Our critical review of the range and reliability of methods for estimating plant-soil BCR offers insight for environmental scientists who must interpret and apply plant-uptake estimates obtained from models and experiments. In this assessment we have emphasized the importance of confronting uncertainties at each stage of the model development and application. We see that uncertainty emerges at the conceptual model stage as well as during mathematical model formulation and calibration and in model applications. The results above show that uncertainty is not simply a variance propagation or “Monte Carlo” assessment that is used to propagate parameter variance at the model application stage. Instead it is a process that begins at the earliest stage of model development and accrues through model formulation and specific applications. In the studies reviewed in this paper, we find that important uncertainties arise at the first stage of model development—the concept formulation. For plant uptake models that address competing soil-root-leaf and soil-air-leaf pathways, conceptual uncertainties remain a dominant source of overall uncertainty. An important contributor to this conceptual uncertainty is the lack of a consistent definition of BCR for soil uptake in both experiments and models. This leads to confusion and inconsistency in the use of BCR. Because this type of uncertainty is difficult, if not impossible, to quantify, we must develop qualitative methods and classifications to communicate this important source of uncertainty. In evaluating model formulation, we observe large differences among models in their predictions of BCR,livestock fodder system but we discovered no clear basis for selecting one model as more accurate than another. The residual errors reported for many of the models in fitting their calibration data leads us to believe that we do not yet have sufficient data to formulate accurate models. But our review of experiments for the single chemical RDX reveals that much of the uncertainty is attributable to the lack of precision and variability of experimental data used to obtain the BCR values used to calibrate models.

It appears that there are advantages to using more than one compartment in formulating BCR models. But lack of experimental data and the poor state of conceptual knowledge suggests that model uncertainty cannot be reduced by adding large numbers of compartments to the models. Too many plant components in our models lead to over specification. But a single compartment model can miss the combined effect of root and shoot uptake processes. The interaction among measurements, conceptual models, and models leads to the conclusion, contrary to our initial expectation, that it may not be possible to distinguish the relative contributions of overall uncertainty from conceptual uncertainty, measurement variability, and model uncertainty. For example, the variations in the value of BCR obtained from the 81 experiments we considered for RDX span a rather wide range. But it is not clear how much of this variation is attributable to experimental uncertainty and how much to inadequate conceptual models. Often the conceptual model is used to design experiments so that an incorrect conceptual model leads to measurements that are difficult to interpret when they are inconsistent with the concept. Perhaps the variance in the experimental values would be much lower if we understood better how BCR is affected by variables whose impact is not yet fully understood—for example temperature, soil properties, etc. Similarly if we really had a complete and thorough conceptual understanding of the process of uptake, then choosing a mathematical equation would likely be less uncertain. That is, the mathematical model formulation may only appear uncertain because we are using mostly-empirical mathematical relationships to describe a process that we do not understand well enough at a conceptual level. So it is not clear whether we classify this as uncertainty in the mathematical model formulation, or as uncertainty in the conceptual model. In applying model performance evaluation to plant uptake modeling, the results and discussion above lead us to a number of key findings. These include: The conceptual formulation of the bio-concentration ratio has an important, but at this point difficult to quantify, contribution to overall uncertainty. In particular, the concept of different plant components, the selection of dry- versus fresh-mass concentrations, and the use of dynamic or steady state concentration ratio strongly impact the reliability and uncertainty of the resulting BCR model.

When we consider both the performance of models with respect to their calibration experiments and also compare different models, we find that quantitative results for any randomly selected organic chemical have very large model uncertainties. We estimate that in the absence of specific experimental information, the expected uncertainty of a BCR model can be represented by a log normal distribution with a GSD of 10 . This means that without additional information on plant species or without plant- and site- specific measurements, we can only expect a model to predict a BCR within ±1 log units such that there is a 66% likelihood that the actual BCR value is 10 times higher or lower than the value obtained from a model. Based on consideration of a large number of experiments for a single, well-studied compound, RDX, we find that experimental measurements of BCR have large experimental variability and that this experimental variability can be represented by a log normal distribution with a GSD of 3.5 . This indicates much of our observed model uncertainty most likely derives from experimental variability. This leads to the observation that controlled measurements cannot necessarily remove the large uncertainties that derive from BCR models. Comparison for RDX of the relative contributions of model uncertainty and experimental variability to uncertainty in BCR estimates indicates that a large fraction of model uncertainty can be attributed to experimental variability. The variability and complexity of the uptake and transport of chemicals in vegetation cannot be captured by a point-value for BCF, but requires the use of ranges and confidence intervals to communicate the large uncertainties associated with estimating BCRs. In any plant-uptake model used to estimate a BCR, we must develop a process for communicating both the magnitude of the result and the confidence that can be placed in this number. On the part of the assessor this requires a presentation of both qualitative and quantitative uncertainties. Heavy metals such as iron , zinc , copper , and manganese are essential micro-nutrients for all organisms, acting as co-factors in a variety of biological processes. These heavy metals are extremely reactive and can become toxic at high concentrations; therefore, the intracellular concentration of these essential metals must be tightly regulated . Other heavy metals such as cadmium , lead, mercury, and the metalloid arsenic do not have biological functions in plants and are toxic even in trace amounts, disrupting several biochemical activities by displacing essential metals from their respective binding sites .

In humans, Cd exposure has been linked to cancer in the kidneys, lungs, and prostate, and severe Cd poisonings can result in neurological disorders and pulmonary and renal failure . While occupational exposure and tobacco products are associated with a high risk of Cd poisoning, consumption of contaminated plant-based foods represents the major source of Cd exposure in the general public . Many cases of widespread cadmium poisonings have been attributed to consumption of contaminated seeds in Thailand, China, Japan, and Australia . However, the molecular mechanisms and genes mediating the loading of both essential and nonessential heavy metals into seeds remain largely unknown. Metal accumulation and distribution in plants consist of several mechanisms, including: metal uptake into roots, xylem-loading and transport to the shoot, and phloem-mediated redistribution of metals from mature leaves to sink tissues, including younger leaves, roots, and seeds . Cadmium enters the root through the Fe transporter IRT1, which shows broad substrate specificity towards divalent metals including Fe2+, Zn2+, Mn2+, and Cd2+ . Once inside the cell, metals bind to different ligands, according to specific affinities, and these metal–ligand complexes can be stored in different cellular compartments or distributed to other tissues through the vasculature . Because of the broad substrate specificity of IRT1 for divalent metals, transcriptional regulation of the Fe-deficiency response,fodder system trays including up-regulation of IRT1, will also have an impact on the uptake of non-essential heavy metals such as Cd. In plants, the root iron-deficiency response is regulated by local signals within the root and also by systemic signals originating from leaves . Two major transcriptional networks have been identified to mediate the Fe-deficiency response at the root level in Arabidopsis: the FIT network and the POPEYE network .The identification of mutants showing a constitutive Fe-deficiency response even when Fe is supplied in sufficient amounts plus experiments where the constitutive root response is restored by foliar application of Fe suggest that mobile Fe is required for proper shoot-to-root signaling . However, the transporters, ligands, and the chemical speciation of the putative phloem-mobile molecule mediating the systemic Fe signaling have not yet been clearly identified. Here, we report that opt3-2, an Arabidopsis mutant carrying an insertion in the 5’ UTR of the oligopeptide transporter gene OPT3 , over-accumulates significant levels of Cd in seeds. We present evidence suggesting that this Cd over-accumulation may be the result of an enhanced transport of Cd through the plant, making opt3-2 a suitable background for studying long-distance transport of non-essential heavy metals. We further show that OPT3 is targeted to the plasma membrane and is preferentially expressed in the phloem.

The Fe/Zn/Mn uptake transporter IRT1 and other ironstarvation-induced genes are constitutively up-regulated in opt3-2. Interestingly, shoot-specific expression of OPT3 restores metal homeostasis and IRT1 up-regulation in roots showing that OPT3 is the first identified molecular component of the network transferring information on the iron status from leaves to roots. Moreover, Fe mobilization between leaves is impaired in opt3-2, suggesting that OPT3 mediates the movement of Fe out of the leaves, and this transport is required for proper communication between leaves and roots and maintenance of the trace-metal homeostasis in Arabidopsis. Understanding phloem-mediated signaling, transport, and seed-loading mechanisms of both essential and non-essential heavy metals will help to develop strategies for excluding toxic metals from seeds and enhance the nutritional value of grains and plant-based products.Members of the Arabidopsis oligopeptide transporter family have been shown to mediate the transport of a broad spectrum of peptides . Glutathione and phytochelatins are peptides that mediate tolerance and long-distance transport of heavy metals ; therefore, we screened mutants in the Arabidopsis OPT family for differential accumulation of Cd in seeds. A mutant of the Arabidopsis OPT3 gene, opt3-2, showed the strongest over-accumulation of Cd in seeds . To test whether this Cd over-accumulation had an effect on seedling growth, assays were performed on plates in the presence and absence of Cd. Figure 1B shows that opt3-2 is hypersensitive to Cd when grown on medium containing 50 μM CdCl2. To determine whether the increased Cd concentration in opt3-2 seeds was due to a systemic over accumulation of Cd throughout the plant, opt3-2 seedlings were grown hydroponically for 6 weeks, exposed to 20 μM CdCl2 for 72h and the metal concentration of roots and leaves was measured by ICP–OES . The roots of opt3-2 over-accumulated Cd compared to wild-type; however, unexpectedly, Cd concentrations in leaves were almost five-fold less than those of wild-type plants . Conversely, seeds of opt3-2 plants show a large increase in Cd levels compared to wild-type seeds .To determine whether the altered distribution of Cd in opt3-2 correlated with the distribution of essential metals in plant tissues, the levels of Zn, Fe, and Mn in opt3- 2 were also measured and compared to wild-type plants . No dramatic differences in the concentration of Zn and Mn in seeds were found between wild-type and opt3-2 . However, in contrast to Cd accumulation, opt3-2 over-accumulated significant levels of Zn and Fe in leaves compared to wild-type . In roots, the concentration of Fe, Zn, and Mn was increased in opt3-2 compared to wild-type . The different distribution of Cd in aerial parts of the plants suggests that the mechanisms mediating accumulation of metals in opt3-2 leaves is different for Cd compared to the essential metals Fe, Zn, and Mn.