3.2. Characteristics of the Seaweed Silages
The characteristics of the
P. umbilicalis and
S. latissima silages are presented in
Table 2 and
Table 3, respectively. As expected, the DM content was greater (
p < 0.001) in the pre-wilted than in the unwilted silages for both seaweeds, but pre-wilting × silage type interactions were detected (
p < 0.001 and 0.020 for
P. umbilicalis and
S. latissima, respectively). No differences (
p > 0.05) in DM content were detected among the unwilted silages of both seaweeds, but when seaweeds were pre-wilted before ensiling the CON and WAS silages had the greatest (
p < 0.05) DM content for
P. umbilicalis and
S. latissima, respectively. The DM content of all unwilted silages was below 190 g/kg, and was similar to that of the fresh seaweeds (
Table 1). For the pre-wilted silages, the seaweeds were dried to achieve a target content of 300 g DM/kg, but the actual content was greater than that in all samples. A minimum DM content of 250 g/kg has been proposed as necessary for ensuring a good silage process [
41], but the DM of the fresh seaweeds was lower than 216 g/kg (
Table 1) and that could have affected the silage process negatively in the unwilted silages.
Apart from preserving the nutritional value of the ensiled feed to the greatest extent possible, the handling of silages must be convenient for the silage of seaweeds to be of interest for the industry, and the presence of effluent complicates silage management. Effluent release usually occurs when the initial forage DM content is below 250–300 g/kg [
41], causing a loss of nutrients and contribution to environmental pollution. Silage effluent is formed from the surface moisture of the plants and cell juice, which is released due to cell lysis during silage and washes out valuable highly-digestible compounds such as soluble carbohydrates, minerals, organic acids and alcohols [
42]. Because of the low DM content of the seaweeds, effluents were present in all unwilted silages of both seaweeds, being greater for
P. umbilicalis than for
S. latissima (25.5 and 10.8% of total weight, respectively; values averaged across silage types). In contrast, no effluent was observed in any of the pre-wilted silages, which is consistent with the greater (
p = 0.001) DM content of these silages compared with those unwilted.
For both seaweeds, the pH values in the pre-wilted silages were greater (
p < 0.001) than in those unwilted, and pre-wilting × silage type interactions were detected (
p < 0.001). As expected, the addition of FAC decreased the pH in all silages (
p < 0.05) resulting in pH values below 4.0. According to Woolford and Pahlow [
43], a pH lower than 4.2 is required for an effective silage process of samples containing about 200 g DM/kg, but due to the low DM content of the seaweeds even lower pH values might be required to prevent clostridial fermentation and the production of butyric acid. In the present study, only the FAC-silages reached the required pH. The pH of the pre-wilted silages without FAC (CON and WAS) was greater than the values reported by Cabrita et al. [
25] for
S. latissima silages when the seaweed was pre-wilted at 18–20 °C for 24 h (4.48 and 4.10 for silages with and without a microbial inoculant, respectively). Our values were also greater than those observed in
S. latissima silages by Campbell et al. [
26] when the seaweed was pre-wilted for 24 h (3.90) and by Herrman et al. [
44] when the seaweed was not pre-wilted before ensiling (3.80). To our best knowledge this is the first report on
P. umbilicalis silages, but the pH values (5.75 and 4.92 for pre-wilted CON and WAS, respectively) were in the range of those reported in previous studies [
15,
16] for silages of pre-wilted
Ulva rigida (5.10),
Gracilaria vermiculophylla (5.20) and
Fucus vesiculosus (4.90) without any silage additive. Differences observed among seaweeds may be due to different content in easily fermentable components, but also to the silage conditions used in the different studies. As already discussed, ensiling seaweeds is challenging due to their low levels of water-soluble carbohydrates and of epiphytic lactic acid bacteria [
44] and high buffering capacity [
45], that can cause the poor reduction in pH during the silage fermentation process [
44,
46].
Concentrations of NH
3-N were greater (
p < 0.001) in the unwilted than in pre-wilted silages for both seaweeds, and CON silages showed the greatest (
p < 0.05) NH
3-N concentrations except for the pre-wilted silages of
P. umbilicalis which showed no differences among them. The concentration of NH
3-N is a common indicator of the proteolytic activity during silage fermentation [
41], and our results indicate that pre-wilting decreased the proteolysis in both seaweeds. In general, the NH
3-N concentrations were low, especially in the pre-wilted silages (<0.80 g/kg DM), suggesting a low proteolytic activity. When the NH
3-N content was expressed as proportion of total N content in the silages, the NH
3-N in unwilted
S. latissima silages was 6.61, 3.24 and 3.39% of total N content for CON, WAS and WASFAC, respectively, and only 2.42, 0.89 and 1.88% of total N for the CON, WAS and WASFAC pre-wilted silages. These proportions are lower than the values ranging between 4.7 and 6.7% of total N reported previously for pre-wilted
S. latissima silages [
25,
26]. The NH
3-N content in the unwilted
P. umbilicalis silages, expressed as % of total N, reached greater values (15.1, 7.32 and 5.11% for CON, WAS and WASFAC, respectively), but most of them were lower than the 10.7 to 100% values reported in other studies for pre-wilted silages of
Ulva rigida,
F. vesiculosus and
G. vermiculophylla [
25,
26]. Content of NH
3-N in
P. umbilicalis pre-wilted silages was low, reaching only 0.49, 2.19 and 1.17% of total N for CON, WAS and WASFAC, respectively. In accordance with the greater NH
3-N concentrations observed for the unwilted silages, free amino acids content was in general greater in the unwilted silages compared with those pre-wilted for both seaweeds. Significant pre-wilting × silage type interactions were detected for valine in
P. umbilicalis and for most amino acids in
S. latissima silages. For
P. umbilicalis, the WAS silage had greater (
p < 0.05) concentrations of free amino acids than CON and WASFAC, whereas for
S. latissima CON silages had the greatest content in free amino acids.
Glucose concentrations were low in the
P. umbilicalis silages (<0.760 g/kg DM) and were not affected (
p ≥ 0.110) by either pre-wilting or silage type, but glucose concentrations in
S. latissima silages were greater (6.22 to 78.8 g/kg DM) and were affected (
p ≤ 0.002) by both pre-wilting treatment and silage type.
S. latissima, also known as “sugar kelp”, is rich in high fermentable sugars [
44], which might help to explain the greater glucose concentrations in its silages. Pre-wilting decreased (
p = 0.002) glucose concentrations in
S. latissima silages. Similarly, the use of FAC reduced (
p < 0.05) the glucose concentration in the silages compared with those untreated with FAC.
Pre-wilting tended to decrease (
p = 0.070) DL-lactate concentrations in
P. umbilicalis silages, and the WAS silages showed greater (
p < 0.05) concentrations of DL-lactate than CON and WASFAC regardless of pre-wilting. Pre-wilting also decreased (
p < 0.001) DL-lactate concentrations in
S. latissima silages, but a pre-wilting × silage type interaction was detected (
p = 0.005). For the unwilted silages, WASFAC had lower (
p < 0.05) DL-lactate concentrations than CON and WAS, whereas no differences among silage types were detected in the pre-wilted silages. In agreement with our results, low lactate concentrations (<14.0 g/kg DM) have been reported for silages of
G. vermiculophylla [
25],
F. vesiculosus [
26],
U. lactuca [
47] and
Ascophyllum nodosum [
44], but much greater lactate concentrations (up to 228 g/kg DM) have been shown for pre-wilted silages of
S. latissima [
25,
26,
44]. In well-preserved silages, lactic acid bacteria are active in degrading water-soluble carbohydrates and the lactate produced causes a decrease in pH that preserves the silage [
41]. The low lactate contents observed in our study are consistent with the greater pH values compared with those reported in previous studies with
S. latissima silages previously discussed [
25,
26,
44]. Altogether, these results indicate that the activity of lactic acid bacteria in our silages was scarce, and that the good ensilability characteristics of
S. latissima reported by other authors were not observed in our study.
In unwilted S. latissima silages, the average L:DL-lactate ratio was 0.15 indicating that bacteria mainly producing D-lactate were responsible for most of the lactate production in these silages. By comparison, average L:DL-lactate ratio was 0.57 in unwilted P. umbilicalis silages. In silages made from terrestrial plants like grass, clover and maize whole crop, a L:DL-lactate ratio close to 0.50 (racemic) is found, and this was also expected for seaweed silages. Substances in the complex seaweed matrix might have inhibited the membrane oxidase used in L-lactate analysis causing the low L-lactate concentrations in S. latissimi; however, the recovery test of added L-lactate to seaweed silage samples showed complete recovery.
Pre-wilting of
P. umbilicalis silages decreased (
p ≤ 0.004) the production of all VFA detected (
Table 2), except for valerate and caproate. In general, washing the seaweeds and adding FAC decreased (
p < 0.05) concentrations of acetate, propionate, butyrate, isobutyrate and isovalerate of the unwilted silages. Similar or even greater VFA concentrations have been reported for silages of
G. vermiculophylla [
25],
U. lactuta [
44]
Laminaria digitata [
44]. The high VFA content (especially of butyrate) of the unwilted silages, together with their great NH
3-N concentrations, seems to indicate a clostridial fermentation with proteolytic clostridia degrading the proteins to NH
3-N [
43]. However, in the pre-wilted silages acetate was practically the only VFA noticed, as the other VFA were detected in negligible amounts and only for some silage types. The low butyrate and NH
3-N concentrations (<0.21 and 0.80 g/kg DM, respectively) would not indicate a clostridial fermentation in these silages.
In contrast to that observed for
P. umbilicalis, acetate and butyrate were the only VFA detected in
S. latissima silages, and butyrate was only detected in the unwilted CON silages (
Table 3). The CON silage also showed the greatest (
p < 0.05) acetate concentration for unwilted silages, whereas there were no differences among silage types when
S. latissima was pre-wilted. These results, together with the low NH
3-N concentrations (<0.64 g/kg DM), seem to preclude a clostridial fermentation in
S. latissima silages. Our results agree well with previous studies showing acetate concentrations ranging from 2.2 to 3.9 g/kg DM and very low or null levels of other VFA in pre-wilted
S. latissima silages [
25,
26].
Ethanol concentrations in
P. umbilicalis silages were decreased (
p < 0.001) by pre-wilting to low values (≤0.36 g/kg DM), but they were not affected (
p = 0.250) by silage type. In addition, very low levels of propanol, 2-butanol, ethylacetate and propylacetate (≤1.04 g/kg DM) were observed in all
P. umbilicalis silages, being undetectable in most of the pre-wilted silages. Ethanol concentrations in
S. latissima ranged from 0.050 to 2.14 g/kg DM, were not affected by either pre-wilting or silage type, and were lower than those reported previously (up to 47 g/kg DM) for
S. latissima silages [
26,
44]. Similar to that observed for
P. umbilicalis silages, concentrations of other alcohols and esters were negligible (≤0.018 g/kg DM) in
S. latissima silages. In agreement with our results, ethanol was the main alcohol in silages of
F. vesiculosus,
L. digitata and
S. latissima silages, and concentrations of other alcohols were minor in previous studies [
26,
44]. Ethanol content in silages is usually attributed to the activity of epiphytic, salt-tolerant yeast populations [
47].
The differences observed in the characteristics of
S. latissima and
P. umbilicalis silages are yet to be elucidated, but they indicate different ensilability possibly due to differences in chemical composition, content in lactic acid bacteria, and buffering capacity, among others [
44,
47,
48].
3.3. Chemical Composition and In Vitro Gas Production Parameters of the Seaweed Silages
Pre-wilting promoted limited changes on chemical composition of silages (
Table 4 and
Table 5), as the only changes observed were an increase in the NDF content for
P. umbilicalis (
p < 0.001; 156 vs. 334 g/kg DM; values averaged across silage types) and in the N content for
S. latissima (
p = 0.036; 8.90 vs. 9.10 g/kg DM). The increase in NDF content of
P. umbilicalis might be due to losses of small particles through the holes of the net used in the pre-wilting process, as these particles probably had lower NDF content. In contrast, there were marked differences among silages type, and pre-wilting × silage type interactions (
p ≤ 0.043) were observed for all chemical fractions of
P. umbilicalis. As expected, washing decreased (
p < 0.05) ash content, and for pre-wilted
P. umbilicalis resulted in greater (
p < 0.05) N, ADIN and ADF content and lower (
p < 0.05) NDF content; however, when
P. umbilicalis was not pre-wilted washing only decreased (
p < 0.05) N and NDF content of the silages. The inclusion of FAC did not affect chemical composition of
P. umbilicalis, with the exception of a decrease (
p < 0.05; 324 vs. 234 g NDF/kg DM) in the NDF content of pre-wilted silages. Previous studies [
49,
50] have shown that formic acid can cause a decrease in NDF concentrations of ensiled plants due to acid hydrolysis.
Differences among the
S. latissima silages were less marked than those observed for
P. umbilicalis silages, and pre-wilting × silage type interactions were only observed for N content (
p = 0.014). Washing decreased (
p < 0.05) both ash and N contents, although differences for unwilted
S. latissima silages did not reach the significance level. The use of FAC did not change chemical composition of
S. latissima silages. Previous studies showed that ensiling
S. latissima decreased their ash content [
26,
44], whereas Cabrita et al. [
25] reported an increase in ash content in the silage. In our study, the ash content of both seaweeds in the silage was similar than that observed in the fresh biomass (
Table 1). The N content in the silages of both seaweeds was similar to that of the fresh biomass (34.5 and 36.0 g N/kg DM for silages and fresh biomass of
P. umbilicalis, respectively; 9.00 and 8.71 g N/kg DM for
S. latissima; values averaged across treatments). This indicates that no N losses occurred during silage fermentation and it is in accordance with the low NH
3-N content observed in the silage extracts. In agreement with our results, Cabrita et al. [
25] reported no CP losses during silage fermentation of pre-wilted
S. latissima for 9 weeks.
The NDF content of
P. umbilicalis silages was lower than that in the fresh seaweed, especially in the unwilted silages. Cabrita et al. [
25] observed similar NDF decreases during silage fermentation of the red seaweed
G. vermiculophylla. These results, together with the lack of decrease in ADF content, might suggest that hemicellulose-like components were degraded during the silage fermentation process. Cabrita et al. [
25] observed increases in acetate and butyrate content in the silages showing reduced NDF levels, which was attributed to fibre degradation during silage fermentation. In accordance with these results, greater concentrations of both acids were observed in the silages that showed decreased NDF levels compared with fresh
P. umbilicalis (all silages except the pre-wilted CON;
Table 2). The negative correlations between NDF content and both acetate (r = 0.699;
p = 0.002; n = 18) and butyrate (r = 0.479;
p = 0.050; n = 18) concentrations in the extracts of
P. umbilicalis silages support this hypothesis.
On the contrary, the NDF content in all
S. latissima silages was similar to that in the fresh biomass, and no relationships between NDF content and concentrations of acetate (
p = 0.911) and butyrate (
p = 0.344) were detected. Similarly, Cabrita et al. [
25] observed no changes in NDF content during the silage fermentation of
S. latissima. On the contrary, Campbell et al. [
26] reported that NDF and ADF contents were 4.5 and 2.4 times lower, respectively, in
S. latissima silages than in the fresh seaweed. These authors attributed the NDF and ADF losses to the degradation of laminarin and/or mannitol during the silage fermentation process, as they are the main storage carbohydrates in brown seaweeds [
51] and can be used as substrate by the lactic acid bacteria during silage fermentation [
52]. It should be noticed that the NDF and ADF content of the
S. latissima ensiled in the study of Campbell et al. [
26] was unusually high (384 and 184 g/kg DM, respectively), and the samples used by Cabrita et al. [
25] and in our study had lower NDF and ADF levels (<130 and 66 g/kg DM, respectively). Differences in the carbohydrate content and profile of
P. umbilicalis and
S. latissima might be responsible for the different changes in fiber content observed in our study for both seaweeds, and for the contrasting results among studies in which the same seaweed was ensiled. Finally, it is worth to mention that lignin contents in
S. latissima silages were similar to those reported previously [
25,
26], despite no lignin was detected in the fresh biomass of this seaweed. The presence of lignin in the silages might be due to protein-fiber interactions occurring during silage fermentation, similar to those produced during heating that result in increased ADIN proportions [
35,
53]. The greater ADIN proportions in the
S.latissima silages compared with the fresh seaweed are consistent with this hypothesis.
The gas production kinetics of silages was measured as an index for their ruminal degradation, as the amount of gas produced is closely related to the amount of organic matter degraded by ruminal microorganism [
37]. As shown in
Table 4, pre-wilting of
P. umbilicalis increased (
p ≤ 0.022) the fractional rate of gas production (
c), AGPR and DMED, and pre-wilting × silage type interactions were detected for these parameters (
p ≤ 0.093). The greater PGP,
c, AGPR and DMED values of the WASFAC silages compared with CON and WAS would indicate that FAC treatment increased the ruminal degradation of silages. These results are in accordance with the lower NDF levels in WASFAC compared with CON silages, as NDF has usually lower ruminal degradation than other compounds such as protein or soluble carbohydrates [
53]. The gas production parameters of the silages were similar to those of the fresh
P. umbilicalis for CON and WAS silages, but WASFAC silages had greater PGP, c, AGPR, DMED values, which indicates that FAC treatment increased the ruminal fermentation of the silages DM.
In agreement with the results observed for
P. umbilicalis, pre-wilting of
S. latissima increased (
p ≤ 0.006)
c and AGPR values, but pre-wilting × silage type interactions (
p ≤ 0.092) were detected for all gas production parameters excepting
c (
Table 5). Pre-wilting also increased (
p = 0.026) the time until gas production started (
lag) and decreased DMED (
p = 0.029; 462 vs. 452 g/kg; values averaged across silage types) and PDMD values (915 vs. 894 g/kg), which might be due to the loss of small particles of high-degradable material during pre-wilting. However, the differences were low and no relevant in the practice. Including FAC in the
S. latissima silages increased (
p < 0.05) AGPR and DMED compared with the FAC-untreated silages. Altogether the results indicated that FAC could increase the ruminal degradation of seaweeds silages.
Samples of two good-quality silages [
54] of perennial rye-grass were included in the in vitro incubations to be used as a reference (
Table 6). Chemical composition of rye-grass silages differed markedly from that of seaweeds silages, as they had greater NDF and ADF content, their N content was lower than in
P. umbilicalis silages but greater than in
S. latissima silages. Chemical composition of the perennial rye-grass silages was similar to that previously reported for high-quality rye-grass silages [
55,
56]. The lower DMED of the seaweed silages compared with that of the rye-grass silages (59.9 and 86.1% of the rye-grass silage values for
P. umbilicalis and
S. latissima, respectively) indicates lesser nutritional quality of the seaweed silages, especially of those from
P. umbilicalis. The differences observed in the DMED of the two seaweeds silages are probably related to their chemical composition, as brown seaweeds are richer in non-structural carbohydrates than red and green seaweeds and are therefore more degradable in the rumen [
12,
21]. In addition, the presence of complex carbohydrates and biocompounds in seaweeds might limit their ruminal fermentation, as specific enzymes are needed for their degradation [
12], and other factors, such as the structure of the cell wall, can also limit the access of microbial enzymes to the substrates [
57]. Interestingly, the DMPD of the
S. latissima silages was similar to that of the rye-grass silages (905 and 915 g/kg, respectively; averaged values), although that
of P. umbilicalis silages was lower (804 g/kg). These results disagree with the fact that PGP of the rye-grass silages was 3.0 and 1.4 times greater than in the
P. umbilicalis and
S. latissima silages, respectively (averaged values, 273, 90.3 and 196 mL/g DM). All seaweed silages had high content of ash (>110 and 252 g/kg DM for
P. umbilicalis and
S. latissima silages, respectively), that can disappear instantly from the bags during the incubations in the rumen or in vitro, causing an overestimation of their rumen degradability [
58]. In contrast, ash is not fermented in the rumen and therefore does not contribute to gas production as measured in the present study.