Plant nutrient uptake from soil is mainly governed by diffusion and transpirationally induced mass flow, but the current methods for assessing the relative importance of these processes are indirect.
We developed a microdialysis method using solutions of different osmotic potentials as perfusates to simulate diffusion and mass flow processes, and assessed how induced mass flow affected fluxes of nitrogen (N) compounds in solution and in boreal forest soil.
Varying the osmotic potential of perfusates induced vertical fluxes in the direction of the dialysis membranes at rates of between 1 × 10−8 and 3 × 10−7 m s−1, thus covering the estimated range of water velocities perpendicular to root surfaces and induced by transpiration.
Mass flow increased N fluxes in solution but even more so in soil. This effect was explained by an indirect effect of mass flow on rates of diffusive fluxes, possibly caused by the formation of steeper gradients in concentrations of N compounds from membrane surfaces out in the soil. Our results suggest that transpiration may be an essential driver of plant N acquisition.
Plant nutrient acquisition usually starts with the movement of nutrients from the surrounding soil to the surfaces of roots and mycorrhizal hyphae. By the 1960s and 1970s, classical works by Peter Nye and Bernard Tinker had shown that these movements constitute the limiting step for plant nutrient acquisition (e.g. Nye & Marriott, 1969; Nye & Tinker, 1977; Tinker & Nye, 2000). More recent studies have also concluded that such fluxes of nutrients are essential to plant performance (Chapin, 1980; Clarkson, 1985; Robinson, 1986; Marschner, 1995; Leadley et al., 1997; Lambers et al., 2008). Despite the acknowledged central role of nutrient fluxes, the availability to plants of nutrients in general, and nitrogen (N) in particular, is usually expressed in terms of amounts or concentrations in the soil (Marschner, 1995).
Movement of N from the soil towards plant roots or mycorrhizal hyphae occurs mainly through two processes: diffusion and mass flow. Both of these processes can supply N to plants, but their relative importance depends on soil conditions (Barber, 1995; Comerford, 2005), species-specific plant characteristics and their interaction with soil (Jungk & Claassen, 1997). Diffusion occurs as a consequence of concentration gradients resulting from the active uptake of N compounds at the surface of roots and hyphae. Transpiring plants induce mass flow of water from the soil solution towards root surfaces; this mass transfer of water also induces mass transfer of nutrients. Gain of N compounds by plant roots through mass flow also differs from that through diffusion in that the latter (though not the former) process is strongly affected by molecular size. The soil solution N composition of boreal forest soils is dominated by organic N compounds (Inselsbacher & Näsholm, 2012a) and may also contain large amounts of macromolecular N, such as peptides and proteins (e.g. Andersson & Berggren, 2005). The strong effect of molecular weight on diffusion rates, but not on mass flow rates, implies that transpiring plants may encounter a different composition of N compounds in the soil, with greater contributions of large N molecules, compared with plants relying solely on diffusion for their N gain.
In nutrient-poor soils, it is generally believed that diffusion is the main driver of N fluxes from the surrounding soil to the plant roots, while mass flow may be of greater importance in nutrient-rich soils (Barber, 1995; Smethurst, 2000; Comerford, 2005). Recent studies have also suggested a potential direct interaction between transpiration and soil nutrient availability, implicating nutritional regulation of whole-plant transpiration (Cramer et al., 2008; Raven, 2008) as well as a more direct interaction between root hydraulic resistance and nitrate availability in the root medium (Gloser et al., 2007; Gorska et al., 2008). These studies indicate a direct link between soil N availability and plant transpiration, suggesting that transpiration may be an important driver of N acquisition by plants.
While some experimental studies have suggested that mass flow is important for plant N acquisition (Strebel & Duynisveld, 1989; Plhák, 2003), several studies based on modeling have concluded that the contribution of mass flow should be negligible (Yanai, 1994; BassiriRad et al., 2008). Based on a theoretical consideration of mass flow and diffusion processes, BassiriRad et al. (2008) state that transpirationally induced mass flow should not lead to a net increase in nutrient gain for plants, because the effect will be cancelled out by a concomitant decrease in diffusion rates. The lack of a net effect of mass flow on root nutrient uptake would hence result from a flattening (or even reversing) of the diffusional gradient caused by the inflow of nutrients into the concentration gradient around the root. Earlier work by Nye & Marriott (1969) acknowledged that mass flow may be of importance both in itself and through its positive effect on rates of diffusion. The authors considered the latter effect small unless rates of water movement were very high, but also recognized that data were missing on the potential interaction effect at realistic rates of water flux.
Loss of water has been deemed an unavoidable disadvantage of the need for CO2 uptake by plants. However, several suggestions for additional functions and benefits of water loss through transpiration have been put forward (see Raven, 2008). It was speculated that one potential up-side of transpiration would be an increased nutrient gain. Conversely, environmental factors that limit transpiration, such as increased atmospheric CO2 concentrations, have been speculated to aggravate nutrient limitations of plants (Conroy, 1992; Conroy & Hocking, 1993; McDonald et al., 2002; McGrath & Lobell, 2013). The potential dependency of plant nutrient acquisition on transpiration is thus linked to the role of mass flow for root nutrient uptake, highlighting the need for accurate estimation of the transpiration–mass flow–N acquisition chain.
The classical method for estimating the relative contributions of mass flow (F) and diffusion (D) to plant N acquisition is based on comparisons of total N uptake by a plant over a period of time (Nup) with an integrated measure of transpiration (w) multiplied by concentrations of N in the soil solution averaged over the same period (c). Thus,
These calculations are believed to provide estimates of the contribution of mass flow; the contribution of diffusion is then calculated as the difference between total plant N uptake and the mass flow contribution (Jungk & Claassen, 1997; Lambers et al., 2008). Some of the challenges associated with this method are that N uptake by mycorrhizas is not included (Chapin et al., 2011), that no account is taken of the uncertainties in the assessment of transpiration or in the assessment of soil N concentrations, and that spatial and temporal variation in both soil water content and soil solution N concentrations are ignored. Furthermore, this method is unable to provide information about the potential interaction between mass flow and diffusion (sensu Nye & Marriott, 1969). We conclude that a direct and more robust method of assessing soil N supply by diffusion and mass flow is required.
Microdialysis has been developed as a technique that enables monitoring of the diffusion of N compounds in soils (Inselsbacher et al., 2011; Inselsbacher & Näsholm, 2012a,b). This method induces diffusion of dissolved compounds along a concentration gradient from the soil solution over a nonselective dialysis membrane into the perfusate (high-purity deionized water; Miro & Frenzel, 2004, 2005). The small size of the membranes allows noninvasive monitoring of the dynamics of induced soil N fluxes in situ (Inselsbacher & Näsholm, 2012a,b). Since osmotic potentials of the perfusate and the study system (the soil) are similar, concentration gradients between the two compartments drive fluxes of compounds, but not water, over the membrane and this setup therefore provides an estimate of diffusion. We speculated that the microdialysis technique could be further developed to estimate the importance of mass flow of soil water for flux rates of N compounds as well. Our hypothesis was that mass flow of soil solution towards the dialysis membrane could be induced by using a perfusate with a lower water potential than the soil solution. This may be achieved by using a perfusate containing a macromolecule of larger size (here dextran) than the molecular weight cutoff of the microdialysis membrane. Hence we hypothesized that both diffusion and mass flow processes as drivers of plant N acquisition can be studied directly in soil (Fig. 1). To accomplish this, we set up a series of experiments testing various perfusates containing different osmotically active compounds in order to assess their suitability to drive water flux over the microdialysis membranes in solutions with known concentrations of N compounds, as well as to examine various settings of the dialysis system under controlled conditions. The range of mass flow rates reported to occur for transpiring plants could be spanned by varying the concentration of osmotically active compounds in the perfusate. After establishing the optimal settings for inducing mass flow under controlled conditions in solutions with known concentrations of N compounds, we estimated diffusion and mass flow of N compounds in soil in order to study the importance of each process, as well as the interactions between them, on rates of N delivered to an imaginary plant fine root represented by the microdialysis probe.
Materials and Methods
Microdialysis set up and calibration
Two microdialysis systems were used in parallel and set up as described previously (Inselsbacher et al., 2011). Each system consisted of a syringe infusion pump (CMA 400) equipped with four gas-tight glass syringes (5 ml; Hamilton, Bonaduz, Switzerland), four microdialysis probes (CMA 20) with a 10-mm-long polyarylethersulphone membrane (molecular weight cutoff, 20 kDa; 400 μm inner and 500 μm outer diameter), and a refrigerated fraction collector (CMA 470). All equipment is commercially available and was purchased from CMA Microdialysis AB (Solna, Sweden). The four syringes delivered solution (‘perfusate’) to the microdialysis probes at constant flow rates and the efflux from the probes (‘dialysate’) was collected with a fraction collector.
To ensure the uniform performance of the probes throughout the experiments, each microdialysis probe was calibrated before and after the experiments according to the general calibration method (Bungay et al., 1990; Torto et al., 2001; Nandi & Lunte, 2009) and as previously described in detail for low-molecular-weight N compounds (Inselsbacher et al., 2011). Briefly, probes were inserted into a standard solution containing 100 μM of NH4+, NO3− and 19 amino acids (histidine, asparagine, arginine, serine, glutamine, glycine, aspartic acid, glutamic acid, threonine, alanine, proline, lysine, cysteine, tyrosine, methionine, valine, isoleucine, leucine and phenyalanine). The standard solution was kept at a constant temperature of 22°C and stirred with a magnetic stirrer throughout the calibration period to prevent the formation of a depletion zone around the probe surface (Inselsbacher et al., 2011). The probes were perfused with high-purity deionized (MilliQ; Merck Millipore Corp., Billerica, MA, USA) water at a flow rate of 1.0 μl min−1 for 8 h. Samples were collected at 2 h intervals and were immediately prepared for chemical analysis (described later). The relative recoveries (RRs) of the individual N compounds by each probe were calculated as given in Eqn 4:
where Cdial is the concentration of the measured N compound in the dialysate and Cstd is the concentration of the compound in the standard solution.
Induction of mass flow using the microdialysis system
Based on the study by Rosdahl et al. (2000), dextran (from Leuconostoc spp.; Sigma-Aldrich) was considered to be the compound of choice for lowering the osmotic potential of the perfusate and, thus, driving mass flow of water from the external environment (solution or soil). Dextran is a complex polymer of glucopyranos units with 1-6, α-glycosidic bonds often branched at the second, third or fourth C atom and with molecular weights spanning the range 3–2000 kDa. In a series of pre-experiments, the suitability of different molecular sizes and concentrations of dextran was tested to determine the water potential of each solution. In addition, we tested the effect of pump flow rate on mass flow of MilliQ water over the membrane surface. In detail, solutions of dextran with different molecular weights (20, 40 and 70 kDa, hereafter referred to as Dextran 20, Dextran 40 and Dextran 70) were prepared at five concentrations (1, 5, 10, 15 and 20%; w/v) and used as the perfusate at five different pump flow rates (1, 3, 5, 7 and 10 μl min−1). The water potentials of the five concentrations of Dextran 40 were calculated according to Michel et al. (1983), as given in Eqn 5:
where ΨDEX is water potential of the dextran solution and [DEX] is the concentration of Dextran 40 in g g−1 H2O.
It should be noted that the concentration of Dextran 40 will gradually decrease, and thus the water potential gradually increase, as the perfusate passes below the membrane surface of the probe. This dilution of the perfusate can be calculated, as mass flow rate and pump rate are known.
Microdialysis membranes were submerged in MilliQ water, which was kept at a constant temperature of 22°C. Dialysates were collected by an autosampler in preweighed vials (300 μl volume) and weighed again immediately after sampling. The amount of MilliQ water that passed through the membrane as a result of mass flow (VF) at each individual setting was estimated as follows:
where Vtot is the total volume of dialysate and Vpump is the volume provided at each individual pump flow rate.
In order to verify the accuracy of the microdialysis pump, that is, that Vpump equals the flow rate (μl min−1) multiplied by sampling time (h), Vpump was quantified empirically in a series of control experiments. Mass flow was inhibited by using the same solution as both the perfusate and the dialysate, resulting in equal osmotic potentials on both sides of the membrane and Vtot = Vpump. The dialysate volume (Vtot) was weighed immediately after each sampling period to confirm Vpump.
Estimating diffusive flux and mass flow of individual N forms in aqueous solution
Microdialysis membranes were inserted into a stirred standard solution containing 100 μM of NH4+, NO3− and 19 amino acids (as described earlier). The solution was kept at a constant temperature of 22°C to prevent temperature-induced shifts in diffusive fluxes (Inselsbacher & Näsholm, 2012b). MilliQ water was used as the perfusate for estimating diffusion of individual N forms over the membrane surface. The optimum pump flow rate for this study was determined in the preliminary experiments and set at 1.0 μl min−1. Samples were collected at 2 h intervals during the entire experimental period of 8 h and were immediately prepared for chemical analyses. Diffusive fluxes (D) over the probe membrane were estimated by calculating the total amount of each N compound diffusing over the membrane surface (15.9 mm2) during each sampling period (Inselsbacher & Näsholm, 2012a,b) and were then expressed as nmol m−2 s−1:
where Ctot is the concentration of each N compound in the dialysate, Aprobe is the membrane surface area, and t is the sampling time.
Based on the results from our preliminary experiments, the mass flow of individual N forms was estimated using 20% (w/v) Dextran 40 as the perfusate and a pump flow rate of 1.0 μl min−1. Samples were collected at 2 h intervals during the entire experimental period of 8 h and were immediately prepared for chemical analyses (described later). The mass flow (F) of each N compound was calculated as:
where Cstd is the concentration of each N form in the standard solution and VF is the volume of solution derived from the mass flow (Eqn 6). The term (Cstd × VF) represents the amount of each compound (nmol) in a sample derived from mass flow.
Diffusive fluxes when mass flow occurred simultaneously (DF) when Dextran 40 was used as the perfusate were subsequently estimated as:
Site description and soil sample preparation
Soil was collected in May 2012 from a Scots pine heath forest at the Rosinedal Research area, near Umeå, Sweden (64°10′20″N, 19°44′30″E). The annual mean precipitation is 587 mm and the annual mean air temperature is 1.9°C. The forest soil is nutrient-poor and classified as a sandy glacial till Haplic podzol (FAO, 1998) with 2% silt, 97% sand, 1% gravel and an organic layer depth of 5–10 cm. The organic layer has a C : N ratio of 38.7 and soil pH (H2O) of 5.2. Total wet and dry N deposition in the study area is c. 2 kg N ha−1 yr−1.
Soil samples were taken from the uppermost organic soil layer (0–10 cm) and immediately transferred to the laboratory for further processing. Coarse debris, needles and stones were removed manually and the remaining unsieved soil was mixed.
Estimates of diffusion and mass flow in soil samples
Samples (15 g) of field-moist soil were placed in 100 ml glass beakers and 25 ml of MilliQ water was added to achieve ≥ 100% of the water-holding capacity. Soil water potentials at water saturation were therefore expected to be approaching 0 MPa (Lambers et al., 2008). Probes were inserted vertically into soil to a depth of 1.5 cm below the surface. The probes were perfused with MilliQ water and 20% Dextran 40 for estimating diffusive and mass flow, respectively, at a constant pump rate of 1.0 μl min−1. The samples were collected over a period of 8 h at 2 h intervals and were immediately prepared for analysis of NH4+, NO3− and amino acids as described in the following section. Diffusive fluxes over the microdialysis probe membranes were calculated as described earlier. Calculations of mass flow of N compounds in soil were based on the soil solution concentrations estimated from the concentrations in dialysates and the recovery factors determined for the individual compounds (see Eqn 4).
Before chemical analysis, dextran in the samples obtained from the mass flow experiments was precipitated with ethanol (cf. Behravan et al., 2003). This step was essential, as high dextran concentrations were found to interfere with the derivatization of amino acids. This was achieved by adding 100 μl of 98% ethanol to 100 μl of sample. The mixture was vortexed for a few min and the dextran-precipitate was spun down through centrifugation at 20 800 g and 4°C for 15 min. The supernatant was collected and immediately processed for chemical analysis. We also performed tests of potential effects of the removal of dextran through ethanol-precipitation on the recovery of NH4+, NO3− and amino acids by comparing concentrations of standards before and after precipitation.
All samples were analyzed for NH4+, NO3− and amino acids as described previously (Inselsbacher et al., 2011). In detail, NH4+ and amino acids were analyzed by reversed-phase LC using a Waters (Milford, MA, USA) Ultra High Performance Liquid Chromatography (UPLC) system with a Waters Tunable UV (TUV) detector. Aliquots of sample (20 μl) were derivatized with a Waters AccQ-Tag™ Ultra Derivatization kit for amino acid analyses. Individual amino acids were separated on an AccQ-Tag™ Ultra column by elution with a mixture of 0.1% formic acid (solution A) and 10% acetonitrile (solution B) using the following gradient: 0–5.74 min isocratic 99.9% solution A, declining to 90.9% solution A from 5.74 to 7.74 min, to 78.8% solution A at 8.24 min and then to 40.4% solution A at 8.74 min, before re-equilibration with 99.9% solution A from 8.74 to 9.54 min. The flow rate was 0.6 ml min−1 and the column temperature was 55°C. Nitrate was analyzed by the vanadium (III) chloride (VCl3) and Griess method as described by Hood-Novotny et al. (2010), based on the technique described by Miranda et al. (2001).
All results were examined using Mann–Whitney U-test performed in SAS 9.3 for Windows (SAS Institute Inc., Cary, NC, USA). Differences were considered statistically significant at P <0.05.
Choice of perfusate and pump flow rate for induction of mass flow
Dextran 40, forming a clear solution when dissolved in MilliQ water, was the most effective perfusate for inducing mass flow of water into the microdialysis probes. By contrast, Dextran 20 was only partially soluble in water and formed a turbid solution, and tests involving Dextran 70 were unsuccessful in inducing mass flow. We used a pump rate of 1 μl min−1 as this rate allowed for a measurable increase in sample volumes with the dextran solution. Five different concentrations of Dextran 40 were tested with respect to their effectiveness to drive mass flow of water into the microdialysis probes. The osmotic potential of the five different concentrations of Dextran 40 solutions were calculated according to Michel et al. (1983), and ranged from −0.002 MPa for 1% Dextran 40 to −0.1 MPa for 20% Dextran 40. Root water potentials may vary within a larger range; at a water potential of −0.2 MPa, roots are able to extract more than two-thirds of the storable water of sandy soils, and root water potentials of trees may reach as low as −2 to −4 MPa (see Larcher, 2003). The water potentials of the perfusate of the microdialysis probe in the current study are thus in the upper range of those found in roots. No detectable mass flow was achieved for the two lowest concentrations (1 and 5%), while a monotonic increase in mass flow rates was observed for the three highest concentrations (Fig. 2). Thus, in this experiment, the maximum mass flow rates of water over microdialysis probe membranes amounted to 2.2 × 10−7 m s−1 (Fig. 2).
Based on these results, a perfusate of 20% (w/v) Dextran 40, corresponding to an osmotic potential of −0.1 Mpa, and a pump flow rate of 1 μl min−1 were used for estimating the effect of mass flow on flux rates of N compounds in standard solution and in soil.
Diffusive and mass flow fluxes of N compounds in solution
Fluxes of NH4+, NO3−, and total amino acids were, respectively, 58, 63 and 34% higher when dextran was used as the perfusate than when water was used (Fig. 3a,b, Supporting Information Table S1). Similarly, the flux rates of total N (sum of NH4+, NO3− and amino acids) over the microdialysis membranes were c. 37% higher when using dextran instead of water as perfusate (P <0.0001, Fig. 3b).
Nitrogen fluxes pertaining to mass flow amounted to similar fractions of total fluxes for the different N forms, namely 19 and 20% for NH4+ and NO3−, and 25 and 24% for total amino acids and total N, respectively (Fig. 3a,b). Moreover, NH4+ and NO3− diffusion rates over dialysis membranes in the presence of mass flow were, respectively, c. 28 and 31% higher than the corresponding diffusion rates in the absence of mass flow (P <0.001, Fig. 3a). The total flux of amino acids supplied by diffusion in the presence of mass flow was not significantly different from that by diffusion in the absence of mass flow (P >0.05, Fig. 3b).
Diffusive and mass flow fluxes of N forms in forest soil samples
Flux rates for NH4+ in forest soil samples were more than three times higher (14.9 vs 4.5 nmol m−2 s−1) when using Dextran 40 than when using water as the perfusate (P <0.001, Fig. 4a, Table S1). The concentrations of NO3− in the dialysates with water as the perfusate were below detection limits, while flux rates of NO3− achieved when Dextran 40 was used amounted to c. 3.9 nmol m−2 s−1 (Fig. 4a, Table S1). Total amino acids dominated the N fluxes of forest soil samples and were c. 72% higher in Dextran 40 dialysates than in water dialysates (P <0.001, Fig. 4b, Table S1). Flux rates of total N in forest soil samples were > 90% higher when Dextran 40 was used as the perfusate than when water was used (P <0.0001, Fig. 4b).
Calculating partial (mass flow and diffusive) fluxes, we estimated that N fluxes pertaining to mass flow amounted to a relatively small share of total N fluxes: 6, 12 and 14% for NH4+, total amino acids and total N, respectively (Fig. 4). Separating total flux into these partial fluxes was not possible for NO3−, as data on soil solution NO3− concentrations were not available. Diffusive fluxes of NH4+ were significantly higher in the presence of mass flow and this effect was the main cause of the large increase in total NH4+ flux in Dextran 40 dialysates (P <0.001, Fig. 4a). Similarly, diffusive fluxes of total amino acids were significantly higher in the presence of mass flow than when it was absent and the increase amounted to c. 55%; this was the main reason for the increased rate of total flux of amino acids (P <0.001, Fig. 4b). The sum of diffusive fluxes of the different N forms (NO3−, NH4+ and amino acids) exhibited an increase of 67% in the presence of mass flow, as compared with the absence of mass flow (P <0.001, Fig. 4b).
Plant nutrient acquisition may often be limited by the flux of soil nutrients in the direction of the surface of roots and mycorrhizal hyphae (Nye & Marriott, 1969; Nye & Tinker, 1977; Tinker & Nye, 2000) and diffusion and mass flow are the two major drivers underpinning this flux of nutrients (Nye, 1977). Estimates of the relative contributions of these two processes to plant nutrient supply have, to date, been based on indirect methods (Jungk & Claassen, 1997; Lambers et al., 2008). As stated earlier, mass flow of nutrients from soil towards roots is driven by transpiration, and hence assessment of the importance of mass flow for plant nutrient gain is also important for an accounting of the pros and cons of plant water loss through transpiration (Raven, 2008). Mass flow affects nutrient gain both directly, through delivering nutrients to surfaces of roots, and indirectly, through affecting the concentration gradient of nutrients from root surfaces out in the soil (see the concentration gradient in Fig. 1). Because plants, but not microbes, can sustain high rates of mass flow, this process is also of importance for plant competition for nutrients such as N against soil microbes. We conclude that research on plant–water relations, plant–soil interactions and plant–microbial competition for N would greatly benefit from a better understanding of how nutrient fluxes depend on diffusion and mass flow.
The current report describes an experimental procedure for direct assessment of diffusion and mass flow rates of different N sources in solution and in soil. We employed and modified the microdialysis technique (Inselsbacher et al., 2011) using an osmotically active solution as the perfusate instead of pure water. Our rationale was that mass flow towards the probe membrane could be induced by using a perfusate with a lower water potential than the system of study (standard solution and soil solution in this case) and containing macromolecules of larger size than the molecular weight cutoff of the microdialysis membrane (Fig. 1), and that the rate of this flux would be proportional to the difference in water potential between the interior and exterior of the probe membrane (Fig. 2). In medical applications of the microdialysis technique, Dextran 70 has been used in situations when microdialysis probes are mounted in tissues like human muscles (e.g. Rosdahl et al., 2000), but in such applications the aim is to avoid net losses of perfusate during microdialysis sampling, while our aim was to drive a net influx of solution over the probe membranes. For this purpose, our tests showed that Dextran 40 is the preferred choice of osmotically active compound.
Rates of water flux towards roots have been estimated to be in the range 0–10−7 m s−1 (Tinker & Nye, 2000; BassiriRad et al., 2008), a range that can be covered by using different concentrations of Dextran 40 as the perfusate (Fig. 2). In our tests with a high concentration of Dextran 40 (20% w/v), the velocity of radial flux of water through the microdialysis probe membrane averaged 3.0 × 10−7 m s−1 when submerged in standard solution and 1.8 × 10−7 m s−1 in soil. Thus, the tests of mass flow effects on fluxes of N compounds in the present study can be considered to represent situations where there are high transpiration rates.
In standard solution, mass flow of water resulted in a concomitant increase in N flux of, on average, 37% (Fig. 3, Table S1). All individual N compounds adhered to the general pattern of increased flux rates under induced mass flow. In soil, on the other hand, where mass flow of water was lower, average rates of N flux across probe membranes were increased by 183% (Fig. 4, Table S1). This increase in N flux was the result of increased fluxes of NH4+, NO3− and total amino acids (Fig. 4). Separating soil N fluxes into diffusion and mass flow pointed to a strong indirect effect of mass flow on diffusion rates of N compounds and a much smaller direct effect through water inflow (Fig. 4, Table S2). Diffusion rates of N compounds are generally much lower in soil than in solution, as a result of both chemical interactions between compounds in solution and the solid phase of soils and of the flow path tortuosity of soils. The strong effect of mass flow on diffusion rates of N compounds in soil, but not in solution, may be caused by the continuous delivery of N towards and into the depletion zone that forms around the probe membrane surface. This supply of N compounds via mass flow would keep the concentration gradient between the inside and outside of the membrane steep, leading to higher diffusive fluxes (see the concentration gradient in Fig. 1).
One unexpected finding in the soil used in our study was that three of the amino acids present in low concentrations (glutamic and aspartic acids, and glutamine) exhibited the opposite pattern, that is, lower flux rates under conditions of mass flow (see Table S1). The reason for this peculiar pattern is presently unclear and will require special attention in additional studies, including a wider range of soils.
Recent as well as older studies have highlighted the potential connection between plant transpiration and nutrition, and the implicated role of mass flow as a driver of nutrient acquisition (Nye & Marriott, 1969; Nye, 1977; Robinson, 1986; Raven et al., 1992; Tinker & Nye, 2000; Cramer et al., 2008, 2009; Raven, 2008). The negative effect of increasing atmospheric CO2 concentrations on stomatal openings, and thereby on plant transpiration, has been deemed to result in reduced rates of nutrient acquisition in both trees (Conroy, 1992; Conroy & Hocking, 1993; McDonald et al., 2002) and annual crops (McGrath & Lobell, 2013). This would imply that reduced transpiration rates could be a driver of progressive nitrogen limitation (Luo et al., 2004). By contrast, model calculations point to a negligible role of mass flow (and thereby transpiration) as a driver of nutrient acquisition (Yanai, 1994; BassiriRad et al., 2008), suggesting that increasing CO2 concentrations will not have a negative impact on nutrient acquisition, at least not as a result of effects on stomata. Resolving these issues requires a more direct assessment of the importance of mass flow and diffusion for plant nutrient acquisition. We propose that the experimental setup involving the microdialysis system presented in this study has the potential to deliver such information by direct measurements. Analysis of the formation of depletion zones around probe membranes using variable mass flow conditions may provide additional information to resolve this issue. Also, direct tests of the impact of transpiration on plant N acquisition may help in clarifying the potential interdependency of these two processes.
Obviously, there are clear limitations to the microdialysis method as well; some of the most obvious are the lack of specificity of probe membranes, while roots may regulate uptake of both water and solutes. Also, roots grow through soils while the microdialysis probes are stationary, meaning that the formation of depletion zones are more accentuated for probes. Future studies need to cover a range of induced mass flow rates in order to better describe the links among transpiration, mass flow of water towards root surfaces and potential increases in delivery rates of nutrients.
In conclusion, we have presented a method of estimating rates of delivery of soil N compounds to roots driven either by pure diffusion or by a combination of diffusion and mass flow. A wide range of mass flow rates can be achieved by varying the osmotic potential of the perfusate, thus covering the full range of water mass flow rates supposedly occurring in plants growing in soil (e.g. Tinker & Nye, 2000). This method is applicable to soil sampling in the laboratory as well as in field studies (cf. Inselsbacher & Näsholm, 2012a), and may also be used to study the effect of temperature on the relative proportions of N gained through mass flow and diffusion (cf. Inselsbacher & Näsholm, 2012b). We chose to focus our study on amino acids and mineral N compounds, but the method should be equally applicable to other N sources (cf. Warren, 2013) as well as to other nutrients (e.g. K, P, S etc.). The indications of a strong interaction between mass flow and diffusion, the former leading to a strong increase in the rate of the latter (Fig. 4), call for further studies using a range of mass flow rates, as well as highlighting the need for a deeper understanding gained through modeling the interaction between mass flow and diffusion processes in relation to plant nutrient acquisition.
We are grateful to Margareta Zetherström for technical assistance in the dextran precipitation and amino acid analyses. We also thank Hyungwoo Lim and Bright Kumordzi for their contributions during the statistical analyses. Financial support from Formas, VR, SLU and the Kempe foundation is acknowledged.