Fermentation Strategies to Enhance Feed Nutritional Value and Optimize Industry Resources

1. Introduction

This Editorial concludes the Special Issue “Fermentation Strategies to Enhance Feed Nutritional Value and Optimize Industry Resources”. The Special Issue examined how fermentation may improve feed quality and resource use through microbial processing, feed preservation, agro-industrial by-product valorization, and the evaluation of candidate feed microorganisms. Its eight contributions comprise one review, two laboratory-scale by-product fermentation studies, one 90-day cattle feeding trial, three laboratory-silo studies, and one in vitro strain-screening study followed by a controlled chick microbiota experiment. This range broadens the evidence base but also limits collection-wide generalization: each finding must be interpreted within the substrate, species, dose, season, storage period, and experimental scale evaluated [1,2,3]. This scope is relevant to broader concerns about feed–food competition, the productive use of food-system leftovers and grass resources, and the resource context of global agriculture [4,5,6].

2. Overview of Published Articles

Collectively, the papers address four linked knowledge gaps: comparative guidance on fermentation platforms; evidence for converting or substituting underused feed resources; dose- and context-specific regulation of ensiling; and methods for screening microbes or optimizing complex bioprocesses. They do not, however, support a single universal conclusion or establish one industrial solution. Contributions 2 and 8 provide laboratory compositional evidence; Contributions 4–6 evaluate small experimental silos; Contribution 3 supplies animal and economic evidence for one cattle system; and Contribution 7 reports microbiota associations without an untreated control or host-performance endpoints. The summaries below therefore separate reported results from interpretation and identify the principal evidence boundaries.

Zhang et al. [Contribution 1] reviewed solid-state fermentation (SSF) and submerged fermentation (SmF) in feed production. The review reports examples in which SSF achieved 2–3-fold higher volumetric productivity and 70–90% lower water use than SmF for solid residues, whereas SmF offered greater process control. It also compiles context-dependent values for antinutritional-factor reduction, including up to 95.5% for phytate and 98.8% for tannins, and reports examples of mycotoxin reduction and vitamin B12 enrichment. These figures originate from different studies and should not be treated as universally attainable benchmarks. The review discusses the Nrf2–KEAP1 pathway and other mechanisms but contributes no new experimental mechanism data. It also identifies heat and mass-transfer constraints in SSF scale-up and variation among regulatory frameworks [7].

Ibarra-Rondón et al. [Contribution 2] optimized Pleurotus ostreatus cultivation using a central composite design in liquid culture and an extreme-vertex design for oil-palm by-products. A medium containing 18.72 g/L glucose and 0.39 g/L urea yielded 8 g/L fungal biomass. After laboratory-scale fermentation of palm kernel cake, lignin decreased from 14.5% to 7.7% (46.9%), and crude protein increased from 15.9% to 27.0% (69.8%). Transcriptomic analysis identified frequent CAZy domains (GH3, GH18, CBM1, AA1, and AA5) with functions relevant to lignocellulose conversion. These results support compositional change under the tested conditions. They do not demonstrate ruminal digestibility, palatability, methane mitigation, animal performance, safety, or commercial-scale feasibility, none of which was directly measured.

Brunetto et al. [Contribution 3] compared complete replacement of soybean meal with dried distillers’ grains with solubles (DDGS) in 24 intact male Holstein calves (12 per treatment) during a 90-day feedlot study. No statistically significant differences were detected in body weight, dry matter intake, feed efficiency, or average daily gain (1.25 versus 1.28 kg/day; p = 0.92). This was not an equivalence or non-inferiority test. Moreover, neutral performance outcomes were accompanied by lower NDF and ADF digestibility (p = 0.03 and p = 0.05), higher ether-extract digestibility (p = 0.02), higher serum cholesterol (p = 0.03), and higher total ruminal short-chain fatty acids (p = 0.01). Three cases of acidosis occurred, and neither group reached the projected 1.5 kg/day gain. Under the prices used, the DDGS diet was 10.5% cheaper and profitability was higher; these economic estimates remain location- and price-dependent.

Amaral et al. [Contribution 4] evaluated lasalocid sodium, limonene essential oil, and a cinnamaldehyde–carvacrol blend in a 4 × 2 design using 40 laboratory TMR silos across summer and autumn. After 110 days, dry matter recovery depended on additive and season; the blend reached 97.24% in autumn. The same treatment showed aerobic-stability times of 125.30 h in summer and 149.30 h in autumn, compared with 102.87 and 121.13 h in the respective controls. The responses were not uniformly favorable: limonene increased pH in autumn, and the blend was associated with higher propionic and butyric acids and higher ADF and lignin in that season. The study measured fermentation and composition endpoints, not microbial taxa, feed intake, digestibility, milk production, or animal health; selective microbial inhibition and productive benefits therefore remain hypotheses.

In two independent dose–response trials, Amaral et al. [Contribution 5] tested 0, 200, 400, and 600 mg/kg dry matter of limonene or a cinnamaldehyde–carvacrol blend in laboratory TMR silos, with five replicates per dose and season. Quadratic models estimated limonene maxima of 97.9% dry matter recovery at 473 mg/kg, 5.87% dry matter lactic acid at 456 mg/kg, and 114 h aerobic stability at 203.7 mg/kg; stability declined at 600 mg/kg. No significant limonene effects were detected for pH, NDF, ADF, crude protein, or non-fiber carbohydrates. For the blend, modeled maxima were 98.2% dry matter recovery at 548 mg/kg and 131 h stability at 359 mg/kg. However, NDF and ADF increased, whereas non-fiber carbohydrates and total digestible nutrients decreased linearly with dose; higher doses also increased ammoniacal nitrogen and isobutyric acid. These trade-offs preclude a dose-independent claim of nutritional improvement.

Andrade et al. [Contribution 6] used 30 laboratory corn-silage units in a 5 × 2 factorial design (control, two doses of each essential-oil treatment, and 150 or 200 days of storage). Additives increased lactic acid, and in vitro dry matter digestibility showed an additive-by-time interaction (p < 0.01); the maximum, 81.5%, occurred with limonene at 150 days, whereas the cited 76.0% control value was measured at 200 days. Storage period, not additive, significantly affected dry matter recovery (92.46% at 150 days versus 84.02% at 200 days). Limonene also reduced ether-extract content. After 72 h of air exposure, treated silages had a smaller pH increase than the control, but temperature-based aerobic stability could not be interpreted because ambient conditions were unsuitable. Microbial populations were not measured. The data therefore support reduced deacidification under the test conditions, not a demonstrated selective inhibition mechanism or farm-scale stability [8].

Qin et al. [Contribution 7] screened 70 lactic acid bacteria and selected two Lactiplantibacillus pentosus strains, h8-c and p15-c, for biofilm formation, stress tolerance, and in vitro inhibition of five indicator bacteria. Fifty specific-pathogen-free chicks were then divided into two groups of 25, each receiving one strain for three days, followed by three weeks without supplementation. Fecal Lactobacillus relative abundance increased at day 3 and remained above baseline at day 21. However, there was no untreated control, and amplicon sequencing did not track the administered strains; these observations do not establish strain-specific colonization or causality. The p15-c group also showed transient enrichment of Gibberella and a more complex network including opportunistic fungi. Epithelial adhesion, barrier function, metabolites, immune responses, growth, and clinical health were not measured, so long-term intestinal-health claims are not supported by the reported endpoints [9].

Zhang et al. [Contribution 8] used 80 laboratory fermentation observations and repeated five-fold cross-validation to compare artificial neural-network (ANN) and random-forest (RF) models for two-stage fermentation of sweet-potato waste. For crude protein, RF achieved R² = 0.61 ± 0.04, compared with 0.12 ± 0.15 for ANN; RF performance for viable lactic acid bacteria was lower (R² = 0.34 ± 0.08), and ANN was slightly better for lactic acid (0.91 versus 0.89). An RF-based optimizer predicted 25.0% crude protein, and six laboratory replicates yielded 24.6 ± 0.4%. Measured total amino acids increased from 84.9 to 181.8 mg/g (114.1%), and essential amino acids increased by 123.9%. These findings demonstrate optimization within one bench-scale substrate–organism system. They do not yet establish external predictive validity, animal feeding value, process economics, or industrial scalability.

3. Synthesis and Research Priorities

3.1. Collective Insights and Evidence Boundaries

Taken together, the collection is strongest in showing that fermentation variables can alter substrate composition and ensiling indicators under specified experimental conditions. It also provides one small animal trial in which several performance measures did not differ significantly between protein sources, alongside metabolic, digestibility, health, and economic observations that were not uniformly neutral. The papers are complementary because they connect a field-level review with laboratory bioconversion, animal feeding, silage preservation, microbial screening, and data-driven optimization. The collection does not collectively demonstrate industrial applicability, equivalence of feed ingredients, persistent probiotic colonization, long-term physiological benefit, or the causal mechanisms proposed for antioxidant and selective antimicrobial effects. Those boundaries define the remaining knowledge gaps rather than diminish the value of the reported context-specific findings.

3.2. Immediate Research Needs

Immediate priorities are independent replication with prespecified primary outcomes, effect sizes, and confidence intervals. If equivalence of DDGS and soybean meal is the objective, an adequately powered equivalence or non-inferiority design is required. The palm-kernel-cake and sweet-potato processes need pilot-scale mass and energy balances, safety assessment, standardized compositional methods, animal digestibility and palatability tests, and techno-economic and life-cycle analyses. Silage additives require farm-scale validation across feedstocks, climates, silo types, and feed-out conditions, together with direct microbial quantification. Probiotic studies require untreated controls, strain-resolved tracking, host-health and performance endpoints, and larger trials under commercial conditions. Cross-study reporting standards would improve comparability of recovery, fermentation, microbiological, and economic outcomes.

3.3. Longer-Term Directions

Longer-term work may integrate real-time sensors and externally validated models into digital twins for process monitoring. Strain engineering and designed consortia could target specific substrates, provided that genetic stability, biosafety, containment, and regulatory requirements are addressed. Multi-omics may help identify candidate pathways linking fermented feed, microbial communities, and host responses, but causal claims will still require targeted perturbation experiments and independent validation. These directions should build on the replication, standardization, scale-up, and practical testing priorities above.