Transferring network analysis to the study of potential biogeochemical interactions of phosphorus-relevant elements in floodplain subsoils - A new use case for the Soilscape Network Approach (SNAp)

Abstract

Subsurface phosphorus (P) loss from deep P stocks in floodplain subsoils can contribute to eutrophication of freshwaters. To date, knowledge on the complex biogeochemical interactions of P in floodplain subsoils is too scarce to enable targeted P management to mitigate subsurface P loss from deep P stocks. We propose using graph theory and the Soilscape Network Approach (SNAp) based on correlations between P-relevant elements to study these complex biogeochemical interactions in the soilscape. Complex interactions of several elements in soils are difficult to investigate from a holistic perspective with conventional data analysis. We translated soil element data from topsoils and subsoils of terrestrial sites, proximal and distal floodplain sites into relational data and analyzed network structure, centrality, and modularity. The results indicate that a higher frequency of groundwater level fluctuations in distal subsoils and proximal topsoils could result in 24-44 % less biogeochemical interaction compared to sites with stable conditions. Impeded microbial processes on the frequently disturbed sites may explain this finding. Our analyses suggest biogeochemical differences between floodplain topsoils and subsoils expressed in 24 % lower and 75 % higher network connectivity in distal and proximal subsoils (respectively). We also found 22 % lower network connectivity in distal than proximal floodplain subsoils, suggesting biogeochemical differences between both soil sections. These findings imply that floodplain P management should not take a whole-floodplain approach but a 3D-approach, which differentiates laterally between floodplain zones and vertically between soil sections. In addition, SNAp indicated that Fe(II) oxides are important in P biogeochemistry of floodplain subsoils but are not the key element. Instead, labile P forms are suggested to have different major associations in distal (Al_ox, Fe_ox) versus proximal deep P stocks (Al_ox, Mn, Ca). Our study provides new insights into the biogeochemistry of deep P stocks in floodplain subsoils which require targeted validation by other methods.

Keywords: Correlation network; Deep phosphorus stock; Eutrophication; Graph theory; Phosphorus loss; Wetland soil.