Stimulated Microbial Reductive Dechlorination following Surfactant

A pilot-scale demonstration of surfactant-enhanced aquifer remediation (SEAR) was conducted in July 2000 at the Bachman Road site located in Oscoda, M...
0 downloads 0 Views 542KB Size
Environ. Sci. Technol. 2004, 38, 5902-5914

Stimulated Microbial Reductive Dechlorination following Surfactant Treatment at the Bachman Road Site† C . A N D R E W R A M S B U R G , * ,‡ LINDA M. ABRIOLA,‡ KURT D. PENNELL,§ FRANK E. LO ¨ FFLER,§ MATTHEW GAMACHE,| BENJAMIN K. AMOS,§ AND ERIK A. PETROVSKIS⊥ Department of Civil and Environmental Engineering, Tufts University, 200 College Avenue, 113 Anderson Hall, Medford, Massachusetts 01255, School of Civil and Environmental Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta, Georgia 30332, CDM, Inc., One Cambridge Place, 50 Hampshire Street, Cambridge, Massachusetts 02139, and GeoSyntec Consultants, 4312 Tara Court, Ann Arbor, Michigan 48103

A pilot-scale demonstration of surfactant-enhanced aquifer remediation (SEAR) was conducted in July 2000 at the Bachman Road site located in Oscoda, MI. The Bachman aquifer is a shallow, relatively homogeneous, unconfined aquifer formation composed primarily of sandy glacial outwash with relatively low organic carbon content (0.02 wt %). A 6 wt % aqueous solution of Tween 80 (a nonionic, food-grade surfactant) was flushed through a localized dense nonaqueous phase liquid (DNAPL) source zone to recover approximately 19 L of tetrachloroethene (PCE). Post-treatment monitoring revealed PCE concentrations were reduced by up to 2 orders of magnitude within the source zone, and there was no evidence of concentration rebound after more than 450 d. Concentrations of PCE dechlorination products (trichloroethene, cis-1,2dichloroethene) 450 d after SEAR operations ceased were more than 2 orders of magnitude greater than pretreatment values, suggesting stimulation of native dechlorination activity. Post-treatment monitoring detected increased concentrations of volatile fatty acids generated from the fermentation of residual-level Tween 80 surfactant. These field data suggest that Tween 80 not only induced and maintained anaerobiosis but also provided reducing equivalents to reductively dechlorinating populations present in the oligotrophic Bachman aquifer. Experience from this site supports application of staged treatment strategies that couple SEAR and microbial reductive dechlorination to enhance mass removal and reduce contaminant mass flux emanating from treated source zones.

* Corresponding author telephone: (617)627-3211; fax: (617)6273994; e-mail: [email protected]. † This paper is part of the Walter J. Weber Jr. tribute issue. ‡ Tufts University. § Georgia Institute of Technology. | CDM, Inc. ⊥ GeoSyntec Consultants. 5902

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004

Introduction Groundwater contamination by chlorinated solvents is a serious problem at both large- and small-scale hazardous waste sites across the United States and other industrialized countries (1, 2). The presence of dense nonaqueous phase liquids (DNAPLs) at numerous contaminated groundwater sites is one of the most difficult and consequently costly challenges facing the remediation community (1, 3, 4). Over the past decade, considerable effort has been directed toward remediation of sites impacted by chlorinated solvents. Recently, concerns related to the effectiveness and longterm success of aggressive remediation have renewed debate on the constitution of appropriate cleanup goals and the metrics used for performance assessment (5-11). The region of an aquifer containing DNAPL often serves as a long-term source for dissolved-phase contaminant plumes and, hence, is frequently referred to as the contaminant source zone. Realization that conventional pumpand-treat remediation of DNAPL source zones is only effective for containment has shifted focus toward development of innovative methods to recover DNAPL mass from the subsurface (12, 13). Many innovative technologies have now transitioned from well controlled laboratory studies to successful field-scale demonstrations of contaminant mass recovery (e.g., refs 14 and 15). For example, alcohol and surfactant flushing have been demonstrated to recover significant contaminant mass from DNAPL source zones (1519). While DNAPL mass recovery has been the focus of a number of demonstrations, comparatively less attention has been directed toward examining effects of remedial fluids on the long-term natural attenuation of the low-level contamination that remains following aggressive source-zone treatments (20-23). The past decade has seen a rapid increase in the understanding of the microbiology involved in efficient reductive dechlorination of tetrachloroethene (PCE) and its daughter products (24-30). All major bacterial phyla have members that capture energy for growth from reductive dechlorination reactions in a process known as chlororespiration (also known as dechlororespiration and choidogenesis). Chlororespiration requires a supply of reducing equivalents (electron donor), but in many oligotrophic subsurface environments electron donor supply limits microbial activity. In such environments, reductive dechlorination rates can often be enhanced by the addition of suitable electron donors (biostimulation) (31). All chlororespiring bacteria use hydrogen and/or acetate as direct electron donors, and thus the goal of biostimulation is to increase the availability of hydrogen and acetate to enhance reductive dechlorination activity (32). A number of simple and complex substrates have been shown to indirectly satisfy electron donor requirements of chlororespiring populations (28, 31-38). Fermentation processes, catalyzed by indigenous subsurface bacteria, contribute to the desired increased flux of hydrogen and acetate, which in turn serve as direct electron donors for the chlororespiring bacteria. Compounds resulting in a sustained slow release of hydrogen have been demonstrated to be ideal candidates for application in biostimulation treatments (28, 29, 32, 35-42). While biostimulation remediation strategies can enhance natural dechlorination processes within the dissolved-contaminant plume with limited capital expenditure, difficulties with electron donor delivery and solvent toxicity limit this approach in DNAPL source zones. The limitations of source-zone biostimulation result in source longevities that are typically measured in decades (43). Hence, long-term operational and maintenance costs 10.1021/es049675i CCC: $27.50

 2004 American Chemical Society Published on Web 10/19/2004

FIGURE 1. Location of multilevel (ML) sampling bundles within pilot-scale treatment zone. Sampler identification is alphabetical with depth (i.e., point A represents the most shallow sampler in each bundle). Approximate depths (m bgs) are given at each sampling location based on the angle (noted in parentheses) of each boring. Reference point (~) shown for orientation with Figures 9 and SI-1. Water and surfactant injection wells used during treatment are labeled W1-W3 and S1- S3, respectively. The extraction well used during treatment is denoted by EW (19). associated with biostimulation may offset any cost savings realized by the comparatively lower fixed-capital expenditures. In contrast, aggressive source-zone mass removal technologies significantly shorten source longevity but may require a staged treatment approach (i.e., a second stage “polishing step”) to achieve regulatory objectives. Combination of aggressive mass removal with stimulation of chlororespiring populations may provide a staged treatment approach that reduces contaminant mass flux to a level that ensures plume containment and ultimately site closure (43). Results from one cosolvent flushing field test (Sages site, Jacksonville, FL) suggest that use of 95% ethanol for sourcezone treatment (17) may have stimulated chlororespiring populations already present at the site (23). While the toxicity of high ethanol concentrations to subsurface microorganisms was a concern at Sages, sulfate-reducing, methanogenic, and chlororespiring populations rebounded downgradient from the treated zone (23, 44). Concerns related to the toxicity of remedial fluids to subsurface microbes within a treatment zone, however, may be alleviated through selection of remedial approaches that have limited impact on native dechlorinating populations (21, 45). This paper explores reductive dechlorination activity after flushing a PCE-DNAPL source zone at the Bachman Road site with an aqueous solution of readily biodegradable, food-grade, nonionic surfactant (Tween 80). Detailed descriptions of the pilotscale test design and implementation at this site are presented elsewhere (19, 46, 47). Here, the stimulation of chlororespiring bacteria indigenous to the Bachman aquifer is investigated through comparisons of pre- and post-treatment monitoring data, in conjunction with mathematical modeling of chemical transport.

Site History In late 1979, the Michigan Department of Public Health detected dissolved-phase organic compounds in water samples from private supply wells along Bachman Road, a residential area located near Lake Huron in Oscoda, MI. Analysis conducted in support of a remedial investigation initiated by the Michigan Department of Natural Resources showed groundwater in the surrounding area to be impacted by four contaminant plumes (Supporting Information, Figure SI-1). The source of the PCE contamination in plume B was associated with a former dry cleaning operation. In 1997, the Michigan Department of Environmental Quality (MDEQ), in conjunction with the Great Lakes and Mid-Atlantic Hazardous Substance Research Center (directed by Professor Walter J. Weber, Jr.), funded a feasibility study for a pilot-scale demonstration of surfactant-enhanced aquifer remediation (SEAR) of the suspected source. Results obtained during phase I site investigations indicated that the PCE source zone was located underneath the building (47). The Bachman Road site PCE source zone is located approximately 230 m west of Lake Huron (47). Regional groundwater flow is generally from the west (i.e., flowing toward Lake Huron) at an estimated velocity of 0.13 m/d (48). The unconfined contaminated aquifer is composed of relatively homogeneous, medium- to fine-grained, glacial outwash sands, having a total carbon (TC) content of approximately 0.21 wt %. The total organic carbon (TOC) content was determined to be 0.02 wt %, indicating the presence of a large fraction of inorganic carbon (47). The aquifer formation is underlain by a clay confining layer, consisting primarily of illite and smectite, located apVOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5903

FIGURE 2. Concentrations of PCE and Tween 80 during site monitoring for location ML-1. Light gray portions of each bar indicate the quantification level of that analysis. proximately 7.6 m below ground surface (bgs). Lenses of lower permeability media consisting of silty material were observed in two zones, immediately above and 4.2 m above the clay confining layer. The depth to groundwater varies seasonally between 2.4 and 3.0 m, and thus the unconfined aquifer thickness ranges between 4.6 and 5.2 m.

Summary of Pilot-Scale SEAR Demonstration Drive point samples and soil cores taken from underneath the northwest corner of the building (angled boring techniques) confirmed the presence of DNAPL. These data, however, were insufficient to quantify DNAPL volumes (47). On the basis of these characterization efforts, a 4.3 m by 6.7 m area beneath the building was selected for pilot-scale SEAR treatment. Single- and multi-phase transport models were employed to aid development of a unique remedial design intended to minimize disruption of ongoing commercial and residential activities inside the structure, while treating a localized volume of the aquifer lying below the building. On the basis of laboratory treatability tests and cost analyses (49), a commercially available nonionic surfactant (Tween 5904

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004

80) was selected for application (47). The average molecular weight and critical micelle concentration (CMC) of Tween 80 (polyethoxylated (20) sorbitan monooleate) were reported by Pennell et al. (50) to be 1310 g/mol and 9.9 µM, respectively. Regulatory authorization for the use of Tween 80 was obtained based on United States Food and Drug Administration (USFDA) approval of this surfactant for numerous food-grade applications. Available studies demonstrated no long-term adverse human health effects associated with the use of Tween 80 in foods, pharmaceuticals, or cosmetics (51). Furthermore, Tween 80 is known to be biodegradable under both aerobic and anaerobic conditions (52, 53), suggesting that negative impacts of residual surfactant on subsurface and aquatic ecosystems would be unlikely. Prior to surfactant flushing, five bundles of multilevel samplers (ML1-ML5) were installed beneath the building using angled boring methods for sampling of the pilot-scale treatment zone (Figure 1). After establishing hydraulic control (necessary for recovery of the injected surfactant solution and solubilized PCE) and completing tracer tests, a solution of 45.8 mM (6 wt %) Tween 80 was introduced over the entire

FIGURE 3. Concentrations of PCE and Tween 80 during site monitoring for location ML-2. Light gray portions of each bar indicate the quantification level of that analysis. aquifer thickness for 5 d (wells S1, S2, and S3 in Figure 1) (19). Surfactant injection continued for a second 5-d period through wells screened over the top and bottom 1.2 m of saturated depth (S2 and S3 in Figure 1). This design was employed to target regions suspected to be highly contaminated, while reducing surfactant usage and, consequently, costs (details available in ref 47). Water injection wells (W1W3) remained in operation for an additional 2 d, and the extraction well operated for 37 d after the cessation of surfactant flushing. Overall, approximately 95% of the injected surfactant was recovered, with 19 L of PCE extracted during pilot-scale operations (19).

Post-treatment Monitoring of Treated Zone Following the surfactant flush, efforts focused on monitoring of aqueous-phase concentrations both within the treated zone and at downgradient locations. Sampling techniques and analytical methods for quantifying PCE and Tween 80 are described in detail by Ramsburg et al. (19). Varying quantification limits at different sampling times resulted from utilization of methods tailored for rapid quantification within

a specific concentration range for purposes of high sample throughput during the pilot-scale test. Quantitative analysis of PCE degradation products was performed following U.S. EPA Method 8260 by contracted laboratories. Pre-SEAR samples were analyzed by laboratories associated with the National Center for Integrated Bioremediation Research and Development (NCIBRD), Ann Arbor, MI. Analysis of postSEAR samples was performed by the MDEQ analytical facility located in Lansing, MI. Relevant properties of PCE and its degradation products trichloroethene (TCE), cis-1,2-dichloroethene (cis-DCE), trans-1,2-dichloroethene (trans-DCE), and vinyl chloride (VC) are listed in Table SI-1 in the Supporting Information (54-56). Treated-zone monitoring was accomplished by sampling all multilevel wells installed in support of the pilot-scale demonstration at 14, 56, 270, and 450 d after the end of the SEAR test (Figure 1). Monitoring results for PCE and Tween 80 concentrations are presented on a log scale in Figures 2-6. In these figures, light gray portions of each bar indicate the quantification level of that analysis. Examination of these figures reveals that PCE concentrations within the swept VOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5905

FIGURE 4. Concentrations of PCE and Tween 80 during site monitoring for location ML-3. Light gray portions of each bar indicate the quantification level of that analysis. Monitoring location 3-A was located near the water table and did not produce sufficient water for analysis during the period “End SEAR” through “56 days post-treatment”. zone were reduced significantly following SEAR and did not rebound through a 450-d period after SEAR operations had ceased. As depicted in Figures 2-6, PCE concentrations decreased by more than 2 orders of magnitude at many locations from those levels measured immediately prior to SEAR (Start SEAR), and generally continued to decrease over time. Surfactant concentrations shown in Figures 2-6 exhibit a similar steady decline over time. At most sampling points Tween 80 was not quantifiable (nq) at the 12 µM level after 270 d. A noteworthy exception to these general trends was observed at ML3-E, where PCE and Tween 80 concentrations remained elevated (15 µM PCE and 2100 µM Tween 80) 450 d following the end of SEAR operations. It is hypothesized that this sampling location is within a low-flow or nearly stagnant region that was penetrated during SEAR, when flow gradients were altered, and subsequently isolated as site hydraulics returned to natural gradient conditions. The appearance of elevated surfactant concentrations, in ML1-E and ML2-E, beginning at 56 d post-treatment is consistent with this hypothesis suggesting a slow migration of the 5906

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004

residual surfactant plume in the deepest portion of the aquifer to these observation points downgradient of ML3-E. In contrast to the trace amounts of PCE daughter products measured prior to treatment, the presence of TCE in samples collected from bundle ML3 at 270 d post-treatment suggested that intrinsic reductive dechlorination had occurred. On the basis of these data, samples collected at the 450-d monitoring interval were more thoroughly analyzed for the formation of PCE daughter products (Table 1). The relative change between measured concentrations of daughter products before and after treatment is presented in Figure 7a-e. Here, relative concentration is defined as the ratio of the concentration at 450 d post-treatment (see Table 1) to the concentration at 320 d prior to the commencement of pilot-scale operations. These data reveal increased concentrations of PCE degradation products at several locations after treatment. Overall, the data shown in Figure 7 indicate that concentrations of cis-DCE and other PCE daughter products increased by up to 5 orders of magnitude in the treatment zone when compared to pretreatment concentration levels. While the

FIGURE 5. Concentrations of PCE and Tween 80 during site monitoring for location ML-4. Light gray portions of each bar indicate the quantification level of that analysis. Monitoring location 4-A was located near the water table and did not produce sufficient water for analysis during the period “End SEAR” through “450 days post-treatment”. logarithm of relative concentration effectively highlights concentration changes, this metric does not provide insight into the magnitude of actual concentrations. Thus, data shown in Figure 7 should be interpreted in conjunction with concentration data reported in Table 1. For example, the TCE concentration in ML1-C increased by almost 1 order of magnitude. The measured TCE concentration at this location, however, was only 0.06 µΜ. In contrast, the approximately 2 order of magnitude increase in TCE concentration observed at ML1-B corresponds to a measured value of 4.26 µΜ. Concentrations of cis-DCE within the treatment zone ranged up to 1032 µΜ (Table 1) and are indicative of cis-DCE accumulation. VC was only detected in three locations within the treatment zone (ML1-E, ML2-E, and ML3-E) and at low concentrations (100 µΜ) at location ML3-E.

The presence of fermentation products, the decrease in PCE concentrations, and the increase in dechlorination products TCE, cis-DCE, trans-DCE, and VC within the treatment zone are all consistent with the hypothesis that microbial reductive dechlorination had occurred. Anaerobic bacteria capable of metabolic reductive dechlorination of chlorinated ethenes are known to be present in the Bachman aquifer (30-32, 57-59). Specifically, Desulfuromonas michiganensis strain BRS1, an acetate-oxidizing PCE-to-cis-DCE dechlorinating organism (58), and Dehalococcoides sp. strain BAV1, a hydrogenotrophic DCE- and VC-respiring population, were isolated from the Bachman Road site (30). Also derived from the Bachman aquifer was a PCE-to-ethene dechlorinating consortium, containing strain BAV1 (i.e., the Bachman culture) that was used successfully for bioaugmentation and biobarrier establishment within a nearby plume (denoted “Plume A-Halorespiration Site” in Figure SI-1) (31). To further support the hypothesis that residual surfactant enhances in situ reductive dechlorination, laboratory experiments were undertaken to explore the influence of Tween 80 on dechlorination endpoints and rates (see Supporting Information for experimental methods). D. michiganensis VOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5907

FIGURE 6. Concentrations of PCE and Tween 80 during site monitoring for location ML5. Light gray portions of each bar indicate the quantification level of that analysis. strain BB1, an organism indistinguishable from strain BRS1 (58), was used in surfactant-amended cultures to examine dechlorination performance in the presence of Tween 80. The range of Tween 80 concentrations used in these experiments (0-3800 µM) was similar to that measured within the treated zone following the cessation of flushing at the Bachman Road site (Figures 2-6). Surfactant-amended cultures exhibited the same dechlorination end point (i.e., cis-DCE) as control cultures without Tween 80, and no significant differences (P > 0.05) in cis-DCE formation rates were observed (Figure 8). Note that, while similar amounts of PCE were added to each experimental system, the presence of Tween 80 altered the equilibrium partitioning between the aqueous and vapor phases, resulting in dissimilar initial aqueous-phase PCE concentrations for different Tween 80 concentrations. For example, Kibbey et al. (60) reported that the presence of approximately 5500 µM Tween 80 reduces the Henry’s law coefficient for PCE by approximately 98%. It should be emphasized, however, that this effect is much less significant for cis-DCE (Figure 8b) where partitioning from the aqueous phase into the headspace is substantially less (Supporting Information, Table SI-1). The small amount 5908

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004

of cis-DCE initially present (∼10 µM) was introduced with the inoculum (2% vol/vol). Results from these batch culture studies demonstrate that residual Tween 80 does not inhibit microbial reductive dechlorination of PCE to cis-DCE by D. michiganensis. While reductive dechlorination was shown to occur in the presence of Tween 80, it is unlikely that Tween 80 can directly support dechlorinating populations (22, 45, 53). The formation of organic acids within the treated zone indicates that anaerobiosis was established and that fermentation of residual Tween 80 occurred following physical-chemical DNAPL mass removal. Tween 80 represents a mixture of ethoxylated sorbitan monooleate compounds that possess on average 20 ethylene oxide groups that contribute hydrophilicity to this surfactant. Fermentation of poly(ethylene glycol) (polymerized ethylene oxide) to equal parts acetate and ethanol has been shown to occur under anaerobic conditions in a series of studies conducted by Schink and Stieb (61) and Dwyer and Tiedje (62, 63). In these studies, ethanol was subsequently fermented to acetate and hydrogen. Thus, the detection of acetate and formate within the treatment zone is significant as these organic acids are

FIGURE 7. Relative concentration of PCE, TCE, cis-DCE, trans-DCE, and VC in (a) ML1, (b) ML2, (c) ML3, (d) ML4, and (e) ML5. Monitoring locations ML3-A and ML4-A were located near the water table and did not produce sufficient water for analysis during the periods “End SEAR” through “56 days post-treatment” and “End SEAR” through “450 days post-treatment”, respectively. hydrogen precursors or directly support the reductively dechlorinating populations present in the Bachman aquifer (30, 32, 58, 59). The absence of detectable acetate at a given observation point does not necessarily indicate that reductive dechlorination is electron donor limited. He et al. (32) observed continued degradation of PCE to cis-DCE with no measurable acetate present in hydrogen-fed co-cultures of D. michiganensis strain BB1 (acetotrophic dechlorinator) and Sporomusa ovata (H2/CO2 acetogen). Their study indicates that H2/CO2 acetogenesis and acetate consumption by acetotrophic dechlorinators may occur at comparable rates, and thus steady-state acetate concentrations may remain too low to be detected using standard analytical procedures. Detection of VC at the Bachman Road site was associated with higher Tween 80 (39-2100 µM) and acetate (18204600 µΜ) concentrations in the same samples (Table 1). Similar observations have been made by McGuire and

Hughes, who observed increased accumulation of VC in PCEfed laboratory microcosms containing Tween 80 (45). It should also be noted that, of all the surfactants used by McGuire and Hughes (45) (i.e., Tween 80, Neodol 25-7, Steol CS-330, alkyl polyglycoside, R-oletin sulfonate 14-16, Aerosol MA 80-I, and CTAB), Tween 80 had the least detrimental effect on microbial reductive dechlorination of PCE, which is consistent with earlier studies of microbial reductive dechlorination of chlorobenzenes (21, 53). The three locations within the treated zone that contained measurable quantities of Tween 80 after 450 d are located in the region of a sand-silt-clay transition zone immediately above the clay layer present at approximately 7.6 m bgs (47). This material has a substantially lower hydraulic conductivity than that of the overlying sand (∼2 m/d vs ∼17 m/d). Clay collected from below this transitional zone had markedly greater maximum capacity for surfactant sorption than the VOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5909

TABLE 1. Monitoring Concentrations (µM) within Treated Zone at 450 d Post-treatmenta location ID

PCE

TCE

cisDCE

transDCE

A 3.5 m B 4.9 m C 6.0 m D 6.6 m E 6.8 m

0.19 31.4 0.38 0.78 0.34

0.14 4.26 0.06 0.18 nq

4.33 2.79 0.52 0.78 258

A 3.4 m B 4.7 m C 5.9 m D 6.5 m E 7.3 m

1.15 0.25 0.24 0.23 0.27

0.09 0.07 nq nq nq

2.27 2.17 1.34 1.13 186

A 3.0 m B 4.3 m C 5.5 m D 6.6 m E 6.8 m

10.9 35.6 3.14 2.77 15.7

A 2.5 m B 3.7 m C 5.0 m D 6.2 m E 6.8 m A 2.9 m B 4.1 m C 5.3 m D 5.9 m E 6.5 m F 7.2 m

VC

Tween 80

acetate

ML1 nq nq nq nq 0.87

nq nq nq nq 1.26

nq nq nq nq 39.8

nq nq nq nq 1820

ML2 nq nq nq nq 0.58

nq nq nq nq 1.30

nq nq nq nq 38.6

nq nq 286 500 2150

0.72 91.3 3.27 1.67 30.4

ML3 6.81 nq 175 2.79 3.09 nq 1.75 nq 1032 5.57

nq nq nq nq 6.56

nq nq nq nq 2100

nq 730 nq nq 4600

dry 0.96 0.28 0.17 0.20

dry 0.05 0.03 0.05 0.05

dry 0.47 0.45 0.17 0.24

ML4 dry 0.01 nq nq nq

dry nq nq nq nq

dry nq nq nq nq

dry nq nq nq nq

0.13 1.03 1.45 0.12 2.41 0.11

0.91 0.40 0.24 0.05 0.31 nq

4.85 5.88 1.13 0.18 1.24 12.4

ML5 nq nq nq nq nq nq

nq nq nq nq nq nq

nq nq nq nq nq nq

nq 485 nq nq nq 1140

a Not quantifiable (nq):