the trees the

cutters cannot see

Problem statement

Cacao swollen shoot disease (CSSD), caused by a complex of mealybug-transmitted badnaviruses, has been fought in Ghana for over seventy years with a single main tool: find infected trees and cut them out. The Cocoa Research Institute of Ghana (CRIG) reports the disease now has an estimated prevalence above 30% and can reduce yield by 30–50% or kill trees within two to three years of infection. The cutting-out program's weakness is that it depends on visual symptoms, and CRIG's own diagnostics work shows why that fails: latently infected trees show no symptoms and are "counted as non-infected," while the ten most-used CSSV-specific and degenerate PCR primers, tested on 189 field isolates from across Ghana, detected only 4% to 23% of them, because the viruses are genetically diverse, occur at low titre, and keep producing new variants. There is no reliable, affordable way to tell which symptomless cocoa trees are carrying the virus — so removal campaigns leave reservoirs behind, and breeders screening for resistance cannot be sure which trees are truly uninfected.

Why this matters

Ghana is one of the world's two largest cocoa producers and CSSD is concentrated in the western regions "where the bulk of Ghana's cacao beans are produced," so the disease bears directly on national export earnings and hundreds of thousands of smallholder livelihoods. Every missed latent tree defeats the purpose of cutting out its symptomatic neighbours — the mealybug vector simply reinfects replanted areas — which is one reason CRIG describes the program as having "persistently encountered numerous challenges emanating from farmer opposition, limited funding, and frequent discontinuities." Farmers who lose healthy-looking trees to a campaign that then fails to stop the disease have every reason to resist the next round. The same diagnostic gap slows the long-term solution: resistance breeding needs to distinguish resistant clones from clones that are merely latently infected.

What’s been tried and why it hasn’t worked

Eradication by rogueing has run intermittently since the 1940s and works only when infected trees are actually identified; latent and missed infections, especially on farms planted with tolerant varieties that mask symptoms, are named by CRIG among the causes of the program's inefficiency. Molecular diagnostics were meant to close that gap — full genome sequencing produced several CSSV-specific and degenerate primer sets — but CRIG's 2021 evaluation found "a highly variable and poor efficiency of the primers" (its 2023 review summarises the record as "erratic"), with the four best-performing primers reaching only 19–23% positivity, so a negative PCR result cannot clear a tree. Mild-strain cross-protection was tried and protects for a while, but CRIG reports the protection "breaks down between 15 years and 20 years post inoculation," and repeat inoculation of mature farms has no practical delivery method. Biocontrol of the mealybug vector largely failed (only one introduced parasitoid established and did not appreciably reduce populations), and more than 14 mealybug species are involved, two of which make up about 90% of the population. The knowledge gaps CRIG lists — the role of endogenous viral elements, mixed and co-infections, alternative forest and weed hosts, and vector population dynamics — mean the epidemiology needed to design smarter surveillance is also incomplete.

What would unlock progress

A field-usable test that reliably flags latent CSSV infection across Ghana's variant landscape would let cutting-out target true infection foci and let breeders verify their material; candidate routes are pan-badnavirus or region-tuned degenerate primer panels validated on the isolate collection, isothermal (LAMP/RPA) assays for field use, or hyperspectral / leaf-reflectance detection of pre-symptomatic infection, which has been demonstrated for other plant viruses. Adjacent precedent: cassava mosaic and citrus greening programs faced the same "asymptomatic reservoir" problem and moved to sequence-informed multiplex assays plus vector-based surveillance. Better characterisation of the variant population (which of the several species and variants dominate where) would let diagnostics be designed against what is actually in the field rather than against reference genomes.

Entry points for student teams

A bioinformatics team could take the published Ghanaian CSSV genome and variant sequences and design and in-silico validate a minimal degenerate primer set or CRISPR-based detection guide set covering the known diversity, then propose a validation protocol against CRIG's isolate panel. A sensing team could prototype a low-cost leaf-spectroscopy or smartphone-camera classifier for pre-symptomatic infection using published spectral signatures of viral stress, framed as a triage tool that tells surveyors which symptomless trees to sample. An operations-research team could model cutting-out effectiveness under different latent-detection sensitivities to show COCOBOD what test performance is actually needed to make rogueing pay for farmers. Relevant skills: molecular biology and primer design, spectroscopy and machine learning, epidemiological modelling, and agricultural extension.