The Effect of Soil Heterogeneity on the Vadose Zone Transport of Bacteria for Bioaugmentation

Heterogeneity in hydraulic, physical and chemical properties of porous media can not only limit microbial dispersion but may also complicate the quantification of microbial transport processes and resultant microbial activities. The objectives of this research were to determine the potential for bacterial transport through an unsaturated soil block under transient flow conditions and to determine the influences of soil properties and phosphate additions. Despite the block consisting of >99% sand and appearing to be completely homogeneous, i.e., structure-less, flow was extremely heterogeneous as only 616% of the cross-sectional area exhibited flow with 88% of this flow occurring through just 4% of the area. The preferential flow paths exhibiting high and moderate flow rates were spatially consistent among rainfall events suggesting that soil properties caused the heterogeneity in flow rather than unstable wetting fronts. Transport of the GFP bacteria was extremely rapid with breakthrough occurring at the initiation of flow (0.1 h). Bimodal breakthroughs of GFP bacteria were observed for fast flow areas. The soil texture, rather than porosity, was the most significant property controlling the microbial transport as areas dominated by fine sand trapped the GFP bacteria. These findings demonstrate how apparent homogeneity in media properties does not equate with homogeneity in flow or transport of solutes and colloids. While bioremedial feasibility studies often center on soil chemical properties, this study indicates that consideration should be given to the physical and hydraulic properties of the soil as well. INTRODUCTION Most bioremedial practives involve the utilization of indigenous populations to degrade existing contaminants, however, recent studies have suggested that genetically engineered microorganisms (GEM’s) could increase the efficiency of bioremediation. Substrate concentrations and cell density are important factors that determine the kinetics of contaminant biodegradation (Alexander and Scow, 1989). To insure contaminant degradation, bacteria and bio-stimulators (i.e. substrates) are often required to be transported throughout the contaminated area. Sediment heterogeneity (i.e. physical, chemical, and hydraulic properties variation) plays a significant role in determining the fate of microbial populations during transport. Water movement through soils with preferential flow pathways or immobile-regions could significantly decrease bioremediation efficiency by preventing contaminant degrading bacteria from reaching the contaminants that are dispersed within the soil matrix. Objective. The objectives of this research were to determine the potential for bacterial transport through an unsaturated soil block under transient conditions and to determine the influences of soil properties and phosphate additions on bacterial transport and retention under unsaturated conditions. Site Description. An undisturbed block was obtained from the 5.9 m depth, i.e the top of the groundwater table, at a DOE research site approximately 4 miles east of the township of Cheriton in Northhampton County, Virginia on the Chesapeake Bay peninsula. The soil at the sampling location was the Molena series (Sandy, mixed, thermic Psammentic Hapludults) and consists of coarse texture sediments. This soil is a strong brown loamy sand (0.71 m thick), with a substratum of strong brown sand to a depth of 1.8 m or more (USDA, 1989). MATERIALS AND METHODS The lack of soil structure in the underlying sand allowed a 35 cm x 35 cm x 70 cm pedon to be easily shaped in the face of a borrow pit at the site. A stainless steel box (32 cm x 32 cm x 50 cm), with open ends, was constructed to encase the undisturbed block. The stainless steel box was hydraulically pressed into the exposed pedon with negligible soil disturbance (Kinsall et al., 1997). A stainless steel plate with a cutting edge was driven horizontally across the bottom to sever the bottom of the encased block from the soil pedestal. An acrylic grid of 64 flow collection chambers (3.75 cm x 3.75 cm) inside a 1 cm wide border flow collection area was fitted and sealed to the bottom of the stainless steel block by insetting it 1.5 cm into the soil. Each collection chamber had been partially filled with sterilized coarse sand to establish a hydrologic continuity between the soil and collection chambers. Two holes were drilled within the 1.0 cm annulus between the steel box and the 30 x 30 cm flow collection chambers to allow for collection of drainage from the border of the encased block. A rain simulator positioned above the soil block consisted of 64 application drippers spaced in a grid pattern 3.75 cm. Each dripper was controlled by a valve, which allowed for rainfall application rates to be easily adjusted as needed. Simulated Rain Events and Effluent Sampling. A 0.001 M CaCl2 solution, similar to the ionic strength of rainfall was applied to the surface of the blocks at a continuous rate of 5400 cm h for a duration of one hour, thereby simulating a 60 mm h rainfall. The soil block received four rainfall applications, at least 2 days apart, prior to the start of the experiment to wet the soil to field capacity. Collection grids, containing Nalgene polypropylene (autoclavable) bottles, were used to collect effluent from the soil into each respective collection chamber. When the first bottle was filled to capacity, the entire collection grid of 64 bottles was removed and replaced with another grid containing 64 empty bottles, and the time recorded. Effluent was continually collected over a period of two days after each rainfall application until free drainage ceased. Each bottle containing effluent was weighed to obtain the flow rate. A total of two rainfall applications were made on the block. During the first rainfall application, a solution containing a fluorescing bacteria, Pseudomonas putida (1 x 10 CFU L ), was applied to the surface of the block. This particular P. putida strain has been modified by the addition of the Green Fluorescent Protein (GFP) gene, which allows detection by fluorescent signal (Burlage et al., 1996). The bacterial solution was followed by two additional applications of 0.001 M CaCl2 , which was used to determine the flushing of phosphate and P. putida. Effluent samples from a fast, moderate, and a slow collection cell were analyzed with respect to time for microbial and chemical constituents Soil Sampling and Analysis. Following the completion of the rainfall applications, the soil block was dissected into 5 cm depth increments with 64 individual samples (3.75 cm x 3.75 cm cell area) per depth increment. Horizontal slits, spaced at 5 cm increments, were made in the front face of the steel box at construction. These slits allowed the encased block to be sectioned into 5 cm depth increments through the use of a 3.2 mm thick stainless-steel plate. After the sheet had been pressed through the entire width of the block, the encased 5 cm layer was carefully lifted out of the steel box. The layer was then sectioned into 64, 3.75 cm x 3.75 cm area increments in a grid pattern that was aligned with the grid of 64 effluent collection cells. Samples from the 3-8 cm, 23-28 cm, and 43-48 cm depths were utilized for complete analysis. P. putida analysis for effluent and soilextraction samples was conducted by the fluorescent signal given off by the GFP bacteria. A fluorescence spectrometer was used to determine the transmittance at a wavelength of 509 nm, with an excitable wavelength of 395 nm. The colony forming units (CFU) was determined by direct count of colonies growing on Luria-Bertani selective media which contained 50 mg tetracycline L. The correlation between microbial concentration, i.e., CFU cm of effluent, and the fluorescent signal was used to estimate the cells cm of each sample. The lower limit detection in this analysis was approximately 10 cells cm. After subsamples of wet soil had been removed for microbial analysis, all samples were analyzed for bulk density(ρb) and gravimetric (θg) water content. To determine (θg), a 10g subsample was removed, oven dried at 105 °C and reweighed. The remaining sample was weighed wet, air-dried for two days and then reweighed. The (θg) of the air dried sample was also determined from a 10 g subsample. Bulk densities, volumetric and gravimetric water contents were calculated by using the samples cube volume of 3.75cm x 3.75cm x 5cm, and the wet and oven-dried weights as described by Gardner (1986). Percent porosity was calculated in each sample by using an expanded version of Danielson and Sutherland’s method (1986), which involved the use bulk density and an estimated particle density of 2.65 g cm. This method can be better described in the equation of: % Porosity= (1ρb/ 2.65 g cm) x 100 (1) Particle size analysis of the sand block samples was conducted by using the Dry Sieving Method for fractionation of sand particles (Gee and Bauder, 1986), based on the USDA scheme for particle sizes. Specific Surface area’s were calculated by using the summation equation of: Specific surface (am) = (6/ps) ∑(ci/di)(2) (2) where ci is the mass fraction of particles of average diameter di, and ps is a particle density of approximately 2.65 g cm (Hillel, 1980). Soil chemistry was determined after soil-water extraction. Three grams of soil were combined with 27 ml of deionized water and placed on a shaker for 20 minutes. Soil solutions were thoroughly mixed and then filtered through a 0.45 μm Acrodisc filter. Effluent samples (10 ml) were analyzed for anions on the Ion Chromatograph (IC), and cations on the ICAP. A 1:9 ratio of soil/water was used for soil extractions to determine soil pH. Both soil and effluent pH was determined on an expandable ion analyzer, which is discussed by McLean (1982). Total phosphate concentrations were determined from soil extractions on the ICP and Ion Chromatograph (IC) respectively. Pearson correlation analysis was conducted using SAS on all variables for effluent according to rain event, and for all soil property variables according to layer. Cumulative flow and effluent bacteria concentrations were also analyzed for correlation, by layer and rain event, to all soil properties of their vertically aligned samples. RESULTS AND DISCUSSION Spatial Distribution of Flow. During the 4 rain events, only 6-16% of the 64 collection cells exhibited outflow. The numbers of collection cells where flow was observed included 3 fast flowing cells(>400 cm/rain event), 1 medium flowing cell (200400 cm/rain event) and 5-7 slow flowing cells (<200 cm/rain event). Spatial distributions of flow among the collection cells were consistent for all rain events, which indicates that the preferential flow paths were controlled by soil properties. The funnel flow concept, which is the occurrence of preferential flow paths through structure-less media, appears to be supported by these findings (Selker et al., 1992). Flow volumes during the first rainfall were highest of the four rainfall events (Figure 1). Approximately 98% of the applied solution was collected from the base of the block indicating that the block was at field capacity before the event. The second rainfall event, however, produced the lowest effluent volumes, with 88% of the applied solution collected from the base of the block. The two remaining events produced effluent volumes similar to the first event with Figure 1.Total Flow vs. Averaged Effluent Microbial 0.0E+00 5.0E+05 1.0E+06 1.5E+06 2.0E+06 2.5E+06 3.0E+06 3.5E+06 1 2 3 4 5 Rain Event P . p ut id a (c el ls c m -3 ) 4700 4800 4900 5000 5100 5200 5300 5400