Phytoremediation of a Petroleum-Hydrocarbon Contaminated Shallow Aquifer in Elizabeth City, North Carolina, USA



A former bulk fuel terminal in North Carolina is a groundwater phytoremediation demonstration site where 3,250 hybrid poplars, willows, and pine trees were planted from 2006 to 2008 over approximately 579,000 L of residual gasoline, diesel, and jet fuel. Since 2011, the groundwater altitude is lower in the area with trees than outside the planted area. Soil-gas analyses showed a 95 percent mass loss for total petroleum hydrocarbons (TPH) and a 99 percent mass loss for benzene, toluene, ethylbenzene, and xylenes (BTEX). BTEX and methyl tert-butyl ether concentrations have decreased in groundwater. Interpolations of free-phase, fuel product gauging data show reduced thicknesses across the site and pooling of fuel product where poplar biomass is greatest. Isolated clusters of tree mortalities have persisted in areas with high TPH and BTEX mass. Toxicity assays showed impaired water use for willows and poplars exposed to the site's fuel product, but Populus survival was higher than the willows or pines on-site, even in a noncontaminated control area. All four Populus clones survived well at the site. © 2014 Wiley Periodicals, Inc.*


Peer-reviewed studies have reported the successful use of phytoremediation tree systems to hydraulically contain and mitigate petroleum hydrocarbon contamination in the subsurface (Arnold et al., 2007; Barac et al., 2009; El-Gendy et al., 2009; Ferro et al., 2001, 2013; Hong et al., 2001; Landmeyer, 2001). Groundwater phytoremediation systems best decrease contaminant mass for more soluble analytes of petroleum hydrocarbons such as benzene, ethylbenzene, toluene, and xylenes (BTEX; Barac et al., 2009; El-Gendy et al., 2009; Ferro et al., 2001, 2013; Landmeyer, 2001) and fuel oxygenates such as methyl-tert-butyl ether (MTBE; Arnold et al., 2007; Hong et al., 2001). Most of the aforementioned studies have been implemented at sites with shallow, contaminated groundwater where depth to groundwater is less than 3 m. A few studies have used trees to reach greater groundwater depths at 4 m to 5 m below ground surface (Barac et al., 2009; Ferro et al., 2013) for hydraulic-containment and mass reduction of mixed gasoline and diesel fuels.

Most groundwater phytoremediation systems utilize hybrid poplar trees (Populus) or willow trees (Salix) because they grow quickly, have deep-rooting characteristics, and can use water directly from the saturated zone (Ferro et al., 2013; Landmeyer, 2001). Populus and Salix are deciduous trees from the same family, Salicaceae, that transpire large quantities of water. Groundwater extraction by these trees will depend on the water balance of the site, depth to groundwater, tree age, tree density, and potential phytotoxicity of the contaminated plume. The cumulative transpiration of a stand of trees can alter both horizontal and vertical flow of groundwater within a tree stand (Landmeyer, 2001; Steward & Ahring, 2009). Estimated water use (per tree) for hybrid poplars range from 0.31 L/day to 6 L/day (one-year-old trees) to 1.9 L/day to 42 L/day (two years of age) to 23 L/day to 59 L/day for four-year-old trees (Ferro et al., 2001, 2013; Landmeyer, 2001). Ferro et al. (2013) recently reported that a mixed stand of four-year-old willow and poplar trees extracted 493 L/day of groundwater which was almost sufficient for complete hydraulic control (660 L/day) at the site according to groundwater flow model (MODFLOW) calculations. At this site, trees were planted as 2.4-m-long poles in 7.5-m-deep boreholes designed to develop deep root growth down to the groundwater at 5 m to 6 m below ground surface.

Using trees to remove pore soil water and extract groundwater can supply oxygen to the vadose zone. Both field and laboratory studies have shown increased concentrations of oxygen at deeper soil depths due to groundwater extraction by trees and the release of oxygen from tree roots (El-Gendy et al., 2010; Weishaar et al., 2009). Greater oxygen input to soils resulted in increased abundance of BTEX degrading microorganisms (Weishaar et al., 2009). Barac et al. (2009) observed increased abundance of toluene-degrading bacteria in rhizosphere soils and endophytic bacteria once poplar roots reached groundwater contaminated with BTEX; the abundance of these bacteria declined as BTEX concentrations declined.

Gasoline range organics (GRO) removal by groundwater phytoremediation is greater than diesel range organics (DRO) removal due to the more hydrophobic, higher molecular weight range of DROs (El-Gendy et al., 2009). GROs, such as BTEX and two- to three-ringed polycyclic aromatic hydrocarbons (PAHs), are dissipated primarily by biodegradation in the root zone of trees (Ferro et al., 2013). However, their lower log Kow ranges (1.5–2.9) can facilitate their uptake by trees. GRO can be taken up by roots, metabolized in tree tissues, and transpired to the atmosphere (Burken & Schnoor, 1999), although these latter pathways are not primary routes of removal. At steady state conditions, GROs are found primarily in woody stem tissue, particularly lower stem tissues (Burken & Schnoor, 1999). More field quantification of DRO removal in groundwater phytoremediation systems is needed because studies have focused primarily on grasses for removal of heavier aromatic hydrocarbons with log Kow ranges greater than 2.9 (Cook & Hesterberg, 2013).

Tree roots will interact with contamination as roots penetrate smear zones or if precipitation events elevate groundwater levels and displace light nonaqueous phase liquid (LNAPL) fuel product upward through the vadose zone to the tree roots. Trees may respond by closing stomata and reducing sap flow in order to avoid the uptake of toxicants (Ferro et al., 2013). Trapp et al. (2001) evaluated the phytotoxicity of weathered fuel product using transpiration as an endpoint. Soils containing either weathered fuel product or freshly added gasoline and diesel fuel were added to hydroponic solutions containing rooted willows or poplar trees. Strong correlations between DRO concentration and impaired transpiration were observed with an effective concentration of 3,910 mg/kg for 50 percent transpiration reduction. Willows were less sensitive than poplars to fresh diesel-polluted soils at 1,000 mg/kg. Soil amended with fresh DRO (10,000 mg/kg) were acutely toxic to willows and poplars. Soils spiked with fresh gasoline (1,000 mg/kg) were lethal to all tree species and were more toxic than weathered gasoline or diesel fuel. Ferro et al. (1999) did not observe any toxic response of poplar trees irrigated with water containing 169 mg/L BTEX and other hydrocarbon volatile organic compounds (VOCs). Likewise, poplars showed no phytotoxicity to groundwater with BTEX concentrations at 1 mg/L (Barac et al., 2009). Published data on field-site toxicity of petroleum hydrocarbons to trees is limited for groundwater phytoremediation systems.

For fuel oxygenates like MTBE, trees primarily translocate and transpire MTBE as the main pathway of removal with some storage in old growth stems (Hong et al., 2001; Ma et al., 2004). Both laboratory and field studies observed tree uptake and volatilization of MTBE from groundwater with MTBE concentrations ranging from 40 mg/L to 230 mg/L (Arnold et al., 2007; Hong et al., 2001). Impairment of transpiration, as measured by normalized relative transpiration (NRT), occurred for willows exposed to 400 mg/L MTBE in hydroponic solutions (Yu & Gu, 2006). Ma et al. (2004) observed reduced transpiration rates for poplars exposed to 593 mg/L and 1,186 mg/L MTBE. In both studies, trees showed no visible signs of toxicity such as leaf discoloration or wilting.

The authors report herein the remedial performance and tree survival of a phytoremediation system planted from 2006 to 2008 at a former bulk fuel terminal at the United States Coast Guard Base Support Unit, Elizabeth City, North Carolina. Information on site establishment costs and methods have already been reported (Cook et al., 2010). This article summarizes quarterly groundwater monitoring data, LNAPL thickness gauging, and semi-annual soil-gas monitoring to demonstrate the system's effectiveness in decreasing petroleum hydrocarbon mass in the subsurface. Biennial inventories of tree survival across the site and phytotoxicity studies using LNAPL product from the site provided field-relevant responses of willow and poplar trees to mixed GRO and DRO fuels.


Site Description

The study area is 2 hectares and approximately 170 m from the Pasquotank River. The depth to groundwater fluctuates from 1.2 m to 2.1 m below surface with an estimated groundwater flow velocity of 0.16 m/day (ARCADIS, 2009). Aircraft fuels were stored at the site from 1942 to 1991 in aboveground and underground storage tanks. Seven leaking steel-underground storage tanks and two aboveground storage tanks were removed from the site between 1991 and 1993 (ARCADIS, 2009). The total petroleum product released to the subsurface is unknown, but an estimated 352,000 L of LNAPL were recovered between 1991 and 1993. An LNAPL recovery system, installed in 1994, removed an additional 109,561 L of free-phase petroleum product from June 1994 to March 2006 (ARCADIS, 2009). The recovery system was decommissioned in March 2006. Although recovery wells and substructure remain in place on the site, the recovery system is no longer functional due to damage from the tree installation.

Tree Plantation Characteristics

A groundwater phytoremediation system was installed from 2006 to 2008 (Cook et al., 2010) using willow (Salix), loblolly pine (Pinus taeda), and four hybrid poplar clones (Populus). Hybrid poplar clones OP-367 and DN-34 are natural crosses between Populus nigra (European black poplar) and Populus deltoides (eastern cottonwood USA), and clones 15 to 29 and 49 to 177 are natural crosses between Populus trichocarpa (western cottonwood USA) and P. deltoides. In 2006 and 2007, trees were planted on 3 m × 3 m centers while remaining trees were planted on 2 m × 2 m centers in 2007 and 2008. At the time of planting, an estimated 579,000 L of LNAPL product remained on site. Details for site location, estimated distribution of the total petroleum hydrocarbon (TPH) plume in 2007, soil quality, tree selection, tree-planting approaches, site maintenance, and tree mortality are provided in Cook et al. (2010). The site is not irrigated nor has it received fertilization treatments.

In April 2010 and 2012, all tree-planting locations were inventoried for tree survival, tree height, and tree diameter at breast height (DBH). Tree height and DBH were determined for live trees using tree height poles, Suunto clinometers, logger tapes, and Lufkin Executive DBH measuring tapes. Tree metric data were recorded on-site using Trimble GeoExplorer 2008 Series GPS rovers and mapped using ArcGIS Geostatistical Analyst (ESRI, Redlands, CA). Aboveground biomass was calculated from tree height and DBH as wood density times volume. Tree densities in kilograms per cubic meter (kg/m3) were determined for each species using a wood density database (Carsan et al., 2012) where medium density or the average of high and low density times 1.7 was used to determine density at 50 percent moisture content. Tree volume was determined using the equation:

display math

Contaminant Characterization

Annual groundwater monitoring activities for the dissolved-phase petroleum hydrocarbons began in June 1993 and continues to date (ARCADIS, 2012; Exhibit 1). During the installation of the phytoremediation design in 2007, 10 additional groundwater monitoring wells were installed by the U.S. Geological Survey (USGS). Since 2002, annual groundwater sampling for BTEX and MTBE has occurred at 10 monitoring wells. Collection of groundwater samples for BTEX and MTBE analysis followed standard field measurement procedures for collection of VOCs (USEPA, 2011b) and were analyzed using USEPA Method 8260B by a certified environmental laboratory (certification # 269). Groundwater collection used dedicated tubing while all nondedicated sampling equipment was decontaminated before field work as well as between wells to avoid cross-contamination.

Exhibit 1.

Locations of groundwater wellsa (A) and soil-gas wells (B)

Note: aWells shown have been routinely monitored for BTEX and MTBE since 2002. Other well locations contained either petroleum free product and could not be sampled or had hydrocarbon and MTBE concentrations below detection limits.

Water and free phase petroleum product levels (free-product thickness) were gauged quarterly by measuring depth-to-water and depth-to-fuel product levels from the top of well casings in 28 to 30 wells and six recovery wells (Exhibit 1) from 2005 to 2012. In brief, the well cap was opened, and the static water level was allowed to equilibrate with atmospheric pressure after which an electronic interface probe was lowered into the well to determine free-phase fuel product and groundwater levels. The difference in altitude of groundwater and the LNAPL determined the fuel product thickness (m).

In 2006 and 2007, 87 soil-gas monitoring points were installed to evaluate the extent of petroleum hydrocarbon contamination across the site and to monitor changes in soil-gas contaminant mass in winter when trees are dormant and in summer when trees are actively transpiring (Exhibit 1). Soil-gas installation and analyses were performed by Amplified Geochemical Imaging, LLC (AGI™; formerly W.L. Gore and Associates, Inc., Newark, Delaware). Soil-gas wells were installed every 30.5 m to a depth of 0.4 m. The AGI™ soil-gas samplers were deployed for 10 to 14 days then retrieved and analyzed by thermal desorption gas chromatography/mass spectrometry using modified USEPA Method 8260 (SPG-WI-0292); soil-gas samplers found submerged in water or with evidence of LNAPL were not analyzed. In February 2007, 68 soil-gas monitoring locations were sampled after which 29 to 41 locations were sampled on a semi-annual basis. Exhibit 1 shows the 29 soil-gas locations that were consistently sampled from February 2007 to December 2012 for all sampling periods.

Weathered Fuel Product Toxicity to Trees

During precipitation events, groundwater levels may rise to the ground surface and provide opportunities for LNAPL to interact with tree roots, thus, hindering early tree development and survival. We evaluated the toxicity of LNAPL from the site to trees by measuring changes in water use using a hydroponic assay system (Trapp et al., 2000). Tree cuttings (∼20 cm in length, 10 mm in diameter) were obtained from a healthy, six-year-old hybrid poplar tree (P. deltoides × P. nigra, OP367) at the site and LNAPL was collected from monitoring well 19 (MW 19; Exhibit 1). Cuttings were placed in tap water after removal from the tree for transport, then stored in tap water with rooting hormone under lights for 14 hr until substantial roots and leaves developed (about six weeks). Once successful rooting began in the first two weeks, cuttings were pulled through a rubber stopper, and the orifice was sealed with silicone. Each stoppered cutting was placed in a 250-m amber glass jar containing Ionic® Grow (100 percent of recommended mixing by manufacturer) and tap water for another four to five weeks to allow sufficient leaf and shoot growth prior to phytotoxicity testing.

Prior to phytotoxicity testing, trees were allowed to acclimate to a 25 percent Ionic® Grow solution at a 24 hr photoperiod for 72 hr. Baseline water use was then determined for all trees by determining the mass of each apparatus (glass jar, tree, and hydroponic solution) with an analytical balance at 24 hr and 48 hr. Reductions in mass were attributed to water use by trees. The same apparatus with a dead control tree had no mass loss over the same period. Trees were randomized and placed into groups (n = 4) for toxicity testing with LNAPL collected from MW 19 in 2011. Fuel product treatments were prepared by adding either 50 µL, 100 µL, 200 µL, 500 µL, and 1000 µL of LNAPL to 225 mL of 25 percent Ionic® Grow solutions. A control group, consisting of four trees with no LNAPL added, were also evaluated for water use. The mass of each apparatus (tree, stopper, jar, and solution) was determined after adding the fuel product. The apparatus mass was determined every 24 hr for 72 hr during the toxicity study to evaluate water use over time.

Data Analysis

Fuel product thickness measurements were interpolated by inverse-distance weighting using ArcGIS. For each soil-gas sampling period, soil-gas mass (µg) was interpolated for BTEX and TPH (C4–C20 aliphatic hydrocarbons) using a minimum curvature algorithm that was performed in log space using grid cells that were set to 3.05 m. For comparison purposes, maximum color contours were set to initial levels in 2007 of which the lower boundaries were determined by the higher value of either the method detection limits or method blank levels at that time. Like inverse distance weighting (IDW), the spline method for surface interpolation is considered a deterministic approach because it assigns values based on surrounding measured values. The spline method can create a smoother surface than IDW, and was used for soil-gas contours and surface elevation maps (Exhibit 2). Tension was set to 100 and values were corrected with a map calculator that fell into a negative range to zero.

Exhibit 2.

Locations of tree mortalities relative to 2007 soil-gas contours (A) and surface elevation (left) for 2010 and 2012 (B)

For phytotoxicity assays, absolute water-use data were transformed to normalized relative transpiration because water use can vary between plants of different mass; however, absolute transpiration, which quantifies the rate of water usage per tree in grams per hour, was used for statistical analyses. NRT normalizes treatments to the initial transpiration and transpiration of nonexposed control trees and eliminates water use variability due to the initial size and growth of trees (Trapp et al., 2000). NRT was calculated using the formula below:

display math

where C is concentration (mg/L), t is time period (hr, 0–72), T is absolute transpiration (g/hr), i is replicate 1, 2, …, n, and j is control 1, 2, …, m. Statistical significances for water use and biomass change between fuel product treatment and control trees were determined using a paired Student's t-test at the 95 percent significance level.

Quality Assurance and Quality Control

To evaluate field data precision for tree mortality, 10 percent (248) of all live trees (2,488) were measured twice as field duplicates. The relative percent difference (RPD) for mortality was less than 10 percent. Samples collected for volatile organic analysis (VOA) in groundwater included one to two trip blanks, a field duplicate, a matrix spike, and a matrix spike duplicate following standard field QC procedures (USEPA, 2011c). For groundwater VOC analyses, instrument calibrations were within 25 percent relative standard deviations (RSDs). Relative percent deviations were less than 3 percent for VOC field duplicates. Analytical percent recoveries for three known reference standards were between 90 percent and 110 percent and were well within acceptable ranges between 70 percent and 130 percent while percent RSDs were less than 5 percent. Each soil-gas collection trip included three to four trip blanks. For soil-gas analyses, reference standard recoveries were between 93 percent and 111 percent with percent RSDs less than 6 percent.


Observations of Tree Mortality and Relevance to Site Characterization

Reports of initial tree mortality during stand establishment in 2006 and 2007 were provided by Cook et al. (2010). The tree mortality observed in 2006 occurred in areas not yet characterized for petroleum hydrocarbon contamination at that time. An expanded comprehensive assessment of the site was completed in February 2007 using AGI™ soil-gas samplers (Exhibits 1B and 2A) prior to planting trees across the majority of the site in March 2007. Areas of clustered tree mortality observed at the end of the growing season in 2008 have persisted (Exhibits 2A and 2B) even after replacing dead trees with new tree cuttings in 2009 and 2013. Tree mortality clusters are in areas where TPH mass was the greatest (Exhibit 2A) which is related to the original location of fuel bunkers and plume migration (area of highest elevation, Exhibit 2B) and a former stream bed area in the eastern portion of the site where elevation is lowest (Exhibit 2B). The incidence of tree mortality in 2012 (white dots) most likely represents tree loss due to Hurricane Irene in 2011. Current studies are evaluating use of grasses in mortality-dense areas and are not yet complete.

Hybrid poplar clones were planted in large blocks across the site while willow and loblolly pine trees were more randomly distributed. Hybrid poplar clones survived better than willow or loblolly pine trees (Exhibit 3) in the phytoremediation area. The same trend was observed in a control area where no significant contamination existed and where all tree species and clones were planted randomly. In the control area, pine and willow tree survival improved to 57 percent and 78 percent, respectively, for 2012. The survival of hybrid poplar clones in the control area also improved relative to survival in the phytoremediation area. Survival was 100 percent (OP-367), 93 percent (15–29), 100 percent (DN-34), and 93 percent (49–177) in the control area which suggests that all clones could survive in noncontaminated soils at this site. The lower survival for clone OP-367 (66 percent) in the phytoremediation area reflects the fact that the block area for this clone was located where TPH mass continues to be most prevalent (Exhibit 2).

Exhibit 3.

Percent survival of Populus clones, loblolly pine (P. taeda), and willow trees (Salix) on phytoremediation site

Weathered Fuel Product Toxicity to Hybrid Poplar and Willow Trees

The toxicity of free-phase, fuel product, collected from MW19 in 2011, is shown in Exhibit 4. The response of trees is presented as NRT wherein control trees are always 100 percent NRT. Treatments under 100 percent represent inhibition of water use for exposed trees while an NRT response greater than 100 percent indicates more water use by exposed trees relative to control trees. NRT is not amenable to statistical comparisons but visually demonstrates water use gains and losses relative to control trees.

Exhibit 4.

Toxicity of fuel product from well 19 (2011) to water use by Populus and willows in hydroponic systems represented as percent normalized relative transpiration (NRT). Asterisks (*) indicate significance differences (p < 0.05, Student's t-test, n = 4) between fuel product treatment versus control trees for each species

All fuel product treatments resulted in lower percent NRT values for hybrid poplars but only treatments of 500 µL and 1,000 µL to 225 mL of solution were toxic to hybrid poplars based on statistical analysis of absolute water use (paired Student's t-test, p < 0.05; Exhibit 4). These two exposure concentrations resulted in 100 percent mortality in hybrid poplars after the 72 hr assay period. The same two treatments were also significantly toxic to willow trees, but willow trees did not die after 72 hr. Interestingly, lower fuel product doses in willow trees caused greater percent NRT relative to control willow trees (Exhibit 4). This response may be hormesis which is observed when low doses of toxicants cause excessive repair responses in exposed organisms (Calabrese & Baldwin, 2003).

Changes in TPH and BTEX Contaminant Mass for Soil-Gas Measurements

The mass of volatile analytes of petroleum hydrocarbons in soils was determined across the site from 2007 to present using AGI passive soil-gas samplers. Twenty-nine soil-gas wells were consistently analyzed in winter and summer, and their cumulative sums of detected TPH and BTEX masses are provided in Exhibit 5. A consistent loss of TPH and BTEX mass is evident after trees were initially planted in 2007 with a final mass loss of 99 percent for BTEX and 95 percent for TPH as of July 2013. Exhibit 6 provides interpolated maps for TPH and BTEX mass distributions across the site which shows the slower removal of TPH and BTEX from areas with tree mortality clusters (Exhibit 2), principally the former fuel bunker and storage tank areas to the northwest and the low elevation area to the southeast.

Exhibit 5.

Total soil-gas mass (µg) of total petroleum hydrocarbons (TPH) and benzene, toluene, ethylbenzenes, and xylenes (BTEX) for 29 soil-gas wells sampled each summer and wintera

Note: aJanuary 2008 and July 2012 samples were not collected.

bSoil-gas wells were reduced to 18 wells; nonsampled wells had no mass detected for three prior collections.

Exhibit 6.

Interpolated surfaces of soil-gas mass (µg) for TPH and BTEX in winter and summer

Changes in Free-Phase Fuel Product Thickness

Interpolations of free-phase fuel product thickness are provided in Exhibit 7 for 13 wells monitored quarterly from 2008 to 2012. We have provided interpolations for only winter and summer periods for 2008, 2010, and 2012. Over time, fuel-product has concentrated to its greatest thickness (1.6 m) at downgradient wells, primarily well 36 (Exhibit 7) which is close to MW 34 that does not contain fuel product. Well MW 36 is upgradient and adjacent to the largest hybrid poplar trees on site that were planted in 2006 (Exhibit 7, biomass inset). These results suggest that fuel product thickness has declined in most of the other 12 wells (Exhibit 1) across the phytoremediation site since the installation of trees.

Exhibit 7.

Inverse distance weighting of free-phase fuel product thickness (meters) at 13 groundwater wells on site and tree biomass

Changes in Benzene and MTBE Concentrations in Groundwater Monitoring Wells

Concentrations of benzene and MTBE in groundwater are shown in Exhibit 8 before and after installation of the phytoremediation system. The wells are listed from the most distant downgradient well (MW 16) near the Pasquotank River to the most upgradient well on the phytoremediation site (MW 18) as shown in Exhibit 1. Funding and contract issues altered annual sampling times for 2012 and 2013, but overall, concentrations for benzene have declined as of December 2012. Historical data from 2003 to 2005 would suggest that most of the MTBE had migrated beyond the footprint of the site toward the Pasquotank River by the time of tree plantings in 2006. However, well MW 20 had increased MTBE concentrations in 2012, which may reflect latent releases from historical contamination or more recent releases from a fuel farm terminal upgradient of the site near well MW 7 (Exhibit 1A).

Exhibit 8. Concentrations (µg/L) of benzene and MTBE in groundwater for each year.a Boldface numbers are concentrations in groundwater above the NCAC 2Lb standards of 1 µg/L benzene and 20 µg/L MTBE

 June 2003Dec 2003May 2004May 2005June 2006May 2007June 2008July 2009July 2010July 2011Jan 2012Dec 2012
  1. a

    NS: not sampled.

  2. b

    Title 15A North Carolina Administrative Code Subchapter 2L (NCAC 2L) Groundwater Standards.

Benzene    Trees planted
MW 16<1.0<1.0<1.0<<1.0<1.0<1.0<1.0NS<1.0
MW 10<1.0<1.0<1.0<1.0<1.3<1.0<1.0<1.0<1.0<1.0NS<1.0
MW 14160190400210<1.33.9<1.0385203.8<1.0<1.0
MW 20NSNSNSNS1000280075026003000260096260
MW 12.4<1.0<1.0<1.0<1.34.2<1.0<1.0<1.0<1.0NS<1.0
MW 34NSNSNSNSNSNSNS<1.0<1.03.1<1.0<1.0
MW 18<1.0<1.0<1.0<1.0<1.01.8<1.0<1.0<1.0<1.0NS<1.0
MW 1645496111046041033021016025NS27
MW 10583038366637360.50113.9NS0.77
MW 14360340130130170160510.5040132277
MW 20NSNSNSNS250200630.500.501258170
MW 12617852501901805.
MW 180.500.502.41.414235.05.0164.5NS0.74

Evidence of Hydraulic Containment for Groundwater in Phytoremediation Area

Exhibit 9 shows the groundwater altitude (mean sea level) for wells shown in Exhibit 1A. Dashed lines indicate wells outside of the phytoremediation area while solid lines are wells within the site. Since 2011, wells within the phytoremediation area (wells 5, 3, 17, 37, and 39) have had lower groundwater altitudes than wells at the downgradient border of the site such as wells 1 and 20. These lower groundwater altitudes have occurred during the last two years even in the presence of normal to above-normal rainfall (Exhibit 9).

Exhibit 9.

Groundwater elevation levels for wells outside of the phytoremediation area (dashed lines) and wells within the phytoremediation area (solid lines)


Establishing a successful tree stand and maintaining tree health over time is important to remedial performance and does require expertise and routine maintenance to manage weed control, irrigation (if provided), fertilization (if desired), tree disease, tree pests, and tree replacement if mortality occurs. One advantage of working with hybrid poplars or willows is that dead trees can be replaced with cuttings from other trees on site; thus, the site is biologically self-sustaining. All of the four hybrid poplar clones survived well, but the willows and pine trees did not due to either water competition with the hybrid poplars or other site condition reasons. There are a variety of hybrid poplar clones available for purchase; careful selection of clones to site conditions and climate is important as well as site preparation and actual planting as noted in Cook et al. (2010).

Few published studies have used passive soil-gas sampling to monitor the effectiveness of a groundwater phytoremediation system for petroleum hydrocarbon contamination even though soil-gas mass data correlate well with groundwater concentrations (ITRC, 2006). These data are particularly compelling for this demonstration site and better reflect the impact of the trees on contaminant removal than annual groundwater monitoring data. Not only do these data help evaluate system performance, but soil-gas data were crucial to understanding the biological response observed at the site with regards to tree mortality. One could argue that the distribution of observed tree mortality is as accurate as the soil-gas maps for delineating subsurface contamination. However, tree mortality may result from various environmental factors such as planting methods, excess or insufficient rainfall, disease/pests, weed competition, and contamination. Combining tree survival inventories with soil-gas data provided a much more accurate understanding of contaminant distribution and removal at the site. In all likelihood, increased groundwater monitoring frequency would still not have provided the comprehensive assessment that soil-gas monitoring and tree survival inventories provided.

Measurements of LNAPL phytotoxicity, collected from the site, to willows and poplar trees show that both tree species were susceptible to impaired transpiration at similar concentrations. Species response to lower exposure concentrations did differ and would suggest poplar trees are less tolerant to lower fuel product concentrations which has been observed elsewhere (Trapp et al., 2001). If systems are designed to hydraulically impact contaminated groundwater, understanding concentrations that inhibit transpiration are important to system effectiveness and performance. For MTBE, groundwater concentrations toxic to trees are available in the published literature, but more information is needed for petroleum hydrocarbons.

Groundwater phytoremediation systems generally utilize water budgets as a metric to evaluate site performance. Since 2011, groundwater elevations have declined in the planted area which suggests groundwater extraction by trees at four and five years of growth even during recent years of above normal rainfall (2012 and 2013). Sap flow measurements combined with biomass data for each tree on site will enable detailed water budget estimations. Those efforts are ongoing for later publication.

Recently, Compernolle et al. (2012) integrated an economic decision model with remedial cost and effectiveness to compare groundwater phytoremediation of petroleum hydrocarbon contamination to other conventional technologies at a known site. The costs of conventional technologies were based on a deterministic groundwater flow model which did not address uncertainties of subsurface heterogeneity and transport dynamics; Compernolle et al. (2012) noted the need to incorporate the value of groundwater modeling (deterministic or stochastic) as part of design costs in their analyses. The conventional technologies included pump and treat (PT), vertical engineered barrier (VEB), permeable reactive barrier (PRB), and noncontainment (NC).

If technologies were compared on average cost effectiveness per year of operation time (regardless of total remediation time needed), then the more cost-effective technologies were first NC then phytoremediation then VEB, PT, and PRB. If technologies were compared on average cost for mass removed during operation time, then the more cost effective technologies were NC–Phytoremediation–PT–PRB–VEB. For incremental cost effectiveness (faster cleanup or more mass removed at extra cost), costs were compared for remediation time or mass removed/contained during remediation time. For the latter, VEB, PT, and PRB were more effective but more expensive. If cost was measured in remediation time, phytoremediation dominated VEB, PRB, and PT. At this point, the value or willingness to pay for incremental change (mass removed or remediation time) has to be determined, presumably by the stakeholder of interest. For their study site, the authors found that PRB and NC were preferred for incremental mass removal over phytoremediation regardless of the willingness to pay value.

When remediation time is monetized, phytoremediation was preferred if the value of meeting remedial goals one year sooner exceeded $14,000 (US). A value lower than $14,000 US favored NC. For our site, Cook et al. (2010) estimated establishment costs at $12/tree for best survival. If one assumes monitoring costs are equivalent for any remedial technology, then differences in cost between technologies would focus on installation and maintenance costs. As noted above, groundwater phytoremediation is competitive for establishment costs, but the value of remediation time differentiates technologies from each other.

Clearly, the intersection of decision making with cost benefit analyses is an area that merits further attention, particularly in times of economic constriction. Interestingly, groundwater phytoremediation sites comprise only 5 percent of all in situ groundwater treatment projects (USEPA, 2010). This situation may reflect slower integration of the maturing practice of groundwater phytoremediation, poor dissemination of results for successful sites, or other economic interests and priorities in remedial design. The potential to utilize groundwater phytoremediation systems as primary or secondary treatment systems will hopefully be considered more in the future as part of an evolving, iterative management strategy for complex groundwater sites (Simon, 2013) and to meet future interests to “green” remediation sites and offset other energy-consumptive remedial technologies (USEPA, 2011a).


This work was supported by USEPA/NC DENR Division of Water Quality 319 NPS Pollution Control Grant #EW06028 and #2862, the U.S. Coast Guard, B.P. (North America) Inc., and the U.S. Geological Survey Cooperative Water and Toxic Substances Hydrology Programs. Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.


  • Elizabeth Guthrie Nichols, PhD, is an associate professor in the environmental technology and management program in the department of Forest and Environmental Resources at North Carolina State University. She has a BS in biology from Emory University and an MS and PhD in environmental science and engineering from the University of North Carolina at Chapel Hill. Her primary research interests are how trees, in natural or managed systems, mitigate and remove contaminants and pollutants from soils, groundwater, and sediments.

  • Rachel L. Cook, PhD, is an assistant professor of soil fertility at Southern Illinois University, Carbondale, IL. Her research interests include improving soil management practices and fertilizer use for improved crop production and protection of environmental resources. She received her BS in biology from Saint Louis University and her master's in natural resources and PhD in forestry and environmental resources from North Carolina State University.

  • James E. Landmeyer, PhD, has been a research hydrologist with the U.S. Geological Survey, South Carolina Water Science Center, in Columbia, SC, since 1990. Landmeyer received his BS from Allegheny College in 1989, and his MS and PhD from the University of South Carolina in 1991 and 1995, respectively. He has been the author or co-author of more than 70 peer-reviewed publications, and in 2011 authored the textbook Introduction to Phytoremediation of Contaminated Groundwater. His research interests include the interaction between plants, microbes, and pristine and contaminated groundwater and surface-water systems.

  • Brad Atkinson is a Brownfields project manager for the North Carolina Department of Environment and Natural Resources. His background is in oil and gas exploration in the Midcontinent, and he has approximately 22 years of experience as a hydrogeologist in North Carolina. He received his BS in geology from the University of Kansas and his focus includes cost-effective assessment and remediation technologies.

  • Donald R. Malone, P.E., is a principal engineer and senior project manager of the environmental division at ARCADIS U.S., Inc. He has approximately 25 years of environmental consulting experience coordinating and directing the successful completion of soil and groundwater remediation projects under the Resource Conservation and Recovery Act (RCRA) and North Carolina regulatory programs. His areas of expertise include project management, and corrective actions designs and implementation.

  • George Shaw is responsible for worldwide sales of Amplified Geochemical Imaging (AGI) passive sampling service for soil-gas, air, water, and sediment located in Newark, Delaware. He has over 21 years identifying and proving new uses for passive sampling. He received a BS at the University of Tulsa in education with minors in math and science.

  • Leilani Woods, P.E., is an environmental engineer with the US Coast Guard in Elizabeth City, NC. Her focus includes implementing various innovative remediation and compliance technologies at stations along the East Coast. Ms. Woods received her BS in civil engineering from the University of Nebraska and her MS in civil engineering there as well.