Water-saving irrigation practices for rice yield information and nitrogen use ef ﬁ ciency under sub-tropical monsoon climate

Simple and practical water-saving irrigation practices (WIP) with nitrogen-reduction are bene ﬁ cial to the development of rice cultivation technology with promotion of resource-conservation and environmental friendliness. The effects of WIP with nitrogen-reduction on population quality, annual yield and nitrogen use ef ﬁ ciency were studied by a ﬁ eld experiment. WIP could maintain or increase the annual yield of rice production models. The highest annual yield of more-water-saving irrigation practice (WIP150) was 8.42 t hm (cid:1) 2 for the double-season rice production model and 12.71 t hm (cid:1) 2 for the ratoon rice production model, respectively. Compared with non-application of nitrogen, the annual yield of nitrogen-reducing practice (NRP) and farms ’ fertilizer practice (FFP) increased signi ﬁ cantly ( p < 0.01), while a non-signi ﬁ cant difference of annual yield between the FFP and NRP was observed; the annual yield of the NRP and FFP was 9.73 and 10.02 t hm (cid:1) 2 of the double-season rice production model, and 12.84 and 14.34 t hm (cid:1) 2 of the ratoon rice production model, respectively. AE N , PE N , PFP N and RUE N of the NRP were higher than those of the FFP. Therefore, observing the change of water layer in the soil layer via a simple self-made PVC indicator tube, reducing about 20% nitrogen quantity was a feasible and simple cultivation technique for water-saving and nitrogen-reduction in the rice production models.


INTRODUCTION
Scarcity of water for agricultural production is becoming a major problem in many countries, particularly the world's

Test site and materials
The experiment was carried out in the Datong Lake district of Yiyang city in 2016 and Yongan town of Liuyang city in 2017, and the climate of the site was sub-tropical monsoon climate. The average temperature was 18.3 C (multi-year average 17.3 C) in the Datong Lake district and 18.3 C (multi-year average 17.5 C) in Yongan town; total precipitation from April to October was 1,443 mm (multi-year average 1,240 mm) in the Datong Lake district and 1,365 mm (multi-year average 1,172 mm) in Yongan town.
The tested varieties were Huanghuazhan (inbred rice) in the ratoon rice production model, Zhongzao 39 (inbred rice), Lingliangyou 104 (hybrid rice) for the early season and Wushansimiao (inbred rice) and Wuyou308 (hybrid rice) for the late season in the double-season rice production model.

Experimental design
Three water regimes were applied during the rice-growing seasons: (1) conventional irrigation practice (CIP): according to the local farmer's habit of irrigation, 30-50 mm water layer at the re-greening stage, 20-30 mm water layer continuously flooded at the tillering stage, and draining at late tillering and maturity stages; (2) water-saving irrigation practice (WIP100): after rice transplanting, 30-50 mm water layer at the re-greening stage, and the water is irrigated and dried naturally while the groundwater level is lower than 100 mm; (3) more-water-saving irrigation practice (WIP150): (W3): after rice transplanting, 30-50 mm water layer at the re-greening stage, and the water is irrigated and dried naturally while the groundwater level is lower than 150 mm. The selfmade PVC indicator tube was used for observation (Figure 1), and the rest of the time there was no irrigation.
Three kinds of nitrogen quantity were adopted in the experiment: (1) non-application of nitrogen (N0); (2) farms' fertilizer practice (FFP): annual quantity of applying nitrogen in the ratoon rice and double rice model was 375 and 330 kg N hm À2 , respectively; and (3) nitrogen-reducing practice (NRP): based on the FFP, the annual quantity of nitrogen applied in the ratoon rice and double rice was reduced 24% and 20%, respectively.
A randomized complete block design was employed with three replicates; each plot area was 30 m 2 (5 m × 6 m) and separated by ridges (0.2 m high × 0.3 m wide). In order to prevent water movement between adjacent plots, the ridges were covered with plastic sheet inserted into the soil at a depth of 0.5 m.
In 2016, the annual quantity of nitrogen applied in the ratoon rice production model was with 60% in the main season and 40% in the ratoon season. In the main season, nitrogen fertilizer was split in four with 40% as basal fertilizer, 30% as tillering fertilizer, 20% as panicle fertilizer and 10% as budding fertilizer; phosphate fertilizer was applied at 105 kg P 2 O 5 hm À2 as the base fertilizer, and potassium fertilizer was applied at 225 kg K 2 O hm À2 (40% as base fertilizer, 40% as panicle fertilizer, 20% as budding fertilizer).
In the ratoon season, all nitrogen fertilizer was applied 3 days after the main season rice was harvested; potassium fertilizer was applied at 45 kg K 2 O hm À2 .
In 2017, the annual quantity of nitrogen applied in the double-season rice production model was with 45% in the early season and 55% in the late season. Nitrogen fertilizer was split in three with 50% as basal fertilizer, 30% as tillering fertilizer, 20% as panicle fertilizer; phosphorus fertilizer as basal fertilizer was applied at 75 kg P 2 O 5 hm À2 in the early season and 90 kg P 2 O 5 hm À2 in the late season; potassium fertilizer was applied at 120 kg K 2 O hm À2 in the early season and 144 kg P 2 O 5 hm À2 in the late season (50% as base fertilizer, 50% as panicle fertilizer). Weeds, diseases, and insects were intensively controlled throughout the entire growing season in both rice production models.

Aboveground biomass at the full heading stage
At the full heading stage, eight hills were sampled per replicate at different stages to determine the aboveground biomass after oven-drying at 70 C to a constant weight.
Leaf area index (LAI) was determined by a LICOR-3100 leaf area analyzer with specific leaf weight.

Yield and aboveground biomass at the mature stage
Ten hills were sampled diagonally from a 5 m 2 harvest area for each replicate at maturity to determine the panicle number per hill, aboveground total biomass, harvest index (HI), and yield components, as described by Zhang et al.
(). Rice yield was determined from a 5 m 2 area per replicate and adjusted to a water content of 0.14 g g À1 fresh weight.

Nitrogen content in different parts of plants
Straw, filling grain and unfilling grain for each replicate at maturity were crushed by a micro-mill and stored in a vacuum bag. Total N was determined with the Kjeldahl method, involving two steps: (1) digestion of the sample to convert organic N into NH-N and (2) determination of NH-N in the digested sample by a Skalar San þ flow injection analyzer.

Data processing and statistical analysis
Nitrogen use efficiency, including nitrogen agronomic efficiency (AE N , kg kg À1 N), physiological efficiency (PE N , kg kg À1 N), nitrogen partial factor productivity (PFP N , kg kg À1 N), and nitrogen recovery use efficiency (RUE N , kg kg À1 N), was calculated with the method of Xue et al. (). The transport rate of biomass before the full-heading stage (%) was the biomass difference of vegetative organs (stem and sheath, and leaf) between the full-heading stage and mature stage divided by the biomass of vegetative organs (stem and sheath, and leaf) at the full-heading stage; the contribution rate of biomass before the full-heading stage (%) was the biomass difference of vegetative organs (stem and sheath, and leaf) between the full-heading stage and mature stage divided by the rice grain yield. Statistical analyses were carried out using Statistix ver. 8.0 (2004).

Annual yield of different rice production models
There was no significant difference in the annual yield of the ratoon rice production model and the double rice model treated with different water regimes ( Table 1). The highest annual yield for water-saving irrigation practice (WIP150) was 8.42 t hm À2 with the double-season rice production model and 12.71 t hm À2 with the ratoon rice production model, respectively. Compared with non-application of nitrogen (N0), annual yields of NRP and FFP increased significantly (p < 0.01), while no significant difference of annual yield between the FFP and the NRP was observed: the highest annual yield of the FFP was 10.02 t hm À2 with the double-season rice production model and 14.34 t hm À2 with the ratoon rice production model, respectively. There was no interaction between water treatment and nitrogen treatment.

Aboveground biomass of different rice production models
There was no effect on aboveground biomass at full-heading and mature stage under water-saving irrigation practices (WIP100 and WIP150) in the ratoon rice production model and double rice production model (Tables 2 and 3). There were differences of aboveground biomass at the mature stage between rice cultivars in the double-season rice production model, aboveground biomass of the WIP150 was lower than that of the CIP in the early season, and aboveground biomass of the WIP150 with the late rice cultivar WY308 was higher than that of the CIP. Similarly, there was no significant difference of aboveground biomass between the FFP and the NRP under different rice production models, and it was significantly higher than that of the N0 treatment. Other indicators, including leaf area index and crop growth rate, showed a similar trend with aboveground biomass.
Nitrogen use efficiency of different rice production models Under the ratoon rice production model (Table 4), the highest nitrogen agronomic efficiency (AE N ) of the main and ratoon seasons was 15.60 kg kg À1 N and 16.19 kg kg À1 N with the water-saving irrigation practices (WIP150), respectively. A  Under the double rice production model (

DISCUSSION
In this study, the variation of water layer in the soil layer was determined by a simple self-made PVC indicator tube, which was used to determine whether artificial irrigation was needed; the water was irrigated and dried naturally under the water-saving irrigation practice (WIP100) and more-water-saving irrigation practice (WIP150) while the groundwater level was lower than 100 mm or 150 mm.
Our study found that water-saving irrigation practice could guide irrigation practice during the whole rice pro-    Our results indicated that nitrogen fertilizer reduction by 80% and 76% in the above models compared with the FFP was a more reasonable range of nitrogen-reduction.
Meanwhile, from the point of nitrogen use efficiency, AE N , PE N , PFP N and RUE N of the NRP were higher than those of the FFP, but there were differences of rice cultivars between the NRP and the FFP under the double-season rice production model.
There is an interaction effect between water conditions and fertilizer, namely, the changing of soil fertility and nutrient absorbability of the rice with the changing of water conditions in the paddy field, which will affect rice's growth and development. Liu et al. () reported that synergistic interaction between site-specific nitrogen management and AWD occurs in yield formation, and such an interaction could increase not only grain yield, but also resource-use efficiency in super rice, which could reduce the nutrient and water used in production of unproductive tillers and