Diurnal variation in the isotope composition of plant xylem 1 water biases the depth of root-water uptake estimates 2

21 1. Stable isotopologues of water are a widely used tool to derive the depth of root water 22 uptake (RWU) in lignified plants. Uniform isotope composition of plant xylem water 23 (i-H2O-xyl) along the stem length is a central assumption, which has never been properly 24 evaluated. 25 https://doi.org/10.5194/bg-2019-512 Preprint. Discussion started: 23 January 2020 c © Author(s) 2020. CC BY 4.0 License.


44
The use of stable isotope composition of water has greatly enhanced ecohydrology studies by 45 providing insights into phenomena that are otherwise challenging to observe, such as depth of invariance of the isotope composition of xylem water has, to our best knowledge, never been 75 assessed. 76 In principle, temporal variance in i-H2O-xyl within a plant during a day or along its height 77 can be expected on first principles. Here we hypothesize that it is in fact likely that various plant 78 physiological processes, ranging from very simple to more complex mechanisms could  multi-source mixing model (Phillips & Gregg, 2003). We subsequently consider vertical water 116 transport within the tree, which relates to the established sap flow pattern. Note that the model 117 presented here, focusses on deuterium but can easily be used to study stable oxygen 118 isotopologues. To ensure consistency and clarity in variable declarations we maintain the 119 following notation in the subscripts of variables: uppercase roman to distinguish the medium 120 through which water travels (X for xylem, R for root, S for soil) and lowercase for units of time  where , is the fraction of water taken up at the i th soil layer (Fig. 1a) defined as: Where ki is the plant specific total soil-to-root conductance, ,0, is the water potential at the 143 base of the plant stem and , , is the soil water matric potential (Fig. 1a). Total plant water 144 https://doi.org/10.5194/bg-2019-512 Preprint. Discussion started: 23 January 2020 c Author(s) 2020. CC BY 4.0 License. potential is generally defined as the sum of the pressure, gravity and matrix potential. Hence, 145 Ψ ,0, represents the xylem pressure potential. The term z i is the gravimetric water potential 146 necessary to lift the water from depth z i to the base of the stem, assuming a hydrostatic gradient 147 in the transporting roots. The model considers z i to be a positive value (zero at the surface) , thus 148 z i is subtracted from Ψ , , . AR,i is the absorptive root area distribution over soil layer i (Fig. 1a).

149
This parameter can be derived from plant allometric relations (Čermák et al., 2006) which is 150 subsequently distributed over the different soil layers via Jackson et al. (1995).

151
The total soil-to-root conductance is calculated assuming the root and soil resistances are 152 connected in series (Fig. 1a): where k R is the effective root radial conductivity (assumed constant and uniform), and = 155 , /ℓ is the conductance associated with the radial water flow between soil and root surface. where the root to soil water potential gradient is represented as ∆Ψ , = Ψ ,0, − (Ψ , , − ).

220
All parameters (e.g. RWU) of the four analyses are given in Table S1.  Table S1). The following sensitivity analyses were considered:  (Table 2).

297
Sampling was performed between 9am and 2pm to assure high sap flow. Liana and tree 298 sampling allowed highly contrasted sap flux density (Gartner et al., 1990).  where R is the heavy to light isotope ratio measured in the sample or standard. We calculate  At the stem base, simulated δ²HX,0,t displays a diurnal fluctuation (Fig. 2) that corresponds to 368 the daily sap flow pattern (Fig. 2c). This pattern is caused by shifting diurnal RWU depth. Early 369 in the morning, when transpiration is low, most of the RWU occurs in deeper layers, where soil 370 water potential is less negative and isotopic composition of soil water is dominated by depleted 371 deuterium ( Fig. S2a-b). As transpiration increases during the day, a significant proportion of 372 RWU is extracted from the drier shallow layers, which have an enriched isotopic composition.

373
In the afternoon, as transpiration declines, isotopic composition reflects again the composition  Fig. 2d. Here, the piston flow dynamics in SWIFT originate from lateral translation of the 395 δ²HX fluctuation at δ²HX,0,t. In addition to sampling height, analysis A3 depicts the importance 396 of sampling time (Fig. 1b).

397
Analysis B outputs predict the occurrence and width of the δ²HX-baseline drop as a function 398 of SFV (Fig. 1c). Moreover, depending on SFV, the isotopic signal can take hours or days to sampling an non-representative area (Fig. 1c).

440
Root properties, i.e. root membrane permeability (Fig. S4c) (Table 2). produces variability in sampled δ 2 HX that is much larger than the expected measuring error. In 465 addition, empirical field data show excessive i-H2O-xyl variance along the stem length (Fig. 4) 466 and over a short time frame (i.e. sub-daily, Fig. 5). Therefore, the assumption of uniform δ 2 HX 467 along the length of a lignified plant is rejected, both theoretically and empirically. Importantly, we show that violation of this assumption results in incorrect assessment 471 of differences in RWU depths between plants. Differences do not necessarily result from 472 variability in RWU depth, but may result from monitoring plants at different heights (Fig. 2), 473 at different times (Fig. 1b) or by comparing individuals which have different SFV (Fig. 1c). Our illustrates that these considerations are non-trivial.

506
Note that, based on our model, we expect that soil isotopic enrichment experiments will 507 generate extensive δ²HX variation along the length of trees whenever diurnal RWU fluctuations 508 cause water extraction to shift between labeled and unlabeled soil layers. Furthermore, when 509 enrichment experiments target trees with different hydraulic properties (such as SFV) care 510 should be taken as to determine when and where to sample these trees in order to assess an 511 enriched isotope composition. Researchers should be certain the signal will be present at the 512 sample height (Fig. 1-2).       Table   835 S1.  Table   S1. Panel