Planta (2001) 214: 220±234 DOI 10.1007/s004250100602 O R I GI N A L A R T IC L E J.P. Comstock Steady-state isotopic fractionation in branched pathways using plant uptake of NO3± as an example Received: 26 October 2000 / Accepted: 13 February 2001 / Published online: 11 August 2001 Ó Springer-Verlag 2001 Abstract A steady-state isotopic discrimination model is developed for material transfer between a single source and two distinct sinks arising from an internal branching of the uptake pathway. Previous analyses of isotopic discrimination in multistep processes are extended to include the eects of the interacting sinks. The theory is ®rst developed as a set of generic expressions allowing for ¯exibility in the de®nition of intermediate pools in all parts of the branched transport pathways, and then applied to a case study of nitrate uptake by plants. The isotopic composition of assimilated nitrate may be evaluated with the model for either contrasting root versus shoot assimilate pools, each with a unique isotopic signature, or as a single mean value for wholeplant nitrate reduction. The theory is further developed to indicate how isotopic measurements may be used to infer (i) eux:in¯ux ratios at the root plasma membrane, (ii) partitioning of assimilate capture between root and shoot reduction sites, and (iii) mixing of root and shoot assimilate pools in sink tissues due to whole-plant circulation of organic nitrogen in both xylem and phloem. Keywords Eux:in¯ux ratio á Isotopic discrimination á Isotopic heterogeneity á Nitrate assimilation á Nitrogen partitioning á Steady-state fractionation Introduction A concise theory of isotopic fractionation during carbon capture by plants has been of tremendous value in understanding carbon dynamics in physiological, J.P. Comstock Cornell and Boyce Thompson Institute Stable Isotope Laboratory (CoBSIL), Boyce Thompson Institute for Plant Research, Tower Road, Ithaca, NY 14853, USA E-mail: [email protected] Fax: +1-607-2541242 ecological, and ecosystem studies, but a similar theoretical grounding has not yet been developed with regards to nitrogen assimilation (Handley et al. 1998). This has been the case in part because plants can utilize multiple forms of nitrogen in the environment, and no single uptake process may fully account for the isotopic compositions observed. Also important, however, is the fact that the physiological processes of nitrogen assimilation within the plant have a more complex pattern of internal compartmentation than is the case with carbon. In the case of nitrate, the single most important form of mineral nitrogen for most plants, compartmentation includes uptake by the root system and an internal branching of the nitrate ¯ux between separate assimilation sites in the root and shoot tissues (Fig. 1, sinks 1 and 2). The branching of the nitrate assimilation pathway at the root cytoplasm with partial reduction of nitrate to organic products and partial transfer of unreacted nitrate to the shoot creates a single uptake process with intermediate pools distributed among multiple organs within the plant. The need to account for the integrated dynamics of such a branched uptake pathway has prevented direct transfer of algorithms developed to describe carbon isotope discrimination to the understanding of nitrogen isotopic behavior in plants. Models have been presented for uptake to a single sink (Mariotti et al. 1982; Shearer et al. 1991), but these have not been expanded to include the multiple sinks found in most plants. Robinson et al. (1998) presented a compartment-based process model for numerical solution of the ¯uxes controlling nitrogen isotopic composition in plants, but simple expressions relating isotopic behavior to uptake characteristics are still lacking. A theoretical analysis of isotopic fractionation during transfer from a single source to multiple sinks by a branched uptake pathway is presented below. The ®rst half of this paper provides a general treatment to describe the isotopic fractionation in a bifurcated pathway. The theory is then speci®cally applied to the model of nitrate uptake for plants growing under pure nitrate 221 d Rsample Rstd Rstd 1 where Rstd refers to a known reference material. If an isotope eect is present, then the product of a reaction takes on a dierent isotopic abundance ratio than the reactant. The magnitude of the isotope eect is described by the fractionation factor (a): a Rreactant Rproduct 2 It is often useful to de®ne discrimination as the deviation of a from 1. In the treatments below, the value a±1 associated with a unidirectional ¯ux (i.e. single transport step or discreet reaction) is a constant designated by a lower-case letter (a±z), while discrimination by an entire process is designated by the Greek letter D. Despite similar de®nitions, these two variable types have an important distinction in that D is not a constant, but varies with ¯ux dynamics. Note that Farquhar et al. (1989) is followed in omitting the factor of 1,000 associated with units of &(per million) from the mathematical de®nitions of D and discrimination constants. Fig. 1 Modeled ¯ow-chart of nitrogen exchanges during the uptake of nitrate from a source pool in the soil (ellipse) and reductive assimilation into organic sink pools (trapezoids) in both root and leaf tissues. Intermediate nitrate pools in various tissues (boxes), are all considered to be in a steady state with respect to both total mass and isotopic composition. Polygons represent the bulk organic nitrogen of root and shoot tissues. In all cases, the pool number is the same as the subscripted index of R, the isotopic abundance ratio of each respective pool. Arrows refer to total unidirectional ¯uxes of both isotopes, with magnitude Fi. Lowercase letters refer to discrimination constants nutrition, and allows prediction of the isotopic ratios of the assimilatory sinks as well as intermediate pools (Fig. 1, R1±R5). The predicted values of the isotopic composition of the immediate assimilatory products also permits an estimate of the mixing of root and shoot assimilation in the bulk organic nitrogen of various tissues. The speci®c application of the model to nitrate assimilation makes the most parsimonious interpretation possible of the number of subpools within the plant required to explain the isotopic behavior. The theoretical background is presented with sucient generality, however, to facilitate adjustment of the discrimination model for dierent hypotheses of internal compartmentation, or variation in compartmentation among dierent plants. General discrimination theory Terminology and basic concepts Variation in isotopic composition is characterized by the isotopic abundance ratio, R, de®ned as [heavy]/[light] isotopes in a sample. R is often reported using d notation (Farquhar et al. 1989; see Table 1 for detailed de®nitions): Assumptions and applicability The basic question is: ``How does variation in discrimination constants among dierent steps of multi-component processes interact with ¯ux dynamics to generate variable discrimination (d source:sink Rproduct =Rsource 1) in the overall movement from source to ®nal sink(s) (Figs. 2, 3, ellipses and trapezoids, respectively)?'' The processes are modeled as series of pools, with gross, unidirectional ¯uxes passing in both directions between consecutive pools (Fig. 2). Each unidirectional ¯ux (Fig. 2, arrows) is associated with a discrimination constant (e.g. a=aa±1). The assumption of a steady state applies to both total mass and isotopic composition of any intermediate pools (Fig. 2, boxes), which are considered to be internally well mixed with respect to the isotopic eects of all ¯uxes. A net ¯ux occurs only from the initial or source pool (Fig. 2, ellipses), and into the ®nal sink (Fig. 2, trapezoid). It is also assumed that, at steady state, {Ri} of intermediate pools may have any needed values due to previous, transient enrichment or depletion, but is then fully stable at the predicted steadystate solution. The mass balance of transient conditions leading to this steady state is not treated in the current analyses. R0 is assumed to have a stable value, but the mass balance of the source pool is not incorporated into the following models. It is thus assumed either that the source pool size is very large relative to the ¯uxes modeled, such that it behaves as an `in®nite source' over a limited time interval, or that other, unlabeled processes actually balance the ¯uxes described such that R0 is really in a steady state like other modeled pools. The predicted {Ri} for all intermediate pools, as well as the 222 Table 1 List of variables used in the text Variables De®nitions Ri Ratio of heavy to light isotopes in a given pool. Subscripts 1, 2, 3... used with R are the same as the pool numbers, and pools are shown as closed ®gures: circles, squares, and trapezoids, in the ®gures. Pool numbers may sometimes be replaced with speci®c names such as source or sink. In a few cases the subscript net is used to indicate R of material transferred in a net ¯ux rather than resident in a speci®c pool Isotopic composition of a given sample expressed relative to an arbitrary standard (see Eq. 1 for de®nition). Subscript i, when used, refers to same pools as Ri Concentration of molecule of interest in dierent pools. Subscripts match Ri to designate speci®c pools The gross, unidirectional ¯ux indicated by an arrow between two pools in any of the ®gures. Subscripts 1, 2, 3...refer to the respective ¯ux arrows in the ®gure under discussion. The subscript net refers to the dierence between opposing gross ¯uxes between the same two pools Fractionation factor describing the isotope eect inherent in a single-step unidirectional process. It is de®ned as Rreactant/Rproduct. Subscripts a, b, c..., when used, refer to fractionation in distinct steps. Most rigorously, this implies a unique a for each gross ¯ux arrow in a ®gure (e.g. Fig. 1, pathway D), but often the forward and backward transfers between adjacent pools are taken to have the same a (e.g. Fig. 1a±c). Discrimination constants of a single unidirectional transfer or reaction de®ned as the deviation of a from unity (i.e. a±1). The symbols match the subscripts of the a notation when referring to the same steps Discrimination, as de®ned above for a, b, c¼, only referring to a process with at least one reversible step in which component discrimination constants may have variable expression dependent on ¯ux dynamics. De®ned as Rsource/Rsink±1, where subscripts i and j refer to source and sink pools, respectively (e.g. ®gures). When numeric subscripts are used, they match the subscripts of Ri in designating the same pools. Several speci®c formulae are developed in the text to de®ne D in terms of component ¯uxes and discrimination constants for dierent transport pathways A shorthand notation referring to a speci®c formula previously de®ned for D in unbranched chains of pools, but used as part of a more complex formula for branched pathways in which H(i:j) does not de®ne the full isotopic relationship between pools i and j Branchpoint Interaction Factor. Subscripts i and j refer to the pool numbers of the source and branchpoint, respectively. B(i...j) de®nes the probablitity (0±1) that a heavy isotope excluded from the branch to sink 1 by discrimination processes, will end up in sink 2 as opposed to euxing back to the source. B(i...j) depends on the number of distinct pools and their ¯ux relationships, but not on the values of any discrimination constants Basal Cross-Product Factor de®ned by discrimination constants and ¯uxes in shared portion of a branched pathway, and used in the most precise formulation of D (Eq. B6). Subscripts i and j refer to source and branchpoint, respectively. See Eq. B7 and Table 2 for numeric calculation of F(i...j). The cross-product eects of the discrimination constants are minor compared to the direct additive eects, and F(i...j) is deleted from the generally more useful approximation in Eq. 11 in the main text `Eux product'/`in¯ux product'. The eux product calculated for each pool in a pathway is the product of the chain of unidirectional ¯uxes leading back from that pool to the original source pool of the material transfer. The in¯ux product is the product of all unidirectional ¯uxes leading inward from the source to the designated pool. The value of EP:IP is always between 0 and 1, and its contrasting values for each pool in a pathway determines the relative expression of all component discrimination constants in D (e.g. Eq. 7). di Ci Fi ai a, b, c¼ D(i:j) H(i:j) B(i¼j) F(i¼j) EP:IP sink, is inherent in the model solutions based on massbalance analyses. Discrimination in an unbranched pathway The simplest ¯ow pathway has just source and sink with no intermediates (Fig. 2, pathways A.1 and A.2). The derivation of D(0:1) for this system begins with the massbalance relationship between net transfer and component ¯uxes: F net F 1 F2 3 where `Fi' are total ¯uxes including both heavy and light isotopes and subscripts refer to Fig. 2, pathway A.1. This system must exhibit not only conservation of total mass, but also conservation of overall isotopic abundance (Hayes 1982) expressed by multiplying each of the ¯uxes from Eq. 3 by its expected isotopic ratio (Fig. 2, pathway A.2): R0 R1 Fnet R1 F1 F2 4 1a 1a R0 and R1 are the isotopic ratios of source and sink, respectively, and `a' is the discrimination associated with the transfer in either direction. The left-hand term is simply net ¯ux times Rsink. The isotopic compositions of the individual ¯uxes F1 and F2, however, are subtler. They are calculated using Eq. 2 with aa equal to 1+a, and Rsource being that of the originating pools (i.e. R0 and R1 for F1 and F2, respectively) (Fig. 2, pathway A.2). To derive the expression for D0:1, Eqs. 3 and 4 are solved simultaneously by ®rst substituting F2±F1 for Fnet in Eq. 4, and then collecting all terms for R0 and R1: R1 F1 F2 F2 1a R0 F1 1a 5 Algebraic rearrangement and the de®nition of D(0:1) as Rsource/Rproduct±1 leads directly to: D 0:1 R0 R1 1a 1 F2 F1 6 Discrimination by the whole pathway has a maximum value of a, the discrimination inherent in unidirectional 223 Fig. 2 Linear unbranched pathways. Shapes representing pools and all symbols have the same meanings as in Fig. 1. A.1 Simplest possible pathway with only source (ellipse) and sink (trapezoid) pools. A.2 The same as A.1, but indicating the isotopic composition of the instantaneous ¯uxes F1 and F2. The solid-edged ellipse and trapezoid labeled R0 and R1 represent distinct pools, while the boxes with dashed outlines are just mathematical concepts of the isotopic composition of the instantaneous ¯uxes without any necessary assumption that they exist as separate, discreet pools. B Similar to A.1, but with an inde®nite number of potential intermediate pools (square rectangles). C The same as B, but with ¯uxes all determined by pool sizes and ®rst-order rate constants. D Same as B only with dierent fractionation constants for forward and backward reactions at each transfer step transfer. This is fairly intuitive since the inward ¯ux contributing to R1 is isotopically lighter than R0 to a degree determined by `a', but the back-reaction discriminates in the same manner, leaving the heavy isotopes behind to enrich R1. If F2=F1, the eects cancel, and D=0. Note that Eq. 6 (as well as all subsequent algorithms for D) could be rewritten using d notation as: D 0:1 d0 d1 d0 1 d1 d1 a 1 F2 F1 6b In all subsequent derivations, we begin as illustrated above with two equations to describe (i) conservation of total mass and (ii) isotopic conservation during steady state, and solve them jointly. Expressions analogous to Eq. 6, but for linear pathways with one or two intermediate pools, are derived in Appendix AI, and it is concluded by induction that D for a pathway with n pools beyond the source (Fig. 2, pathway B) is: Fig. 3 A pathway with one source, two potential sinks, and a branchpoint. Symbols labeling individual ¯uxes and pools have the same de®nitions as in Fig. 1 D 0:n a 1 F2 F2 F2 F4 F2 F4 F2 F4 F6 b c F1 F1 F1 F3 F1 F3 F1 F3 F5 F2 F4 F6 . . . F2 n 1 F2 F4 F6 . . . F2n ...z F1 F3 F5 . . . F2 n 1 1 F1 F3 F5 . . . F2n 1 7 Expressions related to Eq. 7 can be found in several places in the literature (Farquhar et al. 1982 , 1989; Handley and Raven 1992). The form given here is highly generic. Some important points can be made about Eq. 7. The equation for D(source:sink) across a linear string of pools is the set of associated discrimination constants weighted by a predictable series of progressive `eux product'/'in¯ux product' (EP:IP). For any transfer step in the chain, full expression of its particular discrimination constant in D(source:sink) requires, ®rst of all, that the EP:IP of the immediate source pool in the transfer must be near 1. Further, if the EP:IP calculated for the recipient pool of the same transfer is also near 1, then the potential discrimination of that step cancels out (see explanation of Eq. 6). A general way to think about this is that, whenever the EP:IP of a pool approaches 1, bidirectional mixing, including any 224 pools between it and the source, is the dominant process, and the pool approaches equilibrium with R0. This behavior is also consistent with the observation that full expression of the dierent discrimination constants for dierent steps in a pathway is mutually exclusive, and irreversible steps suppress all subsequent potential for discrimination in the ®nal product (O'Leary 1981). Also implicit in the derivation of D is the {Ri} for all intermediate pools. For illustration, consider a linear pathway with a source (R0), two intermediate pools (R1 and R2) and a ®nal sink (R3) (i.e. Fig. 2, pathway B, with R3 taken as the terminal sink). From known {'ai±1'} and {Fi}, the value of R3 is calculated as R0/(1+D0:3) using Eq. 7. R1 and R2 are then calculated relative to R3, treating each respective intermediate as the `source pool' while employing Eq. 7. This is consistent with earlier comments that R0 might itself represent a steady-state intermediate in a larger process. We can thus use appropriate forms of Eq. 7 to work backwards, giving R3á(1+D1:3) and R3á(1+D2:3), for R1 and R2, respectively (i.e. R3á(1+ b(1±F4/F3)+c(F4/F3±F4F6/F3F5)) and R3á (1+c(1±F6/ F5), respectively). Equation 7 is the most general form, developed in regard to the gross ¯uxes. However, if (i) ¯uxes are dependent on the concentrations of the respective pools, (ii) the forward and backward rate constants at a given transfer step are equal, and, ®nally, (iii) all the rate constants are ®rst order (see Fig. 2, pathway C), then Eq. 7 reduces to an important special case: C0 C 1 C1 C2 D 0:n a b C0 C0 C n 1 Cn C2 C3 c ...z 8 C0 C0 In familiar contexts such as a diusion gradient, the discrimination constants most expressed by the overall pathway will be those associated with the largest concentration drops (i.e. most limiting resistances to ¯ux) (Farquhar et al. 1982). Finally, in many processes the discrimination constants may not be equal for the forward and backward ¯uxes between two contiguous pools (Fig. 2, pathway D). The analogue to Eq. 7 developed with unique discrimination constants for each unidirectional ¯ux is: aa F2 0 aa F 2 0 aa ab F2 F4 D 0:n a a b b aa0 F1 aa0 F1 aa0 ab0 F1 F3 aa ab F2 F4 a a a F F 0 a b c 2 4 F6 c c aa0 ab0 F1 F3 aa0 ab0 ac0 F1 F3 F5 aa ab F2 F4 ::::F2 n 1 ::::: z aa0 ab0 F1 F3 :::F 2 n 1 1 aa ab :::az F2 F4 :::F2n 9 z0 aa0 ab0 :::az0 F1 F3 :::F2 n 1 where aa and aa' etc. are 1+a and 1+a', respectively. Because they are close to 1, these extra factors have only a small in¯uence on the ®nal value of D. Omitting them is often acceptable (see `model validation' below) and leads to: F2 F2 F2 F4 D 0:n a a0 b0 b F1 F1 F1 F3 F2 F4 F2 F4 F6 c c0 F1 F3 F1 F3 F5 F2 F4 :::F2 n 1 F2 F4 :::F2n 10 ::::: z z0 F1 F3 :::F 2 n 1 1 F1 F3 :::F2 n 1 Despite the similarity to Eq. 7, the forward and backwards discriminations at a given transfer step need not cancel out when EP:IP equals 1. This could represent equilibrium rather than kinetic isotope eects, or any real system where forward and backward ¯ux arrows represent inherently dierent processes (e.g. C4 photosynthesis with a 4-carbon pump into the bundle sheath and a leakage of CO2 back-diusing). Equations 1±10 summarize a very general treatment of fractionation in unbranched pathways, but do not represent radically new expressions for fractionation. For example, the well-known models of discrimination during C3 photosynthesis (Farquhar et al. 1982, 1989) are equivalent to Eqs. 7 and 8, while the commonly used expressions for C4 photosynthesis (Farquhar 1983) and during the `leaky pump' assimilation of HCO3± by algae (Sharkey and Berry 1985) are closely related to Eqs. 9 and 10. The same approaches can be extended to include novel expressions for discrimination in branched pathways. Branched pathways with multiple sinks Consider a generic pathway as before, only with a branch point at some intermediate pool and then separate branches leading to two distinct sinks (Fig. 3). The derivation for either sink alone is developed in Appendix B culminating in Eq. A.II.6. An often more useful, very close approximation to the full equation is: D source:sink1 H source:sink1 B source:::bp D bp:sink1 D bp:sink2 11 where bp stands for branchpoint (e.g., pool 4 in Fig. 3), B() refers to a branchpoint interaction factor with a range of possible values from zero to one (Table 2), and H and D both represent a similar algebraic shorthand to be expanded for calculation by application of Eqs. 7 or 9 (D for an unbranched pathway) with source:sink de®ned by the respective subscripts. The distinction of the two symbols, H and D, is conceptual in that, where D is used, the expression is not only part of the needed total equation, but also expresses a necessarily true isotopic relationship between the subscripted pools, while H is used to represent a similar algebraic expression from 225 Table 2 Branch interaction factors as aected by the length of the common path below the branchpoint. The numbering of ¯uxes and isotope ratios corresponds to that of Fig. 3. The ®rst column gives the notation used in Eqs. 11 and B6 to refer to branch interaction factors for basal stem pathways of dierent length. The middle two columns provide orientation between the table and Fig. 3 Notation used in Eq. 11 Source pool R Number of intermediate pools in basal path Branch interaction factor, B() B(3..4) R3 0 F15 B(2..4) R2 1 B(1..4) R1c 2 B(0..4) R0 3 B(source..branchpoint) Rsource n F16 1 F15 F16 F6 1 F5 F7 F15 F16 F4 F4 F6 1 F5 F5 F7 F3 F15 F16 F2 F2 F4 F2 F4 F6 1 F1 F3 F3 F5 F3 F5 F7 Fnet:other:sink 1 n flux quotients primaryinflux F1 Eqs. 7 or 9 when that expression does not necessarily de®ne the full isotopic relationship. The branchpoint interaction factor, B(), describes the relative likelihood (potential values range 0±1) for heavy isotopes excluded from one branch to end up in the alternative sink rather than euxing back to the original source. The precise value of B() depends on two component factors (Table 2). The ®rst is the ratio of the net ¯ux into the alternative branch (making the value of B() sink-speci®c) relative to the total, unidirectional in¯ux entering the bottom of the pathway from the ultimate source, and the second is a progressive sum of ¯ux quotients related to the shared pathway and the same for both sinks. Reduced total eux, longer shared pathways, and/or ¯ux dynamics in the shared pathway that inhibit transference through the multiple mixing steps of the intervening pools make it more likely for heavy isotopes excluded from one branch to end up in the other sink, and this results in a value of B() closer to 1. The exact expression for B() is not fully intuitive, but follows a regular pattern that can be expanded for a common basal pathway of any length (Table 2). Equation 11 indicates that, for two sinks originating from a common source, the expected isotope ratio of just one of the two sinks considered alone is related, ®rst of all, to the discrimination between the source and that sink as would be predicted by Eq. 7 while disregarding the presence of the other branch [i.e. H(source:sink1)]. To this is added a second term consisting of the branch interaction factor [B(source..bp); Table 2] multiplied by the dierence in discrimination expected in the two diverged paths beyond the branchpoint. The positive sign associated with D(bp:sink1) in Eq. 11 indicates that the expected fractionation past the branchpoint leading to the sink under evaluation (i.e. sink 1) can be enhanced by the presence of branch 2. This can be understood as an eective increase in the EP:IP for sink 1 since material moving from the branchpoint to sink 2 can carry with it some of the excess heavy isotopes failing to enter sink 1, removing them from the system and allowing the pathway to sink 1 to more fully express its potential discrimination. In contrast, D(bp:sink2), has a negative sign in Eq. 11, because any heavy isotopes left preferentially at the branchpoint by the ¯ux leading to sink 2 will necessarily tend to isotopically enrich the ¯ux to sink 1. If D(bp:sink1) and D(bp:sink2) are equal in value, all interactive eects will cancel regardless of the partitioning of ¯uxes between sinks or other ¯ux dynamics contributing to B(). As discussed previously in regard to Eq. 7, the values of R for all intermediate pools are also implicit in Eq. 11 for a branched pathway. The required formulas and order of calculation to predict R for both sinks 1 and 2 and all intermediate pools in Fig. 3 are summarized in Table 3. Interestingly, the subtler interactions between the branches cancel out in the mean D of the combined sinks de®ned as: D source:combined sinks 1&2 .R sink1 Fnet:sink1 Rsink2 Fnet:sink2 R0 Fnet:total 1 12 The derivation is again left to Appendix B, but the ®nal formula, based on Eq. 11 to de®ne Rsink, is: D source:combined sinks 1&2 Fnet:sink1 H source:sink1 Fnet:sink2 H source:sink2 Fnet:total 13 where H has the same de®nition as in Eq. 11. The reciprocal interactions of the branches, which were important in Eq. 11 for de®ning D of individual sinks, cancel out in this weighted mean of the respective H values. Model validation and application Validation of isotopic theory The process of isotopic discrimination is inherently expressed, from ®rst principles, as a change in the 226 Table 3 Calculation of {Ri} for Fig. 3 with the order of calculations and appropriate formulae. R0, all discrimination constants (a±i), and {Fi} are given as inputs. The order of calculation is: (i) R for sink 1 is calculated using the appropriate form of Eq. 11, (ii) R of intermediates between sink 1 and branchpoint (including the branchpoint itself) using Eq. 7, (iii) calculation of all basal stem pools relative to sink 1 using Eq. 11 with the appropriate form of the branch interaction factor (Table 2), (iv) calculation of sink 2 relative to the branchpoint with Eq. 7, and (v) R of second branch intermediates using Eq. 7. Both Eq. 7 and Eq. 11 are used to ®rst calculate R of terminal sinks relative to a source, and also to back-calculate R for intermediate pools. In each instance where an intermediate pool is treated as Rsource, D source:sink Rsource =Rsink 1, and therefore Rsource Rsink 1 D source:sink where D(source:sink) is de®ned by the appropriate equation identi®ed in the right-hand column Pool label Source Branchpoint Sink 1 Pool R No. 0 1 2 3 4 5 6 7 8 9 Sink 2 10 Calculation D calculated order using: De®ned input R7 1 D 1:7 7 R7 1 D 2:7 6 R7 1 D 3:7 5 4 R7 1 D 4:7 R7 1 D 5:7 3 2 R7 1 D 6:7 R0 1 1 D 0:7 R10 1 D 8:10 10 R10 1 D 9:10 9 R4 8 1 D 4:10 Eq. Eq. Eq. Eq. Eq. Eq. 11 11 11 7 7 7 Eq. 11 Eq. 7 Eq. 7 Eq. 7 abundance ratio, [heavy]/[light], as one isotopic form reacts faster than another, but de®nition of precise massbalance relationships requires the use of atom fractions, [heavy]/([heavy]+[light]). Discrimination algorithms combining both features are needlessly complex, however, and, when working with natural abundance levels, the error generated by the approximation of treating an expression based on the abundance ratio as if it expressed a true mass balance is usually well below the resolution of measurement (Hayes 1982). The initial equations above constraining `conservation of isotopic abundance' (e.g. Eq. 4 or A.I.2) incorporate this common, minor imprecision. All ®nal algorithms discussed above were tested for mass-balance consistency. The de®ned inputs in all simulations were: (i) Rsource, (ii) {Fi}, and (3) {a, b, c, ...}. The value of {Ri} for all pools was then predicted from the algebraic solutions (Table 3 for Fig. 3, Table 4 for Fig. 1). All intermediate pools were then tested for mass Table 4 List of equations used to de®ne the isotopic composition of all pools for nitrate assimilation (Fig. 1) balance using two approaches, illustrated below for the ®rst intermediate pool of any straight-chain (e.g. Fig. 2, pathway B, or Fig. 3): R0 R1 R1 R2 F1 F2 F3 F4 0 1 a 1 a 1 b 1 b 14 and R0 R1 F2 A F1 A 1 a 1 a R1 R2 F4 A 0 F3 A 1 b 1 b 15 where the quotients inside the angled brackets are R¯ux.i, and each A Rflux:i is to be calculated as R¯ux.i/(R¯ux.i+1) (i.e. the conversion from isotope ratio to the corresponding atom fraction). Equation 14 is a `mass balance' based on isotopic ratios. It is not perfectly accurate, but matches the assumptions of the derivations. Equations 7, 9, and A.II.6 all had zero residuals for all pools when tested by Eq 14, indicating algebraic accuracy in the derivations. Equation 15 always indicated small but non-zero residuals. To evaluate the true accuracy of the predicted {Ri}, a set of equations analogous to Eq. 15 for each intermediate pool in the pathway was solved simultaneously by numerical techniques, along with a related statement that the isotopic composition of the product pool must equal the net contribution from the source. This numerical solution identi®ed the unique {Ri} for which all mass-balance residuals were <1á10±16. These were compared to the predicted {Ri} for (i) sensitivity analyses of accuracy across a range of input parameters, and (ii) testing the accuracy of algebraically `simpli®ed' expressions from which `trivial' terms and factors had been deleted (i.e. Eqs. 10 and 11). For most parameterizations relevant to natural abundance contexts, the maximum error in d units for the predicted R of any pool in a pathway modeled by Eq. 7, 9, or A.II.6, was in the order of 0.001±0.01&. Dependence of accuracy on {Fi} was negligible. Dependency of accuracy on the value of R0 from 0.001 to 0.5 was also almost negligible when {ai} values were in the range observed in biological transformations. A small dependency was evident on the magnitude of the {ai}, but with Rsource de®ned at a value realistic for natural abundance nitrogen (0.0036), errors as large as 0.1& were likely only when discrimination constants were over 300&. If {ai} and Rsource both had high values, Pool label Pool No. R Soil NO3± (source) Root NO3± (branchpoint) Root assimilated N (sink 1) Xylem NO3± Leaf NO3± Leaf assimilated N (sink 2) 0 1 2 3 4 5 De®ned input =R5=R3 =R0/(1+D0:2) =R5=R1 =R5(1+D5:4) =R0/(1+D0:5) Calculation order D calculated using: 3 1 3 4 2 Eq. Eq. Eq. Eq. Eq. 7 19 7 7 20 227 the resulting variability in R among dierent pools seriously compromised its suitability as a proxy for massbalance calculations and led to inaccurate predictions, but such parameterization is not relevant to actual natural abundance studies in nitrogen or carbon. The simpli®ed equations with `trivial' terms deleted also performed well, but rigorous tests revealed some caveats. Equation 10 is a simpli®ed version of Eq. 9, and lacks the ratios aa/aa', ab/ab' etc. The magnitude of error associated with this deletion is sensitive to the EP:IPs. If, for example, a in Fig. 2, pathway D, diered from a' by as much as 20&, and the ®rst step was at full equilibrium (i.e. F 2 =F 1 1), then the expected error (a worst-case scenario with respect to F2/F1) could be as large as 0.4&. Moreover, if several such steps with unequal fractionation constants for forward and backwards reactions occur in sequence (and with the same polarity in terms of the unequal isotope eects), the errors are multiplicative and even larger. However, although the errors might seem signi®cant in such a case, approaching a few per million, several additive equilibrium fractionations of 20& would give an overall all-isotope eect almost two orders of magnitude larger than this error. In most practical cases, the use of Eq. 10 gives values correct to within 0.1&, but when EP:IP ratios approach 1 and very precise values are needed, it is better to use Eq. 9. Equation 11 is also a simpli®ed approximation, in this case of Eq. A.II.6. Similar caveats hold as discussed for Eqs. 9 and 10. A worst-case scenario includes (i) a large value for Dbp:sink2 which forms the factor outside the square brackets in A.II.6, (ii) little or no eux back to the source, (iii) ¯uxes into the branches largely irreversible, and (iv) most net ¯ux into the alternate pathway. Quantitative examples are given later in regard to the nitrate-assimilation model described below (Table 5). Table 5 Predicted isotopic behavior of plant pools during nitrate assimilation. Subscripted variables refer to Fig. 1, and it is assumed that F5=F7=F9 at all times, that only the assimilation step shows discrimination (i.e. a=c=d=0), and that b=e=16&. The upper ®ve rows show variable inputs de®ning contrasting ¯ux dynamics. d calculations follow Table 4, only Eq. B6 was used instead of Eq. 11. The `maximum d error' gives the maximum In¯ux (F1) Eux (F2) Root assimilation (F3) Leaf assimilation (F9) d Soil NO3± (R0), & d Root NO3± (R1), & d Root assimilate (R2), & d Xylem NO3± (R3), & d Leaf NO3± (R4), & d Leaf assimilate (R5), & d Plant ((F3R2+F9R5)/(F3+F9)), & d maximum error for any pool, & Eux/in¯ux (F2/F1) Eux/in¯ux % Error using Eq. 22 Assimilation in root (F3/(F3+F9)) Root assimilation % Error using Eq. 23 Expected isotopic behavior during uptake of NO3± The processes of nitrate uptake and reduction are likely to be near to the isotopic steady state most of the time and thus amenable to the kind of discrimination model developed above. Cytoplasmic pools are small, with turnover half-life estimates most commonly between 2 and 5 min (Lee and Clarkson 1986; Siddiqi et al. 1991). Labeled nitrate applied externally to the root can be traced through the root cytoplasmic pool into the xylem sap and leaf cytoplasmic pools all on a timescale of minutes in herbaceous species (Lazof et al. 1992; Hayashi et al. 1997). Discrimination against 15N during nitrate assimilation in plants can be described by applying Eqs. 11 and 13 to the conceptual ¯ow chart describing uptake (Fig. 1). Several key steps are assumed to be irreversible, including the actual reduction of NO3± to organic nitrogen in either root or shoot, and the transport of NO3± in the xylem. Since NO3± is not phloem mobile, it is assumed that there is no backtransport of NO3± from the shoot back to the root (Pate 1973; Smirno and Stewart 1985; Jeschke and Pate 1991). This means EP:IP from the leaf nitrate nitrate pool is zero, and there can be no expression of discrimination by leaf NO3± reductase activity against the soil NO3± pool signature (i.e. all NO3± entering the leaf is eventually ®xed or retained in the vacuolar NO3± pool). The NO3± pool acting as substrate for the root nitrate reductase, however, can exhibit eux back to the soil, and thus allow for the possibility of net discrimination during plant uptake of NO3±. From Eq. 11 and the diagram of Fig. 1 we therefore state that discrimination by the individual root and leaf sinks, respectively, must be: discrepancy when the same table is compiled using Eq. 11. Based on presumed measurement of the isotopic values in an experiment, the theoretical accuracy of Eqs. 22 and 23 is assessed for evaluating the underlying ¯ux dynamics. Five cases are considered: 1±3 have high eux:in¯ux but vary in fraction of assimilation in root, while 4 and 5 show the same contrasts at low eux:in¯ux Case 1 Case 2 Case 3 Case 4 Case 5 1 0.9 0.09 0.01 0 1.42 ±14.35 1.42 17.44 1.42 ±12.77 0.02 0.9 0.0 0.9 1.3 1 0.9 0.05 0.05 0 0.79 ±14.97 0.79 16.80 0.79 ±7.09 0.06 0.9 0.0 0.5 0.7 1 0.9 0.01 0.09 0 0.16 ±15.59 0.16 16.16 0.16 ±1.42 0.02 0.9 0.0 0.1 0.2 1 0.1 0.81 0.09 0 12.92 ±3.03 12.92 29.13 12.92 ±1.44 0.21 0.1 0.0 0.9 0.3 1 0.1 0.09 0.81 0 1.42 ±14.35 1.42 17.44 1.42 ±0.16 0.02 0.1 0.0 0.1 0.1 228 D 0:2 a 1 F2 F1 and D 0:5 a 1 F2 F1 F2 F5 b b F1 F1 c 16 F2 F3 c c F1 F1 b 17 and, by Eq. 13, the net fractionation by the whole plant to: F3 a 1 F2 F2 b F5 a 1 F2 F2 c F1 F1 F1 F1 D 0:2&5 F1 F2 18 These expressions cover the full range of possible fractionation in a system with this branched structure, but most of these steps are not thought to show substantial isotope eects during nitrate assimilation. Diusion of a charged ionic species in water is little in¯uenced by a single isotopic substitution because of the large sphere of hydration. No current evidence yet supports discrimination by either initial uptake from the soil NO3± pool across the plasma membrane in root tissues, or during xylem loading (Mariotti et al. 1982; Handley et al. 1998). The only established point of expressed discrimination during uptake and assimilation of NO3± is thought to be the ®rst step of reduction from NO3± to nitrite (Handley et al. 1998; Yoneyama et al. 1998). Both the further reduction of nitrite and incorporation into an organic molecule may be discriminating reactions, but are tightly linked to the ®rst irreversible step such that all nitrite formed is converted to organic product (Ledgard et al. 1985). If we set a=c=d=0, then Eqs. 16, 17 and 18 further reduce to: F2 F5 D 0:2 b 19 F1 D 0:5 F3 b F1 20 and D 0:2&5 F3 F1 F2 b F2 F1 21 Equations 19±21 permit rather simple interpretations. Expression of the potential discrimination associated with NO3± assimilation in the root (Eq. 19) is dependent directly on the total e¯ux:in¯ux for the root NO3± pool, with the added insight that transport up to the shoot in the xylem (F5 in Fig. 1) counts similarly to actual eux back to the soil NO3± pool (F2). This makes sense, since either ¯ux can remove accumulating heavy isotopes. Discrimination between leaf and soil pools (Eq. 20) is actually a balancing re¯ection of root discrimination in this model. If there is no discrimination in the irreversible xylem transport process, then the ¯ux into leaf assimilate must, at steady state, equal the signature of what is loaded into the xylem, which is R1. R1 is enriched rather than depleted in the heavy isotope by a degree determined by the fraction of total in¯ux from the soil that is assimilated in the root (F3/F1). A tendency for the shoot to have a more positive d15N than the root is a common feature in plants growing under pure nitrate nutrition (Yoneyama et al. 1998). Finally, the mean isotopic discrimination by the whole plant (Eq. 21) is a weighted average of Eqs. 19 and 20. Since Eqs. 19 and 20 are both re¯ections of the root assimilation process, but have unlike signs, mean plant discrimination will always be less than that by the root sink alone. The ®nal result (Eq. 21) indicates that expression of b is related to two factors: (i) the EP:IP of the root NO3± pool with respect to the soil only and excluding xylem loading as an eux (in contrast to Eq. 19), and (ii) the fraction of all assimilation that actually occurs in the root. Quantitative examples of how such contrasting ¯ux dynamics aect the total isotopic behavior of the plantsoil system (Fig. 1) are given in Table 5. The {Ri} for ®ve dierent scenarios are given as d values. Calculations follow Table 4, but displayed values are the result of using Eq. A.II.6 directly, and not the simpli®ed Eq. 11 (i.e. Eqs. 19 and 20 for this model pathway and speci®cally for root and leaf sinks, respectively). The maximum error for any pool in the model resulting from using Eq. 11 is given in the last line of the d values of Table 5, and is only about 0.2& in the worst-case scenario. This error arises exclusively from Eq. 20, D(soil NO3:leaf assimilate). This is because a=c=0 in Fig. 1 causes all of the cross-products inside the square brackets of Eq. A.II.6 to be zero, and only when calculating D(soil NO3:leaf assimilate) does D(bp:alternate sink) have a non-zero value (i.e. `b') causing the outer factor in Eq. A.II.6 to be other than 1 and giving a dierent result than Eq. 11. The fully precise version of Eq. 20 would have another factor of 1/(1+b). While this might seem simple enough to include, it would needlessly complicate the expression developed in Eq. 21 for the whole plant, and other expressions to be developed below, and the added accuracy is barely within most measurement precisions. Discussion This model allows for clari®cation of several points regarding 15N discrimination during plant assimilation of NO3±. First, as has been emphasized by previous authors (Mariotti et al. 1982; Shearer et al. 1991; Handley and Raven 1992), there can only be net discrimination with uptake at the whole-plant level if there is a substantial eux term between the root tissue and the soil NO3± pool 229 (Table 5, Cases 1 vs. 4). This is because the discriminating step is the ®rst reduction step, which utilizes an intermediate pool, cytoplasmic NO3±, as its substrate. If there is no eux then the intermediate root NO3± pool undergoes transient enrichment until the elevated R cancels out the discrimination of reduction during steady-state uptake. For the same reason, net discrimination during assimilation is only possible when a substantial fraction of the NO3± reduction occurs in the root, because only the root NO3± pool can have eux back to the soil (Table 5, Cases 1 vs. 3). This is a consequence of the nonmobility of NO3± in phloem. Actual discrimination at the whole-plant level depends, therefore, on both eux:in¯ux at the root plasma membrane, and the partitioning of total reduction between root and shoot tissues (Eq. 21). Many dierent combinations of these two factors could give the same overall net discrimination, but with dierent implications for isotopic heterogeneity between root and shoot (Table 5, Cases 3 vs. 4). This isotopic behavior can potentially be used to advantage in future studies of nitrogen metabolism in plants. If it is assumed that (i) the plant is dependent solely on NO3± uptake for its nutrition, (ii) the bulk root NO3± pool accurately re¯ects the mean isotopic composition of the root nitrate-reductase substrate pool, and (iii) the plant has grown for an extended period of time under constant nutritional conditions so that the isotopic composition of accumulated biomass is consistent with current physiological activity, then we can use isotopic measures of standing biomass to evaluate mean values of eux/in¯ux (F2/F1 in Fig. 1), partitioning of overall NO3± reduction between root and shoot (F3/(F3+F5) in Fig. 1), and an estimate of the net mixing of the organic nitrogen pools originally reduced in root versus shoot due to the circulation of organic nitrogen in both phloem and xylem (Fig. 1, transfer from leaf and root assimilatory NO3± sinks to bulk organic root and shoot pools)(see Appendix C for derivations). efflux dsoil:NO3 influx droot:NO3 dplant dplant 22 The formula is given in terms of d to emphasize the nature and number of empirical measurements needed. dplant refers to the bulked organic nitrogen of the whole plant. Only if all reduction is con®ned to the foliage is there no isotopic record of the eux:in¯ux ratio at the root. The partitioning of NO3± reduction between root and shoot is highly variable among species and environmental conditions (Smirno and Stewart 1985; Andrews 1986). By this model it can be estimated from isotopic data as: root assimilated N droot:NO3 total assimilated N b dplant 23 where b is the discrimination constant for NO3± reduction (Fig. 1). Estimates of this parameter in the literature are somewhat variable (Handley and Raven 1992), but average between 15 and 17&. Equations 22 and 23 are developed based on Eq. 11, which is an approximation as discussed above. The theoretical limits of accuracy due to this are illustrated for various ¯ux dynamics in Table 5. Predicting the isotopic composition of the root and leaf assimilate pools requires measurement of only droot.NO3. The irreversible NO3± reduction reaction in the root should produce an assimilate pool which diers from droot.NO3 by b. Since the root cytoplasmic pool is both the substrate for reduction and the branchpoint for xylem loading (which does not discriminate), the NO3± transported into the leaf should have the same signature as the root cytoplasmic pool. In the leaf, d of residual NO3± should become further enriched, again by b, while the nitrogen ¯ux assimilated into organic compounds has the same signature as xylem NO3± input at steady state. This leads to: droot assimilate droot:NO3 b 24 and dleaf: assimilate droot:NO3 25 Assuming that transport of organic nitrogen occurs without discrimination, the degree to which leaf and root assimilate pools are mixed in the growth of new tissues throughout the plant can then be estimated as a two-ended mixing model after measuring d for the sink of interest: dsink droot:assimilate 1 rdleaf: assimilate 26 which can be solved for the only unknown, r, the fraction of nitrogen in the sink originating from root reduction. The data of Yoneyama and Kaneko (1989) and Evans et al. (1996) for Brassica campestris (L.) var. rapa and Lycopersicon esculentum Mill. cv. T-5), respectively, allow us to evaluate the above expressions for some empirical measurements. Based on Eq. 23, the fraction of NO3± assimilated in the root in these studies was 13 and 58%, respectively, and this root assimilate pool contributed 36 and 53% (Eq. 26), respectively, to the total nitrogen in the organic fraction of the root. These numbers indicate extensive xylem and phloem circulation and mixing of leaf and root assimilate pools. It is of interest that the data of Yoneyama and Kaneko (1989) have an isotopic pattern throughout the whole plant that is largely consistent with the model presented here. This includes a leaf NO3± pool enriched by the leaf reductase discrimination constant over the root NO3± pool (26.1 vs. 14.2&, compare with Table 5). In contrast, the data of Evans et al. (1996) indicate an enrichment of only 3& for leaf NO3± over the root NO3± pool, much less than predicted. The latter observation could indicate either a spatial or temporal compartmentation of the leaf nitrate pool in L. esculentum that is not accounted for in the present model, such as a storage 230 pool in isotopic equilibrium with incoming xylem nitrate more than the leaf reductase substrate pool. Equations 22±25 may give only rough estimates of the parameters in question, and are the subject of current experiments. The key pool to all of these expressions is the root cytoplasmic NO3±, the branchpoint pool in the model (Fig. 1) and the substrate for root nitrate reductase (Granstedt and Huaker 1982). This pool, however, is small, and very dicult to measure directly. Measures of bulk root tissue NO3± will generally be dominated by the much larger vacuolar pool (Ferrari et al. 1973; Martinoia et al. 1973; Granstedt and Huaker 1982; Miller and Smith 1996). Estimates for turnover rates of vacuolar nitrate have been in the order of 10±14 h (MacKown et al. 1981; Yoneyama et al. 1987; Volk and Jackson 1993) though these values could vary dramatically depending on nutritional status and the variable vacuolar pool size. Since the cytoplasmic pool is the branchpoint for vacuolar exchange as well, the vacuolar pool should re¯ect the cytoplasmic pool isotopically over time. The slower exchange rate and turnover times of the vacuolar pool should result in a time-averaged record of the isotopic value of the cytoplasmic pool. This may be advantageous since the cytoplasmic pool might vary substantially in response to diurnal patterns in nitrate uptake and assimilation (Scaife and Schloemer 1994; Delhon et al. 1995). While there are numerous questions to be answered regarding the accuracy of this average, it can potentially give a value more appropriate for comparison with organic nitrogen in bulk plant tissues than would direct measurement of the more labile cytoplasmic pool. These relationships are currently being evaluated in comparative time courses of d15N for bulk root NO3± and xylem NO3±. The latter should re¯ect the full variation of any diurnal pattern in d15N for the cytoplasmic pool itself. The most comparable model in the literature predicting d15N patterns in plants with multiple nitrate reductase sinks is a compartment-based process model solved by numerical techniques (Robinson et al. 1998). Their model goes beyond just the NO3±assimilation process, and includes explicit terms for storage, both phloem and xylem transport of reduced nitrogen compounds, and the hypothesized excretion of organic N from the root to the soil, but it is consistent with assumptions made above in that only the reduction of nitrate was treated as exhibiting isotopic discrimination. Numerical solution in Robinson et al. (1998) was based on a simultaneous best-®t of the entire {Ri} as opposed to the small number of selected Ri evaluated above in Eqs. 22±24 and yielding speci®c insights into the nitrogen metabolism. The method used by Robinson et al., when applied to the data of Evans et al. (1996) and Yoneyama and Kaneko (1989), resulted in multiple solutions, at least in part because they included in the analysis unmeasured ¯uxes and Ri, such as the excretion of organic N. Of current interest is that the model solutions they concluded were most `biologically rele- vant' were in close agreement the estimates developed here and reported above for these same datasets. The numerically solved process-model approach permits virtually unlimited amounts of detail to be included, and has essential applications in rigorously testing integrated models of nitrogen metabolism and quantifying speci®c hypotheses. An example is the evaluation of the putative excretion of organic N discussed in Robinson et al. (1998). The algebraic solutions presented here have dierent inherent utility, and are likely to be most insightful when keeping the number of recognized pools to a necessary minimum. The simple expressions in Eqs. 22±26, for example, indicate important insights that may be gained from a limited number of measurements, and might be more useful in evaluating larger numbers of plants for the selected characteristics. In the context of the model presented above, the plant nitrate pools are treated as intermediates of small size that can be ignored in the overall de®nition of massbalance between soil and plant, and only organic nitrogen is allowed to be a net sink. The isotopic values of these intermediate pools are predicted, but not accounted for. For most plants, nitrate storage accounts for less than 1% of total plant nitrogen content (Smirno and Stewart 1985; Pate et al. 1993; Grindlay 1997). In such plants, this simplifying assumption is probably justi®ed, but in some species grown at high external nitrate, vacuolar nitrate pools can be quite variable in size and can account for up to 10% of total plant nitrogen. In such plants, the model would need to be expanded to recognize the vacuolar nitrate pool as a separate sink for mass balance. Acknowledgements I thank Dave Evans for ®rst interesting me in this problem, Lou Derry, Dennis Swaney, Roman Pausch, and David Robinson for helpful comments on the manuscript, and David Robinson for also sharing the Excel spreadsheet on isotopic fractionation during nitrate assimilation related to Robinson et al. (1998). The author was partially supported by EPA grant #826531-01-0 during preparation of this manuscript. Appendix A. Unbranched pathways One intermediate pool Consider pathway B in Fig. 2, but let pool 2 be the ®nal sink (i.e. F5=F6=0). Setting the net ¯ux moving from pool 0 to pool 1 equal to that from pool 1 to pool 2 de®nes a steady state for total mass in the transport pathway: F1 F2 F3 F4 A1 Isotopic steady state for the same process is de®ned by multiplying each ¯ux term in Eq. A2 by its isotopic abundance ratio (see previous explanation for Eq. 4): F1 R0 1a F2 R1 F 3 R1 1a 1b F4 R2 1b A2 231 To derive an expression for R0/R2 from Eq. A2, R1 must ®rst be eliminated. Based on Eq. 6, the following expression holds between pools 1 and 2: F4 R1 R2 1 b 1 A3 F3 Equation A3 treats the intermediate pool as the source during application of Eq. 6. Substituting A3 into A2 yields: h i3 2 F4 F4 R F 1b 1 F 1b 1 3 2 F 1 R0 4 2 F3 F3 5R2 1a 1a 1b A4 The numerator over 1+b further simpli®es to (F3±F4)(1+b), and (1+b) cancels. From Eq. A1, replace (F3±F4) with (F1±F2): 2 3 F4 F 1b 1 2 F 1 R0 4 F3 F1 F2 5R2 A5 1a 1a Multiply both sides by (1+a) and divide by F1R2: h i F4 F F a F F F 1b 1 1 2 1 2 2 R0 F3 A6 R2 F1 The ®rst term on the right equals 1, which we subtract from both sides, and the F2 terms not being multiplied by either a or b cancel out: R0 F2 F2 F2 F4 D 0:2 1a 1 b A7 R2 F1 F1 F1 F3 and from Eq. A6: F4 F4 R1 R3 1 b 1 c F3 F3 F4 F6 F3 F5 A11 The substitutions and collection of terms proceeds exactly as above, only the analogue of Eq. A4 has an additional term on the right-hand side over the denominator (1+c). Simpli®cation and reduction leads to: " F4 c F4 F4 F6 2F 1 b 1 2 2F1 R0 F3 F3 F3 F5 1 a 1 a # F3 F4 F5 F 6 R3 A12 From Eq. A8, substitute 2(F1±F2) for (F3±F4+F5±F6) and then proceed as before to the ®nal expression: R0 F2 F2 F2 F4 1a 1 D 0:3 b R3 F1 F1 F1 F3 F2 F4 F2 F4 F6 A13 c F1 F3 F1 F3 F5 Appendix B. Derivations of branched pathways First consider the simplest possible branched pathway (Fig. 4: note that pools and ¯uxes are labeled to be consistent with the larger pathway of Fig. 3 for later comparison). The goal is to derive an expression for D3:5, the discrimination between the source and sink 1 alone. Derivation for two intermediate pools As in Eq. A1, the starting point is a de®nition of massbalance including all intermediate pools: 2 F 1 F 2 F 3 F 4 F 5 F 6 A8 and an isotopic mass balance de®ned by multiplying A8 by the isotopic ratios of each unidirectional ¯ux: F1 R0 F2 R1 F 3 R1 F4 R2 F5 R2 F6 R3 2 1a 1a 1b 1b 1c 1c A9 This time, substitutions are needed for both R1 and R2 in terms of R3. As before, from Eq. 6: F6 R2 R3 1 c 1 A10 F5 Fig. 4 Reproduction of central 4 pools surrounding the branchpoint in Fig. 3. Source and sink relationships (ellipse and trapezoids), however, are rede®ned as required in the derivation leading to Eq. B5. Additional pools were then added to branches and basal shared pathway, and source, intermediate, and sink pools rede®ned as appropriate to each stage, in a sequential manner to derive the general expressions in Eqs. B6 and B11, which apply to the entirety of Fig. 3 232 Begin as before with statements of total mass balance: F7 F8 F9 F 10 F 15 F 16 B1 and isotopic conservation: F7 R3 1 d F 8 R4 F9 R4 1 d 1 e F10 R5 F15 R4 1 e 1 h F16 R8 1 h B2 Similar derivations must be repeated many times while sequentially adding more intermediate pools in the branches towards each individual sink, and within the shared basal portion of the pathway prior to the branchpoint. Each solution provides additional identities needed to eliminate all R but the source and sink of interest from the equation. The ®nal result is a general expression applicable to Fig. 3 or even larger systems: 2 D source:sink1 H source:sink1 B source...bp D bp:sink1 6 Fnet:sink2 U source...bp D bp:sink1 6 primary influx 6 Fnet:sink1 4 primary influx U source...bp D bp:sink2 efflux product BP influx product D bp:sink1 D bp:sink2 To derive an expression for R3/R5, R4 and R8 must be eliminated. These substitutions are done sequentially using Eq. 7, ®rst eliminating R8 in terms of R4: F15 B3 R8 R4 F15 F15 F16 h and then R4 in terms of R5: R4 R5 F9 F9 F10 e F9 B4 Since more-detailed algebra has been presented in previous examples, the steps are merely outlined here. After making the indicated substitutions for R4 and R8 into Eq. B2, it is helpful to ®rst group the terms with common factors of 1/(e+1) and 1/(h+1). This permits some reduction of the denominators. Then combine all terms over a common denominator, isolate the quotient R3/R5 on one side of the equality, and subtract 1 from each side. Expand all products, collect all terms multiplied by d, e, or h, and reduce each of the three resulting expressions using repeated ¯ux substitutions from Eq. B1 to give: d 1 F8 F8 F15 F16 e 1 F7 F7 F15 F16 de 1 F10 =F9 F7 D 3:5 F9 F10 F7 F8 e 1 F7 dh 1 F16 =F15 F10 =F9 h 1 F16 =F15 F10 F9 D bp:sink2 3 7 1 7 7 5 1 D bp:sink2 B6 where H and D both refer to equivalent algebraic expansions using Eq. 7 and the indicated subscripted pools for `source':sink. The essential dierence between H and D is that while D indicates a true isotopic relationship between two pools in a subsection of the overall pathway, H alone does not. The term `primary in¯ux' refers to the unidirectional ¯ux from the ultimate source in the model to the ®rst intermediate pool of the uptake pathway, `BP(eux product)/(in¯ux product)' is the now familiar eux/in¯ux product calculated between the ultimate source pool and the branchpoint, and B() and F() refer to a `branchpoint interaction factor' and a new function analogous to Eq. 7 for D, respectively, both to be discussed in detail below. In the case of the pathway in Fig. 4, B(source..bp) is equal to Fnet_sink2/primary in¯ux (i.e. [F15±F16]/F7) and F(source..bp) is d, the discrimination constant of the transfer step from source to branchpoint. Using these identities, Eqs. B5 and B6 are identical algorithms for D(3:5) in Fig. 4. However, both B() and F() add additional terms with each additional intermediate pool added to the shared basal path in multi-pool pathways. The form of B() for all possible starting points before F15 F16 F7 h 1 F16 F15 1 1 h 1 F16 =F15 B5 233 the branch point in Fig. 3 are given in Table 3. B() actually has two component factors, one of which is always Fnet_sink2/primary in¯ux while the other is a progressive sum of ¯ux quotients. This sum of quotients is not intuitive in form, but has a predictable pattern for any number of pools (Table 2). F() is calculated by taking the same terms listed in Table 2, and weighting each one by the sequence of discrimination constants in the basal shared path. For Fig. 4 this gave simply d, and for the full basal path shown in Fig. 3 this gives: U 0:4 a F2 F2 F4 F2 F4 F6 b c d F3 F3 F5 F3 F5 F7 B7 Happily, much of the complexity of Eq. B6 has an almost negligible eect on the ®nal calculated value. The factor 1/(1+D(bp:sink2)) outside the large square brackets is always very close to 1, while the last three of the four terms inside the square brackets are all discrimination cross-products with magnitudes similar to D2, which is close to zero. Moreover, these subtle eects of the outer factor and inner cross-products tend to cancel each other. Eliminating these extraneous terms and factors is often justi®ed and leads to Eq. 11 in the main text. The mean D for a branched pathway This de®nition in Eq. 12 is essentially equivalent to: D 0:combined:sinks Fnet:sink1 D 0:sink1 Fnet:sink2 D 0:sink2 Fnet:total B8 Substituting for D(source:sink) using Eq. 11 for each of the two sinks into Eq. B8 gives: Fnet:sink1 H 0:sink1 B 0::bp D 0:sink1 D 0:sink2 Fnet:total net:sink2 H 0:sink2 B 0::bp D 0:sink2 D 0:sink1 Fnet:total D 0:sink1&2 F B9 All terms associated with the branch interaction factor cancel out, leaving us with a simple ¯ux-weighted average of H for each branch as shown in Eq. 13 of the main text. D soil:nitrate:mean:plant F2 D source:nitrate:leaf:assimlate F5 F3 C1 As discussed in the main text, D(source.nitrate:leaf.assimilate) is equivalent to D(source.nitrate:root nitrate). F3+F5, the total assimilation by both sinks, is equivalent to F1±F2, the net uptake of nitrate at the root surface. These two substitutions into C1 give: ! F2 F1 D soil:nitrate:mean:plant D soil:nitrate:root:nitrate F2 1 F1 C2 Solving for F2/F1: D soil:nitrate:mean:plant F2 F1 D soil:nitrate:mean:plant D soil:nitrate:root:nitrate Using d dsource C3 dproduct (see Eq. 6b) gives Eq. 22. Partitioning NO3± reduction between root and shoot Again substituting into Eq. 21 de®ning net discrimination at the whole-plant level, but this time replacing F2/ F1 based on Eq. C3 gives: D soil:nitrate:mean:plant D soil:nitrate:mean:plant D soil:nitrate:mean:plant D soil:nitrate:root:nitrate F3 b C4 F3 F5 F3/(F3+F5) is the fraction of total nitrate assimilation taking place in the root (Fig. 1). Solving for this ¯ux quotient yields: root assimilation total assimilation D soil:nitrate:mean:plant D soil:nitrate:root:nitrate b C5 Equation 23 is rewritten from Eq. C5 in terms of d based on D dsource dproduct . References Appendix C. Isotopic expressions for plant nitrogen ¯uxes Root:soil eux:in¯ux ratio To de®ne an isotopic measure of F2/F1, the eux:in¯ux ratio at the root plasma membranes between soil and cytoplasmic NO3± pools (Fig. 1), ®rst substitute D0:5 from Eq. 20 into Eq. 21 giving: Andrews M (1986) The partitioning of nitrate assimilation between root and shoot of higher plants. Plant Cell Environ 9:511±520 Delhon P, Gojon A, Tillard P, Passama L (1995) Diurnal regulation of NO3± uptake in soybean plants: I. Changes in NO3± in¯ux, eux, and N utilization in the plant during the day/night cycle. J Exp Bot 46:1585±1594 Evans RD, Bloom AJ, Sukrapanna SS, Ehleringer JR (1996) Nitrogen isotope composition of tomato (Lycopersicon esculentum Mill, cv. T-5) grown under ammonium or nitrate nutrition. Plant Cell Environ 19:1317±1323 234 Farquhar GD (1983) On the nature of carbon isotope discrimination in C4 species. Aust J Plant Physiol 10: 205-226 Farquhar GD, O'Leary MH, Berry JA (1982) On the relationship between carbon isotope discrimination and the intercellular carbon dioxide concentration in leaves. Aust J Plant Physiol 9:121±137 Farquhar GD, Ehleringer JR, Hubick KT (1989) Carbon isotope discrimination and photosynthesis. Annu Rev Plant Physiol Plant Mol Biol 40:503±537 Ferrari TE, Yoder OC, Filner P (1973) Anaerobic nitrite production by plant cells and tissues: evidence for two pools. Plant Physiol 51:423±431 Granstedt RC, Huaker RC (1982) Identi®cation of the leaf vacuole as major nitrate storage pool. Plant Physiol 70:410±413 Grindlay DJC (1997) Toward an explanation of crop nitrogen demand based on the optimization of leaf nitrogen per unit leaf area. J Agric Sci 128:377±396 Handley LL, Raven JA (1992) The use of natural abundance of nitrogen isotopes in plant physiology and ecology. Plant Cell Environ 15:965±985 Handley LL, Scrimgeour CM, Raven JA (1998) 15N at natural abundance levels in terrestrial vascular plants: a precis. In: Griths H (ed) Stable isotopes, integration of biological, ecological and geochemical processes. BIOS Scienti®c Publishers, Oxford, pp 89±98 Hayashi H, Okada Y, Mano H, Kume T, Matsuhashi S, S-Ishioka N, Uchida H, Chino M (1997) Detection and characterization of nitrogen circulation through the sieve tubes and xylem vessels of rice plants. Plant Soil 196:233±237 Hayes JM (1982) Fractionation, et al.: an introduction to isotopic measurements and terminology. Spectra 8:3±8 Jeschke WD, Pate JS (1991) Modelling of the partitioning, assimilation and storage of nitrate within root and shoot organs of castor bean (Ricinus communis L.). J Exp Bot 42:1091±1103 Lazof DB, Rufty TW Jr, Redinbaugh MG (1992) Localization of nitrate absorption and translocation within morphological regions of the corn root. Plant Physiol 100:1251±1258 Ledgard SF, Woo KC, Bergersen FJ (1985) Isotopic fractionation during reduction of nitrate and nitrite by extracts of spinach (Spinacia oleracea L.) leaves. Aust J Plant Physiol 12:631±640 Lee RB, Clarkson DT (1986) Nitrogen-13 studies of nitrate ¯uxes in barley roots. I. Compartmental analysis from measurements of 13N eux. J Exp Bot 37:1753±1767 MacKown CT, Volk RJ, Jackson WA (1981) Nitrate accumulation, assimilation, and transport by decapitated corn roots. Plant Physiol 68:133±138 Mariotti A, Mariotti F, Champigny M, Amarger N, Moyse A (1982) Nitrogen isotope fractionation associated with nitrate reductase activity and uptake of NO3± by pearl millet. Plant Physiol 69:880±884 Martinoia E, Heck U, Wiemken A (1973) Vacuoles as storage compartments for nitrate in barley leaves. Nature 289:292±293 Miller AJ, Smith SJ (1996) Nitrate transport and compartmentation in cereal root cells. J Exp Bot 47:843±854 O'Leary MH (1981) Carbon isotope fractionation in plants. Phytochemistry 20:553±567 Pate JS (1973) Uptake, assimilation and transport of nitrogen compounds by plants. Soil Biol Biochem 5:109±119 Pate JS, Stewart GR, Unkovich M (1993) 15N natural abundance of plant and soil components of a Banksia woodland ecosystem in relation to nitrate utilization, life form, mycorrhizal status and N2-®xing abilities of component species. Plant Cell Environ 16:365±373 Robinson D, Handley LL, Scrimgeour CM (1998) A theory for 15 N/14N fractionation in nitrate-grown vascular plants. Planta 205:397±406 Scaife A, Schloemer S (1994) The diurnal pattern of nitrate uptake and reduction by spinach (Spinacia oleracea L.). Ann Bot 73:337±343 Sharkey TD, Berry JA (1985) Carbon isotope fractionation of algae as in¯uenced by an inducible CO2 concentrating mechanism. In: Lucas WJ, Berry JA (eds) Inorganic carbon uptake by aquatic photosynthetic organisms. American Society of Plant Physiologists, Rockville, Md, USA, pp 389±401 Shearer G, Schneider JD, Kohl DH (1991) Separating the eux and in¯ux components of net nitrate uptake by Synechococcus R2 under steady-state conditions. J Gen Microbiol 137:1179± 1184 Siddiqi MY, Glass ADM, Ruth RL (1991) Studies on the uptake of nitrate in barley. III. Compartmentation of nitrate. J Exp Bot 42:1455±1463 Smirno N, Stewart GR (1985) Nitrate assimilation and translocation by higher plants: comparative physiology and ecological consequences. Physiol Plant 64:133±140 Volk RJ, Jackson WA (1993) Compartmental analysis of nitrate eux from intact maize roots after continuous loading with 14NO3± or pulse loading with 15NO3±. Plant Physiol 102:S-136 Yoneyama T, Kaneko A (1989) Variations in the natural abundance of 15N in nitrogenous fractions of Komatsuna plants supplied with nitrate. Plant Cell Physiol 30:957±962 Yoneyama T, Yamada H, Yamagata M, Kouchi H (1987) Nitrate reduction and partitioning of nitrogen in Komatsuna (Brassica campetstris L. var. rapa) plants: compartmental analysis in combination with 15N tracer experiments. Plant Cell Environ 28:679±696 Yoneyama T, Fujiwara H, Wilson JM (1998) Variations in fractionation of carbon and nitrogen isotopes in higher plants: N metabolism and partitioning in phloem and xylem. In: Griths H (ed) Stable isotopes. BIOS Scienti®c Publishers, Oxford, pp 99±109
© Copyright 2026 Paperzz