REVISED DRIVERS OF VARIABILITY AND HAZARDS AND SAFETY CONSTRAINTS

Drivers of variability

The largest driver of variability in cassava pulp/pomace is the processing plant’s starch recovery efficiency. Even modest shifts in screening, rasping, washing intensity, and separator performance can redistribute carbohydrate between the starch product and the residue, producing large swings in residual starch versus structural carbohydrate (and therefore metabolizable energy/NEL potential and rumen fermentation kinetics) [30]. Dilution and wash water carryover further contribute to inconsistency by altering as-fed DM and concentrating or diluting soluble fractions, which in turn affect the delivered nutrient density and the actual inclusion rate in a TMR on an as-fed basis, an error mode that is disproportionately important for wet streams [83–87]. Soil contamination and inadequate cleaning/peeling increase ash (particularly insoluble ash), depressing energy density and potentially introducing undesirable inorganic load; this is especially relevant for residues handled on the ground or dried under conditions that promote contamination.

Feed resource / stream | Processing condition | Key composition data (DM basis unless stated) | Mean / range / CV* | Dairy relevance
Cassava pulp | Wet pulp before fermentation | DM 165.5 g/kg as-fed; CP 25.94 g/kg DM; NDF 438.87 g/kg DM; ADF 246.79 g/kg DM | Single-source baseline; CV not reported | Very wet material with high fiber and low CP; useful baseline for interpreting spoilage risk, dilution of nutrient density, and the need for rapid stabilization [24]
Cassava pulp silage | 21-d fermentation with L. casei TH14, urea, and/or molasses (8 treatment combinations) | DM 166.10–177.62 g/kg as-fed; CP 21.34–154.69 g/kg DM; NDF 334.71–394.37 g/kg DM; ADF 198.84–222.94 g/kg DM | Mean DM 171.65 g/kg, CV 2.7%; mean CP 80.66 g/kg DM, CV 77.8%; mean NDF 351.35 g/kg DM, CV 5.5% | Shows that fermentation additives can dramatically change apparent CP and modestly shift fiber fractions even within the same cassava pulp stream; supports using equivalent crude protein / NPN language and not relying on starch or CP alone to predict NEL. Aerobic stability was >120 h after silo opening across additive treatments [24]
Cassava peel | Fresh peel vs hot-water control peel vs enzyme-treated peel | Fresh peel NDF 61.01% DM, ADF 57.00% DM, CF 35.01% DM, cyanide 308.35 μg/g; control peel NDF 35.13% DM, ADF 31.55% DM, cyanide 281.75 μg/g; enzyme-treated peel NDF 33.16–33.18% DM, ADF 24.01–28.91% DM, cyanide 126.61–142.17 μg/g | Clear processing-driven range rather than pooled mean/CV | Strong example of how processing changes both safety and fiber burden; peels are not nutritionally interchangeable with pulp and should be treated as a more fibrous, more cyanide-sensitive stream [54]
REAM (residue from extraction of cassava starch) | Included in diets of multiparous Holstein cows | Inclusion levels 0, 8, 16, 24, and 32% of diet DM | Performance range rather than composition range | At 32% of diet DM, milk yield was reduced by about 15%; authors concluded that up to 16% of diet DM could be included, whereas higher levels reduced productivity. This is a useful recent anchor linking residue variability and ration fit to biologically meaningful dairy outcomes [22]

Upstream agricultural variation also matters: cultivar, agronomic conditions, and season influence root composition (starch, fiber, and cyanogenic potential), and these differences propagate into waste streams, particularly when factories draw from mixed-cultivar pools across seasons [88–92]. Finally, post-processing choices (draining time, sun-drying vs. mechanical drying, pelletizing, ensiling inoculants, urea/molasses use) can shift both apparent nutrient composition and functional value; for example, urea addition predictably elevates analyzed CP in cassava pulp silage largely as non-protein nitrogen (NPN), creating a product that is analytically higher protein but biologically still distinct from true-protein feeds [24].

**How variability translates to ration uncertainty (fermentability, effective ** fiber , milk fat risk)

For dairy cows, the practical consequence of compositional spread is ration uncertainty in three linked domains: (i) ruminal fermentability (rate and extent of starch/fiber digestion), (ii) physically effective fiber (chewing stimulation and rumen pH stability), and (iii) downstream risks of SARA and milk fat depression (MFD). When cassava pulp/pomace arrives with higher-than-expected starch and lower effective fiber contribution, it can behave more like a rapidly fermentable concentrate, increasing acid load and compromising rumen pH unless physically effective NDF (peNDF) and buffering capacity are protected at the ration level [93]. This matters because the milk fat response is sensitive not only to how much starch is available but also to how fast and where it ferments, interacting with the unsaturated lipid supply and rumen biohydrogenation pathways that produce MFD-associated fatty acid intermediates [94]. Conversely, when NDF in cassava pulp/residue is elevated (or particle characteristics change with drying/pelleting), the ingredient may contribute more to fiber supply but not necessarily to physically effective fiber unless particle size distribution and fragility are known, an issue increasingly emphasized in modern dairy feed characterization frameworks [95]. Recent in vitro work with ensiled cassava pulp further illustrates that processing and additive choices can alter digestibility and fermentation traits, meaning that cassava pulp silage is not one feed but a family of feeds whose risk profile depends on how it was produced and how consistently it is described analytically [35]. The contribution of cassava pulp to physically effective fiber should be interpreted cautiously, because NDF concentration alone does not ensure adequate peNDF if particles are small, fragile, or further reduced during TMR mixing. More broadly, dairy evidence shows that peNDF influences chewing activity, rumen fermentation, plasma metabolites, and milk production in high-concentrate diets, supporting the need to evaluate cassava residues in the context of whole-ration particle size and fermentability rather than as isolated chemical ingredients. In practical terms, variability in starch accessibility, particle characteristics, and contamination burden should be expected to affect not only estimated energy value but also rumen pH stability, volatile fatty acid pattern, chewing response, and ultimately milk component responses when cassava residues are incorporated into lactating cow diets. Recent dairy cow evidence also suggests that these uncertainties are biologically meaningful: residue from cassava starch extraction included at 0, 8, 16, 24, and 32% of diet DM reduced milk yield at the highest inclusion level, linking compositional and physical variability more directly to performance risk [22, 23].

Hazards and safety constraints

The safe use of cassava processing residues in dairy systems depends on recognizing that hazards arise from both intrinsic plant chemistry (cyanogenic glycosides), and extrinsic contamination and preservation failures (fungal toxins, pathogenic microorganisms, inorganic contaminants, and chemical residues). Because high-producing lactating cows are routinely fed high-concentrate diets and large daily DM intakes, even moderate deviations in hazard load or preservation quality can translate into clinically relevant outcomes, reduced milk performance, or milk safety concerns. A hazard-focused synthesis is therefore essential to move cassava residues from opportunistic by-products to reliable circular feed ingredients governed by specification-based procurement and process control.

CYANOGENIC GLYCOSIDES AND HCN: A RESIDUE-SPECIFIC TOXICOLOGICAL CONSTRAINT

Cassava tissues contain cyanogenic glycosides (primarily linamarin and lotaustralin) that can be converted to HCN when plant cells are disrupted and enzymatic hydrolysis proceeds, processes that are intensified by chopping, grinding, or maceration and by moist conditions [54]. From a dairy feeding perspective, the risk is stream-dependent: peels- and foliage-derived streams are typically a greater concern than many starch extraction pulps, and the risk increases when fresh materials are fed soon after processing (i.e., before detoxification via drying or fermentation) [54]. In ruminants, low-level cyanide exposure can be detoxified primarily by conversion to thiocyanate [96–100], but rapid absorption of large amounts can overwhelm detoxification capacity and lead to acute poisoning; rumen conditions (including pH) influence the rate and extent of cyanide release [101]. These mechanistic considerations matter operationally because they imply that safe inclusion cannot be defined by a single percentage of diet DM; it must account for stream identity, processing state, and the rate at which cyanide can be released and absorbed.

Recent research has progressed from descriptive hazard recognition to practical mitigation strategies directly relevant to dairy supply chains. Ensiling is particularly important because it can simultaneously address preservation and detoxification, and recent studies demonstrate that inoculation strategies can alter residual cyanide outcomes. A study evaluated cassava root silage inoculated with cyanide-utilizing bacteria isolated from the bovine rumen and observed measurable reductions in total cyanide, along with changes in fermentation quality and in vitro rumen fermentation indices [15]. Complementary evidence indicates that supplementing cattle receiving fresh cassava root with cyanide-utilizing bacteria and sulfur can improve cyanide degradation and modify rumen fermentation and microbiome profiles [102], supporting a plausible biological basis for integrating targeted microbial and nutritional interventions into cassava-feeding programs. More broadly, a recent applied review on cassava byproduct valorization emphasizes that detoxification is a prerequisite for wider utilization of peels and leaves, reinforcing the need for consistent hazard control rather than ad hoc feeding decisions [54].

From a dairy feeding perspective, HCN risk is stream- and process-dependent: peels- and foliage-derived streams are generally of greater concern than many starch extraction pulps, and risk increases when fresh materials are fed before being dried or fermented. Although high-producing dairy cow-specific residual HCN thresholds remain insufficiently standardized, published lactation studies suggest that dietary exposures of 35–70 ppm HCN from fresh cassava pulp and approximately 75 ppm HCN from fresh cassava peel can be tolerated under the reported study conditions without adverse effects on milk yield or milk composition. These values, however, should not be interpreted as universal safety limits for high-producing herds, because cyanide risk depends on stream identity, intake rate, sulfur status, rumen conditions, and the speed of cyanide release and absorption. Recent ensiling work also shows that detoxification efficacy remains incomplete and process-sensitive: after 30 days of ensiling, cyanide removal from cassava root silage was about 39% in untreated silage and improved to approximately 47–51% with cyanide-utilizing inoculants and/or cellulase, indicating that longer-term efficacy and the interaction with rumen microbial adaptation remain important research gaps rather than settled control measures. Low-level cyanide exposure is detoxified primarily to thiocyanate, and dairy studies show that milk thiocyanate can increase when fresh cassava materials are fed; for example, milk thiocyanate concentrations of 1.92–14.58 mg/dL have been reported in cows fed fresh cassava root. Accordingly, HCN should be managed as a process-controlled hazard, with preference for documented detoxification routes, conservative use of fresh high-cyanogen streams in high-producing cows, and residual cyanide testing where feasible [15, 103–105].

MYCOTOXINS AND FUNGAL SPOILAGE: HAZARDS AMPLIFIED BY WET RESIDUES AND POOR DRYING/ENSILING

Cassava residues, particularly when handled as wet pulps or inadequately dried products, can be vulnerable to fungal growth and mycotoxin contamination, especially in humid tropical conditions and in supply chains with prolonged storage, intermittent drying, or re-wetting [51, 52, 106–108]. Cassava-specific survey data indicate that this is not merely a theoretical concern. In a recent value-chain study in Uganda, 192 cassava product samples (flour and chips) were screened for multiple mycotoxins, and the detected profile included aflatoxins, fumonisins, ochratoxin A, deoxynivalenol, zearalenone, and citrinin; all positive samples exceeded the EU threshold of 5 μg/kg for Aflatoxin B1 (AFB1). These findings support the view that cassava matrixes may harbor multi-mycotoxin contamination, not just isolated aflatoxin events, and therefore require a broader surveillance approach than single-toxin screening alone. This is especially relevant for dairy systems because aflatoxin B1 in feed can be converted to aflatoxin M1 (AFM1) in milk. Recent review evidence indicates that AFB1 carryover into milk is typically about 1–2% on average but can reach approximately 6% in high-yielding cows, meaning that cows may show no obvious clinical signs while milk still exceeds regulatory limits. More broadly, a recent meta-analysis of bovine feeds indicates that aflatoxins are especially prevalent in warmer, drier climates and that climate change may shift current and emerging mycotoxin risks, which strengthens the case for routine surveillance when cassava residues are introduced into lactation diets [109–111]. A major step toward standardizing prevention is the Codex Code of Practice (CXC 82-2023) for the prevention and reduction of mycotoxins in cassava and cassava-based products, which explicitly recognizes that toxigenic fungi are associated with soil/dust, crop residues, and post-harvest storage/processing environments, and that risk profiles vary by region and handling practices [112]. Although this Codex guidance is written primarily for food chains, the same hazard logic applies to feed chains, as cassava residues often share upstream storage and processing nodes.

Recent empirical work underscores the relevance of multi-mycotoxin contamination along cassava value chains. A study reported the presence of multiple regulated and emerging mycotoxins (including aflatoxins, fumonisins, ochratoxin A, deoxynivalenol, zearalenone, citrinin) in cassava products sampled across a national value-chain, illustrating that cassava matrixes can exhibit complex co-contamination patterns rather than single-toxin profiles [109]. From the dairy nutrition angle, this matters because lactating cows are sensitive to chronic mycotoxin exposure through impacts on intake, immune function, reproduction, and milk production, and because aflatoxin B1 in feed is a well-recognized precursor to AFM1 in milk (risk magnitude depends on exposure and carryover factors) [110, 113, 114]. A recent environmental health synthesis on mycotoxins in bovine feed highlights that mycotoxin exposure remains widespread across feed components and that climate variability may shift prevalence, underscoring the need for routine surveillance and risk-based testing strategies [111]. Climate variability and warming trends should also be considered, as recent global evidence indicates that aflatoxin risk is favored by warmer, drier conditions, implying that cassava-based feed chains in tropical regions may face increasing contamination pressure unless drying, storage, and surveillance are improved [111]. Practical guidance for livestock producers emphasizes that risk depends on toxin concentration, intake rate, animal class, and exposure duration, conditions that are directly relevant when introducing variable by-products into high intake lactation diets [115].

Furthermore, fresh or insufficiently stabilized cassava pulp should be treated as highly vulnerable to aerobic deterioration. Once exposed to air, yeasts initiate oxidation of residual sugars and fermentation acids, which manifests as heating, rising pH, and loss of palatability before visible mold becomes apparent. In practical terms, heating of the pulp pile should be interpreted as an early warning sign of yeast-driven deterioration rather than a benign temperature fluctuation, because pH elevation and acid depletion create conditions that favor broader spoilage microbiota and, over time, mold proliferation and hygiene failure. For field monitoring, aerobic stability is commonly defined as the time until silage temperature rises 2°C above ambient. In cassava pulp silage studies, additive-treated silages have been reported to maintain>120 h of aerobic stability after opening, whereas cassava residue silages with pH below 4.0 are generally described as well preserved. These values are not universal guarantees, but they provide practical benchmarks for quality control in cooperatives and smallholder systems [24, 116, 117].