Abiotic stresses particularly drought, salinity, heat, cold, nutrient imbalance, and waterlogging remain among the most persistent constraints on plant growth, productivity, and yield stability across agricultural systems [1,2,3]. Their importance has grown sharply under accelerating climate change, which is intensifying the frequency, duration, and co-occurrence of environmental stresses, and is expected to depress global yields of major staple crops by 10–25% before mid-century unless adaptive strategies are successfully deployed [4]. Crop losses attributable to drought and heat alone have more than tripled over the past five decades, with combinatorial stress events particularly drought plus heat, or waterlogging plus heat causing disproportionately large reductions that cannot be predicted from single-stress responses [5]. Before proceeding, it is important to clarify the terminological framework used in this review. Stress response refers to any immediate molecular, biochemical, or physiological reaction triggered by an abiotic stress stimulus, including changes in gene expression, protein abundance, or metabolite levels that occur within minutes to hours of stress onset. Acclimation denotes a reversible, phenotypically sustained adjustment that develops over a longer timescale and improves organismal performance under a given stress condition without permanent genetic change. Stress priming describes the phenomenon whereby a prior sub-lethal stress exposure enhances the speed, magnitude, or efficiency of the response to a subsequent stress event. Stress memory refers to the molecular and epigenetic mechanisms, including persistent histone modifications, DNA methylation changes, and small RNA accumulation, that underlie the retention of priming effects beyond the initial stress exposure, in some cases across cell generations or developmental stages. Long-term stress tolerance describes constitutively or adaptively fixed capacities that allow a species or genotype to maintain growth and reproduction under chronic stress conditions. These distinctions matter because not every stress-responsive transcript, protein, or metabolite is functionally related to acclimation many represent transient damage-response or signaling events whose downstream contribution to sustained performance remains unresolved. Throughout this review, these terms are used with these specific meanings, and claims about acclimation are restricted to evidence that goes beyond immediate stress-responsive changes. In crop science and plant molecular biology alike, the central issue is therefore no longer whether plants respond to stress, but how they coordinate molecular and biochemical adjustment, physiological protection under compound stress, and longer-term acclimation [6].

Over the past two decades, advances in high-throughput sequencing, mass spectrometry, and computational biology have transformed the study of plant stress responses [7,8,9]. Transcriptomics first expanded the scale at which stress-responsive genes could be identified, whereas proteomics, metabolomics, epigenomics, and integrative systems analysis later showed that stress acclimation cannot be understood from transcript-level data alone [10,11,12]. Transcript abundance, protein accumulation, metabolite turnover, and regulatory persistence often diverge substantially in both timing and magnitude, particularly under prolonged or combined abiotic stress [5,13,14]. Recent multifactorial-stress analyses in tomato, for example, demonstrate that transcriptomic, proteomic, and metabolomic layers respond with distinct kinetics and only partially overlapping sets of regulated entities, a pattern also observed in potato under heat, drought, and waterlogging. For this reason, multi-omics integration has become a core direction in contemporary plant stress biology, linking molecular profiling with systems-level interpretation [15].

The value of multi-omics approaches lies precisely in this capacity to move beyond description. Single-omics studies can identify lists of responsive genes, proteins, or metabolites, but they often provide limited insight into causal relationships, pathway coordination, or the regulatory architecture underlying acclimation [4,10,11]. By contrast, integrative analysis relates transcriptional reprogramming to protein-level buffering and metabolic adjustment [14,16], while also connecting these layers to hormonal control, chromatin dynamics, and stress memory [14,17]. Contemporary reviews have therefore emphasized that multi-omics is not merely an accumulation of datasets, but an analytical framework for linking molecular layers that are otherwise interpreted in isolation [15].

Despite this progress, literature remains highly uneven in its taxonomic and agricultural coverage [11,18,19]. Broad syntheses of plant abiotic stress biology are still dominated by model plants and major crops, particularly Arabidopsis thaliana, rice (Oryza sativa), maize (Zea mays), wheat (Triticum aestivum), and soybean (Glycine max) [1,6,7]. This is understandable since these systems benefit from mature genomic resources, standardized phenotyping pipelines, comprehensive annotation databases, and extensively characterized regulatory networks [20,21]. Yet it has narrowed the comparative base from which stress-acclimation mechanisms are generalized. In many recent reviews, tropical and underutilized plants appear only marginally, usually as isolated examples rather than as central systems for integrative synthesis.

That imbalance is increasingly difficult to justify. Tropical and underutilized plant species are not merely peripheral to plant science; many are directly relevant to food and nutritional security, diversified agriculture, resilient production systems, and region-specific adaptation to environmental instability [18,19,22]. They also include species with distinctive metabolic profiles, unusual ecological tolerances, and underexplored regulatory biology [23]. Recent systematic review work has emphasized that underutilized crops deserve greater scientific attention because of their potential contribution to resilient and healthy agri-food systems, yet molecular studies on these species remain fragmented compared with those on major crops [24]. In practical terms, this means that potentially important insights into antioxidant defense, osmotic adjustment, metabolic plasticity, and stress-associated regulatory networks remain under-synthesized in precisely the species that may prove increasingly valuable under intensifying climatic stress.

This issue is especially important in tropical systems. Tropical fruit, medicinal, and underutilized plants often accumulate distinctive classes of specialized metabolites and show stress responses shaped by evolutionary histories different from those of temperate model crops [23,25]. Yet the omics literature on these taxa is frequently scattered across case studies, generated under non-comparable experimental conditions, or limited to a single molecular layer [11,18,24]. Incomplete genomes and weak annotation reduce functional interpretation [20,21,26], while poor cross-study comparability and limited follow-up validation further slow synthesis [27]. A dedicated review is therefore needed not because omics data on tropical and underutilized plants do not exist, but because these data have not yet been integrated into a coherent framework for understanding abiotic stress acclimation [10,19].

The present review addresses that gap by examining how transcriptomics, proteomics, metabolomics, epigenomics, and integrative multi-omics approaches have been applied to abiotic stress acclimation in selected tropical and underutilized plants. Rather than providing another broad summary of plant stress omics, the review focuses specifically on species and systems that remain comparatively underrepresented in mainstream syntheses [18,19,24]. It evaluates what multi-omics approaches have already clarified in these plants; where interpretation remains limited by methodological and resource constraints; and how future work can move from descriptive response profiling toward stronger biological inference [6,7,16] and translational payoff [14,20]. By situating tropical and underutilized species more centrally within the stress-omics literature, the review aims to broaden the conceptual base of plant stress biology and highlight a more inclusive framework for understanding acclimation in a changing environment [10].

This review was conducted as a narrative synthesis of the primary and review literature on multi-omics approaches to abiotic stress acclimation in tropical and underutilized plants. Literature was retrieved through systematic searches of Web of Science, Scopus, PubMed, and Google Scholar, using keyword combinations including “multi-omics”, “abiotic stress”, “tropical plants”, “underutilized crops”, “transcriptomics”, “proteomics”, “metabolomics”, “epigenomics”, “stress acclimation”, and related species-specific terms. The primary search timeframe was 2015–2026, with foundational earlier studies included where necessary to establish methodological or conceptual context. Species were selected on the basis of two criteria: “tropical” refers to species whose cultivation range falls predominantly within the Köppen tropical climate zones (Af, Am, Aw/As), and “underutilized” follows the Bioversity International definition crops with recognized nutritional, economic, or ecological value that remain underrepresented in mainstream scientific research relative to their potential. Coverage depth was intentionally uneven and reflects the actual state of the literature; species with richer multi-omics datasets receive more detailed treatment, while less-studied species are highlighted to indicate research gaps. This review is distinguished from existing multi-omics stress biology syntheses by its deliberate centering of tropical and underutilized species as the primary analytical frame rather than as peripheral examples within model-crop-focused narratives.

The underrepresentation of tropical and underutilized plants in mainstream stress-omics synthesis is not simply a consequence of scientific neglect at the level of individual species. Rather, it reflects a broader asymmetry in plant research infrastructure [20,21,28]. Major crops and model plants dominate omics-based stress biology because they benefit from a long-standing accumulation of genomic resources, curated annotations, standardized reference datasets, and analytical pipelines that make systems-level interpretation comparatively efficient [26,29]. Tropical and underutilized species, by contrast, often enter the literature only through isolated case studies, and their datasets are frequently insufficiently connected to broader comparative frameworks [19,24].

A major part of this imbalance lies in genome-resource availability [20,21,26]. Functional omics interpretation depends heavily on the quality of the underlying genome reference. In species lacking well-assembled and well-annotated genomes, transcriptomic profiling becomes more difficult to interpret, proteomic peptide assignment becomes less complete, and cross-layer integration becomes inherently less reliable [29]. Recent reviews on plant pangenomes and orphan crop genomics have emphasized that genomic underrepresentation is not merely a technical inconvenience; it is a limiting factor that shapes which species can meaningfully participate in comparative biology, trait dissection, and translational crop improvement [30]. As a result, many tropical and underutilized plants remain visible only at the level of fragmented discovery rather than full functional interpretation [18,24].

This resource disparity also reinforces publication and funding bias. Once a species becomes genomically tractable, it is more likely to attract further studies, higher-quality functional analyses, and broader review inclusion [18,26,28]. Species lacking such infrastructure remain trapped in a low-visibility cycle: fewer genomic tools lead to fewer mechanistic studies, which in turn lead to less synthesis and less future investment [19,24]. Recent studies on orphan and underutilized crops have highlighted this structural problem repeatedly, while broader crop-diversification and pangenome perspectives argue that resilience and nutrition security will depend in part on widening the crop research base [30,31].

Analytical constraints further amplify this imbalance. Multi-omics research in non-model plants is not limited by sequencing alone. Annotation quality, orthology inference, metabolite identification, and data harmonization all become more difficult in species with incomplete references or sparse prior literature [21,26,32]. Metabolomics is particularly affected because many tropical and underutilized plants produce lineage-specific or poorly catalogued compounds that remain weakly represented in standard spectral libraries [22,33,34]. Similarly, transcriptomic and proteomic datasets generated from different tissues, developmental stages, and stress regimes are difficult to compare when common reference frameworks are lacking [20,24,29]. This means that even when omics studies are available, they often remain descriptive rather than mechanistically integrative [18].

Language and geography reinforce these technical barriers. Tropical crop research is concentrated in institutions across South and Southeast Asia, Sub-Saharan Africa, and Latin America, where access to large sequencing centres, proprietary metabolomics platforms, and long-term computational infrastructure is often more limited than in temperate-country institutions [19,24,28]. Much of the resulting work is published in regional journals or specialty outlets that are less frequently screened by global systematic reviews, further compounding invisibility [18]. Addressing this imbalance therefore requires not only better tools but also deliberate curatorial effort to incorporate regional literature into comparative synthesis.

For these reasons, the absence of tropical and underutilized plants from mainstream stress-omics synthesis should not be understood simply as a gap in coverage; it reflects a deeper imbalance in how plant functional genomics has been organized [18,24,28]. A review focused specifically on abiotic stress acclimation in these species is therefore justified not only as a matter of representation, but as a necessary step toward broadening the comparative base of plant stress biology. Bringing these systems into clearer analytical focus may reveal conserved stress-acclimation mechanisms [19], as well as lineage-specific strategies that remain obscured when synthesis is restricted to a small set of well-resourced crops [30,31].

Transcriptomics has been the most widely adopted omics approach in tropical and underutilized plant stress biology, largely because RNA-seq can be deployed even in species where proteomic, metabolomic, and genome-scale functional resources remain limited [6,10,16]. In many tropical species, transcriptomic profiling has provided the first system-wide view of abiotic stress responses, identifying candidate transcription factors, signaling components, transporters, and metabolic genes associated with acclimation [3,9,35]. At the same time, the interpretive value of these datasets depends heavily on experimental design, reference quality [21,29], and whether transcript-level patterns are later integrated with other biological layers. Representative transcriptome-led examples are summarized in Table 1.

Among tropical crops, drought remains the most extensively studied abiotic stress at the transcriptomic level. Early RNA-seq work in banana showed that drought-tolerant and drought-sensitive cultivars differed not only in the magnitude of their transcriptional response but also in the distribution of regulatory genes induced under water deficit [36]. In a comparative study of the tolerant cultivar ‘Saba’ and the sensitive cultivar ‘Grand Naine’, drought treatment altered the expression of 2268 and 2963 genes, respectively, with transcription factor families especially MYB, WRKY, bHLH, and NAC members accounting for approximately 15.9% of differentially expressed genes (DEGs) alongside protein kinases, lipid metabolism genes, carbohydrate remodeling enzymes, and secondary-metabolite-related pathways. That study remains important because it demonstrated that even within a single tropical crop, drought-responsive transcriptional reprogramming is genotype-dependent, suggesting that acclimation strategies may differ between cultivars, though functional confirmation of these differences remains limited.

More recent work has extended this picture by moving from static transcript catalogues toward more targeted regulatory interpretation. A study in Musa acuminata used time-course transcriptomics and downstream molecular analyses to identify MaGME777 and MabHLH770 as candidate regulators associated with drought-responsive expression, though their causal roles in tolerance remain to be validated functionally MaGME777 MabHLH770, illustrating the field’s gradual shift from transcript discovery toward candidate-gene resolution [40]. Complementary work on banana aquaporins has identified MaPIP11 as a stress-responsive water-transport component whose expression is associated with drought and salinity conditions, suggesting a potential role in acclimation, further refining the regulatory vocabulary of dehydration acclimation in Musa [41]. Although this remains far from a full multi-omics framework, it shows how transcriptomics in tropical crops is now beginning to generate more targeted candidate nominations rather than only descriptive DEG lists, even if functional validation of most candidates remains outstanding.

Cacao (Theobroma cacao) provides another strong example of the increasing maturity of transcriptomic stress research in tropical species. A 2026 integrated physiological and transcriptomic study of drought response in two cacao hybrid clones showed that drought tolerance may be achieved through distinct molecular routes rather than through a single common expression pattern: one clone relied predominantly on stomatal regulation and photoprotection, whereas a second achieved equivalent tolerance through enhanced osmotic adjustment and antioxidant buffering [37]. This is important for review framing because it emphasizes that transcriptomic variation among tolerant tropical genotypes may reflect alternative acclimation strategies, not merely stronger or weaker activation of the same pathways.

While drought has received the greatest attention, transcriptomic studies of other abiotic stresses are also increasing in tropical systems. In cassava (Manihot esculenta), a tropical staple crop of major importance in climate-vulnerable regions, heat-stress transcriptomics identified marked tissue-dependent differences in gene expression patterns. A 2023 study showed that heat-responsive molecular programs differed substantially between vasculature-associated tissues and leaf mesophyll: leaves preferentially enhanced pyruvate synthesis while mid-vein tissues showed stronger remodeling of sucrose and starch metabolism, carbon export, and phloem-associated transporters [38]. This observation is especially relevant in non-model tropical crops because many stress studies still rely on pooled tissue strategies that can obscure biologically meaningful cell- or tissue-level differences; emerging single-cell and spatial multi-omics approaches reinforce this point by showing how averaged tissue profiles can mask localized regulation [42] Complementary work has identified cassava A20AN1 zinc-finger proteins as candidates that interact with abiotic-stress signaling modules and are associated with drought and cold responses, providing a testable regulatory handle for future functional validation [43].

Oil palm (Elaeis guineensis) provides a useful example of transcriptomics at a broader stress-spectrum scale. A 2024 RNA-seq study examined leaf responses under drought, salinity, waterlogging, heat, and cold, and identified 19,834 DEGs in total across the five stresses, including 588 common DEGs shared across all five conditions [39]. The shared core set was dominated by kinases, NAC and ERF transcription factors, heat shock proteins, ubiquitin-related components, and genes associated with osmolyte biosynthesis, redox homeostasis, and secondary-metabolite pathways. The value of this work lies not only in the stress-specific DEG sets, but in the identification of a core set of stress-responsive transcripts shared across conditions, which provides a preliminary comparative framework for exploring how tropical crops may partition stress-general and stress-specific transcriptional responses. Similar comparative frameworks are beginning to emerge in other tropical crops, including date palm and pineapple, supporting the idea that stress-general regulatory backbones may be conserved across multiple tropical lineages [44].

Salinity and waterlogging studies in oil palm also highlight a broader point relevant to tropical plants: transcriptomic acclimation is often intertwined with environmental regimes that are especially important in humid tropical production systems, such as fluctuating rainfall, poor drainage, and combined osmotic stress [45]. A recent plantation-based transcriptomic study of oil palm stem responses to waterlogging showed that precipitation seasonality and field context shape expression outcomes in ways not easily reproduced in simplified controlled-environment experiments. This finding resonates with recent work in other tropical waterlogging-prone systems, including rice and banana, where root-oxygen availability and timing of flooding strongly influence transcriptomic outcomes [41,46]. It serves as a reminder that tropical stress transcriptomics often sit closer to ecological realism than many model-system studies but becomes correspondingly more difficult to standardize and compare.

Papaya (Carica papaya L.) represents another tropical fruit crop of direct relevance to abiotic stress transcriptomics, yet one that remains underexplored in proteomic and metabolomic terms. A tissue-resolved RNA-seq study of papaya under mild and severe drought—profiling leaves, sap, and roots across 18 cDNA libraries identified 8549 drought-responsive genes through reference-based analysis and 6089 drought-responsive unigenes through de novo assembly, with GO enrichment revealing that responses to ABA signaling, hormone signaling, sucrose metabolism, and suberin biosynthesis were strongly activated in roots under moderate drought stress, while oxidation-reduction and abiotic stress response pathways were enriched across all tissues under severe drought [47]. Co-expression network analysis of the same dataset identified 17 stress-related transcription factors as main regulatory hubs in leaves and roots, including WRKY70, MYB94, bHLH-ICE1, ABI5, ANAC072, and bZIP1 homologues, organized into tissue-specific and stress-severity-specific co-expression communities [47]. A complementary comparative transcriptomic study of a drought-tolerant wild papaya genotype against a susceptible commercial cultivar confirmed that six transcription factor genes CpHSF, CpMYB, CpNAC, CpNFY-A, CpERF, and CpWRKY were consistently more highly expressed in the tolerant genotype after 14 days of water deficit stress, with RT-qPCR validation confirming up to 47-fold higher expression levels for CpWRKY and 39-fold higher levels for CpHSF and CpMYB in the tolerant versus susceptible genotype [48]. Despite this growing transcriptomic foundation, proteomic and metabolomic characterization of Carica papaya stress responses is almost entirely absent from the current literature, positioning this species as a high-priority target for multi-omics integration among tropical underutilized fruit crops.

Underutilized cereals such as finger millet (Eleusine coracana), pearl millet (Cenchrus americanus), foxtail millet (Setaria italica), and proso millet (Panicum miliaceum) have likewise emerged as an important domain for tropical stress transcriptomics, combining intrinsic tolerance to drought, heat, and marginal soils with increasingly mature genome and transcriptome resources [49,50,51]. Transcriptomic studies in pearl millet and foxtail millet have highlighted drought- and heat-responsive networks involving hormone signaling, redox regulation, photosynthetic adjustment, and osmotic protection, reinforcing the view that millets provide tractable comparative systems for abiotic-stress biology. These transcriptomic foundations are revisited in a deeper case-study treatment in Section 7.5, where the broader multi-omics landscape in millets is considered alongside proteomic and metabolomic evidence.

Taken together, these studies show that transcriptomics has already generated substantial insight into abiotic stress acclimation in tropical species. Across crops such as banana [36,40,41], cassava [38,43], cacao [37], oil palm [39,45], and tropical millets [50,52,53], several recurring patterns are evident: strong reprogramming of regulatory genes, repeated involvement of hormone-responsive and redox-associated pathways, induction of transcription factor families such as MYB, WRKY, NAC, bHLH, and ERF, and substantial modulation of carbohydrate and secondary-metabolite metabolism. These are not trivial findings. They demonstrate that transcriptomics has been essential for bringing tropical species into molecular stress biology and for identifying the regulatory vocabulary of acclimation in species that historically lacked deep genomic resources.

At the same time, transcriptomics alone remains insufficient for a fully mechanistic account of stress acclimation. Differential expression does not necessarily predict protein abundance, enzyme activity, metabolite accumulation, or long-term physiological consequence [4,10,13]. This limitation is especially important in tropical and underutilized plants, where reference annotations remain incomplete [21,29] and many transcriptomic studies still end at pathway enrichment and candidate-gene nomination. As recent multi-omics reviews have emphasized, transcript-level studies are most informative when they serve as one layer in a broader integrative framework rather than as stand-alone evidence of regulatory mechanism [11,15]. For this reason, transcriptomics should be viewed not as the endpoint of stress-acclimation analysis in tropical plants, but as the foundation on which proteomic, metabolomic, and regulatory-level interpretation must be built.

Transcriptomics has provided the largest body of evidence for abiotic stress responses in tropical plants, but it captures only one layer of acclimation. Protein abundance and metabolite composition are more directly linked to stress physiology because they reflect the biochemical state of the cell at the point where regulation is enacted rather than only at the stage of transcript accumulation [13,33,54]. For this reason, proteomics and metabolomics are essential for moving beyond gene-level response catalogues toward a more functionally resolved view of acclimation. In tropical and underutilized plants, these two layers are especially important because many stress-related traits osmotic adjustment, antioxidant buffering, membrane stability, and secondary-metabolite remodeling are not adequately explained by RNA profiles alone [49,55].

Proteomic studies in tropical species remain fewer than transcriptomic studies, but they often provide stronger evidence of immediate functional adjustment under stress [54,55]. A representative example comes from banana, where iTRAQ-based proteomic analysis under salinity stress identified broad protein-level remodeling involving ROS-scavenging enzymes (SOD, APX, catalase), stress-related chaperones (HSP70, HSP90, sHSPs), photosynthetic machinery (RuBisCO small subunit, PSII proteins), ribosomal components, and proteins associated with energy metabolism [56]. This kind of evidence is important because it shows that acclimation in tropical crops is not simply a matter of altered transcript abundance, but also of selective stabilization, turnover, and reallocation of proteins directly tied to cellular protection and metabolic continuity. Tissue-specific proteomic work in pearl millet under drought stress has shown that protein-level reprogramming captures protective processes that are not always evident from transcript profiles alone, reinforcing the idea that proteome-level data can detect acclimation processes invisible at the transcriptome level [57].

Recent synthesis work in oil palm further reinforces this point. A 2024 review of low-temperature stress in oil palm concluded that proteomics has helped reveal post-transcriptional and biochemical adjustments not fully captured by transcriptomics, particularly in pathways related to redox homeostasis, membrane stability, photosynthesis, and stress-defense proteins [58]. Although much of the tropical stress literature is still transcript-heavy, the oil palm case illustrates that even in resource-limited perennial crops, proteomic data are beginning to refine the interpretation of acclimation by identifying molecular responses more closely linked to phenotype. Complementary proteomic analyses in cassava under combined drought and heat have revealed selective stabilization of cytosolic glutathione S-transferases and late embryogenesis abundant (LEA) proteins, extending the list of candidate acclimation proteins for tropical staples [59].

Metabolomics contributes a different kind of insight. While transcriptomics identifies candidate regulatory programs and proteomics captures parts of the active molecular machinery, metabolomics records the biochemical outcome of stress adjustment [22,32,33]. This is particularly relevant in tropical and underutilized plants, many of which are characterized by distinctive primary and secondary metabolite repertoires [23]. Metabolomics is especially useful for understanding osmotic adjustment, antioxidant status, central carbon metabolism, amino acid remodeling, and stress-associated specialized metabolites under drought, salinity, temperature stress, and related oxidative conditions [55,60].

An example is mango (Mangifera indica). In a combined metabolomic-transcriptomic study of cold stress in three mango cultivars, more than 1323 metabolites were detected and 22,526 DEGs were identified; the tolerant genotype showed stronger accumulation of flavonoids, terpenoids, lignans, coumarins, alkaloids, and multiple amino acid-related compounds under prolonged cold exposure [61]. These metabolite shifts were accompanied by transcriptomic evidence for altered phytohormone signaling, MAPK-related responses, and ICE–CBF–COR-associated cold-response pathways, but the metabolomic data were critical in showing the biochemical direction of acclimation rather than only its regulatory potential. The study is especially useful for this review because it demonstrates how tropical fruit stress biology can be interpreted through coordinated changes in both primary and secondary metabolisms.

Another example comes from coconut (Cocos nucifera). An integrated transcriptomic and metabolomic study of cold-stressed coconut seedlings identified 9968 DEGs and marked changes in amino acids, flavonoids, lipids, and phenolic-related compounds, together with stress-linked pathway activation in carbon metabolism, hormone signaling, and oxidative defense [62]. Although the study itself is integrative, its metabolomic dimension is especially relevant here because it shows how tropical perennial species may rely on metabolite reconfiguration as a major component of cold acclimation, even when their transcriptomic responses alone would be difficult to interpret fully. In tropical crops where growth and productivity are tightly constrained by relatively narrow thermal windows, such biochemical data are essential for understanding why some genotypes or species show stronger acclimatory capacity than others. Parallel metabolomics work on underutilized melon genotypes has similarly highlighted genotype-dependent accumulation of organic acids, amino acids, and flavonoid glycosides as correlates of water-deficit tolerance [60].

Proteomics and metabolomics are particularly valuable in tropical and underutilized species because they partly compensate for the interpretive limitations of incomplete genome annotation. In species where transcriptomic datasets are available, but gene models remain uncertain, protein-level evidence can validate which stress-associated pathways are functionally deployed, while metabolomic profiling can reveal whether those pathways converge on meaningful physiological outcomes such as osmoprotection, redox buffering, or secondary-metabolite adjustment [21,32,58]. Recent methodological reviews therefore argue that metabolomics and proteomics should not be treated as optional add-ons to transcriptomics, but as essential components of serious plant stress-acclimation analysis [4,16,33].

At the same time, these two approaches present challenges in tropical plants. Protein extraction is often complicated by high levels of polysaccharides, polyphenols, and interfering secondary compounds, requiring optimized TCA–acetone or phenol-based protocols that are still not widely benchmarked across tropical species [54,58]. Metabolite annotation remains difficult when tropical species produce lineage-specific compounds absent from standard databases such as HMDB, METLIN, or MassBank, often leading to large fractions of ‘unknown’ features in untargeted datasets [32,33,60]. Recent advances in computational metabolomics, including MS2 spectral prediction, in silico fragmentation, and deep-learning-based structure elucidation tools such as SIRIUS and MS2DeepScore, are beginning to narrow this gap, but their validation in tropical species is still preliminary. These technical constraints explain why transcriptomics still dominates literature, but they do not reduce the importance of proteomic and metabolomic data. On the contrary, they underscore the need for stronger biochemical characterization in tropical systems if abiotic stress acclimation is to be interpreted mechanistically rather than only descriptively.

Taken together, current evidence shows that proteomics and metabolomics already add substantial interpretive depth to tropical plant stress biology. Proteomics clarifies which stress-responsive proteins are accumulated or remodeled [56,57,58], while metabolomics reveals the biochemical phenotype of acclimation across central metabolism, osmotic adjustment, antioxidant defense, and specialized metabolite pathways [33,60,61]. In non-model and underutilized tropical species, these approaches are still developing, but they already demonstrate why transcriptomics alone cannot serve as a sufficient proxy for stress acclimation [13]. This transition from isolated molecular layers to integrated biological inference, illustrated with species-specific examples from the present review, is summarized in Fig. 1.

The central promise of multi-omics integration lies not in the accumulation of additional datasets, but in the transition from molecular description to biological interpretation [10,11,16]. In plant stress biology, transcriptomic studies have been highly effective in identifying responsive genes and candidate regulatory pathways, yet they rarely resolve whether those changes translate into altered protein abundance, biochemical adjustment, or meaningful physiological acclimation [13,15]. Multi-omics integration is therefore valuable because it allows evidence from different biological layers to be considered together, increasing confidence in candidate mechanisms and reducing the tendency to over-interpret single-omics patterns [63]. In tropical and underutilized species, where genome annotation is often incomplete and functional validation remains limited, this integrative logic becomes even more important [21,29].

A useful conceptual starting point is provided by the systematic multi-omics integration (MOI) framework proposed by Jamil and colleagues, which distinguishes between element-based, pathway-based, and mathematical integration strategies [4]. Element-based integration links features across layers through shared identifiers (for example, mapping transcripts to proteins to metabolites within a single pathway), pathway-based integration aligns layer-specific enrichment results, and mathematical integration applies dimensionality-reduction or network inference methods such as MOFA+, DIABLO, or multi-omics factor analysis to identify latent factors that span datasets [11,64]. This framework is particularly relevant to non-model plants because it recognizes that not all studies need to achieve full-scale data fusion to be biologically useful. Even partial integration such as connecting transcriptomic and metabolomic patterns at the pathway level, or interpreting transcriptomic data alongside physiological measurements can substantially strengthen inference when reference resources are limited [63]. In the context of tropical stress biology, this is an important point: mechanistic insight often emerges not from perfect data completeness, but from careful alignment of complementary evidence. Representative tropical case studies are summarized in Table 2.

Integrated studies in tropical species already illustrate this advantage. In coconut, combined transcriptomic and metabolomic analysis under cold stress identified coordinated changes in phytohormone signaling, amino acid metabolism, flavonoid biosynthesis, and lipid-associated pathways, producing a more interpretively complete picture of candidate stress responses than either layer could provide on its own [62]. The value of that study was not merely that it produced two datasets, but that metabolite shifts helped prioritize which transcriptomic changes were most plausibly connected to biochemically enacted stress responses under cold exposure. Similarly, in mango, integrated transcriptome–metabolome profiling across cultivars with different cold tolerance showed that differential accumulation of amino acids, flavonoids, terpenoids, and other compounds could be interpreted together with transcriptional changes in ICE–CBF–COR signaling, phytohormone responses, and MAPK-associated pathways [61]. In both cases, integration moved the analysis from what changed at the transcript level to which coordinated molecular processes are most plausibly associated with tolerance, though causal confirmation requires downstream functional validation.

In tropical crops, an important intermediate form of integration involves combining molecular profiling with physiological data. Although such designs are not always presented as full multi-omics studies, they often provide stronger biological interpretation than omics datasets alone [14,15,67]. For example, recent work in oil palm integrated transcriptomic profiling with physiological drought-response measurements and showed that molecular signatures can be interpreted more usefully when considered alongside growth, water status, and stress-associated functional traits, though this does not on its own resolve causal mechanisms [68]. Likewise, a 2026 cacao study demonstrated that two drought-tolerant hybrid clones could achieve tolerance through different combinations of physiological behavior and transcriptomic reprogramming, including osmoprotection, antioxidant regulation, photoprotection, and ABA-associated responses [37]. Integrated root-ionome, transcriptome, and metabolome analyses in the mangrove Avicennia marina (Table 2) have similarly shown that low-salinity conditions were associated with HAK8-mediated K uptake and ABA signaling at the transcript and metabolite levels, whereas high-salinity conditions were associated with Na sequestration, glycine betaine accumulation, and polyphenol remodeling—a pattern consistent with tiered acclimation strategies, though direct functional validation of these pathway assignments is still needed [65]. These studies are especially relevant to the present review because they show that in tropical perennial and halophytic species, acclimation cannot be reduced to differential expression alone; interpretation becomes much stronger when gene-level changes are considered together with plant-level performance under stress.

This broader view of integration is important for tropical and underutilized plants because fully developed transcriptome–proteome-metabolome datasets are still relatively uncommon in many of these systems [18,24,28]. In practice, the most informative studies are often those that combine whichever molecular layer is currently available with well-designed physiological or biochemical datasets. From a review perspective, this means integration should not be defined too narrowly. If the goal is to explain stress acclimation rather than only to profile response, then physiological and phenotypic evidence must be considered part of the integrative framework, especially in species where deeper resource development is still in progress [37,65,68].

Multifactorial and combinatorial-stress studies offer a particularly strong argument for integration [2,5,13]. A 2025 analysis of potato responses to combined heat, drought, and waterlogging showed that only 9% of stress-regulated genes were shared across all three conditions, but that combined stress triggered a distinct multi-layer signature involving ABA–ethylene crosstalk, proline and trehalose accumulation, and remodeled lipid composition that could not have been predicted from single-stress datasets. Analogous tomato multifactorial studies have shown that integrated analysis of transcript, protein, metabolite, and ionomic data reveals emergent response modules absent from any single stress. These findings matter for tropical biology because combinatorial stresses, especially drought plus heat, waterlogging plus salinity, and heat plus cold-night transitions dominate many tropical production environments.

A second major contribution of integration is candidate prioritization. In non-model tropical species, stress studies often generate long lists of differentially expressed genes, abundant proteins, or responsive metabolites, but relatively few of these candidates can be followed up experimentally [4,11,63]. Multi-omics integration helps reduce this problem by identifying points of convergence across evidence layers. Candidates supported by transcript abundance, pathway placement, metabolite association, and, where possible, physiological correlation are more likely to be biologically meaningful than those emerging from a single dataset alone [64]. This is one of the most important reasons why integration is not simply technical refinement but a strategic necessity in resource-limited systems.

Recent tropical studies show how this works in practice. In coconut, integrated analyses narrowed broad stress-response datasets toward specific pathways and compounds likely to be involved in tolerance, including amino acid metabolism, flavonoid remodeling, and hormone-associated regulatory circuits [62]. Comparable integrative work in mango highlighted coordinated metabolite and transcript shifts linked to cold tolerance [61]. In oil palm, comparative transcriptome work across drought, salinity, waterlogging, heat, and cold identified a core abiotic-stress network that may provide a useful scaffold for interpreting pathway convergence across stress types [39]. In pearl millet, comparative physiological and proteomic analysis highlighted stay-green behavior, root morphological regulation, and photosynthetic protection as key features of drought resilience [52]. Together, these studies suggest that the future value of multi-omics in tropical plants lies less in producing ever larger stress-response inventories than in generating more selective and biologically justified candidate sets for downstream validation.

Despite these advances, the transition from descriptive profiles to biological inference remains uneven. Multi-omics integration in tropical plants is still constrained by small sample numbers, inconsistent stress protocols, incomplete annotations, low metabolite-identification confidence for lineage-specific compounds, and limited cross-study comparability [4,32,33]. These constraints are not trivial. They affect the quality of network inference, the reproducibility of pathway-level conclusions, and the confidence with which candidate regulators can be prioritized [63,64]. In many tropical studies, integration still occurs after separate layer-specific analyses rather than through deeper model-based frameworks. That limitation does not invalidate current work, but it does mean that the literature remains stronger at identifying stress-associated molecular signatures than at fully resolving causal regulatory architecture [11].

For this reason, the current value of integrative studies in tropical and underutilized plants lies in two related contributions. First, they improve interpretation within individual species by connecting molecular layers that would otherwise remain isolated [4,16]. Second, they provide a basis for more rigorous comparative synthesis, allowing tropical systems to be discussed in terms of coordinated stress biology rather than only as disconnected omics case studies [18,24,28]. As more such datasets accumulate, the field will be better positioned to distinguish conserved acclimation strategies from lineage-specific responses, which is ultimately the main comparative contribution that tropical and underutilized species can make to plant stress multi-omics [19,30,31]. This general workflow is summarized in Fig. 2.

The transcriptome, proteome, and metabolome represent downstream layers of a broader regulatory system that determines how stress signals are perceived, interpreted, and translated into acclimatory responses [8,17,69]. In plants, these regulatory layers include transcription factor activity, RNA-based post-transcriptional control, chromatin-associated regulation, and various forms of stress memory [70,71,72]. Although these processes are now recognized as central to abiotic stress biology, their characterization remains highly uneven across the plant kingdom. In tropical and underutilized species, the regulatory dimension of stress acclimation is still less developed than the descriptive omics literature, with many datasets pointing toward likely regulators but relatively few studies progressing to mechanistic validation [35].

Case studies are most useful in this review when they do more than illustrate species diversity. They should also show how far stress multi-omics has progressed in different tropical and underutilized systems, and where the main bottlenecks remain [18,24,28]. The examples selected here therefore represent different stages of analytical maturity: resource-enriched but stress-underdeveloped fruit systems such as Garcinia; tropical staple crops moving from transcriptomics toward broader systems-level integration such as cassava; fruit crops with recent integrative cold-stress datasets such as coconut and mango; ecologically specialized halophytic systems in which multi-omics has begun to illuminate salinity acclimation directly, as in Avicennia marina; and underutilized cereals such as tropical millets. Together, these cases show that tropical plant research is moving beyond data absence [19], but still faces the challenge of converting fragmented datasets into coherent biological inference [30,31].

A persistent problem in tropical stress-omics is that many species still lack reference genomes of sufficient quality for dependable multi-layer interpretation. Even when assemblies are available, structural annotation often remains incomplete because plant genomes are frequently large, repetitive, heterozygous, or polyploid [21,26,29]. A 2024 analysis of crop genomes and gene prediction accuracy showed that annotation quality has direct implications for downstream proteomics, with typical peptide-to-gene mapping rates varying from 65% to over 90% depending on annotation completeness. This is especially relevant in tropical and underutilized plants, where stress studies often rely on draft genomes, incomplete gene models, or de novo transcript assemblies that are informative but not fully stable platforms for deeper multi-omics integration. Without stronger genome assemblies and better gene annotation, apparent biological gaps can easily reflect technical under-resolution rather than real absence of pathway components.

Recent pangenome work further complicates the picture in a productive way. Reviews now emphasize that single-reference genomes frequently miss structurally variable, presence–absence, and lineage-specific regions that can be important for stress-associated traits [20,26,28]. In underutilized species, where diversity is often poorly sampled and reference bias is strong, the absence of pangenomic resources means that biologically meaningful variation may remain invisible to standard transcriptomic or comparative analyses. Building a plant pangenome for a tropical species typically requires 10–30 high-quality accession genomes, which is achievable for rice and wheat but remains a stretch for orphan crops. In other words, incomplete reference frameworks do not just weaken annotation, they narrow the range of stress biology that can even be asked.

A second major bottleneck lies in stress phenotyping. Molecular datasets gain interpretive value only when they are linked to carefully measured physiological and phenotypic outcomes, yet this remains one of the weakest aspects of much plant stress literature [100,101,102]. In tropical and underutilized species, omics studies often include only limited endpoint measures, making it difficult to distinguish general stress response from meaningful acclimation [14]. Recent translational genomics perspectives stress that even with strong sequence resources, moving toward practical or mechanistic trait interpretation requires integration with robust phenotyping [103]. This is even more critical in species that are less experimentally standardized than major crops.

The challenge is not simply technical throughput. High-throughput phenotyping has advanced considerably through unmanned aerial vehicle (UAV)-mounted multispectral imaging, automated greenhouse platforms, and infrared and chlorophyll-fluorescence-based stress diagnostics, but review work in this area shows that the real difficulty lies in designing phenotyping systems that are biologically relevant, reproducible, and comparable across experiments [100,101]. Under abiotic stress, plant responses are strongly influenced by developmental stage, tissue selection, timing of sampling, and environmental context. If these variables are not aligned with molecular sampling, multi-omics datasets can become analytically rich but biologically ambiguous. In tropical species, where greenhouse control may be weaker and field realism often matters more, this problem is especially acute [45]. Standardized minimum information guidelines—analogous to MIAME or MIAPE—tailored for plant stress multi-omics would substantially improve comparability.

Metabolomics presents a distinct but equally serious challenge. A 2025 review on plant metabolomics in the multi-omics era makes clear that metabolite annotation remains one of the major bottlenecks in plant systems biology, especially when species produce lineage-specific compounds weakly represented in existing databases and spectral libraries [32]. Typical untargeted LC-MS datasets for tropical species still contain 40–70% unannotated features, and confidence levels of identification (MSI Level 1–4) are rarely reported consistently [33]. This issue is particularly relevant in tropical and underutilized plants, which often accumulate unusual or poorly catalogued specialized metabolites including xanthones (Garcinia), prenylflavonoids, lignans, and polyphenols [22]. In such cases, metabolomics can reveal clear biochemical change without always providing confident compound identity or pathway context. Emerging AI-assisted spectral-library tools and metabolite-identification platforms such as SIRIUS 5, MS2DeepScore, and MolDiscovery are beginning to close this gap, but validation in non-model tropical systems is still preliminary.

This annotation problem affects integration directly. Transcriptome–metabolome or proteome–metabolome analyses are most informative when pathway identities are reliable and metabolite assignments can be linked to specific biochemical processes [4,32,33]. In tropical systems, however, integration often occurs under conditions of incomplete metabolite confidence, incomplete genome annotation, and incomplete orthology mapping. As a result, current integrative studies are often stronger at showing coordination across layers than at demonstrating mechanism. This does not diminish their value, but it does mean that pathway-level inference in tropical non-model systems should be treated cautiously when metabolite identity, genome annotation [21,29], and cross-layer mapping remain incomplete.

Protein extraction and proteomic quantification present parallel challenges. Tropical tissues often contain high concentrations of polysaccharides, polyphenols, terpenoids, and specialized secondary metabolites that interfere with standard proteomic workflows [54,58]. TCA–acetone precipitation, phenol-based extraction, and detergent-based buffers each have trade-offs that are still not systematically benchmarked across tropical species, and data-independent acquisition (DIA) mass spectrometry, although promising, remains less widely adopted in tropical labs than data-dependent acquisition. Strengthening the proteomic pipeline for tropical species via standardized reference spectra, harmonized quantification protocols, and shared spectral libraries would substantially accelerate cross-layer interpretation.

A further challenge is that literature is still difficult to compare across species, even when omics datasets exist. Tropical and underutilized plants are often studied through isolated projects built around different stress protocols, tissues, sequencing depths, metabolomics platforms, and analytical pipelines [18,24,28]. This weakens the possibility of genuine comparative synthesis. Recent work on underutilized crop genomics argues that inclusivity in crop science depends not only on sequencing more species, but on developing more interoperable frameworks for using those data [19]. Community-level initiatives such as the African Orphan Crops Consortium, the Global Open Data for Agriculture and Nutrition network, and FAIR data practices for plant multi-omics are emerging as important structural correctives. Without stronger comparability, tropical stress multi-omics will continue to accumulate case studies faster than it builds generalizable biological understanding.

Finally, analytical infrastructure remains unevenly distributed. Computational demands of multi-omics integration including genome assembly, variant calling, multi-dimensional dimensionality reduction, and deep-learning-based metabolite annotation are substantial, and many tropical institutions operate without sustained access to high-performance computing, long-term data storage, or stable bioinformatic pipelines [30,103,104]. Cloud-based bioinformatic platforms (e.g., Galaxy, Terra, CyVerse, and the Plant Sciences Cloud) and regional consortium-based computing arrangements partially mitigate this constraint, but training and retention of bioinformaticians remain persistent challenges in tropical research ecosystems.

These methodological limitations do not imply that tropical and underutilized plant multi-omics is immature or unproductive. Rather, they define the conditions under which the field must now progress. Better genomes and stronger annotations [21,103], better-aligned phenotyping [100,101,102], more careful metabolite interpretation [32], standardized proteomic workflows, and more interoperable analytical design [24,104] are not secondary improvements; they are the main requirements for shifting the literature from fragmented molecular response profiles toward stronger frameworks of abiotic stress acclimation. Until these bottlenecks are addressed more systematically, the field will remain rich in signals but uneven in mechanism [4]. The major methodological bottlenecks constraining multi-omics interpretation in tropical and underutilized plants are summarized in Fig. 3.

The next phase of research on abiotic stress acclimation in tropical and underutilized plants will depend less on the continued expansion of isolated omics datasets and more on whether those datasets can be converted into robust comparative frameworks [14,16,20] and translational pipelines [103,105]. In major crops, recent work has shown that the real frontier is no longer simple data accumulation, but the coordinated use of genomic resources, high-resolution phenotyping, multimodal integration, and predictive analytics to generate biologically meaningful and agriculturally useful hypotheses. For tropical and underutilized species, this transition has only begun. Future progress will therefore depend on how effectively the field can close the gap between molecular profiling and explanatory power.

The first priority is the continued development of genome resources that are adequate for true multi-layer interpretation. In underutilized and tropical plants, fragmented or incomplete reference genomes remain a major obstacle to transcript annotation, peptide assignment, orthology inference, and comparative pathway analysis [21,26,29]. Recent work on orphan crop improvement and plant pangenomes makes clear that future gains will depend not only on sequencing representative genomes, but also on capturing structural variation, presence–absence polymorphisms, and gene-space diversity that single references cannot represent [20,28,30]. Long-read sequencing platforms (PacBio HiFi, Oxford Nanopore) and Hi-C scaffolding have reduced the cost of chromosome-level tropical crop assemblies from months to weeks, while telomere-to-telomere (T2T) assemblies are now achievable for selected tropical staples. In practice, future stress multi-omics in tropical plants should increasingly move toward pangenomic and diversity-aware frameworks rather than rely solely on a single accession or draft reference. Such a shift would make comparative stress biology more realistic and would reduce the risk of missing lineage-specific stress-associated genes and regulatory elements.

The second priority is to improve spatial and cellular resolution in stress studies. One of the major limitations of current tropical stress multi-omics is that most datasets are still generated from bulk tissues, which inevitably compress cellular heterogeneity into averaged profiles [106,107]. Recent reviews on single-cell and spatial omics in plants show that these technologies can resolve cell-type-specific dynamics [108], tissue-level organization [109], and the spatial logic of stress responses [110]. A 2024 best-practices review on single-cell RNA sequencing (scRNA-seq) emphasized that protoplast isolation efficiency, transcript dropout rates, and cell-type annotation consistency are key methodological parameters that must be reported transparently for reproducibility. Applications in rice, maize, and Arabidopsis have already revealed cell-type-specific stress-response atlases that reconfigure our understanding of drought, salinity, and heat acclimation. In the context of abiotic stress acclimation, this matters because many key processes including ion transport, vascular signaling, meristem maintenance, and localized antioxidant responses are unlikely to be understood fully from bulk RNA or metabolite profiles alone.

Spatial transcriptomics and emerging spatial metabolomics approaches for example, matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) and desorption electrospray ionization mass spectrometry imaging (DESI-MSI) further extend this logic by mapping expression and metabolite distributions in situ, enabling interpretation of stress responses in their anatomical context [109,110]. For tropical and underutilized plants, the immediate goal may not be routine single-cell deployment across all species, but rather the gradual introduction of these approaches in strategically chosen systems such as cassava roots under drought [106], banana leaves under combined heat and drought [107], or A. marina root tissues under fluctuating salinity, where they can reveal stress-response architecture masked by bulk omics.

A third direction is methodological rather than purely experimental. As omics datasets increase in size and complexity, integrative analysis will increasingly depend on machine-learning frameworks capable of extracting cross-layer patterns from genomic, transcriptomic, metabolomic, and phenotypic data [30,105,111]. Recent work suggests that these tools may be especially useful in resource-limited species because transfer learning and multimodal ranking can help prioritize candidates where direct validation capacity remains limited. Transfer learning from Arabidopsis or rice stress datasets may therefore be useful, but only if models are recalibrated carefully for tropical species with distinct genomes, metabolites, and ecological contexts.

This direction is especially relevant in tropical and underutilized plants, where sparse datasets, incomplete annotations, and small sample sizes often limit conventional integration. In this setting, AI-assisted tools are best used for feature ranking, transfer of information from better-annotated crops, and cross-layer pattern discovery, but only when predictions remain interpretable and are tested against independent biological evidence [30,105,111]. The future of tropical stress multi-omics therefore lies not simply in adopting AI as a fashionable add-on, but in using it to strengthen candidate prioritization, comparative interpretation, and biologically grounded hypothesis generation.

The fourth priority is to develop validation-oriented experimental pipelines. Many tropical stress omics datasets currently end at candidate nomination, with limited follow-up functional work [4,10,35]. Stress-focused validation should combine transient expression systems and stable transformation where feasible [14], CRISPR/Cas9-based editing and emerging base- or prime-editing tools [112], and natural-variation panels such as cassava GWAS populations [94] and pearl millet association panels [99]. These approaches would move tropical stress studies from candidate prioritization toward functional confirmation, especially where transformation remains inefficient.

The fifth priority is conceptual. Much of the existing literature still treats omics as a way to identify stress signatures, whereas the next phase should use multi-omics to define translationally relevant mechanisms. This means linking molecular responses to measurable physiological traits, candidate regulators, and eventually to selection or improvement strategies [14,103,105]. In underutilized crops especially, recent reviews argue that genomics and associated computational methods should no longer be viewed only as descriptive tools but as part of broader pathways toward crop improvement, diversification, and resilience [19,28,30]. For tropical systems, this will require a stronger culture of experimental comparability, more meaningful phenotyping, and a clearer pipeline from candidate identification to functional testing. Without that transition, the field risks remaining rich in omics data but poor in mechanistic and translational synthesis.

The final priority is intellectual rather than technical. Tropical and underutilized plants should not be treated simply as extensions of model-crop biology, but as systems capable of contributing their own comparative insights into stress acclimation. As work on underutilized crops increasingly shows, these species are important not only for resilience and diversification in agri-food systems, but also because they expand the biological range through which adaptation, metabolic plasticity, and stress-response diversity can be understood [19,24,28]. Future multi-omics research will be most valuable when it does not merely transfer tools from major crops to minor species, but uses these species to broaden the conceptual foundation of plant stress biology itself. In this sense, the future of tropical and underutilized plant multi-omics is not peripheral to the field; it is central to making that field more biologically inclusive and more relevant to global agricultural realities [30]. These interconnected five priorities are illustrated in Fig. 4.

Across the tropical and underutilized species examined here, convergent multi-omics evidence points to ABA signaling, ROS buffering, heat shock protein induction, and osmolyte accumulation as conserved acclimation mechanisms, while lineage-specific responses in Garcinia mangostana, Avicennia marina, cassava, and pearl millet highlight the additional biochemical diversity that a broader comparative framework can reveal.

At the same time, the review also shows that the field remains constrained by incomplete reference genomes, inconsistent phenotyping, limited metabolite annotation, uneven validation pipelines, and weak cross-study comparability. These limitations do not diminish the value of existing work, but they do explain why many studies remain stronger at identifying stress-associated signatures than at resolving regulatory mechanisms. Future progress will require stronger genome and pangenome resources, better-aligned experimental design, and more rigorous cross-layer integration. It will also require greater emphasis on regulatory and epigenetic interpretation and deliberate adoption of single-cell, spatial, and AI-assisted framework.

The broader significance of this field is not only agricultural but conceptual. Tropical and underutilized plants should not be treated merely as peripheral additions to model-crop biology. They represent biologically informative systems capable of expanding current understanding of stress acclimation, metabolic plasticity, and adaptive diversity. A more inclusive multi-omics framework one that places these species more centrally within comparative stress research and pangenome-aware crop biology will improve not only the study of resilient crops, but also the broader biological interpretation of plant acclimation under environmental change.