Why the phosphorus retention of lakes does not necessarily depend on the oxygen supply to their sediment surface

In order to improve the trophic state of Lake Sempach, a eutrophied lake in central Switzerland, its external phosphorus (P) load has been decreased and its hypolimnion has been artificially oxygenated to lower the lake-internal P recycling. Based on more than 15 yr of experience, we conclude that the reduction of the external P load resulted in a corresponding decrease of the lake’s P concentration. However, contrary to initial expectations, increased hypolimnetic dissolved oxygen concentrations neither (1) reduced the P release from sediments during summer nor (2) resulted in an increased permanent P retention. These observations warrant a reevaluation of the well-accepted management strategy of decreasing the lake internal P cycling by maintaining an aerobic hypolimnion and sediment surface. We hypothesize that oxygenation only results in an increased permanent benthic P burial if, because of the depressed sulfide production, more ferrous phosphate (e.g., vivianite) and less FeS is deposited in the anoxic sediment. Hence, it is not the oxic sediment surface that directly affects the permanent redox-dependent sediment P retention but the molar ratio of the available reactive Fe(II) : S22 : PO4 in the anoxic sediment. This ratio is driven by the settling rate and the nature of organic matter and particulate iron, as well as the supply of oxygen, nitrate, and sulfate to the sediment. It is well known and generally accepted that oxic lake sediments retain phosphorus (P) more efficiently than anoxic sediments; that is, formerly oxic sediments release vast amounts of phosphate when they become anoxic. Based on such observations, Einsele (1936a) and Mortimer (1941, 1971) developed the classical concept of the coupled cycling of iron (Fe) and P. According to that concept, under oxic conditions and at about neutral pH, iron precipitates as a trivalent iron oxyhydroxide that has a large capacity to coprecipitate and/or adsorb phosphate. Under anoxic conditions, the reductive dissolution of the solid Fe(III) oxyhydroxide results in a parallel dissolution of the previously bound phosphate. More recently, it has been proposed that benthic bacteria can supplement this abiotic redox-dependent fixation and release of P (Gächter and Meyer 1993; Hupfer et al. 1995). In the presence of O2, some facultative aerobic bacteria deposit P as polyphosphate in their cells but release it as ortho-P under anoxic conditions. Advanced wastewater treatment technology profits from both processes to enhance P removal from wastewater. At first glance, this traditional concept of aquatic sediment P biogeochemistry suggests lower P release rates from the sediment and hence lower P concentrations in the water column, provided that oxic conditions are permanently maintained at the sediment surface. However, based on their experience with artificial hypolimnetic oxygenation of Lake Sempach for more than 15 yr, Gächter and Wehrli (1998) concluded that artificial maintenance of oxic conditions in the hypolimnion did not affect P cycling at the sediment– water interface (Fig. 1). These results obtained from longterm, large-scale experiments, as well as considerable older literature showing that anoxia does not necessarily result in P release from sediments (e.g., Schindler et al. 1973, 1977; Levine et al. 1986), suggest that a modified concept of the redox-controlled P retention in lake sediments is needed. To our knowledge, Einsele (1936a) was the first to describe an interesting relation between the iron and P cycles in eutrophic lakes. In essence, ferrous iron (Fe21 ) and phosphate accumulate simultaneously in the anoxic hypolimnion during stagnation periods. However, during the next turnover (when O2 is introduced) a ‘‘ferric phosphate’’ is precipitated. According to Tessenow (1974) and Gunnars et al. (2002) this ‘‘ferric phosphate’’ is a Fe(III) oxyhydroxide that coprecipitates phosphate. The minimum molar stoichiometric Fe : P ratio of this solid is about 2. Thus, if in the anoxic hypolimnion the ratio Fe21/(HPO 1 H2PO ) $ 2, Fe as 22 2 4 4 well as phosphate precipitate nearly quantitatively after a sufficient amount of O2 is introduced to oxidize and precipitate all the iron. If the ratio Fe21/(HPO 1 H2PO ) , 2, 22 2 4 4 the excess phosphate remains in solution (Gunnars et al. 2002). In Mortimer’s (1941, 1971) classic laboratory experiments on Fe and P cycling in water–sediment systems, sedimentation of organic material was stopped and the sediment overlying water was maintained permanently oxic. Under these conditions, ferric (hydr)oxides formed in the uppermost sediment and trapped the P. If a eutrophic lake behaved similarly, similar processes should occur at its sediment– water interface. However, as sketched in Fig. 2, real lakes differ from such laboratory experiments because the oxic sediment surface at any specified time will eventually be buried by settling material. Thus, during further anoxic diagenesis, the precipitated ‘‘ferric phosphate’’ will be reduced in the buried sediment. Contrary to the situation at the anoxic sediment surface, the resulting Fe21 and HPO / 22 4 H2PO ions will not be diluted quickly into a vast hypolim4 nion. The newly formed concentration maximum in the pore-water profile causes P and Fe(II) to diffuse in the sediment upward, downward, or both according to the respective concentration gradients. In such a narrow production zone of Fe21 and HPO /H2PO , the solubility product of a 22 2 4 4 ferrous phosphate can be exceeded, leading to the precipi-