lable at ScienceDirect

Journal of Environmental Radioactivity 129 (2014) 157e168

Contents lists avai

Journal of Environmental Radioactivity

journal homepage: www.elsevier .com/locate/ jenvrad

OBT/HTO ratio in agricultural produce subject to routine atmosphericreleases of tritium

V.Y. Korolevych a,*, S.B. Kim a, P.A. Davis b

a Environmental Technologies Branch, Nuclear Sciences Division, Chalk River Labs, AECL, Chalk River, ON, Canada K0J 1P0bRad.Safe Inc., Canada

a r t i c l e i n f o

Article history:Received 6 November 2012Received in revised form17 December 2013Accepted 18 December 2013Available online 4 February 2014

Keywords:TritiumHTOOBTVegetation-atmosphere exchange

* Corresponding author. Tel.: þ1 613 584 8811.E-mail addresses: korolevv@aecl.ca, vladi_kor@yah

0265-931X/$ e see front matter Crown Copyright �http://dx.doi.org/10.1016/j.jenvrad.2013.12.014

a b s t r a c t

The mean expected value of the OBT/HTO ratio (i.e. generic ratio) is derived in this study on the jointbasis of a long-term study conducted at Atomic Energy of Canada Limited (AECL)’s Chalk River Labora-tories (CRL), model simulations targeted at filling gaps in a yet incomplete timeline of CRL measurementsand a reference dataset comprised of numerous experiments reported in the literature. Cultivar vari-ability and disparity in site-specific settings are covered by the reference dataset. Dynamical variabilitycaused by meteorology has been a specific target of the long-term experimental campaign at CRL, wherethe former two types of variability were eliminated. The distribution of OBT/HTO ratios observed at CRLappears to be a fairly good match to the distribution of OBT/HTO ratios from the literature. This impliesthat dynamical variability appears important in both cases. Dynamics of atmospheric HTO at CRL iscomprised of a sequence of episodes of atmospheric HTO uptake and re-emission of plant HTO. The OBT/HTO ratio appears sensitive to the proportion of the duration of these two episodes: the lesser thefrequency (and duration) of plume arrivals, the higher the expected mean OBT/HTO ratio. With theplume arrival frequency defined by the typical wind rose, one would encounter a mean OBT/HTO ratioclose to 2. It is important to note that this number is seen both in the reference dataset, and in thecontinuous timeline of HTO and OBT reconstructed from CRL observations by dynamical interpolation(modelling). Many datasets (including that of CRL) targeted at the OBT/HTO ratio are biased highcompared to the suggested number. This could be explained by scarce measurements of the low OBT/HTO ratios in the short phase of uptake of atmospheric HTO by the plant.

Crown Copyright � 2014 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Organically bound tritium (OBT) in plants is responsible for anappreciable portion of the ingestion dose during chronic releases oftritium (Evans, 1969; Hisamatsu et al., 1989; Gulden and Raskob,1992; Kim and Han, 1999; Kotzer and Trivedi, 2001; Peterson andDavis, 2002). The cost of frequent monitoring of OBT makes itlogical to look at the plant tissue-free water tritium (HTO) as itsproxy, provided the OBT/HTO ratio is known. Conversely, the ratiocould be used for probabilistic quantification of plant HTO on thebasis of OBT (and especially non-exchangeable OBT), which doesnot fluctuate as much as plant HTO does. OBT/HTO ratio also in-dicates the ability of plants to concentrate tritium into the organicfraction (Okada and Momoshima, 1993; Boyer et al., 2009).

oo.com (V.Y. Korolevych).

2014 Published by Elsevier Ltd. All

Quantification of the OBT/HTO ratio and interpretation of mea-surements, however, appears difficult (CNSC, 2011). The majordifficulty is with explanation of measurements. According to pre-vailing considerations of the Specific Activity model presently un-derlying all regulatory algorithms (CSA, 1987), the OBT/HTO ratio isexpected to be less than unity. However, field observations do notsupport this suggestion (Bogen andWelford, 1976; Hisamatsu et al.,1987, 1989, 1990, 1992; Brown, 1988, 1995; Pointurier et al., 2004;Inoue and Iwakura, 1990; Momoshima et al., 2000; Baglan et al.,2005; CNSC, 2011). Unambiguous measurements are difficult pri-marily due to the high cost, which prevents the analysis of largeenough samples. Ambiguity inherent to small samples stems fromthe natural spatio-temporal variability in the ecosystem (cultivar,climatic, meteorological, hydrological, etc.) and analytical un-certainties (Diabate and Strack, 1993; Baumgärtner and Donhärl,2004; Pointurier et al., 2004). This study targets the temporalvariability caused by the meteorology and the associated scatter

rights reserved.

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168158

observed in the OBT/HTO ratios (Kim and Davis, 2008; Korolevych,2013).

Most tritium released to the atmosphere from CANDU reactorsis in the form of tritiated water vapour (HTO). The tritium in theeffluent exchanges readily with hydrogen in air moisture, precipi-tation and soil water. Some of this tritium appears in plants viadiffusion from the air through the stomata and via root uptakethrough the transpiration stream (Belot, 1986). This tritium isdistributed throughout the free water portion of the plant freshtissue, in which form it is referred to as either tissue-free watertritium (TFWT) or HTO. The HTO designation will be used here.Some of this tritium can be incorporated into organic compoundsto form organically bound tritium (OBT) (Diabate and Strack, 1993).OBT is formed primarily in the leaves and can be translocated to theedible parts of the plant.

The exchange of tritium (T) and hydrogen (H) atoms in thevarious free water compartments of the environment is very rapid.HTO concentrations in plants have a biological half-life of a fewhours (Strack et al., 2005) and tend to track changes in air con-centrations. In contrast, OBT formation is a chemical rather than anexchange process and occurs over the lifetime of the plant whenthe relevant metabolic processes are operative. An OBT measure-ment at a point in time reflects the integrated effects of the dailyrates of dry matter production in the previous weeks and the HTOconcentration in the plant water that took part in the production.Tritium in OBT is fixed in the plant to the same extent as carbon,with a biological half-life of about 25 days (NCRP, 1978). It does notimmediately change in response to changes in the air concentrationor plant HTO concentration. If the tritium source is removed, theOBT concentration in the plant decreases slowly by conversion toHTO as the dry matter breaks down through catabolic processesand by dilution with new, uncontaminated dry matter as the plantcontinues to grow.

The great mobility of tritium in the HTO form has led to thebelief that the T/H ratio (or equivalently, the HTO concentration)is the same in all interacting water compartments of the envi-ronment, which is the basis of the specific activity (SA) model.The SA model underlies almost all environmental tritium models(Evans, 1969). It is expected that SA concepts apply to OBT as wellsince the OBT formed by a given plant process at a given time hasa T/H ratio that reflects the ratio in the water that enters into thatprocess. In theory, OBT concentrations are lower than HTO con-centrations because the large difference in mass betweenhydrogen and tritium gives rise to significant isotopic effects inOBT formation. The oxygen-tritium bonds of tritiated water aresplit less frequently than oxygen-hydrogen bonds during photo-synthesis, so less tritium than hydrogen is incorporated intoorganic molecules. In contrast, carbon-tritium bonds in the plantdry matter are severed more slowly than carbon-hydrogen bonds.The net result is an OBT concentration that is slightly lower thanthe HTO concentration. This can be interpreted as an isotopicdiscrimination factor (IDp) and also as the OBT/HTO ratio in theplant.

Observed values of the OBT/HTO ratio in plants show largevariations that are not consistent with the value of IDp (e.g.IDp ¼ 0.8 as per recommendations of CSA, 1987) expected fromthe SA model (Bogen and Welford, 1976; Hisamatsu et al., 1987,1989, 1990, 1992; Brown, 1995; Pointurier et al., 2003, 2004;Inoue and Iwakura, 1990; Momoshima et al., 2000; Baglan et al.,2005). At first glance, observations such as these suggest thatprocesses other than simple isotopic fractionation are at play inconcentrating or diluting tritium in organic tissue (Baumgärtnerand Donhärl, 2004). The existence of such processes would castdoubt on the SA model, which is the basis for all regulatory doseassessments of chronic tritium releases. In order to maintain

confidence in the model, the large ratios need to be explained interms of a process, condition or experimental procedure that vi-olates SA assumptions of steady state, but not the principles setforth by the SA approach.

Murphy (1993) analyzed the HTO concentrations in air, plantsand soil, and noticed that all these variables are in a highly dynamicstate. When the atmospheric plume is present, air concentrationsare high, plant HTO concentrations build up rapidly and then leveloff, and surface soil concentrations increase steadily. When theatmospheric plumemoves off, air concentrations decrease abruptlybut remain well above the undisturbed background due to re-emission of deposited tritium. Similarly, plant concentrationsdrop off rapidly to levels that reflect concentrations in the tran-spiration stream. Surface soil concentrations also decrease quickly,although concentrations at deeper levels change only slowly withtime. The cycle repeats itself when the wind brings the plume backover the location in question. In general, concentrations in any onepart of the system are not in equilibriumwith concentrations in anyother part at any time. This creates problems in interpreting tritiumfield measurements. For example, the HTO concentration in vege-tation can be much lower than the air concentration (if the plantsample is taken just after the airborne plume reaches the site) ormuch higher (if the sample is taken just after the plume moves offthe site). These are not examples of isotopic effects or tritium bio-accumulation, but simply analytical artefacts pertinent to transientsystem.

The key part of the dynamical exchange process is the succes-sion of phases of HTO uptake and re-emission. The proportionbetween the average length of deposition (when the plume ispresent) and the average length of re-emission (when the wind isblowing the plume away from the receptor) is site-specific and iswell-defined by a wind rose. The difference between instantaneousmeasurements of HTO and OBT in a plant is defined by a stage of thedepuration process, and is quantified by a difference between therate of HTO re-emission and the OBT depuration rate. OBT/HTOratios measured and reported worldwide, as well as at CRL, wereassembled into a database and compared with results of calcula-tions. Comparison revealed consistency in the generic OBT/HTOratio and showed that the uncertainty of the ratio is lower than thecurrently known scatter of experimental data. The comparison alsoshowed that the OBT/HTO ratio is sensitive to uncertainties inplant-atmosphere exchange rates and the frequency of plumepresence (as per wind rose).

In this study, changes in air HTO concentration are hypothesizedto be the sole cause of persistently non-unit values of the OBT/HTOratio in plants. This implies that normal vegetation sampling pro-cedures inevitably lead to values of the plant OBT/HTO ratiosignificantly different from one, and in most cases greater than one.However, whether the occurrence of high values of the OBT/HTOratio arise simply as a result of normal plant sampling procedures,or these values provide a cause to doubt the validity of the SAprinciples and are an indication of the bioaccumulation of tritiumin plants, is not yet clear. It is therefore the objective of the presentstudy to investigate the consequences of random sampling and thedynamics of tritium transfer processes that occur under typicalmeteorological conditions and subsequently result in changingHTO concentration in the air.

In Section 2, the analytical techniques used to determine HTOand OBT concentrations in plants are discussed. The SA model isdescribed in more detail in Section 3, along with a theoreticaldiscussion of the conditions and processes leading to extremevalues for OBT/HTO ratio. Measured values of the ratio from theliterature are reported in Section 4. Values significantly differentfrom one are explained in terms of the conditions and mechanismsidentified in the theoretical part of the study.

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168 159

2. Methods

2.1. Definition of OBT

Tritium can be bound to organic compounds either by exchangereactions or by enzymatically-catalyzed reactions (Diabate andStrack, 1993; Belot, 1986). In exchange reactions, tritium bonds tooxygen, sulphur, phosphorous or nitrogen atoms as hydroxides,thiols, phosphides or amines, respectively. Conventionally, this istermed exchangeable or labile OBT. Exchangeable OBT is consid-ered to be in equilibrium with HTO in the plant or animal inquestion. In enzymatically-catalyzed reactions, tritium bonds to thecarbon chain of an organic molecule as fixed or non-exchangeableOBT. Such bonds are strong and can be dissolved only duringcatabolic reactions, meaning that non-exchangeable OBT has alonger retention time in the body than HTO or exchangeable OBT.

The literature contains differing views onwhether the definitionof OBT should include exchangeable OBT. Diabate and Strack (1993)state that OBT should mean only non-exchangeable OBT, and this isadopted by the EMRAS Tritium Workgroup (IAEA, 2008). Morerecently, Baumgärtner and Kim (2000), Baumgärtner et al. (2001),Baumgärtner and Donhärl (2004) and Baumgärtner (2005) haveidentified another form of OBT called buried tritium. Buried tritiumis defined as the tritium in exchangeable positions in large bio-molecules in dry matter that is not removed by rinsing withtritium-free water. Such tritium is not carbon-bound, but is simplyfolded into large molecules that are not accessible for exchangewith tritium-free water. Baumgärtner and Donhärl (2004) suggestthat buried tritium makes up 50% or more of what is traditionallymeasured as OBT.

The definition from the Environmental Agency (2001) and thatof the Canadian Standards Association (1987) is wider and includesany organicmatter containing tritium, either exchangeable or fixed.This definition ensures that all forms of OBT are taken into accountin dose assessments.

2.2. Analytical considerations

The measurement of the HTO concentration in plants beginswith the extraction of water from the fresh sample. This is usuallydone either by azeotropic distillation with toluene or by freeze-drying. The extracted water is mixed with scintillation fluid andplaced in a liquid scintillation (LS) counter, which measures theactivity in the sample.

To measure the non-exchangeable OBT concentration in plants,the freewater is first removed from the sample. This is best done byfreeze-drying since the organic residue left on the dry matter afterazeotropic distillation can affect the measurement. The dry sampleis washed repeatedly with tritium-free water to removeexchangeable OBT and then dried. The dry material is oxidized,typically in a Parr bomb, and the combustionwater is collected. Thisis mixed with scintillation fluid and counted in a LS counter. Ifrinsing is omitted, exchangeable OBT remains.

Errors in the measured HTO and OBTconcentrations can enter atmany points in the analytical process. HTO is relatively easy toextract from the plant matrix and counting errors can be kept low ifthe sample is counted for a sufficiently long period of time. Caremust be taken to prevent the sample from becoming cross-contaminated. If analytical procedures are carefully followed, theuncertainty in the measured HTO concentration in plants can bekept around 10e20% (Kim and Baumgärtner,1994; Fuma and Inoue,1995).

The measurement of OBT presents additional challenges(Pointurier et al., 2004). It is difficult to remove all of the labile OBTfrom the sample and small residual amounts will result in an

overestimate of the concentration of non-exchangeable OBT.Further uncertainties arise in combusting the dry sample, and inextracting and counting the combustionwater, which is often acidicand contains numerous impurities. These problems are exacer-bated by small sample sizes. There are other uncertainties associ-ated with LS counting which are applicable to HTO measurementsas well. The tritium concentration in the sample is determined asthe difference between the gross count rate and the count rate of a“tritium-free” water sample, or blank. In practice, it is difficult toobtain water that is truly free of tritium. At environmental levels,the counts from the sample and the blank may be similar and thedifference between the two may be subject to large errors.

Estimated uncertainties associated with the various steps inmeasuring OBT concentrations in environmental samples result inan overall uncertainty of about a factor of 2 (Fuma and Inoue, 1995;Garland and Ameen,1979). This estimate is supported by the resultsof inter-laboratory comparisons using OBT “CRMs” in lieu of stan-dard (Workman et al., 2005).

2.3. Theoretical considerations

2.3.1. Specific activity modelThe fundamental assumption of the SA model is that the T/H

ratio is the same within interacting environmental compartments.In the present context, SA equilibrium implies that the tritiumconcentration in the aqueous phase of the plant (the HTO con-centration) is equal to the tritium concentration in the combustionwater of the plant dry matter (the OBT concentration) when bothconcentrations are expressed in Bq L�1.

SA equilibrium should hold in areas where HTO concentrationsin air and plant water are constant over time. Since OBT formsdirectly from HTO, concentrations in the dry matter should be thesame as those in the plant water, apart from isotopic effects. In thevicinity of a local atmospheric tritium source, soil and air moistureconcentrations fluctuate in time because of changes in meteoro-logical conditions, particularly wind direction. These will induceassociated fluctuations in plant HTO concentrations that will not bereflected in the OBT concentrations, which have a longer residencetime. But even under these conditions, SA equilibrium should beachieved if the OBTconcentration at a given time is compared to theHTO concentration averaged over the month or two prior to thattime, given that the HTO concentration is known. The SA modeldoes not apply to a short-term release.

In theory, isotopic discrimination between tritium andhydrogen results in OBT/HTO ratios somewhat less than one atequilibrium. Theoretical estimates of the ratio are not availablebecause fractionation effects are difficult to quantify in plantmetabolic processes, and because it is not completely clear whatprocesses are involved in OBT formation. However, a “theoretical”ratio can be obtained by measuring the OBT/HTO values in plantsgrown in a fully equilibrated environment. The available empiricaldata suggests that the OBT/HTO ratio lies in the range 0.7e0.8under equilibrium conditions (Section 3.1).

2.3.2. Dynamical change of air HTO leading to the non-unit value ofthe OBT/HTO ratio

OBT and HTO have significantly different formation and clear-ance times in plants. An HTOmeasurement at a single point in timereflects the instantaneous HTO concentration at that time. Incontrast, a single OBT measurement reflects an average over theprevious several weeks. If the plume is present when the samplesare taken, the HTO concentration will be larger than the OBT con-centration and the ratio will be less than 1. If the plume is notpresent, the HTO concentration will be less than the OBT concen-tration and the ratio will be greater than 1. Since the wind tends to

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168160

blow at most 15% of the time into a given sector/direction and therest of the time blows elsewhere, it is likely that the plume will notbe present at a randomly chosen sampling time. Thus, ratios greaterthan 1 will be observed more frequently than ratios less than 1.Large ratios are also to be expected in store-bought produce, whichwill retain its OBT concentration during harvest, shipping andstorage, but lose some of its HTO. Values substantially differentfrom one may also be observed if the HTO release rate from thelocal source varies in time. Ratios greater than (less than) one willoccur if the release rate in effect at the time of sampling is less than(greater than) that seen by the plant a fewweeks prior to sampling.

The dynamical effect of the changes in air HTO concentrationson OBT/HTO ratios can be quantified by the following example.Consider a site at which the HTO concentration in the plant is1 Bq L�1 when the airborne plume is present. When the plume isabsent, the concentrationwill be much lower, but greater than zerobecause of residual HTO in the soil water. A value of 0.2 Bq L�1 willbe used here for illustrative purposes. Assuming the plume ispresent 15% of the time, the long-term average HTO concentrationwill be 0.32 Bq L�1. Under the assumption that SA equilibrium holdslocally, the OBT concentration will also be 0.32 Bq L�1 (neglectingisotopic effects). This leads to an OBT/HTO ratio of 0.32 if the plantis sampled when the plume is present and a ratio of 1.6 if the plantis sampled when the plume is not present. The expected value in arandomly collected representative samplewill therefore be equal to1.48.

Except in extremely remote areas of the world, tritium con-centrations will lie above nominal background levels if the windblows from a regional source of tritium toward the site. Fluctuatingweather conditions will occur on longer time scales than those seennear local tritium sources, but will have a similar effect on OBT/HTOratios. Measurements made at a point in time will reflect theinstantaneous concentration of HTO, but the time-averaged con-centration of OBT. Thus, the ratio will deviate from its equilibriumunit value.

OBT/HTO ratios significantly different from one are just ascommon in background areas as they are in areas close to tritiumsources. In neither case can true equilibrium conditions be ex-pected to hold. Normal sampling techniques in which the HTOconcentration is determined in a single plant sample takenrandomly in time will not yield a value representative of SA con-ditions. Values obtained if the plant is grown under controlledconditions in the laboratory, where the HTO concentration is heldconstant or monitored continuously, would be different from thatin field samples. This is due to a mismatch in the times when theplume is present and when it is not, as well as (to a lesser extent)missing the field-scale holding capacity of soil water.

Table 1OBT/HTO ratios measured near tritium sources.

Year Average � SD Range Samplesize (n)

Reference

1981 1.83 � 0.67 0.79e3.18 22 Strack and Konig (1981)1983 1.52 n/a 33 Momoshima et al. (1991)1992 1.34 � 0.09 1.21e1.46 75 Kim and Han (1999)1996 1.35 � 0.49 0.88e2.52 59 Kim et al. (2000)1997 1.00 � 0.50 0.25e1.48 6 OPG (1999)1997 0.60 � 0.20 0.48e1.00 4 OPG (1999)2002 n/a 0.90e12.00 n/a ATSDR (2002)2003 1.30 � 0.34 0.89e1.79 6 OPG (2004)2004 1.60 � 0.80 0.70e2.92 5 OPG (2005)n/a 8.36 � 12.15 0.13e45.50 21 Golubev et al. (2002)2005-2009 2e3 n/a n/a CNSC (2011)n/a 1.92a � 1.42 0.1e4.8 458 Jean-Baptiste et al. (2011)

a This number comes from a compilation of literature data where “a substantialfraction of the dataset pertains to sites close to local tritium sources”.

2.4. Approach

Certain limitations are inherent to small samples, samplescollected at the single site and to modelling. Since these limitationsare essentially independent and different, we analyzed the OBT/HTO ratio using the combination of the information available in thefirst case (small samples) with the data collected at one site ChalkRiver Laboratories (CRL) and generated by the model, i.e.:

� OBT/HTO ratio from experiments pertaining to measurementsof the OBT/HTO ratio performed worldwide and reported in theliterature (collected in the integrated “reference dataset”);

� OBT/HTO ratio from new CRL field experiments, accompaniedby the extensive record of ambient conditions and performed inset conditions on a fine half-hourly scale (CRL observationsdataset);

� OBT/HTO ratio from carbon-based modelling of plant OBT andHTO deployed for interpolation of new CRL data.

The aim is to find similarities between these different OBT/HTOdistributions and proceed with the quantification of generic OBT/HTO ratio.

The reason for a combined observations-modelling approachstems from the results of a pilot study that analyzed the datacollected at CRL during the growing seasons of 1995 and 2008e2011. CRL data does not appear much different in size from someother studies (Kim and Han, 1999; Pointurier et al., 2004) and yet,there appears to be no consistency between the results of thesestudies, each based on large samples. We assume that the syner-gistic analysis deploying both observations and modelling will helpremove the noted ambiguity.

2.5. Observations

2.5.1. Reference datasetA review was carried out to identify measured plant OBT/HTO

ratios in the literature. Since the work of Murphy (1993) was per-formed in the vicinity of the source, it remains unclear whetherdynamical processes equally prevail in the far field. For this reason,each reported value obtained in the vicinity of a tritium source wasset apart from those obtained in the far field, traditionally perceivedto be background areas. The values found represent all times of thegrowing season, conditions in Europe, Asia and Canada, naturalplants (tree leaves, needles, lichen and grass) and agricultural crops(leafy vegetables, fruit, grain, roots). The measurements are sum-marized in Tables 1and 2.

As much information as possible, on the conditions under whicheach OBT/HTO ratio was obtained, was extracted from the studyreports referenced in Tables 1and 2 In particular, information on thetime of measurement, the averaging time, the locationwith respectto tritium sources and the uncertainty in the measured HTO andOBT concentrations was noted where it was available. These con-ditions were used to assign the measurement to one of the twocategories defined above.

In order to confirm that no bioaccumulation occurs, we alsoreviewed the OBT/HTO ratios obtained in ostensibly similar con-ditions, but with the local source of nearly constant magnitude(near waste management areas, where the source is the soil water).In this particular category, the equilibrium assumption underlyingthe SA model holds and the SA model in its traditional form isdecidedly valid. Therefore, strictly speaking, this is the only cate-gory where the deviation of the OBT/HTO ratio from the smallerthan one isotopic discrimination ratio (CSA, 1987) would clearly

Table 2OBT/HTO ratios in the far field away from the source.

Year Mean � SD Range Samplesize (n)

Reference

1970 3.25 � 0.86 1.87e4.37 7 Bogen and Welford (1976)1970 3.7 � 0.40 3.39e3.96 2 Bogen et al. (1979)1970 3.15 3.15 1 Clemente et al. (1979)1983 1.46 � 0.31 n/a 21 Takashima et al. (1987)1983 1.43 n/a 20 Momoshima et al. (1991)1985 1.74 � 1.21 0.57e4.30 7 Hisamatsu et al. (1987)1986 1.54 � 0.57 0.67e2.60 14 Hisamatsu et al. (1989)1987 1.30 � 0.85 0.69e3.00 6 Hisamatsu et al. (1989)1987 1.19 � 0.18 0.80e1.40 7 Hisamatsu et al. (1992)1987e88 1.09 � 0.15 0.95e1.30 4 Hisamatsu et al. (1990)1987e88 1.08 � 0.41 0.49e1.90 14 Hisamatsu et al. (1992)1991 0.89 n/a 6 Momoshima et al. (2000)1991 0.73 n/a 6 Momoshima et al. (2000)1997 1.00 � 0.30 0.63e1.48 6 OPG (1999), analysis

by CRL2001 1.46 � 0.89 0.34e3.93 18 OPG (1999), analysis

by CRL2002 7.10 � 6.60 1.33e24.60 31 Bruce (2003), analysis

by CRL2003 24.00 � 35.00 4.00e141.00 14 OPG (2002)2004 5.90 � 1.65 3.90e8.00 5 OPG (2004)n/a 0.83 � 0.31 n/a n/a Inoue and Iwakura (1990)n/a 0.62 � 0.05 n/a n/a Inoue and Iwakura (1990)n/a n/a 0.10e1.60 n/a Inoue et al. (1992)n/a n/a 0.90e2.10 n/a Takashima (1987)n/a 1.60 � 0.5 n/a 70 Pointurier et al. (2004)n/a 3.00 � 1.5 n/a 63 Pointurier et al. (2004)n/a 2.20 � 0.64 1.57e2.86 3 Baglan et al. (2005)2002e2005 1.90 � 0.90 0.43e3.94 14 Jean-Baptiste et al. (2007)

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168 161

indicate bioaccumulation of some kind. This category of OBT/HTOratios was compared to ratios obtained in controlled laboratoryexperiments.

2.5.2. CRL datasetSome causes of the large scatter of the OBT/HTO ratio, and

especially the frequency of occurrence of extreme ratios, remainunclear. This resulted in two primary motives for funding the newexperimental campaign. Firstly, the detailed ambient data isneeded for an explanation of the OBT/HTO ratios scatter (hourlymeteorology, soil temperature, etc.) and yet, it was often not re-ported alongside collected OBT/HTO ratios from the referencedataset. Dynamic measurements of OBT/HTO ratios in a constantsetting and on a much finer scale than regular monitoring weresubsequently started at CRL in 2008.

Second is the disparity of mean values of the OBT/HTO ratioencountered in different experiments, which led us to believe thatsome parts of the dynamical HTO exchange process are frequentlymissing. The beginning of the re-emission phase (when the plumehas just departed) was specifically suspected because of the highestOBT/HTO ratios associated with it, as per a pilot study conducted atCRL in 1995 (Davis, 2007). Subsequently the follow-up measure-ments at CRL in 2008e2011were targeted at the re-emission phase.

The CRL site is located in Ontario on the south shore of theOttawa River, about 200 km northwest of Ottawa. Low amounts ofradioactive airborne and liquid releases occur from CRL duringnormal operations. Tritium is released routinely by nuclear facil-ities, with HTO being the most abundant form of tritium. Differentlocations with homogeneous land cover and vegetation subject totritium deposition were analysed at CRL and three sampling loca-tions were chosen: Building 600, SE of the NRU reactor; Building513, S of the NRU reactor and SE of the reactor stack; and the AcidRain Site (ARS), within 2.5 km NW of the NRU reactor and thereactor stack (Fig. 1). The ARS was chosen for the experimental

garden plot because the NWdirection coincides with the prevailingwinds at CRL, and planning sampling campaigns in-line with thearrival and the departure of a plume was more convenient than inother sectors. The biological samples included (Figs. 1 and 2):

� grass (Agrostis perennans) e collected at Building 600,� dandelion (Taraxacum officinale), American Beech (Fagus gran-difolia) e collected at Building 513,

� rush (Juncus tenuis), upland bent grass (A. perennans), witch orcouch grass (Agropyron repens), potato plants (S. tuberosum,Russet Burbank) tomato plants (S. lycopersicum, BeefsteakSlicing), green and red pepper plants (Capsicum) e collected atthe ARS.

Experiments included the overnight period, lasted several hoursand samples were collected at sub-hourly intervals. In addition tothis, onemorning sample and one noon sample of not ripe, but fullygrown tomatoes and pepper were collected. At the same time ofday over the course of a typical harvest period (within an interval ofa few days), ripened potato, tomato and pepper were collected. Inearly spring, one part of the experimental garden plot was strippedof the organic and top-soil layers, and covered by the exportedtritium-free mixture of the organic soil and sand to the consistencyof the soil in the control portion of the garden plot. Clean soil wascovered with a tarp and plant seedlings were planted in small holesin the tarp. The neighbouring control part of the garden was leftexposed to the atmosphere. This was done in order to evaluate therole of soil HTO in the formation of plant OBT, and particularly fruitsand tuber OBT (Fig. 2c and d).

Each sample was a composite of a minimum of five fruits (tu-bers) of the same species collected randomly over the whole gar-den. Samples from the tarp-covered part were collected separatelyfrom those grown on the control part in ambient setting.

2.6. Dynamical interpolation using modelling

Measurements precisely at the beginning of the uptake phaseand observation of ultimately low OBT/HTO values remain prob-lematic because this phase is very narrow. The corresponding lowOBT/HTO ratios are frequently missing in the five years of CRLmeasurements and thus, it was decided to support observationswith modelling.

Themodelling has been performed over a relatively short periodof time comprised of a few uptake-depuration episodes. Modelresults were aligned with observations of HTO and OBT in plant.The use of a model in the reconstruction of a continuous timeline ofa given variable in between observations is termed dynamicalinterpolation.

2.6.1. Reconstructing continuous timeline of driving air HTOThe methodology of reconstruction of the continuous timeline

of driving air HTO from the high-frequency readings of the Ar-41 bythe ambient gamma-radiation monitor is presented in Korolevychand Kim (2013). Empirical correlation of the measured Ar-41 withair HTO accounts for Ar-41 decay occurring during the plume travel.The travel time is defined by the wind speed and intensity of plumemeandering measured by sigma-theta (wind direction standarddeviation). Ambient humidity is taken into account and the dif-ferences between transport of water vapour (HTO) and tracer gas(Ar-41) appear minor. The average travel time combines twosources of tritium atmospheric emissions (about half of air HTO isreleased from the CRL NRU reactor stack and most of the other halfe from the reactor building roof vents). The data from gamma-monitor located at ARS was pre-processed and substituted with ahalf-hourly running mean.

Fig. 1. The map of sampling sites (marked by stars) at CRL.

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168162

The ARS was chosen for empirical reconstruction of air HTObecause it is located on the pass of ESE prevailing winds bringingthe plume both from roof vents and the stack of the NRU reactor,thus making the atmospheric dispersion of Ar-41 and HTO (whenthey reach ARS) similar. The prevailing nature of winds in thissector makes the frequency of plume occurrence at the ARS high(�16%) and a plume tangible every time the wind blows towardsthe ARS. The uncertainty factor of 2 is associated with the recon-struction of air HTO from Ar-41.

2.6.2. ModellingTo outline the role of explicit modelling of the carbon cycle,

which is critical for correct quantification of the OBT concentration,the process-based (carbon-water-energy) model with HTO transferported to its water cycle, and OBT formation and translocationported to its carbon cycle was deployed. The model is outlined inAppendix A.

The model was driven by the realistic continuous atmosphericconcentration based on real weather conditions and ambient

Fig. 2. a) The continuous air sampling system and the Beech tree that was sampled at CRL (Bthe impervious pavement. c) The experimental garden plot at CRL (ARS), potato plants areportion of the plot had initially clean and protected soil. d) The experimental garden plot attomato plants are to the right.

gamma-radiation monitoring. The typical sequence is comprised ofa real intermittent plume (when wind blows from the source),followed by the real tails approaching actually observed localbackground (when the wind blows elsewhere).

3. Results and discussion

3.1. Overview of published results (reference dataset)

Table 1 shows the OBT/HTO ratios reported for various plants inthe vicinity of nuclear facilities worldwide. It is worth noting, thatKim and Han (1999) and Kim et al. (2000) present similar meanresults obtained for similar plants (rice, vegetables and pine nee-dles), but the latter reports much larger range of observations,comparable to that of measurements made in Germany (Strack andKonig, 1981) on tree leaves, apples and the leaves of Brusselssprouts. A larger range reaching ratios of 12 was observed in grassand leaves at Four Mile Creek on the Savannah River Site in theUnited States (ATSDR, 2002) and much greater variation (0.13e

513). b) The dandelion plant that was sampled at CRL (B513). The plant is surrounded byin front and tomato plants are behind. Soileatmosphere exchange is reduced, as thisCRL (ARS), which is a natural setting with exposed soil. Potato plants are to the left and

Table 3OBT/HTO ratios at CRL.

Year Average � SD Range Sample size (n) Location

1995 1.45 � 1.87 0.38e11.38 54 B600, ARS2008 4.02 � 11.95 0.30e88.78 131 B513, ARS2009 2.16 � 0.57 1.48e3.11 10 ARS2010 5.87 � 3.93 2.69e19.10 15 ARS2011 1.17 � 0.78 0.52e2.64 10 ARS

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168 163

45.50) was measured on lichen samples from Russia (Golubev et al.,2002). The recent study by CNSC (2011) confirmed the large vari-ability of OBT/HTO ratios near the tritium source. Substantialvariability is summarized by Jean Baptiste et al. (2011).

Table 2 summarizes OBT/HTO ratios in various plants in the farfield away from the source in what was supposed to be backgroundareas of theworld. The earliest results were obtained in the U.S. andItaly in the late 1970s. Additional measurements were carried out inJapan in the 1980s and 1990s, and more recent data have becomeavailable from an extensive study in France. However, given thelarge number of nuclear facilities in all of these countries, it isunlikely that these measurements represent true background(equilibrium) levels. From this standpoint the occasional encounterof large OBT/HTO ratios in the far field is expected, e.g. 4.37 inBogen and Welford (1976), 4.30 (without rinsing to removeexchangeable OBT from the dry matter) in some of the Japanesedata (Hisamatsu et al., 1987) and 3.94 in southeast France data(Jean-Baptiste et al., 2007). Baglan et al. (2005) reported a smallerrange, but a consistently large mean ratio of 2.2 � 0.64 in leafsamples, although they noted that the associated uncertainty wasquite high.

Canadian results add to the noted tendency. Since 1997, OntarioPower Generation (OPG) and Bruce Power have measured HTO andOBT concentrations at a number of background sites (Barrie, Ban-croft, Lambton, Kingston, Sarnia, Picton, Apsley and Brockville) inOntario as part of their Radiological Environmental MonitoringPrograms (REMP). At a given site, sampling was done duringdaylight hours on a single day toward the end of the growingseason. Most OBT measurements were made on composite sam-ples, which may have included mixed natural plants or vegetables.It is worth noting the high variability (from 2.5 to 6.4 Bq L�1) of HTOconcentrations in plants at the sites in the far field traditionallyperceived as background areas. Moreover, the range of HTO values(fromminimum tomaximum) in a given year showedmuch greatervariation than the average values from year to year. In particular, in2003, some atypically low HTO values (down to 0.2 Bq L�1) wereobserved. The variability of average OBT concentrations in plants inthe far field was even larger and ranged from 5.8 to 27.8 Bq L�1.Brown (1988) suggested the high atmospheric HTO (as related tosoil HTO) being the reason of the noted deviation of OBT/HTO ratiofrom the value prescribed by equilibrium considerations (Jean-Baptiste et al., 2011). There are many uses to this assumption. Forexample we can deploy it for correction of the Pointurier et al.(2003, 2004) dataset, which is based on HTO in precipitation andnot TFWT (plant HTO). To support the elevated OBT we back-calculated the (not measured) intermittently elevated air HTOand obtained the corrected OBT/HTO ratios in plant tissue: 1.10(n ¼ 70, Study A) and 2.06 (n ¼ 63, Study B). Regarding all othermeasurements, the generalized proposition of Brown (1988) mayhold and the highly variable ambient HTO load (air HTO) can be areason of variability seen in Table 2.

The other important comment pertains to the role of theexchangeable OBT (E-OBT) retained in some measurements andremoved in others. It can be shown, that notwithstanding the fact,that activity concentration in E-OBT can exceed that in non-exchangeable OBT by as much as 57% (Kim and Korolevych,2013), the mean OBT/HTO would remain not very sensitive to thepresence of E-OBT. The reason is in the labile E-OBT fluctuations inaccord with HTO, which results mostly in narrowing the range ofOBT/HTO and to a lesser extent in change to mean OBT/HTO value.

With respect to the constant HTO load causing near-equilibriumconditions, we present here results of a recent study of OBT/HTOratio carried out at Duke Swamp, a relatively large wetland on CRLproperty that contains elevated levels of tritium due to releasesfrom a nearby waste management area (Kim and Davis, 2008; Kim

et al., 2011). HTO and OBT concentrations were measured in a va-riety of natural vegetation at five locations in the swamp during the2005 growing season. The HTO concentrations varied spatially (by afactor of 3 over distances of a few tens of metres) and temporally(by a factor of 2 over the growing season), but conditions werelikely close to equilibrium locally. On a landscape scale quasi-equilibrium was found. The average OBT/HTO ratio was less thanone for all plant species with the exception of lichen. The meanratio over all samples in this quasi-equilibrium case was0.74 � 0.42. The constant HTO load can also be achieved incontrolled conditions (McFarlane, 1976; Garland and Ameen, 1979;Kim and Baumgärtner, 1994), inwhich the HTO concentration in airwas held constant over a period of a few weeks. The resulting meanOBT/HTO ratio of 0.70 � 0.12 was observed for the selected crops,much in-line with the Duke Swamp example and a general state-ment from Boyer et al. (2009) and Jean-Baptiste et al. (2011).

It should be reiterated here that the analysis has been performedseparately for a near field and for far field in order to evaluatewhichone appears closer to equilibrium. However, no significant differ-ence in the range of OBT/HTO ratios encountered in the far and inthe near field was observed. On the other hand, both the far and thenear field datasets appear clearly different from the equilibriumand quasi-equilibrium data. This implies an appreciable degree ofdynamics recorded in both datasets and a similarity in implicitdynamical processes.

3.2. CRL dataset of OBT/HTO ratios collected in 1995 and 2008e2011

Measurements at CRL were collected at the former Acid RainSite either when the plume was observed at the neighbouringgamma-radiation monitoring station, or routinely around noon inthe pre-harvest period. The measurements reflected numeroushigh OBT/HTO ratios (reaching 9.5 and higher) and one very highrecord of 88.78, which is not a result of an analytical error. Smallvalues approaching 0.03 were also recorded, but not frequentlyenough to have the distribution complete on the low end, with theresult being a mean of 3.31 and a standard deviation of 8.50. Table 3presents the summary of OBT/HTO ratios measured in differentyears of the CRL study.

Tritium measurements in rinsed and non-rinsed dried sampleswere compared in 15% of cases to assess the difference in tritiumactivity concentration due to exchangeable OBT. Differences in OBTactivity concentration in excess of 25% were not encountered andthe role of exchangeable OBT was not analyzed in this study anyfurther.

Regarding particularly large values of the OBT/HTO ratio, whichcontributed to the summary shown in Table 3, it is important toreiterate that an explanation of large numbers was one of the keyobjectives of the study and in 2008 many attempts were made tomeasure tritium in plants when the plume had just departed.Measurements were performed on a fine scale (half-hourly) andthis allowed recording of the essentially non-equilibrium phase ofplant HTO re-emission back to atmosphere. For this reason, thenumber of cases with OBT/HTO>>1 was disproportionally large. To

Fig. 3. Comparison of ranked observations of the OBT/HTO ratio from the referencedataset with ranked OBT/HTO ratio from CRL observations (qeq plot).

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168164

avoid this bias towards high values, random sampling was per-formed in 2009 and onwards. Persistently high ratios neverthelessreappeared. In 2009 and 2010, this was caused by a different reasone the NRU reactor shut-down ewhich drove the whole ecosystemout of equilibrium. In 2009, the consequence of source interruptionwas a much slower depletion of plant OBT as compared to plantHTO (and hence e persistence of high OBT/HTO ratios). In 2010, itwas a very low depleted HTO background to which plant HTOreverted each time the wind did not blow from the stack (againcausing the persistence of high OBT/HTO ratios).

3.3. Comparison of distributions of OBT/HTO ratios in the referenceand CRL datasets

The distribution of the OBT/HTO ratios in the CRL datasets wascompared to that of the reference dataset. The former was collectedat a single site, while the latter was comprised of experimentsperformed worldwide and thus has various factors of variabilityreflected in it (i.e. variability of ratios caused by the variability incultivar, climatic conditions, land use, etc). None of these factors ofvariability is present in the CRL dataset. Comparison, therefore,reveals the net effect of these types of variability. Since the twodatasets have similar distributions (clearly seen in the scatter plotin Fig. 3), or even probably populate the same distribution, it seems,that inter-species (cultivar) variability, variability due to differentclimatic conditions and due to different parameters (soil types, etc.)all have a relatively random nature and largely cancel out whenlumped together into a single eclectic reference dataset. What re-mains captured in both datasets is about three orders of variabilityin the magnitude of the OBT/HTO ratio, which is caused by thedynamical transient processes in ecosystems under non-equilibrium (e.g. by the differences between the uptake and re-emission phases at CRL).

3.4. Comparison of OBT/HTO ratios interpolated via modelling withratios from observations dataset of CRL

3.4.1. Example of reconstruction of plant HTO and modelling of OBT/HTO ratios

Fig. 4 shows a typical example of a single uptake edepurationsequence following a single plume arrival episode. The typicaltimeline of plant HTO and plant OBT concentrations are presentedin Fig. 4a. Fig. 4b provides the typical timeline of the OBT/HTO ratio,clearly showing that the depuration (re-emission) phase isresponsible for the high ratios. Fluctuations in the air HTO localbackground are particularly important, as the concentration of theplant HTO quickly follows that of the air HTO and resides in thedenominator. Low OBT/HTO ratios are seen forming in the uptakephase, when the plume is present.

3.4.2. Comparison of results of carbon-based model of OBTformation and HTO exchange with CRL observations

Auxiliarymodelling in this study has been used for interpolationof HTO and OBT measurements performed at certain moments oftime. The procedure is termed dynamical interpolation (as opposedto extrapolation/prediction) and is illustrated in Fig. 5, showing thedynamically interpolated (modelled) timeline of leaf HTO concen-tration and observations. The model was driven by the realisticsequence of plumes (continuous air HTO record) and reflectsnumerous feedbacks between the constituents of the water, energyand carbon cycles. The important feature of the model is thedynamical calculation of the exchange HTO flux and subsequentquantification of the HTO uptake and depuration rates on the basisof the carbon assimilation rate (via stomata opening) and plantgrowth. Following themethodology in Section 2.6.1, the continuous

record of air HTO concentration at the ARS garden plot wasreconstructed. The air HTO concentration timeline is not shownhere for the sake of brevity, but the adequate timing of the resultingleaf HTO is seen properly reconstructed in Fig. 5.

OBT/HTO ratios retrieved from the carbon-based model ofatmosphere-tritium exchange were ranked and pair-wisecompared with ranked OBT/HTO ratios observed in CRL field ex-periments. The resulting scatter-plot (known as a qeq plot tech-nique, Chambers et al., 1983) is shown in Fig. 6 and provides agraphical comparison of distributions these OBT/HTO ratios popu-late. When two distributions are placed on a single scatter-plot (qeq plot) a number of differences between important statistical in-dicators in datasets are revealed (Chambers et al., 1983). In our casethe presence of heavier tails (more frequently encountered extremevalues) either in observations or in modelling is worth noting(Fig. 6).

The primary purpose of the model was the reproduction of theuptake phase often missing in CRL observations, and subsequentreproduction of missing low ratios. The overall accuracy of themodel at the uptake phase (the lower left corner of Fig. 6) appearssatisfactory. Notwithstanding the fact that the model was validatedagainst CRL observations, the modelled OBT/HTO ratio clearly de-viates from observations during the period immediately afterplume departure (the middle part of the distribution in Fig. 6). Themodel contains approximately two times more sampling points forthe OBT/HTO ratio than the observations dataset in the vicinity of aunit ratio. This implies that the observed plant HTO concentrationdrops about two times faster than that in themodel. The other partsof the exchange process appear relatively close to observations.

Having compared and analyzed the modelled OBT/HTO ratioswith the observed ratios, we can proceed with averaging of thecollected ratios.

3.5. Generic OBT/HTO ratio

3.5.1. Quantification of generic OBT/HTO ratioThe averaging of OBT/HTO ratios from two observational and

one interpolated (modelled) datasets results in different values andconsiderable scatter in all cases. The summary is presented inTable 4.

Fig. 4. Example of a single uptake-depuration sequence corresponding to a realistic air HTO input: a e the timeline of HTO and OBT concentrations; b e the timeline of the OBT/HTOratio.

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168 165

The random sampling of the plant reflects sharp dynamicchanges inplant tissue freewater tritium (HTO),which resides in thedenominator of the analyzed ratio e these changes are reflected inall four cases. The comments in Table 4 are meant to caution fromdirectly using the presented numbers. For example, based on ourexperiencewith the CRL dataset,whereweknow that the lowendofthe distribution of the OBT/HTO ratios is missing, we can suggestthat the similar underrepresentation of the lowend is true about theReference Dataset. Representative sampling appears possible onlywhen the CRL observations dataset is extended andmade completewith the help of the modelled timeline of the OBT/HTO ratio.

3.5.2. Further comments on analyzed OBT/HTO ratiosThe wind blows toward a given sampling site for a relatively

small fraction of the time and this causes the deviation of arandomly sampled and measured OBT/HTO ratio from unity. Thiseffect equally holds true at sites near tritium sources and far away.It is worth noting, that remote sites have traditionally beenassumed to have constant background levels of tritium in theenvironment. Yet, concentrations at nominal background locationsvary by as much as a factor of five depending on which way the

Fig. 5. Dynamic interpolation (modelling, solid line) of leaf HTO observations at CRLARS (markers); for reference the set of observations is complemented with HTO intomatoes (circles) collected at times when leaf HTO was not available.

wind blows. This implies that OBT/HTO ratios measured at back-ground sites are subject to most of the same influences as those atnear-field sites and differ from one for the same reasons.

Two-thirds of the ratios measured close to local tritium sourcesshow values greater than one, with amean of 2.0� 4.2. Backgroundsites show a higher mean (6.9 � 16.1), although these results areinfluenced by the high ratios obtained from 2002 to 2005 in theOPG and Bruce Power REMP programs. The sites in the far field alsoshow a higher proportion (85%) of values greater than one, whichcould be an indication of far field being farther from equilibriumthan the near field. If this is the case, it could be due to a lessefficient mixing of secondary plumes of re-emitted HTO in the farfield. This possibility merits examination. It should also be noted

Fig. 6. QQ plot: Comparison of the ranked OBT/HTO ratios based on the results of thecarbon-based model with the ranked OBT/HTO ratios from the CRL observations.

Table 4Summary of OBT/HTO ratios.

Source Average � SD Range na Comments

ReferenceDataset

3.94 � 14.06 0.10e140.50 527 Low end of distributioncould be incomplete.High bias possibleb.

CRLDataset

3.31 � 8.50 0.38e88.78 220 Low end of distributionis known to be incomplete.Biased high.

Model 2.07 � 1.78 0.06e12.57 8000c Not biased. Distributionresembles that of observations.

a Total number of sampled OBT/HTO ratios.b Suggestion is based on the analysis of the modelled timeline of the OBT/HTO

ratio and the experience with the CRL dataset (see the comment pertaining to CRLdataset).

c The number is arbitrary - 8000 points were chosen from a continuous record tocover all parts of the dynamical exchange process with sufficient redundancy.

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168166

that it is difficult to measure low-level HTO and OBTconcentrationsin environmental samples precisely and the uncertainties in themeasurements are large. This contributes to the variability in themeasured ratios.

Finally, it shall be noted, that when local HTO concentrationapproaches the historical background values (ca 1 Bq L�1), temporalfluctuations due to redistribution of spatially inhomogeneous HTOby wind would dominate tritium from the remote anthropogenicsources and the proximity of the soil-plant system to quasi-equilibrium shall be reassessed.

4. Conclusions

Most high (and low) values of the OBT/HTO ratio arise simply asa result of normal plant sampling procedures. In the empirical datathere is neither an evidence of processes that result in the bio-accumulation of tritium in plants, nor any cause to doubt the val-idity of Specific Activity principles. However, in its present form, theSpecific Activity model based on these principles does not providean explanation for the variability in observed OBT/HTO ratios. Highvalues of the ratio could be associatedwith lowHTO concentrationsrather than with high OBT concentrations. Values much less thanone accompany the arrival of the plume, reside in a narrow intervaland therefore are rarely recorded in the field. The value of 0.74 forthe OBT/HTO ratio corresponds to equilibrium seen in the fieldunder special circumstances (e.g. at Duke Swamp of CRL) and isconsistently recorded only under laboratory conditions, where theair HTO concentration could be held constant.

It is confirmed, thatmeasurement of the OBT/HTO ratio at a singlesite results in ambiguity unless an extensive dataset representative ofall aspects of dynamically changing conditions is collected. Limita-tions intrinsic to experimental assessments and that of modelling inthe particular case of the OBT/HTO ratio assessments appear effi-ciently removed by the synergistic approach combining observationsand interpolation via modelling. As a result, the quantification of ageneric OBT/HTO ratio appears feasible, despite the large variability.An estimate close to 2.0 for the OBT/HTO ratio reflects conditionsencountered at CRL. The close proximity of the distribution of OBT/HTO ratios observed at CRL to that of the world-wide referencedataset led us to believe that the value of 2.0 also reflects the genericOBT/HTO ratio (mean, expected to be encountered elsewhere).

Acknowledgements

This study was conducted within the AECL R&D Project RD-1.5.5.1-4142 and was funded by COG WP-3072 and WP-3073. Sig-nificant parts of this paper have been taken directly from COG-06-3053-R1 (Kim and Davis, 2008) with the full consent of COG and in

compliance with COG’s procedures. The authors confirm that thismaterial has not previously been published in the open literature.The authors would like to acknowledge the help of Evan McNa-mara, Wayne Workman and Mike Bredlaw with tritium measure-ments; Oscar Petersons, Mark Ranjram, Evan Burgess, John Clarkand Grigoriy Kimaev with model development; and Bruce Reaviewith meteorological and radiation measurements and data pro-cessing. The help offered by Environment Canada (EC) and partic-ularly by prof. V. Arorawho provided the CTEMþCLASS code of EC isvery much appreciated.The authors also appreciate helpful com-ments of two anonymous reviewers.

Appendix A

Carbon-based model of tritium transfer (CTEM-CLASS-TT)

The role of modelling in this study is auxiliary and only a simpleinterpolation of HTO and OBT measurements has been required. Yet,the modelling of plant tritium (particularly OBT) in between obser-vations when many ambient drivers change simultaneously(including soil moisture and thermodynamics) presents a challenge(Korolevych and Kim, 2011). In this study we deploy the assumption,that tritium behaves as a simple tracer within the soil-plant system:this means that HTO simply follows (labels) the water transferredbetween the water-holding compartments of the soil-plant system,and that OBT simply follows (labels) the carbon cycle. The labelling,as traditionally performed in botany, is based on specific activity oftritium in each compartment and in each component of the waterand carbon cycles. The tritium transfer thus merely represents inexplicit form processes already implemented in so called land-surface schemes of the operational weather predicting modelsroutinely used around the world. These process-based land surfaceschemes usually have a carbon allocation algorithm. Inclusion of thelatter seems particularly beneficial for our purposes as it helpsbridging typically sparse OBT observations in plants.

The plant-soil model defining the tritium transfer, as proposedin Korolevych and Kim (2011) is Canadian LAnd Surface SchemeCLASS 2.7 (Verseghy, 1991, 2000; Verseghy et al., 1993; Kothavalaet al., 2005) complemented by the daily phenology (plant growth)offered by the Canadian Terrestrial Ecosystem Model CTEM-CLASS(Arora, 2003). The model of photosynthesis is based on the Farqu-har photochemistry (Farquhar et al., 1980; Collatz et al., 1991, 1992)and uses either BalleBerry (Ball et al., 1987) or Leuning (1995)coupling procedure linking photosynthesis, stomatal conductanceand components of energy balance (Arora, 2003).

The scheme of tritium transfer as per water and carbon routes ofCTEM-CLASS model (Verseghy, 1991, 2000; Verseghy et al., 1993;Arora, 2003) is presented in Figure A1. As noted, this study restson the assumption of tritium being amere tracer in plant tissues andsubsequently the only new process introduced for quantification ofairborne tritium transfer is its diffusion at the leaf and ground sur-face, parameterized by the exchange velocity. Surface diffusion dueto air-surface HTO concentration gradient governs HTO uptake intothe soil-plant system and HTO re-emission. It is assumed, that in allother respects HTO behaves as a traditional simple tracer: it arrivesin every water-holding compartment with the incoming water, be-comes diluted there as per the specific activity approach (stoichi-ometry) and then leaves with the departing water.

HTO allocated according to the size of the constituents of thewater cycle of CTEM-CLASS further enters carbohydrates formed inthe leaf via photosynthesis according to the current concentrationof HTO in the leaf. Total activity of so formed OBT in the newbiomass adds to total OBT inventory, which therefore changesdynamically. From this entry point, OBT follows the carbon cycle ofCTEM-CLASS and eventually exits the system with the respiration

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168 167

stream back in the form of HTO. It also remains in the OBT form inthe litter and organic matter of the soil, where it waits for microbesto decompose it into HTO and CO2 in years to come as parameter-ized by the heterotrophic respiration. The fate of OBT thus simplyduplicates the carbon cycle of CTEM-CLASS.

The uptake of air HTO in canopy is controlled by stomatalconductance, formation of OBT from HTO is based upon calculatedrate of photosynthesis, stoichiometry and isotopic discriminationfactor. OBT allocation directly corresponds to allocation of carbonwithin the five prognostic carbon pools of CTEM-CLASS (leaf, stem,root, litter and organic soil); this includes maintenance and respi-ration (Korolevych and Kim, 2011). HTO uptake at the ground surfaceis parameterized by the exchange velocity in dry conditions and bywet deposition with precipitation. HTO follows the plant respirationroutes and re-emits back to atmosphere via evapotranspiration.

Similarly to plant, the soil HTO is modelled as a simple tracer ofsoil moisture already calculated in CTEM-CLASS (Verseghy, 1991).The soil system of CTEM-CLASS comprises of three regular soillayers and a topmost litter and organic soil layer. In the formerthree the exchange of HTO is relatively slow due to their size, whilein the latter one (the topmost) the exchange of tritium is fast due todirect exchange with rapidly varying atmospheric HTO via wetdeposition and dry exchange flux and a relatively small size of acompartment. HTO enters soil compartments with infiltratingwater, root autotrophic respiration flux and heterotrophic respira-tion of organic matter. Independent diffusion of tritiumwithin andbetween soil layers is presently not taken into account, whichmeans no other exchange of HTO between layers except notedabove and also the complete mixing of soil water infiltrated at thesoil top throughout the soil layer during each time step. HTO leavessoil compartments with infiltrating water, with drainage and withevapotranspiration flux, all of them having the average HTO con-centration of the corresponding soil layer at the current time step.

Like in all other models of this class, CTEM-CLASS predictionsreflect site specifics through local parameterization of hydrologicalcycle, winds (driving plume concentration and occurrence fre-quency), precipitation, and other meteorological drivers, and sodoes the tracing tritium. Amongmany distinctive features of CTEM-CLASS described elsewhere (Arora, 2003) it worth noting, that theplant body mass production and leaf area index in CTEM-CLASS areprognostic rather than prescribed variables and this governs thedynamically changing OBT production rate.

The proposed approach is a manageably detailed research tool.However, notwithstanding the conceptual simplicity of handlingHTO as a tracer in CTEM-CLASS, the inventory of constituentcomponents of water cycle appears large. Subsequently thedetailed explanation of tritium transfer and presentation of relatedimportant parameters of CTEM-CLASS appears exceedingly longand beyond the scope of this study and as such is postponed untilthe next publication.

Fig. A1. Block scheme of tritium transfer as defined by the water and carbon cycles ofthe CTEM-CLASS model

References

Arora, V.K., 2003. Simulating energy and carbon fluxes over winter wheat usingcoupled land surface and terrestrial ecosystem models. Agric. For. Meteorol. 118(1e2), 21e47.

ATSDR, 2002. Health Consultation-LLNL. http://www.atsdr.cdc.gov/HAC/PHA/livermore2/liv_p1.html.

Baglan, N., Alanic, G., Pointurier, F., 2005. Tritium determination at trace level:which strategy to determine accurately HTO and OBT in environmental sam-ples? Fusion Sci. Technol. 48, 749e754.

Ball, J.T., Woodrow, I.E., Berry, J., 1987. A model predicting stomatal conductance andits contribution to the control of photosynthesis under different environmentalconditions. Prog. Photosynth. Res. 4, 221.

Baumgärtner, F., Kardinal, Ch., Mullen, G., 2001. Distribution of tritium betweenwater and exchangeable hydrogen bridges of biomolecules. J. Radioanal. Nucl.Chem. 249, 513e517.

Baumgärtner, F., 2005. Accumulative tritium transfer from water into biosystems.Fusion Sci. Technol. 48, 787e790.

Baumgärtner, F., Donhärl, W., 2004. Non-exchangeable organically bound tritium(OBT): its real nature. Anal. Bioanal. Chem. 379, 204e209.

Baumgärtner, F., Kim, M.-A., 2000. Tritium/protium fractionation near and insideDNA. J. Radioanal. Nucl. Chem. 243, 295e298.

Belot, Y., 1986. Tritium in plants: a review. Radiat. Prot. Dosim. 16, 101e105.Bogen, D.C., Welford, G.A., 1976. “Fallout tritium” distribution in the environment.

Health Phys. 30, 203e208.Bogen, D.C., Welford, G.A., White, C.G., 1979. Tritium Distribution in Man and His

Environment, pp. 567e574. IAEA-SM-232/75.Boyer, C., Vichot, L., Fromm, M., Losset, Y., Tatin-Froux, F., Guétat, P., Badot, P.M.,

2009. Tritium in plants: a review of current knowledge. Environ. Exp. Botany 67(1), 34e51.

Brown, R.M., 1988. Assessment of the Significance of Organically Bound Tritium inEnvironmental Samples. Atomic Energy Control Board. INFO-0283.

Brown, R.M., 1995. The Measurement of Tritium in Canadian Food Items. AtomicEnergy Control Board. INFO-0499.

Bruce Power, 2003. Annual Summary and Assessment of Environmental Radio-logical Data for 2002. B-REP-03419-00003-R00.

Chambers, J.M., Cleveland, W.S., Kleiner, B., 1983. Graphical Methods for DataAnalysis. Wadsworth Statistics/probability Series.

Clemente, G.F., Belloni, P., Di Pietro, S., Santori, G., 1979. Tritium and PlutoniumContent of the Italian Diet and Transfer to Man, pp. 257e263. IAEA-SM-237/64.

CNSC, 2011. Tritium Studies Project Synthesis Report. Published by the CanadianNuclear Safety Commission (CNSC). Catalogue number INFO-0800 Revision 1.http://nuclearsafety.gc.ca/eng/pdfs/Reports/CNSC_Tritium_Studies_Project_Synthesis_Report_e.pdf.

Collatz, G.J., Ball, J.T., Grivet, C., Berry, J.A., 1991. Physiological and environmentalregulation of stomatal conductance, photosynthesis and transpiration: a modelthat includes a laminar boundary layer. Agric. For. Meteorol. 54 (2), 107e136.

Collatz, G.J., Ribas-Carbo, M., Berry, J.A., 1992. Coupled photosynthesis-stomatalconductance model for leaves of C4 plants. Funct. Plant Biol. 19 (5), 519e538.

CSA (Canadian Standard Association), 1987. Guideline for Calculating DerivedRelease Limits for Radioactive Material in Airborne and Liquid Effluents forNormal Operations of Nuclear Facilities. CAN/CSA-N288.1eM87. Rexdale,Ontario, Canada.

Davis, P.A., 2007. HTO and OBT in Grass Samples Collected at CRL during 1995.Private communication.

Diabate, S., Strack, S., 1993. Organically bound tritium. Health Phys. 65 (6), 698e712.Environmental Agency, 2001. Potential for Bio-accumulation of Organically Bound

Tritium in the Environment: Review of Monitoring Data. National ComplianceAssessment Service. Technical Report, NCAS/TR/2000/026.

Evans, A.G., 1969. New dose estimates from chronic tritium exposures. Health Phys.16, 57e63.

Farquhar, G.D., von Caemmerer, S.V., Berry, J.A., 1980. A biochemical model ofphotosynthetic CO2 assimilation in leaves of C3 species. Planta 149 (1), 78e90.

Fuma, S., Inoue, Y., 1995. Simplified and sensitive analysis of organically boundtritium in tree rings to retrospect environmental tritium levels. Appl. Radiat.Isot. 46, 991e997.

Garland, J.A., Ameen, M., 1979. Incorporation of tritium in grain plants. Health Phys.36, 35e38.

Golubev, A.V., Belovodsky, L.F., Golubeva, V.N., Kosheleva, T.A., Kuznetsova, V.F.,Mavrin, S.V., Surano, K., Hoppes, W., 2002. Application of lichens for assessmentof atmospheric pollution by tritium. Fusion Sci. Technol. 41, 409e412.

Gulden, W., Raskob, W., 1992. Accidental tritium doses based on realistic modelling.Fusion Technol. 21, 536e543.

Hisamatsu, S., Hashimoto, T., Takizawa, Y., 1990. Tritium concentration in someEuropean foods. J. Radioanal. Nucl. Chem. Lett. 146, 245e253.

Hisamatsu, S., Inoue, Y., Takizawa, Y., 1989. Tritium concentration in some Europeanfoods. J. Environ. Radioact. 10, 251e255.

Hisamatsu, S., Katsumata, T., Takizawa, Y., 1992. tritium concentration in Akita CityDiet. Health Phys. 63, 393e397.

Hisamatsu, S., Takizawa, Y., Abe, T., Katsumata, T., 1987. Fallout 3H ingestion in Akita,Japan. Health Phys. 53, 287e293.

IAEA, 2008. Modelling the Environmental Transfer of Tritium and Carbon-14 toBiota and Man. Report of the Tritium and Carbon-14 Working Group of EMRASTheme 1. Environmental Modelling for RAdiation Safety (EMRAS) Programme.

V.Y. Korolevych et al. / Journal of Environmental Radioactivity 129 (2014) 157e168168

International Atomic Energy Agency (IAEA), Vienna. http://www-ns.iaea.org/downloads/rw/projects/emras/draft-final-reports/emras-tritium-wg.pdf.

Inoue, Y., Iwakura, T., 1990. Tritium concentration in Japanese Rice. J. Radiat. Res. 31,311e323.

Inoue, Y., Tanaka-Miyamoto, K., Iwakura, T., 1992. Uptake of atmospheric tritium bymarket foods. Fusion Technol. 21, 494e499.

Jean-Baptiste, P., Baumier, D., Fourre, E., Dapoigny, A., Clavel, B., 2007. The distri-bution of tritium in the terrestrial and aquatic environments of the Creys-Malville nuclear power plant (2002e2005). J. Environ. Radioact. 94 (2), 107e118.

Jean-Baptiste, P., Fourré, E., Baumier, D., Dapoigny, A., 2011. Environmental OBT/TFWT ratios revisited. Fusion Sci. Technol. 60 (4), 1248e1251.

Kim, C.K., Han, M.J., 1999. Dose assessment and behaviour of tritium in environ-mental samples around Wolsong nuclear power plant. Appl. Radiat. Isot. 50,783e791.

Kim, C.K., Lee, S.K., Rho, B.H., Lee, Y.G., 2000. Environmental distribution andbehavior of 3H and 14C around Wolsong nuclear power plants. Health Phys. 78,693e699.

Kim, M.-A., Baumgärtner, F., 1994. Equilibrium and non-equilibrium partition oftritium between organics and tissue water of different biological systems. Appl.Radiat. Isot. 45, 353e360.

Kim, S.B., Korolevych, V., 2013. Quantification of exchangeable and non-exchangeable organically bound tritium (OBT) in vegetation. J. Environ. Radi-oact. 118, 9e14.

Kim, S.B., Davis, P.A., 2008. OBT/HTO Ratios in Plants. CANDU Owners Group, Tor-onto. Report COG-06-3053-R1.

Kim, S.B., Chouhan, S.L., Davis, P.A., 2011. Observed and modelled tritium in thewetland ecosystem in Duke swamp near a nuclear waste management area.Fusion Sci. Technol. 60 (3), 960e963.

Korolevych, V.Y., 2013. Generic OBT/HTO Ratio in Agricultural Produce Subject toRoutine Atmospheric Releases of Tritium. CANDU Owners Group, Toronto.Report COG-11e3059.

Korolevych, V.Y., Kim, S.B., 2011. Modelling and validation of OBT formation in to-mato and potato plants. Fusion Sci. Technol. 60 (4), 1288e1291.

Korolevych, V.Y., Kim, S.B., 2013. Relation between the tritium in continuous at-mospheric release and the tritium contents of fruits and tubers. J. Environ.Radioact. 118, 113e120.

Kothavala, Z., Arain, M.A., Black, T.A., Verseghy, D., 2005. The simulation of energy,water vapor and carbon dioxide fluxes over common crops by the CanadianLand Surface Scheme (CLASS). Agric. For. Meteorol. 133 (1), 89e108.

Kotzer, T., Trivedi, A., 2001. Dosimetric implications of atmospheric dispersalof tritium near a heavy-water research reactor facility. Radiat. Prot. Dosim. 93,61e66.

Leuning, R., 1995. A critical appraisal of a combined stomatal-photosynthesis modelfor C3 plants. Plant, Cell. Environ. 18 (4), 339e355.

McFarlane, J.C., 1976. Tritium fractionation in plants. Environ. Exp. Bot. 16, 9e14.

Momoshima, N., Kakiuchi, Okai, T., Hisamatsu, S., Maeda, Y., 2000. Tritium in pineforest ecosystem: relation between fresh pine needles, organic materials on aforest floor and atmosphere. J. Radioanal. Nucl. Chem. 243, 474e482.

Momoshima, N., Okai, T., Kaji, T., Takashima, Y., 1991. Distribution and trans-formation of various chemical forms of tritium in the environment. Radiochim.Acta 54, 129.

Murphy, C.E., 1993. Tritium transport and cycling in the environment. Health Phys.65 (6), 683e697.

NCRP (National Council on Radiation Measurement), 1978. Tritium in the Envi-ronment. NCRP Report No. 62, Bethesda, Maryland, U.S.A.

Okada, S., Momoshima, N., 1993. Overview of tritium: characteristics, sources, andproblems. Health Phys. 65 (6), 595.

Ontario Power Generation, 1999. Annual Summary and Assessment of Environ-mental Radiological Data for 1998. OPG report, N-REP-03419e0597 R00.

Ontario Power Generation, 2002. Annual Summary and Assessment of Environ-mental Radiological Data for 2001. OPG report, B-REP-03419e00002.

Ontario Power Generation, 2004. Annual Summary and Assessment of Environ-mental Radiological Data for 2003. OPG report, N-REP-03481e10002.

Ontario Power Generation, 2005. Annual Summary and Assessment of Environ-mental Radiological Data for 2004. OPG report, N-REP-03481e10003.

Peterson, S.-R., Davis, P.A., 2002. Tritium doses from chronic atmospheric releases: anew approach proposed for regulatory compliance. Health Phys. 82, 213e225.

Pointurier, F., Baglan, N., Alanic, G., 2004. A method for the determination of low-level organic-bound tritium activities in environmental samples. Appl. Radiat.Isot. 61, 293e298.

Pointurier, F., Baglan, N., Alanic, G., Chiappini, R., 2003. Determination of organicallybound tritium background level in biological samples from a wide area in thesouth-west of France. J. Environ. Radioact. 68, 171e189.

Strack, S., Diabate, S., Raskob, W., 2005. Organically bound tritium in Plants: Insightgained by long-term experience in experimental and modelling research.Fusion Sci. Technol. 48, 767e770.

Strack, S., Konig, L.A., 1981. Determination of Organically Bound Tritium in Envi-ronmental Samples by Application of the Oxidizing Plasma Technique. Ker-nforschungszentrum Karlsruhe. KfK 3249.

Takashima, Y., Momoshima, N., Inoue, M., Nakamura, Y., 1987. Tritium in Pine Needlesand its significant sources in the environment. Int. J. Radiat. Isot. 38, 255e261.

Verseghy, D.L., 1991. CLASSdA Canadian land surface scheme for GCMs. I. Soilmodel. Int. J. Climatol. 11 (2), 111e133.

Verseghy, D.L., 2000. The Canadian land surface scheme (CLASS): Its history andfuture. Atmos. Ocean 38 (1), 1e13.

Verseghy, D.L., McFarlane, N.A., Lazare, M., 1993. CLASSdA Canadian land surfacescheme for GCMs, II. Vegetation model and coupled runs. Int. J. Climatol. 13 (4),347e370.

Workman, W.J.G., Kim, S.B., Kotzer, T.G., 2005. Interlaboratory comparison oforganically bound tritium measurements in environmental samples. Fusion Sci.Technol. 48, 763e766.