Introduction

In Durango, Mexico, grass species and varieties -with higher forage productivity and quality than those of currently used grasses- are needed to meet the forage demand mentioned in the State’s cattle inventory, particularly during the dry season (from winter to early summer) (^{Rosales et al., 2016}). In Durango, the evaluation of species has been taken up again, for grasses with the following characteristics: forage productivity (˃27 t of green forage per ha per harvest; ^{INFOSIAP, 2017}) and high quality (crude protein ˃12 %, neutral detergent fiber (FDN)=55 % and digestibility ˃50 %) (^{Ball et al., 2001}). This is the case of annual ryegrass (Lolium multiflorum), perennial ryegrass (L. perenne), and prairie grass (Bromus willdenowii), mainly of the Matua variety (^{Jiménez et al., 2014}).

Perennial and annual ryegrass are two of the most frequently used species in Durango (^{Reyes et al., 2000}). Evaluating grass adaptation, forage quality, and protein content in Durango is important, because protein is the most expensive ingredient of the supplements fed to cattle during their growth and final stages. Using different varieties and mixtures of grasses with high protein content can reduce the costs of feeding growing cattle which require high doses of supplementary protein (16.0 %) (^{NRC, 2001}). During the dry season, cattle are mainly fed low protein forage. In order to select better forage, the quality (chemical composition and digestibility) and its components (raw protein, NDF, ADF, lignin, and dry matter digestibility) must be analyzed. The forage quality analysis determines its nutritional value, based on its protein, fat, ashes, and structural component content (^{Van-Soest et al., 1991}), as well as its digestibility (^{Giraldo et al., 2007}).

Forage digestibility diminished as temperature increased and plants ripened (^{Wilson and Minson, 1983}; ^{Bal et al., 2001}), as a result of changes to the blade-stem ratio and the increase of senescent tissue (^{Wilson et al., 1986}). Hence, forage digestibility is usually higher during winter and gradually diminishes in spring and summer (^{Mueller and Orloff, 1994}).

Forage yield and quality depend on the phenological stage at harvest time, botanical composition of the population sown, soil fertility, weather, and agronomical crop management (^{Ball et al., 2001}; ^{Aganga et al., 2004}). In Durango, the aim is to make the most of the balance between biomass productivity and forage quality of grasses (particularly, protein content). The objective of this research was to evaluate the forage productivity and quality of several varieties of grass and of a commercial mix of the grasses grown during the dry season in Durango. The hypothesis of this study was that grass varieties and mixes have similar forage productivity and quality.

Materials and Methods

The study was carried out for two consecutive years. The sowing was carried out in the 2013-2014 autumn-winter cycle (Cycle 1) and in the 2014-2015 autumn-winter cycle (Cycle 2), on the first and the second year, respectively. Hercules and Maximus grass of annual ryegrass (L. multiflorum), a commercial mix of perennial ryegrass (L. perenne; perennial green mix), and Matua prairie grass were the varieties used as treatments. In both cycles, grasses were sown on October, 4^th in the Campo Experimental Valle de Guadiana, INIFAP, Durango, located in km 4.5 carretera Durango-El Mezquital 23° 59’ 21” N, 104° 37’ 33” W and 1877 m height).

The soil type was sandy loam, with a 30 % field capacity, 16 % permanent wilting point, 1.3 g cm^-3 apparent density, intermediate humidity retention capacity, average depth, 0 to 2 % slope, 7.9 pH, and low organic matter, phosphorus, and nitrogen content. The regional weather is temperate, semiarid, and extreme; it has an annual rainfall of 476 mm -mainly between June and September (^{Medina et al., 2005})-; and an annual average temperature of 17.4 °C. Its weather formula was BS₁ kw (w) (e) (^{García, 1987}).

In Cycle 1 of each treatment, the sowing was carried out in 100 m long and 10 m wide paired strips (1000-m² experimental plot), with rows of plants every 15 cm and a sowing density of 35 kg of seed ha^-1. An Aitchinson GrassFarmer 1414C^® sowing machine was used for this purpose. The 180-60-00 doses (N-P₂O₅-K₂O) were distributed in three applications: at sowing (60-60-00), 94 days after the sowing (DAS) (60-60-00) and 168 DAS (60-00-00). Besides rainfall (102 mm) (Table 1), nine auxiliary irrigations were applied throughout this study. A 2, 4-D (dimethylamine salt of the 2, 4-dichlorophenoxyacetic acid) herbicide was applied to control weeds.

In Cycle 2 the sowing density was 40 kg of seed ha^-1 and a Brillion SSPT 604-5^® was used. The fertilization dose was 96-70-00, distributed in two applications: one during the sowing (27-70-00) and the second after the second harvest (69-70-00) at 177 DAS. The experimental plot was 2 m wide and 100 m long (200 m²), with eight replicates per treatment. Weed control was carried out with 2, 4-D.

Four auxiliary irrigations were applied and the cumulative rainfall in the cycle reached 333 mm (Table 1). Humidity was higher than in Cycle 1. The irrigation-constant rainfall mix between sowings and in the first and second harvests hindered the application of fertilizers and herbicides.

Productivity was determined using forage yield evaluations in each harvest; a 50 cmX50 cm metallic frame (2500 cm²) was used to determine the sample area. Plants were cut 5 cm above the ground. Green and dry forage samples were weighted in a paper bag, using a scale with an accuracy of 0.01g. Forage was dried at 60 °C, using a forced air circulation system stove. Forage yield was expressed in t ha^-1. The remaining grass of each experimental plot was harvested (cut 5 cm above the ground), after the sampling stage, using a John Deere 530^® harvester. The grass was left to sprout freely again until the following harvest.

Forage was harvested six and five times in cycle 1 (between 75 and 241 DAS), and in Cycle 2, respectively. This difference between cycles was the result of the abundant, atypical and constant winter rainfall (260.8 mm; Table 1) that occurred after irrigation, which slowed the growth of the grass. Therefore, harvests started 118 DAS and continued until 262 DAS. Water excess reduces plant growth (^{NDS, 2007}), mainly due to hypoxia and its negative effect in water and nutrimental absorption (^{Akhatar and Nazir, 2013}).

In Cycle 1, six equidistant samples of each variety were harvested (systematic sampling with six replicates) from the strip of land. In Cycle 2, two samples of each species and variety were obtained from the four central strips of land (systematic sampling with eight replicates). The higher number of replicates in Cycle 2 was the result of the plot condition and the greater experimental surface.

Forage quality was established based on its chemical composition, fibers, lignin, and in vitro dry matter degradation (IVDMD). Dehydrated samples were ground using a Wiley^® grinder, with a 1 mm sieve. Dry matter (DM) content was obtained drying the sample at 100 °C, at constant weight; crude protein (CP) was obtained using the Kjeldahl method (^{AOAC, 1990}); fibers were obtained using NDF and ADF; and lignin (L) was obtained using ^{ANKOM (2005)}.

For IVDMD, the rumen fluid of two 700-kg (live weight) male cattle with cannula was used. The cattle were fed alfalfa hay and commercial concentrate (12 % protein). The fermentation phase was carried out in a Daisy^II incubator (ANKOM Technology Corp. Macedon, NY, USA), using the protocol suggested by the manufacturer (^{ANKOM, 2011}).

For each cycle and harvest date, green and dry forage yield were analyzed separately using ANOVA, with a completely random model with six and eight replicates, in Cycle 1 and 2, respectively. Forage quality variables were analyzed using four field replicates. When statistical differences were obtained, a Tukey test (p≤0.05) was applied to compare means. A correlation analysis was applied to green and dry forage, with maximum and minimum temperatures.

Results and Discussion

Forage yield

In Cycle 1, green forage yield varied significantly (p≤0.05) in the first, fourth, and sixth harvests (Table 2). In the first harvest, Maximus and Matua had the greatest and lowest yield (Figure 1A). In the sixth harvest, perennial green and Hercules green forage had the highest and lowest yield, respectively. In Cycle 2, green forage yield varied (p≤0.05) in the three initial harvests (Table 3). Perennial green and Hercules stood out during the first harvest (Figure 1B).

Table 2 Mean squares of the variance analysis of the green and dry forage yield for grasses grown during the dry season of the 2013-2014 cycle, in Durango, Mexico. 

†F.V.	gl	Corte 1		Corte 2		Corte 3		Corte 4		Corte 5		Corte 6
		FV	FS	FV	FS	FV	FS	FV	FS	FV	FS	FV	FS
Variedad	3	1452.0	13.1	197.3	13.8	286.3	6.8	304.6	0.5	28.3	3.4	41.0	1.1
¶p		<0.0001	0.0025	0.1047	0.0036	0.0705	0.0894	0.0093	0.8551	0.4486	0.0718	0.0242	0.3021
Error	20	110	2	85	2	105	3	61	2	31	1	11	1
§C.V. (%)		26	25	22	20	21	19	21	22	20	22	18	24

^†F. V.: variance source; gl: degrees of freedom; FV: green forage; FS: dry forage; ^¶p: probability; and ^§C.V.: variance coefficient.

Figure 1 Forage yield for grasses grown in Durango Mexico. A) Green forage, 2013-2014; B) green forage, 2014-2015; C) dry forage, 2013-2014; and D) dry forage 2014-2015. 

Table 3 Mean squares of the variance analysis of the green and dry forage yield for grasses grown during the dry season (2014-2015), in Durango, Mexico. 

†F.V.	gl	Corte 1		Corte 2		Corte 3		Corte 4		Corte 5
		FV	FS	FV	FS	FV	FS	FV	FS	FV	FS
Variedad	3	181.7	1.7	3.4	206.1	436.8	10.2	29.3	2.3	14.9	1.4
¶p		<0.0001	<0.0003	<0.0001	<0.0007	<0.0001	<0.0001	0.0713	0.0602	0.1230	0.0185
Error	28	465.0	5.6	12.7	406.9	303.4	8.6	314.5	23.4	198.9	10.1
§C.V. (%)		19.4	9.5	14.8	26.4	17.7	20.0	22.1	28.5	22.2	27.3

^†F. V.: variance source; gl: degrees of freedom; FV: green forage; FS: dry forage; ^¶p: probability; and ^§C.V.: variance coefficient.

There were differences in dry forage yield in the first two harvests of Cycle 1 (p≤0.05) (Table 2). Maximus and Hercules stood out in the first harvest, while Matua and perennial green stood out in the second. All varieties showed green and dry forage yield losses in late winter (Figures 1A and 1C).

In Cycle 2 there were differences between green forage varieties (p≤0.05) (Table 3), in the first three harvests; there were differences between dry forage varieties, in all harvests (Figures 1B and 1D). As a result of their green forage yield, perennial green and Hercules stood out in the first and second harvests. During the third harvest, the yield of Maximus and Hercules increased.

In Cycle 2, green and dry Mantua forage yield was reduced during the first to fourth harvest, as a result of the damage caused by herbicides. Therefore, the remaining varieties had a greater yield (Figures 1B and 1D). The accumulated dry forage yield for Mantua was 14.4 t ha^-1 [with an interval of 17.8 t ha^-1 (Green Perenne) to 19.5 t ha^-1 (Maximus)].

Cycle 2 had different results from Cycle 1. In Cycle 1 the cumulative yields of green and dry forage had cumulative yields 242 and 39 t ha^-1, with the same grass varieties. Differences were the result of weather conditions (Table 1) and crop handling. In Cycle 2, heavy rains were recorded before the first harvest and this hindered the application of fertilizers and herbicides to control broadleaf weed.

In Cycle 1, the minimum temperature had a negative correlation with green (r=-0.73) and dry (r=-0.83) forage yield, and maximum temperature was also correlated with green (r=-0.58) and dry (r=-0.57) forage yield. In Cycle 2, dry forage yield had a higher correlation coefficient with minimum (r=-0.90) and maximum (r=-0.89) temperatures; meanwhile, and correlation coefficients for green forage yield decreased with minimum (r=-0.66) and maximum (r=-0.49) temperatures. Therefore, temperature affected the cultivars’ forage productivity. This effect was reported in other studies: minimum temperature affected the breathing rate and reduced biomass accumulation (^{Hatfield and Prueger, 2015}).

The grasses evaluated in this study yielded up to 249 t ha^-1 and 49 t ha^-1 of green and dry forage, respectively.

Forage quality

In Cycle 1, the varieties had different protein content (p≤0.0009, p≤0.0024) in most of the harvests (Table 4). Matua stood out in the first harvest (21.8 %), owing to the lower growth of plants; Maximus, Hercules, and the perennial green mix showed statistically lower and equal content among them (Figure 2A). Additionally, Maximus had a high forage yield, with regard to the requirements of the growing cattle (16.0 %) (^{NRC, 2001}).

Table 4 Mean squares of the variance analysis of the forage quality from grass grown during the dry season (2013-2014), in Durango, Mexico. 

†F. V.	gl	Corte 1	Corte 2	Corte 3	Corte 4	Corte 5	Corte 6
		Proteína Cruda
Variedad	3	7.4	6.5	8.6	8.0	4.4	6.7
¶p		0.0024	0.0158	0.0021	0.0019	0.1515	0.0009
Error	4	0.2	1.0	0.9	0.9	1.4	0.1
§C. V. (%)		2.4	5.9	6.1	5.8	9.2	3.2
		Fibra detergente neutro (FDN)
Variedad	3	16.6	60.7	33.8	19.8	19.2	10.2
p		0.2153	0.0777	0.1219	0.0228	0.0115	0.0210
Error	4	7.1	12.2	9.3	1.9	1.2	0.9
C. V. (%)		5.8	6.9	5.9	2.5	2.0	1.6
		Fibra detergente ácido (FDA)
Variedad	3	15.6	0.6	13.2	5.9	19.6	5.6
p		0.1864	0.6247	0.0270	0.0399	0.0389	0.1583
Error	4	6.0	0.9	1.4	0.8	2.6	1.9
C. V. (%)		10.1	4.6	4.5	3.2	5.6	16.8
		Lignina
Variedad	3	3.7	0.8	1.5	1.6	1.1	0.2
p		0.4820	0.1206	0.1403	0.3594	0.1287	0.9453
Error	4	3.8	0.2	0.5	1.1	0.3	1.4
C. V. (%)		35.5	11.8	13.6	27.7	10.2	18.6
		Digestibilidad in vitro de la materia seca (DIVMS)
Variedad	3	13.2	6.2	5.4	7.6	20.6	5.8
p		0.6037	0.6528	0.8189	0.4285	0.3707	0.7674
Error	4	19.2	10.5	17.4	6.5	14.9	14.8
C. V. (%)		6.3	4.8	6.4	4.2	6.4	7.4

^†F. V. variation source; gl: degrees of freedom; p: probability; ^§C.V.: coefficient of variation.

Figure 2 Forage quality of three varieties and a commercial mix of grasses grown in Durango, Mexico. Cycle 1: A) protein; C) NDF (neutral detergent fiber); E) ADF (acid detergent fiber); G) lignin; and I) IVDMD (in vitro dry matter digestibility). Cycle 2: B) protein; D) NDF; F) ADF; H) lignin; and J) IVDMD. 

The lower protein content of Maximus (compared with Matua) was related with its earliness and the delay in forage harvesting -carried out based on the growth of all the types of grass studied. Maximus had high protein content during the third and fourth harvests. Therefore, it overtook all other varieties (third harvest) and Hercules and Matua (fourth harvest). Protein content was lower during the last harvest: around 8.8 % for Matua and 12.9 % for perennial green.

In Cycle 2, the protein content was different (p≤0.05) between varieties. During the first harvest, the average content was 28.2 % (22.3 % and 33.2 %). The values diminished and had a lower average during the fourth and fifth harvests (13.2 % and 13.6 %). Hercules had the lowest value (p≤0.05) (Figure 2B). During the warm months (April and May), the protein content diminished for all varieties, plant development was sped up, the blades were lignified, and spikes were formed (^{Aganga, 2004}; ^{Velasco et al., 2005}). Nevertheless, in several harvests, protein content was higher than 12 %. This situation reduces the supplementation requirements of cattle in Durango, with regard to other forages, such as maralfalfa grass (PC 5.1-12.7 %) (^{Ortiz et al., 2016}).

Neutral detergent fiber content was similar (p˃0.05) between varieties, but there were differences during the fourth, fifth, and sixth harvests of Cycle 1 (p≤0.05); Cycle 2 only had different content during the third harvest. This matched higher environmental temperatures (Figure 2C and 2D). Neutral detergent fiber content in Cycle 1 increased from 45.6 % (first harvest) to 60.7 % (last harvest). The highest mean value in Cycle 2 was 49.3 % (first harvest) and it was higher than during Cycle 1, as a result of the delay in the development of the weather conditions from the sowing to the first harvest. The highest NDF value was recorded during the last harvest, which was statistically similar (p˃0.05) in all varieties (average: 64 %). The NDF content in all varieties increased during the warmer months (April and May), as a result of the accelerated development and the accumulation of structural carbohydrates which support reproductive structures (^{NRC, 2001}).

In Cycle 1, ADF content was similar (p˃0.05) between varieties in the first, second, and sixth harvests; however, in Cycle 2, differences were detected in the first through fourth harvests (Figure 2E and 2F). ADF content increased along with the number of days after sowing. In Cycle 1, the mean values increased from 24.2 % to 34.3 % (Figure 2E) and, in Cycle 2 (Figure 2F), they increased from 26.8 % to 36.9 %. The increase of ADF content had a negative impact in forage degradation. Therefore, grass must be cut every 21 d, reducing the accumulation of structural carbohydrates in blades.

NDF and ADF content in Hercules and Maximus was similar to the data reported by ^{Lozano et al. (2002)} for Oregon annual ryegrass with mean values of 42 % and 27.3 % for NDF and ADF, respectively. Both contents showed a tendency to increase as the plant ripened (^{Aganga, 2004}), although “good quality” values were achieved, during the first three harvests of Cycle 1 and first two harvests of Cycle 2. The maximum limits are 52 % for NDF (Hercules and perennial green) and 32 % for ADF (all varieties) (^{Van-Soest, 1965}).

In both cycles and for most harvests, the lignin content did not show differences (p˃0.05) between varieties. Mean values increased during the warmer months of the year (Figure 2G and 2H). Therefore, the mean values in Cycle 1 increased from 5.5 % (first harvest) to 6.3 % (last harvest) (241 DAS). Lignin content showed great fluctuation, particularly in the Maximus and Hercules varieties, mainly due to their high growth capacity, timely harvesting, and variation in the ripeness of plant structure. In Cycle 2, mean lignin values increased from 4.4 % to 5.9 % and less fluctuation between harvests was observed (Figure 2H); nevertheless, during the fifth harvest, lignin content increased significantly in Maximus (5.8 %), perennial green (6.3 %), and Hercules (6.9 %). Temperature increase had a positive effect in lignin accumulation in all varieties, reducing forage quality.

In vitro dry matter degradation showed differences (p≤0.05) between varieties, only during the first harvest of Cycle 2 (Figure 2I and 2J). In Cycle 1, mean IVDMD decreased from 69.2 % (first harvest) to 51.8 % (last harvest) (241 DAS) (Figure 2I). In Cycle 2, IVDMD decreased from 74.2 % (first harvest) to 62.8 % (262 DAS harvest) (Figure 2J). Temperature increase reduced IVDMD. Therefore, during the warm months, diminishing the period between harvests is not advisable. This would avoid the decrease of forage quality mainly caused by lignin accumulation.

Conclusions

The varieties of annual ryegrass and the commercial perennial ryegrass mix can be used to produce high-quality forage during the dry season in Durango. The perennial green mix and the Matua prairie grass can be used to obtain green forage, with low productivity during the first harvests and high productivity during the last harvests. The protein complement for growing cattle will be low, as a result of the high-quality of the forage that was evaluated.