Volume 37, Issue 1 p. 21-33
Critical Review
Free Access

A review of measured bioaccumulation data on terrestrial plants for organic chemicals: Metrics, variability, and the need for standardized measurement protocols

William J. Doucette

Corresponding Author

William J. Doucette

Utah Water Research Laboratory, Utah State University, Logan, Utah, USA

Address correspondence to [email protected]

Search for more papers by this author
Chubashini Shunthirasingham

Chubashini Shunthirasingham

Air Quality Processes Research Section, Environment and Climate Change Canada, Toronto, Ontario, Canada

Search for more papers by this author
Erik M. Dettenmaier

Erik M. Dettenmaier

Restoration Installation Support Team, Hill Air Force Base, Utah, USA

Search for more papers by this author
Rosemary T. Zaleski

Rosemary T. Zaleski

ExxonMobil Biomedical Sciences, Occupational and Public Health, Annandale, New Jersey, USA

Search for more papers by this author
Peter Fantke

Peter Fantke

Quantitative Sustainability Assessment Division, Department of Management Engineering, Technical University of Denmark, Lyngby, Denmark

Search for more papers by this author
Jon A. Arnot

Jon A. Arnot

ARC Arnot Research and Consulting, Toronto, Ontario, Canada

Department of Physical and Environmental Sciences, University of Toronto at Scarborough, Toronto, Ontario, Canada

Department of Pharmacology and Toxicology, University of Toronto, Toronto, Ontario, Canada

Search for more papers by this author
First published: 04 October 2017
Citations: 61


Quantifying the transfer of organic chemicals from the environment into terrestrial plants is essential for assessing human and ecological risks, using plants as environmental contamination biomonitors, and predicting phytoremediation effectiveness. Experimental data describing chemical uptake by plants are often expressed as ratios of chemical concentrations in the plant compartments of interest (e.g., leaves, shoots, roots, xylem sap) to those in the exposure medium (e.g., soil, soil porewater, hydroponic solution, air). These ratios are generally referred to as “bioconcentration factors” but have also been named for the specific plant compartment sampled, such as “root concentration factors,” “leaf concentration factors,” or “transpiration stream (xylem sap) concentrations factors.” We reviewed over 350 articles to develop a database with 7049 entries of measured bioaccumulation data for 310 organic chemicals and 112 terrestrial plant species. Various experimental approaches have been used; therefore, interstudy comparisons and data-quality evaluations are difficult. Key exposure and plant growth conditions were often missing, and units were often unclear or not reported. The lack of comparable high-confidence data also limits model evaluation and development. Standard test protocols or, at a minimum, standard reporting guidelines for the measurement of plant uptake data are recommended to generate comparable, high-quality data that will improve mechanistic understanding of organic chemical uptake by plants. Environ Toxicol Chem 2018;37:21–33. © 2017 SETAC


Plants are the largest component of the earth's biomass and the base of all food webs. Organic chemical contaminants can directly contact and accumulate in aboveground plant tissues (shoots) through vapor and particle deposition, or in belowground tissues via the roots. The movement and distribution of the contaminants within the plant are determined by the properties of contaminants and plants. Plants and associated microorganisms (e.g., endophytes, rhizosphere organisms) can also transform organic chemicals, impacting their environmental fate and transfer to higher trophic levels. Quantifying and predicting the transfer of chemicals from the physical environment into terrestrial plants are important for assessing human and ecological risks, evaluating the use of plants as biomonitors of environmental contamination, and predicting the effectiveness of phytoremediation.

Contact of organic contaminants with aboveground vegetation (shoot) occurs through gas exchange and deposition at leaf surfaces, with the dominant pathway and kinetics dependent on the properties of the contaminant and leaf cuticle and the environmental conditions 1-3. The cuticle consists of several lipid or lipid-like components including cutin, cutan, and extractable waxes that exhibit varying affinities for organic contaminants 1, 4, 5. It is also possible, but less likely to be significant for neutral hydrophobic chemicals (1 < log10 octanol–water partition coefficient [KOW] < 8), that organic contaminants can sorb to the nonlipid organic matter fractions of leaves. Stomatal uptake might also be important for some chemicals 1. Contaminants accumulating in shoot tissues during periods of high atmospheric concentration can also be released when exposure concentrations decrease 3, 6, 7.

Root uptake of most organic contaminants is passive. In passive transport, the more water transpired, the greater the amount of organic contaminants that moves into the plant. Once a chemical passes through the root membrane, it can be transported to other parts of the plant, depending on its properties. For neutral compounds, hydrophobicity is one of the key transport factors 8-10, whereas for ionizable organics, the movement and distribution also depend on the dissociation constant (pKa), charge of the chemical, and pH of the various plant compartments 11. Xylem channels conduct the flow of water, nutrients, and contaminants from roots to the photosynthetic sections of the plant, whereas phloem distributes sugars and other photosynthetic products throughout the plant. Xylem transport rates are directly related to transpiration rates, whereas phloem transport rates are governed by differences in solute concentrations between sites of synthesis and consumption 12. Daytime xylem transport rates are generally 10 times greater than phloem transport rates 13, but their relative contributions to specific compartments can vary. For example, although xylem and phloem contribute approximately equally to apple fruit growth early in the growing season, from the middle to late in the growing season, the phloem dominates 13, 14. Within the xylem, lateral movement to adjacent cells may provide a pathway for contaminants to move into the phloem 12, 15.

During transport within the plant, organic chemicals can be metabolized, sequestered within various plant tissues, and volatilized from the plant surfaces. Plants contain enzymatic systems such as cytochrome P-450s, and evidence regarding the biotransformation capabilities of plants comes from cell culture and intact plant studies 16. Plants can degrade a wide range of organic contaminants, from highly polar herbicides like glyphosate to very hydrophobic chemicals such as dichlorodiphenyltrichloroethane (DDT) and hexachlorobenzene 17-19. Plant biotransformation capacity may also be chemical- and plant species-specific. For example, Liu et al. 20 showed that polychlorinated biphenyl (PCB) 77 was hydroxylated by poplar plants but not by switchgrass during the experimental period. Endophytic organisms can also contribute to the metabolism of organic chemicals within plants 21. Plant biotransformation rates are available for a few pesticides 17, 22, 23, pharmaceutical chemicals 24, and plasticizers 25, 26. Furthermore, plant dissipation rates (degradation and transport loss mechanisms) can vary considerably between different plant compartments and environmental conditions 27, 28. For chemicals with a high Henry's law constant (KH, Pa.m3/mol or KAW, dimensionless) that have been translocated to shoots, volatilization from leaves and stems, often referred to as “phytovolatilization,” can be a measurable loss mechanism 29, 30.

Differences between plant species can also be important in the root uptake of organic chemicals. For example, partitioning into the roots is a function of root lipid content in addition to the hydrophobicity of the chemical 31. Another example of species differences is the enhanced ability of “gold rush” zucchini (Cucurbita pepo, subspecies pepo) to transfer hydrophobic organics from roots to aboveground tissues, relative to other plants 32-35. This enhanced root to shoot transfer is thought to be partly attributable to interactions with various proteins that increase the solubility of these compounds in the xylem sap 36, 37.

Empirical data describing the extent of chemical uptake by plants can be expressed as ratios of chemical concentrations in the plant compartment of interest (e.g., leaves, shoots, roots, xylem sap) to those in the exposure medium (air, soil, soil porewater, hydroponic solution) measured at the time the samples are collected 38. These ratios are generally referred to as bioconcentration factors (BCFs) that implicitly assume that steady state is approximated. The most commonly reported plant BCFs are for roots, stems or wood, leaves, shoots (aboveground tissues), and xylem sap. The chemical concentration ratio between xylem sap and the external exposure solution (usually a hydroponic solution) is specifically referred to as a transpiration stream concentration factor (TSCF). When plant biomass increases faster than contaminant uptake, growth dilution can influence BCF values. The importance of growth dilution in predicting foliar and root uptake of organic contaminants has been well documented 28, 39-41.

Reliable, high-quality plant bioaccumulation data are critical in developing and evaluating plant bioaccumulation models, assessing human and ecological risks, using plants as environmental contamination biomonitors, and predicting phytoremediation effectiveness. Earlier reviews, focusing specifically on modeling plant bioaccumulation 42-45 and more generally on terrestrial bioaccumulation 46, 47, have highlighted substantial experimental data gaps and limitations.

The objectives of the present study were to 1) summarize the common metrics for quantifying organic chemical bioaccumulation in terrestrial plants; 2) review and compile measured plant bioaccumulation data, including uptake and biotransformation rates published in the peer-reviewed literature; 3) assess the general consistency of information reported for plant bioaccumulation studies; 4) evaluate the applicability of some previously reported relationships between plant bioaccumulation metrics and physical–chemical properties (e.g., KOW and octanol–air [KOA] partition coefficients) to the newly compiled data set; 5) illustrate possible data contradictions; 6) use simple theoretical partitioning models to provide a frame of reference for examining and comparing the different types of measured bioaccumulation data; and 7) provide preliminary direction for future plant bioaccumulation studies.


Plant bioaccumulation concepts and assessment metrics

Figure 1 provides a conceptual overview of the key processes associated with organic chemical bioaccumulation in a plant and the typical metrics or test endpoints used to describe these processes. Bioaccumulation metrics are reported using a variety of units depending on how the chemical concentrations are expressed. Chemical concentrations in the plant tissues are commonly specified on either a fresh or a dry weight basis. Occasionally, measured lipid contents of the plant tissues (mainly root and leaves) are reported (i.e., kilograms lipid wt per kilogram dry wt), and the chemical concentrations are expressed on a lipid weight basis. Contaminant concentrations in the environment surrounding the plant (exposure media) can also be reported in various ways. Chemical concentrations in air can be reported as total, including gaseous and particulate phases, or separately. Concentrations of chemicals in soils are typically reported on a dry weight basis; however, if the organic carbon content of the soil is measured (i.e., kilograms organic carbon per kilogram dry wt), the results can also be expressed on an organic carbon-normalized basis. When the concentrations for neutral organic chemicals are lipid- and organic carbon-normalized, chemical equilibrium partitioning approaches have been proposed to better understand chemical fate and plant bioaccumulation 48.

Details are in the caption following the image
A conceptual overview of key plant bioaccumulation processes and assessment metrics; a further summary of calculated endpoints is provided in Table 1. FCF = fruit concentration factor; LCF = leaf concentration factor; SCF = stem concentration factor; TSCF = transpiration stream concentration factor; RCF = root concentration factor; KH = Henry's law constant (or KAW, dimensionless); KOC = organic carbon–normalized sorption coefficient.

Table 1 lists definitions and summary calculations for the 3 general plant bioaccumulation metrics (endpoints) included in the present review. Supplemental Data, Table S1, provides a more detailed summary for all of the different specific metrics (n = 21) included in the database. The first general plant bioaccumulation metric is defined as chemical concentration ratios between belowground vegetation (i.e., roots, tubers) and appropriate environmental compartments (i.e., soil or other solid-phase medium, soil porewater solution, or hydroponic solution). Tuber concentration factors and root concentration factors (RCFs) are specific examples of belowground vegetation concentration factors. These factors can be further subclassified for cores and peels or total belowground tissues and whether or not the source phase is solid (e.g., soil) or liquid (e.g., hydroponic). The second general metric characterizes chemical concentration ratios between aboveground vegetation (i.e., foliage, shoot) or a specific aboveground plant compartment (i.e., stems, leaf, fruit) and appropriate environmental compartments (i.e., air, water, soil, or soil solution). Leaf concentration factors (LCFs), shoot concentration factors, stem concentration factors, seed concentration factors, and fruit concentration factors are specific examples of aboveground vegetation BCFs in the database. The third general bioaccumulation metric is the TSCF, defined as the ratio of chemical concentration in the xylem sap to the chemical concentration in water (hydroponic or soil solution) taken up by the plant. Ideally, bioaccumulation endpoints should be determined under steady-state conditions (i.e., concentration ratios between plant compartments and exposure media are constant over time) or kinetically using uptake and elimination rates for consistency and comparability.

Table 1. General plant bioaccumulation factors
General bioaccumulation endpoint Summary calculations
Belowground vegetation concentration factor urn:x-wiley:14381656:media:etc3992:etc3992-math-0001
Aboveground vegetation concentration factor urn:x-wiley:14381656:media:etc3992:etc3992-math-0002
Transpiration stream concentration factor urn:x-wiley:14381656:media:etc3992:etc3992-math-0003
  •  aConcentration in soils or tissues expressed on a wet (fresh) or dry basis.
  •  bExposure solution = hydroponic solution (measured) or soil solution (calculated or estimated).
  •  cTotal or dissolved.

Plant bioaccumulation database development

The peer-reviewed literature was searched using the list of plant bioaccumulation metrics outlined in the Introduction (BCFs, RCFs, TSCFs, etc.) and biotransformation rates or corresponding half-lives in whole plants or individual plant compartments as keywords. Only plant bioaccumulation metrics based on measured data were included in the database. Reported bioaccumulation metrics have various units depending on the exposure medium. If the units were documented in the study, this was noted in the database. However, in some cases, the units were not explicitly reported but could be inferred from the data provided. This was also noted in the database. In other cases, however, the units could not be readily inferred. Data from these studies were included, but notes were added to show that the units were not available. When available, other key parameters reported from the experiments were also included in the database, such as exposure concentration, (air) temperature, lipid and water content of sampled plant material, sampled plant mass, duration of light exposure, soil organic carbon content, and soil water content. The primary reference was also stated for each bioaccumulation value entered into the database.

Physical–chemical properties

The following physical–chemical properties were obtained for all chemicals in the database: molar mass (M, grams per mole); KOW, KOA, and KAW (dimensionless); and organic carbon-normalized sorption coefficients (KOC; liters of water divided by kilograms of organic carbon). For consistency, most physical–chemical property data were obtained from the US Environmental Protection Agency's (USEPA's) EPI Suite™ 49 program. Measured chemical property data were preferentially selected over predicted values. The database also includes SMILES notations 50 and Chemical Abstracts Service registration numbers. Chemicals that contain recognized ionizable functional groups were identified and flagged in the database as ionizable organic chemicals. Future database additions should include dissociation constants for ionizable organic chemicals and more explicit evaluation of the data with respect to this chemical property.

Relationships between plant bioaccumulation metrics and organic chemical properties

Many correlations between physical–chemical properties and various plant BCFs have been reported and used in various models to predict plant bioaccumulation 8, 51, 52. For example, RCFs 8, 53, stem or wood concentration factors 30, 54-57; and aboveground or shoot concentration factors 33, 43, 58 have been related to log KOW. Similarly, based on the assumption that the lipophilic cuticle is the major plant component governing air–plant interactions, simple regression models have been developed that relate air–shoot BCFs to KOA 59-62 or to a combination of KAW (dimensionless) and KOW 63.

Two different types of general relationships between log KOW and TSCF have also been reported. Bell-shaped curves relating TSCF to log KOW 8, 64, 65 suggest that compounds that are either highly water-soluble or highly hydrophobic will not be significantly taken up by plants. More recently, Dettenmaier et al. 9 presented an empirical relationship between TSCF and log KOW that indicates that non-ionizable, polar, highly water-soluble organic compounds are most likely to be taken up by plant roots and translocated to shoot tissue. Highly water-soluble compounds were also predicted to be the most likely type of organic chemicals transferred from soil to aboveground plant compartments 45, 66.

Despite being developed from relatively small data sets over 25 yr ago, both the Travis and Arms BCF–KOW regression 58 and the Briggs et al. TSCF–KOW 8 relationship are still used by regulatory agencies in North America. The USEPA lists the Travis and Arms relationship in its 2005 “Human Health Risk Assessment Protocol for Hazardous Waste Combustion Facilities” 67, and the Briggs relationship is used in the 2014 “Guidance for Assessing Pesticide Risks to Bees” 68 developed by the USEPA, the Health Canada Pest Management Regulatory Agency, and the State of California Department of Pesticide Regulation.

Plant bioaccumulation database evaluation

A general requirement for assessing data quality is that the reported test procedures and designs comply with, or are comparable to, published testing guidelines or are conducted with accepted methods that are described in sufficient detail (e.g., Registration, Evaluation, Authorisation and Restriction of Chemicals [REACH]). Klimisch et al. 69 and others 70, 71 have developed data quality–assessment methods for ecotoxicology and fish bioaccumulation testing data, respectively, based on nationally or internationally recognized testing guidance. Although plant-testing guidance exists for toxicity (e.g., USEPA 72), few standard methods (e.g., USEPA 73) exist for plant bioaccumulation testing, making data-quality assessment difficult 74.

Using the guiding principles from “Klimisch-type” data quality–assessment methods 69, we developed some preliminary criteria (Table 2) and tentatively assigned each database entry into one of 3 screening-level data confidence categories: high, medium, or low. Because of the wide variety of experimental methods used to develop the data and the inconsistent reporting of supporting information, data-quality assessment was challenging. All of the criteria listed for the particular category had to be met for the entry to be assigned to that confidence category. Unless there was a clear indication from the publication that the criteria were not met, we assumed that they were. Thus, data screened as “high confidence” may still have significant error. In addition, because this is a relatively simple screening exercise, data screened as “medium” or “low” confidence may still be reliable, particularly in specific contexts. For example, bioaccumulation metrics reported for homologs are considered to be of low confidence because chemical-specific identification and metrics are lacking. However, if the reported homologs (or mixtures) have similar partitioning and degradation properties, the data may still be valuable for certain contexts such as model development and testing. We encourage interested users of the database to consider our proposed criteria for assigning screening-level data confidence categorizations (Table 2) but to critically examine data entries of interest according to their context-specific objectives.

Table 2. Screening-level criteria for assigning tentative data confidence categories (high, medium, and low) to the database entriesa
Category Criteria
High Chemical (Chemical Abstracts Service registration no. or name) and plant species (botanical and common name) identified
Units for reported bioaccumulation metric(s) are explicitly defined
Chemical-specific analysis for exposure medium and plant
Chemical concentration in exposure media measured at the beginning, during, and at the end of the exposure period; or rate constants (or associated half-lives) are used to calculate kinetic bioaccumulation metrics from measured concentrations
No apparent toxicity to the exposed plant
Reasonable growth conditions typical for the selected plant species (i.e., hours of light and provision of nutrients)
A negative control included
Measures of key parameters influencing chemical partitioning between the exposure medium and the plant or plant compartment such as organic carbon content in the soil, dissolved organic carbon in hydroponic systems, pH for ionogenic organics, dissolved concentrations
Composition of the plant compartments (i.e., lipid and water contents), plant mass, growth rate, and amount of water transpired
Medium Chemical (Chemical Abstracts Service registration no. or name) and plant species (botanical or common name) identified
Bioaccumulation metric units can be readily inferred from reported concentrations
Bioaccumulation values are estimated from a graph or figure
Concentrations in exposure media are measured at some point in time
Concentration in plant measured at end of exposure period
Chemical-specific analysis
No apparent toxicity to the exposed plant
Low Chemical (Chemical Abstracts Service registration no. or name) and plant species (botanical or common name) identified
Bioaccumulation metric units are not clearly defined or readily interpretable
Unclear nature of test substance or use of homologs or mixtures
Only assumed nominal concentrations in exposure medium and plant reported
Aqueous exposure concentrations exceeding solubility in hydroponic test systems
Documented toxicity to plant
  • a A database entry screened as “high confidence” satisfies (or is assumed to satisfy) all of the criteria outlined for high confidence. A database entry that does not meet all of the high-confidence criteria but meets all of the medium-confidence criteria is considered to be of “medium confidence.” Data that satisfy at least one of the criteria outlined for “low confidence” are considered to be of low confidence.


Database summary

More than 350 articles (published ca. 1955–2014) were initially collected and reviewed. Suitable values and parameters reported or inferred from the reviewed studies were added into a database. The final database includes 7049 unique entries for plant bioaccumulation metrics or rate constants for 310 organic chemicals measured in 112 plant species from 156 peer-reviewed publications. In some cases, the articles did not report specific bioaccumulation metrics but provided sufficient measured data necessary to calculate them. In other articles, bioaccumulation metrics calculated from the measured data were displayed in figures but specific values were not reported. The bioaccumulation metrics derived from figures are noted in the database. When values of the same experimental study were reported by different literature sources, only the original source data were included in the database. The database and summary information compiled from the experiments and the selected physical–chemical properties for each entry are included in the Supplemental Data in the form of a Microsoft Excel® spreadsheet. The spreadsheet database is also available at www.arnotresearch.com.

The database includes 2728 belowground concentration factors (BGCFs) for 237 chemicals, 3947 aboveground concentration factors (AGCFs) for 208 chemicals, and 309 TSCFs for 104 chemicals. The database includes plant bioaccumulation metrics for polycyclic aromatic hydrocarbons (PAHs), pesticides, pharmaceutical and veterinary chemicals, polybrominated diphenyl ethers (PBDEs), PCBs, per- and polyfluoroalkyl substances (PFASs), polychlorinated dibenzo-p-dioxins, and polychlorinated dibenzofurans (PCDFs). Pesticides and PCBs represent the majority of the compounds, and the majority of data entries are for PAHs and pesticides. Pesticides, PAHs, pharmaceutical and veterinary compounds, and PFASs make up approximately 26, 27, 8, and 4% of the total data entries, respectively. The number of data entries for PBDEs, PCBs, and polychlorinated dibenzo-p-dioxins/PCDFs are approximately 4, 15, and 10%, respectively. Chemicals not associated with the aforementioned chemical classes represent the remainder of the total entries. Supplemental Data, Table S2, provides a summary of values for some of the physical–chemical properties for the chemicals in the database. Approximately 30% of the 310 chemicals were identified as ionizable organic chemicals.

Several test parameters could be considered essential to fully interpret the data and evaluate the quality of the measured values, (e.g., Supplemental Data, Table S3); however, these parameters were not always reported. For example, some studies did not clearly state whether the data were reported on a fresh weight or a dry weight basis. Some did not mention what specific tissue types (roots, shoots, stems, leaves, fruits, or tubers) were sampled, and several did not report the soil concentrations or key soil properties. Clear reporting of the units for the bioaccumulation metrics was missing in approximately 20% of the data entries. Although it may be possible to infer or assume the units for the concentration ratios, this may lead to significant uncertainties in data interpretation. For example, RCFs are expressed as both root to soil (dry wt basis) and root to soil solution or hydroponic solution (e.g., liters per kilogram or milliliters per gram) ratios. When sufficient details were reported to differentiate these types of RCFs, we reported them as RCFsolid and RCFaq, respectively.

Attainment of steady-state conditions, dependent on chemical-, plant-, and exposure-specific variables, is generally not verified experimentally and/or reported for plant uptake studies. This can limit data interpretation and interstudy comparisons. Reported exposure durations were included in the database but were not used to assign screening-level data confidence scores. Finally, none of the reviewed articles reported plant growth rates, and measurements for chemical concentrations in the exposure medium at the end of the exposure period were uncommon. Approximately 16% of the database entries were assessed to be of high confidence, 12% of low confidence, and the vast majority (∼ 72%) of medium confidence.

In the following sections, the relationships between several plant bioaccumulation metrics and physical–chemical properties such as log KOW and log KOA are examined using the new database. The database can test refined model hypotheses that make more explicit assumptions for chemical partitioning (e.g., distribution ratios for ionizable organic chemicals). We selected KOW and KOA over other potential descriptors such as KOC, calculated distribution coefficients, membrane water coefficients, and other coefficients because they are widely available and are the most frequently used chemical descriptors in organic chemical plant uptake models. We then qualitatively compare these data to previously reported empirical relationships and simple theoretical partitioning models. If the water content of the plant tissue was not reported, we assumed a factor of 0.2 (dry wt divided by wet wt) for converting dry weight plant values to wet weight (or vice versa) for plotting bioaccumulation metrics in the figures that follow. This assumed conversion factor is the median for database entries based on studies that reported the water content in plant tissues (reported water content values range from 70–95%).


There are 2728 BGCFs in the database, with various units, spanning over 7 orders of magnitude for chemicals with log KOWs ranging from –3.1 to 9.1 and molecular weight ranging from 78.1 to 959.2 g/mol. The BGCFs include different environmental exposure media (solid phase or aqueous) and different plant material analyzed (i.e., roots, tubers, peels, cores).

The most widely cited relationships with log KOW, for example, Briggs et al. 8, have expressed RCF values with units of liters of water per kilogram wet root weight (RCFaqs) based on exposures via hydroponic solution or soil porewater. Figure 2 summarizes 723 RCFaqs in the database spanning approximately 7 orders of magnitude reported from 40 articles representing a range of experimental conditions and methods. For neutral chemicals with log KOW >1, RCFaq increases with KOW (excluding the low-confidence data; slope = 0.73, r2 = 0.65). For the ionizable organic chemicals, there is still a positive but weaker correlation for increasing RCFaq as a function of KOW (log KOW > ∼ 1, slope = 0.24, r2 = 0.26). The extent of ionization is generally uncertain because the pH of the test systems was rarely reported.

Details are in the caption following the image
Measured root concentration factors derived from porewater or solution exposure media (RCFaq) as a function of chemical octanol–water partition coefficient (KOW). No tubers were included in this analysis. The figure on the left is for neutral compounds, and that on the right is for ionizable organic chemicals. “Low,” “medium,” and “high” refer to the data confidence screening assessment for the measured data. For ionizable organic chemicals the KOW for the neutral form of the chemical is used. The Briggs et al. [8] RCF model is presented along with a simple equilibrium partitioning (EqP) model (RCFaq = 0.8 + 0.015KOW). A kinetic adjustment factor is also included in a version of the EqP model 75. Equilibrium partitioning models with and without a kinetic adjustment factor are presented for reference. EqP_Kin = EqP model with a kinetic adjustment factor. IOC = ionizable organic chemical.

Briggs et al. 8 proposed that 2 components determine the total amount of chemical in the roots: the chemical associated with the water in the roots and the chemical associated with the lipophilic root solids, leading to the correlation illustrated in Figure 2. For comparison, the results of 2 partitioning models 75 also used to calculate RCFaqs are shown in Figure 2. The first equilibrium partitioning (EqP) RCFaq model assumes the root is 80% water and the remaining organic solids have an assumed “octanol equivalence” of 1.5% for partitioning to the roots. The second model (EqP_Kin) also includes a kinetic adjustment factor to address conditions in which the root does not reach equilibrium with the surrounding water. This can occur when growth rates exceed uptake rates (e.g., “growth dilution” effect) or the exposure duration is too short to approximate equilibrium 75. The Briggs regression model and the partition-based models show similar RCFaq predictions for chemicals in a log KOW range of ∼ –2 to 3, but model divergence is greater with higher log KOW chemicals. There is a large scatter over the entire range of hydrophobicity, and separating the neutrals (left) from the ionizable organic chemicals (right) improves the fit of all 3 models for the neutral chemicals. For some of the ionizable organic chemicals the RCFaqs are higher than EqP at lower KOW, perhaps attributable to electrostatic interactions. The KOW-based models were not specifically developed to include ionizable organic chemicals.

Database values of RCFaqs for chemicals having similar log KOW can span several orders of magnitude. For example, the 9 poplar RCFaqs for trichloroethylene (log KOW = 2.4) that were scored as low-confidence values range from 5.1 to 570, whereas a single medium-confidence RCF value was 2.5. The 9 lower-confidence RCFs were determined using a non-chemical-specific 14C analysis. A single, high-confidence RCFaq for phenanthrene (log KOW = 4.5) is 6.5. Thirty-five medium-confidence RCFaqss for phenanthrene measured in various plants from various experimental designs range from 5.0 to 1730. Variability is also notable for ionizable organic chemicals where uncertainties in exposure water pH may be particularly significant. Only 154 of the 481 RCFaq entries for ionizable organic chemicals reported pH. For example, for triclosan (log KOW = 4.8 [neutral form], pKa = 7.8), 9 RCFaqs considered medium confidence range from 0.08 to 100 and 4 RCFaqs considered low confidence range from 1.4 to 90. For diclofenac (log KOW = 4.5 [neutral form], pKa = 4.2), 7 RCFaqs considered medium confidence range from 0.08 to 0.84 using chemical-specific analysis, whereas 2 RCFaqs were derived using total radioactivity and hence considered low confidence range from 28 to 105.

Figure 3 illustrates the relationship between log KOW and 1199 RCFsolid values in the database expressed in units of kilograms dry weight soil per kilogram dry root weight. The RCFs are calculated and presented using dry weight soil concentrations rather than organic carbon-normalized soil concentrations because organic carbon content in soil was not regularly reported (0.02 kg organic carbon/kg dry wt is the median value in the database, when reported in the studies). The RCFsolid values span approximately 5 orders of magnitude and show a large scatter over the entire range of hydrophobicity. The illustrative EqP model calculations shown in the figure suggest that the large majority of RCFsolid values have not reached equilibrium and the large variance in RCFsolid is not driven by variability in soil organic carbon content.

Details are in the caption following the image
Measured root concentration factors derived from solid exposure media (RCFsolid) as a function of chemical octanol–water partition coefficient (KOW). The figure on the left is for neutral compounds, and that on the right is for ionizable organic chemicals. “Low,” “medium,” and “high” refer to the data confidence screening assessment for the measured data. For ionizable organic chemicals the KOW for the neutral form of the chemical is used. Equilibrium partitioning (EqP) models with and without a kinetic adjustment factor are presented for reference. Variability in predicted RCFsolid for a range of soil organic carbon contents is demonstrated. EqP_Kin = EqP model with a kinetic adjustment factor; OC = organic carbon; IOC = ionizable organic chemical.

An additional 175 BGCFs in the database (not shown in Figure 3) classified as tuber concentration factors and expressed as total tuber biomass (dry wt) are displayed in Figure 4 (left). The tuber concentration factors span approximately 3.5 orders of magnitude, and there is no significant relationship with KOW. The tuber concentration factors are lower than would be expected based on estimated EqP. An additional 292 peel concentration factors and 220 core concentrations factors for belowground edible vegetable concentration factors (a mixture of tubers and roots: carrots, radishes, and potatoes) for neutral organic chemicals for data entries that were not included in any of the previous figures are displayed in Figure 4 (right). Most of the “total,” peel, and core data are from the same study 76. On average, partitioning to the peel is approximately 7 times greater than partitioning to the core of the edible vegetable. The average peel concentrations approximate the average “total” concentrations because the peel concentrations are highest. No significant relationships exist between the belowground edible vegetable concentration factors and log KOW from approximately 2 to approximately 9. The edible vegetable concentrations are well below the EqP calculations, and the assumption for a kinetic correction that suggests a decrease in concentration at higher KOW does not capture the general trend observed in the measured data. In general, the EqP model with the assumed 1.5% lipid equivalence assumption for the vegetable and the 2% organic carbon content for the soil overpredicts the median of “total” measured edible vegetable concentrations (variable lipid contents and organic carbon contents) by a factor of approximately 60. However, most of these data (>95%) are from a single field-based study 76 involving long-term “aged” chemical contamination rather than the shorter-term exposures commonly found in laboratory-based studies or in a sewage sludge–amended soil situation.

Details are in the caption following the image
Measured tuber concentration factors (TCFs; left) for potatoes and belowground edible vegetable concentration factors (BGEVCFs; right) for carrots, potatoes, and radishes as a function of chemical octanol–water partition coefficient (KOW). “Low” and “medium” refer to the data confidence screening assessment for the measured data. The BGEVCFs are separated into data analyzed for peels, cores, and “total” belowground vegetables. Equilibrium partitioning (EqP) models with and without a kinetic adjustment factor are presented for reference. EqP_Kin = EqP model with a kinetic adjustment factor; OC = organic carbon; IOC = ionizable organic chemical.

The EqP model calculations in Figures 2, 3, and 4 provide a frame of reference for examining and comparing the different types of measured bioaccumulation data. Differences between EqP model calculations and experimental data indicate that laboratory hydroponic exposures often approximate equilibrium (medium- and high-confidence data in Figure 2, left) for belowground plant concentration ratios, but most field data for neutral organic chemicals with log KOW >2 do not (Figures 3 and 4). These differences may reflect reduced plant bioavailability from solid phases under typical field conditions (not sewage sludge–amended soil). Further examination of porewater and solid-phase concentrations (and chemical activities) and plant uptake from field studies seems warranted to better ascertain possible differences between laboratory (“dissolved exposure conditions”) and field BCFs. Passive sampling may provide insights into these issues of “bioavailability.” Higher field growth rates and differences in loss kinetics (e.g., metabolism, volatilization) between laboratory and field could also play a role. Supplemental Data, Figure S1, shows that there is no apparent decrease in root vegetable bioconcentration with increasing molecular weight. However, the database contained only one value for a chemical with molecular weight >500 g/mol, indicating that additional data are needed to better understand the uptake of larger molecules.


The 3947 AGCFs in the database have various units and exposure routes (air, soil, and water) for chemicals with log KOWs ranging from –3.1 to 8.9 and molecular weight ranging from 78.1 to 959.2 g/mol. The AGCFs include a variety of environmental exposure media (solid phase or aqueous) and plant material analyzed (e.g., stem, fruit, leaves, whole plant). Plant–air concentration factors were predominantly LCFs, and chemical concentrations in the air were expressed as gaseous (“dissolved”) or total (gaseous and particulate phases).

BCFs for total aboveground plant compartments (root to shoot pathways)

The 2956 entries for BCFs describe the root uptake of chemicals into aboveground vegetation reported from 95 different articles. The database includes a combination of field- and laboratory-determined BCFs. Of these BCFs, 981 are based on exposures from an aqueous phase (i.e., hydroponic or soil solution) and 1975 are based on exposures from a solid phase (i.e., soil). An additional 75 values in the database were derived from both soil and air exposures.

Figure 5 shows the relationship between log KOW and 1888 BCFsolids (i.e., plant/soil concentration ratio) for a variety of plant tissues (e.g., stems, fruit, seeds, leaves). The BCFsolids span approximately 8 orders of magnitude and show a substantial variability throughout the range of log KOW values. The widely used relationship of Travis and Arms 58, developed mainly from literature field data (29 neutral organics), is also displayed as a reference.

Details are in the caption following the image
Measured bioconcentration factors derived from solid-phase exposure media (BCFsolid) as a function of the chemical octanol–water partition coefficient (KOW). The figure on the left is for neutral compounds, and that on the right is for ionizable organic chemicals. “Low,” “medium,” and “high” refer to the data confidence screening assessment for the measured data. For ionizable organic chemicals the KOW for the neutral form of the chemical is used. The Travis and Arms [58] relationship for aboveground edible plant parts and soil as a function of KOW derived in 1982 is also displayed (BCF = 1.58 − 0.58 × log KOW, n = 29). IOC = ionizable organic chemical.

The Travis and Arms model underestimates, by approximately 4 orders of magnitude, some of the BCFs categorized as being of higher data confidence. Furthermore, Supplemental Data, Figure S2 (subset of data from Figure 5), emphasizes that the Travis and Arms model 58 underestimates concentrations in aboveground edible vegetation, highlighting the potential for significant error in applications of the model for risk assessments. Supplemental Data, Figure S3, indicates that there is no apparent decline in BCFsolid for high–molecular weight chemicals (i.e., molecular weight >500 g/mol).

Figure 6 shows 927 BCFaqs (i.e., plant/solution) for a range of plant tissues (e.g., shoots, stems, fruit, seeds) as related to log KOW. The BCFaqs span almost 8 orders of magnitude, with substantial variability at any particular KOW and no clear relationship between BCFaq and KOW over a wide range of chemical hydrophobicity. As observed for RCFs, BCFs for chemicals having the same or similar log KOW can span several orders of magnitude. For example, BCFaqs (liters per kilogram wet wt) in 4 species determined in the same study 77 and categorized in the medium-confidence range for N,N-diethyl-meta-toluamide (log KOW = 2.18) range from approximately 0.04 to 25 (n = 7). Another study 78 reports that BCFaqs for 2,3,7,8-tetrachlorodibenzo-p-dioxin (log KOW = 6.80) in 2 species considered to be medium confidence ranged from approximately 0.03 to 1.7 (n = 16). Large variability is also observed for ionizable organic chemicals where exposure water pH is a critical variable, yet it was reported in only 7 of the 675 BCFaq entries. For perfluorobutanesulfonic acid (estimated log KOW of 1.82 for the neutral form, estimated pKa ∼ –3) BCFaqs range from approximately 0.15 to 70 (n = 37). Similarly for perfluoro-n-undecanoic acid (estimated log KOW of ∼ 6.82 for the neutral form and a pKa ∼ 0.5), BCFaqs range from 0.0056 to 17 (n = 33). Both examples are from the same study 79 and were tentatively assigned medium-confidence scores. The differences in BCF values could be the result of species variability but could also be attributable to differences in the aboveground tissues of the plant selected in the calculation, plant growth rates, and the amount of water transpired.

Details are in the caption following the image
Measured bioconcentration factors derived from aqueous media (BCFaq) as a function of the chemical octanol–water partition coefficient (KOW). The figure on the left is for neutral compounds, and that on the right is for ionizable organic chemicals. “Low,” “medium,” and “high” refer to the data confidence screening assessment for the measured data. For ionizable organic chemicals the KOW for the neutral form of the chemical is used. IOC = ionizable organic chemical.

Plant–air concentration factors and relationships with KOA values

Figure 7 summarizes 916 plant–air concentrations factors, predominantly LCFs, included in the database as a function of KOA. The reported LCFs were determined from either total or dissolved (gaseous) air concentrations. The observed scatter is generally less than the other plant endpoints that have been previously discussed. This is likely attributable to the smaller chemical domain of compounds tested and/or the use of similar measurement methods, because most of the data were generated by a small number of investigators (data are from 10 publications). However, the variability is still notable, and the difference between gas– and total air–based concentration ratios becomes quite large at high KOA. As mentioned in the Introduction, it has also been suggested that some of the interspecies variability can be attributed to differences in the sorptive leaf phase (i.e., cutin, cutan, and waxes) 80, 81.

Details are in the caption following the image
A summary of 916 plant–air measurements included in the database as a function of the chemical octanol–air partition coefficient. Measurements are based on either bulk air or gas phase (no particles) estimates. “Low,” “medium,” and “high” refer to the data confidence screening assessment assignments for the measured data. An equilibrium partitioning (EqP) model is included that assumes that sampled plant material has 1.5% (kilograms per kilogram) octanol equivalence 75. A kinetic adjustment factor is also included in the EqP model (EqP_kin1) to account for competing rates of chemical uptake and growth. A second adjustment factor is included in the EqP_kin2 model that accounts for particle deposition to the plant surface to illustrate the potential role of these processes on measured data. KOA = octanol–air partition coefficient.

The solid line in Figure 7 represents equilibrium partitioning between the gas phase and the leaf, assuming the leaf has an octanol equivalence of 1.5% (kilograms per kilogram, assumed). The dashed line (EqP_kin1) includes a disequilibrium factor attributable to growth rates exceeding chemical uptake rates, and, in addition, the dotted line (EqP_kin2) considers particle deposition to the plant (leaf) surface for very high KOA chemicals 75. The simplified models shown in Figure 7 (details in the Supplemental Data) approximate mechanistic processes for chemical uptake in plants detailed by McLachlan 81. The database contains no plant–air concentration factors for ionizable organic chemicals.


Of the 299 TSCFs shown in Figure 8, 238 were generated using hydroponic exposure systems. Hydroponics is used more often than soil systems because the exposure concentrations are more easily measured and controlled. Assuming the passive uptake, TSCF values should have a theoretical maximum of 1 82. Thus, reported values of TSCF >1 were categorized as low confidence in this screening exercise. The empirical relationships previously developed by Briggs et al. 8 and Dettenmaier et al. 9 for neutrals or chemicals in their non-ionized form under the conditions of the experiment are also shown in Figure 8.

Details are in the caption following the image
Measured transpiration stream concentration factor (TSCF) values as a function of octanol–water partition coefficient (KOW) with the relationships developed by Briggs et al. 8 and Dettenmaier et al. 9. The figure on the left is for neutral compounds, and that on the right is for ionizable organic chemicals. “Low,” “medium,” and “high” refer to the data confidence screening assessment for the measured data. For ionizable organic chemicals the KOW for the neutral form of the chemical is used. Under the conditions of the experiment, ionizable organic chemicals may or may not be ionized. IOC = ionizable organic chemical.

Briggs et al. 8 determined TSCF values by indirectly calculating xylem sap concentrations from the total mass of the compound analyzed in the shoots divided by the volume of water transpired. The plants were grown hydroponically in a solution containing a known concentration of the chemical. Out of the 299 total reported values of TSCF, 100 were measured using this type of approach. Dettenmaier et al. 9 directly collected xylem sap from a detopped plant (i.e., a plant whose aboveground tissues have been removed) sealed in a pressurized chamber containing a solution of a known concentration of the chemical. Because the chamber is pressurized, xylem sap is collected as it exits the stem, and is then analyzed. The TSCF is calculated from the steady-state ratio of the measured xylem sap and root exposure concentrations. The database contains 54 values measured using the pressure chamber approach.

The TSCF values for chemicals having the same or similar log KOW varied widely. For example, TSCF values for trichloroethylene (log KOW = 2.4) range from 0.02 to 0.59 for 14 high-confidence data entries and 0.02 to 0.22 for 9 low-confidence data entries (radiolabeled, not corrected for parent chemical), with a single medium-confidence entry of 0.81. Variability is also notable for ionizable organic chemicals where uncertainties in exposure water pH may be particularly significant because only 96 of the 183 TSCF entries for ionizable organic chemicals reported exposure water pH.

Kinetic data

One hundred sixty-four of the database entries include some information on uptake rate, chemical half-life, or biotransformation rate of the organic chemicals in different plant species. Maddalena et al. 83 reported release rates and plant–air partition coefficients for phenanthrene, anthracene, fluoranthene, and pyrene. From studies such as these, the uptake rates can be estimated. Uptake rates from air for PCBs, pesticides (including hexachlorobenzene, hexachlorocyclohexanes, DDT-related compounds, alachlor, dieldrin, trifluralin, mirex, thionazin, sulfotep, and bromacil), phenols, and nitrobenzene were reported in a few studies. A few articles indicated that plants were able to metabolize organic chemicals; however, these studies did not report biotransformation rates. For the others, measured biotransformation rates ranged from approximately 0.0003 to approximately 0.02/h, corresponding to biotransformation half-lives of approximately 3000 to approximately 30 h. A review of 4513 measurement-based dissipation half-lives (possible biotransformation and transport loss processes) for 346 pesticides applied to 183 plant species has shown that, typically, only overall dissipation from plants is reported, whereas metabolism rates contributing to dissipation are reported in only <2% of all included studies 17. However, dissipation half-lives can often be used as a first proxy for biotransformation in plants under field conditions 84 because biotransformation is generally the dominating process contributing to overall dissipation 41.


Substantial variability exists in plant bioaccumulation data. For chemicals having more than one literature BCF, the values often range over several orders of magnitude. In our opinion, much of the variability in BCF values is associated with the wide variety of experimental approaches used to conduct the plant uptake studies. Often, the experimental objective of the study was to determine relative plant uptake and/or distribution within the plant, not specifically to determine a BCF or TSCF value. Thus, a BCF or TSCF was reported as the outcome of the study, but because it was not the main objective, key supporting data describing environmental conditions (e.g., light, temperature, and humidity) and plant growth- and health-related parameters were not reported or collected. Without this information, it is difficult to determine why chemical uptake data vary so much between studies.

The variation in testing methods and data reporting also prevented any meaningful global examination of the influence of variables related to environmental conditions (e.g., temperature, light, humidity, pH) and plant properties (e.g., growth rates, metabolism, age). Studies systematically quantifying the importance of these variables are needed.

Developing and applying data confidence assessment criteria to the values in the database were also challenging because of the lack of standard methods for plant uptake studies. In some cases, potential sources of uncertainty in the analytical methods can be identified, such as the use of a radiolabeled test chemical without parent chemical–specific quantification. In other cases, the experimental approach itself was the biggest potential source of data uncertainty. For example, the TSCF is defined as the xylem/exposure solution ratio. However, in most cases, the xylem sap concentration was determined indirectly from the total mass in the shoot/volume of water transpired. To assign a high-quality assessment to a TSCF value that was measured indirectly is probably not appropriate even though the study might have been carefully done. Similarly, BCFs can be calculated from ratios of tissue and exposure concentrations; however, without knowing the volume of water transpired and plant growth rate, the relevance of the data is still uncertain, even if the overall quality of the study as determined from other criteria and considerations was high.

For these reasons it is suggested that plant bioaccumulation and in vivo biotransformation rate estimation testing and data reporting guidance be developed and recommended for future experimental testing. Without providing such a perspective and technical guidance to the scientific community, it is unlikely that data and models used in exposure assessments will improve to any significant extent in the foreseeable future.

Despite the variability in experimental data regarding the uptake of organic chemicals by plants, some generalizations can be made. Current empirical and equilibrium models using KOW generally overestimate bioaccumulation in belowground edible vegetables (based on soil exposures) and underestimate bioaccumulation in aboveground edible vegetables (based on soil exposures). This is not surprising because it is unlikely that equilibrium or even steady-state conditions are attained during the relatively short-term exposure experiments typically conducted. Mechanistic models incorporating multiple exposure pathways (i.e., air to aboveground plant tissues) and kinetic factors associated with plant growth and metabolism are more likely to explain the available data if study-specific information is available to parameterize the models to the test conditions. Despite most existing plant uptake models predicting a decrease in contaminant uptake by plants as a function of size and hydrophobicity, this was not observed in the available data.

In summary, we have developed a database of values quantifying the uptake of organic chemicals into terrestrial plants, and developed and applied data-quality criteria to preliminarily evaluate confidence in the database entries. We emphasize that the current screening-level criteria cannot explicitly identify all possible data-quality issues (good or bad). Database users are encouraged to consider how these guiding principles were developed and review the original source before using the data for their particular application. A detailed strategy for developing standardized test methods for plant bioaccumulation testing has recently been proposed 74. Some of the more commonly applied regression models and basic theoretical partitioning models were used to compare with database values; however, these analyses are not intended to be comprehensive model evaluations. We hope the database will provide opportunities for more comprehensive model evaluations. By including a broader spectrum of models and a more thorough comparison, chemicals and plants that are not well predicted or are lacking measurements could be identified and used to guide future research efforts. Differences in biotransformation between species and chemical concentration differences within plant tissues may contribute to the overall variability of the reviewed plant bioaccumulation data. However, the potential contributions of these factors are impossible to reconcile because of uncertainty in the data and variability in test methods.

Supplemental Data

The Supplemental Data are available on the Wiley Online Library at DOI: 10.1002/etc.3992.


We thank ExxonMobil Biomedical Sciences for funding support to C. Shunthirasingham and J.A. Arnot. E.M. Dettenmaier's database development was supported by the Utah Water Research Laboratory at Utah State University and the Graduate Assistantships in Areas of National Need (GAANN) program.

    Data availability

    Data, associated metadata, and calculation tools are available from the corresponding author ([email protected]).