Global importance, patterns, and controls of dissolved silica retention in lakes and reservoirs

Lentic water bodies (lakes and reservoirs) offer favorable conditions for silica (SiO2) burial in sediments. Recent global estimates suggest that (1) lentic SiO2 trapping is a globally important SiO2 flux, and (2) through reservoir construction, humans have dramatically altered river dissolved SiO2 (DSi) transport and coastal DSi delivery. However, regional to global scale patterns and controls of DSi removal in lentic systems are poorly constrained. Here we use 27 published lake and reservoir DSi budgets to develop insights into patterns and controls of lentic DSi retention and to develop a new, spatially explicit, global model of lentic DSi removal called SiRReLa (Silica Retention in Reservoirs and Lakes). In our analysis, lentic DSi removal (kg SiO2 yr(-1)) was significantly and positively related to DSi loading (P < 0.0001; r(2) = 0.98), and DSi removal efficiency was significantly and positively related to water residence time (P < 0.0001; r(2) = 0.68). In addition, DSi settling rates were, on average, 6.5-fold higher in eutrophic systems than in non-eutrophic systems (median settling velocities: 11.1 and 1.7 m yr(-1) for eutrophic and non-eutrophic systems, respectively; P < 0.01). SiRReLa, which incorporates these insights, performed quite well in predicting both total DSi removal (kg SiO2 yr(-1); Nash Sutcliffe Efficiency (N.S.E) = 0.88) and DSi removal efficiency (% Si removed; N.S.E. = 0.75), with no detectable bias in the model. Global application of SiRReLa confirms that lentic systems are important sinks for DSi, removing 89.1 Tg DSi yr(-1) from watersheds globally, roughly 19-38% of all DSi inputs to surface waters. Small lakes and reservoirs (<50 km(2)) were critical in the analysis, retaining 81% (72 Tg DSi yr(-1)) of the globally retained total. Furthermore, although reservoirs occupy just 6% of the global lentic surface area, they retained approximately 35% of the total DSi removed by lentic systems. Regional hot spots for lentic DSi removal were identified and imply that lentic systems can remove the vast majority of DSi across a large fraction of Earth's land surface. Finally, a sensitivity analysis indicates that future improvements in DSi trapping and transport models should focus on improving estimates of DSi input to surface waters.