Forage yield

To assess forage yield, we randomly selected two fixed 0.25 m² squares per repetition. Forage in each box was harvested a day before the cut, at a 5 cm height. Samples were then deposited in paper bags, rinsed and dehydrated until constant weight in an oven with forced air circulation (Memmert model UF 260). The weight of the dried samples was recorded on a precision scale (Sartorius model Entris 323-S), to assess the yield per surface unit (kg DM ha^-1).

Leaf: stem ratio

The data for the leaf:stem ratio (H/T) was obtained from the morphological composition of leaves and stems and expressed in kg DM ha^-1.

Weight per stem

One day before each cut, we randomly selected 10 crowns, and in each one, harvested a stem randomly at ground level. The stems with leaves were then dehydrated in an oven with forced air circulation (Memmert model UF 260) until constant weight; their weight was recorded in a precision scale. The average weight per stem was calculated.

Stem density

At the beginning of the study, at random, two 20x20 cm fixed frame were placed on the floor, on each repetition of each variety. The number of stems was monthly recorded, and the changes in their density recorded and seasonally averaged.

Plant density

In a fixed frame of 1 m², placed on the ground at the start of the study, in each repetition by variety, the number of plants was assessed every month and their changes in population density during the year recorded.

Botanical and morphological composition

One day before each cut, of the forage we harvested samples from a sub-sample of approximately 20 %, alfalfa was separated from weeds to determine the botanical and morphological composition of each sample. From a second sample, we separated the alfalfa in leaves and stems. Each component was dehydrated in a forced air the stove, at 55 °C for 72 h. Dry biomass was assessed with a precision scale; data was were averaged by season.

Climate data

Minimum and maximum temperature and precipitation data from the study period were obtained from the agro-meteorological station at the Colegio de Postgraduados, which is 100 m from the experimental area (Figure 1). The maximum recorded temperatures were in July 2010, and from March to June 2011, with a 28 °C average in spring and summer. The minimum temperature (-1 °C) was recorded during the winter season, in December 2010, and January and February 2011. The highest precipitation (404 mm) was in summer and autumn, in July, August, September and November 2010 and June 2011. During the winter and spring season irrigation was provided every 15 d.

Statistical analysis

An ANOVA was performed with the mixed models procedure (^{SAS, 2009}), on a randomized complete block design with four replications. The variables were: forage yield, H/T ratio, weight per stem, stem population per m², plant population per m², and botanical and morphological composition. The comparison of means was performed with the adjusted Tukey test (p≤0. 05).

Figure 1 Maximum monthly average temperature, minimum and cumulative monthly precipitation during the study period (June 2010 to June 2011). 

Forage yield

The highest annual yield (p≤0.05) was that of Júpiter variety and the lowest of Cuf 101 variety. The distribution was: 37 % summer > 26 % spring > 25 % autumn > 13 % winter (Table 1). This gradient coincided with highest spring and summer temperatures, favoring alfalfa development (^{Guimire et al., 2014}), since its growth temperature fluctuates between 15 and 25 °C. Our results were similar to those reported by ^{Villegas et al. (2004)} for 21 600, 21 400, 20 000 and 20 100 kg MS ha^-1 with Oaxaca, Tlacolula, Valencia and Moapa varieties.

Table 1 Seasonal and annual yield (kg DM ha^-1) of alfalfa. 

abc: Means with different lowercase letter in a row are statistically different (p≤0.05). ABC: Means with different capital letter in a column are statistically different (p≤0.05).

^{Idris and Adam (2013)} reported lower yields and noted that, regardless of the variety and the cut frequency, on average, obtained an annual yield of 10 552 kg MS ha^-1. Our results are similar to those of ^{Avci et al. (2010)} with eleven lines and two alfalfa varieties, averaging 20 615 kg MS ha^-1. ^{Abusuwar and Daur (2014)} reported higher annual alfalfa yield on the Cuf-101 and lower in the Hegazi varieties (18 065 and 17 545 kg MS ha^-1).

^{Mendoza et al. (2010)} reported a greater total cumulative forage production in the San Miguelito variety, harvested every 6 and 7 weeks (34 457 kg MS ha^-1), with a seasonal distribution of 31, 26, 23 and 20 % in summer, spring, autumn, and winter. ^{Rivas et al. (2005)} recorded an annual average yield of 33 864 kg DM MS ha^-1 in five alfalfa varieties, with the same cutting intervals than those in this research. Those yields are higher than those in our research, probably because the crop had more than 2 years from its establishment, so its persistence and yield decreased in the time after planting.

Leaf: stem ratio

The H/T ratio showed a significant effect (p≤0.05) between varieties and seasons (Table 2). The Júpiter variety showed the lowest H/T ratio (p≤0.05). Among the other varieties there were no statistical differences (average 1.22). Regardless of the variety, in autumn and winter the H/T ratio was the highest (1.49) compared to summer and spring (0.92 and 0.94). That ratio between forages is an indirect measure of quality: if the value is greater one, it indicates better fodder quality compared to values of less than a unit (^{Hernández-Garay et al., 2000}; ^{Villegas et al., 2004}; ^{Morales et al., 2006}).

Table 2 Seasonal changes in the leaf:stem ratio of five varieties of alfalfa. 

abc: Means with different lowercase letter in a row are statistically different (p≤0.05). ABC: Means with different capital letter in a column are statistically different (p≤0.05).

The H/T ratio in our study, 1.05 and 0.62 in January and November (p≤0.05), was partially different from the four varieties of alfalfa, with two intensities of court; of them, Moapa and Tlacolula showed the best and worst H/T ratio, with 1.4 and 1.1 (^{Villegas et al., 2006}). In contrast, ^{Rivas et al. (2005)} observed lower values (0.79) to those above. ^{Morales et al. (2006)} recorded 0.68 in fourteen varieties of alfalfa and ^{Zaragoza et al. (2009)} recorded in the fifth week in summer, autumn and spring values of 0.8, 0.7, and 0.6 and 0.8 in winter, in the sixth week.

Weight per stem

Weight per stem was not different (p>0.05) between varieties (Table 3). In all the varieties, there were seasonal effect (p≤0.05), 0.92 g per stem spring > 0.78 summer > 0.64 fall > 0.38 winter. Lower values in winter were associated with the negative effect of low temperatures (Figure 1). Although, according to ^{Chen et al. (2012)}, might also be effect of cutting frequencies; they assessed that frequency and weight determined by stem minor and major of 0.27 and 0.45 g, for the lowest and highest frequency.

Table 3 Seasonal changes in the stem (g) weight of alfalfa. 

abc: Means with different lowercase letter in a row are statistically different (p≤0.05). ABC: Means with different capital letter in a column are statistically different (p≤0.05).

Stem weight increased at 15 cm (1.1 g per stem) with cut frequency, and related to the leaf area index and higher yield. These results are similar to those obtained by ^{Meuriot et al. (2005)} in spring with the varieties of Júpiter and San Miguelito. The decrease in weight per stem coincided with the increase in the density of stems (Table 4). This coincided with the decline in individual weight of stems increasing its density per unit area obtained by ^{Hernandez-Garay et al. (1992)}. The previous effect is explained by the self-thinning law (^{Hernandez-Garay et al., 2000}).

Table 4 Seasonal changes in the stems density (stems per m²) of five alfalfa varieties. 

abc: Means with different lowercase letter in a row are statistically different (p≤0.05). ABC: Means with different capital letter in a column are statistically different (p≤0.05).

Stem density

Statistical differences in higher stem density, between varieties (p≤0.05), were found in Júpiter and Atlixco with an average of 641 stems per m²; in contrast Cuf 101 presented lower density, with 417 stems per m² (Table 4). The seasonal average density showed differences (p≤0.05) and decreased with time: 677 autumn > 584 summer > 524 winter > 460 spring (Figure 1).

^{Stanisavljević et al. (2012)} observed a similar response in four varieties of alfalfa: the higher density of stems, 518 per m², in the first year and the lower, 140 stems per m², in the fourth year. ^{Chen et al. (2013)} indicated that stem density increases often cut to a point in decline, regardless of the variety and year of evaluation, 645, 734 and 688 stems per m² with frequencies of 30, 40 and 50 d, which is highly related to the yield.

Organic fertilization on alfalfa increased stem density, plant height and yield (^{Al-Juhaimi et al., 2014}). Temperature and soil moisture are the main climatic factors that influence the density and weight of stems, they are related to the environmental conditions, and if these are favorable, stem production is constant and increases biomass production (^{Ventroni et al., 2010}).

^{Hernández-Garay et al. (1992)} noted the inverse relationship between the density of stems and dry matter production: the greatest number of stems results in less forage yield, possibly for the low individual weight of each one of them. To relate the stem weight and tiller density was observed following the self-discharge law, described by ^{Hernández-Garay et al. (2000)}, since the greater weight, with lower stem density was presented in spring.

Plant density

Plant density decreased with time in all varieties (Table 5), from 32 plants per m² in summer to 22 in spring (p≤0.05). Greater and lesser plants mean density corresponded to Júpiter and Cuf 101, with 33 and 21 plants per m², and they lost 10 and 11 plants, respectively.

Table 5 Seasonal changes in the plant density of alfalfa (plants per m²). 

abc: Means with different lowercase letter in a row are statistically different (p≤0.05). ABC: Means with different capital letter in a column are statistically different (p≤0.05).

The increase in time of establishment in a meadow of alfalfa stabilizes coverage and plants density; but, at certain time they decrease, depending on the variety and the site (^{Mortenson et al., 2005}). ^{Dolling et al. (2011)} in alfalfa with low density, eight plants per m², obtained the higher yield of dry matter compared to 39 plants per m², due to the fact that the crown develops a greater number of stems by a reduced intraspecific competition.

Alfalfa higher yield in spring was related to greater intercepted solar radiation (95 %) at distances of 10, 15, 20, 25 and 30 cm between plants; but, in summer and winter it reached 95 % of intercepted radiation only when distance between plants was 10 to 15 cm (^{Mattera et al., 2013}). The smaller the separation between plants, the higher the yield (^{Stanisavljević et al., 2012}; ^{Mattera et al., 2013}; ^{Baldissera et al., 2014}).

Botanical and morphological composition

Alfalfa represented more than 90 % of the desirable species on the prairie during the entire period of study (Figure 2). Only in spring and summer there was the greater weed presence (p≤0.05) (Aristida stricta, Bromus inermis and Malva neglecta were predominant). The differences between varieties throughout the year were significant (p≤0.05). Cuf 101, with 5 %, presented greater weed invasion. Statistical differences were observed between seasons of the year on the contribution of leaves to yield (p≤0.05) regardless of the variety; autumn and winter, with 57 and 59 % of leaf, surpassed spring (47 %) and summer (45 %). Dead material was not presented in the experimental period since alfalfa defoliates senescent leaves.

Figure 2 Seasonal changes in botanical and morphological composition (%) of five alfalfa varieties. 

The higher percentage of weeds in spring and summer might be due to higher temperature in these seasons, which favored its competition for light, water and nutrients (^{Mortenson et al., 2005}). The greater weed variety in Cuf 101 was due to its lower plant density. This coincided with what documented ^{Celebi et al. (2010)} and ^{Mendoza et al. (2010)}.

^{Rivas et al. (2005)} assessed five varieties of alfalfa and recorded the highest annual leaf percentage (p≤0.05) (35 and 33 %) in the varieties of Oaxaca and San Miguel, and the lower in Moapa 30 %. In this regard, ^{Hernández-Garay et al. (2012)} observed that with more frequent harvest the nutrimental quality of alfalfa increases, because it increases the leaf percentage; but frequent cuts quickly eliminate the perennial due to depletion of carbohydrate reserves, since the regrowth of forage species occurs by translocation of carbohydrates in the roots and base of aerial meristems remaining stems; this causes the rate of regrowth to be slow and the weeds invade the prairie.