Carbon Fluxes Over an Even-Aged Pure Masson pine (Pinus Massoniana) Stand

: Masson pine ( Pinus massoniana ) is a tree species widely planted in central and south 12 China. In the present pioneer study, we reported about our two years of carbon flux 13 observations over an even-aged pure Masson pine stand. Light intensity could explain nearly 14 half (47%) of the variance in daytime net ecosystem exchange (NEE). Daytime dark respiration 15 was lower than nighttime NEE, suggesting a possible effect of light inhibition on respiration. 16 The mean annual NEE was -557 g C m -2 yr -1 , which indicated that this stand is a medium to 17 large carbon sink. This NEE estimates were defensive because we checked the data with 18 thorough quality controls and in consistency with previous independent estimates. An 19 unexpected seasonal pattern of NEE was observed with a clear reduction around the 20 transitional period between summer and autumn (around July). This NEE reduction is 21 probably a consequence of water stress induced stomatal control, and not of a decrease in light 22 intensity. The updated optimal stomatal theory did not provide the best description of 23 stomatal control in relation to photosynthesis. Whether this is a new emergent property of the 24 ecosystem scale needs further investigation.


45
The eddy covariance technique, which is based on the theory of turbulent mass and 46 energy transport, became popular in the 1990s [6] . This method could overcome the deficiency   4 adopted the open-path EC driving by solar energy for the study. We used the open-path EC 70 model EC150 (Campbell Scientific Inc., Logan, UT, USA), which was installed at the height of 71 30 m above the ground. The sampling frequency for EC was the usual 10 Hz, which was 72 controlled by a data logger (model CR3000, Campbell Scientific, USA).

73
The profile system can be separated into two parts, the above-and below-ground profile.

74
For the above-ground profile, we measured the humidity and temperature (sensor type: 75 HMP45C, Campbell Scientific Inc., USA), wind speed (sensor type: 014A-L, Met One,

81
A radiometer that accounts for downward and upward short-wave and long-wave 82 radiation was used to measure and derive net radiation (Rn) (sensor type: CNR2, Apogee, 83 USA). The soil heat flux (G) was measured using a plate flux sensor (sensor type: HFT3,

88
were calculated as covariance between vertical wind velocity and target scalars with an 89 averaging period of 30 min [7] . We carried out coordinate rotation and WPL corrections to the 90 flux data with the source code written by Xuhui Lee of Yale University. Net ecosystem 91 exchange (NEE), which indicates the carbon sink or source strength for a target ecosystem, 92 was calculated as the sum of eddy transferred carbon flux (Fc) and storage flux (Fs). Fs was 93 calculated using CO2 concentration recordings [8] .

94
We discarded the NEE over 30 μmol m -2 s -1 or under -50 μmol m -2 s -1 as well as latent heat 95 flux over 800 wm -2 and under -200 w m -2 as hard spikes. After removing the hard spikes, we 96 adopted a rule in which values more than three times higher than the 95% confidential interval 97 at the same time point were considered outliers, with a window size of 10 days.

98
The calm air-related nighttime flux underestimation was corrected using the u* filtering 99 method (u* indicates friction velocity) [9] . As shown in the results section, we adopted u* 100 threshold value of 0.25 m s -1 for our study. We omitted all values on NEE below this threshold 101 value for both daytime and nighttime flux. The present study differed from previous studies 102 because they have only considered nighttime flux u* filtering.

103
In order to obtain the defensible annual sum for NEE, we filled the data gap caused by 104 instrument failure, spike detection/omission, and u* filtering with nonlinear regression. We 105 adopted a modified daytime data model from Lasslop and a nighttime data model from van't 106 Hoff [10,11] . The van't Hoff model is expressed as follows: 107 108 where A and B are fitted parameters, and T is temperature, here specified to air  where α is apparent quantum yield, Pm is light saturated photosynthesis rate, I is light 113 intensity (in this case specified to solar radiation), and Rd is dark respiration. In the original 5 version of Lasslop [10] , they related Pm to water vapor deficit (D) to exhibit a decline in Pm 115 above 1 kPa. We excluded this term for two major reasons. The first was that very few data 116 points were observed under D above 1 kPa. In addition, the inclusion of this term has limited 117 impact on the fitted results (cf. Figure s1). The second reason was that introducing one more 118 variable will inevitably lead to numerical instability in curve fitting. Finally, we modified Clearly, this is a combination of a rectangular hyperbola and van't Hoff model. In general, 122 this model was numerically stable when applied to our datasets.

123
In order to calculate gross primary production (GPP), we adopted the daytime flux 124 partition method, as suggested by Lasslop [10] . The GPP was calculated as:

126
where Rd is dark respiration derived from the light response, which equals to .

129
(for details see cf [12] ). Because the studied forest with a stand age of 30 years had a closed 130 canopy, we selected the dry canopy Gs (determined in observations when there was no 131 rainfall in the past three hours) to calculate bulk stomatal conductance [13] .

132
According to the Ball-Berry stomatal model [14] , GPP can be expressed as a function of Gs

141
The optimal stomatal theory has suggested the following relationship [16] :

143
where G0 and G1 are the fitted parameters. The term is always >1, and it tends to 144 dominate the term [16] . Therefore, GPP can be expressed as:

146
where k and b are linear parameters, as shown above.

147
We also calculated the Jarvis and McNaughton decoupling factors (Ω) [17] . The Ω varied 148 between 0 and 1. When Ω approached 1, canopy was fully decoupled from overhead 149 atmosphere, and evapotranspiration was dominated by available energy.

150
Notably, the decoupling factor could be related to the crop water stress index (CWSI) 151 proposed by Jackson [18] , in case Gs approximates infinity (cf. detailed equation

179
The manner in which nighttime NEE increases with u* can be roughly expressed by an 180 equation that describes exponential growth to maximum (Figure 4). In the present study, the 181 nighttime NEE (respiration rate) stabilized at approximately 2 μmol m -2 s -1 . Sixteen bins were 182 distributed as u* <0.25 m s -1 for a total of 37 bins. This indicated that 43.24% of the nighttime 183 data were omitted through u*-filtering. The calm air-induced NEE underestimation can occur 184 irrespective of daytime or nighttime; thus, we also removed daytime NEE values when u* was 185 below this threshold.

186
Light intensity is the dominant controller in daytime NEE, similar to that found in leaf 187 photosynthesis (Table 1 and Figure 5). In contrast to leaf levels, the dark respiration term at the 188 ecosystem scale was significantly larger, which includes respiration terms from leaf, branch, 189 stem, root, and soil organic matter. As enzyme-catalyzed processes, such as respiration, have 190 strong temperature dependency, we included the van't Hoff model into a regular 191 photosynthesis light response model (Equation 3). We predicted that this could improve the 192 goodness-of-fit, but not to a large extent (see the determinant coefficient listed in Table 1). The daytime-based respiration. This might be a consequence of the Kok effect, which is related to 208 light inhibition of ecosystem respiration and has been reported previously [19] .

209
The NEE values after u*-filtering and gap filling are shown in Figure 8. During this study,

235
This classification based on growth stages differed from that based on successional stages, in 236 which 10 years old stands are classified as young stands, 10-40 years old stands are classified 237 as middle-aged stands, and 40+ years old stands are classified as mature forests [21] . In case of 238 even-aged pure plantations, the classification based on growth stages might be more suitable 239 than that based on successional stages.

240
The studied plantation is at its middle-age growth stage (ca. 22 years old when the eddy 241 flux measuring started). The average NEE was -557 g C m -2 yr -1 , which indicated that the 242 plantation is a medium to large carbon sink (Figure 8). This value was very close to that 243 reported in a 20-year-old mixed coniferous plantation (553 to 645 g C m -2 yr -1 ) in Qianyanzhou 244 (QYZ), Jiangxi, China [22] . This mixed coniferous plantation comprised the slash pine, Masson 245 pine, and Chinese fir. Ma compared the vegetation carbon storage and biomass increase 246 between exotic slash pine and Masson pine stands and found no significant differences [23] . Our 247 results are also very close to that of the study on 100-year-old mixed pine and broadleaved

280
We wanted to investigate the possible reasons for the reduction in NEE around July. As 281 NEE is the net balance between GPP and ecosystem respiration, the reduction in NEE could be 282 contributed by either reduction in GPP or by increase in ecosystem respiration. As listed in 283 Table 1 and shown in Figure 8, a sharp increase in daytime based dark respiration (Rd) or 284 night-based ecosystem respiration was not detected around July. In contrast, there was a sharp 285 reduction in the light-saturated photosynthesis rate (Table 1) and in daily peak NEE (Figure 8)

286
around July. Thus, the reduction in NEE was most likely caused by declining GPP.

287
We illustrated that light intensity ( Figure 6) and stomatal conductance (Figure 10) are two 288 major factors controlling ecosystem photosynthesis in the studied stand. After examining the 289 light intensity (Rg) (Figure s2), we excluded it as a major controller causing the NEE reduction 9 around July. Notably, there was a clear reduction in Ω during the period around July in both 291 2012 and 2013. The term 1-Ω is an approximation of the crop water stress index [18] . The 292 reduction in Ω indicated an increase in water stress (1-Ω). The value of Ω is derived based on 293 the energy flux components, which are independent of carbon fluxes. The synchronization of 294 1-Ω and NEE suggested that water stress-related stomatal reduction might be the critical 295 reason for NEE reduction around July. This idea also received support from the results of a 296 study conducted in QYZ site in which the researchers attributed carbon sequestration 297 reduction to summer drought [22] . It should be noted that the soil water content did not exhibit 298 the lowest value around July (Figure s3). In other words, we were not able to detect ecosystem 299 water stress through soil water observations. We thus recommend the 1-Ω (or CWSI) for water 300 stress analysis at the eddy flux site.

322
The way in which Gs influences photosynthesis at the ecosystem level also needs special 323 attention. In the present study, the Gs control on GPP predicted by the optimal stomatal 324 theory was better than that predicted by the Ball-Berry model, but worse than that predicted 325 by the Leuning model (Figure 10). This result was unexpected because the optimal stomatal 326 theory we used was regarded as an updated version of the Leuning model [16] . It is still not clear 327 whether this is a common case or a case specific for the studied stand. Given that this is a

357
The mean value (black circles) and standard deviation (error bars) of each bin were calculated. An exponential growth to maximum equation was fitted to the dataset.  there are variants of Pinus massoniana which were also not included in the study. The study site is indicated by a red star. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.   Dependence of nighttime net ecosystem exchange (NEE) on friction velocity (u*). The data were binned as the same number of data points per bin after sorting as u*. The mean value (black circles) and standard deviation (error bars) of each bin were calculated. An exponential growth to maximum equation was tted to the dataset.  Observed and simulated daytime net ecosystem exchanges (NEE) for the investigated Masson pine evenaged pure stand. Kernel density plot was used to show the distribution of data points. Linear regression was performed as shown by the white line.    Gross primary production (GPP) expressed as a function of surface conductance (Gs) and water vapor de cit (D) or relative humidity (RH). GPP is expressed in μmol m-2 s-1, Gs is expressed in m s-1, Ca is the ambient carbon dioxide concentration and is expressed in ppm, RH is expressed in %, and D is expressed in kPa. Linear regression was applied to the dataset, which is indicated by the black line ((a) y=122.6700+5.8480, r2=0.1967; (b) y=4.6969x+4.6071,r2=0.4148; (c) y=3.9000x+4.9045, r2=0.3235) .