WEATHER, YIELDS AND PLANT VARIABLES

The summer of June, July and August 2019 was drier than the average level of the past 30-year (see Table 4). Meanwhile, July was cooler while June and august were hotter. The annual yields of 2019 were 312 to 1040 g m-2 from N0 to N300, which had wider range than the recorded range of 630 to 830 g m-2 based on different fertilizer amounts, harvest times and grass ages (Nissinen and Hakkola, 1994). While the yields of N0, N150 and N300 are comparable to the corresponding yields in the first harvest year of the latest study by Termonen (2020). Fertilizer significantly improved the yields since both active soil carbon cycling and soil nutrient releasing are enhanced by fertilization (Lu et al., 2011; Lind et al., 2016).

Table 4 Temperature and precipitation comparison between 30-year mean values (1981~2010) and 2019

June July August

30-year scale 2019 30-year scale 2019 30-year scale 2019

Temperature (℃) 14.1 16.1 17.0 15.2 14.5 15.1

Precipitation (mm) 66 43 77 20 75 61

Plant variables (LAI) on 3rd and 12th June were unexpected lower compared to the later pre-harvest measurements. the biomass harvested of each treatment on 13th June was more than the later ones on 25th July or 3th September. Therefore, the plant variables on pre-harvest days of 3rd and 11th June should be comparable concerning plant variables to those on 16th and 22nd July at least. The reason might attribute to measuring mistakes or weather conditions such as clouds and wind.

5.2. NITROUS OXIDE EMISSIONS

The significant higher N2O cumulative emissions from N150 and N300 in this experiment are in agreement with the study of Thornton & Valente (1996). N2O cumulative emissions from N300 and N150 are 145% and 68.1% more than that of N0 while that from N300 is 45.7% improved than that of N150. The improvement might because N2O emissions from nitrification and denitrification depend on N content in soil (Akiyama, Tsuruta and Watanabe, 2000). Addition of N fertilizer significantly improved both the NH4+-N and NO3--N in soil thus facilitated N2O emissions (Mosier,

2001; Khalil, Mary and Renault, 2004; Carmo et al., 2005; Liu et al., 2005; Ruser et al., 2006;

Zanatta et al., 2010).

The drastic N2O emission spiking on 17th June (N applied on 13th June) and gradually dwindled of N300 and N150 are in consistent with previous studies that the largest N2O emissions occur within the first two weeks after fertilizer application (Bergstrom, Tenuta and Beauchamp, 2001; Liu et al., 2005, 2006; Schils et al., 2008). However, N2O emissions of N300 and N150 after the second fertilization did not boost unexpectedly. It might because soil organic carbon becomes a limiting factor for microbial activities thus affecting N2O emissions (Brentrup et al., 2000) after vigorous biomass production in both treatments. Secondly, SOC decomposition might be suppressed by N addition in low C:N ratio (<15) soils (Lu et al., 2011). In addition, soil carbon replenishment from root exudates might be insufficient to meet microbial daily demands and even worse when the aboveground biomass is removed. Since the aboveground biomass removal would induce less carbon allocation to roots (Craine, Wedin and Chapin, 1999; Kuzyakov, 2006) and the reduced NEE would lead to a decreased root exudates or root turnover (Luo and Zhou, 2006). N2O emissions are governed by both the organic carbon availability and the competence of the nitrification and denitrification microbial groups (Kim et al., 2004; Morley and Baggs, 2010). It seems that both nitrification and denitrification are probably prohibited by insufficient organic carbon. This is also viable to explain the overall down turned trend of all treatments after the first fertilization till the end of August.

Fertilization induced N2O flux turbulence was also well captured. N2O emissions from N300 had the largest coefficient of variation with 136% compared to 101% from N150 and 81.6% from N0.

N300 also had the highest interquartile range which was 3.2 and 1.5 times than those from N0 and N150. Emissions on 17th June were clearly partitioned by fertilizer rates from the largest emission of N300 to the lowest of N0, while their difference shrank subsequently till 30th July, after which the flux difference was even smaller onwards till the end of August. The N2O emission turbulence reflects the activities of soil microbes. Thus, N300 had the most responsive microbial groups to the emission related environmental changes such as temperature, moisture and soil nutrients (Davidson and Swank, 1986; Wolf and Brumme, 2002; Zhang and Han, 2008). However, it is difficult to differentiate the single individual factors from this experiment.

The N2O promotion on 17th June from N0 might attribute to rain event. Liu (2006) observed the markedly increased N2O diffusion from soil to atmosphere just after rain and downturned back to normal level three days later. The rain induced N2O promotion attributes to the increase of soil moisture (Baggs et al., 2000; Giacomini et al., 2006). Therefore, on contrast, emissions of N0 on 26th July (one day after the second fertilization) did not burst similarly.

5.3. METHANE UPTAKE

All the three treatments were sinks of atmospheric CH4 during June to August 2019. This is typical finding in a mineral agricultural soil in Finland (Regina et al., 2007; Maljanen et al., 2010).

Although there was not statistically difference in the CH4 uptake capacity amongst treatments during the three months, the N300 showed a stronger potential to take up more CH4. This observation is against some studies that fertilizer inhibit soil CH4 sink capacity (Hütsch W., Webster and Powlson, 1993; Hütsch W., 1996; Kravchenko et al., 2002; Seghers et al., 2003). One possible reason might be methanotroph’s community composition is mostly controlled by soil types thus leading to various CH4 oxidation potentials (Seghers et al., 2005). Secondly, the mechanism of prohibition of soil CH4 oxidation varies in response to different fertilizer practice (Seghers et al., 2003). Thereby, the soil microbial community from the field might have been shifted into a low CH4 oxidation status and adopted to fertilizer because of yearly cultivation and fertilization prior to this experiment. Similar results were reported by some studies (Adamsen and King, 1993; Seghers et al., 2003, 2005). On top of this, the microbial activities might be facilitated by the fertilizer legacy effect thus CH4 uptake in N150 and N300 were higher compared to that of N0 (Hahn, Arth and Frenzel, 2000; Paul et al., 2000; Schimel, 2000; Dan et al., 2001).

While the similarity of cumulative CH4 sink between N0 and N150 might happen because, on one hand, the reasons mentioned above complies to N150. On the other hand, the fertilizer amount of N150 might insufficient to stimulate CH4 oxidation comparably to that of N300. This might further explain the entangled cumulative lines of N0 and N150, which do not set apart until 12th August (see Figure 8). The tiny difference of daily emissions between N0 and N150 are probably amplified by accumulation.

5.4. CARBON DIOXIDE EXCHANGE

Various studies found that grassland is a net CO2 sink (Ciais et al., 2010; Chang et al., 2015;

Gomez‐Casanovas et al., 2016) and nitrogen addition could stimulate NEE (Cheng et al., 2009).

This is in agreement with our study. N addition promoted cumulative NEEs of N300 and N150 by 104% and 97.1% compared to that of N0 during the period. Although cumulative TER were also stimulated concurrently, the NEEs were overridden by GPPs, which is also found in the study from Xia and Wan (2008). The mechanism might be that fertilization facilitates both GPP and TER.

(Evans, 1989; Sinclair and Horie, 1989; Muchow and Sinclair, 1994; Luo and Zhou, 2006), the enhanced soil respiration encourages root turnover by secreting more root exudates to soil, microbial activities are thus stimulated to decompose more SOM as a positive response (Hanson et al., 2000). Ultimately, more biomass is produced accompanied by higher CO2 emissions under fertilization (Lam et al., 2011; Yadav and Wang, 2017).

The measured GPPs and NEEs during the summer did not correlate to PAR or temperature while It has been proved that GPP correlates to PAR and TER corelates to temperature, respectively (Marushchak, 2014; Eckhardt et al., 2019; Lind et al., 2020). Actually, there was not any correlation between GHG fluxes including N2O and CH4 to any environmental variables (i.e. air temperature, soil volumetric water content etc.) or plant variables (i.e. LAI and plant height, etc.) during the period. The possible reasons might be, firstly, the small sample size which only comprise three-month data of summer was lack of change (i.e. temperature seasonal switch) and short to catch any correlation. In addition, the summer was dry thus affected the correlation and also the results. Last, the disruptions of harvest also interfere with the results.