Thirty Years of Teak Data Reveal Which Clones Deserve to Stay

In the quiet research plots of the Tropical Forest Research Institute in Jabalpur, India, a living experiment has been quietly maturing for more than three decades. Planted in the early 1990s, a clonal trial of teak (Tectona grandis) containing genetic material collected from across the country has now grown into one of the most valuable long-term datasets in Indian forest genetics. A team of researchers from the institute’s Genetics and Tree Improvement Division, led by Divya Prakash and including Naseer Mohammad, Nikhil Verma, Kaushal Tripathi and Pramod Kumar, has finally subjected this mature plantation to a rigorous statistical interrogation, and the results, published in the Indian Journal of Genetics and Plant Breeding, offer both a blueprint for improving teak seed orchards and a striking demonstration of how modern statistical methods can extract genetic insight from decades-old field trials.

Teak is not an ordinary tree. Often called the king of timbers, it commands some of the highest prices in the global wood market thanks to its durability, water resistance, and dimensional stability. India hosts a large share of the world’s natural teak forests, and the species underpins enormous plantation economies across South and Southeast Asia. Because teak is so valuable, foresters have long sought to improve it through selective breeding, and clonal seed orchards, plantations established from grafted or rooted copies of elite trees, are the workhorse of that effort. The logic is simple: if you plant copies of the best trees together, they will cross-pollinate and produce seed that carries superior genes into the next generation of plantations. But the logic only works if the clones in the orchard really are the best, and that is precisely what the Jabalpur team set out to verify.

The trial they examined is remarkable for its age and provenance diversity. It contains ramets, the individual vegetative copies of each clone, sourced from different teak-growing regions of India, and the trees have now passed the thirty-year mark. That maturity matters enormously. Early evaluations of clonal trials, conducted when trees are five or ten years old, can be misleading because growth trajectories diverge over time, competition reshapes stands, and traits like bole straightness or forking only express their full economic significance in mature stems. A thirty-year-old trial captures the traits that actually determine timber value at harvest, making it a far more reliable guide to which clones should father the next generation of commercial plantations.

The researchers evaluated both quantitative growth traits and qualitative characteristics. The growth traits included clear bole height, the length of trunk free of branches and therefore the most valuable sawn timber; girth at breast height, a standard forestry measurement taken at 1.37 meters; total tree height; and computed wood volume. The qualitative traits covered straightness of the bole, compactness of the crown, the presence or absence of forking, overall tree health, and flowering intensity. Each of these matters in its own way: forked stems reduce usable timber, sprawling crowns signal poor growing-space efficiency, and flowering behavior in a seed orchard directly influences how much genetic contribution each clone makes to the seed crop.

What sets this study apart methodologically is its use of a linear mixed model, or LMM, a statistical framework that has become the gold standard in genetics and plant breeding but is still underused in some forestry contexts. In an LMM, some effects are treated as fixed, meaning the researcher wants to estimate and compare them directly, while others are treated as random, meaning they represent a sample from a larger population and account for background variability. Here, the clones were modeled as fixed effects, since the team wanted to rank and compare each one, while replication within the trial was modeled as a random effect, absorbing the environmental noise of position in the field. Variance components were estimated using restricted maximum likelihood, or REML, a technique that produces unbiased estimates even in unbalanced designs where some trees have died or grown unevenly over three decades, an almost inevitable feature of any long-term field trial.

From the fitted model, the researchers calculated estimated marginal means, or EMMs, for each clone, which are model-adjusted averages that level the playing field across the trial’s spatial variability. Pairwise comparisons between clones were then carried out with Sidak’s post hoc test, a correction that controls the risk of false positives when many comparisons are made simultaneously. The analysis revealed statistically significant variation among clones in every growth trait measured: clear bole height, girth at breast height, total height, and wood volume all differed meaningfully from clone to clone. Just as importantly, all the growth traits were significantly correlated with one another, meaning that clones that grew tall also tended to be thick and voluminous. That correlation is good news for breeders, because it suggests that selecting for one growth trait will tend to pull the others along with it, simplifying the selection process.

To visualize the overall genetic landscape, the team applied hierarchical clustering, a technique that groups clones based on the similarity of their measured traits. The clustering sorted the collection into five distinct clusters, and the pattern that emerged was striking. Cluster V contained the stars of the trial: clones APNPL-4, KLK-1, ST-16, and MHALP-1, which posted the strongest values across the growth traits. At the other end of the spectrum, cluster IV contained clones ORPDP-29 and MYSA-2, which showed low values not only for growth but also for one or more of the qualitative traits. The practical recommendation that follows is one of the most actionable findings in the study: the underperforming clones in cluster IV should be rogued out of the orchard, meaning physically removed, so that their inferior genes no longer contaminate the orchard’s seed crop. Roguing is a standard but powerful tool in seed orchard management, and every inferior clone removed translates directly into increased genetic gain in the progeny raised from the remaining trees.

Among the top performers, one clone stood out as nearly ideal. KLK-1 recorded high estimated marginal means for all the growth traits and showed desirable values for every qualitative trait except flowering. That single caveat is worth pausing on, because in a seed orchard, flowering is not a cosmetic trait; it is the mechanism by which a clone transmits its genes to the next generation. A clone that grows superbly but flowers sparsely will contribute less pollen or fewer flowers to the seed crop than its genetic merit would justify, diluting its impact on the progeny. The finding illustrates a recurring tension in seed orchard genetics: growth performance and reproductive output must both be considered when deciding which clones to retain, and a clone that excels in one dimension may need supplemental management, such as crown exposure or flowering stimulation, to pull its full weight in the orchard.

Perhaps the most conceptually interesting result concerns geography. A long-standing assumption in provenance research is that material collected from sites environmentally similar to the planting site should perform better, an idea rooted in the logic of local adaptation. Yet in this trial, the nearness of the test site in Jabalpur to the place of origin of each clone turned out to be inconsequential for growth traits. Clones from distant regions performed just as well, or as poorly, as clones from nearby ones, once the statistical model accounted for the trial’s structure. For orchard managers, this is liberating news: it suggests that the search for elite teak germplasm need not be constrained geographically, and that provenance trials across India can be mined for superior clones regardless of where the planting site sits on the map.

Beyond its immediate practical recommendations, the study is a quiet argument for patience in forest genetics. Trees operate on decadal timescales, and the traits that matter most, mature wood volume, stem form, long-term health, cannot be reliably judged in saplings. By revisiting a thirty-year-old clonal trial with contemporary statistical tools, the Jabalpur team has shown that old plantations are not merely aging assets but data goldmines, capable of answering questions their planters could scarcely have framed when the first ramets went into the ground. As India and other teak-growing nations face rising timber demand, shrinking natural forests, and a changing climate, the ability to squeeze maximum genetic information from long-term field trials, and to act on it by roguing the weak and multiplying the strong, may prove one of the most cost-effective tools in the entire forestry toolkit. The next generation of teak plantations, and the markets that depend on them, will be quietly shaped by decisions like these made among thirty-year-old trees in central India.

Subject of Research: Genetic evaluation of growth and qualitative traits in a mature teak clonal seed orchard using linear mixed models

Article Title: Genetic Evaluation of Teak Clonal Seed Orchard for Key Traits Using Linear Mixed Model

Article References: Prakash, D., Mohammad, N., Verma, N., Tripathi, K., & Kumar, P. (2026). Genetic Evaluation of Teak Clonal Seed Orchard for Key Traits Using Linear Mixed Model. Indian Journal of Genetics and Plant Breeding, 86(2), 241-251. https://doi.org/10.1007/s44489-026-00016-1

Image Credits: AI Generated

DOI: 10.1007/s44489-026-00016-1

Keywords: teak, Tectona grandis, clonal seed orchard, linear mixed model, REML, genetic gain, hierarchical clustering, growth traits, roguing, forest genetics, tree breeding, India