Biogeochemistry DOI 10.1007/s10533-013-9944-z SYNTHESIS AND EMERGING IDEAS Lack of steady-state in the global biogeochemical Si cycle: emerging evidence from lake Si sequestration Patrick J. Frings • Wim Clymans • Erik Jeppesen • Torben L. Lauridsen • Eric Struyf • Daniel J. Conley Received: 4 September 2013 / Accepted: 17 December 2013 Ó Springer Science+Business Media Dordrecht 2014 Abstract Weathering of silicate minerals releases lentic systems accumulate Si via biological conversion dissolved silicate (DSi) to the soil-vegetation system. of DSi to biogenic silica (BSi). An analysis of new and Accumulation and recycling of this DSi by terrestrial published data for nearly 700 systems is presented to ecosystems creates a pool of reactive Si on the assess their contribution to the accumulating conti- continents that buffers DSi export to the ocean. nental pool. Surface sediment BSi concentrations Human perturbations to the functioning of the buffer (n = 692) vary between zero and [60 % SiO2 by have been a recent research focus, yet a common weight, apparently independently of lake size, location assumption is that the continental Si cycle is at steady- or water chemistry. Using sediment core BSi accu- state. However, we have no good idea of the mulation rates (n = 109), still no relationships are timescales of ecosystem Si pool equilibration with found with lake or catchment parameters. However, their environments. A review of modelling and issues associated with single-core accumulation rates geochemical considerations suggests the modern con- should in any case preclude their use in elemental tinental Si cycle is in fact characterised in the long- accumulation calculations. Based on lake/reservoir term by an active accumulation of reactive Si, at least mass-balances (n = 34), our best global-scale esti- partially attributable to lakes and reservoirs. These mate of combined lake and reservoir Si retention is 1.53 TMol year-1, or 21–27 % of river DSi export. Again, no scalable relationships are apparent, sug- gesting Si retention is a complex process that varies from catchment to catchment. The lake Si sink has Responsible Editor: Cory Cleveland implications for estimation of weathering flux gener- ation from river chemistry. The size of the total P. J. Frings (&) W. Clymans D. J. Conley Department of Geology,Á Lund University,Á So¨lvegaten 12, continental Si pool is poorly constrained, as is its 22362 Lund, Sweden accumulation rate, but lakes clearly contribute sub- e-mail: [email protected] stantially. A corollary to this emerging understanding is that the flux and isotopic composition of DSi E. Jeppesen T. L. Lauridsen Department ofÁ Bioscience, Freshwater Ecology, Vejlsøvej delivered to the ocean has likely varied over time, 25, 8600 Silkeborg, Denmark partly mediated by a fluctuating continental pool, including in lakes. E. Struyf Department of Biology, Research Group Ecosystem Keywords Silica cycle Biogenic silica Lake Management, University of Antwerp, Universiteitsplein 1, Á Á 2610 Wilrijk, Belgium retention Silicon isotopes Á 123 Biogeochemistry Introduction corresponding to *40 % of oceanic primary produc- tion (Yool and Tyrrell 2003). After death, the BSi is an The global biogeochemical Si cycle is characterised efficient exporter of organic C to the deep ocean. by two discrete sub-cycles of comparable magnitude Annually, *3 % escapes dissolution and is perma- on the continents and in the oceans. Rivers are the nently removed from the ocean sub-cycle (Tre´guer principal link between these sub-cycles, supplying and De La Rocha 2013). *85 % of annual inputs of dissolved silicate (DSi) to These two linkages—silicate weathering and BSi the ocean, after its mobilisation from terrestrial soils export in the ocean—tightly couple the Si and C and bedrock. The long-term Si cycle consists of (1) cycles. DSi release from bedrock and subsequent release of dissolved silicate (DSi) and particulate transport to the ocean are key elements in the Earth silicates from regolith at the Earth’s surface, (2) System. They are commonly assumed to be equivalent cycling and partial transfer of DSi and particulates and are typically assessed through measurement of from land to ocean and (3) permanent burial in ocean stream DSi concentrations and discharge. Yet such an sediments (see Struyf et al. 2009a and Tre´guer and de assessment assumes steady-state conditions in the la Rocha 2013 for reviews). All stages are driven by continental Si cycle. On human (\100 years) time- biological activity and include various reservoirs that scales, steady-state depends on the constancy of Si act to buffer or even temporarily reverse the long-term pools in the ecosystem filter, where different ecosys- unidirectional flux from land to ocean. tems have different weathering, internal recycling and Silicate weathering releases DSi and consumes CO2 release rates. An emerging paradigm asserts humans at a temperature dependent rate, thereby acting as a have perturbed the balance between these processes, negative feedback for atmospheric CO2 (Berner et al. and ecosystem Si pools are growing or depleting 1983). Weathering is strongly biologically mediated following land use change, agricultural Si export, through root exudation of organic acids, subsurface eutrophication or climate change (Clymans et al. pCO2 enrichment, physical breakage and facilitation 2011; Struyf et al. 2010a; Conley et al. 2008; Sommer of soil water flow (Berner 1992; Berner et al. 2005). A et al. 2013). We now know the modern Si cycle is substantial fraction of the mobilised DSi enters the highly perturbed, so steady-state should not be soil-vegetation system, where it contributes to the assumed in the Anthropocene. production of biogenic silica (BSi) structures in On longer timescales, the Si cycle would be in continental vegetation that convey structural and steady-state if permanent burial in the oceans equals ecological benefits (Epstein 1999). Continental vege- the sum of Si mobilised by weathering and erosion on tation BSi production is estimated at *84 9 1012 mol the continents. On these timescales, and distinct from Si year-1 (Carey and Fulweiler 2012) which is too anthropogenic activity, we hypothesise the continental large to be supported entirely from newly mobilised Si cycle is not at steady-state and instead is charac- DSi. Instead, it derives from recycling of a pool of terised by gradual aggradation or depletion of the ASi reactive amorphous Si (ASi) compounds (including pool held in continental soils, sediments and deposits, BSi) in ecosystem soils. This soil–vegetation Si pool is in response to changing environmental forcings. This orders of magnitude larger than annual Si mobilisation has implications for the ultimate flux of DSi from land and has been termed the ecosystem silica filter (Struyf to ocean and the strength of the silicate weathering- and Conley 2012). This filter buffers DSi export from pCO2 feedback. catchments such that actual Si export will equal Si In this contribution, we summarise evidence for mobilisation from bedrock only when the filter non-steady state functioning of the global Si cycle functions at steady-state. independent of human activity (‘‘An accumulating After cycling through continental ecosystems, continental Si pool’’). Meybeck and Vo¨ro¨smarty approximately 6 TMol DSi year-1 (1 TMol = 1 9 (2005) conceptualise the fluvial system from soils to 1012 mol) is delivered to the oceans via the fluvial estuaries as a series of filters that ‘recycle, store, system. Annual ocean BSi production of *240 TMol remobilise and transform’ material. Within this frame- year-1 (Tre´guer and De La Rocha 2013) is therefore work, we focus on a key element—lentic systems— similarly supported by recycling of previous produc- and present a meta-analysis of a compilation of lake/ tion. The diatoms dominate ocean BSi production, reservoir sediment BSi and associated catchment data 123 Biogeochemistry for nearly 700 systems to better understand where and especially in tectonically active areas (Allison et al. why Si is being sequestered (‘‘Lake and reservoir Si 1998). In particular, large river floodplains appear to accumulation: a meta-analysis’’). Our synthesis and buffer sediment delivery on timescales greater than interpretation of this dataset highlights uncertainties Quaternary environmental changes (Me´tivier and and flaws in current approaches and the assumption of Gaudemer 1999; Blo¨the and Korup 2013). Similarly, steady-state. We place our data in the context of our colluvial and aeolian sediments are accumulating in understanding of past and present Si cycling and many regions, particularly those with agriculturally demonstrate that river DSi fluxes are variable in disturbed sediment budgets (Trimble 1983). Reser- magnitude, isotopic composition and spatial distribu- voirs annually accumulate about 20 % of the global tion over a range of timescales (‘‘Lake Si accumula- river sediment flux delivered to the ocean by rivers tion: uncertainties and implications for the global Si (Syvitski et al. 2005) and the number of established cycle’’). Implications of a non-steady-state continental reservoirs are increasing markedly (Lehner et al. Si pool are highlighted and recommendations formu- 2011). Lakes and ponds presumably have a similar lated for future research priorities. function (Hay 1998; Houser et al. 2010). Such Throughout, we discriminate between particulate accumulation is not a realistic long-term situation. silicates and amorphous Si (ASi). Here, particulate