Natural rubber remains one of the world’s most vital agricultural commodities, essential for industries ranging from automotive manufacturing to healthcare.
With global production exceeding 14.5 million tonnes annually, the demand for high-yielding, resilient rubber trees (Hevea brasiliensis) has never been greater. However, traditional breeding methods, which rely on slow and labor-intensive processes, struggle to keep pace with this demand.
A groundbreaking study published in Industrial Crops & Products in 2025 offers a transformative solution: genomic selection (GS), a DNA-based tool that accelerates breeding by predicting plant potential early in their lifecycle.
The Challenges of Rubber Tree Breeding
Rubber trees are notoriously difficult to breed. They require six years to reach maturity, and their low female fertility limits genetic diversity.
For decades, farmers and scientists relied on clonal propagation—a method where branches from high-yielding trees are grafted onto rootstocks—to maintain productivity. While effective, this method is time-consuming and fails to address deeper genetic limitations.
In West Africa, where smallholder farmers manage 85% of rubber plantations, these challenges are particularly acute.
Côte d’Ivoire, now the world’s third-largest rubber producer, has seen rising demand for faster, more efficient breeding techniques to sustain its growing industry.
Genomic selection (GS) emerged as a promising alternative. By analyzing DNA markers (specific sequences in the genome that highlight genetic variations) across the genome, GS predicts traits like latex yield or disease resistance in seedlings, bypassing years of field trials.
Previous studies in rubber trees focused narrowly on single families or traits, but this research breaks new ground. It evaluates two genetically distinct families across four diverse sites in Côte d’Ivoire and Nigeria, offering insights into how genetic relationships and environmental factors influence breeding success.
Genomic Selection Transforms Rubber Tree Breeding Methods
The research centered on two biparental rubber tree families—groups of trees created by crossing two parent plants. The first family combined PB260, a high-yielding female clone, with RRIM600, a male clone prized for its adaptability. The second family paired PB260 with RRIC100, a male clone known for disease resistance.
These families were planted across four experimental sites: two in Côte d’Ivoire and two in Nigeria. Each site differed in climate, soil type, and management practices, allowing researchers to assess how environmental conditions affected trait expression.
A total of 674 trees were genotyped (analyzed for genetic differences) using 12,960 single nucleotide polymorphisms (SNPs)—DNA markers that act like genetic signposts, highlighting variations in the genome.
SNPs are critical in genomic studies because they help identify regions of the genome associated with specific traits. Phenotypic data (physical measurements of traits) including rubber production, sucrose content, and tree girth, were collected over several years.
Rubber yield, for instance, was measured by tapping trees every two days and analyzing latex output adjusted for environmental variables like tree size and block effects. Sucrose content, which influences latex quality, was assessed chemically, while tree girth provided insights into growth rates.
The study employed a genomic prediction model called Genomic Best Linear Unbiased Prediction (GBLUP). This model estimates breeding values (the genetic potential of an individual to pass desirable traits to offspring) by linking phenotypic traits to genome-wide markers. Researchers tested four validation approaches:
- Cross-validation within sites (using 80% of data to predict the remaining 20%).
- Full-sib predictions (training models on one site to predict another within the same family; full-sibs are siblings sharing both parents).
- Half-sib predictions (training across families with shared parentage; half-sibs share only one parent).
- Mixed training (combining full-sib and half-sib data).
Additionally, the team explored whether integrating quantitative trait loci (QTLs)—specific DNA regions statistically linked to traits—could enhance accuracy.
Using a Bayesian model called BayesC (a statistical method that estimates probabilities based on prior knowledge), they identified top SNPs (markers) associated with rubber production and incorporated them into the GBLUP framework.
Genetic Relationships Boost Rubber Tree Breeding Success
One of the study’s most striking revelations was the critical role of genetic relatedness (how closely individuals are related by ancestry) in prediction accuracy.
When models were trained and validated on full-sib populations (trees from the same parents), accuracy soared. For rubber production, full-sib predictions achieved an accuracy of 0.54, compared to just 0.17 for half-sibs. Similarly, sucrose content predictions reached 0.36 for full-sibs but dropped to 0.21 for half-sibs.
These results underscore a fundamental truth: closely related populations share more genetic information, enabling models to make reliable predictions.
However, the study also found that mixing full-sib and half-sib data in training populations yielded nearly comparable results. For rubber production, mixed models achieved an accuracy of 0.52—only slightly lower than full-sib-only models.
This suggests breeders could simplify logistics by using open-pollinated populations (where trees are pollinated naturally, resulting in a mix of half-sibs) without sacrificing precision. For resource-strapped programs, especially in regions like West Africa, this approach could democratize access to advanced breeding tools.
Environmental Factors in Rubber Tree Breeding Outcomes
Prediction accuracy fluctuated dramatically across sites, highlighting the impact of environmental conditions. At Site 1 in Côte d’Ivoire, rubber production predictions achieved an accuracy of 0.63—the highest in the study.
In contrast, Site 2, located in a different agroecological zone, saw accuracy plummet to 0.33. Similar disparities emerged for sucrose content, with Site 2 recording 0.51 accuracy versus 0.30 at Site 3 in Nigeria. Tree girth predictions also varied, ranging from 0.45 at Site 3 to 0.34 at Site 4.
These variations stem from genotype-by-environment (G×E) interactions—the phenomenon where a tree’s genetic potential interacts with local conditions like soil, climate, or management practices.
For example, a tree genetically predisposed to high yield might underperform in drought-prone areas without matching adaptive traits.
The study’s authors emphasize that breeders must tailor models to regional conditions. A one-size-fits-all approach risks poor performance, particularly in heterogeneous landscapes like West Africa.
QTL Limitations in Rubber Tree Genomic Selection
Despite high hopes, integrating quantitative trait loci (QTLs)—DNA regions linked to specific traits—into prediction models backfired. Adding markers linked to rubber production reduced accuracy across all sites. At Site 1, accuracy dipped from 0.63 to 0.62; at Site 4, it fell from 0.40 to 0.34.
This counterintuitive outcome stems from rubber production’s polygenic nature—the trait is influenced by hundreds of small-effect genes rather than a few large-effect QTLs. When models prioritize a handful of markers, they overlook the cumulative impact of countless minor genes, leading to incomplete predictions.
Furthermore, QTLs identified in one family or site often failed to generalize. For instance, a marker explaining 17% of phenotypic variation at Site 1 had negligible impact at Site 2.
This instability, compounded by environmental differences, renders QTLs unreliable for complex traits. The study concludes that genome-wide approaches like GBLUP, which consider all markers equally, remain superior for polygenic traits.
Trait Heritability in Effective Rubber Tree Breeding
While QTLs disappointed, the study reaffirmed the value of heritability—the proportion of trait variation explained by genetics.
Rubber production showed remarkably high heritability (up to 94%), meaning genetics play a dominant role in yield differences.
This makes it an ideal candidate for genomic selection. Sucrose content, with moderate heritability (56–69%), proved more challenging, as environmental factors like soil nutrients and rainfall exerted greater influence. Tree girth, though highly heritable (65–85%), still faced accuracy hurdles due to site-specific growth conditions.
These findings guide breeders toward prioritizing traits with strong genetic control. For example, investing in GS for rubber production could yield rapid gains, while sucrose content might require complementary strategies like environmental monitoring.
Implications for Farmers and Breeders
For smallholders in Côte d’Ivoire and Nigeria, this research offers tangible hope. By adopting genomic selection, breeding programs could slash evaluation times from decades to years.
Seedlings with high genetic potential could be identified at 1–2 years old, bypassing the six-year wait for maturity. This acceleration is critical for climate resilience; as pests, diseases, and droughts intensify, rapid trait improvement becomes a lifeline for vulnerable farms.
The study also advocates for collaborative breeding networks. By pooling data across regions, programs could develop robust models that account for environmental diversity. For instance, a model trained on data from both Côte d’Ivoire and Nigeria might better predict performance under varying rainfall patterns.
Future Research in Advanced Rubber Tree Breeding
Looking ahead, the authors outline several priorities. Expanding training populations (the group of individuals used to develop prediction models) to include more families and environments could enhance model reliability.
Integrating omics data—such as transcriptomics (study of gene expression) or metabolomics (analysis of metabolic profiles)—might uncover hidden genetic mechanisms. Machine learning tools could also refine predictions by detecting non-linear relationships between genes and traits.
Climate resilience remains a pressing focus. Identifying markers linked to drought tolerance or disease resistance could help breeders develop “future-proof” varieties. For example, a gene conferring resistance to Corynespora leaf fall, a devastating fungal disease, could be prioritized in high-risk areas.
Conclusion
This study marks a watershed moment for rubber tree improvement. By demystifying the roles of genetic relatedness and environmental variability, it equips breeders with tools to navigate the complexities of modern agriculture. While challenges like polygenic traits and climate uncertainty persist, genomic selection offers a path toward sustainable, high-yielding rubber production.
As global demand for natural rubber grows, the lessons from this research extend far beyond laboratories and test plots. For farmers in Côte d’Ivoire, Nigeria, and beyond, these advancements promise not just better harvests, but a brighter, more resilient future.
Key Terms and Concepts
Genomic Selection (GS):
A breeding method that uses DNA markers across an organism’s genome to predict traits like yield or disease resistance. Instead of waiting years for trees to mature, breeders can select the best plants early using their genetic data. In the study, GS predicted rubber production and sucrose content in rubber trees. Importance: Speeds up breeding cycles and reduces costs. Example: Using SNP markers to identify high-yield rubber clones.
Quantitative Trait Locus (QTL):
Specific regions of DNA linked to traits like latex production. QTLs help identify genes influencing complex traits. The study tested whether adding QTL data improved prediction accuracy. Importance: Guides targeted breeding. Example: A QTL on chromosome 24 was linked to higher rubber yield.
Full-Sib Families:
Plants or animals bred from the same two parents (e.g., siblings). In the study, full-sibs from crosses like PB260 × RRIM600 were used. Importance: Genetic similarity improves prediction accuracy. Example: Full-sib trees showed higher prediction accuracy (0.54) than half-sibs.
Half-Sib Families:
Individuals sharing one parent but not both. For example, trees from PB260 × RRIM600 and PB260 × RRIC100 are half-sibs. Importance: Tests genetic diversity. Use: Predicting traits across less-related populations. Accuracy was lower (0.17) compared to full-sibs.
Training Population:
A group of individuals with both genetic (DNA) and trait data (e.g., yield) used to build prediction models. In the study, training populations included full-sibs, half-sibs, or a mix. Importance: Determines model accuracy. Example: Mixing full-sibs and half-sibs gave accuracy comparable to full-sibs alone.
Prediction Accuracy:
How well a model’s predictions match real-world data. Measured using correlations (e.g., 0.54 for rubber yield in full-sibs). Importance: Higher accuracy means better breeding decisions. Formula: Pearson’s correlation coefficient between predicted and observed values.
Heritability:
The proportion of a trait’s variation caused by genetics (vs. environment). In the study, rubber yield had 85–94% heritability. Importance: High heritability means traits are strongly genetic. Example: Selecting high-heritability traits speeds up breeding.
Best Linear Unbiased Prediction (BLUP):
A statistical method to estimate genetic potential by adjusting for environmental factors. Used in the study to standardize rubber yield data. Formula: �=��+��+�, where � represents genetic effects.
Genomic BLUP (GBLUP):
A version of BLUP that uses DNA markers to calculate genetic relationships. The study applied GBLUP for genomic predictions. Formula: �=1�+��+�, where � is genomic effects.
Single Nucleotide Polymorphism (SNP):
DNA variations where a single base (e.g., A, T, C, G) differs between individuals. The study used 12,960 SNPs. Importance: SNPs act as genetic markers for traits. Example: SNPs on chromosome 24 predicted latex yield.
Genotyping-by-Sequencing (GBS):
A low-cost method to identify SNPs by sequencing parts of the genome. Used in the study to genotype rubber trees. Importance: Enables large-scale genetic studies. Example: GBS revealed 14,143 SNPs in Hevea brasiliensis.
Linkage Disequilibrium (LD):
When certain DNA markers are inherited together more often than expected. Important for GS because markers must be linked to trait-influencing genes. Example: LD between SNPs and QTLs affects prediction accuracy.
Genotype-by-Environment Interaction (G×E):
When a plant’s performance depends on environmental conditions (e.g., soil, climate). The study found G×E reduced QTL usefulness. Example: A QTL effective in Côte d’Ivoire might not work in Nigeria.
Phenotypic Data:
Measurable traits like latex yield or tree girth. The study collected standardized data across four sites. Importance: Links genetics to real-world performance. Example: Latex production ranged from 1,658 to 48,573 cg.
Genetic Relatedness:
How closely individuals are related genetically. The study compared full-sibs (high relatedness) and half-sibs (lower). Importance: Closer relatives improve prediction models.
Cross-Validation:
Testing prediction models by splitting data into training and validation sets. The study used 80% training and 20% validation. Example: Cross-validation accuracy for rubber yield was 0.63 at Site 1.
Minor Allele Frequency (MAF):
The frequency of the less common DNA variant at a locus. The study filtered SNPs with MAF < 0.05. Importance: Rare alleles reduce prediction reliability.
Imputation:
Filling in missing genetic data using statistical methods. The study used Beagle software for imputation. Importance: Ensures complete datasets for analysis.
Bayesian Methods:
Statistical approaches that update predictions as new data arrives. The study used BayesC for QTL detection. Example: BayesC calculated SNP inclusion probabilities.
Polygenic Trait:
A trait influenced by many genes (e.g., rubber yield). The study found QTL integration didn’t help due to polygenic effects. Importance: Requires genome-wide marker data.
Vulcanized Rubber:
Rubber treated with sulfur to improve durability. The end product of latex from Hevea trees. Example: Used in tires, industrial products.
Clonal Selection:
Breeding by copying high-performing trees through grafting. The study compared clonal vs. genomic selection. Importance: Traditional but slow (25–30 years per cycle).
Latex Yield:
The amount of latex a tree produces. A key trait in the study, measured in centigrams (cg). Example: PB260 clones had rapid latex increase after tapping.
Ethophon Stimulation:
A chemical treatment to boost latex flow. Not used in the study to avoid skewing natural yield data. Importance: Farmers often use it commercially.
Pseudochromosome:
A reconstructed chromosome using genome assembly tools. The study mapped SNPs to 27 pseudochromosomes. Example: QTLs on pseudochromosome ptg000024 influenced yield.
Reference:
Kouassi, D. K., Daval, A., Le Guen, V., Clément-Demange, A., Lopez, D., Mournet, P., Bonal, F., Hofs, J.-L., Soumahoro, M., Akaffou, D. S., & Cros, D. (2025). Enhancing genomic selection in rubber tree (Hevea brasiliensis): Exploring the impact of genetic relatedness and QTL integration. Industrial Crops and Products, 228, 120908. https://doi.org/10.1016/j.indcrop.2025.120908
