Soil factors controlling mineral N uptake by Picea engelmannii seedlings: the importance of gross NH₄⁺ production rates

Summary

• • 
  Hydroponic studies suggest that plant nitrogen (N) demand determines the rate of mineral N uptake; however, field observations show N limitation to be widespread. Field experiments are needed to understand soil factors controlling mineral N uptake.
• • 
  We planted Picea engelmannii seedlings that had initially been grown from sterilized seeds, on a recently clearcut site. We applied a hybrid isotope dilution/pulse labelling technique to compare the gross production rate, concomitantly to the plant uptake rate, of soil mineral N. We also measured mineral N concentrations, microbial N, and percent ectomycorrhizal root tips.
• • 
  Gross NH₄⁺ production rate was the most important determinant of plant uptake rate. Exploratory path analysis suggested that plant uptake was also determined by microbial N, which was, in turn, determined by soil mineral N concentrations. Percent ectomycorrhizal root tips was negatively related to gross NO₃⁻ production rate and microbial N concentrations.
• • 
  We conclude that nutrient flux density is important in controlling plant uptake. Mycorrhizal colonization may alter N dynamics in the rhizosphere without affecting mineral N uptake by seedlings.

Introduction

Uptake of mineral nitrogen (N) by higher plants has been studied under hydroponic conditions that ensure a tight control over fixed experimental factors. For example, Glass et al. (2002) reported on a series of experiments that have elucidated molecular level mechanisms that regulate root transport of NH₄⁺ and NO₃⁻ as a function of solution concentrations of these two ions. Their findings are significant from an ecological perspective, because they suggest a plant's ability to adapt to fluctuating soil environments in order to optimize mineral N acquisition. This has been corroborated by split root hydroponic experiments showing that mineral N deprivation to half the root system of annual ryegrass (Lolium multiflorum Lam.) plants results in the compensatory doubling of mineral N influx by the other half of the root system within 24 h (Laine et al., 1998). Why is it, then, that in spite of these adaptive physiological traits, N remains the limiting nutrient for plant growth in many humus rich ecosystems (e.g. Pastor et al., 1987), or that mineral N uptake rates can vary substantially among neighbouring individuals (Kelly et al., 2004)? Clearly, intrinsic physiological controls on plant N uptake cannot completely override external soil factors controlling mineral N availability. Field experiments on mineral N uptake rates are needed to compensate for the limitations of hydroponic studies.

Boreal forest soils, including organic rich humus layers, typically display mineral N (i.e. NH₄⁺ + NO₃⁻) concentrations below 10 µg g⁻¹ d. wt (Lapointe et al., 2005). Although these concentrations are small relative to mineral N concentrations measured in many agricultural soils, once converted to approximate bulk soil solution concentrations, they remain within the range of mineral N concentrations (i.e. 100 µm−10 mm) over which high affinity root transport proteins regulate the influx of these two ions under hydroponic conditions (Engels et al., 2000; Glass et al., 2002). Thus, within this range of concentrations, N uptake is expected to obey N demand generated by the growth rate of the plant, and be independent of external mineral N concentrations. However, the influx of mineral N creates a mineral N depletion zone near the root surface, with concentrations several orders of magnitude lower than in the bulk solution (Ingestad & Agren, 1988). Under hydroponic conditions, it may be assumed that this boundary layer is negligible, but under field conditions, soil moisture content is normally below saturation resulting in nutrient diffusion gradients that extend the length of thin and tortuous water films along electrostatically charged soil particles. Thus, the difference in mineral N uptake rate between neighboring plants should depend on external soil factors that ensure the replenishment rate of mineral N in the depletion zone near the root surface.

In the present study, we investigated four properties of rhizosphere soil that we suspected were causally linked to mineral N uptake rates of outplanted Engelmann spruce (Picea engelmannii Parry ex. Engelm.) seedlings. Firstly, we hypothesized that the gross rate at which mineral N is produced in the rhizosphere represents the amount of mineral N that the root ‘can see’ per unit time. Many studies have shown that net mineralization rates greatly underestimate gross turnover rates of mineral N (e.g. Davidson et al., 1992; Stark & Hart, 1997; Stottlemyer & Toczydlowski, 1999), and no study to our knowledge has ever tested for a relationship between gross production rates (GPR) and plant uptake rates (PUR) of mineral N. The second factor that interested us was the mineral N concentration, mainly from the perspective that hydroponic studies predict a poor correlation between mineral N concentrations and PUR, due to the up- or down-regulation of transport systems into the root (Glass et al., 2002). The third rhizosphere property that we investigated was the microbial N pool, generally considered the most labile of soil organic N pools. Myrold (1987) related microbial N to the amount of N that can potentially mineralize and be made available to plants. Furthermore, root derived carbon (C) can maintain a large nonsymbiotic microbial biomass in the rhizosphere, compared with the bulk soil, and promote microbial acquisition of N from soil organic matter (Bradley & Fyles, 1995). Finally, the fourth rhizosphere property we investigated was the soil's ectomycorrhizal (ECM) inoculating potential, which is expected to vary randomly over short distances (Ettema & Wardle, 2002). Thus, we hypothesized a positive relationship between the percentage of root tips colonized by ECM fungi and mineral N uptake by seedlings.

In order to study the dependence of PUR on these four rhizosphere properties (referred to as ‘independent variables’), we conducted a field experiment using Engelmann spruce seedlings planted on a recently clearcut site. A disturbed site was preferred because it was expected to provide a wide spread of values for GPR, mineral N concentration, and microbial N. For the same reason, the seedlings had initially been grown from sterilized seeds and then outplanted for 1 yr, as this was expected to create a wide range of % ECM root tips. We developed a hybrid isotope dilution/pulse labelling approach to compare the GPR concomitantly to the PUR of mineral N. Other variables were measured according to routine methods. Finally, we used exploratory path analysis to investigate possible causal structures between independent variables and PUR.

Materials and Methods

Site description

The study was conducted on a 4-yr-old clearcut plot (1 ha) on Mount Mara, located near the town of Sicamous (51° N, 119° W) in the southern interior of British Columbia (BC). The site is located within the Wet Cold Engelmann Spruce – Subalpine Fir (ESSFwc) bio-geo-climatic zone (Meidinger & Pojar, 1991). Mean annual temperature is 1.2°C, and mean annual precipitation is 930 mm, of which about two-thirds falls as snow between late October and late May The surrounding forest is dominated by a 300-yr-old stand of Engelmann spruce and subalpine fir (Abies lasiocarpa (Hook.) Nutt.). Undisturbed soils are Humo-Ferric Podzols overlain by a 4 cm deep forest floor. The understory shrub and herb community includes white-flowered rhododendron (Rhododendron albiflorum Hook.), black huckleberry (Vaccinium mebranaceum Dougl. ex. Torr.), oval-leaved blueberry (V. ovalifolium Sm.), Sitka valerian (Valeriana sitchensis Bong.), oak fern (Gymnocarpium dryopteris (L.) Newman), foamflower (Tiarella unifoliata Hook.) and mountain arnica (Arnica latifolia Bong.).

Seedlings

Seedlings used in the present study were prepared as part of a larger investigation on the effects of ECM diversity on the nutrition of conifer seedlings after disturbance. The strategy involved planting nonmycorrhizal Engelman spruce seedlings in a clearcut, and allowing naturally occurring ECM fungi to colonize the roots. Briefly, seeds of spruce were surface sterilized and planted in a sterilized peat and vermiculite mixture (1 : 1 v/v). Seedlings were grown with photoperiod extension to 18 h in a glasshouse for c. 6 wk, and then transferred to the study site in early July 1999. The 800 nonmycorrhizal seedlings were planted in rows at regular 0.5 m spacing, at least 10 m from the edge of the clearcut. Seedlings were carefully removed from the potting mix and planted in a small furrow using a hand spade. The depth of the forest floor in which each seedling was planted varied between 0 and 2 cm, because of soil disturbance that had occurred during harvest operations. At time of planting, the shoots of seedlings (1–2 cm high) were covered by 10 cm high cages made of 5 mm wire mesh to protect them from herbivores. Seedlings were left to grow for 1 yr.

In situ¹⁵N labelling

In mid-June 2000, immediately following snowmelt, a plastic sleeve (10 cm dia. × 15 cm high) was inserted into the soil around 150 randomly chosen seedlings using a Gidding corer (Giddings Machine Co., Ft. Collins, CO, USA). Each soil core, complete with seedling and sleeve, was lifted out of the ground, the bottom covered with a plastic cap, and the capped core was replaced into the hole with soil packed around it to eliminate any air space. The purpose of the sleeve and cap was to contain the ¹⁵N-labelled solution that would be subsequently applied, within a defined soil volume. At the time of coring, the shoots of plants growing within a 15 cm radius of each seedling were excised to eliminate any shading. The coring process caused little disturbance to the soil around the seedlings because the soil was soft and moist. Seedlings were left to grow for another 4 wk before labelling, to allow roots and hyphae that may have been damaged during the coring process to re-establish within the core. By mid-July, when isotope solutions were applied, all seedlings had broken bud.

Fifty cores were injected with 20 ml of aqueous 1 mm (¹⁵NH₄)₂SO₄ solution (139.6 mg l⁻¹ at 99 atom %¹⁵N), another 50 cores with 20 ml of aqueous 2 mm K¹⁵NO₃ solution (216.6 mg l⁻¹ at 99 atom %¹⁵N), and another 50 cores with 20 ml distilled water. Solutions in each core were injected in five 4-ml aliquots at points located equidistant between the stem and the edge of the core. In order to distribute the solutions evenly throughout the cores, we used a syringe equipped with a 15-cm needle having four openings pointing outward. The needle was inserted 10 cm deep and the plunger was depressed gently as the needle was slowly pulled back up through the soil.

Fifteen minutes following injection (t = 0), half the cores allocated to each solution were destructively sampled. The cores were removed from the ground, the soil was extruded onto clean plastic sheeting using disposable rubber gloves, seedlings were carefully extracted from the soil, the shoots were excised from the roots, and both were rinsed with 0.05 m CaCl₂ and placed in separate plastic bags on ice. The remaining soil was sieved through a 5.6-mm mesh and placed in a plastic bag on ice. All instruments were rinsed with 0.05 m CaCl₂ and deionized water between the sampling of each core. The procedure was repeated for the remaining cores at 26 h after injection (t = 26). Soil and tissue samples were returned to the laboratory and stored at 4°C overnight.

Gross mineral N production rates

Gross production rates of NH₄⁺ and NO₃⁻ were estimated following the principles of isotope dilution (Kirkham & Bartholomew, 1954). Approximately 30 g of fresh soil from each core was weighed in an aluminum dish, dried at 105°C for 48 h, and reweighed to determine moisture content. A second 30-g subsample from each core was weighed in a 500-ml Mason jar, 100 ml of 1 m KCl solution was added to the jar, the jar was sealed and the mixture placed on a reciprocal shaker (130 cycles min⁻¹) for 60 min. The supernatant was filtered through a Whatman 42 cellulose filter paper than had been preleached with 1 m KCl. Mineral N concentration in each soil extract was measured colorimetrically using a Technicon II AutoAnalyser (Pulse Instrumentation Ltd, Saskatoon, Canada), with nitroprusside-salicylate reagent for determining NH₄ ± N, and sulfanilamide color reagent with a Cu-coated Cd reduction column for determining NO₃–N. Extracts were prepared for mass spectrometry by diffusion on acid traps according to Brooks et al. (1989), with the amount of N diffused adjusted to 300 µg-N with expected enrichment of c. 1% N, as suggested by Bradley & Fyles (1996). Devarda's alloy was added to extracts from soil samples injected with ¹⁵NO₃⁻ before diffusion to convert NO₃⁻ to NH₄⁺. Samples were encapsulated in Sn capsules and sent to the Stable Isotope Laboratory, University of California at Davis, for isotopic analysis.

Gross mineral N production rates can normally be calculated using zero-order equations derived by Kirkham & Bartholomew (1954), as modified by Hart et al. (1994):

• (Eqn 1)

where GPR = gross production rate

t = time

APE₀ = atom percentage excess of the labelled pool at t = 0

APE_t = atom percentage excess of the labelled pool at t = 26

[N]₀ = N concentration of the labelled pool at t = 0

[N]_t = N concentration of the labelled pool at t = 26.

However, for this equation to be valid, each t = 0 sample must be paired to one t = 26 sample, and both must be subsamples of the same experimental unit. In our study, given that t = 0 and t = 26 samples were from different cores, we estimated APE₀ and [N]₀ for each sample extracted at t = 26 as follows. First, we calculated the fraction of ¹⁵N recovered () in all t = 0 cores according to the following equation (Hart et al., 1994):

• (Eqn 2)

Because differences in values among t = 0 cores were very small relative to differences in values between t = 0 and t = 26 cores, we used the average value of all t = 0 cores ( = 0.59 and 0.65, for NH₄⁺ and NO₃⁻, respectively) to estimate APE₀ in each core extracted at t = 26 using the following equation (Hart et al., 1994):

• ( Eqn 3)

Similarly, we used [N]_t as an estimate of [N]₀ for each core extracted at t = 26, given that prior soil tests had shown very little net mineralization over the assay period.

This made eqn 1 invalid, however, and required that we use the following equation to calculate gross mineral N production rates (Kirkham & Bartholomew, 1954):

• (Eqn 4)

where M₀ = Mass of tracer plus nontracer of the labelled pool

H₀ = Mass of tracer of the labelled pool at t = 0.

H = Mass of tracer of the labelled pool at t = 26.

Mineral N

The mean concentration of NH₄⁺ and NO₃⁻ in each core during the assay was estimated as [N]_t.

Microbial N

The concentration of microbial N was measured by the chloroform fumigation – direct extraction method (Brookes et al., 1985). For chloroform fumigation, 50 ml of glass-distilled CHCl₃ (prefiltered through 2.5 g AlO₃) was placed with boiling chips in a glass dish on the bottom of a pressure cooker. A 30 g (f. wt) subsample of soil from each core was placed in a Mason jar, the jars were placed inside the pressure cooker on shelves lined with moist paper towel and fumigated under vacuum (21.3 kPa) for 24 h. After removal from the pressure cooker, fumigated as well as nonfumigated soils from each core were extracted in 1 m KCl solution, as described above, and the extracts were immediately frozen at −20°C until analyzed. Upon thawing, 5-ml aliquots of fumigated and nonfumigated soil extracts were syringe filtered through a 0.45-µm membrane filter, oxidized with 10 ml of alkaline K₂S₂O₈, and autoclaved at 121°C for 45 min converting all N present in a sample to NO₃⁻ (D’Elia et al., 1977). Total NO₃⁻ N was measured colorimetrically, as described above, and microbial N in each core was calculated as the difference between fumigated and nonfumigated subsamples divided by the total soil dry mass in the core.

Percent ECM root tips

The root system of each seedling was examined for ECM colonization under × 50 and × 400 magnification. The % ECM root tips of each seedling was noted.

Mineral N uptake rates

Seedling roots and shoots, including ECM fungal sheaths, were oven dried at 60°C for 72 h, ground in a Wiley Mill and passed through a 800 µm sieve. Subsamples were packaged in Sn capsules and sent to the Stable Isotope Laboratory (UC at Davis) for ¹⁵N analysis by continuous flow mass spectrometry. Assuming that isotope discrimination by the roots would be minor relative to the enrichment caused by labelling of the soil, plant uptake (PU) of each mineral N form was calculated as:

• (Eqn 5)

where SNP = soil N pool = (soil dry mass) × [N]_t

• 

EIP = excess isotope in the plant = (plant dry mass) × (atom percentage ¹⁵N excess in the plant).

Each PU value of each mineral N form was then converted to plant uptake rate (PUR) by dividing by time (i.e. 26 h = 1.083 d). Given that PUR may be biased by the amount of roots, we generated a new variable, which we called the relative plant uptake rate (RPUR), by dividing PUR by the dry mass of roots in each core.

Data analysis

Given the one-time sampling opportunity that was implicit to this novel experimental protocol, 18 potential data points were lost as a result of low [N]_t-values. Four additional data points were lost due to a manipulation error during microbial-N determinations, eight data points were deleted from percentage ECM counts because seedlings had less than one fully colonized root tip, and three data points representing either PUR or RPUR were deleted because of failed analyses at the isotope laboratory. The remaining 264 data points were used to determine the dependence of PUR and RPUR on the four independent variables. Simple linear regressions between each pair of variables were estimated independently for NH₄⁺ and NO₃⁻ by the least squares difference method using SPSS 11.0 software (SPSS Inc, Chicago, IL, USA). Regression parameters were considered different from zero when P < 0.05. Before analysis, all measured variables were log transformed to meet assumptions of normality and homoscedasticity, except for % ECM root tips, which was arcsin–square root transformed.

When two or more variables related significantly to plant N uptake, exploratory path analysis was used to investigate possible causal structures between the variables, using the SGS algorithm (Spirtes et al., 1993; Shipley, 2000a) as implemented in the EPA2 program (Shipley, 1997). A detailed explanation of the algorithm is given by Shipley (2000b) and only an intuitive description is given here. The first part of the algorithm constructs an undirected dependency graph, which links two variables if, and only if, they are dependent when all possible subsets of other variables in the model are mathematically fixed. The second part of the algorithm attempts to orient the undirected dependency graph based on ‘unshielded colliders’ in the graph. This is described in detail in Shipley (2000a) and the mathematical proof is given in Spirtes et al. (1993).

Results

The observed ranges of values measured across all cores in each assay are given in Fig. 1 as ‘box and whisker’ plots. The GPR of NH₄⁺ was significantly and positively related to both PUR (P = 0.001) and to RPUR (P < 0.001) (Fig. 2). PUR was also positively related to NH₄⁺ concentration (P = 0.026) and to microbial N (P = 0.045). The soil NH₄⁺ concentration and microbial N were significantly related to each other (P < 0.001), as were PUR and RPUR (P < 0.001). Given that three predictor variables (GPR, NH₄⁺ concentration, and microbial N) were each significantly related to the one response variable (PUR), exploratory path analysis was used to compare competing path models that would reflect all possible causal structures binding these four variables. It was found that only the ordering of the variables as shown in Fig. 3 could have produced the data (P < 0.05), although the small sample size resulted in no directed path diagram from being excluded at this alpha level. Assuming that PUR could only be a dependent variable, the accepted path model is therefore the one shown in Fig. 3, in which causal links are shown by the arrows and path coefficients express the standardized regression slope. This directed path model (P < 0.15) reveals that GPR was the most important determinant of PUR (i.e. regression weight of 0.55), but that PUR was also determined by microbial N, which was determined, in turn, by NH₄⁺ concentration. Thus, NH₄⁺ concentration had an indirect effect on PUR.

Figure 1. Box-plots showing the median and 25% quartiles (i.e. spread) for each of four independent (a, b, c, d) and two dependent (e, f) variables, in both the NH₄⁺ (hatched boxes) and NO₃⁻ (open boxes) assays. Outliers (black dots) are values greater than 1.5 times the spread. GPR, gross (NH₄⁺ or NO₃⁻) production rate; [N] (NH₄⁺ or NO₃⁻)-N concentration; PUR, plant (NH₄⁺ or NO₃⁻) uptake rate; RPUR, relative plant (NH₄⁺ or NO₃⁻) uptake rate.

Figure 2. Scatterplot matrix showing a separate plot (below the diagonal) for each possible pair of variables derived from the ¹⁵NH₄⁺ assay; PUR, plant NH₄⁺ uptake rate; RPUR, relative plant NH₄⁺ uptake rate. Coefficients of determination (R²) and associated P-values are shown (above the diagonal) for significant relationships only; ns, not significant (P > 0.05).

Figure 3. Directed path diagram showing the most likely causal structure (P < 0.15) between gross NH₄⁺ production rate (GPR), seedling NH₄⁺ uptake (PUR), microbial N, and NH₄⁺ pool size, obtained by exploratory path analysis. Structural path coefficients, displayed above their respective arrows, are the standardized slopes of the linear regressions and represent the effect sizes calculated by the model estimation program.

The GPR of NO₃⁻ was positively related to microbial N (P = 0.004) and negatively related to % ECM root tips (P < 0.027) (Fig. 4). The soil NO₃⁻ concentration and microbial N were significantly related to each other (P < 0.013), as were PUR and RPUR (P < 0.001). When data from all cores were pooled together, there was a significant negative relationship between microbial N and % ECM root tips (Fig. 5).

Figure 4. Scatterplot matrix showing a separate plot (below the diagonal) for each possible pair of variables derived from the ¹⁵NO₃⁻ assay; GPR, gross NO₃⁻ production rate; PUR, plant NO₃⁻ uptake rate; RPUR, relative plant NO₃⁻ uptake rate. Coefficients of determination (R²) and associated P-values are shown (above the diagonal) for significant relationships only; ns, not significant (P > 0.05).

Figure 5. Relationship between the percentage of mycorrhizal roots of seedlings (arcsin–square root transformed) and microbial N (log transformed) with data pooled from both ¹⁵NH₄⁺ and ¹⁵NO₃⁻ labelling assayst. B_std represents the standardized slope of the regression.

Discussion

NH₄⁺ uptake: the importance of GPR

For the NH₄⁺ uptake assay, a significant positive relationship was found between PUR and three rhizosphere properties, namely GPR, NH₄⁺ concentration, and microbial N. The proportion of the total variability in PUR attributable to GPR (i.e. coefficient of determination) was higher than that attributable to either NH₄⁺ concentration or microbial N. In addition, GPR was the only independent variable to show a significant positive relationship with RPUR. Given, as well, that GPR was orthogonal with both NH₄⁺ concentration and microbial N, we conclude that there is a strong and direct dependence of PUR on GPR. This relationship concurs with Ingestad & Agren's (1988) argument that nutrient uptake models based on external concentrations as the driving variable may be flawed, that nutrient flux density (i.e. the amount of nutrient made available per unit time) may be more important in controlling plant uptake. Their argument is based on data from a model system in which plants attained optimal growth rates when roots were fed with a rapid spray of very low nutrient solution. This is analogous to the situation found in many forest soils where mineral N concentrations and net N mineralization rates may be quite low compared with GPR (Stark & Hart, 1997). For example, Stottlemyer & Toczydlowski (1999) estimated that the GPRs of NH₄⁺ and NO₃⁻ in a boreal forest watershed were, respectively, 23 and 19 times greater than net production rates. In our study, the net amount of mineral N produced in each core over the 26-h assay was negligible, but the gross amount of NH₄⁺-N produced over the same time frame (based on an average soil dry mass of 200 g per core) was about 100 times the amount of NH₄⁺-N absorbed by seedlings.

Exploratory path analysis is an extension of the regression model used to compare the fit of the correlation matrix against two or more causal models. The SGS algorithm (Spirtes et al., 1993; Shipley, 2000b), as implemented in the EPA2 program (Shipley, 1997), is probably correct assuming a large sample size (> 100) and a homogeneous causal process generating the data. Given the smaller than optimum sample size in this study, we make use of an experiment-wise rejection level of 0.15 to select the best-fitting model for the advancement of theory, but the conclusions we draw must be considered provisional and subject to further verification. All the same, the path model shown in Fig. 2 corroborates the importance of GPR in controlling the uptake rate of NH₄⁺ by spruce seedlings. The second, less significant, causal path relevant to PUR was decomposed to include an indirect effect of NH₄⁺ concentration. In a linear system such as this one, the indirect effect is calculated by multiplying the two intermediary path coefficients. Thus NH₄⁺ concentration had a regression weight of 0.32 on PUR. It is conceptually more difficult to link microbial N directly to PUR, than it is to link GPR to PUR, as one is a pool and the other a rate. Adding more variables to the model will not change the order or the direction of this second causal path, but it may introduce new intermediary variables. Our results suggest therefore that microbial N is either a direct cause of PUR, or a parent of another cause.

A study by Zak et al. (1994) showed that microbial biomass changed predictably across broad gradients of above-ground net primary production, but ours is the first study to show a relationship between microbial N and plant NH₄⁺ uptake rates. That soil NH₄⁺ concentration would control microbial N suggests that NH₄⁺ is an important nutrient form for microbes in this soil and that, unlike seedling uptake rates, microbial uptake rates are directly dependent on external concentrations.

NO₃⁻ uptake

The salient feature of Fig. 4 is the lack of a significant relationship between PUR (or RPUR) of NO₃⁻ and any of the four independent variables. This may be due to the low NO₃⁻ uptake capacity generally reported for various spruce species. For example, Kronzucker et al. (1997) found that uptake rates of NH₄⁺ by nonmycorrhizal white spruce seedlings growing under hydroponic conditions could be 20 times greater than that of NO₃⁻ from equimolar solution. We see from Fig. 1 that the GPR of NO₃⁻, soil NO₃⁻ concentration, and PUR (or RPUR) of NO₃⁻ were, indeed, approximately one order of magnitude lower than the same variables in the NH₄⁺ assay. These lower amounts in the NO₃⁻ assay inevitably result in a higher coefficient of variation and a higher risk of committing Type II errors. We did not expect, however, that microbial N would relate differently to PUR in each assay, because this relationship is assumed to be impartial to which ion is being tested. Indeed, Fig. 1(c) reveals very little difference in the median concentration of microbial-N in either assay. It must be recognized that, in the NH₄⁺ assay, this relationship was significant but rather weak (P < 0.045), and could therefore have been missed in the NO₃⁻ assay due to the low sample size.

Soil NO₃⁻ concentrations and the GPR of NO₃⁻ were both positively related to microbial N. Our data might thus suggest that NO₃⁻ is an important nutrient form for microbes in this soil. Although microbiologists have generally affirmed that the preferred microbial mineral-N source is NH₄⁺ (Lin & Stewart, 1998), isotope dilution studies have repeatedly suggested that gross microbial uptake rates of soil NO₃⁻ are relatively high (e.g. Davidson et al., 1992; Stark & Hart, 1997), sometimes exceeding those of NH₄⁺ (Hart et al., 1997).

% ECM root tips

Certain isolates of ECM fungi are known to improve plant mineral N uptake by mediating the translocation of mineral N across the depletion zones that occur around the roots (Finlay et al., 1988). By contrast, there was no significant relationship between % ECM root tips and PUR (or RPUR) in the present study. Likewise, we found no relationship between the absolute number of ECM root tips per seedling (i.e. [% ECM root tips]×[# fine roots]) and PUR (data not shown). It may be that N taken up by external hyphae is slowly transported to the roots, such that a 26-h assay period is inappropriate to measure N uptake as a function of % ECM colonization. We do not believe this was the case, however, as the external fungal sheaths, which comprise most of the ECM biomass, were digested along with the plant material to determine PUR. There could be more plausible reasons for this. Firstly, it is conceivable that ECM fungi do not provide a ‘return’ to spruce seedlings in terms of nutrient acquisition until a certain age. Alternatively, it is increasingly being recognized that not all ECM fungal species and strains occupy the same niche; they may thus differ in their nutritional effects on the host plant (Read, 2002). This hypothesis will be tested and discussed in a subsequent paper emanating from this project (M. D. Jones et al., in preparation). A third hypothesis states that organic N, rather than mineral N, is the more important N source for some ECM strains (Michelsen et al., 1998). This seems unlikely in this case, however, because subsequent work with ¹⁵N-labelled amino acids at this site indicated that uptake of NH₄⁺ and NO₃⁻ by seedlings was equal to, or higher than, N uptake from organic sources (M. D. Jones et al., in preparation).

Read & Perez-Moreno (2003) have discussed the potential of some ECM fungi to depolymerize N-containing polymers, which is likely to regulate overall N cycling and plant N uptake (Bradley et al., 1997; Schimel & Bennett, 2004). Oddly, our results suggested that ECM root tips actually interfered with some aspects of soil N cycling, as exemplified by the negative relationship between % ECM root tips and the GPR of NO₃⁻ (Fig. 4), or between % ECM root tips and microbial N concentrations (Fig. 5). Our results are consistent therefore with the few studies that have shown inhibitory effects of ECM hyphae on nonsymbiotic soil microorganisms (Gadgil & Gadgil, 1971; Rasanayagam & Jeffries, 1992; Olsson et al., 1996). Bending (2003) reminds us that this is an issue of soil microbiology that remains contentious and that attempts to corroborate negative interactions between ECM fungi and soil decomposers have been inconsistent. For example, Olsson & Wallander (1998) have shown that an inhibitory effect of ECM fungi on soil microbial communities may depend on specific soil conditions, and should not be considered a general characteristic. This is further supported by Koide & Wu (2003) who provided evidence that ECM fungi inhibit soil decomposers specifically in dry weather conditions. We can only conclude that, in the present study, the extent of ECM colonization in young spruce seedlings altered N dynamics in the rhizosphere, but we have no evidence that this affected mineral N uptake by spruce seedlings.

PUR vs RPUR

The variable RPUR was derived in order to correct for differences in root mass between cores. PUR and RPUR were closely related indicating small differences in root mass between cores. To our surprise, correcting for root mass actually weakened the relationship between plant N uptake and microbial N, or between plant N uptake and soil NH₄⁺ concentration (Fig. 2). We conclude that plant N uptake was by and large controlled by N cycling characteristics of the rhizosphere soil.

Future research

Previously, hydroponic studies have suggested that the PUR of mineral N obeys N demand generated by growth rates, rather than growth rates obeying PUR generated by mineral N availability (Glass et al., 2002). Our study has shown that, under field conditions, PUR of NH₄⁺ by Engelmann spruce seedlings is quite variable and is simultaneously related to several rhizosphere properties. The extent to which the GPR and PUR of NH₄⁺ remain related under different field situations needs to be verified. For example, future research should test the relative importance of biophysical factors such as heat, water, and sunlight, in modifying rhizosphere properties and limiting nutrient uptake by roots.

Acknowledgements

We gratefully acknowledge financial support from the Natural Sciences and Engineering Research Council of Canada, Strategic Research Program Grants.