Forage yield

Júpiter showed higher annual forage accumulation (14,510 kg DM ha^-1) and San Miguelito, Cuf-101 and Valenciana the lowest yields (8,020 kg DM ha^-1, average), with 81 % less than Júpiter (Table 1). Regardless of the variety, the highest seasonal yield was in summer (3,508 kg DM ha^-1), whereas in fall and winter it was 58 % lower (2,214 kg DM ha^-1). All the varieties showed seasonality with the tendency: summer (33 %) > spring (26 %) > winter (21 %) > fall (20 %). This behavior depended on the direct and close relation between the forage yield and the optimal temperatures for alfalfa growth (^{Rojas et al., 2012}), which were present in summer (Figure 1).

^{Mendoza et al. (2010)} reported a similar seasonal behavior in the forage yield, with the variety San Miguelito and cutting interval of 6 to 7 weeks, and the higher production was in summer (10,638 kg DM ha^-1) with a seasonal distribution of 31, 26, 23 and 20 % in summer, spring, fall and winter. ^{Teixeira et al. (2007)} evaluated the Kaituna variety, with cutting interval of 42 d, and the lowest values were in winter and beginning of spring (2,000 and 3,000 kg DM ha^-1, respectively) as compared to spring (3000 kg DM ha^-1). However, ^{Morales et al. (2006)} obtained the highest yields (5,500 kg DM ha^-1) at the end of winter (February and March) and beginning of fall (September and October), and the lowest (2,000 kg MS ha^-1) at the beginning of summer (June-July). They indicated that these yields coincided with the months of highest and lowest temperature in the study region (Mixteca Region, Oaxaca). The values from our study were lower than those by ^{Villegas et al. (2004)} in the Valenciana variety, which accumulated more fodder (4,700 kg DM ha^-1) in spring. The lower yields in our study, as compared to those by ^{Villegas et al. (2004)}, ^{Morales et al. (2006)} and ^{Teixeira et al. (2007)}, may be due to the fact that the grassland was in the fourth year of production, with lower persistence, higher competition from undergrowth, and depletion of carbohydrate reserves. In this regard, ^{Mendoza et al. (2010)} mentioned that the depletion in carbohydrate reserves in plants reduces the yield.

Botanical and morphological composition

Alfalfa was 56 % of the accumulated annual yield, and 44 % was grasses, undergrowth and dead material (Figure 2). The highest production per season (60 %) was obtained in fall and spring, and the lowest in summer (25 %). Júpiter showed the highest annual average (60 %) of leaf and stem, and the lowest of undergrowth (33 %) and dead material (7 %). The highest average annual invasion was found in Chipilo (56 %). Regardless of the variety, the contribution of alfalfa to the yield decreased progressively with the time of study (Figure 2); in contrast, the weed increased from 30 to 63 %. The highest amounts of leaf and stem were found in spring (31 and 29 % in average), and in fall and summer those of dead material (14 %) and undergrowth (63 %), respectively. In the grassland, the persistence of alfalfa decreased due to the interspecific competition over light, water and nutrients with other species (^{Villegas et al., 2006}), which depletes the carbohydrate reserves and decreases the fodder yield (^{Mendoza et al., 2010}). ^{Cinar and Hatipoglu (2014)} observed contributions of 52 % from alfalfa to the fodder yield on the third year after being established and confirmed that this species does not maintain the yield potential between years; they also mentioned that the contribution of alfalfa in mixed grasslands decreases significantly with the age of the grassland. Our results were similar to those of ^{Sevilla et al. (2002)}, who reported death of more plants during spring-summer and that the alfalfa yield is maintained if the minimum plant density is 30 m^-2, which was on the fourth year of the grassland being established. According to ^{Teixeira et al. (2007)}, the percentage of undergrowth in the grassland increased with the age of the alfalfa cultivation.

Figure 2 Seasonal changes in the botanical and morphological composition (%) of alfalfa (Medicago sativa L.) varieties, harvested at cutting intervals defined seasonally in spring-summer, fall and winter: 28, 35 and 42 d, from September 2011 to September 2012. MM: dead material. 

The leaf contributed less (p ≤ 0.05) to the yield in summer and more in fall, winter and spring (31, 31 and 29 %; p ≤ 0.05). The stem showed a higher contribution in spring (29 %) and lower in summer (15 %). The dead material was 14 % more abundant in fall compared to winter (8 %), spring (7 %) and summer (6 %) (p ≤ 0.05). In this regard, ^{Mendoza et al. (2010)} found a higher amount of leaf in winter and of dead material in summer.

Leaf:stem rate

Significant differences in the leaf:stem rate were found between varieties in all the seasons of the year and in the annual average (p ≤ 0.05). The varieties with the highest annual average of leaf:stem rate were Júpiter and Oaxaca (1.8 and 1.4). The highest leaf:stem rate between varieties (1.6) was found in fall and Júpiter stood out for the highest value (3.3). The seasonal tendency showed the following order: fall > winter > spring > summer (1.6, 1.4, 1.1 and 1.0 respectively). The Chipilo, Oaxaca, Milenia, San Miguelito and Valencia varieties did not show differences between seasons (Table 2). According to ^{Villegas et al. (2004)}, the seasonal behavior was: fall > summer > winter > spring; besides, the decrease in the leaf:stem rate during the warmer months is related to the higher production of dry matter, due to the increment in individual weight of the stems. ^{Lamb et al. (2014)} observed a higher leaf:stem rate in June (beginning of summer), and they related it to the early age of regrowth and, besides, the lowest leaf:stem rate was during the later stages of flowering. ^{Halim et al. (1989)} reported values of 0.7 with 21 d of regrowth and 0.6 with 49 d, and in our study the cutting intervals are similar to those results. According to ^{Morales et al. (2006)}, the leaf:stem rate had an inverse behavior to the yield in dry matter and the growth rate, primarily in the warmest months of the year, which coincides with the accelerated replacement of tissue and causes a higher leaf fall. However, the value recorded in the Júpiter variety of our study (Table 2) was similar to that obtained by ^{Villegas et al. (2006)} in the Valenciana variety (3.4) in fall.

Population density of stems

The Júpiter and Aragón varieties showed the highest annual averages in population density of stems (578 and 542 stems m^-2), San Miguelito presented the lowest density (p ≤ 0.05) (Table 3), and the differences between seasons were significant (p ≤ 0.05). The highest (546 stems m^-2) and lowest (362 stems m^-2) population amount of stems was observed in winter and spring, and the seasonal order was winter (546 stems m^-2) > fall (450 stems m^-2) > summer (380 stems m^-2) > spring (362 stems m^-2). These results had an inverse relation with the forage yield of all the varieties (Table 1), due to the lower growth during fall and winter (^{Rojas et al., 2016}) and was compensated by the higher density of small stems (^{Rivas et al., 2005}; ^{Rojas et al., 2016}), known as law of auto-clearing (^{Hernández-Garay et al., 1999}). In this regard, ^{Teixeira et al. (2007)} pointed out that in the basal stratum of the plant the environmental factors, such as temperature and light, control the dynamic of stem appearance; thus, when the canopy opens, the light stimulates regrowth and delays senescence. ^{Ventroni et al. (2010)} found densities of 336 to 700 stems m^-2, with average of 435 stems m^-2. ^{Teixeira et al. (2007)} observed higher amounts (780 to 900 stems m^-2) and in alfalfa populations with cutting intervals of 28 and 42 d, the average values can be 784 stems m^-2 and a leaf area index of 2.1.

Plant density

Between varieties and seasons of the year, the average annual plant density changed (p ≤ 0.05; Table 4). Júpiter, Vía Láctea, Milenia, Atlixco and Oaxaca had the highest annual values (17 plants m^-2 average), and the lowest value was for Valenciana (8 plants m^-2). The seasonal averages tended to decrease during the study as follows: fall (20 plants m^-2), winter (16 plants m^-2), spring (13 plants m^-2) and summer (11 plants m^-2). In this regard, ^{Sevilla et al. (2002)} pointed out that the death of plants is higher in spring and summer, and then the plant density tends to increase. The forage yield (Table 1) contrasted with the number of average plants in each season, and the higher yields of forage during spring and summer are due to the favorable conditions of light and temperature (Figure 1) for alfalfa growth (^{Rojas et al., 2012}). And, according to ^{Rojas et al. (2016)}, in five alfalfa varieties the growth rate, intercepted radiation, leaf area index, and plant height were higher in spring-summer. However, ^{Villegas et al. (2006)} pointed out that in order to maintain the persistence of the grassland and the forage yield, the interval of plant exploitation should be defined for each season of the year, based on the speed of growth. According to ^{Améndola et al. (2005)}, the useful life of alfalfa grasslands is three years, because the cutting frequency (9 to 11 cuts) causes a lower persistence of the plant population. ^{Villegas et al. (2006)} explain that frequent cutting or grazing cause a fast disappearance of plants of the desired species due to the depletion of carbohydrate reserves.