According to the correlation analyses, soil CH4flux had significantly high positive correlation with soil moisture over the whole study period (r=0.65 ***, figure 11). In other words, increases in soil moisture levels subsequently led to significant decreases in soil CH4 sink over the study period. Correlation between CH4 flux and soil temperature, however, was small and statistically insignificant (r=0.016 ns), indicating that changes in soil temperature did not lead to a significant change in CH4 sink. Because the correlation between soil moisture and soil temperature was r<|0.70| (r=0.29 ***), they were not considered multicollinear. In the treatment scales over the whole study period, correlations between CH4 flux and soil moisture were highly significant (***) in both main plots (C: r=0.37 , I: r=0.41) and in all subtreatment groups with the exception of IE (**), CA, CO and CT (ns, figure 12). In contrast to all other subtreatment groups, CT and IE had negative correlations (CT r=-0.12 and IE r=-0.46 **), indicating that increases in soil moisture led to increases in CH4 sink in those groups. The highest correlations in the irrigation site were in IT (r=0.6 ***) and CE in the control site (r=0.81 ***). In terms of soil temperature, both main plots had negative correlations (C: r=-0.45 ***, I: r=-0.18 **) which indicated a higher positive dependence of CH4 sink on soil temperature in the control site (figure 12). In the subtreatment level, soil
Figure 11: Correlation matrix of the measured variables. Correlation analyses were carried out using Pear-son’s correlation coefficient. Methane flux (lflux_ch4, µg m−2 h−1) was significantly positively correlated with soil moisture (moist, VWC %), or in other words, soil CH4 uptake rate correlated negatively with soil moisture. Correlation between soil temperature (temp, °C) and CH4 uptake was not significant. Soil temperature and soil moisture were also significantly positively correlated.
temperature correlated significantly with CH4 flux only in CA, CO and CT, meaning that measured CH4 uptake rate in CE and all subtreatment groups in the irrigation site did not significantly follow changes in soil temperature and vice versa.
Correlation between CH4 flux and soil moisture in the exclusion data set was highly signif-icant but somewhat lower than in the pooled data with all subtreatments included (r=0.4
***). Soil moisture in both layers had significant positive correlations with CH4 flux, and negative ones with CH4 sink, in the scale of the whole study site, but when divided into main plots moisture in both layers had significant negative correlations with flux (mineral r=-0.46
**, organic r=-0.33 *) in the irrigation site, the latter indicating an increase in CH4 sink with an increase in soil moisture (figures 12 and 13). The opposite was true for the control site, in which soil moisture had high positive correlations (mineral: r=0.82 ***, organic:
r=0.68 ***), indicating decreases in sink with increases in soil moisture. In contrast to soil
Figure 12: Correlations and frequency densities of the measured variables with linear regression trends in main plots (A and C) and subtreatment groups (B and D) over the whole study period. The plots include lines fit to linear regression at 95% confidence level. Soil moisture had high correlations with CH4 uptake rate in both main plots and nearly all subtreatment groups. Soil temperature in general was not significantly correlated with CH4 uptake rate but had higher correlations in the control site than in the irrigation site in both main plot and subtreatment level, with the exception of CE. Correlations between soil temperature and flux were negative in all except IA and IE, indicating decreases in sink by increases in soil temperature in the latter groups.
moisture, soil temperature had higher overall correlations with CH4 flux than in the pooled data (r=0.19 *). However, temperature only had a significant correlation in organic layers (r=0.24 *). In the main plot level, soil temperature had relatively low negative correlations with CH4 flux, or in other words low positive correlations with CH4 sink, in both layers in the control site (ns), while the corresponding values in the irrigation site were both positive and higher (mineral r=0.31 and organic r=0.29), indicating negative correlations with CH4
sink. In the soil layer data, correlation between soil temperature and soil moisture was higher than in the pooled data (r=0.42 ***). See the full correlation matrix of soil layer data in Appendix H.
In the temporal scale, soil moisture had positive and relatively high correlations with flux
Figure 13: Correlations and frequency densities of the measured variables in the whole study site (A and C) and main plots (B and D) over the whole study period, divided into organic and mineral layers from CE and IE subtreatments. The plots include lines fit to linear regression at 95% confidence level. Soil moisture had significantly high positive correlation with CH4 flux in general, but also in both layers with higher correlation in the organic layer. In other words, soil moisture correlated negatively with CH4 sink.
Correlation in the control site was higher and positive while irrigation site had a lower negative correlation, indicating strong decreases in sink with increases in soil moisture in control site and weak increases in sink with increases in soil moisture in irrigation site. In both sites, soil moisture in mineral layers had higher correlations with higher significance than in organic layer. Soil temperature had a significant and relatively low correlation with CH4flux in the whole study site, temperature in organic layers having higher correlation and significance. Correlations were also higher in the irrigation site in both layers.
in general, highest correlations occurring in July in both main plots (whole site r=0.75
***, C: r=0.43 ***, I: r=0.49 ***). In other words, soil moisture correlated negatively with CH4 uptake rate every month. Between the main plots, irrigation site had higher correlations in all months except in August. In the subtreatment level, there was more variation in the correlations between soil moisture and CH4 sink along months (figure 14).
In contrast to other treatments, IE had high negative correlations between soil moisture and flux in all months, ranging from -0.59 to -0.73, also indicating high positive correlations with CH4 sink. The strong negative correlations of IE held true also in both its organic and mineral layers every month, ranging from -0.86 to -0.37 (organic) and -0.59 to -0.71 (mineral) but were mostly insignificant or of low significance. Highest correlations of organic layers in IE occurred in September (r=-0.86 *) and August in mineral layers (r=-0.71 **).
Also, correlations with flux were negative in CT in June, August and September, the latter having almost perfect correlation of -0.99 (*) and all of them showing very strong positive correlations with CH4 sink. The positive correlation trends between soil moisture and flux in CE were consistent also in both its organic and mineral layers, the coefficients of which ranged from 0.37 to 0.93 (organic) and 0.66 to 0.96 (mineral), the mineral layer having significantly stronger correlations than the organic layer. The strongest correlations of organic layers in CE were found in September (0.93 **) and those of mineral layers in July (0.96 ***), both indicating strong negative correlations between soil moisture and CH4 sink.
Soil temperature had low, mostly statistically insignificant and negative correlations with CH4 flux during the study period, indicating low positive correlations with CH4 sink. The highest correlations between soil temperature and flux in the whole study site occurred in July (r=0.19 *) while in the control site highest correlations were in August (r=-0.6 ***) and in the irrigation site in September (r=-0.56 **). However, as with soil moisture, there was more temporal variation in the subtreatment level (figure 14). In general, there were more positive correlations between temperature and flux in June in nearly all subtreatment groups, after which they turned mostly negative until August. In other words, soil temperature correlated mostly negatively with CH4 sink in June and positively in July and August. IE had low positive correlations between temperature and flux every month while those in CO remained negative, meaning that soil temperature in IE correlated negatively with CH4 sink and in CO positively. Both layers in IE had very weak insignificant correlations every month, mineral layer showing stronger correlations ranging from 0.06 to 0.47 while those of organic layer ranged from 0.12 to 0.36, both indicating negative correlations with uptake rates. The highest correlations were observed in the control site and especially in CT and CO which had minimum correlations of |0.32|. Similar trend was observed also in both layers of CE, except in organic layers in September (0.91 *). The correlations of CE layers ranged from 0.15 to -0.91 in organic and 0.26 to -0.77 in mineral layer, highest correlations of both layers found in September. See temporal correlations of soil layers in Appendix H.
Figure 14: Correlations between the variables, divided into months by main plots (A and B) and subtreatment groups (C and D). The plots include lines fit to linear regression at 95% confidence level. Soil moisture correlated strongly with flux in all months, especially in the irrigation site. In both main plots, an increase in soil moisture led to decrease in uptake rate every month. Correlations were lowest in September and highest in July in both sites. In the subtreatment level, there was more variation in the temporal correlation trends. Soil temperature had generally low negative correlations with flux with low or no significance in all months, irrigation site having higher correlations in June and July, after which soil temperature in control site had stronger correlations until September. Correlation in the whole study site was highest in July and lowest in August. Control site in general had more variation in the correlation trends between soil temperature and flux and thus soil temperature and uptake rate.