JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 8 August 2008
Published in J Environ Qual 37:1701-1709 (2008)
DOI: 10.2134/jeq2007.0521
© 2008 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Anderson, R. H.
Right arrow Articles by Lanno, R. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Anderson, R. H.
Right arrow Articles by Lanno, R. P.
Agricola
Right arrow Articles by Anderson, R. H.
Right arrow Articles by Lanno, R. P.
Related Collections
Right arrow Toxic Trace Metals
Right arrow Ecological Risk Assessment
Right arrow Heavy Metals
Right arrow Soil Pollution
Right arrow Soil Chemistry

TECHNICAL REPORTS

Ecological Risk Assessment

Partitioning Species Variability from Soil Property Effects on Phytotoxicity: ECx Normalization Using a Plant Contaminant Sensitivity Index

R. H. Andersona, N. T. Bastaa,* and R. P. Lannob

a School of Environment and Natural Resources, Ohio State Univ., 2021 Coffey Rd., Columbus, OH 43210
b Dep. of Entomology, Ohio State Univ., 318 West 12th Ave., Columbus, OH 43210. Salaries and support provided by state and federal funds appropriated to the Ohio Agricultural Research and Development Center, The Ohio State Univ., Columbus, OH 43210

* Corresponding author (basta.4{at}osu.edu).

Received for publication October 3, 2007.

   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
Soil properties mitigate hazardous effects of contaminants through soil chemical sequestration and should be considered when evaluating ecological risk from terrestrial contamination. Empirical models that quantify relationships between soil properties and toxicity to ecological receptors are necessary for site-specific adjustments to ecological risk assessments. However, differential sensitivities of test organisms in dose–response studies may limit the utility of such models. We present a novel approach to toxicity estimation that partitions the effect of differential sensitivities of test organisms from that of soil chemical/physical properties. Five soils that ranged in selected properties were spiked with five concentrations of sodium arsenate. Bioassays were conducted where above ground dry matter growth and the corresponding tissue arsenic concentrations were evaluated for three terrestrial plants (Alfalfa, Medicago sativa L.; Perennial ryegrass, Lolium perrene L.; and Japanese millet, Echinochloa crusgalli L.). Estimates were combined into a plant contaminant sensitivity index (PCSI) and used to normalize phytotoxicity parameters to the most sensitive species (i.e., alfalfa) where necessary. Simple linear regression and ANCOVA indicated a 36.5% increase in the explanatory power of the modifying effects of soil properties on phytotoxicity when differential arsenate sensitivities were accounted for by PCSI (r2 = 0.477–0.833). Normalization of ecotoxicity parameters by PCSI is a seemingly effective approach to quantify the modifying effects of soil properties on phytotoxicity endpoints when it is of interest to consider multiple plant species (or varieties within a species) with differential sensitivities to experimental contaminants.

Abbreviations: DMG, dry matter growth • EC, electrical conductivity • ICP-OES, inductively coupled plasma–optical emission spectroscopy • OC, organic carbon • PCSI, plant contaminant sensitivity index • PMUCV, plant metal upper critical value • SRM, standard reference material


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
POTENTIAL hazardous effects of environmental contaminants on nonhuman receptors are the focus of ecological risk assessments. Toxicity and bioaccumulation are dominant processes considered when evaluating risk from exposure of ecological receptors to terrestrial contamination. Various pathways exist in environmental compartments for contact between toxic contaminants and organisms, resulting in variable levels of exposure. However, not all forms of contaminants can be absorbed by organisms (i.e., bioavailable), nullifying exposure as the sole requisite risk to toxicity and bioaccumulation. Furthermore, bioavailability of soil contaminants is modified by soil matrix properties through reactions including precipitation, complexation, adsorption, and chelation. Dominant soil chemical/physical properties known to affect the bioavailability of contaminants are soil pH; organic carbon (OC); cation exchange capacity; clay content; and reactive iron, aluminum, and manganese oxides (Basta et al., 2005; Fairbrother et al., 2007).

Studies have described the modifying effect of soil properties on the bioavailability of contaminants to ecological receptors. Among the most important of the receptors studied are plants because they provide the foundation to the food web for most terrestrial ecosystems and because plant roots are in direct contact with soil. Multiple regression techniques have been used to describe the influence of soil properties on arsenic toxicity to barley and tomato plants (Song et al., 2006; Zhao et al., 2006), cadmium uptake by a hyperaccumulator (Yanai et al., 2006), cadmium toxicity to lettuce plants (Bowers et al., 1997), cadmium and zinc uptake by lettuce plants (Peijnenburg et al., 2000), and copper toxicity to barley and tomato plants (Rooney et al., 2006). Additionally, structural equation modeling via path analysis has been used to describe the collinear associations among soil properties and their combined effect on lead toxicity to lettuce plants (Dayton et al., 2006). Three reviews of soil contaminant bioavailability have summarized additional studies (Basta et al., 2005; Kirkham, 2006; Fairbrother et al., 2007).

Although these studies have described the role of soil properties on contaminant bioavailability through bioaccumulation and phytotoxicity endpoints, a universal approach that quantifies the modifying effect of soil properties on phytotoxicity across different plant species remains operationally difficult. Individual studies are usually applicable only to a particular species because typical dose–response studies are designed using a single "model" species. Problems emerge when the test species does not reflect the tolerance or sensitivity to contaminants of species present on contaminated sites (Lepp et al., 1997). A typical remedy is to use a sensitive test species in dose–response experiments as a conservative estimate. However, when multiple studies with different test species or varieties within a test species are compared, systematic bias due to differential contaminant sensitivities often confounds the results by creating too much additional variability (Clark et al., 2004). Thus, a system of normalization is necessary to integrate dose–response studies comprised of test species with differential contaminant sensitivities to develop more robust predictive phytotoxicity models for ecological risk assessments based on soil property measurements.

We propose a normalization technique that partitions the effect of differential contaminant sensitivities of test plants from that of soil chemical/physical properties. This approach allows combining data from multiple test plant species in dose–response experiments and possibly future meta-analyses creating unified predictive models based on soil property measurements. The fundamental framework for the proposed normalization procedure is a modification of the plant metal upper critical value (PMUCV) concept developed by Beckett and Davis (1977) and tested by Davis and Beckett (1978). Davis and Beckett (1978) described phytotoxicity as a function of increasing contaminant uptake with plateau-linear functions independent of growing conditions. Plateau-linear functions can be fit to toxicity data comprised of multiple soils, as is the case in dose–response studies designed to evaluate the effect of soil properties. The total area under the function can be used as a representative index of the overall sensitivity of a plant to a contaminant or the plant contaminant sensitivity index (PCSI). Once estimated, PCSI can be used to normalize phytotoxicity parameters of species with significantly greater contaminant tolerances to that of a more sensitive species by the ratio of their respective PCSI as illustrated in Fig. 1 according to Eq. [1]:

Formula 1[1]
Thus, the PCSI approach to phytoxicity normalization is subjective. Assumptions associated with the application of the PCSI normalization procedure include the following conditions: (i) Plants should have similar physiological uptake mechanisms or pathways for the contaminant of interest, and (ii) metabolism within the plants should be similar. For example, in the current study, none of the plants were hyperaccumulators, and it was presumed that contaminant uptake followed similar pathways with no differences in root exclusion. It was also assumed that metabolism and intracellular contaminant translocation and accumulation were similar for each species.


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 1. Conceptual framework for the development of a plant contaminant sensitivity index (PCSI) and PCSI normalization of estimated phytotoxicity parameters for robust (i.e., less sensitive) plants to that of a more sensitive plant.

 
Specific objectives of the current study were (i) to demonstrate significant differential contaminant sensitivities in higher plants by evaluating relationships between phytotoxicity parameters and soil properties, (ii) to illustrate the PCSI normalization approach, and (iii) to test PCSI normalized phytotoxicity parameters for collapsibility across plants. The PCSI normalization approach is illustrated for arsenate phytotoxicity to the nonhyperaccumulating higher plants: alfalfa (Medicago sativa L.), perennial ryegrass (Lolium perrene L.), and Japanese millet (Echinochloa crusgalli L.).


   Materials and Methods
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results and Discussion
 Conclusions
 REFERENCES
 
Soil Selection and Characterization
Five uncontaminated soils that varied in selected physical and chemical properties were collected (0–20 cm) from within the USA for use in dose–response experiments (Tables 1 and 2 ). All soils were collected in bulk and processed to obtain uniform material. Soils were air dried and sieved (<2 mm) to remove roots, rocks, and other debris. Once sieved, soil was combined and homogenized for 4 h.


View this table:
[in this window]
[in a new window]

  		Table 1. Selected experimental soils and their USDA classification.

 

View this table:
[in this window]
[in a new window]

  		Table 2. Selected soil chemical and physical properties.

 
Soil Organic Carbon
Acid dichromate digestion was used to determine soil OC content (Heanes, 1984). Soil (0.5 g, <0.15 mm) was weighed into glass digestion tubes with 10 mL of H2SO4 and 5 mL of 0.167 mol L–1 K2Cr2O7. Reagent-grade sucrose was used to prepare calibration standards of OC. Soil and sucrose samples were treated similarly. Two reagent blanks were also digested. Digested samples, standards, and blanks were diluted to 50 mL in volumetric flasks and filtered through 0.45-µm, acid-resistant membrane filters. Absorbance of chromium (III) was determined colorimetrically at 600 nm by a spectrophotometer for digested samples, calibration standards, and blanks. A calibration curve was generated using sucrose and was used to determine the amount of OC in samples. Samples were digested and analyzed in triplicate.

Soil pH
Soil pH was determined using two different extracting solutions and soil/solution ratios according to Thomas (1996). Separately, 10 mL of deionized water and 20 mL of 0.01 mol L–1 CaCl2 solution was added to soil (10 g, <2 mm) inside 50-mL centrifuge tubes. Centrifuge tubes were then placed on a reciprocal shaker at 120 oscillations min–1 for 30 min. Sample suspensions were separated using centrifugation (0.5 h at 3444 x g). Soil pH was measured using a combination pH electrode immersed in the supernatants of separated suspensions. Samples were extracted and measured in duplicate.

Reactive Iron, Aluminum, and Manganese Oxides
Reactive iron (Fe), aluminum (Al), and manganese (Mn) oxide content was determined by acid ammonium oxalate extraction (McKeague and Day, 1993). Soil (5 g, <250 µm) was added to 25 mL of 0.20 mol L–1 acid ammonium oxalate and placed on a reciprocal shaker at 120 oscillations min–1 for 2 h in the dark. Sample suspensions were then separated using centrifugation (0.5 h at 11,158 x g), and supernatants were filtered with acid-tolerant, 0.45-µm filters into 100-mL volumetric flasks. Extracts were further diluted when necessary and analyzed by inductively coupled plasma–optical emission spectroscopy (ICP-OES). Samples were extracted and analyzed in duplicate.

Particle Size Analysis
Particle size analysis was conducted following the pipette method of Kilmer and Alexander (1949) as described in Gee and Bauder (1986). Approximately 50 g of each soil (<2 mm) was weighed into 300-mL glass beakers and pretreated with 50 mL of 30% H2O2 to remove soil organic matter. Approximately 100 mL of water was added to each beaker, and sample suspensions were brought to a gentle boil using a hot plate. Samples were stirred regularly to avoid soil loss from overflow as a result of the oxidation of organic matter. The process was repeated until frothing subsided, indicating that oxidation of organic matter was complete. Samples were then dried at 75°C overnight and crushed to pass a 2-mm sieve. After pretreatment, 10 g of each soil was added to 10 mL of a Na-dispersing agent and 200 mL of deionized water inside a capped 450-mL sedimentation bottle. Bottles were placed on a reciprocal shaker at 120 oscillations min–1 for 16 h. Deionized water was added to bottles until an exact weight of 410 g was obtained, and the resulting suspensions were stirred with an electric mixer at 7000 rpm for 1 min. To determine the clay (<2 µm) fraction, 25-mL aliquots of the stirred suspensions were removed 2 cm from the surface at 210.73 min with an automatic pipette and transferred into 100-mL centrifuge tubes. Suspensions were separated by centrifugation (0.5 h at 3444 x g), and supernatants containing the remaining colloidal suspensions were transferred to preweighed porcelain crucibles. Crucibles were oven dried at 105°C for 24 h and reweighed. Stokes law and a correction for the salt effect was used to calculate % clay content (Gee and Bauder, 1986).

Preparation of Spiked Soils for Dose–Response Experiments
Uncontaminated soils were spiked with 0.5-L solutions prepared from reagent grade H2NaAsO4 to achieve 10, 50, 100, 200, and 300 mg kg–1 nominal soil arsenate concentrations. Solutions were added to 1 kg of each experimental soil in an aluminum pan. Spiking solutions were analyzed by ICP-OES to confirm concentrations. A total of 5 kg was prepared for each soil–concentration combination. Soil was equilibrated with arsenate solutions for 4 h to imbibe aqueous arsenate. If all of the aqueous solution was absorbed, additional deionized distilled water was added such that saturation was achieved, thoroughly mixed, and allowed to re-equilibrate for 4 h. Saturated soils were oven dried at 60°C for 24 h. Spiked and control soils underwent a total of four wet–dry cycles.

Care was taken to reduce artifacts from spiking. For example, trace elements added as salt to soil can result in a "salt effect" where heavy metal availability is greater in spiked soil (Basta et al., 2005). Repeated wet–dry cycles increased the reaction between the soil matrix and the spiked arsenate concentrations, reducing the "salt effect." As a precaution, the electrical conductivity (EC) of the spiked soils was determined after the wet–dry process. Fifty grams of soil was shaken with 50 mL of deionized distilled water for 2 h. Soil EC was then measured from filtered (0.45 µm) solutions. Soils that had an EC >1.5 dS m–1 were leached with deionized water until a soil EC <1.5 dS m–1 was obtained. Spiked soils that had EC <1.5 dS m–1 were not leached. Several of the more clayey soils formed large aggregates during the wet–dry process and were gently broken by tumbling spiked soil in a jar mill until aggregates were <2 mm. Final spiked soil arsenate concentrations were confirmed by microwave-assisted acid digestion using USEPA Method 3051 and ICP-OES. Recovery ranged between 99.1 and 103%. Thus, nominal spike values were used for all statistical evaluations.

Range-Finder Tests
Range-finder tests were conducted for each soil–plant combination of the five spiked concentrations to determine applied soil arsenate concentrations that resulted in complete phytotoxicity for the three experimental plants. Uncontaminated soils served as negative controls. To ensure that all soils had adequate nutrition, macronutrients were tested and adjusted. Plant-available phosphorus (P) and potassium (K) were determined using the Mehlich 3 extraction (Mehlich, 1984), with subsequent analysis by ICP-OES. Plant-available nitrogen (NO3–N and NH4–N) was determined by a 1 mol L–1 KCl extraction followed by automated flow injection analysis (Mulvaney 1996). If necessary, soils were fertilized to ensure a minimum of 60 mg kg–1 N, 30 mg kg–1 P, and 30 mg kg–1 K (Havlin et al., 1999). Spiked soil (800 g) was mixed with 50% (by volume) vermiculite in 1-L polystyrene growing containers. Vermiculite bulking agent was added to maintain aeration and drainage due to the re-packing of sieved soil. Air-filled vermiculite adds little mass to the soil mixture, so the dilution effect (by weight) is minimal. Twenty perennial ryegrass seeds were then counted, placed into each container, and firmly seated. Plants were grown in a controlled environment growth chamber with 18 h of light per day, day temperatures of 20°C, and night temperatures of 18.5°C. Soil moisture was maintained at 75% of each soil's water-holding capacity (data not shown) through biweekly application. Twenty Japanese millet and alfalfa seeds were planted and subject to the same growing conditions and measurements. All plants were grown for 3 wk. Alfalfa seeds were inoculated with N-fixing bacteria before planting.

Definitive Tests
The range of spiked soil arsenate concentrations used in range-finder tests was determined to be appropriate for use in definitive tests. Test conditions and procedures were similar to those for range-finder tests with the exception that plants were grown for 40 d. Each soil was again tested for available N-P-K before planting, and, if needed, soil was fertilized with an equivalent 60 mg kg–1 N, 30 mg kg–1 P, and 30 mg kg–1 K (Havlin et al., 1999). Spiked soil (800 g) was again mixed with 50% (by volume) vermiculite in 1-L polystyrene growing containers.

After preparation of growing the containers, definitive tests were conducted. Fifty seeds were planted, but only 20 were allowed to grow after germination (i.e., post-emergence). Plants were grown in two replicate aliquots of each soil–concentration combination, including negative controls, arranged in a completely randomized design. Uncontaminated, fertilized soils served as negative controls. Plants were grown in a controlled environment growth chamber with 18 h of light per day, daytime temperatures of 20°C, and night-time temperatures of 18.5°C. After 40 d, plants were harvested at the transition between the hypocotyl and root, rinsed in deionized water, oven dried at 70°C for 48 h, and crushed by hand. Dried samples were weighed to determine aboveground dry matter growth (DMG). To account for yield differences due to soil quality (e.g., acidity, texture, etc.), DMG relative to the control or relative dry matter growth (RDMG) was calculated according to Eq. [2]:

Formula 2[2]
Metal hydrolysis associated with spiking (i.e., dosing) soils with cationic transition metal salts can decrease soil pH (Basta and Tabatabai, 1992). However, no consistent trends in soil pH were observed after definitive tests when uncontaminated negative control soils were compared with soils spiked at the highest rate (data not shown). Plant bioassays did not affect the OC content of experimental soils (data not shown). Consequently, soil property measurements taken from negative control soils were used for all statistical evaluations.

Plant Tissue Analysis
Tissue arsenic (As) in the aboveground biomass of plants grown in soils used in the definitive tests was determined by nitric acid digestion (Zarcinas et al., 1987). In this method, 5 mL of 40% (w/v) Mg(NO3)2 solution and 5 mL of concentrated HNO3 were added to 0.25 g of plant material in porcelain crucibles. Crucibles were covered with watch glasses, placed on a hot plate at 70 to 80°C, and allowed to reflux overnight. The next morning, watch glasses were removed, and the temperature of the plate was increased to 200°C and heated until solutions completely evaporated. Crucibles were placed inside a muffle furnace and heated to 150°C. The temperature was slowly increased to 450°C with a ramp time of 0.8°C min–1 and held for 6 h, and crucibles were placed back on the hot plate. Next, 2 mL of deionized water and 2 mL of concentrated HCl were added to the ash-filled crucibles and heated at 100°C for 1 h. Samples were transferred to 25-mL volumetric flasks and diluted to volume with 0.10 mol L–1 HCl. Solutions were transferred to 20-mL vials for later analysis by ICP-OES.

Quality Assurance and Quality Control
Blanks, spikes, and certified standard reference materials (SRMs) were digested and analyzed for quality assurance and quality control of contaminant measurements in soil and plant tissue samples. Blanks, spikes, and SRMs were evaluated for every six samples of soil or plant tissue. Spike recoveries ranged from 92 to 99%. Mean As recoveries in soil SRMs (CRM020–050) ranged from 97 to 99% (relative standard deviation, 1.8%), and mean recoveries in plant SRMs (aquatic plant BCR No. 60) ranged from 92 to 95% (relative standard deviation, 2.8%).

Data Analysis
Phytotoxicity Determination
Aboveground DMG was regressed on the range of replicated spiked soil arsenate concentrations, including negative controls (i.e., 0 mg kg–1 of applied soil arsenate), for each soil–plant combination based on dry soil spike concentrations. Best-fit functions describing relationships were analyzed using nonlinear regression according to Environment Canada (2005). Spiked arsenate concentrations that resulted in both a 20% and 50% reduction in DMG relative to the control response (as estimated by the y-intercept in best-fit functions) were determined (ECx). Estimates of ECx, their associated standard errors, and adjusted R2 values were obtained from the most explanatory of functions [3–6] recommended by Environment Canada (2005):

Formula 3[3]

Formula 4[4]

Formula 5[5]

Formula 6[6]
where yint is the y intercept, x is the spiked soil arsenate concentrations, z is the ECx fraction (i.e., 0.20 or 0.50), b is a scale parameter, and ECx is the toxicity parameter (i.e., EC20 or EC50).

Best-fit functions were determined through a systematic interpretation of results. Data were initially plotted to visually examine relationships between DMG and spiked soil arsenate concentrations. Adjusted R2 values were used to determine the most appropriate function. Data were not modeled if fewer than four spiked soil arsenate concentrations resulted in measurable DMG. Soil–plant combinations that failed to meet this criterion were labeled "No Data."

PCSI Determination
Aboveground dry matter growth was evaluated against the log of tissue As concentrations. A constant (1) was added to each tissue As concentration before log transformations to avoid excluding samples where no tissue As was detected (i.e., negative controls). Samples below detection (i.e., 1.0 mg kg–1) were assumed to contain 0 mg kg–1 tissue As and were evaluated as such. Plant metal upper critical value, slope, plateau, and y-intercept estimates along with their respective standard errors were determined using a segmented nonlinear function with the Marquardt numerical estimation algorithm.

Evaluation of the PCSI Normalization Approach
Relationships between conventional and normalized phytotoxicity parameters and soil property measurements were evaluated as a means to compare the efficacy of the proposed normalization approach. Initially, Pearson's linear correlation coefficients were used to determine the soil property or properties that significantly modified arsenate phytotoxicity. Multiple significant properties were determined and collectively indexed by the first component scores of a principal component analysis to prevent bias from simply selecting a property from which to evaluate relationships. Analysis of covariance was used to test the simultaneous equality of slope and y-intercept estimates among plants for relationships between phytotoxicity parameters and the significant modifying soil properties (i.e., first component scores). Because only five soils were used to evaluate the proposed normalization approach, all analyses were considered significant at {alpha} = 0.10. Phytotoxicity estimation, PCSI estimation, and statistical evaluations of the proposed normalization approach were conducted in SAS version 9.1 for windows (SAS Institute, 2001).


   Results and Discussion
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
Conclusions
 REFERENCES
 
Phytotoxicity
Phytotoxicity was evaluated using the aboveground DMG endpoint as opposed to more sensitive (i.e., sublethal) endpoints, such as root elongation. Preliminary analyses indicated that root elongation was not as effective as the DMG endpoint for our objectives because arsenate phytotoxicity was evaluated over a narrow range (i.e., 0–300 mg kg–1) as oppose to studies that used much larger ranges of soil contaminant concentrations (Song et al., 2006; Rooney et al., 2006). Increasing the range of spiked soil arsenate concentrations would have likely improved dose–response curves for the root elongation endpoint but would have reduced the accuracy of estimated phytotoxicity parameters for the DMG endpoint. Consequently, root elongation trials were discontinued, and DMG was considered the more sensitive endpoint and was used to estimate soil ECx values and to measure internal As concentrations.

Best-fit functions describing relationships between DMG and applied soil arsenate concentrations were determined for each soil–plant combination (Fig. 2 ). Linear, exponential, and logistic functions best described the relationships (Table 3 ). No obvious trends were observed among best-fit functions other than that DMG was negatively affected by increasing applied soil arsenate concentrations for all soil–plant combinations, as was expected from preliminary range-finder tests. Although no obvious trends were observed among best-fit functions, Richfield and Webster soils usually had greater DMG in the uncontaminated negative control soils (i.e., yield potential), followed by Kirkland soil, and Kirkland soil usually had greater yields than the acidic Teller and Sassafras soils (Fig. 2). Differences in yield potential among experimental soils had no consistent effect on the determination of the best-fit function. Insufficient measurable DMG (i.e., less than four spiked soil arsenate concentrations) prevented evaluation of the best-fit function for alfalfa in the acidic Teller soil and is most likely the result of alfalfa's greater sensitivity to Al toxicity exacerbated at low soil pH (Campbell, 1999).


Figure 2
View larger version (20K):
[in this window]
[in a new window]

  		Fig. 2. Dose–response curves between aboveground plant dry matter growth (DMG) and spiked soil arsenate concentrations for all soil–plant combinations.

 

View this table:
[in this window]
[in a new window]

  		Table 3. EC20 and EC50 values{dagger} estimated from aboveground dry matter growth response to As spike concentrations.

 
Spiked soil arsenate concentrations resulting in a 20% (i.e., EC20) and 50% (i.e., EC50) reduction in DMG relative to control responses (the y-intercept in best-fit functions) were determined for each soil–plant combination. Results are presented in Table 3. Estimates of EC20 and EC50 were usually larger for the warm-season species (i.e., Japanese millet) than for either of the cool-season species (i.e., alfalfa and ryegrass), meaning that greater applied soil arsenate concentrations were required for equivalent toxicity, as measured by DMG. No obvious trends emerged between the monocotyledonous (i.e., ryegrass and Japanese Millet) and dicotyledonous (i.e., alfalfa) species. Similar trends in phytotoxicity to trinitrotoluene, trinitrobenzene, and 2,4-DNT for these same three plant species have been observed (Rocheleau et al., 2006).

Several EC50 estimates were outside the range of spiked soil arsenate concentrations used in dose–response experiments. These parameters were labeled "N.A." and are not reported because using extrapolated parameter estimates could lead to erroneous conclusions. Soil–plant combinations where less than 50% phytotoxicity was observed resulted as an inherent artifact of the experimental design. However, a strong correlation (r2 = 0.858) was observed between the remaining (EC20, EC50) pairs (data not shown), suggesting that EC20 estimates are adequate to represent arsenate phytotoxicity in general. In fact, phytotoxicity estimates representing less than a 50% reduction in a measured endpoint are usually recommended in standard test methods (ISO, 2005; ASTM, 2003). Consequently, further evaluations were conducted with EC20 estimates only and considered sufficient for the objectives of the current study.

Effect of Soil Properties on Phytotoxicity
The effect of soil properties on arsenate phytotoxicity, as measured by DMG, was evaluated by investigating quantitative relationships between EC20 estimates and direct soil property measurements. Significant statistical associations were observed. Pearson's linear correlation coefficients (r) and their associated significance are presented in Table 4 . Significant correlations were observed between EC20 estimates and OC, Fe, and Mn oxide content; total oxide content (i.e., the summation of Fe, Mn, and Al oxide fractions); and total clay content for ryegrass and Japanese millet, similar to results observed by Song et al. (2006). However, in our study, total reactive oxide content had higher correlation coefficients than any oxide fraction individually across all three plants (Table 4). Significant correlations were not observed for alfalfa, although consistent trends were observed in the correlation coefficients for all three plants (Table 4). However, because of the lower survival rate for alfalfa due to Al toxicity in the acidic Teller soil, a smaller sample size (n = 4) was available for statistical inference.


View this table:
[in this window]
[in a new window]

  		Table 4. Pearson correlation coefficients and associated significance between selected soil properties and estimated EC20 values.

 
Differential Plant Arsenate Sensitivities
Statistical differences in the relationships between phytotoxicity parameters and soil properties among the experimental plants must exist to address the primary objective of the current study, which is to empirically illustrate the normalization of phytotoxicity estimates to better evaluate the effect of soil properties. Otherwise, there would be no formal distinction among plants allowing the combination of dose–response data without the need for normalization. Multiple soil properties were correlated with EC20 estimates (Table 4), which made the demonstration of any potential significant differential arsenate sensitivity operationally difficult. A logical solution was to select a single property for statistical evaluations, but this would have biased results to the property selected. Also, OC was significantly correlated with the total oxide (p = 0.051) and clay (p = 0.085) contents. Therefore, to evaluate differential arsenate sensitivities among the experimental plants, OC, total oxide, and total clay measurements were collectively indexed by the first component scores from a principal component analysis to eliminate the property bias (Fig. 3 ). Principal component analysis has been used to graphically illustrate dissimilarities among experimental soils (Peijnenburg et al., 2000), but it can also be used to condense multiple correlated variables into a single independent index (i.e., the first component scores) when multicollinearity is experimentally unavoidable (Graham, 2003).


Figure 3
View larger version (18K):
[in this window]
[in a new window]

  		Fig. 3. Principal component analysis used to condense multiple modifying soil properties into a single independent variable, the first component scores.

 
To effectively demonstrate differences (or lack thereof) in the relationships between EC20 estimates and soil properties (i.e., first component scores) among the experimental plants, ANCOVA was used to test the simultaneous equality of slope and y-intercept estimates. A significant departure in parallelism was detected (f2,8; p = 0.028) and was due to Japanese millet because no differences in slope estimates were detected between alfalfa and ryegrass (t8; p = 0.689). No differences in y-intercept estimates were detected between alfalfa and ryegrass (t10; p = 0.790). Because no differences in the slope or y-intercept estimates were detected between alfalfa and ryegrass, data were combined (represented by the dashed regression line in Fig. 4 ). Thus, arsenate phytotoxicity, when evaluated across soils, was less severe for Japanese millet relative to alfalfa and ryegrass where the effect was determined to be statistically equivalent, effectively demonstrating significant differential arsenate sensitivities.


Figure 4
View larger version (14K):
[in this window]
[in a new window]

  		Fig. 4. Relationships between soil properties (first component scores) and EC20 estimates across plants. Analysis of covariance indicated a significant departure in parallelism (f2,8; p = 0.028) due to Japanese millet because no differences in slope estimates were detected between alfalfa and ryegrass (t8; p = 0.689). No differences in y-intercept estimates were subsequently detected between alfalfa and ryegrass (t10; p = 0.790). Because no differences in the slope or y-intercept estimates were detected between alfalfa and ryegrass, data were combined and are represented by the dashed regression line.

 
Proposed Normalization of Toxicity Parameters using the PCSI Approach
Phytotoxicity parameters estimated from dose–response studies reflect not only the contaminant sequestration capacity of the growing medium but also the inherent toxicity of a contaminant to the plant (Hall, 2002; Poynton et al., 2004; Sharma and Dietz, 2006; Clemens, 2006). The primary objective of the current study is to illustrate the normalization of phytotoxicity parameters to more accurately reflect the arsenate sequestration capacity of soil. The normalization approach proposed involves the ratio of a PCSI for two plants where differential contaminant sensitivities are observed. In the current study, we have demonstrated a significantly greater tolerance to arsenate by Japanese millet relative to alfalfa and ryegrass, which resulted in a seemingly greater soil property effect (i.e., slope estimate) for Japanese millet.

Log transformations are a much simpler yet potentially comparable alternative to the proposed PCSI approach that suppresses the distribution of a response variable. Log transformation of toxicity parameters are used to normalize ecotoxicological data for evaluation of environmental factors, such as the effect of soil properties, across multiple species (Song et al., 2006; Zhao et al., 2006). Log10 transformation of the EC20 values eliminated the inconsistency in slope estimates, due to Japanese millet, when re-evaluated against soil properties (i.e., first component scores) because results from ANCOVA suggest a parallel lines model where no difference in slope estimates were detected (f2,8; p = 0.995). However, significantly different y-intercepts between alfalfa and Japanese millet (t10; p = 0.001) and ryegrass and Japanese millet (t10; p = 0.001) were also detected. Log transformation effectively normalized the soil property effect on arsenate phytotoxicity, but variability in EC20 estimates due to species level differences was still evident (Fig. 5 ). Differences in y-intercepts were not detected between alfalfa and ryegrass. Consequently, data were combined (represented by the dashed regression line in Fig. 5).


Figure 5
View larger version (14K):
[in this window]
[in a new window]

  		Fig. 5. Relationships between soil properties (first component scores) and log10 transformed EC20 estimates across plants. Analysis of covariance indicated no departure in parallelism (f2,8; p = 0.995) across plants. A subsequent test determined that y-intercept estimates significantly differed between alfalfa and Japanese millet (t;10; p < 0.001) and between ryegrass and Japanese millet (t;10; p = 0.001). No differences were detected in y-intercept estimates between alfalfa and ryegrass (t10; p = 0.655). Because no differences in the slope or y-intercept estimates were detected between alfalfa and ryegrass, data were combined and are represented by the dashed regression line.

 
To estimate the PCSI of the three experimental plants, plateau-linear functions established under the original PMUCV concept were applied (Beckett and Davis, 1977). Plateau-linear functions are graphically illustrated in Fig. 6 , and the estimated PCSI parameters are shown in Table 5 . Consistent with the relationship observed between EC20 estimates and soil properties (i.e., first component scores), Japanese millet had the highest PCSI because it had the highest PMUCV estimates, which represent threshold tissue contaminant concentrations (Beckett and Davis, 1977), along with the lowest absolute PCSI slope estimate (Table 5). The larger PMUCV estimate for Japanese millet was due to a larger inherent physiological tolerance to arsenate, whereas the substantially smaller absolute PCSI slope estimate was due to a less severe toxic effect per unit increase in tissue As concentration (Meharg and Hartley-Whitaker, 2002). Although no differences were determined between alfalfa and ryegrass for the relationship between their respective EC20 estimates and soil properties (i.e., first component scores), alfalfa had a smaller PCSI than ryegrass (Table 5). However, their PCSI slope estimates were very similar (Table 5).


Figure 6
View larger version (18K):
[in this window]
[in a new window]

  		Fig. 6. Plateau-linear functions representing relationships between dry matter growth relative to the control, relative dry matter growth (RDMG), and log10 transformed tissue As concentrations among plants used to estimate a plant contaminant sensitivity index.

 

View this table:
[in this window]
[in a new window]

  		Table 5. Estimated parameters{dagger} necessary for calculation of the plant contaminant sensitivity index (PCSI).

 
Because there were no statistical differences in the relationship between EC20 estimates and soil properties (i.e., first component scores) for alfalfa and ryegrass, normalization of either species was unnecessary. However, the significantly greater tolerance of Japanese millet to arsenate relative to alfalfa and ryegrass presented an opportunity to illustrate the PCSI normalization approach. Because the PCSI normalization approach is subjective, alfalfa was used to normalize Japanese millet EC20 estimates because alfalfa had the lowest PCSI (Table 5) or the overall greatest sensitivity to arsenate. Consequently, normalized EC20 estimates for Japanese millet were calculated by equating the ratio of the PCSI for alfalfa and Japanese millet to the ratio of normalized EC20 estimates and the conventional EC20 estimates for Japanese millet and solving for the normalized parameters (Fig. 1). Comparison of the relationship between alfalfa normalized Japanese millet EC20 estimates in combination with the EC20 estimates for alfalfa and ryegrass with soil properties (i.e., first component scores) using simple linear regression indicated a 36.5% increase in explanatory power over that from the conventional Japanese millet parameters (Fig. 7 ). Furthermore, results from ANCOVA suggested the simultaneous equality of slope (f2,8; p = 0.733) and y-intercept (f1,6; p = 0.162) estimates among all three experimental plants. Thus, the PCSI approach not only normalized the soil property effect on arsenate phytoxicity, similar to the log transformation, but also normalized species level differences critical to a unified comprehension of studies of soil property effects on phytoxicity comprised of multiple test species.


Figure 7
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 7. Relationships between soil properties (first component scores) and (A) conventional EC20 estimates and (B) plant contaminant sensitivity index normalized EC20 estimates among plants.

 

   Conclusions
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
REFERENCES
 
Relationships between phytotoxicity parameters and soil properties revealed a significantly greater arsenate tolerance for Japanese millet relative to alfalfa and perennial ryegrass. The greater arsenate tolerance of Japanese millet inflated the nominal effect of soil properties on arsenate phytoxicity. The PCSI suggested the inherent phytotoxicity of arsenate to the experimental plants was in the order alfalfa > perennial ryegrass > Japanese millet. Phytotoxicity parameters (i.e., EC20 values) estimated for Japanese millet were normalized by equating the ratio of the PCSI for alfalfa and Japanese millet to the ratio of normalized EC20 estimates and the conventional EC20 estimates for Japanese millet and solving for the normalized parameters. The PCSI approach and log10 transformation of phytoxicity parameters effectively normalized the effect of soil properties on phytotoxicity. However, the PCSI approach also removed the bias from the significantly greater arsenate tolerance of Japanese millet. Because PCSI provides integrated toxicity information, we conclude that it would be beneficial to report as a routine parameter in phytotoxicity studies. Adjustment of ecotoxicity parameters by PCSI is seemingly an effective approach to quantify the modifying effects of soil properties on phytotoxicity endpoints when it is of interest to consider multiple plant species (or varieties within a species) with differential sensitivities to experimental contaminants.


   ACKNOWLEDGMENTS
 
We thank the Technical Editor, the Associate Editor, and anonymous reviewers for their comments, which improved the manuscript. The research in this manuscript was funded by the Strategic Environmental Research and Development Program, Arlington, VA through project CU-1210.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Anderson, R. H.
Right arrow Articles by Lanno, R. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Anderson, R. H.
Right arrow Articles by Lanno, R. P.
Agricola
Right arrow Articles by Anderson, R. H.
Right arrow Articles by Lanno, R. P.
Related Collections
Right arrow Toxic Trace Metals
Right arrow Ecological Risk Assessment
Right arrow Heavy Metals
Right arrow Soil Pollution
Right arrow Soil Chemistry

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome