Keywords: rice; grain yield; CO2 fertilization effect; Kinetin; Sink capacity; exogenous application
Guoyou Zhanga, Hidemitsu Sakaia*, Mayumi Yoshimotoa, Hitomi Wakatsukia, Takeshi Tokidaa, Hiroki Ikawaa, Miwa Araia, Hirofumi Nakamaurac and Toshihiro Hasegawab*
aInstitute for Agro-Environmental Sciences, National Agriculture and Food Research Organization, Tsukuba, Ibaraki 305-8604, Japan; bTohoku Agricultural Research Center, National Agriculture and Food Research Organization, Morioka, Iwate 020-0198, Japan; cTaiyo Keiki Co. Ltd., 1-12-3, Nakajujo, Kita-ku, Tokyo 114-0032, Japan
*corresponding author e-mail: hsakai@affrc.go.jp; thase@affrc.go.jp.
Running title: Kinetin improves the CO2 fertilization effect on rice yield

Introduction

Global food production need to be increased by 70% by 2050 to meet the need of future population \cite{2009,Godfray_2010}. However, yield increase of major crops are currently very slow and limited, for example, through breeding and genetic techniques, rice and wheat yield could be only increased by around 1% annually \cite{Godfray_2010,Ort_2015}. Field management practices, however, have large potential to stimulate crop yield and simultaneously reduce negative environmental impacts \cite{Chen_2014,Yao_2017}. On the other hand, rising atmospheric carbon dioxide concentration (EC) is likely to stimulate plant photosynthesis, plant productivity and crop yield \cite{Kobayashi_2001,Long_2006,Hasegawa_2013,Zhang_2013,Zhang_2015}, which is known as the CO2 fertilization effect (CFE). From long term free-air CO2 enrichment (FACE) studies on rice in Japan, CFE on rice yield (CFEyield) is found to be susceptible to fertilization levels and rice varieties \cite{Kim_2003,Hasegawa_2013,Zhang_2015}, indicating that the CFE could be improved through variety screening and field management practice optimization. 
Through variety screening, larger sink capacity is found to be a key factor to gain larger CFEyield in rice \cite{Hasegawa_2013,Zhang_2015,Nakano_2017}, and a larger sink size cultivar Takanari (Indica) benefited larger CFEyield from EC. The larger sink size in Takanari is due to the large grain number which is stimulated by the cytokinin (CKs) accumulation \cite{imbe2004development,Ashikari_2005}. CKs as well as auxins, gibberellins, ethylene, and ABA are well known phytohormones which influence plant growth, differentiation and development \cite{Peleg_2011,Ha_2012,Zwack_2015,Jameson_2015,Koprna_2016,Yang_2016}. Basing on the positive effects on such as: branching, delaying of senescence, sink strength and seed set, exogenous application of CKs as a tool for field practice and yield management has been tried for cereals, legumes, vegetables, cotton and fruit trees \cite{Koprna_2016,Albrecht_2016}. Among the CKs, Kinetin (KIN) is likely to stimulate rice yield and improve rice adaptations to abiotic stresses \cite{RAY_1981,PATEL_1992,MIYOSHI_1997,Zahir_2001,YANG_2002,Kariali_2007,Ghorbani_Javid_2011,Borah_2016}. Effects of KIN on rice grown at EC, however, is yet to known. As EC stimulate photosynthesis and source supply and KIN increase sink capacity and sink strength, thus, rice yield enhancement by EC will be stimulated by exogenous application of KIN. And rice cultivar which has large sink capacity and benefits more from the CFE may gain more yield enhancement under EC+KIN. In order to check these two hypothesis, we conducted a field experiment at the Tsukuba FACE site \cite{NAKAMURA_2012} to check the combined effect of exogenous application of KIN and atmospheric CO2 enrichment on rice yield. 

Materials and methods

1.     Experimental site

The experiment was conducted at the Tsukuba FACE site (http://www.naro.affrc.go.jp/archive/niaes/outline/face/index.html), Tsukubamirai City, Ibaraki Prefecture, Japan (35o58’N, 139o60’E; 10m above sea level) in 2016. Introduction and performance of Tsukuba FACE was reported before \cite{NAKAMURA_2012}, in general, there are four blocks established in the open paddy fields, and two 240 square meters (17m across) octagonal plots for ambient (AC) and elevated CO2 concentration (EC) were set in each block, respectively. Pure CO2 were released from the windward side through emission tubes equipped on the EC plot perimeters, targeting to increase the CO2 concentration by 200 μmol mol-1 above the AC. The soil at the site has a bulk density of 0.87 mg m-3, and contents of 36% sand, 40% silt and 23% clay \cite{Hasegawa_2013}. The start of CO2 release was on 31st May, ended on 17th September. The season-long daytime average concentration in AC, EC plots were 391 ± 11.6 and 586 ± 21.7 μmol mol-1. Weather conditions in the 2016 growing season were shown in Table 1.

2.     Crop cultivation

2.1  Seed sowing

In late April, seeds of rice cultivars Koshihikari (Japonica) and Takanari (Indica) were soaked in water to germinate and then sowed into seedling trays, which have circular cells (16 mm in diameter and 25 mm in depth, Minoru 448, Minoru Industrial Co. Ltd., Okayama, Japan). Sterilized soil with a fertilizer level of 0.4 g nitrogen (N), 0.35 g phosphorus (P) and 0.5 g potassium (K) per 1 kg soil was filled into the cells of the seedling trays, three seeds then were sowed in each cell by hand. After seed sowing, the seedling trays were covered with the sterilized soil and kept indoor until germination. After the seedling emergence, the seedling trays were transferred to a puddled open filed with the protection of tunnel cloche or floating mulch for 2 weeks.

2.2  Field practice

In early April, a PK compound fertilizer with a rate of 4.36 g m-2 of P and 8.30 g m-2 of K was applied to each plot equally before plowing. The field was kept submerged after late April. In mid-May, a rate of 8 g m-2 of N fertilizers were applied to each plot before puddling. There were three kinds of N fertilizers: urea, 2 g m-2; coated urea controlled-release fertilizer (LP100, JCAM Agri. Co. Ltd), which release 80% of its total N for 100 days, 4 g m-2; Controlled-release fertilizer (LP140, JCAM Agri. Co. Ltd), which release 80% of its total N for 140 days, 2 g m-2. On 25th May 2016, approximate 5-leaf seedlings were manually transplanted into the paddy fields on Tsukuba FACE site, with a density of three seedlings per hill (30 cm x 15 cm) and 22.2 hills in every square meter. The fields were kept flooded until 20 August, about three weeks after heading, then the ponding water was drained in preparation for harvest. After water drainage, flush irrigation was applied to keep the soil moist. Each cultivar was harvested at physiological maturity around 14th September.

2.3  Kinetin application

Kinetin (5 x 10-5 M, Sigma) were sprayed on the entire plants grown at EC and AC plots in the early tillering stage. KIN was sprayed in the late afternoons on 44th, 47th, 51st, 55th, 58th and 64th days after transplanting (DAT), at the rate of 70ml per plant with 0.5% (V: V) tween 20 (Sigma) as the surfactant. In compare with KIN spray treatment (KN), same volume of 0.5% tween 20 solution were sprayed to the non-hormone treatment plants as control (CN).

3.     Sampling and analysis

Yield determination and analysis are reported by Hasegawa et al. (2013). In general, we collected aboveground plants parts from 5 hills in early September (Table 1). Harvested plants were air-dried under a rain shelter. Each rough sample was then split into three sub-samples to determine the rice yield, moisture content, proportion of ripened spikelets and classification of unfilled spikelets through different solutions with a specific gravity (SG). Three kinds of solutions were used for the sorting: 80% ethanol solution with a SG = 0.86; water, SG = 1.0; ammonium sulfate solution, SG = 1.06.

4.     Statistical analysis

Analysis of variance was conducted through a split-split-plot design, in which CO2 concentration was treated as the main factor, Cultivar as the split factor, and KIN as the split-split factor with four replications. The mixed model procedure of the SAS package (SAS Ass-In 7.13 HF3 for Microsoft Office, SAS Institute, Tokyo, Japan) was used to test the statistical significance of each factor. Statistical significance is indicated for ***, P < 0.001; **, P < 0.01; *, P < 0.05; actual P value, 0.05 < P < 0.1 and ns, no significance.

Results and Discussion

Rough rice yield was significantly increased by EC by 11.4% averaged across two cultivars with and without KIN application (KN and CN), moreover, the CFEyield was rice cultivar and KIN dependent (P < 0.05, Table 2). In Takanari, CFEyield was significantly improved by KIN application, which changed from 11.2% (CN) to 34.3% (KN). However, in Koshihikari, KIN reduced the CFEyield from 11.5% at CN to -11.4% at KN, these results partly support our first hypothesis, which is that KIN will stimulate the CFE on rice yield. The second hypothesis, which is that cultivar has larger sink capacity and benefit more from CFE may gain larger CFEyield through KIN application. In this study, Takanari benefited more CFEyield through KIN application, thus supporting our second hypothesis.
The cultivar-dependent responses to EC+KIN could be firstly explained by that the optimal concentrations of KIN are cultivar dependent \cite{Lee_2002}, Takanari may prefer the concentration (5 x 10-5 M) of KIN better than Koshihikari. Secondly, morphological differences may also affect the KIN effect, leaf orientation, leaf shape and leaf texture are quite different between Koshihikari and Takanari \cite{Chen_2014a,Muryono_2017,Taguchi_Shiobara_2015}, morphology-dependent spray coverage and retention affect the efficiency of KIN \cite{Gossen_2008,Holloway_2000}. Thirdly, late cultivar was reported to be able to benefit larger yield enhancement by KIN \cite{malabug2010improving}, Takanari grew slower than Koshihikari as shown by the heading dates, which were 77 DAT (EC) and 78 DAT (AC) in Takanari and 75 DAT (EC) and 76 DAT (AC) in Koshihikari. Thus, phenology-late Takanari gained larger yield stimulation by KIN. Fourthly, the stimulation of KIN on tiller bud elongation prefers low light intensity \cite{JIN_1993,LANGER_1973}, Takanari may have experienced relative lower solar radiation due to the slower growth in compare with Koshihikari (Table 1), thus, the stimulation on panicle number by KIN was more apparent in Takanari (P < 0.05, Table 2), which contribute to the larger yield enhancement.     
KIN increased the panicle density (PD) in Koshihikari at AC (10.5%) but decreased PD at EC (-20.3%, P < 0.05, Table 2), and the CFE on PD (CFEPD) was offset by KIN. In general, KIN stimulates tiller bud elongation and increases rice panicle \cite{LANGER_1973,RAY_1981}, however,  at limited nitrogen (N) condition KIN reduces rice tiller number  and shading exacerbates the reduction \cite{JIN_1993}. In this study, The CO2 dependent responses of PD to KIN maybe caused by the stimulation effect of elevated CO2 on rice leaves and the reduction effect on plant N \cite{Anten_2004,Hikosaka_2005}, which induces the shading and N limitation. Takanari, however, can maintain the plant N and the vertical orientation leaves may help avoiding the EC-induced shading \cite{Chen_2014a}. Thus, KIN increased the PD in Takanari at both of AC and EC conditions. KIN increased the spikelet number per panicle (SPP) in Koshihikari, same with earlier studies \cite{Ray_1983,Ashikari_2005,Ding_2013}, however, KIN reduced SPP in the large panicle size cultivar, Takanari  (= 0.0565, Table 2). SPP was reported to be positively correlated with CKs concentrations in rice panicle \cite{Wu_2017}. Takanari bears vertical leaves which may induce a lower absorbance of KIN, and the larger biomass (Table 3, P <0.05) may have a dilution effect on KIN. Thus, KIN concentrations in the panicle of Takanari maybe lower, which induced the SPP reduction. Spikelet density (SD) is the result of PD x SPP, and SD responded similarly with the yield (Table 2). The effects of EC, KIN and EC+KIN on SD is the results of their effects on PD and SPP. 
Percentage of ripened spikelet (RP) in Koshihikari were enhanced by EC by 2.3% at CN and 10.3% at KN, respectively (Table 2, < 0.05). Similarly, RP in Takanari were enhanced by EC by 5.9% at CN and 12.5% at KN, indicating the CFE on RP (CFERP) was improved by KIN and the effects were marginally larger in Takanari (P = 0.0606). RP is the product of filled spikelet divided by the number of total spikelet per area, indicating the efficiency of seed-set \cite{Tsukaguchi_2016}. Stimulation on RP by EC and KIN was reported before \cite{Yang_2009,Hasegawa_2013,Ray_1983,PATEL_1992,malabug2010improving}, respectively. RP increase is mainly caused by the increased assimilate supply, which could be both achieved  by EC and KIN. Their mechanisms, however, are different: EC stimulates photosynthesis \cite{Yang_2009,Sasaki_2005}, KIN prolongs photosynthesis \cite{malabug2010improving,Biswas_1986}. Thus, EC+KIN further increased the assimilate supply and the RP.                
KIN had a trend to increase the thousand grain weight (TGW), however, EC reduces TGW and the effect was more apparent in Koshihikari (P < 0.05). Changes in TGW further changed the sink capacity. Takanari had larger sink capacity than Koshihikari and the sink capacity stimulation by EC was larger at KN in Takanari (P < 0.05, Fig.1). Similar changes in the rough rice yield and sink capacity under KIN x EC in these two cultivars indicate that the yield response was mainly accomplished through the sink capacity change. Hasegawa et al. (2013) compared rice yield of eight cultivars grown under EC, suggesting sink size is an effective trait to gain more CO2 fertilization effect on yield. Nakano et al. (2017) reported that large sink capacity in rice benefited larger CFE on yield. In this study, KIN amplified the sink capacity of Takanari at EC thus, the yield enhancement by EC was further stimulated by KIN. These results suggest that KIN could be used to stimulate the CFE. However, in consider of the negative response in Koshihikari, further studies on the optimization of KIN application need to be conducted. For both of Koshihikari and Takanari, RP was stimulated by EC, and KIN amplified that effect with a significant interaction between CO2 and KIN (P < 0.05, Table 2). This suggest the stimulated biomass allocation to the grain filling, thus the biomass allocation was further analyzed (Table 3).
The above-ground biomass was stimulated by EC, and the effect was largest in Takanari that was sprayed by KIN (P < 0.05, Table 3). CFE on the biomass (CFEbiomass) was increased by more than twice in Takanari through the exogenous application of KIN, however, in Koshihikari, the CFEbiomass was decreased by KIN. Allocation of biomass to rice yield as shown by the harvest index (HI) was marginally affected by KIN (P = 0.0959, Table 3), but was Cultivar and CO2 dependent (P < 0.01). HI was stimulated by EC in Takanari and the effect of EC was larger under KIN application, however, HI was depressed by EC in Koshihikari and KIN exacerbated the depression . Changes in the panicle dry weight were similar with HI: CEF on panicle dry weight (CFEpanicle) was tripled in Takanari by KIN, however, the KIN negatively affect CFEpanicle in Koshihikari (P < 0.05). Although the panicle dry weight was differently affected by CO2, Cultivar and KIN, the assimilate supply per spikelet (as indicated by the grain filling percentage \cite{Tsukaguchi_2016} was not affected by EC and only marginally affected by Cultivar x KIN (P = 0.0789, Fig. 2).

3.     Conclusion

CO2 fertilization effect was found to be improved through exogenous application of Kinetin. Kinetin increases the sink capacity and stimulates biomass allocation to the grain filling thus improves the CFE on rice yield, however, the effect was rice cultivar dependent. These suggest that the CFE could be adjusted through Kinetin application. Optimizing of the application such as: surfactant, hormone quantities and spray timing, and hormone and rice cultivar selection need to be further studied to maximize the CFE.

Acknowledgements

This work was supported in part by the MAFF through “Development of Technologies for Mitigation and Adaptation to Climate Change in Agriculture, Forestry and Fisheries” and in part by the JSPS Grant-in-Aid for Scientific Research (A) (No. 26252004) and Grant-in-Aid for JSPS Fellows (No. 16F16096).