Experiments were conducted on the cutting field at the Songnen Grassland Ecological Research Station, operated by Northeast Normal University and located at latitude 44˚40'N, longitude 123˚44'E, 150 km northwest of the provincial capital, PR Changchun, China. The study area has a frost-free period of about 140 days. The mean annual temperature ranges from 4.6˚C to 6.4˚C, varying from −16˚C in January to 25˚C in July. The annual precipitation is 350 - 450 mm and. The annual potential evapotranspiration is approximately three times as much as the mean annual precipitation. The main soil type of the area is a meadow chernozem with pH is 7.5 - 9.0 and organic matter is 3.5% - 6% in the surface layer. The grassland is continual cutting field and the dominant species is L. chinensis [1] [3] [17] .

The experiment was from the August 15^th, 2006 to August 11^th, 2007. The cutting date was from August 15^th to October 4^th in 2006 with every 10 days interval. The experiment used a randomized complete block design, the plots area were 5 m × 5 m, 5 replications, and the height of the remaining stubble was about 5 cm using the reaphook. Then we investigated the regrowth character of the different treatments in October at the current year, including the regrowth plant height, leaf number, and aboveground biomass of L. chinensis, constitute of species, and density with the sampling area was 1 m × 1 m in each plot.

We settled two 1 m × 1 m subplots in each plot the on May 10^th in next year after the grassland had been turning green. The first month we measured the length of the second leaf, the height of L. chinensis in each sample, and harvested the aboveground biomass of one of the two samples. We harvested the aboveground biomass of the other settled subplots on August 11^th, 2007, measured the height of L. chinensis, the species composition of the plant community, the aboveground biomass of each species and the stem/leaf of L. chinensis. We also made the background investigation of the grassland on August 16^th, 2006, the plots were 1 m × 1 m with 10 replications.

The method of harvesting aboveground biomass was using the scissors to cut the plants getting closing to the ground. All the samples were oven-dried at 65˚C C for at least 48 hours immediately after harvest and weighted. The crude protein percentage was determined in duplicate using Kijldahl method (Kjeltec Analyzer Unit, FOSS TECATOR).

The similarity of community was calculated by Sørensen Coefficient (IS_s = 2C/(A+B), A and B are the number of the species of the two communities, C is the number of the common species) [24] . The effect of height, leaf number, leaf length, density, aboveground biomass and stem/leaf data were analyzed by ANOVA with Tukey test for multiple comparisons among the treatments. All statistical analyses were performed using SPSS 13.0 (SPSS Inc., Chicago, USA) at 0.05 level.

The species number of the community was few, and there were about nine species in 1 m². The fresh and dry weight of L. chinensis made up 86.26% ± 2.25% and 89.70% ± 1.70% of the plant community aboveground biomass, the plant community fresh and dry weight were 517.17 ± 35.40 g/m² and 297.66 ± 18.99 g/m², separately. The main companion species were Kalimeris integrifolia, Echinochloa crusgalli, Eragrostis pilosa, Setaria viridis, Lathyrus quinquenervius, Carex duriuscula, Scriprs fluriatilis, Eleocharis intersita, and Polygonum sibiricum. The height and density of L. chinensis were 31.47 ± 1.12 cm, and 1176 ± 66 plants/m².

The values of regrowth character of L. chinensis descended as the cutting date putting off (Table 1). The height descended quickly after the first cutting date, which was 12.05 ± 0.52 cm. There were no significance between August 25^th and September 4^th, and between September 4^th and September 14^th, but there was significant difference between August 25^th and September 4^th. And the shortest height was at the last two cutting dates, September 4^th and October 4^th. The leaf number was significant different among the cutting dates (P < 0.05). The cutting dates could be divided into three periods and there were significance in any two of the three periods. The regrowth density of L. chinensis was much less than that without cutting (1176 ± 66 plants/m²), the highest one was just litter more than half of that without cutting, the others were less and less, and L. chinensis density of the last cutting date just was as one fifth of that without cutting. The regrowth of the aboveground biomass were also light, and the heaviest was 18.44 ± 1.44 g/m² after the first cutting date.

The regrowth of other companion species were mainly K. integrifolia and C. duriuscula (Table 2). But the height of the species was short. The species such as Artemisia scoparia just appeared in some polts, and the cutting date on August 15^th had the most regrowth species number.

The L. chinensis was at the third-leaf stage when we measured the plot on May 11^th, 2007 after the grassland turning green, and we measured the second leaf length of L. chinensis (Table 3). The second leaf length were no significance among the first three cutting dates, there were significant difference on August 25^th, September 14^th and September 24^th. And the trends were descending as the cutting dates delayed though most of them were not significant (Table 3).

The height of L. chinensis population was lowest on September 24^th, but most of the treatments were not significant, the significant dates were between August 15^th, August 25^th and September 24^th (Table 3).

Means ± SE with different letters were significantly different at 0.05 level.

--: Meant that the species did not grow in the investigated plots.

Means ± SE with different letters were significantly different at 0.05 level.

Neither the density nor aboveground biomass of L. chinensis were significance during the different cutting dates at the turning green stage (Table 3). The density of the first three cutting dates were less than the remaining treatments. The trend of aboveground biomass was ascending as the cutting date delayed. The maximum value was 27.33 ± 5.20 g/m² at the last cutting date.

The three indexes of L. chinensis population which we had measured were not significant among the different cutting dates at the harvest time (Table 4). The

Means ± SE: with different letters were significantly different at 0.05 level.

highest L. chinensis was on October 4. The character of the density was similar as the turning green stage (Table 3), the first three cutting dates were less than the remaining treatments. The maximum aboveground biomass was 199.69 ± 18.20 g/m² of the cutting on September 14^th, and the minimum was 166.25 ± 11.03 g/m² of the cutting on August 25^th.

The species number of the plant community were also few, the species saturation were 10, 9, 11, 12, 10 and 10, respectively for the cutting dates. The dominant species was L. chinensis, which made up the most proportion of the community both in density and aboveground biomass. The dry weight of L. chinensis contributed to at least 90% of the aboveground biomass (Table 4, Figure 2) .The main companion species were K. integrifolia, L. quinquenervius, C. duriuscula, E. intersita, P. sibiricum. But the aboveground biomass of L. chinensis community was not significant among the cutting dates, and they were about 200 g/m² (Figure 1).

The coefficients of the community similarity (IS_s) were high for the neighboring two cutting dates, and higher than 0.70. The highest coefficient was 0.84 between August 15^th and August 25^th, and, they 8 common species.

The values of stem/leaf of L. chinensis were not significant different during the cutting dates. And they all were lower than 0.70, so the weight of the leaves was 1.43 times as much as stems at least. And the cutting date on August 25^th, the weight of the leaves was next to 2 times as much as stems. The crude protein was form 8.0% to 9.5%, but there was not significant among the treatments，the companion species contain higher crude protein than L. chinensis, which were from 10% to 12% (Figure 2).

The Songnen grassland plays an important role in stock raising in northeast China because of its famous for producing L. chinensis which has high palatability and quality [4] [5] [17] . The species richness (species number) of the community is small, furthermore, the L. chinensis was absolute predominance (Table 4 and Figure 1). Therefore, the communities were extremely similarly with different cutting dates (Table 5). Mowing in succession years made the aboveground biomass and density descending, the percentage of the high quality herbage in the community decreasing, weeds increasing [25] [26] . In our study, the values of height and aboveground biomass of L. chinensis had decreased from

2006 to 2007. But the L. chinensis proportion was improved after cutting, probably because the rainfall was less in 2007 than 2006, and L. chinensis is more tolerance than other species in the community [5] . The regrowth character also reflected the advantage of L. chinensis (Table 1 and Table 2).

The value of stem/leaf was an important index of forage quality, intake, and diet selection [27] , usually the value was smaller and the nutrition was higher, because leaf contains more nutrition and is more palatable than stem [28] [29] . The stem/leaf which we measured showed that there was no difference by the different treatments (Figure 2(a)), and the leaves made up a bigger percentage of the aboveground biomass and the quality was fine. And the crude protein percentage was also not significant both L. chinensis population and community, so we got the conclusion that autumn different cutting dates didn’t influence the quality of the L. chinensis grassland. But the plant community had more crude protein than L. chinensis population (Figure 2(b)), because the companion species contained more crude protein, especially the leguminous forage like L. quinquenervius [5] .

The aim of cutting in grassland was to gain hay or ensilage [30] , thus the harvesting date and forage yield were practical significance [11] . The aboveground biomass of L. chinensis population and community were not significant after different cutting dates (Figure 1), and the height, density and stem/leaf were also not significant. This meant that the concentrated harvest dates had the similar influence on the grassland at the remaining stubble was about 5 cm, so when farmers harvested the hay, they just need to consider the factors of weather, labor force, nutrition content and yield of the grassland. The aboveground biomass of L. chinensis reached hump at the middle ten days in August [9] , but the

crude protein decreased and fiber increased after L. chinensis turning green, in order to get maximum crude protein, the grassland should be cutting in the last ten days of July [10] [14] . Considering the weather condition in Songnen grassland, we advised that harvest dates should be concentrated from the middle ten days of August to the first ten days in September [5] .

Cutting means removing the organics from the grassland [4] [31] , the utilization rate of the grassland was less than 50 percent to keep it healthy [25] , because plant litter played an important role on the grassland ecosystem [11] , and it alters the physical and chemical environment, therefore, it may affect the plant community structure and dynamics [32] . We also investigated the uncut community closed to the experimental plots in 2007, the height, aboveground biomass was significant more than the treatments because of the litter, and the stem/leaf and crude protein were not significant. However, the stand litter would be the obstacle when the farmers mow the grass and decrease the quality of the forage, so the managers need to trade-off the yield and quality, and institute the scientific cutting systems to keep the grassland continuative production.