Feeding Strategies to Increase Nitrogen Retention and Improve Rumen Fermentation and Rumen Microbial Population in Beef Steers Fed with Tropical Forages

: The effect of the inclusion of Leucaena leucocephala and Tithonia diversifolia in Zebu steers receiving a diet based on improved pastures such as Brachiaria decumbens and Brachiaria hybrid cv Cayman on nitrogen (N) excretion, urinary volume and rumen microbial population was evaluated. To determine the dry matter intake and nutrient excretion, eight steers were used in a 4 × 4 Latin square design consisting of four periods and four diets. Four of them were cannulated for ruminal ﬂuid extraction and quantiﬁcation of ruminal microorganisms in three times of grazing (T0, T7 and T15). Forage intake was calculated through the external marker titanium dioxide. Diet including forages with superior protein content generated an increase in the gene copy numbers of Prevotella ruminicola and total bacteria on 15 sampling day ( p < 0.001). Animals receiving diets with the dietary inclusion of Leucaena and Tithonia had daily N intakes of 228 and 113.5g N intake d − 1 , of which they excreted 42% and 61%, respectively. Inclusion of both protein forages increased daily urinary volume (9% and 7% d − 1 ), with respect to the pasture-based diet. This study revealed that the inclusion of 18% Leucaena in a pasture-based diet improves the dry matter intake and N retention in Zebu steers under tropical conditions.


Introduction
Improving productivity is a challenge for the entire agricultural sector, including livestock production, since the growth of the world's population requires meeting the growing demand for more food with fewer resources. Latin America and the Caribbean contribute 30% of the global demand for beef with South American countries (i.e., Brazil, Argentina, Uruguay, Paraguay and Colombia) contributing 80%, and the remaining 20% coming from Mexico and Central America [1]. Although it has all the potential to constitute a sustainable productive system, since its inception cattle production has developed as a productive activity based on grass monocultures, commonly offering diets characterized by high-fiber and low crude protein (CP) contents, specifically in the tropical region [2]. In cattle production systems, a low-protein diet reduces the efficiency of the productive system, since it can be unfavorable for microbial growth, ruminal digestion, availability of nitrogen (N) and carbohydrates by the animals [3]. Ruminants depend mainly on the essential amino acids obtained from the microbial protein and the energy provided by the organic acids produced during ruminal fermentation [4]. A deficit of metabolizable energy and the low CP content of the diets usually consumed by livestock generate losses of N and excretion of nutrients that lead to great economic and environmental costs. The excess to the Köppen-Geiger classification, this sub-region is classified as Climate Tropical dry (As) [19]. The study site was situated in the context of a livestock agroecosystem corresponding to the agroecosystem of mosaic of crops and pastures of temperate semi-arid climate in fans with very severe erosion soil and ustic regime [20].
The average temperature in the research center and the annual rainfall are 28.4 • C and 953 mm, respectively (CIAT meteorological station). In Palmira the average temperature is 24.5 • C and the annual rainfall is in the range of 1000-1500 mm, with the rainiest months occurring between April and October [21]. This study was conducted between the months of February 2017 and January 2018. During the experimental feeding period, the average temperature registered was 25.3 • C (max 30.5 and min 20.2 • C), and the average rainfall was 169.1 mm.

Animals and Experimental Design
This experiment was conducted following protocols that ensured animal welfare. All experimental protocols were reviewed and approved by an external Ethics Committee for Scientific Research of the Center for Research on Sustainable Agricultural Production Systems-CIPAV.
Eight Zebu 20-month-old steers with an initial weight of 282 ± 16.4 kg were used, for the quantification of dry matter intake and nutrient excretion in dung and urine. Four of them were cannulated for extraction of ruminal fluid and quantification of the abundance of ruminal microorganisms.
A 4 × 4 Latin square design was used, with four treatments and four periods. The diets consisted of two grasses: Brachiaria hybrid cv CIAT BR02/1752 (Cayman) and Brachiaria decumbens CIAT 606 cv both in monoculture. Other treatment consisted of a silvopastoral system of Cayman associated with L. leucocephala, and the last treatment consisted of a mix of B. decumbens with T. diversifolia. The evaluation periods for variables such as fecal production, urinary volume, N excretion and dry matter intake consisted of 15 days of grazing, of which 12 days were an adaptation phase to the different diets and 3 days for sample collection. To register microbial changes at the rumen level, the sampling days included day 0, where the animals entered each treatment (plot), followed by day 7 of grazing and finally to day 15 of grazing.
Four experimental plots of 46 × 63 m ha were established, which were directly grazed by steers. In the silvopastoral system, the L. leucocephala was established at a planting density of 1428 plants per hectare. As a contrast plot, Cayman grass was established in monoculture. In another two plots, B. decumbens grass was established in monoculture through sexual seed. A fifth plot of 40 × 46 m ha with T.diversifolia next to the others treatments was established. The T.diversifolia was sown through vegetative material at a planting density of 2 plants per m 2 . The mixed diets were offered in an 80:20 and 65:35 ratio for Cayman/L. leucocephala and B. decumbens/T. diversifolia, respectively. Both leaves and stems of the T. diversifolia forage were supplied in individual feeders every day within the grassland. Each plot with the treatment was divided into strips using electrical fence tape to manage the grazing of the animals. Every three days the animals moved to a new strip.
All plots were fertilized during establishment with di-ammonium phosphate and 46%N-18%P-0%K. Grasses were offered to steers with a regrowth period between 50 and 55 days, while T. diversifolia and L. leucocephala were offered with a regrowth between 50 and 60 days, respectively. Dry matter intake was calculated using the external marker titanium dioxide (TiO 2 ) and the diet digestibility was determined through the internal marker indigestible neutral detergent fiber (iNDF). For 15 consecutive days, animals received 10 g of the external marker in the morning and in the afternoon (20 g TiO 2 per day) with an interval of eight hours between AM and PM supply [22]. In the last three days of marker supply, individual samples of feces were collected twice per day from each animal, which were immediately weighed on a digital scale (SF-400) to determine their fresh weight. They were oven-dried at 60 • C for three days and ground using a 2 mm sieve. All samples per animal were mixed to deliver the marker; a mixture of wheat flour, vegetable oil and water was used, with which a malleable mass was obtained that allowed the wrapping of the 10 g of TiO 2 for each dose. The final weight of the mixture (mass/TiO 2 ) was 32.5 ± 6.5 g. The concentration of the marker in feces was performed through the H 2 O 2 digestion method [18], with amounts of 0, 2, 4, 6, 8 and 10 mg of TiO 2 used as standard concentrations. The analysis obtained a determination coefficient of 0.9951 between TiO 2 added (mg) and TiO 2 analyzed (mg), and the model of the linear equation was Y = 0.1346 × −0.0729.
A first test was carried out where the dung samples were collected in the morning and afternoon, as well as the mixture of both being analyzed separately. This allowed the determination that the level of variation and concentration of the marker was stable with the analysis of the mixture of the samples obtained daily in the morning and in the afternoon. Therefore, three samples per animal in each diet were used to quantify the marker and each sample was quantified in triplicate. The TiO 2 concentration was estimated by calorimetrically measuring absorbance at a wavelength of 410 nm. Likewise, a TiO 2 recovery test to confirm the quantification capacity of the laboratory technique (96.7-98.5% TiO 2 ) was carried out [18]. For the recovery percentage of the marker in the field, an average of 96% was assumed based on the previous reports [23,24].

INDF Determination
The iNDF was determined by means of the in vitro DaisyII technique (ANKOM Technology-08/05), in forage samples of the different treatments, which were collected before the animals were admitted to the pasture. Likewise, the iNDF was determined in the dried dung samples collected during the grazing of the animals within the period of the marker supply. 0.5 g of each sample was deposited in Fiber Filter Bags of 25 micron porosity (F57 filter bags), and incubated for 144 h in triplicate [25]. For the in vitro technique, 400 mL of ruminal fluid were collected from animals grazing on diets corresponding to the diet to be incubated. The DaisyII incubator consists of four jugs with a capacity of two liters each, where each jug was inoculated with the ruminal fluid (filtered twice with gauze) corresponding to each of the four animals that grazed the treatment. The culture medium for in vitro incubation consisted of buffer solution A (10 g L −1 KH 2 PO 4 ; 0.5 g L −1 MgSO 4 7H 2 O; 0.5 g L −1 NaCl; 0.1 g L −1 CaCl 2H 2 O; 0.5 g L −1 Urea), buffer solution B (15 g L −1 C 6 Na 2 CO 3 and 1.0 g L −1 Na 2 S 9H 2 O) and neutral detergent solution. In each digestion jar, solutions A and B were added in a ratio of 1:5 (1330 mL:266 mL) for a total volume of 1600 mL of combined A/B mixture in each digestion jar. After the incubation process was completed, the bags were rinsed in water and the NDF content was subsequently analyzed.

Forage Quality Parameters
The availability and collection of forage samples was made through the double sampling methodology [28] using a quadrant of 0.25 m 2 and 0.5 m × 1 m for pasture and shrub forages, respectively. The estimation of the Cayman/L. leucocephala ratio offered to animals was carried out using the agronomic method [16], where the difference between the amount of fodder offered and the amount remaining at the end of each grazing was quantified. On the other hand, the level inclusion of T. diversifolia was estimated by employing the difference between the amounts of fodder offered to animals and the amount remaining at the feeder at the end of each day. The green biomass supply based on the average forage production was 4,2-ton ha −1 for the B. decumbens per grazing period; 4,3-ton ha −1 for the Cayman in monoculture; 4-ton ha −1 of T. diversifolia and 8.7-ton ha −1 on the silvopatoral system (Cayman 4.5-ton, L. leucocephala 4.2-ton).
Forage samples were collected one day before the animals were allowed to graze of each treatment and were subsequently weighed and dried at 60 • C for 72 h in a forcedair oven to calculate the dry matter content (DM). The dried samples were ground in a 1 mm sieve, for later ash analysis (4 h at 500 • C; ( [29]: method 942.05)), crude protein (CP = N × 6.25; Kjeldahl AN 3001 FOSS; ( [30]: method 984.14)). Neutral detergent fiber (NDF) and acid detergent fiber (ADF) were determined using the methodologies proposed by Van Soest et al. [31], adapted to an Ankom Fiber Analyzer AN 3805 (Ankom ® Technology 2052 O'Neil Rd, Macedon, NY 14502, USA). Forage digestibility was determined as described by Goering and Van Soest [32]. Lignin was analyzed following the procedure described in ISO 13906:08, AOAC 973.18, AN 3804 ANKOM. The main forage nutritional components of diets are showed in Table 1. All laboratory analyses were developed at the Animal Nutrition and Tropical Forages Quality Laboratory of the International Center of Tropical Agriculture (CIAT).

Urine Volume and Nitrogen Balance
A manual collector was used to obtain urine samples from steers during grazing. During the first 12 days of grazing, the steers were accustomed to the manual urine collector and to the presence of humans. On days 13, 14, and 15 urine sub-samples (50 mL) were collected every four hours for a period of 12 h per day to calculate urine volume through creatinine quantification in urine (Creatinine-Kit Biosystems ® , Costa Brava 30, 08030 Barcelona, Spain) [33]. The sub-sample of urine was acidified with 0.5 mL sulfuric acid (1 M) and frozen at −20 • C. The dung samples collected for quantification of total excreta production (item 2.4.1) were also used for the quantification of total N and carbon content. The urine and dung N content was determined by Kjeldahl method [34]. The carbon content was calculated through a dry combustion technique. The nitrogen balance was calculated by the difference between the N intake and the N excreted in dung and urine.
All laboratory analyses were developed at the CIAT Forage Quality and Animal Nutrition Laboratory.

Rumen Fermentation Parameters
On days 0, 7 and 15 of each grazing event, a liter of solid and liquid content mixture of the rumen of each animal was collected through the fistula of four cannulated steers. The collection was carried out 4 h (at 10:30 am) after introducing the animals to a new grazing strip. The ruminal fluid was stored in thermoses preheated to 38 • C, which were transported to CIAT's Forage Quality and Animal Nutrition Laboratory for further processing. The content of each thermos was filtered using five layers of mesh and subsequently the pH was measured. Approximately 50 mL were stored in sterile falcon tubes at −80 • C for subsequent DNA extraction. On the other hand, 10 mL of ruminal fluid were stored in sterile falcon tubes at −80 • C, in order to quantify the ammoniacal concentration N (NH 3 ). Additionally, 50 mL subsamples of ruminal fluid (stored in falcon tubes) were centrifuged at 13,000× g for 30 min. From the resulting supernatant, 4 mL + 1 mL of 25 % metaphosphoric acid were taken. These samples were stored in sterile falcon tubes at −20 • C for subsequent volatile fatty acids (VFA) analysis. The VFA ratios were determined by high-performance liquid chromatography (HPLC; Shimadzu ® series 20 A, 7102 Riverwood Drive, Columbia, MD 21046, U.S.A.) equipped with a UV/VIS detector at a wavelength of 210 nm (SPD-20 AV), and a chromatography column (BIO-Aminex RAD HPX-87 H). The mobile phase was prepared for analysis with 0.005 M H 2 SO 4 with a flow rate of 0.7 mL min −1 and an injection volume of 20 µL. The molar proportions of volatile fatty acids were calculated by retention times and peak areas of commercial standards of acetic, butyric and iso-butyric acids (10, 100, 250, 1000, 1500, 2000 and 3000 ppm) and corrected by molar mass of each acid.

DNA Extraction from Rumen Digesta
The DNA was extracted using 1 mL of ruminal fluid [35]. Once the DNA was extracted, it was treated with RNase ( ® Fair Lawn, NJ, USA) according to the Joint Genome Institute: (i) RNase A treatment: bring the DNA volume to 200 µL in TE buffer (pH 8.0); add 10 µL of RNase A; incubate at 37 • C for 1 h; check small aliquot (5 µL) on an agarose gel with no treatment control; run gel 10-15 min. (ii) Ethanol precipitation: add 1/10 volume of 3 M sodium acetate to RNase A treated DNA; add 2.5 volumes of 100 % ethanol; mix and spin down simply; place at −80 • C for 30 min; spin sample at 4 • C for 20 min to pellet DNA; carefully pour off supernatant; wash pellet with 70% ethanol (cold); spin sample at 4 • C for 3-4 min; take out all ethanol with pipet tips; air-dry the pellet; resuspend pellet with 100 µL of TE. The concentration and purity of extracted DNA was measured with the Nanodrop 2000 spectrophotometer (Thermo Scientific ® , Wilmington, DE, USA). The DNA was diluted, and 20 ng were used as template in quantitative real-time polymerase chain reactions (qPCR). The DNA was stored at −80 • C.

Quantitation of Microbial Populations
SYBR ® Green-based real-time PCR was used for absolute quantification of Fibrobacter succinogenes, Prevotella ruminantium, Selenomonas ruminantium, Protozoa and total bacteria populations. The primer sequences used and the amplification conditions are described in Table 2. Each reaction mix was characterized by containing 10 µL SYBR ® Green (QIAGEN ® , 19300 Germantown Road, Germantown, MD 20874, USA), 1 µL of each primer, 2 µL of DNA samples plus 6 µL of ultrapure water for a total volume of 20 µL per reaction. Standard curves for each microorganism were generated with 101 to 108 copies of recombinant plasmids per µL from the cloning and transformation of competent cells of Escherichia coli, where the Promega commercial purification kit was used (Wizard ® Plus SV Minipreps DNA Purification System, Promega Corporation, 2800 Woods Hollow Road, Madison, WI 53711 USA). Number of copies of the DNA template were calculated using the equation developed by Faseleh et al. [36], and the absolute abundance was expressed as copies mL −1 of culture sample. Estimation of gene copy number for the samples was made from the linear relationship between the threshold amplification (Ct) and the logarithm of DNA copy numbers from the standard (R 2 0.991, with a mean efficiency 97 ± 1.2 %, and a slope value of 3.2).

Statistical Analysis
Following a Latin square design with crossover arrangement, with four diets and four periods, the variables dry matter intake, dung and urine production, N concentration in urine and dung, N balance, and volatile fatty acid production were analyzed, as well as the change in the abundance of ruminal microorganisms.
The variable dry matter intake, dung and urine production, N level in urine and dung, and N balance were evaluated through the following model (Equation (3)): where: The variables, volatile fatty acid production, and microorganism population were analyzed through the following model (Equation (4)): where: yijk = response of the steer k, under diet i and day j µ = the general mean δi = effect of the i-th diet (I = Cayman, Cayman + L. leucocephala, B. decumbens, B. decumbens + T. diversifolia) βk/i = effect associated with the steer k within the diet i Dj = day effect j (j = 0, 7, 15) (δD)ij = effect of diet i interaction and day j εijk = experimental error All data were subjected to analysis of variance and when significant treatment effects occurred, differences between means were evaluated using the Bonferroni test. Pearson correlation coefficients were determined to evaluate linear relationships between variables. Programing language R was used with package 'agricolae' for statistical analyses [40].

Dry Matter Intake and Nitrogen Excretion
The inclusion of L. leucocephala and T. diversifolia in pasture-based diets increased the total dry matter intake (p < 0.0001), and the silvopastoral system resulted in being the diet with the highest dry matter intake from the evaluated treatments. Intake was not different between monoculture grass diets ( Table 3). The inclusion of high protein species did not increase dung production compared to treatments in which the same pasture grass was included. However, the silvopastoral system generated a greater dung production with respect to the B. decumbens and B. decumbens + T. diversifolia diet. Nitrogen intake by steers on diet B. decumbens based increased when T. diversifolia was included in the diet, but there was no difference when steers were fed with Cayman in monoculture. The urine volume changed between diets with higher protein content led to higher urine synthesis ( Table 3). The N excretion in urine and dung changed across diets. The dry matter intake in the silvopastoral system generated higher N excretion in urine, while the diet B. decumbens + T. diversifolia generated a similar N excretion to the Cayman in monoculture. Between the two diets in monoculture, the B. decumbens diet excreted less N in urine and dung, but this diet offered less N as well. There was no difference in the N excretion in dung by the T. diversifolia and L. leucocephala diets that were pasture based. Steers receiving the Cayman diet had higher N excretion in dung ( Table 3).
The T. diversifolia + B. decumbens diet was associated with an increase in N balance compared to the B. decumbens. This same mixture led to greater N balance compared with the Cayman diet, even though N intake with these two diets was the same (Table 3). In terms of the N excreted/N ingested ratio, the silvopastoral system showed a higher N retention. Likewise, the balance (N intake-urine N-dung N) was significantly higher for the steers that grazed on Cayman associated with L. leucocephala. The pastures in monoculture showed a low N balance, and the N balance with B. decumbens was significantly lower Cayman. With both grasses there was improved N balance when a protein source was included (Table 3). Table 4 presents the Pearson correlations between variables related with N flows. Strong correlations were observed among a group of variables, including dry matter intake, N intake, urine volume, N urine and N balance. The urine volume was correlated with dry matter intake, N intake and N balance ( Table 4). The N intake was correlated with dry matter intake (0.76 p < 0.001) and the same time, and the N urine and N balance were highly correlated with N intake (0.95 p < 0.001; 0.96 p < 0.001). The N balance was also correlated with urine volume (0.82 p < 0.001) and N urine (0.87 p < 0.001). The N dung was correlated only with dung production (0.6 p < 0.05), while this was correlated only with the N intake (0.65 p < 0.05).

Fermentation Parameters and Microbial Populations
None of the diets evaluated affected either the pH or the production of propionic, butyric and isobutyric acid in the rumen (Table 5). On the contrary, the N-NH 3 production, acetic acid, and the acetic:propionic ratio were affected by the different diets and an interaction between measurement time and diets was observed. The N-NH 3 production in the B. decumbens diet decreased at days 7 and 15, with the N-NH 3 value at day 15 being lowest compared with the other diets on day 15. When T. diversifolia was combined with B. decumbens, the N-NH 3 level of the diet increased on day 7 with no difference to day 15. The Cayman-alone diet had an increase on day 7 and remained unchanged on day 15. The concentration of N-NH 3 in the rumen was greater with time in the silvopastoral system diet, showing higher N-NH 3 production at day 15 compared with the other diets on the time point (day). The inclusion of T. diversifolia to the B. decumbens diets reduced the acetic acid levels on days 7 and 15 compared with the B. decumbens-alone diet and presented similar levels to the Cayman-alone diet. On day 15, the silvopastoral system diet generated higher acetic acid levels in the rumen compared with the other diets ( Table 5).
The acetic:propionic ratio was lower in the B. decumbens + T. diversifolia and Caymanalone diets, while it was higher under the silvopastoral system and B. decumbens diets ( Table 5).
The inclusion of shrub forages high in protein generated an increase in the gene copy numbers of Prevotella ruminicola and total bacteria on day 15 of sampling ( Figure 1). Meanwhile, with the B. decumbens diet, the population of Prevotella ruminicola was reduced on day 7 of measurement. However, animals grazing on the B. decumbens-alone diet did not show changes in the abundance of total bacteria during the different sampling days. On day 7 of measurement, the total bacterial population decreased compared to the start day of the experiment under the Cayman monoculture diet but increased on day 15 of measurement.  The protozoa population was different in each sampling period, where in all diets there was an increase on the last time point (day 15) of measurement. However, there was no significant difference in the protozoa abundance between different diets. Neither the diets nor the measurement periods affected the population of Fibrobacter succinogenes and Selenomonas ruminantium. (Figure 1, Table 6).

Dry Matter Intake
Cattle systems in the tropics are characterized by meat and milk production under the use of monoculture pastures of low nutritional value, affecting diet consumption parameters and efficiency in the use of the nutrients consumed. In the present study, the inclusion of species such as T. diversifolia and L. leucocephala in diets based on improved pastures contributed to higher food intake and greater N retention. The steers increased their dry matter intake when the diets offered included species with a high protein value and low NDF content, thus achieving an average intakes of 3% and 3.2% of live weight in the B. decumbens + T. diversifolia and in the Cayman + L. leucocephala diets, respectively. These results are in line with previous information reported by other authors [41], who observed higher intake levels as the CP level increased in the diet and aligned with other studies that reported increased intake in diets with added L. leucocephala [42,43].
The inclusion level reached in the T. diversifolia diet was 32.7%, while the L. leucocephala reached an 18.4% inclusion level. These results are consistent with other studies that have reported levels of inclusion of T. diversifolia between 15% and 35% in grazing of lactating cows, without altering the animals' voluntary intake [10,13]. The NDF and ADF The protozoa population was different in each sampling period, where in all diets there was an increase on the last time point (day 15) of measurement. However, there was no significant difference in the protozoa abundance between different diets. Neither the diets nor the measurement periods affected the population of Fibrobacter succinogenes and Selenomonas ruminantium. (Figure 1, Table 6).

Dry Matter Intake
Cattle systems in the tropics are characterized by meat and milk production under the use of monoculture pastures of low nutritional value, affecting diet consumption parameters and efficiency in the use of the nutrients consumed. In the present study, the inclusion of species such as T. diversifolia and L. leucocephala in diets based on improved pastures contributed to higher food intake and greater N retention. The steers increased their dry matter intake when the diets offered included species with a high protein value and low NDF content, thus achieving an average intakes of 3% and 3.2% of live weight in the B. decumbens + T. diversifolia and in the Cayman + L. leucocephala diets, respectively. These results are in line with previous information reported by other authors [41], who observed higher intake levels as the CP level increased in the diet and aligned with other studies that reported increased intake in diets with added L. leucocephala [42,43].
The inclusion level reached in the T. diversifolia diet was 32.7%, while the L. leucocephala reached an 18.4% inclusion level. These results are consistent with other studies that have reported levels of inclusion of T. diversifolia between 15% and 35% in grazing of lactating cows, without altering the animals' voluntary intake [10,13]. The NDF and ADF values reported in this study for T. diversifolia are below those reported in other studies [44] (NDF: 53%; ADF: 48.18%) and above those reported by other investigations [45] (NDF: 33.35% and ADF: 12.21%) for similar regrowth ages (56 and 60 days), and establishment of the shrubby cuttings. Both investigations were carried out in different environmental conditions than those of this study (2456 m.a.s.l and 2905 m.a.s.l, respectively). Chemical composition of this shrub fluctuates in response to altitude and, according to other authors [46], the composition of T. diversifolia is mainly related to factors such as soil fertility, environmental temperature, and the level of solar radiation, which can influence the plant maturation process and structural carbohydrates. This composition of the T. diversifolia did not represent a limiting factor to forage intake with respect to the B. decumbens and Cayman diets.
The L. leucocephala intake values reported in our study are consistent with those reported [47]. A diversified supply of fodder stimulates voluntary intake in ruminants during grazing, since they can select and consume plant materials of the best nutritional quality [15]. It has been reported that increases in voluntary intake by providing L. leucocephala in the diet is associated with a higher DM rate degradation and a higher passage rate [47]. In our study, the forages offered in the Cayman grassland, by including L. leucocephala in rows, was 1.7 times greater (2591 kg DM ha −1 ) than in the Cayman grassland and B. decumbens in monoculture (1513 kg DM ha −1 ; 1481 kg DM ha −1 , respectively), as well as 1.3 times greater than in B. decumbens + T. diversifolia grassland (1965 kg DM ha −1 ). Likewise, the DM and CP L. leucocephala content values are consistent with the quality parameters referenced for this type of legume [48] and with those reported by other studies under similar conditions of the tropics [49,50]. However, the NDF and ADF values found in our study for both L. leucocephala and T. diversifolia are above those reported in other studies at similar sampling ages [43,45,49,50]. According to several studies [51,52] high levels of fiber constitute physical factors that limit the DM intake and degradation rate. However, similar values of NDF and ADF in L. leucocephala to those found in our study have been reported [47], and when included at 20% in the diet did not generate any effect on the DM intake. However, the same level generated a reduction in NDF digestibility. In our experiment, the greater protein contribution in the diet by the L. leucocephala inclusion favored the protein degradation, in part due to supply of adequate energy levels at the rumen, not limiting total DM degradation and DM intake. However, individually, both L. leucocephala and Cayman grass presented acceptable in vitro DM and NDF digestibility values.
The inclusion of L. leucocephala in silvopastoral systems contributes to improving soil fertility and, therefore, pasture productivity through biological N 2 fixation [53]. At the same time, other studies [12,54] indicate that the arrangement in rows can generate a better distribution of the N 2 fixed by L. leucocephala, as well as provide better microclimatic conditions where the grass is less exposed to direct solar radiation and wind effects. The above contributes to the increase in the nutritional quality of the grass in terms of CP and gross fiber [49]. In our study, no differences were found in the nutritional quality of Cayman associated with L. leucocephala, compared to the Cayman in monoculture. This may be due to the L. leucocephala low planting density (1428 seedlings ha −1 ), and to the early age of the crop (one year), due to a not yet significant N contribution by the legume to the grassland, as well as a less homogeneous N distribution in the pasture. The N levels are significantly higher below the rows of L. leucocephala (0.9 t N ha −1 at 0.2 m deep), being 27% more than in grassland [55].
On the other hand, in our study the soil was characterized as a compact soil of low aeration. Additionally, although the soils in this study corresponded to neutral soils [56], and phosphorus fertilization was performed, the P levels in the soil were low compared to other studies in plots established in the same research center [57], and which indicate a moderate P deficiency (5-15 mg kg −1 ) [56]. P deficiency in the soil can directly affect L. leucocephala growth, limit nodular development, and therefore have an indirect effect on the symbiotic N 2 fixation [58]. Maybe, in our study, these physical and chemical characteristics would influence the root development of the legume and a higher planting density will be needed to generate a greater impact on the physical soil structure and the nutritional grass quality associated.
The DM and CP levels found in our study for both plots of B. decumbens are consistent with those reported by other authors at similar ages and under tropical conditions [41]. However, CP levels are low, especially in consideration of the high NDF content found. Therefore, energy may have been a limiting factor to ensure an intake level for these pastures. In general, the growth of grasses in all plots was slow and this led to long resting periods (50 days), which could explain the high NDF values reported in this study. Inclusion of both protein sources in this study generated a greater volume of urine, which is consistent with reports by other authors, who attributed this effect to the greater N ingestion [59], given the increased water intake that is required to digest the proteins and for N metabolism. Furthermore, the contribution of N supplied with the diet of the silvopastoral system, in turn, increased the concentration of N-NH 3 in the rumen and at the same time increased the amount of urine N compared to the other treatments. However, the daily contribution of CP for treatments that included species with high protein value was 84 and 147 g kg DM −1 . These CP values do not exceed the range of requirements for ruminants (100 to 170 g kg DM −1 ; [8]). In this sense, the total amounts of N excreted per N consumed is low. Therefore, the inclusion of T. diversifolia and L. leucocephala in the diet leads to better dietary N use and, in turn, can improve the use of the low-quality forage and lead to better performance in terms of live weight gain [13,16,60,61]. Additionally, other studies have reported that the exponential increase in N excretion through urine is generated with N intakes above 400 g d −1 [62,63]. In our study, the highest N consumed corresponded to 228 g d −1 , and the N excreted/N consumed ratio was 61% and 42% for the T. diversifolia and the silvopastoral system, respectively. Therefore, our results suggested that the mixture of Cayman + L. leucocephala has the potential to favor N retention and thus favor DM degradation under tropical conditions. Finally, this can contribute to improving N use efficiency at the ruminal level [64].
Use of the silvopastoral system and the inclusion of T. diversifolia in the B. decumbens diet generated a lower N excretion in dung compared to the monoculture diets. This is consistent with several studies [62,65] where no direct relationship was found between N ingestion and N excretion in dung. These results suggest that protein forages could have contributed to better amino acid digestibility at the intestinal level since the fecal N corresponds largely to the undigested N of the food [66].
The B. decumbens diet in monoculture had the lowest N retention, which is characteristic of the low-quality diets used in the tropics [67]. The N excretion in dung and urine as a proportion of the ingested N decreased with T. diversifolia inclusion, generating a better N balance (Table 3). These results are consistent with data reported by other authors [68], who found a higher percentage of N retention (35.5%) resulting from the inclusion of 35% T. diversifolia with respect to the 100% grass-based diet. Likewise, other studies found that N excretion in dung, as a proportion of the N ingested, showed linear and quadratic reductions [69].

Ruminal Parameters
In our study, rumen fluid pH values remained within an optimal range for the degradation of dry matter, ranging between 6.3 and 6.7, without being affected by the different treatments. These results are consistent with other studies where animal diets are based on forages [70]. Likewise, in another study where the inclusion of different CP levels were evaluated through the supply of L. leucocephala flour, pH levels were not affected [71]. On the other hand, the ruminal N-NH 3 concentration was affected by the different diets; where the inclusion of protein forages increased the amount of N-NH 3 in the rumen over the days, the silvopastoral system diet presented the larger increase on the 15th day of measurement. The minimum required N-NH 3 concentration for maximum microbial growth in the rumen is 50 mg L −1 [72]. Likewise, another study reported that levels of 50-80 mg L −1 appear to be sufficient to favor fiber digestion, because when they obtained higher N-NH 3 levels they did not significantly increase the degradation characteristics of food in the rumen [73]. According to this information, the N-NH 3 levels reported in our study under L. Leucaena in the Cayman diet would be meeting the minimum requirement at the ruminal N level, which could be used in the synthesis of microbial proteins at the level ruminal [74]. In the same way, the N-NH 3 concentration in the rumen is a gross predictor of the efficiency of the dietary N conversion into microbial N [73]. Although this same diet has generated a higher concentration of urine N, it also generated a greater N retention. Despite this, it is not possible to infer that under the L. Leucocephala inclusion in the diet it has contributed to efficient microbial protein synthesis, since predictive factors such as the DPR and UDPR level, as well as the passage rate, were not quantified in this study.
The production of VFA in the rumen is influenced by dietary carbohydrate fraction and degradability. In our study, the molar acetate concentration was higher with the L. leucocephala inclusion, which could relate to a greater CP availability in the diet, since the CP promotes the growth of ruminal bacteria and with it greater acetate production arising from cellulose degradation [75]. Similar results were reported in the other studies, where an increase in the protein level supplied to lambs increased acetic acid concentration and the acetic:propionic ratio, as well as higher N-NH 3 concentration [76]. In this sense, in our study, with the inclusion T. diversifolia in the diet of B. decumbens, a greater total bacteria production was found at the 15th day of measurement. This was followed by the diet including L. Leucocephala, both of which were the diets with greater CP contribution supplied in this trial. These results are consistent with those reported in other studies [77,78], where a greater total bacteria production in the rumen was evidenced by increasing CP in the diet. On the other hand, L. leucocephala has a non-protein amino acid called mimosine (α-amino, 3-hydroxy-4-pyridinepropanoic acid). This component fluctuates between 40 to 120 g kg DM −1 according to its physiological state and plant fraction [11] and, during ruminal fermentation, can contribute to the increase in acetate production as well as to an increase in the N-NH 3 content and organic matter degradation [79]. On the other hand, if the higher N-NH3 level found in this study under the silvopastoral system would influence the higher growth of ruminal microorganisms, would be promoting cellulose degradation [80], which would contribute to acetic acid synthesis. However, the acetic:propionic ratio in the ruminal fluid was higher in B. decumbens grass in monoculture, but this diet presented quantitatively lower production of propionate.
Other populations quantified in our study, such as F. succinogenes, S. ruminantium, and Protozoa, were not significantly affected by the different diets (p > 0.05), although the Protozoa populations showed a significant increase during 15th day of measurement under the T. diversifolia inclusion in the B. decumbens-based diet (p < 0.001). This result could trigger higher emissions of enteric CH 4 , since, methanogens associated with ciliated protozoa are responsible for 9% to 37% of methane production in the rumen [79]. Therefore, by increasing the population of protozoa, the activity of the methanogenic bacteria associated with them could be increased. These results are contrary to those reported by various studies where T. diversifolia was included and reduced the protozoa population significantly in in vitro conditions [81,82], which is mainly attributed to the content of secondary metabolites such as condensed tannins (CT) and total phenols [83][84][85]. However, in another study that evaluated T. diversifolia between 30 and 60 days of regrowth, the presence of CT and total phenols was low, both in leaves and in the mixture of leaves and stems, mainly finding alkaloid type metabolites [86].
On the other hand, the presence of CT in the plant negatively affects its palatability, thus reducing DM intake [87]. However, in our study, the results indicate an increase in the DM intake with respect to the diet offered in B. decumbens and the Cayman in monoculture, despite there being high levels of the shrub in the diet (35%). Therefore, according to the previous studies, it may be that the concentrations of CT and phenols in the ration were not high enough to generate an effect in reducing the population of protozoa. In contrast, the diet offered has high levels of fiber and did not show variation in pH, which could be a reason to stimulate the population of protozoa in the rumen [88,89].
Better physical and chemical characteristics of the soil than those observed in this study are necessary to improve the performance and the nutritional quality of B. decumbens grass, and thereby lead to better results on ruminal parameters when this grass is associated with protein forages such as T. diversifolia. Likewise, the concentration of tannins and phenols in the plant depends on edaphoclimatic factors [83]. Therefore, further research is necessary to evaluate the tannin concentration in T. diversifolia and its relationship with the population of protozoa and methanogenic bacteria, to identify the role of this shrub more accurately on the population of protozoa and methanogenic bacteria in the rumen.
Prevotella ruminicola 23 is one of the most abundant and important microorganisms for ruminal fermentation; 50% of the population grows well under cultivation in the lab and has a high frequency within the 16S gene [37,38]. Its population was affected by the inclusion of protein forages in the diet, showing an increase on the 15th day of measurement, being slightly higher with the T. diversifolia inclusion for the same period. This result suggests that the diet is providing enough energy for multiplication and microbial growth [34]; this benefit is only evidenced up to 15 days after consuming protein forages (T. diversifolia and L. leucocephala). This result is consistent with those reported by another study [76], which found greater abundance of Prevotella when lactating cows were fed legume hay (Medicago sativa). Likewise, other authors found that the total count of viable bacteria, cellulolytic bacteria, and proteolytic bacteria counts increased when L. leucocephala leaf meal was supplied without thermal processing [71].
The higher acetate production on the 15th day of measurement found that L. leucocephala may be related to the greater abundance of Prevotella, since acetate corresponds to one of its main fermentation products [90]. However, it is known that L. leucocephala has secondary compounds such as condensed tannins whose total concentration can fluctuate between 1.8-4% [91] and can affect the DM intake [50], as well as causing rumen defaunation [92]. In our study, the tannin content in L. leucocephala was not quantified; therefore, the level of this secondary metabolite ingested by the animals during this test is not known. However, under the inclusion L. leucocephala level (18%) obtained, neither the intake parameters nor the total protozoa population were affected. Likewise, in a study where the CT effect extracted from L. leucocephala (CIAT 734) and Desmodium ovalifolium on the parameters of fermentation and cellulolytic bacteria was evaluated, L. leucocephala had a lower effect on digestibility, fibrolytic population bacteria, and total AVF production [93].

Conclusions
The silvopastoral system, with the inclusion of 18% L. leucocephala as a fresh matter basis, provides a source of protein widely accepted by cattle, which led to increased DM intake, greater rumen total bacteria populations and greater N retention in steers. This is a superior advantage for tropical beef production systems, given that greater retention of dietary N may contribute to improving productive parameters such as live weight gain, thereby reducing age at slaughter. This has environmental implications, as reduced grazing time would lead to both reduced lifetime of enteric methane emissions and reduced excreta deposition to the soil, also a major source of nitrous oxide emissions from pastures.
The N-NH 3 rumen concentration in this study when feeding steers with T. diversifolia showed the potential for increased rumen protein degradation, and thus provided the N necessary for the generation of rumen microbial protein.
The incorporation of different species and plant strata in livestock feeding constitutes an effective strategy to generate economically efficient beef production systems with low environmental impact (i.e., GHG emissions). Further studies of the potential of plant protein sources as only food in livestock feeding should undertake to contribute to the environmental sustainability of beef and milk production systems.