The composition of tropical grass silages also showed substantial variability, indicating the broad range of conditions represented in the data set. For example, DM content ranged from extremely low (168 g/kg) to high (466 g/kg), while IVDMD (314 to 709 g/kg DM) and n-butyric acid concentration from 0.6 to 31 g/kg DM, likely reflecting differences in agronomic and ensiling conditions among fields. The absence of correlation between n-butyric acid concentration and IVDMD indicated that fermentation pattern and nutritional value were not directly associated. These findings indicate that complementary strategies are needed to produce highly digestible, well-fermented silage.
4.1. Nutritional value of silages
In this study, silages with the highest digestibility were obtained from production systems that adopted greater doses of nitrogen fertilization, greater cutting frequency, and lower regrowth age at harvest, which together may contribute to greater annual productivity. More frequent cutting usually allows forage production with a higher leaf:stem ratio and, consequently, better nutritional value at harvest. This intensive management, when soil nutrient availability is adequate, promotes faster growth and efficient plant recovery, maximizing the biomass produced. In addition, this practice helps to maintain plants in the vegetative phase, with greater nutritional quality (Mari, 2003). On the other hand, the application of greater doses of N allows faster plant growth while maintaining desirable leaf:stem ratio (Pinto et al., 1994; Oliveira et al., 2020).
An intriguing finding was the dichotomy in fertilization practices. While a higher proportion of farmers in the low in LD applied fertilizer, the actual nitrogen dose was significantly lower than that in HD group. This may suggest that in the LD group, fertilization is done as a standard practice but at suboptimal rates, which might be insufficient to maintain high proportion of leaves and delay stem elongation. Lower application rates of fertilizers combined with a longer regrowth interval observed in LD likely resulted in poorer forage quality. This underscores that the dose of fertilizer is critical for affecting grass productivity and nutritive value.
As the plant matures, the proportion of highly digestible components (i.e., NDS) tends to decrease. On the other hand, the proportions of fibrous fractions increase, resulting in lower digestibility (Moser, 1995; Jung and Allen, 1995). This pattern was evidenced in the present study by the lower concentrations of NDS and higher concentrations of fibrous fractions in the group with lower IVDMD, as well as by the negative correlation between iNDF and IVDMD. Therefore, forage harvested at a more advanced stage of maturity was associated with higher iNDF and lower IVDMD of the silages, as cell wall lignification hinders fiber digestibility by rumen microorganisms (Van Soest, 1994). In this survey, the average grass maturity at harvest was nearly 100 days, consistent with the high iNDF (318 g/kg DM) and low IVDMD (501 g/kg DM) observed. A high proportion of concentrate was reported even in diets for breeding cows and growing cattle, likely to compensate for the low feeding value of the silage.
In grasses from Urochloa and Megathyrsus genus, the IVDMD decreases with maturity, with a more pronounced decline after approximately two months of regrowth (Mari, 2003; Tomaz et al., 2018). The high iNDF:lignin(sa) ratio (averaging 4.96) corroborates that lignification has a greater negative effect on fiber digestibility in tropical grasses than in corn silage and temperate grasses (Daniel et al., 2017; Raffrenato et al., 2019; Moraes et al., 2023). The strong negative correlation between iNDF and IVDMD confirms that, for tropical grass silages, the indigestible fiber fraction is a major driver of overall DM digestibility. Therefore, iNDF may be more informative than total NDF for benchmarking the nutritive value of tropical grass silages.
The association between the Megathyrsus genus and the low IVDMD group (LD) is noteworthy. While both genera are tropical, their morphological and agronomic traits differ. It is plausible that Megathyrsus grasses on these farms were managed more extensively (e.g., lower N fertilization, longer regrowth intervals), leading to more lignified fiber at harvest. This aligns with the lower N application and numerically greater maturity at harvest observed in LD, where Megathyrsus predominated. Conversely, the higher prevalence of Urochloa in HD suggests that, under the management practices adopted on these farms, it may have been harvested at a less mature stage, preserving its nutritive value. This highlights the importance of on-farm management in determining the expression of the genetic potential of these grasses. The HD had 3.00 cuts/yr compared with 1.85 cuts/yr in LD, while regrowth age was lower in HD (83.8 vs 112 d). Such differences reflect not only botanical differences but also potentially distinct forage production systems.
Crop-livestock integration areas accounted for 50% of HD but only 3% of LD, whereas semi-perennial systems predominated in LD (90%). These results suggest that integrated systems may substantially enhance soil fertility, nitrogen cycling, and overall crop productivity (Lemaire et al., 2014), facilitate harvesting at earlier phenological stages, thereby indirectly contributing to higher digestibility.
The concentration of several fermentation products, either considered individually (e.g., acetic acid) or as the sum of compounds (sumFP) were higher for the group with greater IVDMD, indicating a greater extent of fermentation, which must have occurred due to the greater availability of soluble substrates for microorganisms in the younger better-fertilized plants. Despite the difference in maturity, the DM content was similar between silage groups. Previous studies have shown that maturity advancement is not an exclusive factor determining the moisture content of tropical grasses (Wilkinson, 1983; Santos et al., 2014; Tomaz et al., 2018; Moraes et al., 2023).
Although no differences were observed for the use of moisture absorbent (27 and 28% in HD and LD, respectively), the dose of absorbent (126 and 169 g/kg in HD and LD, respectively), and the type of absorbent between digestibility groups, the results from this clustering based on digestibility should be interpreted with caution as the mixture of those ingredients with the grass can modify the silage digestibility.
4.2. Silage fermentation pattern
In this study, most tropical grass silage samples (86%) were considered poorly fermented, with n-butyric acid concentrations > 3 g/kg DM (Kaiser et al., 2002). The relationship between DM content and n-butyric acid concentration indicated that silages with DM contents below 370 g/kg DM were at a high risk of clostridial development, whereas silages with more than 370 g/kg DM had lower risk of butyric fermentation. Recently, Carvalho et al. (2024) confirmed that the FC model proposed by Weissbach et al. (1974) is applicable to tropical grasses, including the Megathyrsus and Urochloa genus. The authors observed that, on average, the minimum DM content necessary to reduce the risk of butyric fermentation was approximately 350 g/kg DM, which is very similar to the value found in the present study. Thus, it can be concluded that the minimum DM value of 250 g/kg mentioned by McDonald et al. (1991) or even the value of 300 g/kg commonly adopted in practice may not be sufficient to restrict butyric fermentation and minimize nutrient losses in tropical grass silages, especially those from the genera Megathyrsus and Urochloa. This finding, derived from an on-farm survey, provides field-based support for the controlled experimental data reported by Carvalho et al. (2024) and suggests a practical benchmark for farmers and consultants. For Megathyrsus and Urochloa grasses, wilting or blending moisture absorbents to achieve a DM concentration above 370 g/kg represents the single most critical factor for mitigating the risk of clostridial fermentation. As the availability of harvesting and forage conservation machinery increases, the production of wilted tropical grass silage is expected to become more widespread.
Although no significant differences in pH were observed between the fermentation quality groups (LB vs. HB) or between the digestibility groups (HD vs. LD), the pH value alone was not a reliable indicator of fermentation quality in this study. Silage pH is influenced by multiple factors, including the concentration of buffering compounds (either originating from the forage or formed during fermentation), the presence of other organic acids, the use of additives, and moisture level. Notably, silages with higher DM content tend to exhibit relatively high pH values even under good fermentation conditions (Muck, 1988). Therefore, relying on pH alone, without profiling fermentation products can be misleading.
The correlation between n-butyric acid and lactic acid concentrations was inverse. The segmented regression model suggests a threshold of approximately 7.5 g/kg DM of lactic acid. Below this level, butyric acid concentration increased sharply. When enough lactic acid is produced in the fermentation process, microbial activity is reduced, mainly due to the effect of the acidic environment, allowing the silage to be stored aerobically (McDonald et al., 1991; Pahlow et al., 2003; Rooke and Hatfield, 2003). On the other hand, the lactic acid produced at the beginning of fermentation may have been degraded by the activity of undesirable microorganisms in silages classified as poorly fermented, especially by bacteria of the genus Clostridium (McDonald et al., 1991; Pahlow et al., 2003; Rooke and Hatfield, 2003). Both mechanisms may have occurred and the real cause of the low lactic acid concentration in silages with high concentration of n-butyric acid cannot be established in this study. Nevertheless, these findings suggest that achieving and maintaining an adequate lactic acid concentration during fermentation may help inhibit clostridial activity. Management strategies, as the use of efficient silage additives, must therefore help promote adequate lactic acid production and favorable fermentation.
In addition to n-butyric acid, several fermentation products can be associated with the development of clostridia in silages, such as i-valeric acid, n-valeric acid, i-butyric acid, 2,3-butanediol, isopropanol, 2-butanol, ethanol, acetic acid, ammonia, and biogenic amines (McDonald et al., 1991; Rooke and Hatfield, 2003). Several of these compounds are also associated with DM losses and reduced feeding value. In this study, n-butyric acid content was positively correlated with propionic acid, NH3-N, and other fermentation products typically produced by clostridia. Furthermore, silages classified as HB had higher concentrations of NH3-N, acetic acid, propionic acid, i-butyric acid, n-valeric acid, i-valeric acid, and 2-butanol.
High concentrations of NH3-N, i-butyric acid, i-valeric acid, and n-valeric acid indicate severe proteolysis and deamination triggered by proteolytic microorganisms (McDonald et al., 1991). The observed differences in NH3-N concentration suggest marked degradation of true protein and a potential reduction in nitrogen-use efficiency in HB silages. Moreover, a higher NH3-N concentration (and likely other hypophagic metabolites) in grass silage has been reported to be inversely related to feed intake in cattle (Krizsan and Randby, 2007), further reducing silage feeding value. These nutritional and economic consequences warrant further investigation in future research. Although propionic acid does not contribute significantly to pH reduction high concentrations of this acid in moist silages typically indicates poor fermentation (Kung et al., 2018).
The development of undesirable microorganisms in tropical grass silages generally results in the degradation of readily available soluble nutrients and decreased digestibility (Gomes et al., 2021; Moraes et al., 2023). In fact, in the present study, poorly fermented silages had lower levels of SDN and higher levels of NDF, but the difference of IVDMD between LB and HB was only numerical (531 to 496 g/kg DM), as delayed harvest was associated with less n-butyric acid primarily through higher DM in the LB group. However, this does not imply that delayed harvest is an appropriate management strategy (Tomaz et al., 2018), because late harvest reduces nutritive value. Therefore, plant composition and cell wall digestibility at harvest likely had a stronger effect on silage IVDMD than did fermentation pattern, supporting the distinction between these two quality attributes in tropical grass silages.
In addition to plant composition, silage management can also influence the conservation process. In our dataset, the group of well-preserved silages showed better physical processing of the forage, as indicated by the lower proportion of particles retained on the 19- and 40-mm sieves. The association between finer particle size and reduced butyric fermentation in the LB group reinforces the importance of mechanical processing. Effective chopping serves a dual purpose: (1) it releases more soluble carbohydrates needed for the growth of lactic acid bacteria (McDonald et al., 1991; Pahlow et al., 2003), promoting rapid lactic acid production and pH drop; and (2) it facilitates better compaction, reducing porosity and creating the anaerobic conditions necessary to inhibit undesirable microorganisms. Hence, particle size seems more relevant to fermentation pattern than to silage digestibility and this dataset supports the concept of independence between fermentation pattern and bromatological composition. In future surveys, silage density and its relationship with particle size and fermentation pattern warrant further characterization.
In the current survey, more than 80% of producers used bacterial inoculants, with a predominance of products containing a combination of homolactic and heterolactic bacteria (81%). Under favorable growth conditions, homolactic bacteria can accelerate the pH decline at the onset of fermentation, inhibiting undesirable microorganisms such as enterobacteria and clostridia (Pahlow et al., 2003). However, in this study, the use of inoculants was not associated with improved fermentation or nutritional value. The effects of bacterial inoculants applied to wet tropical grass silages have been inconsistent (Mühlbach, 2000; Coan, 2005; Tomaz et al., 2018). The high moisture and low soluble carbohydrate content of these forages create an environment in which even inoculated LAB may fail to dominate the epiphytic microbiota. Furthermore, the predominance of inoculants containing a combination of homo- and heterolactic bacteria may not be an optimal strategy for direct-cut wet forages, as heterofermentative bacteria prioritize acetic acid production rather than rapid acidification, potentially delaying pH decline and increasing the risk of clostridial activity (Arriola et al., 2021). This suggests that current commercial inoculant strategies, as applied by farmers, are insufficient to overcome the fundamental ensiling challenges posed by these tropical grasses, highlighting the need for more robust technological interventions.
The use of moisture-absorbing additives (e.g., soybean hulls, soybean residue, and by-products from corn ethanol production), as an alternative to wilting, was adopted by a relatively small proportion of farmers and did not differ between the LB and HB groups. Although some studies have demonstrated positive effect on the fermentation profile of grass silages (Igarasi, 2002; Ribeiro et al., 2007; Avila et al., 2022), the addition of moderate levels of moisture absorbents as the only management strategy does not necessarily ensure adequate fermentation (Tomaz et al., 2018; Gouvêa et al., 2020; Gomes et al., 2021) and can increase the complexity of the ensiling process in large operations. In this study, the lack of efficacy noted for the use of absorbents might be due to the dose, type of absorbent or either initial forage DM.