Banana fibre yield and recovery in diversified farming systems of Uganda: Protocol, allometry and determinants

Abstract

The use of fibre from banana pseudo-stems is gaining traction globally. However, reliable tools for determination of banana fibre stocks and their determinants are lacking. We developed widely applicable allometric models for rapid evaluation of plant-level banana fibre yield and probed determinants for yield and recovery. A total of 346 pseudo-stems were purposively sampled from on-station and farmers’ fields across four banana genomes (AAA, AAA-EA, AAB, ABB) at pre- and post-flowering growth stages. Fibre was mechanically extracted, sun-dried, stored and weighed to determine fibre yield and recovery, whose determinants were assessed using generalized mixed-effects models. Genome-specific allometric models for estimating fibre yield, based on variables measured within (in-situ) and outside the field (ex-situ), were fitted using multiple linear regression analysis. Mean dry fibre yield was 146 g plant− 1 while recovery was 0.8%, highest in genomes AAA and AAB, respectively. Half of the ten developed models had a prediction accuracy (adj. R2) ranging from 65% to 79%, highest for AAA-EA and general models for in-situ and ex-situ conditions, respectively. Genetic and morphological pseudo-stem traits were the top determinants of yield and recovery, while physiological and phenological traits were specific to yield and recovery, respectively. Assuming 0.8% fibre recovery and current machine efficiencies, seven pseudo-stems yield 1 kg of dry fibre. The dominant genome AAA-EA is a potential source of raw material as fibre traits were similar to genomes AAB and ABB, which are currently preferred for high fibre output by fibre producers.