Cassava is grown across the tropics chiefly for its starchy tubers, but every harvest also leaves behind mountains of leafy biomass that are routinely burned or left to rot in the fields. That foliage is far from worthless: it can carry between roughly 16 and 40 percent crude protein, along with essential minerals and vitamins, making it a potentially valuable ruminant feed that does not compete with human food crops. The obstacles are equally real. Cassava leaves contain cyanogenic glycosides that can release toxic hydrogen cyanide, their high moisture drives rapid post-harvest deterioration, and their abundant fiber limits how much can be included in animal diets. A new study published in Bioresources and Bioprocessing reports that a carefully dosed cocktail of fiber-degrading enzymes and lactic acid bacteria can transform this awkward by-product into well-preserved, protein-rich silage with a microbial community dominated by a single desirable species.
The research team, working at the Institute of Subtropical Agriculture in Taoyuan County, Hunan, China, harvested fresh cassava leaves with young stems and petioles, chopped them into 2 to 3 centimeter pieces, and wilted them for four hours until the dry matter moisture reached around 60 percent. They then packed 300 grams of material into sealed plastic bags under vacuum, preparing 60 bags in total across five treatments, three ensiling durations of 0, 21 and 42 days, and four replicates each. The control group, designated CK, received only 2 percent glucose as a fermentable carbohydrate source. Four additive treatments built on that base: T1 added Lactobacillus plantarum at 3 × 10⁸ CFU per gram; T2 added glucanase at 833 U per gram alongside the bacterium; T3 added cellulase at 533 U per gram; and T4 combined both enzymes with the inoculant. The design was completely randomized, and fermentation and nutritional data were analyzed with two-way ANOVA testing treatment, ensiling duration and their interaction.
The clearest early signal came from acidification. Silage pH declined significantly over the ensiling period in all treatments, and the additive groups dropped faster and further than the glucose-only control. Among the treatments, T2, which paired glucanase with L. plantarum, recorded the lowest pH, 3.88 at day 21 and 3.78 at day 42, squarely within the 3.8 to 4.2 range generally considered characteristic of well-fermented silage. Lactic acid concentration rose over time, and by day 42 the treated silages produced between 5.92 and 6.03 percent lactic acid compared with 4.56 percent in the control, with T1 showing the highest value. Acetic acid told a complementary story: treated groups started with slightly higher levels, around 0.43 to 0.49 percent, which then declined sharply by day 21, while the control stayed relatively stable at 0.31 to 0.35 percent. Propionic acid remained low throughout, creeping from roughly 0.010 to 0.013 percent at day 0 to 0.020 to 0.024 percent at day 42, and butyric and iso-butyric acids were not detected in any treatment, all signs the authors interpret as desirable fermentation without secondary or clostridial activity.
Nutritional changes reinforced the fermentation picture. Dry matter declined significantly as ensiling proceeded, from an initial range of 37.9 to 39.2 percent down to 31.9 to 35.1 percent by day 42, with the enzyme-treated groups showing the largest decreases, a pattern consistent with active microbial consumption of soluble substrates. Crude protein, by contrast, increased during ensiling and peaked at 27.7 percent in T4 on the final day, with the enzyme-inoculated groups T2 through T4 generally outperforming both the control and the bacterium-only T1. Water-soluble carbohydrates fell as microbes consumed sugars, but T2 retained a comparatively high 9.84 percent after 42 days while the control dropped to just 2.40 percent. Ether extract rose slightly and ash content shifted significantly, reaching its highest value of 12.8 percent in T4, which the authors attribute partly to concentration effects as organic matter was degraded.
The fiber results are where the enzymes most visibly earned their place. Neutral detergent fiber, acid detergent fiber, cellulose and hemicellulose all decreased markedly with enzyme addition, particularly in T3 and T4, with strong treatment effects at P < 0.001. By day 42, T4, the dual-enzyme treatment, showed the lowest values of all: 46.8 percent NDF, 24.3 percent ADF, 14.8 percent cellulose and 22.5 percent hemicellulose on a dry matter basis. Acid detergent lignin barely moved, holding between 9.08 and 11.4 percent with no significant treatment or interaction effects, which fits the well-established resistance of lignin’s complex phenolic structure to both microbial and enzymatic attack during ensiling. The practical payoff showed up in relative feed value, a calculated index combining digestible dry matter and dry matter intake: it climbed from 68.3 in the control at day 0 to 139 in T4 at day 42.
To understand what was happening microbially, the team extracted DNA from the 42-day silage samples, amplified the bacterial 16S rRNA V3–V4 region and sequenced it on a DNBSEQ-G99 platform, processing the reads in QIIME2 with DADA2 to generate amplicon sequence variants. Alpha diversity analysis revealed a nuanced outcome. Richness metrics such as ACE, Chao1 and observed species did not differ significantly between the control and treated groups, meaning the additives did not reduce how many bacterial species were present. But the Shannon and Simpson indices, which capture diversity and evenness, were significantly lower in all treated groups than in the control, which showed the highest values of both. In ensiling, lower diversity is usually good news: it signals that desirable lactic acid bacteria have taken over and are suppressing unwanted microbes through rapid acidification.
Beta diversity analysis sharpened that interpretation. Non-metric multidimensional scaling of weighted UniFrac distances, with a very low stress value of 0.0185, showed the control forming a distinct, dispersed cluster clearly separated from the treated groups along the first axis, while the four treated groups clustered tightly together with only slight separation among themselves. At the phylum level, Bacillota dominated every group at day 42, with relative abundances of 92.2 to 96.1 percent, higher in treated silages though not statistically significant, while Pseudomonadota, the second-largest phylum at 3.73 to 7.35 percent, was less abundant in treated silages. Cyanobacteriota, Actinomycetota and Deinococcota appeared at low levels but were significantly reduced by treatment, a shift the authors view favorably since Cyanobacteria are associated with poor fermentation.
The species-level data delivered the study’s most striking numbers. Lactobacillus plantarum, the inoculated strain, dominated all treated silages at relative abundances of 93.0 to 94.6 percent, compared with only 13.1 percent in the control, a difference significant at P = 0.036. Early colonizers that thrive in the opening phase of fermentation were largely displaced: Pediococcus acidilactici fell from 35.8 percent in the control to between 0.03 and 0.76 percent in treated silages, Pediococcus pentosaceus dropped from 23.7 percent to 0.00 to 0.61 percent, and Weissella cibaria, Lactobacillus fermentum and Enterococcus mundtii also declined significantly. Notably, two potentially undesirable species, Klebsiella pneumoniae at 1.23 percent and Enterococcus faecium at 0.64 percent, were detected only in the control and were completely absent from every treated group. Small proportions of Bacillus licheniformis appeared in T3 and T4, which the authors suggest may have aided fiber degradation through extracellular enzyme production.
Spearman correlation analysis on day 42 tied these microbial shifts to silage chemistry. Most bacterial species correlated positively with pH and negatively with lactic acid, while L. plantarum showed the opposite pattern, a strong negative association with pH and a positive one with lactic acid, consistent with its role as a primary driver of rapid acidification. Klebsiella pneumoniae and Enterococcus faecium correlated negatively with lactic acid and strongly positively with acetic acid. On the nutritional side, the displaced early colonizers, including the Pediococcus, Weissella and Enterococcus species, correlated positively with dry matter, NDF, ADF, lignin, cellulose and hemicellulose and negatively with crude protein, water-soluble carbohydrates and ether extract, whereas L. plantarum showed exactly the inverse relationships.
Taken together, the results support the study’s central hypothesis that cellulase, glucanase and L. plantarum act synergistically: the enzymes hydrolyze structural polysaccharides into fermentable sugars that feed the bacteria, and the bacteria convert those sugars into lactic acid that locks in the improvements. Among all treatments, T4, combining glucanase at 833 U per gram, cellulase at 533 U per gram, L. plantarum at 3 × 10⁸ CFU per gram and 2 percent glucose, delivered the best overall performance. The authors frame this as a practical strategy to raise the feeding value of cassava leaves for ruminants in tropical and subtropical regions, converting an often-burned agricultural residue into a preserved, protein-rich forage. As with any bag-silage experiment, the findings come from controlled small-scale conditions, and the correlation analyses describe associations rather than proven causal mechanisms, but the convergence of fermentation chemistry, fiber data and microbial sequencing points in one consistent direction.
Subject of Research: Combined fibrolytic enzymes and lactic acid bacteria improve the fermentation quality, fiber degradation and microbial community of cassava leaf silage.
Article Title: Synergistic effects of fibrolytic enzymes and lactic acid bacteria on cassava leaf silage quality and microbial community
Article References: Hossain, M. D., Yan, Q., Wittayakun, S., Napasirth, V., Paengkoum, P., & Tan, Z. (2026). Synergistic effects of fibrolytic enzymes and lactic acid bacteria on cassava leaf silage quality and microbial community. Bioresources and Bioprocessing, 13(1), Article 157. https://doi.org/10.1186/s40643-026-01151-z
Image Credits: AI Generated
DOI: 10.1186/s40643-026-01151-z
Keywords: cassava leaves, silage, lactic acid bacteria, Lactobacillus plantarum, cellulase, glucanase, fiber degradation, fermentation quality, microbial community, ruminant feed, crude protein, relative feed value
News Source: William Thompson. (October 27, 2026). Enzymes and Bacteria Team Up to Turn Cassava Leaves into Better Livestock Feed. Scienmag.
