Plant uptake of organic chemicals: Current developments and recommendations for future research


Published on the Web 8/06/2007.

It is widely recognized that plants may become contaminated with a range of toxic organic chemicals, including poly-chlorinated biphenyls (PCBs), polycyclic aromatic hydrocarbons (PAHs), explosives, and dioxins [1]. These contaminants subsequently may enter food chains, resulting in the exposure of humans and ecosystems to hazard. The uptake, storage, and release of chemicals by vegetation are also critical components of the global cycling of persistent organic pollutants [2]. A knowledge of the uptake and degradation of organic chemicals by plants is also an essential element of any phytoremediation application, and a prior understanding of these processes is implicit in decision-support tools such as the U.S. Environmental Protection Agency (U.S. EPA) online Phyto Decision Tree (

To address these scenarios, a number of plant-based pollutant uptake submodels have been developed within larger model frameworks. Examples include the contaminated land exposure assessment model to assess the risks posed by the development of contaminated sites [3] and the European Union System for the Evaluation of Substances for the registration of chemicals in the European Union [4]. The U.S. EPA currently is using simple models to assess risks from contaminated soils to flora and fauna by developing ecological soil screening levels [5]. Despite the importance of these processes, only a limited number of studies have reviewed comprehensively the range of plant uptake and storage processes. One such work is Trapp and MacFarlane [6] in 1995, but an updated assessment is now timely. It was on this basis that a special session entitled “Plant uptake of organic pollutants—processes and modeling” was organized as part of the SETAC North America Meeting in November 2005.

One component of this special session was a semiquantitative analysis of the key processes in plant-uptake modeling using fate, events, and processes analysis. Internal transfers within the plant, xenobiotic metabolism, and deposition via the soil-air-plant pathway were identified as important processes that we currently cannot quantify with sufficient proficiency to justify adding them to a predictive model. The fate, events, and processes analysis also indicated that fruits were not in equilibrium with the other components of the system, because they influenced the system far less than the system influenced them. The papers that follow discuss uptake processes [7,8], modeling [9,10], and data variability [10] on plant uptake of organic chemicals.

Many of the established models use the equations of Ryan et al. [11] for estimating the transfer of organic chemicals from soil into the plant root and subsequently to the shoot. Deviation from this model has been observed [12,13]. The work of Wild et al. [8], using a novel nondestructive technique to investigate the more hydrophobic PAHs, casts some doubt on the likelihood of any root-to-shoot transfer for this chemical class. Information is scarce on the transfer of organic pollutants via the shoot to edible fruits, which the fate, events, and processes analysis identified as not being in equilibrium with the rest of the system, and it has been suggested that crop-specific models may be required for risk assessments of different crops [14]. Further work therefore is required to understand the internal transfer processes, which was recognized by the fate, events, and processes analysis. However, until more information becomes available, it is possible that equilibrium models are most appropriate for roots and shoots.

An additional complication to predicting the rate of uptake of organic chemicals by plants is inter- and intraspecies variation. A number of reports document wide variation in translocation of persistent organic pollutants in the Curcurbitaceae [15], and this phylogenetic variability is explored further in a series of continuous-flow hydroponic experiments reported by Gent et al. [7]. The modeling studies of the proceeding shows a 10-fold difference between the partition to xylem water in zucchini compared to cucumber, indicating that species differences can have a significant influence on the uptake and transport of persistent organic pollutants [9]. We therefore must proceed with care when assuming crops within botanical families accumulate chemicals at equivalent rates.

When uptake processes are quantified and predictive models are constructed, it is essential that they are validated to test their ability to predict outcomes from a range of scenarios. One reason the simple regression model of Travis and Arms [16] performed well in a recent model intercomparison exercise was that it was calibrated using a range of data rather than a single set of calibration experiments [17]. McKone and Maddelena [10] found high experimental variability (CV = 170%) when investigating soil-to-plant bioconcentration factors for organic pollutants, but the variability between models used to predict the plant bioconcentration factors was even greater (CV = 1,400%). The wide divergence between model predictions brings their applicability into question, and work is required to determine the source of the differences between individual models. It is probable that the variability arises because the models are mechanistic and have not been calibrated and validated sufficiently; these two procedures require a range of high-quality data that are not currently available for all of the processes many models calculate.

The work of Gent et al. [7] and Wild et al. [8] demonstrates that we are beginning to address some of the gaps in our understanding of the processes controlling the internal transfer of organic pollutants in vegetation, but we also must tackle the other processes where our knowledge is limited if we are to improve our ability to model the contamination of vegetation. At present we are some distance from developing a model that can predict the variable data reported for the transfer of organic pollutants to vegetation, principally because of the range of predictions between individual models. Consequently, our ability to predict the human and ecological health impacts from exposure to vegetation grown in contaminated soils is limited. This could impede the development of brownfield sites or result in negative health outcomes when such land is developed.