The biogeochemistry of phosphorus in a temperate coastal lagoon afflicted by recurring cyanobacterial blooms

Estuaries and coastal waters are highly productive ecosystems that provide critical habitats and nurseries for many marine species in additional to other environmental and economic values. Recent research all points to changes in anthropogenic input of nutrients to estuaries as the primary cause of increasing eutrophication. The role of nitrogen (N) and phosphorus (P) as the primary agents of eutrophication and the link to the formation of harmful algal blooms is increasingly being studied. Studies into P dynamics show that P is not uniformly mixed within an ecosystem, but rather partitioned into ‘pools’ (eg. Dissolved within the water column/pore water, and buried within the solid phase). These pools of P exhibit vastly different dynamics and as such offer differing degrees of bioavailbility/lability with only the highly labile pools observed to be contributing the estuarine eutrophication. It is well known that phosphorus within the sediment is closely associated with Fe oxides of aquatic systems and that this Fe associated pool can be released with the onset of anoxic conditions and increased organic matter loading. Very few studies have, however, quantified this pool size and dynamic in estuarine systems and how this relates to cyanobacterial bloom dynamics, which are typically P limited. This research investigated the biogeochemical processes fuel the significant cyanobacterial blooms that occur within the Gippsland Lakes, Australia, a coastal lagoonal system prone to extended periods of anoxia and recurrent diazotrophic cyanobacterial blooms (Nodularia spumigena) through two means: (1) a whole system nutrient budget model that incorporates seasonal effects of nutrient inputs and the resultant impacts on the development of summer time cyanobacterial blooms, and (2) an in-depth investigation into the large fluxes of P released from sediments during summer anoxia, through a series of intact sediment core experiments aimed at identifying the highly labile and bioavailable components of the sedimentary Fe and P cycle. The whole system nutrient budget model (LOICZ) of the Gippsland Lakes was carried out from the 1st June 2010 through to 28th March 2012. The LOICZ budget model (LBM) highlighted a number of key features in the cycling of nutrients within the Gippsland Lakes leading up to and during a significant cyanobacterial bloom. These include a flood event (21st July 2011 – 18th August 2011) during which large loads of both Total Nitrogen (TN - 76.1 Mmol) and Total Phosphorus (TP - 3.4 Mmol) were input into the Gippsland Lakes. In the post flood period (19th August 2011 – 20th November 2011) increased nutrient concentrations, surface water temperatures and light availability favoured phytoplankton growth and both diatom and dinoflagellate species began to bloom, lowering the dissolved inorganic nitrogen (DIN) and dissolved inorganic phosphorus (DIP) concentrations within the surface waters. Following the mixed diatom-dinoflagellate bloom, water column stratification increased and respiration in the bottom waters caused dissolved oxygen concentrations to decline, creating hypoxic/anoxic bottom waters. Upwelling of DIP (0.83 Mmol) across the pycnocline from the nutrient rich bottom waters to the surface waters, occurred late in the post flood period. This upwelling further drove the development of conditions within the surface waters that were favourable to significant diazotrophic cyanobacterial growth (1:1.3 DIN:DIP). During the initial cyanobacterial bloom period (21st November 2011 – 14th December 2011) N-fixation rates of 8.6 Mmol were calculated for the initial growth period and accounted for 23% of the annual (yearly) input and 196 % of the summer riverine TN input into Lake King. Overall, the Gippsland Lakes were found to be a sink of nutrients across the entire study period, with a calculated annual net retention of both TN 8 % (230 tonnes) and TP 5 % (18 tonnes). The budget model also revealed a large gap in the P budget owing to a significant internal source of P (P released from the sediments ~2.5 Mmol). The internal P flux represented 20 % of the calculated annual riverine input of P from the catchment. The storage and release of P from the sediments was further investigated using a sequential extraction scheme that used ascorbate to extract P and Fe associated with the reactive Fe oxide phase. The ascorbate extraction revealed deep pools of the Fe-bound P within the sediment of up to ~5 mol-P g-1 dry sediment down to ~20 cm depth. This pool of P rapidly decreased with the onset of bottom water anoxia, leading to a release of 2.80 mmol P m-2 d-1 for a period of 105 days. Integrated over the periodically anoxic portion (>7 m depth contour, equating to a surface area of 5 km2) of the lake, this equated to a total P release of 1.4 Mmol. Upon re-oxygenation of the bottom waters, a rapid partial regeneration (14-27%) of the deep Fe-bound P pool was observed in intact core experiments. Extraction analysis performed on suspended sediment entering the Gippsland Lakes indicated that ~60 % of the P associated with particulate material could be considered bioavailable to primary producers (extracted by MgCl2 and ascorbate). Combining the extraction analysis from the suspended sediment with the loads of TP calculated by the LBM, the Gippsland Lakes could receive up to ~11.5 Mmol yr-1 (~6.9 Mmol bioavailable P yr 1), approximately twice the amount of P released from the sediment. We speculate however, that most of this catchment derived P transits the lakes, as it is delivered during flood events. A re-oxidation experiment was undertaken using intact sediment cores collected during bottom water anoxia. The water column was subsequently re-oxygenated and cores were extracted using the modified extraction scheme to identify Fe and P concentrations for a period of 40 days. The re-oxidation experiment revealed the presence of the common polychaete worm Capitella capitata within the sediments at the Lake King North study site in densities of 2900 ± 600 individuals m-2. This experiment confirmed the regeneration of the Fe-bound P observed in previously in-situ and suggested a link between the bioirrigating behaviour of C. capitata and the formation of deep sediment pools of P. A further experiment using a two dimensional oxygen sensor (planar optode) showed flushing events by C. capitata could deliver dissolved oxygen concentrations to a sediment depth of up to 4 cm. An inert tracer experiment was used to quantify C. capitata bioirrigation rates (5.9 38.3 L m 2 d-1) and indicated that irrigation could extent to depths greater than observed oxygen (O2) penetration (potentially 5-10cm). I speculate that the production of Nitrate (NO3-) in the presence of O2 and the subsequent transport to depths greater than observed O2 penetration (via bioirrigation) can mediate the regeneration of the reactive iron oxides and subsequent Fe-bound P through the microbial driven Fe(II) oxidation by NO3-. Multiplying the bioirrigation rates and the average pore water P concentrations in the top 5 cm of sediment (98 µmol L-1, based on data from sediment core 16th February 2012), yields fluxes of P ranging from 0.58 – 3.78 mmol m-2 d-1, which encompasses the flux of 2.80 mmol m-2 d-1 calculated via the mass balance. Overall, C. capitata appear to play a significant role in the Gippsland Lake sediments by acting as biogeochemical amplifiers, mediating the enhancement of both the release of the bioavailable Fe-bound P during anoxia and the subsequent rapid regeneration during oxygenated conditions. This study underscores the importance of internal P cycling in sediments colonised by deep irrigating fauna. It allows highlights the potential negative feedback of P released during water column anoxia due to the cessation of bioirrigation, resulting in optimum nutrient conditions in the water column for diazotrophic cyanobacterial growth