWAS have become indispensable in identifying and transferring resistance loci into commercial germplasm [135].

To date, over 400 QTLs linked with disease resistance have been reported across the 26 cotton chromosomes [138]. These loci have been mapped using diverse populations including chromosomal substitution lines, recombinant inbred lines (RILs), F₂ families, and multi-parent advanced generation inter-cross (MAGIC) populations, often with high-density marker systems [101,139,140]. A consensus map meta-analysis of the Cotton Marker Database (CMD) highlighted five resistance “hotspots” spanning chromosomes c16–c23, representing conserved genomic regions useful for breeding [141]. Key QTLs and candidate genes linked to Verticillium wilt, Fusarium wilt, and other major pathogens are summarised in Table 3. Collectively, these studies provide molecular markers and validated loci that accelerate marker-assisted and genomic selection for disease-resilient cotton improvement.

Li, Ma [142] identified 17 SNPs and 22 candidate genes for VW resistance through GWAS of 299 accessions with ~85,000 SNPs, while Zhang et al. (2019) detected a strong FW resistance QTL on chromosome c14 linked to B12 alongside VW loci in 550 MAGIC lines [143]. Using the CottonSNP63K array, Abdelraheem, Zhu [140] mapped 15 QTLs for VW and 13 for FW resistance validated with diagnostic SNP markers. Similarly, Bardak, Çelik [103] confirmed VW-associated markers across a global diversity panel. Recent GWAS incorporating multi-environment data further resolved the genetic architecture of VW resistance, identifying robust candidate loci.

MAS has been effectively deployed in transferring resistance across cotton germplasm. Zhao, Liu [144] used marker-assisted backcrossing (MAB) to introgress wilt resistance QTLs from G. barbadense into G. hirsutum. Likewise, Shahbazi, Ghaffarian [145] evaluated 25 Iranian cultivars using SSR markers and identified three loci (DPL405, DPL866, and DPL0022) strongly associated with VW resistance, with multivariate clustering reliably distinguishing resistant from susceptible cultivars. In Pakistan and India, MAS-driven introgression of CLCuD resistance from G. arboreum has already produced regionally adapted lines now used in farmer fields [146,147], demonstrating successful translation into commercial cultivars. Although the limited number of markers in some programmes reduced mapping resolution, such studies provide practical and locally impactful tools for breeding.

The evolution of molecular marker systems in cotton reflects the broader trajectory of crop genetics. Early RFLP-based markers were successfully used for bacterial blight resistance detection [148], though their cost and complexity limited their application. These were gradually replaced by PCR-based markers (SSR, RAPD, AFLP), which provided higher throughput but often with reduced reproducibility [149,150]. The advent of SNP arrays and GWAS platforms greatly increased mapping resolution and accelerated the identification of stable QTLs. More recently, pan-genome assemblies and resequencing of over 700 accessions have enabled the discovery of structural variants, haplotype diversity, and rare alleles underlying biotic stress resistance [151]. These advances provide breeders with a richer catalogue of allelic variation for resistance deployment. Molecular marker-assisted gene pyramiding (MAGP) represents a key strategy to integrate multiple resistance loci simultaneously, thereby accelerating breeding cycles and ensuring durable resistance [152]. In cotton, MAGP has been used to combine wilt resistance with fibre quality traits from multiple donors, expanding the genetic base and enhancing stability [153]. Similar approaches in wheat and rice have produced durable multi-disease resistance, underscoring its cross-crop potential.

Despite this progress, challenges persist. The limited genetic diversity of cultivated G. hirsutum restricts the available resistance pool, while marker transferability across populations often requires further validation [154]. Many resistance traits display polygenic architecture, demanding integration of genomic prediction and machine learning approaches to complement MAS. Furthermore, the rapid evolution of pathogens, such as new virulent races of Verticillium dahliae, necessitates continuous allele mining. Looking ahead, integrating multi-omics layers (genome, transcriptome, epigenome, metabolome) with AI-driven prediction models will refine resistance gene discovery and improve accuracy in breeding pipelines. Combined with genomic selection and genome editing, MAS and MAGP are expected to play a central role in delivering cotton cultivars with durable, broad-spectrum, and climate-resilient biotic stress resistance.

Table 3. Key Biotic-Stress Resistance Loci in Cotton and Mapping Evidence.

Pathogen/Stress	Key QTLs/Candidate Genes	Population Type	Marker System/Platform	Main Finding/Significance	Reference
Verticillium wilt (VW)	QTL clusters on c16 and c19	376 upland cotton accessions (GWAS panel)	CottonSNP63K array (high density)	Identified 15 VW and 13 FOV4 QTLs; clusters on c16 and c19 were consistent across environments	[153,154,155]
Verticillium wilt (VW)	Major QTLs on c16 (D7) and c23 (D9)	Backcross Inbred Line (BIL) population	SSR-based linkage map (~2895 cM, 392 SSRs)	Ten VW QTLs and 28 clusters identified; hotspots correlated with NBS gene density	[143,156]
Verticillium wilt (VW)	SNPs on A01 and A10 (GhAMT2 candidate)	Multi-parent Advanced Generation Inter-cross (MAGIC) population	GWAS integrated with transcriptomics	Major QTLs identified; GhAMT2 implicated in VW response and resistance signalling	[157,158]
Cotton blue disease (CBD)	Cbd gene on chromosome 10 (0.75 cM)	ΔOpal F2 families	SSR marker (DC20027) + SNP haplotype analysis	Early example of MAS: trait tagging via haplotype-based SNP markers for CBD resistance	[159]
Cotton bacterial blight (CBB)	GhSWEET10 (TAL effector target)	Pathogen–cotton interaction study	Transcriptomics + functional validation	GhSWEET10 induced by TAL effector Avrb6; gene silencing reduced susceptibility	[160]

Table 3. Key Biotic-Stress Resistance Loci in Cotton and Mapping Evidence.

Pathogen/Stress	Key QTLs/Candidate Genes	Population Type	Marker System/Platform	Main Finding/Significance	Reference
Verticillium wilt (VW)	QTL clusters on c16 and c19	376 upland cotton accessions (GWAS panel)	CottonSNP63K array (high density)	Identified 15 VW and 13 FOV4 QTLs; clusters on c16 and c19 were consistent across environments	[153,154,155]
Verticillium wilt (VW)	Major QTLs on c16 (D7) and c23 (D9)	Backcross Inbred Line (BIL) population	SSR-based linkage map (~2895 cM, 392 SSRs)	Ten VW QTLs and 28 clusters identified; hotspots correlated with NBS gene density	[143,156]
Verticillium wilt (VW)	SNPs on A01 and A10 (GhAMT2 candidate)	Multi-parent Advanced Generation Inter-cross (MAGIC) population	GWAS integrated with transcriptomics	Major QTLs identified; GhAMT2 implicated in VW response and resistance signalling	[157,158]
Cotton blue disease (CBD)	Cbd gene on chromosome 10 (0.75 cM)	ΔOpal F2 families	SSR marker (DC20027) + SNP haplotype analysis	Early example of MAS: trait tagging via haplotype-based SNP markers for CBD resistance	[159]
Cotton bacterial blight (CBB)	GhSWEET10 (TAL effector target)	Pathogen–cotton interaction study	Transcriptomics + functional validation	GhSWEET10 induced by TAL effector Avrb6; gene silencing reduced susceptibility	[160]

9. Abiotic Stress

Cotton is predominantly cultivated in regions prone to recurrent abiotic stresses such as drought, salinity, and extreme heat, making resilience to these challenges a strategic priority in modern breeding programmes [161]. To enhance tolerance under increasingly variable climates, it is essential to dissect the genetic and molecular basis of stress adaptation. Accordingly, molecular markers, linkage mapping, and quantitative trait locus (QTL) analysis have been widely deployed, enabling the discovery of stress-associated loci and supporting the development of resilient cultivars [81,88,162].

SSR and SNP markers remain cornerstone tools for stress dissection in Gossypium, with multiple studies reporting stable loci linked to drought, salinity, and heat tolerance (Table 4). For instance, in a G. tomentosum × G. hirsutum population, SSR-based mapping identified 13 consistent drought-related QTL clusters across nine chromosomes [163,164]. Similarly, inbred G. hirsutum populations yielded 165 QTLs for drought tolerance across ten chromosomes using combined SSR and SNP markers, with several loci consistently expressed across environments [165]. More recently, GWAS in a 550-line MAGIC population resolved loci for both drought and salinity tolerance [104]. While CottonSNP70K-based GWAS refined salt-tolerance loci at the germination stage [118].

Transcriptome-derived markers have broadened the genetic basis of stress dissection by linking polymorphisms directly to stress-inducible genes. Dwivedi, Suthar [84] mined drought-responsive transcriptomes to design 1946 EST-SSR primers, of which 15 consistently discriminated tolerant from sensitive genotypes, confirming their value for marker-assisted selection (MAS). Integration of GWAS with RNA-seq has also identified salinity-responsive SNPs co-localising with differentially expressed genes involved in ion transport and reactive oxygen species (ROS) detoxification [166]. Such multi-omics overlays repeatedly highlight aquaporins, ROS scavengers, and stress-signalling regulators as robust candidates for functional validation and breeding [167].

High-density SNP arrays such as CottonSNP63K [90] and CottonSNP70K [168] have been instrumental in dissecting stress loci with greater resolution. Marker-assisted pyramiding strategies using SSRs and AFLPs have introgressed multiple tolerance loci, with significant associations reported for osmotic adjustment, relative water content, and yield stability [169,170]. Furthermore, introgression from wild donors (G. tomentosum, G. barbadense) through GBS and resequencing has uncovered novel alleles conferring stable drought and salinity resilience [171].

Despite these advances, most QTLs remain population- or environment-specific, and only a few have been validated across diverse genetic backgrounds or translated into breeding pipelines. To accelerate deployment, integration of pan-genomics, haplotype-based association mapping, functional transcriptomics, and multi-environment phenotyping is critical. When combined with MAS, genomic selection, and predictive modelling, these tools will underpin the design of climate-resilient cotton ideotypes. Looking ahead, synergy with CRISPR-based allele editing and AI/ML-driven predictions promises to transform stress-resilient breeding by enabling precise pyramiding of tolerance alleles and improving prediction accuracy [172]. Table 4 provides a consolidated overview of representative studies (2000–2025) that employed molecular markers and omics platforms to evaluate abiotic stress tolerance in cotton, highlighting the depth of genomic resources now available for translational breeding.

Table 4. Representative studies employing molecular markers to evaluate abiotic stress tolerance in cotton.

Stress Type	Population/Material	Marker/Platform	Approach	Key Findings/Loci	Candidate Genes	Reference
Drought	F2/F3 (G. hirsutum × G. barbadense)	RFLP, SSR	QTL mapping	33 QTLs for water-use efficiency, photosynthesis, productivity	WUE/photosynthesis pathways	[173]
	Upland cotton (multiple populations)	SSR	QTL mapping	QTLs for physiology, yield, architecture	–	[174]
	Segregating breeding lines	SSR (NAU2715, NAU2954, BNL loci)	QTL mapping, MAS	NAU2715/2954 linked to RWC; BNL loci with osmotic adjustment	–	[93]
	F2:3 (G. tomentosum × G. hirsutum)	SSR	QTL mapping	13 QTL clusters on 9 chromosomes; consistent loci	–	[175]
	Upland cotton diversity panel	SNP (array/GBS)	GWAS	Loci for yield under drought stress	Stress-response pathways	[176]
Salinity	Chinese/U.S. panels (196–323 accessions)	CottonSNP70K	GWAS	Stage-specific salt-tolerance SNPs	Ion transporters, osmotic adjustment genes	[118]
	320 G. hirsutum accessions	SNP array + RNA-seq	GWAS × transcriptomics	33 SNPs, 13 QTLs, 98 candidate genes	ROS scavengers, signalling genes, transporters	[105]
	268 upland cotton accessions	CottonSNP70K	GWAS + transcriptomics	27 SNPs (15 salt-tolerance index)	Gh_D01G0943, Gh_A01G0906	[117]
	Upland panel (multi-environment)	GBS, SNP arrays	GWAS	Consistent drought & salinity QTLs	ABA signalling & ion transport	[177]
Heat	Elite Pakistani breeding lines	SSR, KASP	MAS pyramiding	Improved cell membrane stability & RWC	Heat-shock proteins, membrane stability	[178]
	Natural diversity panel	SNP array/GBS	Multi-locus GWAS	QTNs for heat tolerance across traits	ERF, HSF, HSP families	[179]
Cold	110 G. hirsutum genotypes	SSR (101 markers)	Assoc. mapping	16 marker–trait associations; 10 major loci (PVE > 10%)	BNL0569, CIR081, CIR202	[180]
	Transgenic G. hirsutum lines	Overexpression/genome editing	Functional validation	Genes enhancing drought, cold, and heat tolerance	GhDREB1B, GhKCS13, AtSAP5, AmCBF1	[181]

Table 4. Representative studies employing molecular markers to evaluate abiotic stress tolerance in cotton.

Stress Type	Population/Material	Marker/Platform	Approach	Key Findings/Loci	Candidate Genes	Reference
Drought	F2/F3 (G. hirsutum × G. barbadense)	RFLP, SSR	QTL mapping	33 QTLs for water-use efficiency, photosynthesis, productivity	WUE/photosynthesis pathways	[173]
	Upland cotton (multiple populations)	SSR	QTL mapping	QTLs for physiology, yield, architecture	–	[174]
	Segregating breeding lines	SSR (NAU2715, NAU2954, BNL loci)	QTL mapping, MAS	NAU2715/2954 linked to RWC; BNL loci with osmotic adjustment	–	[93]
	F2:3 (G. tomentosum × G. hirsutum)	SSR	QTL mapping	13 QTL clusters on 9 chromosomes; consistent loci	–	[175]
	Upland cotton diversity panel	SNP (array/GBS)	GWAS	Loci for yield under drought stress	Stress-response pathways	[176]
Salinity	Chinese/U.S. panels (196–323 accessions)	CottonSNP70K	GWAS	Stage-specific salt-tolerance SNPs	Ion transporters, osmotic adjustment genes	[118]
	320 G. hirsutum accessions	SNP array + RNA-seq	GWAS × transcriptomics	33 SNPs, 13 QTLs, 98 candidate genes	ROS scavengers, signalling genes, transporters	[105]
	268 upland cotton accessions	CottonSNP70K	GWAS + transcriptomics	27 SNPs (15 salt-tolerance index)	Gh_D01G0943, Gh_A01G0906	[117]
	Upland panel (multi-environment)	GBS, SNP arrays	GWAS	Consistent drought & salinity QTLs	ABA signalling & ion transport	[177]
Heat	Elite Pakistani breeding lines	SSR, KASP	MAS pyramiding	Improved cell membrane stability & RWC	Heat-shock proteins, membrane stability	[178]
	Natural diversity panel	SNP array/GBS	Multi-locus GWAS	QTNs for heat tolerance across traits	ERF, HSF, HSP families	[179]
Cold	110 G. hirsutum genotypes	SSR (101 markers)	Assoc. mapping	16 marker–trait associations; 10 major loci (PVE > 10%)	BNL0569, CIR081, CIR202	[180]
	Transgenic G. hirsutum lines	Overexpression/genome editing	Functional validation	Genes enhancing drought, cold, and heat tolerance	GhDREB1B, GhKCS13, AtSAP5, AmCBF1	[181]

Taken together, drought tolerance has been the most intensively dissected trait in cotton, with stable QTLs and candidate genes validated across multiple populations and environments, making it the closest to practical deployment through MAS and genomic selection. Salinity tolerance has also advanced rapidly, with GWAS and transcriptomic overlays uncovering ion transporters and ROS-regulating genes as robust targets for functional validation. Heat tolerance, while less mature, is now benefiting from pyramiding efforts in elite germplasm, whereas cold tolerance remains largely confined to early-stage discovery with limited translational uptake. These advances highlight that while substantial progress has been made, particularly for drought and salinity resilience, next-generation strategies such as CRISPR-based allele editing and AI-enabled prediction are required to bridge the gap between gene discovery and climate-resilient cultivar development.

10. Agronomic Traits

Yield, fibre quality, and associated agronomic parameters remain central breeding targets in cotton, yet these traits are typically complex, polygenic, and strongly influenced by the environment [82]. Their low heritability under field conditions has historically limited selection efficiency, posing a major challenge for conventional breeding [182]. Advances in molecular biology and genomics have addressed this bottleneck by enabling the development of diverse DNA marker systems such as SSR, SNP, and EST-SSR, which have been widely applied in QTL mapping and trait dissection. These tools facilitate the precise identification of genomic regions linked to key agronomic traits and allow breeders to introgress favourable alleles from diverse germplasm into elite cultivars [183]. Marker-assisted selection (MAS), when supported by robust marker–trait associations, thus provides a pathway to improve complex traits that are otherwise difficult to enhance through phenotypic selection alone [184].

A primary focus of cotton genomics has been the identification of QTLs for yield, quality, and fibre properties through molecular mapping [185,186,187]. Using microsatellites in a recombinant inbred line (RIL) population, Cao, Wang [188] identified QTLs related to fibre quality, including two stable QTLs for lint percentage and fibre length, thereby confirming earlier findings and underscoring their value for MAS. The advent of high-density SNP arrays has transformed trait dissection. For instance, Sun, Wang [189] performed GWAS for fibre quality in 719 upland cotton accessions using the Illumina 63K SNP array, detecting ~10,500 polymorphic markers and 46 significant SNPs across 15 chromosomes. These associations corresponded to 612 candidate genes involved in polysaccharide biosynthesis, signal transduction, and protein trafficking, including 163 and 120 genes associated with fibre length and strength, respectively, many of which overlapped with candidate genes reported in previous QTL studies. Similarly, resequencing of 419 cotton accessions uncovered 3.6 million SNPs, and GWAS for 13 fibre quality traits identified numerous novel associations [190].

Subsequent studies have strengthened the resolution of fibre-related QTLs. Liu, Song [191] detected 42 SNP markers and 31 QTLs linked to five fibre quality traits, of which 25 QTLs were consistent with prior reports while six were novel. These QTL regions encompassed 822 candidate genes, including two pleiotropic SNPs associated with multiple traits such as fibre elasticity, strength, length, uniformity, and resistance. Thyssen, Jenkins [192], leveraging whole-genome sequencing and GWAS in a MAGIC population, identified seven major regions for six essential fibre quality traits, and revealed 14 candidate genes carrying non-synonymous SNPs. Likewise, Sarfraz, Iqbal [193] employed genomic prediction models alongside association studies to highlight SNPs strongly associated with both agronomic and fibre quality parameters.

Earlier SSR- and EST-SSR-based studies remain foundational. Abdurakhmonov, Buriev [194] mapped QTLs for lint percentage, lint yield, seed yield, and fibre traits in a G. hirsutum × G. barbadense cross, identifying numerous stable loci across multiple chromosomes. Abdurakhmonov, Saha [195] extended this work with genome-wide linkage disequilibrium (LD) mapping in 334 upland cotton accessions from diverse ecotypes, associating 12–22 SSRs with fibre length, strength, and fineness across environments. Cao, Wang [188] further evaluated LD patterns, identifying 97 polymorphic SSRs linked to fibre quality.

The integration of modern breeding accelerators has further enhanced the translation of genomic insights into field applications. Speed breeding protocols shorten generation cycles, while high-throughput phenotyping platforms generate precise trait measurements at scale, enabling efficient validation of marker–trait associations. When combined with AI/ML-driven predictive models, these approaches enhance the accuracy of genomic prediction for yield and fibre traits, particularly under variable environments [196,197]. Such frameworks are now being applied in cotton to refine trait dissection and to accelerate the design of resilient ideotypes that combine productivity with superior fibre properties. Taken together, the convergence of conventional marker systems, high-density SNP arrays, GWAS, and breeding accelerators has greatly advanced the precision of trait dissection in cotton (Table 5). These advances not only strengthen the foundation for marker-assisted selection (MAS) and breeding but also establish a direct bridge to genomic selection and genome editing, which are reshaping the future of agronomic trait improvement.

Table 5. Representative published studies employing molecular markers and genomic tools to dissect yield, fibre quality, and agronomic traits in cotton.

Trait Focus	Population/Material	Marker/Platform	Approach	Key loci/Findings	Candidate Genes (if Reported)	Reference
Fibre quality (strength, length)	Segregating BC1F4 populations	RAPD → SCAR (SCAR1920)	MAS	Marker linked to QTL for fibre strength; improved genotype selection	–	[198]
Fibre strength (QTLFS1)	Segregating populations	SSRs	QTL mapping + MAS	DNA markers linked to QTLFS1	–	[199]
Fibre yield & quality	RIL population (G. hirsutum)	SSRs	QTL mapping	Stable QTLs for lint % and fibre length	–	[200,201]
Fibre quality (multiple traits)	MAGIC population (G. hirsutum)	Whole-genome resequencing SNPs	GWAS	Seven major regions; 14 genes with non-synonymous SNPs	Structural & regulatory genes	[191]
Fibre quality (length, strength, fineness)	334 G. hirsutum accessions	SSRs, EST-SSRs	LD mapping	12–22 markers linked with fibre traits across environments	–	[186]
Fibre quality (yield & properties)	G. hirsutum × G. barbadense (F2)	SSRs	QTL mapping	QTLs for lint %, lint yield, fibre traits	–	[89]
Fibre traits (yield, architecture)	Upland cotton panels	SNP arrays (63K Illumina)	GWAS	~10,500 SNPs; 46 significant loci across 15 chromosomes	612 genes (163 for fibre length, 120 for strength)	[36,202]
Fibre quality (13 traits)	419 accessions resequenced	WGS SNPs (~3.6M)	GWAS	Multiple novel associations	–	[203]
Agronomic + fibre quality traits	Diverse germplasm panels	SNPs + GS models	GWAS + Genomic prediction	SNPs strongly associated with yield + fibre traits	–	[129,204,205,206]

Table 5. Representative published studies employing molecular markers and genomic tools to dissect yield, fibre quality, and agronomic traits in cotton.

Trait Focus	Population/Material	Marker/Platform	Approach	Key loci/Findings	Candidate Genes (if Reported)	Reference
Fibre quality (strength, length)	Segregating BC1F4 populations	RAPD → SCAR (SCAR1920)	MAS	Marker linked to QTL for fibre strength; improved genotype selection	–	[198]
Fibre strength (QTLFS1)	Segregating populations	SSRs	QTL mapping + MAS	DNA markers linked to QTLFS1	–	[199]
Fibre yield & quality	RIL population (G. hirsutum)	SSRs	QTL mapping	Stable QTLs for lint % and fibre length	–	[200,201]
Fibre quality (multiple traits)	MAGIC population (G. hirsutum)	Whole-genome resequencing SNPs	GWAS	Seven major regions; 14 genes with non-synonymous SNPs	Structural & regulatory genes	[191]
Fibre quality (length, strength, fineness)	334 G. hirsutum accessions	SSRs, EST-SSRs	LD mapping	12–22 markers linked with fibre traits across environments	–	[186]
Fibre quality (yield & properties)	G. hirsutum × G. barbadense (F2)	SSRs	QTL mapping	QTLs for lint %, lint yield, fibre traits	–	[89]
Fibre traits (yield, architecture)	Upland cotton panels	SNP arrays (63K Illumina)	GWAS	~10,500 SNPs; 46 significant loci across 15 chromosomes	612 genes (163 for fibre length, 120 for strength)	[36,202]
Fibre quality (13 traits)	419 accessions resequenced	WGS SNPs (~3.6M)	GWAS	Multiple novel associations	–	[203]
Agronomic + fibre quality traits	Diverse germplasm panels	SNPs + GS models	GWAS + Genomic prediction	SNPs strongly associated with yield + fibre traits	–	[129,204,205,206]

11. Leveraging Markers for Cotton Breeding Using Genomic Selection Approach

Cotton improvement has long been constrained by the polygenic nature, strong genotype × environment (G×E) interactions, and low heritability of key traits such as fibre quality, lint yield, and stress resilience. Traditional marker-assisted selection (MAS) is invaluable for major-effect QTLs but falls short in capturing the small yet cumulative contributions of many loci. Genomic selection (GS) overcomes these limitations by leveraging genome-wide marker data to predict genomic estimated breeding values (GEBVs), thus enabling indirect selection without requiring extensive phenotyping in each breeding cycle [207]. Unlike MAS, GS employs dense genome-wide SNP datasets and applies statistical and machine learning models such as genomic best linear unbiased prediction (GBLUP), ridge regression BLUP (rrBLUP), Bayesian frameworks, and ensemble learning algorithms to capture both large- and small-effect loci. The efficacy of GS depends on several parameters, including marker density, training population size, trait heritability, and the model’s ability to capture G×E interactions. High-density genotyping platforms such as CottonSNP63K, CottonSNP70K, and cost-effective genotyping-by-sequencing (GBS) technologies now facilitate the practical implementation of GS in modern breeding programmes [208].

Several studies demonstrate the feasibility of GS in cotton breeding (see Table 6). For example, Shan, Xiong [208] evaluated 550 recombinant inbred lines (RILs) from a MAGIC population and reported that Bayesian models, particularly BayesB, yielded the most accurate predictions for multiple fibre quality traits. Gapare, Liu [209] assessed GS across 215 diverse upland cotton lines grown in multi-environment Australian trials, showing that multi-environment models significantly improved prediction accuracy (0.71 for fibre length and 0.59 for strength). In a large-scale dataset, Li, Liu [129] achieved accuracies of 0.76 for fibre length, 0.65 for strength, and 0.64 for lint yield when combining genomic and pedigree data. Likewise, Billings, Jones [210] demonstrated the deployability of GS across 14 environments, while Souaibou, Yan [211] integrated Bayesian and machine learning models to achieve stable prediction accuracies (0.52–0.72). Many GS-identified loci overlap with QTLs or GWAS hits, reinforcing the continuum between gene discovery and predictive breeding.

To fill a critical knowledge gap, recent studies have extended GS to abiotic stress traits. For instance, a GWAS conducted in a MAGIC population of 550 lines revealed 20 QTLs for drought tolerance and 23 for salt tolerance, identifying 53 candidate genes, several of which overlapped between the two stress responses [161,212]. Integrating such QTLs and candidate genes into GS frameworks holds considerable promise for enhancing selection efficiency under variable climatic conditions.

Challenges remain: prediction accuracy can decline when validation sets diverge genetically from training populations, and genotyping costs still limit widespread adoption, particularly in developing countries. Implementation also requires coordination across breeding programmes to ensure robust model training and data sharing. Recent efforts to enhance GS through the incorporation of GWAS-derived loci, transcriptomic signatures, and phenomic predictors have shown marked improvements in predictive power [205,213]. Beyond research settings, real-world adoption of genomic prediction in cotton is gaining traction through pilot programmes in multiple countries. For example, the CGIAR Excellence in Breeding Initiative and CSIRO in Australia have established multi-location genomic-prediction trials that integrate digital phenotyping and AI-based decision-support systems to refine selection under field conditions [214,215]. Similarly, breeding programmes in India and Brazil have incorporated GS into conventional selection pipelines to accelerate cultivar release, demonstrating that prediction accuracies can be maintained when applied across diverse environments [216]. These initiatives highlight the transition of GS from theoretical modelling to operational breeding, with success depending on collaborative data infrastructure, capacity building, and validation across farmer-managed trials [217].

Looking ahead, GS is poised to play a central role in next-generation cotton improvement. Its integration with genome editing platforms such as CRISPR/Cas, speed breeding, and AI/ML-driven predictive modelling offers the potential to design cotton ideotypes with superior fibre quality, yield stability, and stress tolerance within shortened breeding cycles. By bridging marker discovery and predictive breeding, GS provides a scalable and data-driven framework for accelerating genetic gain in cotton under increasingly variable climates (Figure 3).

Collectively, published evidence indicates that genomic selection (GS) in cotton consistently achieves moderate-to-high predictive accuracy for key fibre traits. For instance, the CSIRO study that combined genomic and pedigree information reported accuracies of 0.76 for fibre length, 0.65 for fibre strength, and 0.64 for lint yield using Bayesian regression models [133]. Similarly, an evaluation of a Upland cotton MAGIC population across three years demonstrated that GS methods including GBLUP, Bayesian frameworks, and reproducing kernel Hilbert space (RKHS) models delivered high predictive performance for multiple fibre quality traits [129]. Prediction accuracy generally improves when multi-environment trials are employed and pedigree information is incorporated, emphasising the importance of modelling genotype × environment (G × E) interactions. In the CSIRO analysis, the inclusion of both genomic and pedigree data contributed substantially to the highest accuracy values observed. Bayesian methods such as BayesB often outperform linear models like GBLUP, particularly for traits of lower heritability or where non-additive effects are significant [133]. Overall, these findings confirm that GS has matured into a practical and high-value tool for cotton breeding, particularly when combined with robust experimental designs (multi-environment and pedigree-informed) and capable statistical models that capture complex trait architectures.

Table 6. Representative Genomic Selection Studies in Cotton.

Trait Focus	Population/Material	Marker/Platform	Approach/Models Compared	Key Findings/Prediction Accuracy	Candidate Genes (if Reported)	Reference
Fibre length & strength	215 upland cotton lines (Australia)	13,330 SNPs (array)	GS with BayesB, GBLUP, rrBLUP, M × E models	Multi-environment GS improved accuracy: 0.71 (length), 0.59 (strength)	–	[202]
Fibre quality traits (6)	550 RILs (MAGIC population)	6292 SNPs (GBS)	GBLUP, rrBLUP, BayesB, Bayes LASSO, RKHS	Prediction accuracy 0.55–0.69; BayesB most effective	–	[192]
Fibre traits & yield	80 elite upland cultivars (14 environments)	CottonSNP63K array	GBLUP, BayesB, GWAS-assisted GS	Accuracy 0.45–0.68 depending on trait; GWAS signals increased predictive power	–	[133]
Fibre length, uniformity, micronaire	1385 CSIRO breeding lines	12,296 SNPs (GBS)	Bayesian ridge regression, BayesB, ML models	Prediction accuracy up to 0.76 (length), 0.65 (strength), 0.64 (yield); combining pedigree + genomic data improved predictions	–	[218]
Fibre yield & quality	300 upland cotton accessions	CottonSNP70K array	GBLUP, BayesCπ, Random Forest (ML + Bayesian)	Prediction accuracy 0.52–0.72; combined models gave stable predictions	–	[219]

Table 6. Representative Genomic Selection Studies in Cotton.

Trait Focus	Population/Material	Marker/Platform	Approach/Models Compared	Key Findings/Prediction Accuracy	Candidate Genes (if Reported)	Reference
Fibre length & strength	215 upland cotton lines (Australia)	13,330 SNPs (array)	GS with BayesB, GBLUP, rrBLUP, M × E models	Multi-environment GS improved accuracy: 0.71 (length), 0.59 (strength)	–	[202]
Fibre quality traits (6)	550 RILs (MAGIC population)	6292 SNPs (GBS)	GBLUP, rrBLUP, BayesB, Bayes LASSO, RKHS	Prediction accuracy 0.55–0.69; BayesB most effective	–	[192]
Fibre traits & yield	80 elite upland cultivars (14 environments)	CottonSNP63K array	GBLUP, BayesB, GWAS-assisted GS	Accuracy 0.45–0.68 depending on trait; GWAS signals increased predictive power	–	[133]
Fibre length, uniformity, micronaire	1385 CSIRO breeding lines	12,296 SNPs (GBS)	Bayesian ridge regression, BayesB, ML models	Prediction accuracy up to 0.76 (length), 0.65 (strength), 0.64 (yield); combining pedigree + genomic data improved predictions	–	[218]
Fibre yield & quality	300 upland cotton accessions	CottonSNP70K array	GBLUP, BayesCπ, Random Forest (ML + Bayesian)	Prediction accuracy 0.52–0.72; combined models gave stable predictions	–	[219]

12. Harnessing Genomics and CRISPR/Cas Systems for Next-Generation Precision Breeding

12.1. Fibre Quality Enhancement

The advent of CRISPR/Cas genome editing represents a paradigm shift in cotton improvement, enabling accurate and predictable modification of complex genomes previously intractable to conventional breeding or transgenic approaches [18]. Cotton fibre quality, yield stability, and stress resilience are governed by polygenic networks and tightly regulated developmental pathways, posing major challenges for traditional breeding due to linkage drag, extended cycles, and pleiotropic effects [41]. Genome editing addresses these limitations by generating heritable, targeted mutations without foreign DNA integration, thereby accelerating gene validation and the design of ideotypes optimised for fibre performance and environmental adaptability [220].

Fibre length is a primary determinant of cotton quality and a long-standing breeding target. Transcriptomic and functional genomic analyses have identified numerous genes preferentially expressed during elongation [221,222]. Yet, the biological roles of many remain unresolved, limiting their deployment in breeding programmes. CRISPR/Cas systems now provide a powerful means to dissect gene function and introduce modifications that directly enhance fibre traits [220,223].

One of the earliest demonstrations targeted GhAlaRP, an alanine-rich protein expressed during fibre elongation [224]. Mutagenesis achieved frequencies of 71–100% in GhAlaRP-A and 93–100% in GhAlaRP-D alleles, mainly deletions, with no off-target effects. Edited plants confirmed GhAlaRP’s role in elongation via cell-wall assembly and secondary metabolism [225]. Similarly, map-based cloning coupled with CRISPR editing identified GhIm, a pentatricopeptide-repeat (PPR) protein essential for mitochondrial nad7 mRNA splicing; its disruption impaired mitochondrial activity, producing immature, non-fluffy fibres with reduced cellulose deposition h [226]. This finding highlighted a mechanistic link between organelle homeostasis and fibre morphogenesis.

Targeted modification has also advanced understanding of secondary cell-wall formation and fibre strength. Editing GhCESA genes, central to cellulose biosynthesis, altered secondary wall thickening and enhanced cellulose content without yield penalties [227]. Likewise, editing GhMYB25-like, a transcription factor controlling trichome and fibre initiation, revealed dual roles in differentiation and wall development, offering opportunities to balance fibre number and strength [228].

Together, these studies demonstrate how genome editing is decoding the regulatory networks of fibre elongation and wall biosynthesis. Precise manipulation of GhAlaRP, GhIm, GhCESA, and GhMYB25-like genes provides a foundation for direct engineering of fibre attributes, moving beyond correlation-based breeding. Future work should prioritise multiplex editing of key pathways, integration with multi-omics datasets, and approaches that enhance fibre quality while preserving yield stability across environments.

12.2. Cotton Oil Enhancement

Cotton (Gossypium spp.) though primarily a fibre crop is also the third most important source of edible oil globally, after soybean and canola. Cottonseed oil and by-products serve as nutritional resources and industrial feedstocks, positioning cotton as a dual-purpose crop [229,230,231]. Conventional breeding has achieved limited progress in oil-quality improvement because of polygenic regulation and linkage drag. Genome editing offers a transformative tool for targeted metabolic re-engineering of lipid biosynthesis, bypassing the constraints of transgenic methods and enabling precise, inheritable edits [232,233].

Oil composition depends largely on desaturase enzymes. The stearoyl-ACP desaturase (SAD) family converts saturated stearic acid (18:0) into monounsaturated oleic acid (18:1), improving nutritional quality and oxidative stability. Comparative genomics has identified 9, 9, 18, and 19 SAD copies in G. arboreum (A₂), G. raimondii (D₅), G. hirsutum (AD₁), and G. barbadense (AD₂), respectively [230]. Transcriptomic profiling of developing seeds revealed several isoforms highly expressed during oil accumulation, marking them as strong candidates for targeted editing.

Functional validation has confirmed the roles of SAD and FAD2 genes in regulating fatty-acid balance. Gene knock-outs and RNAi consistently increased oleic acid while reducing polyunsaturated fractions, enhancing oil stability [231]. A landmark study edited homoeologous GhFAD2 alleles in allotetraploid cotton using multiplex CRISPR/Cas9, achieving high efficiency across both subgenomes [234,235,236]. Edited lines exhibited > 70% oleic acid content and a sharply reduced linoleic fraction, surpassing improvements achievable through conventional breeding (Figure 4). Importantly, these plants are non-transgenic, as no foreign DNA was introduced—simplifying regulatory approval and improving consumer acceptances [234].

Regulatory frameworks for genome-edited crops, however, remain diverse. In the United States, the USDA SECURE rule follows a product-based approach under which genome-edited plants without foreign DNA are generally exempt from GMO regulation [237]. Brazil’s CTNBio Normative Resolution No. 16 (2018) applies a case-by-case evaluation that often treats such edits as conventional varieties [238,239]. In the European Union, although the 2018 European Court of Justice ruling placed all edited plants under GMO legislation, the 2023 European Commission proposal (COM (2023) 411) introduces a differentiated framework for new genomic techniques (NGTs), where Category 1 plants with small, targeted edits may be exempt from full GMO risk assessment [240]. Divergent national policies significantly influence global commercialisation, trade, and deployment of edited crops [241]. Achieving international harmonisation, transparency, and science-based regulation will be essential for equitable deployment of genome-edited cotton varieties worldwide.

Beyond GhFAD2, editing has expanded to other genes affecting oil quality, nutritional value, and industrial use. For example, multiplex editing of SAD, FAD3, and FAE1 homologues further increases oleic acid and reduces undesirable polyunsaturates [242]. Parallel efforts targeting gossypol biosynthesis—such as disruption of δ-cadinene synthase (CYP706B1) or dihydroflavonol-4-reductase—aim to create low-gossypol, edible seeds while retaining natural pest resistance in vegetative tissues [133,243]. These designs exemplify how coordinated genome editing can generate cotton lines with improved oil profiles and reduced anti-nutritional compounds.

Looking ahead, oil-trait editing is shifting from single-gene modification to systems-level optimisation. Integrating multi-omics datasets, metabolic modelling, and AI-based prediction will refine target selection and accelerate design of ideal lipid pathways. Emerging base and prime editors now enable single-nucleotide precision, permitting subtle tuning of enzyme function [244]. Moreover, editing efforts are expanding to engineer oils with tailored fatty acid profiles for industrial feedstocks, such as high-stearate or high-erucic-acid cottonseed oils for lubricants, surfactants, and bioplastics [235]. Editing is also being extended to engineer industrially tailored fatty-acid profiles, such as high-stearate or high-erucic-acid oils for lubricants, surfactants, and bioplastics [233]. Collectively, these advances are transforming cotton from a fibre-centred crop into a sustainable, bio-based resource for the circular economy.

12.3. Fibre Enhancement

Fibre length critically influences yarn strength, spinnability, and commercial value [245]. This trait arises from a complex network of cell-expansion, hormonal, and redox pathways [246]. Traditional breeding has achieved only modest progress because of polygenic control, linkage drag, and the buffering effects of polyploidy. Genome editing now allows precise functional analysis and modification of key regulator [228].

A recent meta-analysis of fibre-development genes identified CRISPR/Cas9 as the most efficient method for dissecting regulatory networks [247]. Two complementary strategies have emerged: knock-out of negative regulators and activation of positive ones. Disrupting negative regulators such as GhHDZ76 [222], GhFAD3-4 and GhFAD7A-1 [248], GhATL68b [249], GhAlaRP [250], and GhARF18 [224,225] shortens fibres, confirming their repressive roles, as summarised in Table 7. Conversely, editing GhKNOX6, GhBZR3, and GhMDHAR1A/GhDHAR2A enhances elongation by maintaining redox balance and cell expansion [227,248,251]. Genes involved in wall biosynthesis, such as GhCesA4 and GhEXPA3-1, also influence fibre strength and fineness [225,252].

Given cotton’s allotetraploid nature, simultaneous editing of At and Dt homoeologues via multiplex sgRNAs is essential for clear phenotypic expression [253]. Combining suppression of negative regulators with activation of positive ones enables balanced improvements in fibre length and structural integrity. Fine-tuning with base or prime editors allows precision modification without full gene disruption. As breeding programmes increasingly incorporate genomic selection and speed breeding, such functional validation expedites translation from gene discovery to cultivar deployment. While genome editing accelerates functional validation and trait enhancement, over-reliance on narrow elite germplasm could erode allelic diversity and long-term adaptability [254]. To mitigate this, future frameworks should integrate CRISPR strategies with pangenomic resources, multi-parent populations, and genomic-selection models that capture polygenic variation [252]. Balancing accelerated genetic gain with conservation of diversity will be vital for durable cotton improvement [38,255].

Table 7. Representative CRISPR-based studies modulating fibre length.

No.	Author (Year)	Target Gene(s)	Effect on Fibre Length
1	Wu et al. (2024) [222]	GhHDZ76	Shortening
2	Wang et al. (2024) [256]	GhFAD3-4	Shortening
3	Jia et al. (2024) [248]	GhFAD7A-1	Shortening
4	Jia et al. (2024) [248]	GhKNOX6	Increasing
5	Li et al. (2024) [249]	GhATL68b	Shortening
6	Jiao et al. (2025) [247]	GhMDHAR1A/GhDHAR2A	Increasing
7	Zhu et al. (2023) [250]	GhEXPA3-1	Shortening
8	Tian et al. (2024) [227]	GhGRF4/GhARF2-GhGASA24	Shortening
9	Zhu et al. (2021) [225]	GhAlaRP	Shortening

Table 7. Representative CRISPR-based studies modulating fibre length.

No.	Author (Year)	Target Gene(s)	Effect on Fibre Length
1	Wu et al. (2024) [222]	GhHDZ76	Shortening
2	Wang et al. (2024) [256]	GhFAD3-4	Shortening
3	Jia et al. (2024) [248]	GhFAD7A-1	Shortening
4	Jia et al. (2024) [248]	GhKNOX6	Increasing
5	Li et al. (2024) [249]	GhATL68b	Shortening
6	Jiao et al. (2025) [247]	GhMDHAR1A/GhDHAR2A	Increasing
7	Zhu et al. (2023) [250]	GhEXPA3-1	Shortening
8	Tian et al. (2024) [227]	GhGRF4/GhARF2-GhGASA24	Shortening
9	Zhu et al. (2021) [225]	GhAlaRP	Shortening

13. Development of Stress-Resistant Cotton

The intensifying challenges of climate variability, pathogen evolution, and soil degradation have elevated stress resilience to a central objective in cotton improvement. Abiotic stresses such as drought, salinity, and heat, together with biotic threats including viral complexes, soil-borne fungi, and insect pests, collectively inflict substantial losses in yield and fibre quality [156,257,258]. Conventional breeding has achieved incremental gains through introgression from wild germplasm, yet progress is constrained by the polygenic architecture of stress adaptation, linkage drag, and limited allelic diversity in cultivated cotton. CRISPR/Cas-mediated genome editing offers a paradigm shift by enabling precise, heritable, and multiplexed modification of stress-related loci, thereby accelerating the development of climate-resilient ideotypes while avoiding many of the regulatory complexities associated with transgenic technologies [259].

Among biotic constraints, Cotton Leaf Curl Disease (CLCuD) remains one of the most destructive. Caused by begomoviruses such as Cotton Leaf Curl Multan Virus and their associated betasatellites, CLCuD continues to devastate production across South Asia and Africa [260]. The rapid evolution of these viral complexes has undermined conventional resistance sources. CRISPR/Cas-based approaches provide a dual defence strategy: (i) editing host susceptibility genes to reinforce innate immunity, and (ii) directly cleaving viral genomes to disrupt replication. Multiplex CRISPR systems targeting conserved DNA-A sequences of CLCuV significantly reduced viral titres and symptom severity, validating host-delivered resistance [261]. Moreover, geminivirus-derived replicons such as pBeYDV-Cas9-KO have been adapted as efficient in planta delivery vehicles for CRISPR/Cas reagents, providing a flexible antiviral platform with broader applicability across Geminiviridae [262].

Soil-borne fungal pathogens present equally formidable challenges. Verticillium wilt (Verticillium dahliae), for instance, causes annual economic losses exceeding USD 250 million in China alone [263]. Unlike viral resistance, which can be achieved via sequence-specific cleavage, knockout of Gh14-3-3d, a susceptibility factor facilitating Verticillium infection, conferred durable resistance in T₂ plants without adverse agronomic penalties [264]). Similarly, targeted disruption of GhCPK33, a peroxisome-localised calcium-dependent protein kinase that destabilises jasmonic acid (JA) biosynthesis by phosphorylating GhOPR3, enhanced JA-mediated defence signalling [265]. These studies highlight how removing negative regulators via genome editing can strengthen immunity against fungal pathogens while maintaining yield stability (Figure 5).

Recent innovations extend CRISPR utility beyond conventional knockouts. Base editing and prime editing now permit single-nucleotide changes or targeted insertions without double-strand breaks, enabling fine-tuning of stress-responsive alleles with reduced off-target risk [266]. In parallel, dCas9-based transcriptional regulators fused to epigenetic modifiers provide reversible, non-heritable modulation of stress-inducible genes, offering conditional defence activation without altering the underlying DNA sequence. Multiplex CRISPR platforms, designed with polycistronic sgRNA cassettes, further allow simultaneous targeting of multiple susceptibility loci, thereby reducing the risk of pathogen adaptation [261]. These advances illustrate how CRISPR-based editing is not only uncovering the genetic architecture of stress responses but also providing tractable routes to engineer durable resistance in cotton. By linking precise genome modifications with functional genomics and breeding pipelines, stress-resilient traits can be developed in a more predictable and environmentally responsive manner. As the following section explores, the power of genome editing is maximised when combined with multi-omics datasets, predictive breeding models, and advanced phenotyping platforms, which together accelerate the translation of molecular insights into resilient, high-yield cultivars.

14. Cross-Talk Between Genomics and Speed Breeding for Fast-Tracked Cotton Improvement

The accelerating pace of climate disruption and fibre market volatility has rendered conventional cotton breeding timelines increasingly untenable [267]. With a single breeding cycle extending over four to six months, the rate of genetic gain remains misaligned with the urgency of deploying cultivars adapted to shifting thermal regimes, erratic rainfall, and emerging pathogen pressures [268]. Speed breeding an approach that exploits controlled-environment optimisation of light, temperature, and photoperiod to compress generation cycles offers a transformative solution to this temporal bottleneck. More than a technical acceleration, it represents a structural re-engineering of breeding architecture, enabling multi-generational advancement within a single year and redefining the operational cadence of genomic selection, genome editing, and predictive ideotype design [197].

Initially demonstrated in temperate cereals, speed breeding has now been successfully adapted to cotton (G. hirsutum), reducing cycle duration from approximately 130 days to 71–85 days under optimised photoperiod and thermal regimes [196]. Recent work by Wang, Sun [269] demonstrated the feasibility of achieving up to five generations per year in cotton under controlled environments by combining photoperiod manipulation, high-intensity LED lighting, genome-based knowledge, and embryo rescue techniques. These innovations have enabled rapid fixation of favourable allele combinations for complex traits, exemplified by the accelerated development of elite lines such as JND24-i3, consolidating superior fibre quality and yield attributes in less than 18 months, a process that traditionally required 5–6 years.

However, speed breeding is not without its constraints. High energy inputs, genotype-specific photoperiod sensitivities, and the physiological cost of rapid cycling remain significant challenges [23]. Moreover, accelerated development under controlled conditions can induce atypical source–sink dynamics, altered hormonal profiles, and potential trade-offs in fibre elongation or boll retention, which may compromise field performance if unaddressed. Standardised, genotype-specific optimisation and integration with stress physiology modelling are therefore imperative to avoid phenotypic penalties associated with artificial acceleration. Equally important are the socioeconomic and regulatory dimensions: the infrastructure and energy requirements of speed-breeding facilities may limit access in low-resource public breeding programmes; intellectual-property and licensing arrangements around genome-edited lines can affect equitable uptake; and regulatory uncertainty over gene-edited versus transgenic cultivars remains a critical bottleneck in many jurisdictions [270,271]. Addressing these issues through capacity building, public–private partnerships, and harmonised biosafety frameworks will be essential to ensure that technological advances translate into farmer-ready cultivars.

Beyond these constraints, scalability remains a major concern. Breeding programmes in developing regions often face prohibitive costs associated with controlled-environment infrastructure and continuous lighting, both of which substantially increase operational energy demands. Although empirical data for cotton-specific speed-breeding facilities remain scarce, cost analyses from analogous cereal and legume breeding systems indicate that genotyping expenses typically range between USD 5 and USD 35 per sample, depending on platform type, marker density, throughput, and service provider [272]. Studies have further shown that achieving per-sample genotyping costs below USD 15 is generally critical for maintaining cost-effective genetic gain. Kompetitive Allele-Specific PCR (KASP) and other low-density marker platforms offer practical, lower-cost solutions when only a limited number of SNPs are required [273]. To enhance scalability and ensure equitable access, strategies such as infrastructure sharing through regional breeding hubs, integration of renewable energy systems, and adoption of open-source bioinformatics pipelines will be essential to bring speed breeding within reach of resource-limited breeding programmes.

The strategic potential of speed breeding is magnified when embedded within multi-layered breeding pipelines. Integration with genomics-assisted breeding (GAB) enhances the rate of allele discovery and fixation for traits governed by complex networks such as lint yield, fibre Micronaire, and abiotic stress resilience. Coupling speed breeding with CRISPR/Cas genome editing creates a rapid feedback loop for functional validation of edited alleles, collapsing validation timelines for genes controlling fibre development, disease resistance, and oil composition from years to months [253]. Similarly, when combined with genomic selection (GS) models and high-throughput phenotyping (HTP), accelerated cycling amplifies selection intensity, improves prediction accuracy, and compresses breeding cycles at an unprecedented scale [274]. Nevertheless, the success of this integration ultimately depends on farmer acceptance, seed-system readiness, and policy support. Without early engagement of growers, extension agencies, and regulatory bodies, accelerated pipelines risk producing cultivars that remain confined to research stations rather than achieving broad commercial impact.

Advances in speed breeding demonstrate how controlled-environment innovations can reshape the temporal limits of cotton improvement, while also underscoring the need for concurrent investment in socio-technical systems that support its adoption. The next step lies in firmly integrating accelerated breeding cycles with genomic selection, genome editing, and high-throughput phenotyping to ensure that rapid generation turnover translates into genetically stable, field-validated performance. Rather than operating as a stand-alone technique, speed breeding must function as a core component of genomics-led, climate-smart breeding pipelines that enable continuous genetic gain under variable environments.

15. Conclusions and Future Perspectives

Cotton breeding is entering a decisive new era shaped by genomics, genome editing, and digital innovation. High-quality reference genomes, pan-genomic resources, and multi-omics integration now enable precise dissection of traits once considered intractable—from fibre architecture to resilience under climate stress. CRISPR/Cas-based editing has advanced the field beyond correlation, offering direct tools to reprogramme fibre development, enhance oil composition, and engineer durable stress resistance without the drag of conventional introgression. Equally transformative accelerators speed breeding, genomic selection, and high-throughput phenotyping are redefining both the pace and precision of selection. Coupled with AI-driven analytics and digital twin simulations, cotton improvement is shifting from empirical breeding to predictive, ideotype-based design. Yet, the long-term success of these technologies depends not only on scientific capacity but also on the socioeconomic and regulatory frameworks in which they operate. Harmonised biosafety policies, transparent differentiation between gene-edited and transgenic cultivars, and equitable intellectual-property management will be essential to facilitate approval, seed multiplication, and farmer access. Infrastructure and energy demands associated with phenotyping and speed-breeding platforms remain prohibitive for many public-sector and low-resource programmes, highlighting the need for capacity building, shared infrastructure, and inclusive public–private partnerships. Early engagement with growers, extension agents, and value-chain stakeholders will be equally vital to ensure market acceptance and equitable benefit distribution.

Beyond technical readiness, the economic feasibility of deploying genome-edited and genomics-assisted cotton must be assessed through the lens of smallholder adoption. Developing nations, where cotton remains a critical livelihood crop, often lack reliable seed delivery systems, access to high-throughput genotyping, or regulatory clarity for genome-edited products. Integrating genomics pipelines with existing seed systems, such as national foundation-seed schemes and regional cooperatives, will be essential for scalability. Initiatives that link molecular-breeding innovations with seed enterprises, farmer field schools, and credit mechanisms can bridge the gap between discovery and deployment. Models from India, Pakistan, and sub-Saharan Africa demonstrate that locally adapted seed networks, supported by public–private–producer alliances, significantly lower entry costs and accelerate adoption of improved cultivars [275,276,277,278]. Strategic investment in breeder training, participatory varietal selection, and low-cost genotyping platforms will ensure that genomics-enabled advances deliver tangible economic benefits to smallholder producers while strengthening national cotton value chains.

Rigorous field validation of genomic-prediction and digital-breeding tools remains crucial. Multi-environment trials and cross-population testing are needed to confirm the stability of prediction models and marker–trait associations under diverse agroecological conditions. Recent initiatives for example, multi-location genomic selection pilots in India and the USA, and the CGIAR Excellence in Breeding Genomic Prediction Toolbox demonstrate how predictive models can be integrated into operational breeding pipelines to accelerate gain while maintaining reliability. Scaling such efforts will require sustained investment in interoperable databases, data-driven breeder training, and collaboration with seed enterprises and extension agencies to ensure that digital tools move from proof-of-concept to practical application.

Globally, major advances in cotton genomics and breeding have emerged through international collaboration. Large-scale pan-genome analyses encompassing nearly two thousand accessions have identified non-reference genes and presence/absence variations linked to fibre quality, yield, and stress tolerance—unlocking new genetic resources for breeding [67]. Functional genomics and editing studies from leading programmes in China, the United States, and India have translated these discoveries into applied outcomes; for example, targeted GhFAD2-1A/D knockout produced high-oleic, non-transgenic cotton without compromising fibre properties [234]. Open-access data platforms such as CottonGen [279], have further strengthened global information exchange and the dissemination of genomic resources. National programmes in India (ICAR–CICR) and Australia (CSIRO and Cotton Breeding Australia) are now integrating genomic selection and speed breeding, while regional resequencing and introgression efforts in Pakistan, Uzbekistan, and Brazil are broadening the Gossypium gene pool for resilience and fibre improvement.

In Iran, national research institutes—including the Agricultural Biotechnology Research Institute of Iran (ABRII) and the Cotton Research Institute (CRII)—are actively integrating genomics and molecular-breeding approaches within pre-breeding pipelines. Transcriptomic and gene-family analyses from these centres have illuminated small RNA and DNA methylation dynamics underlying heterosis in allotetraploid cotton [280,281]. Additional studies have characterised key gene families involved in stress response and fibre development [281,282,283,284,285]. SSR- and SNP-based diversity assessments of Iranian cultivars are now informing marker-assisted and genomic-selection frameworks targeting drought tolerance, fibre fineness, and disease resistance under semi-arid conditions [145]. These initiatives are laying the foundation for Iran’s participation in global cotton-genomics networks and for deploying genomics-assisted breeding under local agroecological realities [119]. With these converging tools, inclusive frameworks, and international collaborations, genomics-assisted cotton breeding is poised to deliver climate-resilient, high-yield cultivars that are not only technologically advanced but also economically viable and socially equitable. As cotton improvement continues to bridge genomics, systems biology, and digital agriculture, it offers a scalable model for integrating scientific innovation with economic sustainability, ensuring global fibre security in an era of accelerating climate change.