Introduction

The low profitability of production systems in tropical regions is due, in part, to livestock activities that depend on the use of grasses as the main source for animal diets (^{Ku-Vera et al.,2013}). Grasses under conditions of monoculture present a low content of digestible protein and high levels of fiber throughout the year (^{Carmona, 2007}). The association of trees, shrubs and forage grasses allows improving the quality of the livestock diet, increases the productivity of agricultural/livestock systems, allows savings in the use of nitrogenous fertilizers, and maintains the proportion of botanical components in space in time (^{Merchant and Solano, 2016}).

Forage production of Cynodon nlemfuensis in monoculture can reach up to 26.9 Mg DM ha^-1 year^-1 (^{Villalobos and Arce, 2013}), and when it is cultivated in association with Leucaena leucocephala it can produce 18.5 Mg DM ha^-1 year^-1 (^{Palma, 2005}). In association of grasses with trees, the availability of forage increases resulting from more nitrogen in the soil under the trees (^{Maya et al., 2005}).

These differences are relevant due to the interest for converting livestock systems into more intensive systems, which allow higher productivity, competitiveness and sustainability, increases. Silvi-pastoral systems are a good option to achieve sustainable livestock production and are characterized by high densities of forage trees (more than 10 000 plants ha^-1) and associated grasses (^{Barros-Rodríguez et al., 2012}; ^{Gaviria-Uribe et al., 2015}). In addition, the plants that grow under shade suffer morphological changes as mechanism of adaptation to the system (^{Sousa et al., 2010}, 2012; ^{Santiago-Hernández et al., 2016}). Under these conditions, ^{Obispo et al. (2008)} found that the high level of shade has a negative effect on yield of Panicum maximum biomass.

The forage shrub Tithonia diversifolia presents accelerated growth, abundant leaf production (^{Pérez et al., 2009}), foliage with 18.52 to 24.13 % of CP, 32.94 to 38.62 % of NDF and 10.33 to 34.48 % of ADF (^{Nieves et al., 2011}; ^{Gallego-Castro et al., 2014}). In monoculture, 23.48, 17.66 and 21.8 Mg DM ha^-1 year^-1 were obtained (^{Gallego-Castro et al., 2014}).

The Gliricidia sepium foliage contains 24.11 % of CP and 38.81 % of NDF (^{da Costa et al., 2009}); it is used as food during the dry period and can substitute concentrate as supplement for livestock food (^{Carmona, 2007}). In Yucatán, with G. sepium in monoculture and cutting at 45, 60 and 75 d, 2.43, 3.5 and 2.26 Mg DM ha^-1 were obtained (^{Ramos et al., 2016}).

The productive behavior in monoculture for the species mentioned is known, although there is no information in agroforestry system (AFS). The objective of this study was to determine the forage production, the percentage of forage components, and the nutritional quality of C. nlemfuensis, T. diversifolia and G. sepium biomass in monoculture and agroforestry system, in Yucatán, Mexico. The hypothesis was that the association of forage species with different growth habit, C. nlemfuensis, T. diversifolia and G. sepium, increases the availability of forage, modifies the percentage of forage components and the quality of the edible biomass in the system throughout the year.

Materials and Methods

The study was performed over a year, in the Technological Institute of Conkal, Yucatán, Mexico (20º 05’ N and 89º 32’ W). The climate of the region is warm sub-humid with summer rains (Aw₀), with mean annual temperature of 26.5 °C, and total annual precipitation of 900 mm and 9 masl (^{INEGI, 2009}). The climate conditions during the evaluation period, January 2005 to May 2006, were recorded (Figure 1). The predominant soils are lithosols with pH 7 to 8 (^{Bautista et al., 2015}). The design was completely random unbalanced with six repetitions in each monoculture and five for AFS; the treatments were: C. nlemfuensis (MCn), T. diversifolia (MTd) and G. sepium (MGs) in monoculture, and AFS in which the three species were associated.

Figure 1 Distribution of the precipitation and average temperatures in the study area during the experimental period (January 2005 to May 2006). 

There were 23 experimental plots, six for each monoculture and five for AFS, established on lithosol soil. The C. nlemfuensis (MCn) plots in monoculture were 16.0 m², divided into four sub-plots of 4.0 m² with a useful area of 1.0 m² in each; the monoculture of T. diversifolia (MTd) was established in a sowing pattern of 1.0 x 1.0 m with 6.25 m² of useful area per plot (10 000 plants per ha), and six plants per plot from the central furrow were evaluated. The experimental units of G. sepium (MGs) monoculture were established with sowing density of 1.0 x 1.0 m between plants and furrows (15 m² plots with 24 plants per plot and useful area of 6.0 m² with six plants); three plants were evaluated randomly and the yield per ha was calculated based on the sowing density. Five plots of 90 m² were established with AFS in a pasture of C. nlemfuensis; each plot had 15 plants of G. sepium, with sowing density of 3.0 x 2.0 m between furrows and plants, 18 plants of T. diversifolia interspersed between the lines of G. sepium at 3.0 x 1.0 m, and C. nlemfuensis was present in the whole area. The plants evaluated were three of G. sepium and 12 of T. diversifolia and C. nlemfuensis from the useful area (15 m²). Before the evaluations, a cut was done to make uniform the experimental plots; both cultivation systems had been established for two years. During the study, the plots were cleaned through manual weeding.

The forage yield (Mg DM ha^-1) and the percentage of leaf (L), stalk (S) and dead tissue (DT) were quantified for C. nlemfuensis in monoculture and in AFS. The samples were harvested at less than 5 cm of ground, every 30 d in the rainy season, and every 42 d with north winds and dry season. The green forage was weighed, mixed, and two subsamples of 100 g were separated randomly, one to determine the yield (Mg DM ha^-1) and another to obtain the fodder components.

In T. diversifolia the plants were harvested at 0.5 m from the ground and G. sepium at 1.0 m in monoculture and in AFS. The fresh biomass was weighed, and then two subsamples of 1 kg were separated to determine the yield and fodder components. The subsamples were dried for 48 h at 55 ºC in a stove with forced air circulation and were weighed to estimate the yield (Mg DM ha^-1).

The forage subsamples from each species in monoculture and in AFS were pulverized with a mill (Thomas Wiley®) with 1.0 mm sieves. The total content of N was determined with the Kjeldahl method and CP content was calculated with the conversion factor of 6.25 (^{AOAC, 2016}), and the NDF content was also defined (^{Van Soest et al., 1991}).

The yield (Mg DM ha^-1) and fodder composition, expressed as percentage of leaf (L), stalk (S) and dead tissue (DT), content of CP and NDF were analyzed with Mixed models (^{SAS, 1999}), which included the treatment as fixed effect, and the sampling plots as random effect and Tukey (p ≤ 0.05) means comparison.

Results and Discussion

The global forage yield (p≤0.05) in AFS was higher than MCn, MTd and MGs. The higher availability of biomass in AFS was in SATd (76.5 %), then SACn (12.8 %) and SAGs (10.7 %) (Figure 2). Between systems, the forage production of MCn was higher than SACn (p ≤ 0.05), SATd was higher than MTd (p ≤ 0.05), but MGs and SAGs were similar (p ≤ 0.05).

Figure 2 Forage yield of Cynodon nlemfuensis (MCn and SACn), Tithonia diversifolia (MTd and SATd) and Gliricidia sepium (MGs and SAGs) in monoculture and agroforestry system, respectively (mean ± standard error). 

In August, February and May a higher forage productivity (p ≤ 0.05) was observed in SAGs and SATd, and the lowest production of biomass was with MGs and SAGs; in August, November and May the productivity of MCn was higher (p ≤ 0.05) than SAGs, SATd, SACn, MGs and MTd (Figure 3). The association of T. diversifolia and G. sepium with the other species did not modify production, except the yield of C. nlemfuensis which decreased, probably because of low tolerance to shade (^{Obispo et al., 2008}; ^{Santiago-Hernández et al., 2016}) and from competition over nutrients, water and light, which interfere directly with the photosynthesis of the grass (^{Moreno, 2008}).

Figure 3 Productive behavior of Cynodon nlemfuensis (MCn), Tithonia diversofolia (MTd) and Gliricidia sepium (MGs) in monoculture and in agroforestry system (AFS)(mean ±standard error). 

A similar behavior was observed in other native grasses under the top of Acacia nilotica, Dalbergia sissoo and Prosopis juliflora (^{Kaur et al., 2002}). The lower yield of C. nlemfuensis in AFS could result primarily from the age, frequency and height of cutting. The rate of dry matter accumulation increases with age, frequency and height of cutting of the grass. This agrees with a report for alfalfa that the frequency of cutting modifies the productive behavior (^{Ventroni et al., 2010}). In the stoloniferous species, biomass production is due to the elongation of leaves and stalks, which vary in function of the age of regrowth and season of the year. In the Brachiaria hybrid, the biomass accumulation increases with older regrowth, from a grazing frequency of 14 to 28 d during the rainy season (^{Cruz-Hernández et al., 2011}).

The cultivation system did not affect the production of forage components in any of the species (p > 0.05), although in G. sepium and T. diversifolia, the availability of edible material was higher in both cultivation systems (Figure 4). That is, both species can be considered for the design of agroforestry systems.

Figure 4 Percentage of forage components of Cynodon nlemfuensis (MCn and SACn), Tithonia diversifolia (MTd and SATd), and Gliricidia sepium (MGs and SAGs) in monoculture and in agroforestry system, respectively. 

These results are similar to those reported in grass production systems in monoculture and associated to trees, in which the production of L and S is similar in both systems (^{Ramírez et al., 2003}). Although the cultivation system did not change the production of forage components, T. diversifolia may be an option with high leaf availability and low proportion of S and DT. This can be due to its physiological adaptability to the cutting frequency of 42 d, because at younger age of cutting the production of L and tender S is higher. This behavior was observed in species with similar growth, such as Morus alba and Cajanus cajun (^{Martínez et al., 2002}).

The high production of leaves and low production of stalks in G. sepium could be attributed to the age of cutting, because this production can be better with cuts every 60 d (^{Ramos-Trejo et al., 2016}), which differs from the results of our study. By increasing the age of regrowth, the proportion of edible material decreases, but the total production of biomass increases regardless of the time of the year. This is similar to what was observed by Angulo et al. (2005) in Acacia mangium which highest biomass production was with a frequency of 80 d, in the dry and rainy seasons, with cutting frequencies of 40, 60 and 80 d, and independent of the season.

The CP content in C. nlemfuensis was similar in monoculture and AFS (p > 0.05). The cultivation system did not affect the NDF content in any species (Figure 5). In G. sepium and T. diversifolia there were high values of CP in both cultivation systems compared to the CP in C. nlemfuensis. The CP and NDF contents of the study were within the interval reported for G. sepium and T. diversifolia (^{Verdecía et al., 2011}; ^{Ramos et al., 2016}).

Figure 5 Forage quality of Cynodon nlemfuensis (MCn and SACn), Tithonia diversifolia (MTd and SATd) and Gliricidia sepium (MGs and SAGs) in monoculture and in agroforestry system. 

In our study, there was no effect of the cultivation system on the CP content in the grass. This agrees with a report by ^{Medinilla-Salinas et al. (2013)} for Megathyrsus maximus, in which the shade did not change the quality of the biomass; however, the quality of the biomass improved under the tree shade in a study of the effect of Melia azedarach L. on the physiology, production and forage quality of Megathyrsus maximus cv. Tanzania and cv. Mombaza, and the Urochloa Oaxaca and Yacare hybrid (^{Santiago-Hernández et al., 2016}). The CP content in the grass in our experiment differed from the report by ^{Obispo et al. (2008)} that the CP content increases in the grasses with shade.

However, the results of G. sepium and T. diversifolia agree with some intervals of values from the same variables recorded by ^{Ramírez et al. (2010)} and ^{Ramos et al. (2016)} that both species have a higher quality than the minimum necessary in the forages for ruminant diets (7 % of CP). The lower values reduce the microbial activity of the rumen because of the lack of N for ruminal microflora (^{Ku et al., 2013}). The results from the analysis of the whole plant (mixture of stems and leaves) could have an effect, because in the stems the conversion of photosynthetic products into structural components is fast, and a decrease in nitrates, proteins and soluble carbohydrates takes place, with an increase of the structural components of the cell wall (^{Soliva et al., 2008}). The results with Morus alba confirm this, since when increasing the age of cutting from one to six months, the L:S rate was increased, indicating an increase in the production of stems with older age of cutting (^{Kabi et al., 2008}). The NDF content in monoculture and ASF was similar and agrees with reports in the literature about tree species similar to G. sepium where there are not large variations of NDF from the effect of regrowth age. The values fluctuate between 31 and 44 % of NDF (Rivera et al., 2004), in star grass with cutting frequency of 28 d there is 76.9 % of NDF (^{Johnson et al., 2001}) and in association of C. nlemfuensis with L. leucocephala there is 73 % of NDF (^{Maya et al., 2005}).

Conclusions

The use of AFS established by C. nlemfuensis, associated with G. sepium and T. diversifolia, produced higher yield of forage in conjunction, without affecting the proportion of the forage components (leaf and stem) or the nutritional quality of the forage from the species.

However, in AFS the forage yield of C. nlemfuensis decreased and the yield of G. sepium changed, although the forage yield of T. diversifolia increased. Precaution is recommended when incorporating species in the design of an agroforestry system, for it requires identifying the species adequately to obtain an optimal result.