Discussion

Results from both the field experiments and the global meta-analysis support the substrate competition theory as a predominant mechanism explaining P alleviation of N-suppression of CH4 uptake in global grassland. A number of studies have concluded that soil NH4+ content is a major driver of CH4 uptake suppression in drylands (Dunfield & Knowles 1995; Gulledge & Schimel 1998). The toxic impact of nitrite on methanotrophic activity might be another reason for this suppression (Dunfield & Knowles 1995). In our field site, however, the toxic impact of nitrite on methanotrophic bacteria (Schnell & King 1994) most likely did not contribute to the inhibited CH4 uptake because no nitrite was detected in the soil, a widespread phenomenon in semi-arid grasslands (Holst et al. 2007; Giese et al.2013; Zhang et al. 2017). Therefore, we conclude that substrate competition remains the best theory to explain the P alleviation of N-suppression of CH4 oxidation.
A recent study in tropical forest postulated that P mitigation of the N-suppressed CH4 oxidation might due to P stimulation of methanotrophic activities, although it further concluded it might be a minor contribution due to the trivial fraction of microbial biomass taken in by CH4 oxidation bacteria (Zhang et al.2011). This mechanism was postulated, but not supported by empirical data (Zhang et al. 2011). Enhanced methanotrophic activity by P-enrichment is thus one possible mechanism, yet it remains to be empirically confirmed as a predominant mechanism. More molecular and genomic analysis of the linkages between microbial mechanisms and ecosystem function remain needed to mechanistically elucidate and predict the CH4 oxidation in a changing environment (Xuet al. 2016). Combined, the field experiment, meta-analysis, and global empirical modeling analysis unambiguously demonstrate that P-stimulated soil N depletion mitigate the N-induced reduction of CH4 uptake.
Both the global meta-analysis and modeling results indicate a widespread impact of N and P on CH4 uptake in grassland. Summarizing all grids with treatment impacts compared with ambient grids, we found that N suppression of CH4 uptake occurs in over 90% of global grasslands; the pervasive N suppression is consistent with reported N suppression of CH4 uptake in many ecosystems (Mosier et al. 2003; Chen et al. 2013; Liuet al. 2017; Zhang et al. 2017). Model results suggest that P alleviation of N-suppression of CH4 uptake occurs in over 89% of the global grasslands (Fig. 4 ). Considering the limited studies on P impacts on CH4 uptake (Veraartet al. 2015), more field experiments on this aspect are needed. Estimates of the global grassland sink for CH4 have been made in a large number of studies, yielding a broad range of 1.9 - 9.3 Tg CH4 yr-1 (Potter et al.1996; Ridgwell et al. 1999; Curry 2007; Zhuang et al.2013; Yu et al. 2017). Our result (5.9 Tg CH4y-1) falls within the range of previous studies. According to our empirical model, increased N deposition reduced the grassland CH4 sink by 11%, which was similar to another modeling study (10%) (Zhuang et al. 2013), but lower than the value reported in a data synthesis study (38%) (Liu & Greaver 2009).
The effects of N and P depositions on CH4 uptake were quantified, yet a few issues should be paid attention to when interpreting the results. First, we quantified the potential CH4 patterns under four N and P deposition scenarios, but did not consider the contributions of other changing environmental factors, such as increasing atmospheric CO2concentrations, and shifting land management practices, which have been shown to be important determinants of CH4 flux (Xuet al. 2010). Second, our global-scale estimates of CH4 uptake were based on a simple empirical model. Intensive data-model integration and model- model inter-comparisons are still needed to better quantify the uncertainties in these CH4 budgets. Third, the meta-analysis was carried based on the most comprehensive dataset for the impacts of N and P additions on CH4 flux in global grassland; however, the uneven distribution of the field observational dataset might cause biases in the global analysis of nutrient impacts. This unequal spatial distribution of field observations exist for eddy covariance towers (Baldocchi et al. 2001), vegetation data (Kattge et al.2011), and soil data (Xu et al. 2013). This problem needs to be resolved for reaching a more accurate global extrapolation and increasing our understanding of the functioning of terrestrial ecosystems. Deposition and addition of N and P take place in many forms, including inorganic and organic, dry and wet deposition. Due to limited data availability (Mahowald et al. 2008) and field experiments (Wang et al. 2015), the impacts of various N and P forms in atmospheric depositions could not be fully investigated in the present study. Understanding the impacts of different N and P forms on CH4 oxidation still remains to be fully investigated.