Forages

Biomass of TD was quantified in the pre-flowering stage (60 d) for a cut made 40 cm above ground level, including leaves and stems, in February 2013. The TD was grown at the experimental farm of the National University of Colombia (Universidad Nacional de Colombia), Palmira (1,000 m asl, 24 °C average annual temperature, 1,020 mm average annual rainfall precipitation and 72 % average annual relative humidity). Collection of PP was done at 75 d of age (10 cm above ground level). Both forages were reduced to a 2-3 cm particle size with a three-blade mill (7.5 hp, 1400 rpm, and 4.5 amps; Gaitan).

Silage preparation

When ensiled, humidity in the TD was 30 % of DM and that in the PP forages was 35 % of DM. Four treatments were used: T1) Control, 100% PP silage no inoculum; T2) mixture of PP (67%) and TD (33%) silage no inoculum; T3) mixture of PP (67%) and TD (33%) silage, LAB inoculum (Lactobacillus paracasei - T735 previously isolated from TD); and T4) mixture of PP (67%) and TD (33%) silage, commercial Sil-All®4x4 inoculum [LAB mixture: Streptococcus faecium (National Collection of Microorganism Cultures - CNCM; I-3236), L. plantarum (CNCM I-3235), Pediococcus acidilactici (CNCM I-3237) and L. salivarius (CNCM I-3238)] (Table 1). The mini silos were done in triplicate. Both inocula were applied at a 104 CFU/g concentration. The forage mixtures (1,000 g each) were vacuum-packed following the Rostock model for silages ¹⁹, and stored for ninety days in darkness at room temperature (25 °C). On d 90, the min silos were opened, and the silages lyophilized and ground (Thomas Wiley Mill 4, with a 1.0 mm sieve).

In vitro ruminal fermentation in RUSITEC system

The RUSITEC system consists of eight fermenter tubes that allow simultaneous evaluation of a maximum of eight treatments. In the present study four treatments, two repetitions per treatment, were analyzed; the run effect was used as a repetition. Experimental period was ten days, of which the first four days corresponded to the adaptation period of the microorganisms to the experimental diets and the following six days were for data collection and sampling.

Ruminal fluid (890 ml) and buffer solution or artificial saliva (110 ml) were added to each of the eight fermenter tubes ²⁰ . Before being added, the ruminal liquid was filtered through four layers of gauze. Artificial saliva flow was controlled with a pump to ensure a continuous supply of 500 ml per day per fermenter, equivalent to a 0.5 per day dilution rate.

At the beginning of each experimental period, 60 g ruminal content and 16 g DM separately packed experimental silage were added to each fermentation tube for incubation. The diets were packed in nylon bags measuring 13.5 x 6.5 cm with 100 µm pores (NItex 03-100 / 32, SEFAR, Heiden, Switzerland). Subsequently, every 24 h one of the bags was replaced by a new one, starting with the ruminal content bag. In this way each bag was incubated for 48 h. After a bag change, air in the fermenter was displaced with N2 to restore anaerobic conditions. The removed bags were washed with cold water until the wastewater ran clear, and stored at -20 °C until analysis.

The fermenters were fed daily. Four hours before feeding, ruminal fluid samples were taken to quantify redox potential, pH, ammonium concentration and microorganism count. Gas collected from the fermenters was stored in 10 l bags (Supel™ Inert Foil Gas Sampling Bags, Screw Cap Valve) and quantified by water displacement.

Chemical analysis

Chemical composition analyses were run on the samples before digestion in the RUSITEC and on the incubation residues after digestion. Bromatological analyses were done following established methods: nutritional quality, NFTA Method 2.1.4 in oven at 105 °C (930.15)²¹; acid detergent fiber (ADF), NFTA Method 4.1 (973.18) ²²; neutral detergent fiber (NDF) ²³,²⁴,²⁵; and crude protein (CP), Kjeldahl (984.13) ²². Estimated dry matter digestibility (EDMD) was calculated with the equation: EDMD = 88.9 - (0.779 x ADF).

Ammonium concentrations were measured following the Ammonia (NH3) Electrode Instruction Manual. Three points (1, 10 and 100 mmolar) were calibrated. For each calibration 20 ml molar solution was stirred and 1 ml sodium hydroxide added, and this repeated for the other two points (10 and 100 mmolar). Once calibrated, a 1 ml inoculum sample was taken, 18 ml distilled water and 1 ml sodium hydroxide added, and the reading taken.

pH and redox potential

Redox potential and pH were measured with a pH/ion meter (SG8, Mettler-Toledo series: B 337764747). After device calibration, a 6 ml sample was taken from each fermenter and the measurement taken with the indicated electrode.

Protozoan count

Protozoa counts were done using 2,000 µL Hayem solution (2.5 g/L HgCl2; 25 g/L Na2SO4; 5.0 g/L NaCl) and 500 µl ruminal fluid (days 0 and 1), or 100 µl Hayem solution and 1,000 µl ruminal fluid (d 2 to 8). The mixtures were placed in plastic tubes, these placed in a 0.1 mm deep Neubauer chamber (Hausser Scientific), and the protozoans present in the entire chamber area counted ²⁶.

Methane

Methane (CH4) was measured with a chromatographer (Shimadzu GC-2014). The column (Shimadzu) had a series of columns packed for methane: 4m H-D 80/100, 0.7m S-Q and 1.5 P-N. Column temperature was 80 °C and detector temperature was FID 25 °C. This device does not employ a traditional injection port and injection is directly connected to a cable, which is at room temperature. The carrier gas was nitrogen at a column flow rate of 30.83 ml/min.

Statistical analysis

Data were analyzed using a completely randomized design (P<0.05). The variables were evaluated with general and mixed linear models for repeated measurements (Infostat, ver. 2010 software). The model included the fixed effects of treatment, measurement time and their interaction. Incubation time was considered a random effect. Different covariance structures were analyzed for each variable and the best model estimated based on Akaike (AIC) and Bayesian (BIC) information criteria.

Ruminal liquid

After 8 d fermentation in the RUSITEC, pH was slightly higher than normal rumen acidity range (7.3 to 7.4), but without differences (P>0.05) between treatments (Figure 1). High pH levels are likely a response to a highly fibrous diet since long fibers stimulate rumination and secretion can increase saliva production ²⁹. Saliva functions as a lubricant for the consumed feed, and can raise pH to an average of 8.2, as well as increase sodium, potassium, bicarbonate and phosphate levels ^29,30). These characteristics are simulated in the artificial saliva used in the RUSITEC, which may initially raise ruminal liquid pH, but its buffer capacity later stabilizes it. In addition to this effect of the artificial saliva, the trend from acidic towards basic pH in the present results may have resulted from a downward trend in organic acids production over time.

T1= Control, 100% PP silage no inoculum; T2= mixture of PP (67%) and TD (33%) silage no inoculum; T3= TD/PP (33/67%) silage, LAB (Lactobacillus paracasei - T735); and T4= TD/PP (33/67%) silage, commercial Sil-All^®4x4.

Figure 1 Rumen acidity in silages made from a mixture of Pennisetum purpureum (PP) and Tithonia diversifolia (TD) in a RUSITEC system 

Balance in the natural rumen environment is maintained thanks to a buffer solution derived from the alkaline salivary secretion. This modulates the acidity of carbohydrate fermentation in the rumen, within a range of 5.8 to 7, generally near neutral²⁹. Of all the factors of the rumen medium, pH is the most susceptible to variation, and ration is the factor that can most affect it. Maintaining ruminal pH within an adequate range is the result of production and neutralization or elimination of protons in the rumen medium. Fermentation of non-structural carbohydrates (NSC) is more energy efficient but it is highly acidic; therefore the proportion of NSC in a diet needs to be limited or counteracted with structural carbohydrates (SC) which provide a buffering capacity to the rumen environment³¹.

Redox potential did not differ (P>0.05) between treatments (Figure 2), which suggests that simulation in the RUSITEC closely approximated a rumen’s anaerobic environment. At times oxygen can be present in the rumen, possibly because it has entered with feed or water ^3,15. Low oxygen concentrations, as indicated by negative oxidation potential (Eh) values between -250 and -450 millivolts (mV), stimulate the growth of anaerobic microorganisms³².

T1= Control, 100% PP silage no inoculum; T2= mixture of PP (67%) and TD (33%) silage no inoculum; T3= TD/PP (33/67%) silage, LAB (Lactobacillus paracasei - T735); and T4= TD/PP (33/67%) silage, commercial Sil-All^®4x4.

Figure 2 Redox potential of silages containing Pennisetum purpureum (PP) and Tithonia diversifolia (TD) and digested in a RUSITEC 

Dry matter degradation

Mean DM degradation over eight days of digestion in the RUSITEC was significantly higher (P<0.0001) in T3 (41.33 ± 5.63) than in T1 (37.88 ± 5.76) and T2 (34.61 ± 4.79) (Figure 3), but did not differ significantly from T4 (39.71 ± 5.90). Longitudinal analysis also identified differences (P<0.0001) between degradation at 8 days and at the beginning of the period. The diets containing TD exhibited higher DM degradation, which can be explained by their lower non-legume protein dicotyledonous fiber content and higher forage digestibility and degradability, both inversely related to NDF content²⁸.

T1= Control, 100% PP silage no inoculum; T2= mixture of PP (67%) and TD (33%) silage no inoculum; T3= TD/PP (33/67%) silage, LAB (Lactobacillus paracasei - T735); and T4= TD/PP (33/67%) silage, commercial Sil-All^®4x4.

Figure 3 Dry matter degradation in silages containing a mixture of Pennisetum purpureum (PP) and Tithonia diversifolia (TD) and digested with a RUSITEC 

Volatile fatty acids (VFA)

Volatile fatty acids (VFA) levels were not affected by treatment in that no differences were present, although VFA levels were slightly lower in T3 (Table 3). In forage-based diets VFA can provide between 50 and 85 % of energy requirements. Moreover, the proportions, relative to total rumen gasses, between VFA and greenhouse gases (GHG) is 65 % for CO₂, 27 % for CH₄ and 0.2 % for free H₂³³.

Fermentation product composition differed between treatments. Though not significant (P= 0.4875), acetate drainage was higher in T1 compared to the treatments containing TD. A similar trend was observed for propionate drainage, although again with no differences (P= 0.3525). Butyrate production did differ between treatments (P<0.0001), with the highest production in the control (T1), and progressively lower levels with addition of TD (T2) and the LAB inocula (T3 and T4).

In ruminants, formation of propionate is more energy efficient than that of butyrate and acetate because production of the latter two leads to release of carbon atoms that cannot be burned in the form of CH₄ (or H atoms that do not convert to VFA)³⁴. This means that grass-only diets lead to high segregation of butyrate as C atoms are released as CH₄. Total VFA was not affected by the treatments in the present study but the VFA profile was notably modified, which coincides with a previous report on diet additives in a RUSITEC³⁵. This is confirmed by the significant differences in the acetate/propionate ratio, which was higher in T4 and T3 (P=0.0009) than in T1. Normal VFA molar proportions in the rumen are acetic acid (65 %); propionic acid (20 %); butyric acid (13%) and others (2 %)³⁶. Though slightly altered, these general proportions were observed in the present results with T4 having the proportions nearest these estimates: 60.7 % (acetate); 26.7 % (propionate); 9.7 % (butyrate) and 2.9 % (others). In diets based entirely on forages the acetate:propionate ratio approaches 3:1, while as diet concentrate percentage increases the ratio narrows to 2:1 or less¹⁶.

Ammoniacal nitrogen (NH₃-N)

Production of NH₃-N (as a measure of dietary N metabolism) was lower in the treatments including TD, suggesting lower dietary protein degradation in these treatments. The higher ammonium level in T1 was probably related to increased proteolysis. Daily NH₃-N production changed over time in all treatments (P<0.0001); generally, values increased on the first day and then gradually decreased, which agrees with previous reports¹⁶.

On day 8, NH₃-N was highest in T1 compared to the treatments containing TD. Ammonium content did not differ among these treatments although it was slightly higher in T3, even though DM digestibility did not differ between them. This occurred despite the greater N input in T3. Previous studies have shown that in high protein forages incorporation of dietary N in microbial N increases as NH₃-N decreases^37,38. This discrepancy indicates that protein degradation may occur at a slower rate in diets containing TD. Ruminal ammonia concentration in the different treatments was much lower than the 3.26 mmol/l reported elsewhere³⁹, but near the 1.76 mmol/l observed in an evaluation of a mixed substrate (1/3 Cratylia argentea, 2/3 Brachiaria dictyoneura) in a RUSITEC system³. However, these levels are still lower than the 3.6 mmol/l suggested for maximum microbial protein synthesis in the rumen⁴⁰.

Methane emission

At 8-days incubation, net CH₄ production per g DM was higher in T1 than in T4, which had the lowest production (Table 3). The present CH₄ production was similar to the 1.47 and 1.61 mmol/g DM reported for a mixture of B. dictyoneura hay and additives³⁹. This also generally agrees with a study using the gas production methodology in which grass-only silages emitted high levels of CH₄ which decreased as TD inclusion levels increased, and where 80 to 88 % of the total gas produced at 60 h incubation was detected at the inflection point⁴¹. In another study of PP/TD forage mixtures CH₄ production decreased as TD inclusion increased from 15 % (33.3 ml) to 30 % (30.1 ml) and 100 % (28.06 ml)⁴². Decreases in CH₄ emission per gram degraded MS are probably due to lower fiber degradation, suggesting that less CH₄ would be emitted per unit of edible animal protein with the TD treatments, even though total CH₄ emission may not decline³⁹.

The lower CH₄ production in the PP/TD silage in the present results may be due in part to the action of condensed tannins (CT), which are known to reduce CH₄ emissions in woody forage species⁴³. Tithonia diversifolia (TD) forage has relatively moderate CT levels⁹, which can vary from 1.0 % in the dry season to 1.4 % in the rainy season⁴⁴; these are not negligible levels.

The higher CH₄ production in T1 was probably due to the greater proportion of cell wall in this all-grass forage. Cell walls contain more NDF, which has low ruminal digestibility, meaning more NDF passes through the rumen into the large intestine where it can be fermented and thus produce CH₄ through the action of methanogenic microbes^35,43.

Protozoa population is a good measure of rumen microbial biomass since they have relatively high weight among the microbiota. Though their net count may be smaller than the bacteria, the protozoa have a greater individual volume, resulting in a protozoan cell mass similar to the bacterial mass¹⁶. The protozoa also colonize and degrade plant tissues in the rumen and produce enzymes capable of degrading plant and hemicellulose polysaccharides²⁹. In the present results the protozoa population changed over time in all treatments with values gradually decreasing during the first three days and then stabilizing during the last four days at average values that did not differ between treatments (P>0.05). Among the TD treatments (T2, T3 and T4), the protozoa count was relatively lowest in T3, which also had the highest CP proportion. This coincides with a study in which degradation was highest in substrates supplemented with a high-quality legume forage³⁹. However, these same forages can have high secondary metabolite (e.g. tannins and saponins) contents which can negatively affect ciliate (protozoa) populations; this would definitely improve nitrogen use but not necessarily reduce methanogenesis.