A Biodegradation Assessment Tool for Decision on Beach Response

Oil from the DWH blowout was deposited in the supratidal zone (i.e., landward of the high tide line) of beaches during major storms in the Gulf of Mexico. The oil got buried during subsequent events as a result of sand deposition on the top of it. Due to its large landward distance from the water line, the reworking by regular storm waves was not expected to physically remove this oil. At opposite ends of the options available for the Response by the Unified Incident Command (UIC) resided mechanical removal and natural recovery. The prior is capable of removing most of the oil but could be costly, and could have side effects such as habitat disturbance and beach erosion. Natural recovery would be adopted if the oil can biodegrade relatively fast (say within a few years). Therefore, the UIC needed to answer the following question: “If no action is taken, how long would it take for the oil left behind to be removed by biodegradation?” If the duration is too long (say 10 years and longer), then aggressive mechanical removal would be adopted, and if the duration is relatively short (say a few years), then the oil could be left in place to biodegrade naturally. To answer these questions, we conducted field measurements and numerical predictions.

Working closely with the UIC, we set up monitoring stations on three sand beaches (Fig. 1) to measure oil properties and the environmental conditions affecting the oil biodegradation (Fig. 1). We also used the fate and transport model BIOMARUN (described below) to model oil biodegradation. We modeled the biodegradation of the long chain alkanes (C30+) and the four-ring polycyclic aromatic hydrocarbons (PAHs) because the low molecular weight components of the oil were generally absent in the buried oil.

The beaches for the study were located at Bon Secour National Wildlife Refuge (Alabama), Fort Pickens National Monument (Florida), and Grand Isle State Park (Louisiana), as shown in Fig. 1. Due to space limitation, we will show only the field setup of Fort Pickens (FP). However, we will report the biodegradation results of the other beaches.

At FP, the oiled sediments were approximately 0.30 m deep (Fig. 2), and were above the water table, which was approximately 1.0 m deep (Fig. 3). The concentration of the long chain alkanes and PAH in the oiled sediments was measured using GC-MS FID (Battelle Labs), and was approximately 2.0 mg/kg of sediments and 1.5 mg/kg of sediments, respectively. These are relatively low concentrations of oil within the sediments, and they reflect the fact that the oil was “smeared” on the sediments and not entrapped within an “oil ball” or a “tar ball” (note Fig. 2). We also measured the water level with the beaches (i.e., water table), water temperature, salinity, nutrient concentration, dissolved oxygen, sulfate concentration, and microbial density (using ATP and MPN).

The water level was measured using pressure transducers (CTD, Schlumberger) placed in galvanized steel pipes (Fig. 4), and water samples were taken from multiple depths through multiport sampling wells (Fig. 4). The latter were used and discussed in Boufadel et al. (2010) and Li and Boufadel (2010), and will not be discussed herein.

Samples were taken using 60 ml syringes and placed in 125 ml propylene bottles. They were shipped frozen to Temple University in Philadelphia for further analysis.

The nutrient compounds were measured using AutoAnalyzer3 (Seal Analytical, Mequon, WI). The frozen samples will be defrosted and kept in the fridge (below 4 °C) in batches of 76 samples, at the time of analysis the samples were taken out of the fridge, hand shaken for 15 seconds and passed through 0.45 micron PTFE membrane filters (Puradisc^™, Whatman, Florham, NJ) into the AutoAnalyzer3 cups. The segmented flow method was used in Autoanalyzer3 and the concentrations were detected by colorimetric analysis. Ammonia was measured using the Berthelot reaction where a blue-green colored complex forms and gets measured at 660 nm wavelength. Nitrate in the solution was reduced to nitrite by a copper-cadmium reactor column (Grasshoff et al. 1999; Seal Analytical 2008).The nitrite was then reacted with sulfanilamide under acid condition to form a purple azo dye. The color was detected in 550 nm wavelength (Grasshoff et al. 1999; Seal Analytical 2008).

Phosphate was measured following the Murphy and Riley method until a blue color is formed by reaction of orthophosphate, molybdate ion and antimony ion followed by reduction with ascorbic acid at a pH<1. The blue complex is read at 880 nm wavelength (Grasshoff et al. 1999; Seal Analytical 2008). The soluble silicate is determined in this method based on reduction of siliconmolybdate in acidic solution to molybdenum blue by ascorbic acid. The complex was read at 820 nm wavelength (Grasshoff et al. 1999; Seal Analytical 2008).

The oxygen was measured using an optical DO probe (RDO, Thermo Scientific) connected to a handheld meter ORION4 Plus (Thermo Scientific).

The salinity of the same pore-water samples was measured using a digital refractometer (Salinity-300035, Sper Scientific, Scottsdale, AZ). The samples were filtered and about 1.5 mL of sample was poured into the measuring cup of the instrument and the salinity was determined based on the refraction index of the sample. The refractive index of the samples is affected by the density of each sample which would be different depending on the salinity.

Sediment samples were taken through a hollow stem auger with sleeves of diameter 1.0 inch. They were placed on ice and shipped to Philadelphia for microbial analysis. For ATP measurements, the sleeves were cut into small pieces and incubated under agitation for 5 min with a lysis buffer. The mixture was then filtered before being analyzed. Aliquots of 0.1 g of sediments were weighted and resuspended in 10 mL of sterile phosphate buffer saline (PBS) by vortexing. The suspension was then decanted for 5 min and an aliquot of supernatant was mixed the lysis buffer and incubated under agitation for 5 min. After incubation, samples were analyzed using BacTiter-Glo™ microbial viability assay (Promega). Bioluminescence was recorded on a Modulus Luminometer 9200 (Turner Biosystems).

Microbial counts were obtained using a most-probable-number (MPN) procedure for hydrocarbon degraders (Wrenn and Venosa 1996). The procedure is described in details there and is not repeated herein.

The BIOMARUN model is a fate and transport model that simulates the biodegradation of two organic compounds in the subsurface of beaches and aquifers. The model can simulate the biodegradation of oil located both below and above the water table, and as such is directly applicable to oil in tidally influenced beaches, in particular the Deepwater Horizon (DWH) oil buried in some of the beaches of the Gulf of Mexico.

The model BIOMARUN has three main modules: Water flow, solute transport, and biodegradation (Fig. 6). The first two modules are the components of MARUN (for marine unsaturated, (Boufadel et al. 1999b). The MARUN model accounts for the effects of water salinity on water density and viscosity, and subsequently on water motion (heavier water tends to sink when surrounded by lighter water). More documentation on the first two modules of BIOMARUN can be found in Boufadel et al., (1999b), Boufadel (2000), Li and Boufadel (2010), and Xia et al., (2010).

Where μ_1max and μ_2max are the maximum growth coefficient (e.g,. 1/day). The terms of the equations have the following designation: N, nutrient concentration (mg-N/L); O, oxygen concentration (mg/L); F₁, food (mg/L of pore water). The terms K_N, K_o, and K_F1 are constants. The coefficient I_θ accounts for moisture inhibition (discussed below). For the nutrients and hydrocarbons, the standard Monod is used, as supported by the literature (Essaid et al. 1995; Nicol et al. 1994). However, a modified Monod is used to account for oxygen effects, as shown in Eq. 2.

The fraction means that if there is no oxygen (O=0.0 mg/L) then no growth would occur, which is correct. As the oxygen concentration increases, the growth rate increases until reaching a plateau. This is reasonable, as other factors would become limiting for growth, and it is not realistic to increase the growth rate indefinitely by increasing the oxygen concentration. Oxygen limitation appears when the oxygen concentration drops below 2.0 mg/L, and it becomes obvious at 1.0 mg/L and lower (Boufadel et al. 2010). Therefore, we adopted the empirical model reported in Fig. 3, which is a modification of the Monod model. The constant K_o is taken as 2.0 (mg/L)⁴. The traditional Monod model cannot account for the sudden decrease in the aerobic rate when the oxygen concentration drops below 1.0 mg/L. Alternately, when the Monod rate is small at concentration less than 1.0 mg/L then it gives small rates at oxygen concentrations larger than 2.0 mg/L, which is not realistic.

Where θ is the water content, which varies between the residual water content θ_r and the porosity ϕ, and θ_m is the optimal value of the water content for bacterial growth, σ_θ is a parameter of the distribution, and a_θ is the maximum inhibition value. The following values for θ_m, σ_θ, and a_θ were used by El-Kadi [2001]: θ_m = 0.228, σ_θ =0.1, a_θ = 5.0 where the porosity was ϕ=0.4 while the lowest soil moisture was θ_r=0.1. We believe the transition needs to be more gradual, as a study by Fallgren et al. (2010) showed that the microbial activity at 30% saturation is even higher than that at 80%, and for this reason we adopt the value σ_θ= 0.3. The resulting function I_θ is plotted in Fig. 8. Note that we are assuming that the inhibition behavior is the same for both alkane and PAH degraders, X1 and X2, respectively, which is reasonable considering the uncertainty in the inhibition function.

The oxygen concentration was measured to be > 4.0 mg/L throughout the beach. Therefore, it is expected that the microbial activity in the vadose zone is aerobic. (If it is aerobic below the water table, it sure is aerobic above the water table). Fig. 9 reports the results of salinity, nitrate, phosphate, and sulfate. It shows that the average nutrient concentration in the upper ports was around 0.2 mg-N/L, and therefore this value could be viewed as limiting maximum biodegradation. Note that the optimal value ranges from 2.0 mg-N/L to 10.0 mg-N/L (Boufadel et al. 1999a; Du et al. 1999). The phosphate concentration was around 0.2 mg-P/L, a large value that implies that the biodegradation is not limited by the availability of phosphorus. A ratio of N/P=10/1 is considered appropriate if N is at the optimal value, i.e., 2.0 to 10.0 mg/L, which gives 0.2 to 1.0mg/L of P (Bragg et al., 1994; Sharifi et al., 2010). But considering that N is at 0.2 mg-N/L, it is clear that nitrogen is the limiting nutrient.

While the BIOMARUN model can simulate the movement of water and chemicals in the beaches, we found after conducting simulations that prediction of the environmental conditions for the next five years would require detailed forecast of rainfall, tide, and water table, which could not be done within the two months window that was provided to provide answers for the UIC. For this reason, we considered the present conditions to persist, and as such, there was no need to model water movement and chemical transport in the pore water. In other words, we used only the biodegradation module of the code BIOMARUN.

The parameter values for the biological module of BIOMARUN are reported in Table 1 along with their comparison with the literature. The concentrations used in the simulations are reported in Table 2, and we explain our selection of these parameters herein.

For both Bon Secour and Fort Pickins sites, we used the oxygen concentration measured in the water below the oil, which was around 8.2 mg/L, close to the solubility limit of oxygen in water in contact with air. This value was used in the simulation because the pore water in the vadose zone is in direct contact with air. For Bon Secour and Fort Pickins, we measured nutrient concentrations of 0.2 mg-N/L in the upper ports but the average concentration in the beaches were 1.2 mg-N/L and 0.8 mg-N/L, respectively. Unlike oxygen in the vadose zone that gets replenished continuously, nutrients are expected to get depleted during oil biodegradation and to be replenished sporadically. The replenishment of nitrogen would occur due to rain whose concentration is 0.8 mg-N/L or by wave splash whose concentration is 0.2 mg-N/L. For these reasons, when modeling the biodegradation of oil at Fort Pickens and Bon Secour, we used constant concentrations of 0.6 mg-N/L and 0.4 mg-N/L for Bon Secour and Fort Pickens, respectively.

The beach at Grand Isle had very low oxygen concentration (approximately 0.4 mg/L) suggesting that aerobic biodegradation of oil is essentially not occurring. The smell of hydrogen sulfide (rotten eggs) when excavating the sediments suggested the presence of a process known as sulfate reduction. This was confirmed with detailed environmental measurements. Therefore, the biodegradation of oil could be occurring during sulfate reduction (i.e., the sulfate plays a similar role to oxygen). However, the rate of oil biodegradation during sulfate reduction is usually small, less than 20% that of aerobic oil biodegradation. To reduce the rate to 20% that of the aerobic condition, we used an oxygen concentration of 1.0 mg/L, and by doing so the maximum rate was multiplied by 0.2. We used the measured nutrient concentration of 2.0 mg-N/L. We believe this approach provides an upper estimate of the oil biodegradation at Grand Isle.

On average, the concentrations of alkane and PAH degraders were 10-fold higher at oiled locations than at unoiled areas within each beach. Thus, to run the model, we used the microbial count at unoiled areas as initial condition for the simulation, assumed to occur in July 2010 when the oil stranded on the shorelines in AL and FL. The simulations at Bon Secour and Fort Pickens gave biomass values comparable to what was measured in December 2010 in the oiled sediments on the three beaches. This indicates that the model was able to capture the increase in hydrocarbon degraders within the six months duration.

Figs. 10, 11, and 12 report the normalized concentration of long chain alkanes and PAHs as function of time at Bon Secour, Fort Pickens, and Grand Isle, respectively. The concentrations were normalized by the values obtained in mid-December 2010. Time zero in these figures is July 2010. Fig. 10 (Bon Secour) shows that the concentration of alkanes reaches 13% of the initial value at 2.5 years, and that the degradation is very slow beyond that time. The PAH concentration decreased slower than that of the alkane reaching 15% of the initial value at 5 years. The model showed that not much degradation occurs after 5 years (not reported for brevity). Due to uncertainty in the estimated parameters, the modeled times provided are accurate within 50%. This means that the decrease of the alkane concentration to 13% should be read as occurring at 2.5 ± 1.25 years. Similarly the decrease of the PAHs to 15% should be viewed as occurring at 5.0 ± 2.5 years.

Fig. 11 (Fort Pickins) shows that the concentration of alkanes reaches 15% of the initial value at 2.5 years, and that the degradation is very slow beyond that time. The PAH concentration decreased slower than that of the alkanes, reaching 20% of the initial value at five years. The model showed that not much degradation occurs after 5 years (not reported for brevity). Similar to the discussion on uncertainty above (for Bon Secour), the times provided are accurate within 50%. This means that the decrease to 15% of the alkanes concentration should be read as occurring at 2.5 ± 1.25 years. Similarly the decrease of the PAHs to 20% should be viewed as occurring at 5.0 ± 2.5 years.

Fig. 12 (Grand Isle) shows that the concentration of alkanes reaches around 80% of the initial value at 5 years, and that the degradation continues at a slow rate causing the concentration to reach 50% of the initial value at 10 years. The concentration of PAHs at 5 years is about 95% of the initial value, and it decreases a very slow rate until reaching 87% at 10 years.

The comparable biodegradation of oil at Bon Secour and Fort Pickens reflected that the overall conditions at these sites are similar. The soil moisture at oiled zones was measured at 20% and 30% of the porosity at Bon Secour and Fort Pickens, respectively. Thus, all factors being equal, the biodegradation at Fort Pickens is expected to be faster. However, the nutrient concentration was 1.2 mg-N/L and 0.80 mg-N/L at oiled zones at Bon Secour and Fort Pickens, respectively. Thus, the higher nutrient concentration at Bon Secour results in higher biodegradation rates (all factors being equal). For this reason, the biodegradation of oil at Bon Secour and Fort Pickens were comparable.

The beach at Grand Isle had very low oxygen concentration (around 0.4 mg/L) suggesting that aerobic biodegradation of oil would be very slow. Therefore, the biodegradation of oil could be occurring during sulfate reduction (i.e., the sulfate plays a similar role to oxygen). However, the rate of oil biodegradation during sulfate reduction is usually less than 20% that of aerobic oil biodegradation, which explains the slow biodegradation of the alkanes and PAHs in Fig. 12. The high soil moisture at the site (90% of the porosity) and the high nutrient concentration (2.0 mg-N/L) could not compensate for the severe deficiency of oxygen.

Cleanup of sand beaches is generally thought to be relatively easy. However, when oil becomes deeply buried and beyond the reach of normal reworking by the beach cycle of erosion and deposition, then responders and resource agencies are faced with the environmental trade-offs between aggressive mechanical treatment methods and natural recovery. This issue was of particular concern on sand beach habitats within properties managed by federal agencies where the ecological services of the habitat were of concern. The key questions were: 1) How long will the oil persist? and 2) What are the ecological risks of the residual oil, compared to those from mechanical treatment? This study was designed to address the first question. The results provided the resource managers a science-based timeframe that they could use in making response decisions. The model results were available in less than 2 months after initiation of field data collection, indicating that it can be considered an operational tool in support of response decisions.

The views expressed herein are those of the authors and not their institutions. In particular, the views are not to be construed as official or reflecting the views of the Commandant or of the U.S. Coast Guard. This work was supported in part through a contract to Temple University from the Unified Command. We acknowledge contribution to this work made by Terry Walden (BP).