The field experiment was initiated in May 2001 at Duntou Town of Lanxi (29˚19'N, 119˚43'E and 72.8 m elevation above sea level) and Dingqiao township of Haining city (30˚26'N, 120˚39'E, 12.8 m elevation above sea level), which located in the Hangjiahu Plain and Jinqu Basin, Zhejiang Province, China. Average annual temperature is 17.7˚C and 15.7˚C, annual precipitation is 1399.8 mm and 1220.3 mm , cumulative temperature above 10˚C is 5532˚C and 4170˚C, annual frost-free days are 265 d and 231 d, total sunshine hours is 1981.6 h and 2021 h, respectively. The soil is referred to as Fec stagnic Anthrosols. Its properties are shown in Table 1.

The experiment was a randomized complete block design with three treatments and three replications. The three treatments were plastic film mulching with no flooding (PM), no plastic film mulching and no flooding (UM), and the traditional flooding management (TF). The plot size was 15 m². The TF plots were isolated from the PM and UM plots by a 2 m wide alley with plastic film vertically buried to a depth of 60 cm . The other plots were separated by 0.5 m wide irrigation furrows with two 0.3 m wide levees at two sides of the furrows. For the TF

treatment, the plots were flood-irrigated every 3 - 5 days to maintain a 3 cm water depth until 2 weeks before rice harvest. For the PM and UM treatments, no permanent water depth on soil surface was maintained during the rice growing season. From transplanting (early May) to tilling stage (early June), limited irrigation (30 - 50 mm per irrigation) was applied on PM and UM when soil water content fell below 80% of soil field capacities. The irrigation amounts and rainfall for the treatments were measured and recorded during rice growing seasons.

Fertilizer rates were 135 kg N ha⁻¹ N as urea, 90 kg P₂O ₅ ha ^{− 1} as calcium phosphate, and 90 kg K₂O ha⁻¹ as potassium chloride. Prior to rice transplanting, fertilizers were broadcasted and incorporated into the top 15 cm soil by plowing. The rice variety was Liangyoupeijiu for Duntou and Bing 0209 for Dingqiao. The rice seedlings were transplanted at a spacing of 20 cm × 28 cm with two seedlings per hill. Plastic film, 0.005 mm thick and 1.7 m wide, was used to cover the soil in the PM treatment.

Soil samples were collected from 0 to 15 cm layer at harvest stage and air-dried and through 0.149 mm sieve for determining organic matter in every year. Organic matter was tested by Walkley-Black. And fresh soil samples at Duntou were also collected at 5 August, 7 September 2006, which were the tilling and heading stages for the rice. Five random cores were taken from the plow layer (0 - 15 cm) of each plot with a 5 cm diameter tube auger. The samples of PM and UM treatments were mixed thoroughly and sieved (2 mm mesh) to remove visible plant roots, respectively. For the TF treatment, plant roots were removed by hand. The soil for the molecular analysis was kept immediately at −20˚C.

The bacterial community structure was also studied by a polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE) approach [9]. Whole-community total DNA was extracted and PCR were performed according to M. Y. Wu et al. [10].

WUE was calculated as the amount of grain yield per unit of water used. Water losses due to runoff and leaching were assumed to be negligible [11].

The difference between different treatments was determined using the least significant difference (LSD) test in the data processing system SAS 8.02 software.

The DGGE fingerprinting of PCR-amplified 16S rDNA was shown in Figure 1(a) and Figure 1(b), which was extracted from the soils with different treatments at tiller and heading stages. Apparently, the band patterns of three replicate samples from PM, UM, and TF were highly similar (>95% similarity). The total number of bands in DGGE profiles of different treatments is shown in Table 2. At tiller stage, the total number of bands was 33.3, 23.7, 27.0 for PM, TF, and UM treatment, respectively (p < 0.05^*). At heading stage, the total number of bands was 31.7, 20.0, 28.3 for PM, TF, and UM treatment, respectively (p < 0.05^*). That is, these changes in the number of bands suggest that PM treatment favored the development of a more total bacterial community compared with TF management. One dominant band showed little change among three treatments, suggesting that it may be a predominant bacterial community. It is possible that low pH in TF management could be responsible for the low numbers of dominant bacterial bands as found on DGGE, because there was an increased pH value in PM management [10] [12].

The decreased trend of organic matter content was presented with continuous plastic film mulching under

Standard error is indicated in parentheses (n = 3).

non-flooding condition. At Duntou sites, organic matter content were declined by 36.7%, 28.5% and 33.9% under PM, UM and TF, respectively, compared with the initiation (Figure 2). At Dingqiao sites, organic matter content under PM, UM and TF were decreased by 51.4%, 51.5% and 40.3% after 6 years, respectively (Figure 3). The reason for these results is that plastic film mulching increased soil temperature during the entire growing seasons, and resulted of the improving of bacterial community activities and diversities, which promoted the decomposition of organic matter. This was similar to the previous studies [13] [14].

No significant trend existed in rice yields among the different treatments (Table 3). PM produced similar rice grain yield as TF at Duntou and Dingqiao sites, the average was 7924 kg∙ha⁻¹ and 7015 kg∙ha⁻¹ for PM and 8150 kg∙ha⁻¹ and 6990 kg∙ha⁻¹ for TF. Grain yields in PM had significantly higher than UM plots for two years at two sites, 2.7%, 6.1% at Duntou and 0.8%, 8.1% at Dingqiao, respectively.

Plastic film mulching reduced rice water consumption. During rice growing seasons in different years, the total water consumption in PM plots was 723 - 890 mm, much lower than that in TF plots, which was 1388 - 1591 mm. Irrigation water used in PM was 103 - 253 mm and in UM 196 - 310 mm, which were much lower than those in TF (718 - 947 mm). The irrigation water use reduced by 72.8% - 85.6% in PM and 66.9 - 72.8% in UM compared to TF treatment. Water use efficiency (WUE) in terms of grain yield per unit consumed water increased by 70.2% - 80.4% for PM and 47.9% - 63.8% for UM. Irrigation water use efficiency in terms of grain yield per unit irrigation water improved by 273.7% - 1300.0% for PM and 171.8% - 585.4% for UM, compared with TF respectively. PM reduced water consumption and increased WUE significantly compared to TF. These results were strongly supported by other studies conducted in the same region or elsewhere [15] [16].