Agro-Industrial Wastes as Substrates for Solid-State Fermentation: Substrate Properties, Microbial Responses and Bioproduct Formation: A Review

Full text

Text extracted from the article PDF for reading and search. Tables, figures and equations are in the PDF, which remains the authoritative version.

1. Introduction

Brans, husks, bagasse, pomaces, press cakes and spent grains are produced as a by-product of agro-industrial processing which cost little when compared to refined fermentation media. Cost of raw materials can account for more than 40% of bioprocess costs (Borkertas et al., 2025) [3] and substrate cost can be assigned as 30-40% of total production cost (Yazid et al., 2017) [28]. The functional performance of a residue is influenced by the polymers to be degraded, the amount of water available, the permeability of the product bed to respiration and inhibitory materials present at the surface of the particle(s).

Solid state fermentation (SSF) is the culture of microorganisms by growing on moist solid particles without a significant amount of free water. The main organisms are filamentous fungi due to the ability of hyphae to penetrate the particles, presence of enzymes and tolerance to low water activity (Borkertas et al., 2025) [3]. Lower capital cost (Borkertas et al., 2025) [3], need for less agitation and aeration equipment, unsterilized substrates and reduced wastewater (Yazid et al., 2017) [28] are some of the attractions reported. Enzymes, organic acids, biosurfactants, pigments, phenolic compounds and aroma compounds are different types of products produced (Soccol et al., 2017; Yazid et al., 2017) [24, 28].

It is still difficult to use the literature for making decisions, however. Most studies are conducted in a flask with one residue–strain combination, the pilot-scale testing is limited (typically 100–10,000 L), and there is not much data available for bioreactor scale testing (typically 0.5–10 L) (Astudillo et al., 2023) [2]. While it is often claimed that yield depends primarily on the composition of the wastes, the reported waste composition is given in mixed units and is not reported in conjunction with physical and water-related properties that determine whether even colonization occurs. This review presents the evidence in a single chain: substrate properties affect the microbial responses, which in turn affect the product formation, selectivity, and reproducibility (Figure 1). Each link is considered as being modified by the pretreatment and process conditions. The objectives are: (1) to identify from experimental evidence which properties of the substrate are important; (2) to outline the available evidence to demonstrate consistency, conflict and the lack of evidence; and (3) to provide a qualitative framework for choosing and conditioning residues for a specific microbial system and product class. It is not a predictive model but rather reflects the synthesis of published work and any claim that it is the first is disclaimed.

Fig 1: Conceptual framework linking substrate properties (A), microbial responses (B) and bioproduct classes (C) in solid-state fermentation. Solid arrows show the direction of influence synthesized in this review; dashed arrows show that pretreatment, blending, moisture adjustment, inoculum and process conditions modify each link. Substrate icons are schematic. Relationships are drawn from the studies discussed in Sections 3–6 (e.g., Astudillo et al., 2023; Borkertas et al., 2025; Costa et al., 2023; Nargotra et al., 2023) [2, 3, 7, 17] and represent qualitative associations, not quantified effects. C:N, carbon-to-nitrogen ratio.

2. Literature Search and Review Scope

This is a critical narrative review, not a systematic review as there is no screening counts reported nor the PRISMA flow. It searched on August to October 2026 using a general web search engine, then repeated several times with a limited search engine that only included those of the publishers (MDPI, Frontiers, BioMed Central, Springer and ScienceDirect) and opening candidate records on the publisher's article page. The retrieved set is not exhaustive, as no subscription indexes (as Scopus, Web of Science, PubMed, etc.) and DOI-registry services were able to be queried. The search words used were “solid-state fermentation” and “agro-industrial residue”, “water activity”, “moisture”, “particle size”, “packed-bed bioreactor”, “pretreatment”, “inhibitors”, “oilseed cake”, “lipase”, “protease”, “xylanase”, “cellulase”, “citric acid”, “biosurfactant”, “pigment”, “mycotoxin” and “techno-economic”.

Studies from 2020-2026 were considered more relevant, and a few older studies were retained, such as those that provided background information for water activity (Corona et al., 2005) [6] or general overview of the field (Soccol et al., 2017; Yazid et al., 2017) [24, 28]. The numbers given in the primary studies are used; numbers from reviews are marked as such. If the title, authors, journal, year and volume or article number could be read on the publisher page, then the source was retained, 28 sources met this criterion. It is here where what has not been seen is not seen, and no pigments or other bioactive metabolites are seen. Comments on the strength of evidence are qualitative judgments.

3. Industrial Wastes and Their Substrate Properties

3.1. Substrate Classes and Composition

Residues are readily separated into the predominant polymer groups (Table 1). Lignocellulosic residues are mainly composed of cellulose, hemicellulose and lignin, with reported ranges ranging from 35–50%, 20–35% and 15–30% (Nargotra et al., 2023) [17] and 35–50%, 25–30% and 25–30% (Yazid et al., 2017) [28], which vary with species, plant part and method. The hydrolysis rate of polymers is strongly influenced by the quantity present as well as the accessibility; the amorphous cellulose is hydrolyzed 3–30 times faster than the crystalline cellulose, and the hydrolytic enzymes do not bind to the lignin in an effective and specific way (Nargotra et al., 2023) [17]. Bagasse is also of practical value that is not reflected in its composition: its fibrous texture contributes to aeration and heat removal since its matrix is composed of lignin, which resists degradation (Martins et al., 2026) [15].Residues rich in starch and sugar have different properties. The water content in fruit and vegetable residues ranges from 70 – 90% while the carbohydrate content is 20 – 30% in citrus pulp and up to 60 – 70% for potato peels in dry weight (Borkertas et al., 2025) [3]. The yields for citric acid were similar: fruit wastes yielded the highest, and cereal coproducts, the lowest (West, 2023) [27]. Wheat bran has been used as a source of hemicellulose (rich in arabinoxylan), protein, minerals and vitamins and also retains moisture (Martins et al., 2026) [15], which are advantageous for production of cellulase from the wheat for similar reasons (de Moura et al., 2026) [9].

The protein content of residues, mainly oilseed cakes and most meals, give preference to products requiring nitrogen and fats. Meals such as rapeseed meal and sunflower meal have a crude protein content of 38.3 and 37.0, respectively (Sentís-Moré et al., 2023) [20]; while soybean meal has been used as a rhamnolipid substrate (Dabaghi et al., 2023) [8]. These materials can contain anti-nutritional components like that gossypol in cottonseed meal and the glucosinolates in rapeseed meal (Wei et al., 2026) [26].

Table 1: Representative agro-industrial wastes and substrate characteristics relevant to solid-state fermentation. Values are as reported in the cited sources; ranges from reviews are labeled as such. dw, dry weight; dm, dry matter. Substrate class Representative waste Relevant properties Suitability considerations Limitations Sources Lignocellulosic Sugarcane bagasse, straw Reported ranges: cellulose 35– 50%, hemicellulose 20–35%, lignin 15–30%; fibrous, porous Bulking agent; improves aeration and heat dissipation; source of hemicellulose for xylanases Crystallinity and lignin limit access to polymers Nargotra et al., 2023; Martins et al., 2026; Mitchell et al., 2023 Cereal bran Wheat bran Arabinoxylan-rich hemicellulose, protein, minerals, vitamins; retains moisture Frequently a strong substrate for fungal hydrolases Alone, packed beds heated by 7–16 °C (low porosity)

Martins et al., 2026; de Moura et al., 2026; Casciatori et al., 2024 Mixed polysaccharide– protein Brewer’s spent grain >60% dw non-starch polysaccharides; 10–26% protein Responds to thermal pretreatment (ohmic heating)

Spoils in storage if not cooled Borkertas et al., 2025; Silva et al., 2025 Sugar-rich, high-moisture Citrus pulp, potato peels, apple pomace Water 70–90%; carbohydrate 20–30% dw (citrus pulp) to 60– 70% dw (potato peels)

Highest citric acid yields were on fruit wastes Moisture basis often unstated; cereal coproducts gave lower acid yields Borkertas et al., 2025; West, 2023 Protein-rich Rapeseed, sunflower and soybean meals Crude protein 38.3 and 37.0 g/100 g dm (rapeseed, sunflower)

Substrate for proteases, lipases and biosurfactants Anti-nutritional factors (e.g., gossypol, glucosinolates)

Sentís-Moré et al., 2023; Wei et al., 2026; Dabaghi et al., 2023 Amended or blended Digestate + food waste + sawdust Growth only when food waste was added; sawdust lowers humidity and raises porosity Blending as a design variable Growth varied among Trichoderma species Alias et al., 2022

3.2. Physical Structure

Whether an advantageous composition can be utilized depends on the movement of air, water and heat through the bed which is dependent on particle size, porosity and packing. However, hemp and rice husk initial porosities have been reported at 0.48 and ~0.85, respectively, and a pilot bed of 50% wheat bran and 50% bagasse had an apparent dry bulk density of 90 kg m−3 (Mitchell et al., 2023) [16]. Myceliophthora thermophila was used to pack endoglucanase and xylanase, generating > 850 and > 3,850 U g−1 temperature rises, respectively, in the sugarcane bagasse fibers/wheat bran (7:3, w/w) pack, while heat rises above 7 °C were obtained in the pure wheat bran, and above 16 °C in power fibre bagasse packs; the lower bed porosity in the latter was responsible for the higher temperature rises. The units for the U g−1 values are not mentioned, which is again discussed in Section 7.

3.3. Water: Moisture Content and Water Activity

The moisture content (water per mass of substrate at a stated moisture content - either on a wet or dry basis) and water activity (aw, the thermodynamic availability of the water) are not the same amount, but they are related by a sorption isotherm that is dependent on the solid matrix. On wheat bran and starch, Gibberella fujikuroi required increasing moisture to maintain the same aw during fermentation with optimal growth and GAA production rate and yield around aw 0.99 (Corona et al., 2005) [6]. The authors noted that superabsorbent polymers provided additional water, which increased biomass, but not spore production, which was thought to be due to water films limiting aeration with extra water, since microbes utilised mainly capillary water, which is lost in SSF.

The moisture content retained in primary studies ranges from 60% (Sentís-Moré et al., 2023) [20] to 85% (Rana et al., 2021) [19], with units being per dry matter (w/w) (Silva et al., 2025) [22] and per initial humidity (Dabaghi et al., 2023) [8]. However, fungi are reported to require 40 to 60% moisture, and its thermophilic fungus produced the maximum xylanase activity at 80% (Yazid et al., 2017) [28], while the bacteria require 60 to 85% moisture (Martins et al., 2026) [15]. Since the moisture optimum is different for each matrix, and neither remains stable throughout the run (Corona et al., 2005) [6], moisture optima can't be transferred from other matrices. For most of the retained primary studies, however, moisture content was not reported along with aw.

3.4. Nutrients, pH and Inhibitory Compounds

Product selectivity and growth are influenced by nutrient balance. The thermophilic fungus above achieved the maximum xylanase activity on water alone, as well as added nitrogen source from the bagasse–straw–wheat bran mixture, and added nitrogen source did not yield more enzyme activity than did the water alone, the authors conclude that this was because the mixture supplied sufficient nutrients (Martins et al., 2026) [15]. Contrary to this, in rhamnolipid production on soybean meal, glycerol had a high effect, humidity, temperature and inoculum size, and in citric acid production, methanol (usually 2-6%) and other supplements were used also (Dabaghi et al., 2023; West, 2023) [8, 27]. C:N ratio, minerals and buffering capacity were seldom studied as specific factors.

There are two sources of inhibitory compounds: inhibitory compounds from pretreatment and inhibitory compounds from lignin. Weak acids and furans degrade microorganisms by various mechanisms such as ATP depletion, cytoplasmic acidification and membrane damage; above a temperature of approximately 110 °C, lignin-derived aromatics such as furfural and 5-hydroxymethylfurfural degrade microorganisms (Sjulander & Kikas, 2020) [23]. However, this literature is based on bioethanol hydrolysates which is a complete contrast to SSF, where untreated residues are used which are not fully documented as to the extent to which these mechanisms are explored. Higher phenolics, protein and reducing sugars were solubilised from the Ohmic-heated brewer's grain, whereas no such effect was observed for the autoclaved grain and the phenolics production was higher which meant that these do not need to inhibit the production of xylanases and proteases for each other (Silva et al., 2025) [22], although the effect of solubilised phenolics was not separated.

3.5 Contamination, Variability and Storage

SSF is typically not completely sterilized (Yazid et al., 2017) [28], making native microbiota and storage spoilage potential issues, as with uncooled spent grain shows (Borkertas et al., 2025) [3]. However, there are costs associated with sterilization; steam or high pressure treatment can reduce protein content in oilseed meals, while in the case of non-sterile fermentation using local strains, it is avoided (Wei et al., 2026) [26]. No quantitative information on contamination levels or yields losses due to contamination was found in the retained sources. The variability of its feedstocks is also difficult to evaluate as composition tables combine different units (% dry basis, g kg−1, mg g−1, g per 100 g) that cannot directly be compared (Astudillo et al., 2023) [2].

4. Microbial Responses to Substrate Characteristics

4.1. Filamentous Fungi

The hybrid dominance in SSF is due to the filamentous fungi which penetrate the particles and have a more wide pH range, being pH 2–9 against 4–5 for yeasts, and a more or less wide range in the tolerance to low water activity, than the latter (Borkertas et al., 2025) [3]. Answers, however, are not uniform or consistent across genera and substrates. Aspergillus is the best studied genus for enzymes, and is the source of enzymes for most of the industrial cellulase platforms, except the industrial strain Trichoderma reesei which produces relatively little β-glucosidase, which can lead to the accumulation of cellobiose; Penicillium is a notable source of enzymes with high levels of β-glucosidase (de Moura et al., 2026) [9]. Mycelium yield estimated by qPCR varied greatly among related fungi: T. reesei RUT-C30 yielded 689.80 ± 80.53 mg g−1 on a digestate–food waste– sawdust mixture while T. harzianum T-22 yielded 174.26 ± 2.87 mg g−1 (Alias et al., 2022) [1].

4.2. Induction, Repression and Nutrient Accessibility

The role of available sugars is unclear. The effects of changing sugar availability are shown by experiments which induce or repress the cellulase genes in Trichoderma by changing conditions (de Moura et al., 2026) [9]. Among the results reported here, the enzymatic hydrolysis of an olive– sunflower cake mixture increased by three times the concentration of reducing sugars and nearly tripled the concentration of yeast cells, while decreasing the lipase activity by 11 times compared with the untreated control, associated with increased in sugar and protease ( Costa et al., 2023) [7]. In the case of Aspergillus oryzae 2095, glucose was added to the fish flour media, with 5% favoring the production of neutral and alkaline proteases, while 10% and 15% inhibited the production of these enzymes, likely because of preferential use of a more easily assimilated carbon source (Londoño-Hernández et al., 2024) [13]. The conditions of product formation and growth can therefore not be related: a condition that optimizes growth may not optimize a product, so repressible products can be suppressed.

4.3. Yeasts and Bacteria

Low moisture content is less tolerable to bacteria, and for bacteria most of the reported SSF requires moisture content between 60–85% which is higher compared to 40–60% required for fungi (Yazid et al., 2017) [28]. Yarrowia lipolytica W29 had 99 ± 1 U g−1 in monoculture SSF using an olive– sunflower cake mixture which did not increase when grown co-cultured with Aspergillus niger ( Costa et al., 2023) [7]. Bacillus subtilis is classified as a Generally Recognized Safe (GRAS) microorganism, as it secretes proteases, amylases and lipases. Although rhamnolipids have been produced by P. aeruginosa on soybean meal (Dabaghi et al., 2023) [8], bacterial surfactant producers are thought to be less safe to use in food products due to pathogenicity concerns, while yeasts like Y. lipolytica are thought to be safe (Vučurović et al., 2024) [25].

4.4. Compatibility, Inhibition and Contamination

It is possible to have safety but not compositional compatibility.Safety does not imply compositional compatibility. The aflatoxigenic (Aflavus, Parasiticus) and ochratoxigenic (Ochraceus, Carbonarius) species of Aspergillus can contaminate feed and food chain (Pócsi et al., 2020) [18], and in the case of A. niger, there is a relatively low risk for mycotoxins and they can be reduced by selecting strains and managing fermentation conditions (Wei et al., 2026) [26]. These statements allow screening at the strain level, rather than blanket assurance. Overall, the retained evidence shows no organism is optimal in all substrates or all products, while matching between inducer availability and water tolerance, nutrient needs and formation mechanism of the product is possible.

5. Substrate–Microbe Relationships in Bioproduct

Formation Table 2 collects outcomes from the retained studies with units as published. Values are not ranked, because strains, substrates, moisture, assays and reporting bases differ and normalizing them would need data the studies do not give. Protease is reported per mL of extract by Londoño-Hernández et al. (2024) [13], Casciatori et al. (2024) [5] do not state the basis of their U g−1 values, and a review has set pectinase in U g−1 for SSF beside U mL−1 for static liquid culture (Sharma et al., 2022) [21].

Table 2: Literature-based substrate–microorganism–bioproduct examples, with units as published. Values are not ranked or normalized. †Values reported in a review and not checked against the original studies. CMCase, carboxymethyl cellulase; U, enzyme unit as defined in each source.

Substrate Microorganism Product Conditions Reported outcome Source Brewer’s spent grain, ohmic-heated (145 °C, 20 min)

Silva et al., Olive cake + sunflower cake (50:50, dry basis) Y. lipolytica W29 Lipase 75% moisture (wet basis); 27 °C; 2 d 99 ± 1 U g−1 (monoculture); 49 ± 7 and 8 ± 2 U g−1 in simultaneous and sequential co-culture with A. niger Costa et al., Bagasse + straw + wheat bran (1:1:1) M. heterothallica Xylanase 75% moisture; 45 °C; 4 d; 5 g flasks 495.2 U g−1 dry substrate Martins et al., 2026 Bagasse fibers + wheat bran (7:3, w/w), 0.1 g cm−3 M. thermophila Endoglucanase; xylanase Packed bed; temperature rise <1 °C >850 and >3,850 U g−1 (basis not stated)

Casciatori Wheat bran + wheat straw; wheat bran + rice bran (1:1)

5.1. Enzymes

The best documented products of the hydrolysis are the ones produced on a hemicellulose-rich substrate (xylanase; Casciatori et al., 2024; Martins et al., 2026) [5, 15] and on a protein- or lipid-containing substrate (protease and lipase; Costa et al., 2023; Silva et al., 2025) [7, 22]. The substrate properties exert a strong influence through the amount and activity of the inducers supplemented (Section 4.2), by the physical characteristics of the substrate (limitation of heat and oxygen, Casciatori et al., 2024). It's important to note that formation is time dependent; protease activity was highest at 44 h decreasing to 72 h (Londoño-Hernández et al., 2024) [13] so comparing with a single end-point could be misleading.

5.2. Organic Acids

The sugar-availability-conditioning thing becomes apparent in citric acid. In a study where A. niger SSF was carried out using agricultural co-products, apple pomace (with rice husk) gave almost 350 g kg−1 whilst those of oranges-juice-processing waste gave nearly 450 g kg−1, and the cereal co-products gave the lowest yields, as described by West (2023) [27]. The result was very different in the case of one strain, 0.03 g g−1 starch on potato waste and 0.79 g g−1 starch on taro. The reported fold changes from the treatments are relative to one of several different controls, strains and scales and should not be pooled, and yields per kg of substrate moisture basis not stated indicate fold change in magnitude, not benchmark performance (West, 2023) [27].

5.3. Biosurfactants

Biosurfactants connect substrate to product via carbon and lipid supply, and are moisture and aeration sensitive. The main factors for the soybean meal were temperature, inoculum size and humidity with P. aeruginosa reaching 19.68 g kg−1 of dry substrate after 10 days at 34.5 °C and 100% of initial humidity with the addition of glycerol in the response surface model (Dabaghi et al., 2023) [8]. Optimal moistures were 70% rhamnolipids and 82% surfactin, and in the intermittent agitation experiment, the yield of sophorolipids increased by 31% (elsewhere yields were presented in g g−1 dry matter, mg g−1 dry substrate, and/or g kg−1 wet matter, so cannot be directly compared; Bustos-Vázquez et al., 2025) [4]. “Recovery matters – as much as formation” – in a techno-economic screening of the production of sophorolipid from oil cake, recovery was the biggest contributor to the operating costs (Martínez et al., 2022) [14].

5.4. Biomass and Protein

In the case of feed or food biomass, the product is the organism itself and substrate effects are manifested as growth, protein quality and protein safety. The crude protein gains come in between about 2.38% with tea residue mixed with wheat bran to 132% with wheat straw and Inonotus obliquus; the residues of the anti-nutrients in the grain remain uncountable, and the agreed safe levels are not available, while another risk lies with the re-synthesis of mycotoxins during storage and accumulation of biogenic amines (He & Cui, 2025) [10].

5.5. Other Metabolites and Product Recovery

Pigments and Phenolic compounds are reported in SSF products (Soccol et al., 2017) [24]. There was not enough quantitative evidence to draw a cross-substrate conclusion for these classes. In one reference case cited by Yazid et al., 2017, up to 74% of the enzyme recovery was achieved after the extraction with ultrafiltration. The amount of water used per gram of substrate for extraction varies, for example 5 g water/80 mL in case of xylanase (Martins et al., 2026) [15] and 0.5% NaCl, with 0.5% Tween 80 for lipase (Sharma et al., 2022) [21], indicating that the reported activities are partially due to extraction efficiency.

6. Pretreatment, Process Optimization and Scale-Up

Figure 2 summarizes the SSF process chain used in this review, from feedstock preparation to residue management, and the four variables (moisture, oxygen supply, temperature and bed structure) through which operating conditions interact with substrate properties.

Fig 2: Solid-state fermentation process (steps 1–8) and intervention points during fermentation: M, moisture; O, oxygen supply; T, temperature; B, bed structure. The packed-bed drawing is schematic, with forced aeration and a water jacket; the shading indicates that bed temperature can rise along the airflow path when heat removal is inadequate (Mitchell et al., 2023; Sentís-Moré et al., 2023). O2, oxygen; CO2, carbon dioxide.

6.1. Conditioning: Pretreatment, Blending, Supplementation and Moisture Adjustment Pretreatment affects particle structure, free sugars, phenolics, and accessibility which in turn affects product formation, depending on what the product requires. Ohmic processing of brewer's spent grain at 145 °C for 20 min released the highest phenolics, protein and reducing sugars with A. niger CECT 2088 (540 U g−1 xylanase activity and 49 U g−1 protease activity) compared to autoclaved spent grain (Silva et al., 2025, 121 °C for 15 min) [22]. In a yeast system, the ultrasound treatment had a significant effect on the lipase only in semi-solid system and the enzymatic hydrolysis had significant reduction of lipase (Costa et al., 2023) [7]. The acid-treated materials resulted in 2- to 2.3-fold greater yield of citric acid than the untreated materials (West, 2023) [27]. This can consequently benefit, cause no benefit, or cause harm depending on the organism, product, system, and general rules regarding pretreatment.

Few SSF studies report the costs of pretreatment. Mechanical treatment is energy consuming and can generate furfural and hydroxymethylfurfural (Kucharska et al., 2018) [12] which represent up to 40% of total costs in biomass processing.

Detoxification steps added: Overliming reduce furans, but may lose up to 70% of sugars at pH 12 (Sjulander & Kikas, 2020) [23]. Comparisons of this sort are, therefore, essential to the interpretation of the results, and chemical inputs, energy values and water and waste streams should be recorded in addition to yield.

Levers that are cheaper to manipulate are blending, supplementation and moisture control. The two residue mixtures (bagasse–straw and wheat bran–straw) were significantly superior to the individual residues in respect of xylanase (Martins et al., 2026) [15], and temperature control (Casciatori et al., 2024) [5] in the low-density packed bed of bagasse–wheat bran. There are also ways of moisture management that can be active: Superabsorbent polymers (SAP) and negative-pressure irrigation (NPI) both increased cellulase activity by 2.47-fold. When combined together, cellulase activity increased to 2.80-fold, and with increased moisture, temperature gradients lower than 0.4 °C cm−1 produced better heat transfer, but no improvement for sporulation (He et al., 2019) [11].

6.2. Temperature, Aeration, Inoculum and Time

Bed structure interacts with temperature, aerobic conditions, inoculum, pH and time. In forced aeration, oxygen transfer and the removal of metabolic heat is of the main concern and thus, superficial air velocity is basically central. Good temperature control was observed in many packed-bed studies with velocities below 1 cm s−1 and even in some, velocities of 6 and 10 cm s−1 maintained the bed temperature within 1 °C of the inlet air temperature (Mitchell et al., 2023) [16]. The inoculum should be reported per gram of dry substrate and the identity of the strain since, the use of an inoculum (size, temperature and humidity) are important factors for rhamnolipid production (Dabaghi et al., 2023) [8].

6.3. Scale-Up

In laboratory SSF, a usual amount of substrate in a flask is between 5–10 g of substrate, and heat and gas exchange are the factors rarely constrained.In a laboratory SSF, usually the amount of substrate in flasks ranges from 5–10 g of substrate, with heat and gas exchange being the factors not limiting. In large solid beds, this is the opposite of the rule: it is difficult to maintain uniform temperature, gas exchange and moisture, and localized overheating, anoxia and desiccation result (Borkertas et al., 2025) [3]. Mitchell et al. (2023) [16] predict that for packed beds the height shall be limited to approximately 1 m and diameter be increased for the same bed with superficial velocities up to 10 cm s−1 for reasonable temperature control. Compared to columns and packed beds, which could be operated with forced aeration, tray reactors are always operated by natural diffusion (Bustos-Vázquez et al., 2025) [4].

Little evidence is there above the bench. Studies on biosurfactant production in reactors with sizes of 100-10,000 L are few (Astudillo et al., 2023) [2] and the reactors with 22 L (3 kg wet matter) and 100 L (10.5 kg wet matter) were a biomass reactor and a biosurfactant reactor, respectively (Bustos-Vázquez et al., 2025) [4]. The try techno-economic screening of the sophorolipid production assumed a 15% production loss with scale-up and identified no existing commercial packed-bed bioreactor with the assumed capacity of 5 m3 (Martínez et al., 2022) [14]. Current monitoring consists mainly of temperature measures, proxies for off-gas and moisture, while low cost sensor arrays have been able to map tray gradients (Sentís-Moré et al., 2023) [20] and biomass is difficult to measure in solids (Yazid et al., 2017) [28].

6.4. Safety, Economics and Sustainability

Safety is dependent on both the organism and the substrate and application. In addition to the Aspergillus toxins (Section 4.4), there are no consensus safe post-fermentation levels for anti-nutrients as gossypol (Zhou et al., 2010), and the risk of re-synthesis of mycotoxins in fermented food stored for an extended period of time is a pathogenicity issue (He & Cui, 2025) [10]; some bacterial producers are not safe to use in food (Vučurović et al., 2024) [25].

The benefits of the economy and the environment are only suggested. The cellulosic enzyme unit cost for SSF and submerged fermentation was USD 15.67 kg−1 and USD 40.36 kg−1, respectively, with SSF economic studies being nonexistent (Yazid et al., 2017) [28]. A screening of the sophorolipid production costs estimated USD 5.1, 5.7 and 6.9 kg−1 for production on wheat straw, rice husk and coconut fiber supports, respectively, but it is only a preliminary estimation whose Monte Carlo net present value ranged from −2.1 to 10.9 million USD and that did not consider the transport cost and several other costs (Martínez et al., 2022) [14]. Thus, the use of a waste does not imply the sustainability of a process: the benefits of pretreatment, drying, transport, sterilization, extraction and waste treatment are not measurable and so any benefits should be limited to what has been measured.

7. Integrated Substrate Selection Framework and

7.1 Selection Framework

The evidence suggests a progression of questions, rather than a prediction formula (Figure 3). The moisture retention capability is the most important parameter for lipase and biosurfactants (Costa et al., 2023; Dabaghi et al., 2023) [7, 8], while the hemicellulose content is relevant for xylanase ( Casciatori et al., 2024; Martins et al., 2026) [5, 15] and the availability of sugar for citric acid (West, 2023) [27]. The organism class is then correlated by water tolerance, enzyme repertoire and documented safety (Section 4) and the substrate is characterized on dry-matter basis, such as polymers, free sugars, nitrogen, inhibitors, moisture, aw, and particle size, or porosity. When the match is not good enough, conditioning is assessed based on energy and waste streams, chemical use, and the characteristics of the conditioned substrate is recharacterized, because the more accessible the sugar, the less of this product that is formed ( Costa et al., 2023) [7]. The process constraints, safety and quality, and scale-up feasibility are then discussed, one by one. There is not uniform support for such measures. Its strongest strength lies within its performance as a bed structure and heat removal in packed beds (Casciatori et al., 2024; Mitchell et al., 2023; Sentís-Moré et al., 2023) [5, 16, 20] and is reasonably consistent for the product-specific effects of accessible sugar ( Costa et al., 2023; de Moura et al., 2026; Londoño-Hernández et al., 2024) [7, 9, 13]. It is moderate for matching product class to substrate (where studies are conducted in incompatible units); and lowest for inhibitors (SSF), contamination, safety, pilot-scale performance and economics.

Fig 3: Evidence-informed framework for selecting and conditioning agro-industrial residues for solid-state fermentation. Numbered boxes are steps, diamonds are decisions, and dashed arrows are return loops. The scheme is a qualitative synthesis of the studies discussed in Sections 3–6; it has not been validated experimentally and is not a predictive model.

7.2. Research Gaps and Priorities

Table 3 converts the key gaps into specific enhancements and four key priorities are measurable from current approaches. First, the effect of increasing the severity of pretreatment should be explored in a graded manner across one substrate– strain pair to see if there is an optimum point for sugar availability for product formation as suggested by Costa et al. (2023) [7] and Londoño-Hernández et al. (2024) [13]. Second, the moisture content and aw should be varied separately and the sorption isotherms obtained before and after the fermentation process (Corona et al., 2005) [6]. Third, packed-bed trials should be carried out with a fixed packing blend and the bed height, packing density, air velocity and temperature profiles should be reported in order to be able to separate the porosity effects from the composition effects (Casciatori et al., 2024) [5]. Fourth, the strain identity, the dry-substrate basis, the definition of the assay and the ratio of the extract need to be reported for any study to allow for the possibility of combining results in forthcoming quantitative syntheses.

Table 3: Critical research gaps and practical research priorities. aw, water activity; TEA, techno-economic analysis; LCA, life-cycle assessment.

Gap Consequence Proposed improvement Evidence basis Incomplete substrate characterization Differences in output cannot be attributed to composition, structure or water Report dry matter, cellulose, hemicellulose, lignin, sugars, starch, protein, lipid, ash, moisture (stated basis), aw, particle size, bulk density and porosity Astudillo et al., 2023; Nargotra et al., 2023; Yazid Nonstandard units and reporting bases Yields cannot be compared across studies Report U g−1 dry substrate with assay definition and extraction ratio; state the basis of g kg−1 values Table 2; Bustos-Vázquez et al., 2025; Sharma et al., 2022 Moisture reported without aw Moisture optima do not transfer between substrates Measure aw and sorption isotherms before and after fermentation; vary moisture and aw separately Corona et al., 2005; He et al., Inadequate strain identification and safety assessment Toxin risk; poor reproducibility Report species, strain and collection number; screen products for relevant toxins Pócsi et al., 2020; Wei et al., 2026; He & Cui, 2025 Pretreatment judged without full accounting Benefits contradictory; costs, inhibitors and waste omitted Compare with an untreated control on the same dry basis; record sugars, phenolics, chemicals, energy and waste Costa et al., 2023; Silva et al., 2025; Kucharska et al., Limited pilot-scale and heat and mass transfer data Scale-up uncertain; flask results may not transfer Report bed height, packing density, air velocity, temperature and moisture profiles Mitchell et al., 2023; Astudillo et al., 2023; Bustos-Vázquez et al., 2025 Missing recovery and techno-economic data Cost and sustainability unquantified Report extraction efficiency; add TEA with sensitivity analysis and LCA Martínez et al., 2022; Yazid et al., 2017; He & Cui, 2025

8. Conclusion

In the retained evidence, composition determines what products are possible, whereas physical structure, water relations and sugar accessibility will dictate if a fermentation will reach them. No organism or pretreatment gave an absolute best performance under all substrates and products and the same pretreatment appeared to increase, keep the same level or decrease the output in different systems. A variety of factors have constrained those conclusions: characterization of the substrates was inconsistent; the units and moisture bases were inconsistent; the strains were not fully identified; there was limited pilot-scale and safety data, and there was not an exhaustive search. The selection framework is designed to guide the next experiments and not a validated predictor. The best way to get the field from case reports into rules that could be transferred is through standardized reporting, and pilot size studies to measure recovery, safety and costs.

Conflict of Interest

The authors declare that no conflict of interest exists.

Funding

This work did not receive any external funding.

Ethical Approval Ethical approval was not required for this review, as it did not involve human participants, animals, or the collection of primary data.

Data Availability No new datasets were generated or analyzed in this study. All sources consulted are cited in the References section.

Author Contributions Hassan Bashir Mukhtar was responsible for conceptualizing the review, designing the study, and writing the Introduction and oversaw the manuscript design. Musab Muhammad Bello, Abubakar Yau Ibrahim, Auwal Abdullahi Zubairu, and Safiyya Ibrahim BB-Farouk contributed to the remaining sections of the manuscript. All authors reviewed the final version and gave their approval.