Limnol. Oceanogr., 44(2), 1999, 425–430

‘‘Peepers’’ (dialysis samplers) rely on exchange across a dialysis membrane between fluid-filled cells and the surrounding sediment in order to sample the pore water at various depths. In the current study, we show that under some realistic circumstances, where there is a large density difference between the pore water and the cell fluid in permeable sediments, the resulting convective motion in the pore water distorts concentration distributions near the peeper. Peepers withdrawn prematurely in these conditions will contain seriously misrepresentative samples. We present numerical model and laboratory experiment results that show that the equilibration dynamics are complex and that the time required for equilibration may be far in excess of the duration predicted by a simple molecular diffusion model. A ‘‘peeper’’ (dialysis sampler) consists of a series of cells assembled in a vertical array that is covered with dialysis membrane (Hesslein, 1976) (Fig. 1). The peeper is embedded into aquatic sediment to sample the pore water in order to allow us to construct a concentration profile. Ions in the pore water diffuse across the membrane into the peeper cells, and the peeper is removed when the cell solution reaches equilibrium with the surrounding solution. A major consideration for peeper use is that deployment lasts long enough to ensure that the concentrations within peeper cells accurately reflect those in the undisturbed sediment. The principal factors influencing the time required to reach equilibrium are the permeation speed of the peeper membrane, the rate of mixing within the peeper cell, and the rate of transport from the surrounding sediment to the membrane surface. Experiments by Webster et al. (1998) demonstrated that a numerical model simulating diffusion through the surrounding sediment was well able to describe the equilibration dynamics of isolated vial peepers. Harper et al. (1997) also modeled peeper equilibration dynamics for three different cases: well buffered (desorption or dissolution from the solid phase to the pore water); diffusion (no resupply from the solid phase); and partial resupply from the solid phase. Using a numerical model to solve the two-dimensional diffusion equation, we simulated the equilibration behavior of a peeper embedded in sediment with a uniform pore-water concentration. This is equivalent to the second case described above (no resupply), which was considered by Harper et al. (1997). The model predictions were compared to measurements made in mesocosm sediments, from which peepers were withdrawn at daily intervals. The measured peeper concentrations increased at a rate that was several times faster than predicted by the model. Concentrations increased more rapidly in the lowest cells, but our diffusion model predicted faster equilibration in only the lowest cell. Analogous laboratory experiments showed the same enhanced equilibration. Clearly, the diffusion model is deficient under these circumstances. A possible explanation is that haline convection in the sediment pore water along the face of the peeper enhances solute transport through the sediment pores. As ions diffuse from the sediment into peeper chambers initially filled with distilled water, the pore water becomes depleted of ions and therefore reduces in density. The lighter pore water floats upward along the face of the peeper, toward the sediment surface, and draws in denser water from further afield. For uniform pore-water concentrations, this shortens the time to equilibrium and brings the lowest cells into equilibrium first. Adler (1977) also observed faster equilibration in the lowest peeper cells, and he too suggested pore-water convection as a possible mechanism. He concluded that peeper water and pore waters should be the same density in order to prevent convection from distorting the results. As yet, this is not normal practice (e.g., Adams 1994). The suggestion has been made that peepers should be filled with a sodium chloride (NaCl) solution that is half the concentration of the NaCl expected at the sampling location, with the intention that a dilute salt solution would reduce osmotic loss from a peeper deployed in a saline environment while still providing a favorable concentration gradient for transporting dissolved species into the peeper (Simon et al. 1985). Clearly this practice would be inappropriate if the sediment is permeable enough to allow pore-water convection. We conducted laboratory experiments using peepers containing 36 dual-windowed cells, each 10-mm deep, 6-mm high, and 60-mm wide, with a center-to-center separation of 10 mm (Fig. 1). The cells were covered with a 0.45-mm polysulfone membrane. This design is similar to that developed by Hesslein (1976), a design that has been used in geochemical and ecotoxicological studies (Gaillard et al. 1986; Tessier et al. 1989; Williamson and Parnell, 1994; Hare et al. 1994). Six peepers were filled with deionized water and placed in a bin of fine sand containing well-mixed artificial seawater made from NaCl, magnesium chloride, calcium chloride, potassium chloride, sodium bicarbonate, and strontium chloride. The peepers were inserted into the sediment and were removed one at a time over the next 10 d, and the cell contents were analyzed using inductively coupled plasma mass spectroscopy (ICP-MS). We will discuss only the measurements of the magnesium concentration, although our findings are applicable to all species. Magnesium comprised 4% of the original dry mass of species dissolved and is treated as a tracer in our experiments. The porosity and permeability of the sediment were measured to be 0.42 and 3.0 3 10211 m2, respectively.