Competition between roots and soil micro-organisms for nutrients from nitrogen-rich patches of varying complexity

Authors


  • ‡Present address: Department of Plant and Soil Sciences, University of Aberdeen, Aberdeen AB24 3UU, UK.

Dr A. Hodge (fax + 44 1904 432860; e-mail ah29@york.ac.uk).

Summary

1 We used Lolium perenne plants grown in microcosms to investigate the responses of root demography, plant N capture, soil fauna populations and microbial community profiles to five organic patches containing the same amount of N but differing in their chemical and physical complexity and C : N ratio. All patches were dual labelled with 15N/13C. Control patches contained the background sand : soil mix only.

2 There was rapid decomposition in, and plant N capture from, the patches of lowest C : N ratio. Early in the experiment 13C was detected in the soil atmosphere and 15N in the shoots. No 13C enrichment was detected in the plant material.

3 The rate of root production was slowest in the most complex patch (L. perenne shoot material) but accelerated when patches were simpler and had lower C : N ratios. There was no difference in root mortality between treatments.

4 Nitrogen concentrations of shoots and roots and shoot biomass were greater in the N-containing patches than controls, except for the most complex patch, while root biomass did not differ with treatments.

5 Total plant N capture was 45–54% of that initially added in patches that had a C : N ratio < 4. However, in the most complex patch (C : N ratio c. 21 : 1) plants captured only 11% of the N added.

6 Biomass of microbial-feeding protozoa was related to soil NO3-N concentration in the patch but not to numbers of microbial-feeding nematodes. Patches of greater complexity increased the metabolic diversity of the microbial community (i.e. the number of substrates used in a Biolog GN plate) and altered the pattern of substrate utilization.

7 At harvest, the amount of patch-derived N estimated to be in the microbial biomass was much smaller (i.e. 7–13%) than in the plant tissues. Thus, plants were highly effective competitors with micro-organisms when capturing N supplied in patches with a low C : N ratio.

Introduction

It is well established that plant roots proliferate and/or alter nutrient uptake kinetics upon encountering nutrient-rich zones or patches (reviewed by Robinson 1994; Robinson & van Vuuren 1998). Proliferation of roots in these nutrient-rich patches is widely believed to be a major mechanism by which roots forage in their environment to enhance nutrient capture (de Kroon & Hutchings 1995). However, most work on root proliferation responses has been conducted under controlled laboratory conditions or has used the addition of inorganic nutrient solution to generate patches. Nutrient availability varies greatly in soil, even at distances of only a few centimetres (Jackson & Caldwell 1993; Stark 1994; Farley & Fitter 1999a). Further, in most ecosystems, nutrients such as N are derived mainly from organic matter introduced into soil as discrete inputs of litter, roots and animal corpses, etc. Consequently, those inputs vary widely in their physical and chemical complexity, their C : N ratio, and the extent to which they create nutrient-rich patches. Fitter (1994) identified three important attributes for patches on a spatial and temporal scale, their extent (size and duration), distribution (pattern and predictability) and number (abundance and frequency), that may influence the response by plant root systems. Although it may seem intuitively obvious that patch attributes can influence nutrient capture from the patch and, consequently, plants’ responses to patches, this has seldom been investigated.

In natural soil systems any nutrient-rich patch will be the site of intense activity of both microbes and roots (Christensen et al. 1992; Griffiths et al. 1994). Thus, root systems will have to compete with micro-organisms for the available nutrients. The release of N from organic patches may be driven largely by the decomposer organisms in the soil food web (de Ruiter et al. 1993). The biochemical breakdown of organic matter is carried out by the microbial (bacterial and fungal) community, while the cycling of N and other plant nutrients is more a function of the microbial-feeding fauna, in particular nematodes and protozoa (Griffiths 1994). Because of the rapid turnover of micro-organisms when active, it is recognized that population sizes of nematodes and protozoa are better indicators of preceding microbial productivity than the microbial populations themselves (Christensen et al. 1992). The microbial community is regulated by the substrates available in soil, and results based on community-level physiological profiling have revealed large changes in microbial community structure with different organic substrates (Bossio & Scow 1995; Lupwayi et al. 1998; Sharma et al. 1998). Community-level physiological profiles of soil microbes indicate the relative ability of the microbial community to utilize a range of individual carbon substrates. Thus, by looking at both microbial community structure and microbial-feeding fauna, a clear picture can be obtained of the microbial involvement in decomposition of organic matter patches. Previously we have suggested that the microbial community is more important than plant attributes in controlling patch exploitation per se when a patch of high C : N ratio (c. 31 : 1) is applied (Hodge et al. 1998). Here, we examined how a range of different patches of varying C : N ratio, chemical and physical complexity influenced patch exploitation by plants when in competition with micro-organisms.

We tested the effect of addition of different organic patches that varied in both their physical and chemical complexity to examine how this affected root proliferation within the patch zone and subsequent plant N capture and total biomass production. Plants were grown in microcosm units to which one of five different types of organic patch was added, providing the same total amount of N to each unit. Three of the patch types (i.e. algal amino acids, algal lyophilized cells and l-lysine) examined had similar C : N ratios, while the other two had a ratio either an order of magnitude lower (urea patches) or higher (Lolium perenne shoots patches). Although the algal lyophilized cells had a similar C : N ratio to that of both the amino acid patches, the algal cell material was both physically and chemically more complex. Some plants may also acquire N as intact amino acids (Reinhold & Kaplan 1984; Lipson & Monson 1998), although the extent to which that happens when these plants have access to multiple N sources is unknown. If quantitatively significant, the direct uptake of amino acids by plants could ‘short-circuit’ the microbial mineralization step. It could lead to greater or faster N capture from patches rich in amino acids compared with from more complex patches. Most N in chemically and physically complex organic patches such as plant and animal remains is in the form of protein. Plants can capture that N only after proteolysis has liberated low molecular weight forms of N, a process normally carried out in soil by micro-organisms.

All organic patches added were dual labelled with 15N and 13C so that plant N capture and patch decomposition (as 13CO2 released) could be followed with time. In addition, patches with the highest and lowest C : N ratios were added to microcosm units that contained no plants (unplanted units) so that the patch decomposition in the absence of roots could be followed. Plant tissue enrichments with 13C would also indicate uptake of organic carbon from the patches. Further, if there was a relationship between 13C and 15N enrichments in the plant tissues this would be good evidence of uptake of intact organic compounds from the patches (Näsholm et al. 1998). Root production and mortality were monitored in situ by the use of mini-rhizotrons, as measurements of root biomass alone are insufficient to determine either total root production (Hendrick & Pregitzer 1992) or successful patch exploitation (Fitter 1994).

We tested the following hypotheses. (i) Root production and mortality rates would both increase as the complexity of the patch decreased, because nutrients would be more available for root capture in the simplest patches, and root turnover generally increases in fertile soils (Aber et al. 1985) so we would expect the same in nutrient-rich patches. (ii) Plant N capture from the patch would decrease as the C : N ratio of the patch increased, because N immobilization by the microbial biomass would occur. In the patches of similar C : N ratio but less physically and chemically complex (i.e. the amino acid patches vs. the algal cell patch), plant N uptake would be greater as microbial decomposition would be bypassed. (iii) Plant growth would be depressed in the most complex patch (L. perenne shoots) due to immobilization of nutrients by the microbial biomass. (iv) Because of a lack of competition from roots, protozoan biomass would be greater in unplanted microcosm tubes; this was tested for the simplest (urea) and most complex (L. perenne shoots) organic patches. (v) Physiological changes in the microbial community would occur as a result of both the nature of the added patch and whether microcosm units were planted or unplanted.

Materials and methods

Experimental design

Full details of the experimental design are given in Hodge et al. (1999a). Briefly, plants were grown in microcosm tubes made out of a section of PVC pipe (length 20 cm, internal diameter 10 cm), which had two holes cut to allow insertion of a glass mini-rhizotron tube (25 cm long × 2.2 cm external diameter) at an angle of 45° to the horizontal. In addition, a smaller hole was cut to allow placement of a solution injection/gas sampling tube (15 cm long × 0.3 cm internal diameter) directly above the mini-rhizotron tube. A 2-cm section (10 cm wide at the top and 7.5 cm wide at the base) was cut from a PVC funnel and was placed in the top of each tube to direct the roots into the middle section where the patch was to be inserted. The mini-rhizotron and gas sampling tubes were sealed with rubber bungs. A wooden pole (15.5 cm long × 2.7 cm external diameter) fitted with a plastic attachment at its base, shaped to fit securely on to the mini-rhizotron tube and cover the base of the gas sampling tube, was placed vertically so that its base was at 10 cm depth (i.e. directly above position 5, corresponding to video image frame 5, of the mini-rhizotron tube). Once the seedlings had developed suitably the pole could be removed to allow precise placement of the patch while ensuring minimal disturbance to the system. Eight microcosm tubes were contained within each of six large (60 × 40 × 30 cm) freely draining insulated boxes containing a mixed turf of Trifolium repens L. (white clover) and Lolium perenne L. (perennial ryegrass) plants, to buffer the microcosm tubes against fluctuations in external temperature and to produce a realistic microclimate around the tubes. The boxes were maintained in a heated glasshouse from the time of seed planting. Each microcosm tube was filled with a 50 : 50 mixture of sand : soil as described in Hodge et al. (1999a). The soil was a loam with a pH of 6.8 (in 0.01 m CaCl2) collected from an experimental garden at the University of York, UK, and sieved through a 2-mm mesh. The background total N concentration of the sand : soil mix was 0.9 mg N g–1.

Twenty-four seeds of Lolium perenne L. cv. Fennema (perennial ryegrass), supplied by Johnson Seeds, Lincolnshire, UK, were planted into each microcosm tube on 18 July 1997 (12 in an outer circle of 6 cm diameter and 12 in an inner circle of 4 cm diameter). All seeds in the microcosm tubes had germinated after 1 week. Seedlings were left to develop for a further 96 days before the experimental treatments were imposed. The experiment ran for 49 days between 29 October and 17 December 1997. The mean temperature over the duration of the experiment was 18.3 °C (SE ± 0.18), with a mean daily maximum of 21.0 °C (SE ± 0.33) and mean daily minimum temperature of 14.0 °C (SE ± 0.10). The plants were grown under 16-h days with natural light supplemented by 400 W halogen bulbs. Photosynthetically active radiation (PAR) flux was recorded weekly and averaged 205 µmol m–2 s–1 at plant level.

Organic patch material

To test the response of L. perenne plants to added organic substrates of differing physical and chemical complexity, and their ability to acquire N during its decomposition, the following treatments were applied: (1) 2.41 g of dual labelled (15N/13C) Lolium perenne L. cv. Miranda shoot material; (2) 1.22 g of algal lyophilized cells; (3) 0.36 g of algal amino acids; (4) 1.5 ml of 1.07 m l-lysine solution; (5) 1.5 ml of 1 m urea; and (6) 1.5 ml of deionized H2O (control treatment). The shoot material for treatment 1 was produced as described in Hodge et al. (1998). For treatments 2–5 all 15N- and 13C-labelled compounds were purchased from Promochem Ltd (Hertfordshire, UK) and, except for the l-lysine solutions (i.e. alpha-15N and 1-13C), were uniformly labelled. The labelled compounds were then mixed with unlabelled compounds, and the atom percentage enrichment, mg 15N, mg C, mg 13C and C : N ratio of the added patches are given in Table 1. All treatments, except the control, were added to provide the same quantity of N (c. 45 mg) per tube. In addition, to follow the decomposition of the highest and lowest C : N ratio patches in the absence of living roots, L. perenne shoot material (i.e. treatment 1 above) and urea (i.e. treatment 5 above) patches were added to microcosm units containing no plants. Thus, there were eight treatments in all (Table 1). One treatment was applied per microcosm tube, hence each of the six boxes contained one replicate of each treatment.

Table 1.  Elemental and isotopic compositions of material in the patch treatments added per microcosm tube. Note that treatments 1–5 contained the same amount of N (45 mg) but differed in their C : N ratio
Patch treatment mg 15NAtom % 15Nmg Cmg 13CAtom % 13CC : N ratio
1. L. perenne shootsA. With plants6.0212.6931.012.41.2320.7 : 1
B. Without plants6.0212.6931.012.41.2320.7 : 1
2. Algal lyophilized cells 3.487.7146.013.910.53.2 : 1
3. Algal amino acid mixture 8.6019.1138.020.714.93.1 : 1
4. l-lysine 0.962.1108.01.361.262.4 : 1
5. UreaA. With plants15.133.619.24.0721.20.4 : 1
B. Without plants15.133.619.24.0721.20.4 : 1
6. Deionized H2O (control) 

The patches were added by removing the wooden pole from the microcosm unit then placing the appropriate patch material (mixed with 7.5 g dry weight of the sand : soil mix) at the base of the space created. The remaining space was filled with the sand : soil mix only. All patches were placed at a depth of 10 cm in the microcosm unit and at 20% moisture content.

Measurements and harvests

To monitor respiratory 13C release from the patch, soil gas samples were taken and analysed as described by Hodge et al. (1999a). Soil gas was sampled 0.04, 0.5, 3, 6, 9, 13, 17, 21, 25, 29, 33, 37, 41, 45 and 49 days after patch addition. These sampling times were chosen to allow the duration of the decomposition of the patch to be followed.

To follow the uptake of N from the patch into the shoots of L. perenne, a single plant was cut at the soil surface in each tube on alternate gas sampling dates starting at 0.5 days, oven-dried at 60 °C to a constant dry weight, weighed, milled and analysed for total N and 15N (see below). Shoots were always sampled after soil gas samples were taken. Sampling individual plant shoots in this way allowed a degree of replication and a frequency of measurements that would have been logistically impossible otherwise. Furthermore, as L. perenne swards are often cut or grazed it was a realistic sampling regime. The shoots of the harvested plants regrew over the experimental period but were not sampled again until the final harvest, when all remaining shoot material, except one individual (day 49 sample), was bulked together for analysis.

On day 49 the microcosm units were removed from the boxes and harvested. Once the mini-rhizotron and gas sampling tube had been removed, each soil core could be removed intact from its tube. The core was then cut into three sections, top, middle (containing the patch zone) and bottom, each of 6 cm depth. The middle section was divided further by separating out the immediate patch zone (i.e. an inner cylinder of 5 cm diameter to allow for diffusion of the patch from the original 2.7-cm hole), which was used for further analysis (Hodge et al. 1999c). The remaining shoots were removed from the top section and oven-dried at 60 °C to a constant dry weight, weighed and analysed as below.

Root demography

After 1 day and at each gas sampling occasion from 6 days, images of roots were recorded every 2 cm along the upper surface of the mini-rhizotron tube using an Olympus OES swing-prism boreoscope with fibre-optic light source and a Sony PVS1 portable video system. Each video frame image was c. 7 mm in diameter.

The video images were captured onto a PC fitted with a frame-grabbing card. Exact overlays of all visible roots were recorded manually using an acetate sheet attached to the screen. Roots were identified on the acetate sheet with a unique code number so survival over with time could be monitored. From the data on individual roots, root birth and death rates were calculated over time. For this analysis, data were gathered from video frame 5 only (10 cm depth, i.e. the site of patch addition) and compared with root dynamics at the same depth in the control treatment.

Plant and soil analysis

The roots extracted from the different sections (top, middle, bottom) at harvest on day 49 were washed thoroughly, oven-dried at 60 °C to a constant dry weight, weighed and milled. The shoot material was also oven-dried and weighed. A subsample of the root and shoot material was analysed for total C, N, 13C and 15N by continuous-flow isotope ratio mass spectrometry (CF-IRMS). The concentration of N from the patch recovered in the shoots harvested with time was calculated as described in Hodge et al. (1999a).

Subsamples of the soil from the different sections (top, middle, bottom) of each tube were used for determination of moisture content (105 °C) and for total C, N, 13C and 15N analysis. Soil from the patch addition zone (i.e. removed from the middle section of the core; see above) was used to determine the inorganic N content of the soil, protozoa and nematode enumeration (as described in detail by Hodge et al. 1998). Briefly, 5 g of soil was suspended in 20 ml sterile Neff's modified amoeba saline (NMAS; Page 1967). An aliquot of the suspension was serially diluted in nutrient solution, incubated at 15 °C and the numbers of protozoa (flagellates, ciliates and naked amoebae) determined microscopically by a most-probable-number technique. Protozoan biomass was calculated assuming an average size of 50 µm3 per flagellate, 400 µm3 per amoeba and 3000 µm3 per ciliate (Stout & Heal 1967) and a dry weight conversion factor of 0.212 pg µm–3 (Griffiths & Ritz 1988). Nematodes were extracted directly from 4 ml soil suspension by flotation in colloidal silica (Griffiths et al. 1990). The soil suspension produced was also used to inoculate a Biolog (Biolog, Hayward, CA) GN plate, to determine the community level physiological profile (CLPP) of the microbial community (Garland & Mills 1991). Sufficient soil suspension was diluted with 25 ml NMAS to give an absorbance of 0.4 at a wavelength of 595 nm, and 150 µl inoculated into each well of a Biolog GN plate. The plates were incubated at 15 °C and the absorbance of each well at 595 nm was measured daily with an automatic plate reader for 5 days.

The percentage of patch N contained within the microbial biomass was estimated from the values obtained for the protozoan biomass using the following equation:

% N in microbial biomass = 100 × (mean protozoan biomass × microbial N derived from urea/protozoan biomass from urea)(1)

This calculation is based on the assumption that protozoan biomass is a reliable indicator of microbial biomass (Christensen et al. 1996) and that all the patch N remaining in the planted urea patch was contained within the microbial biomass. In soil, urea is rapidly converted to ammonium, so that 15N in added urea would be either incorporated into plants and microbes or present as inorganic N either in solution or adsorbed onto soil particles. The latter would be negligible in the presence of roots while the other sinks were measured. The percentage of patch N lost from the system was calculated as:

100 × [amount of N added in the patch – (patch N in plant + patch N left in patch soil)](2)

and the patch N that remained as unaltered substrate in the soil as:

% total patch N left in soil – % contained within the microbial component(3)

Statistical analysis

Data recorded at the final harvest only (day 49) were analysed using a one-way anova. The numbers of nematodes and biomass of protozoa were transformed logarithmically prior to analysis to normalize skewed distributions. Previously, we (Hodge et al. 1998), observed a relationship between patch soil NO3 concentrations and log protozoan biomass after addition of inorganic N patches of varying concentration. Further, as many nematodes feed on protozoa a relationship between these two taxa may be expected. Thus, to examine the nature of these relationships in the present study an analysis of covariance of patch NO3-N concentration and protozoan biomass and protozoan biomass and nematode numbers was conducted.

Data on samples taken over the duration of the experiment (i.e. soil gas samples, individual shoots and root demography) were analysed using the general linear model (repeated measurements) in SPSS v. 7.0 (SPSS 1996). In all cases, a randomized block design was used. Differences referred to in the text were statistically significant with P < 0.05, unless otherwise stated. For comparisons between treatments, the Bonferroni mean comparison test was applied. Root birth rates were estimated from regressions of mean cumulative root births for the period days 17–48, when rates were linear.

Microbial growth on a particular substrate in a Biolog well causes the tetrazolium dye present in each well to turn red in a quantitative relationship. Colour development over time on each substrate therefore gives the potential growth curve for the microbial community on that particular substrate. The time–course profiles of the Biolog data were analysed from the area under the colour development profile (Hackett & Griffiths 1997). The 95 areas-under-the-curve (i.e. one for each substrate) were analysed using both principal component analysis and canonical variate analysis, with GENSTAT v. 5 release 3.2 (Payne et al. 1993), to distinguish between treatments. The number of positive wells, i.e. those with an optical density greater than 1.4 times the control well (which has no added carbon substrate), on each day was analysed using analysis of variance as above.

Results

Root demography

At the early stages of the experiment, more roots were found in the H2O (control) patch zone than in all other treatments except urea. After day 17, the net number of roots in the control did not differ significantly from any of the other treatments (Fig. 1a) and by day 48 was the lowest value of all. The net number of roots visible in the urea patch was significantly greater than numbers in the algal cells from days 9 to 37, except at day 21, and also greater than in the l-lysine patch at day 9 and the L. perenne shoot material at day 37. After day 41 there was no significant difference among treatments in net numbers of roots in the patch addition zone.

Figure 1.

Roots per frame in the patch addition zone (10 cm depth) for (a) net numbers of roots and (b) cumulative root births. Data are means (n = 6) with SE bars. For (b) the regression coefficients of a linear regression fitted to these data for the period 17–48 days were: Lolium shoots, 0.264ab; algal amino acids, 0.354bc; algal lyophilized cells, 0.344c; l-lysine, 0.353c; urea, 0.342abc; H2O, 0.205a. Coefficients not sharing superscripted letters were significantly different at P < 0.05, based on 95% confidence limits. R2 values for regressions were all > 0.95 except for urea (R2 = 0.89).

Cumulative births (Fig. 1b) followed the same pattern as net numbers of roots. Birth rates in the algal amino acids, algal cells and l-lysine patches were all significantly faster than in the H2O control. In the L. perenne shoot patch, root birth rate did not differ from either the control or the algal amino acid patch but was significantly slower than the algal cell and l-lysine patches. Root birth rate in the urea patch was least well described by a linear regression, because of the large number of roots produced between days 20 and 25 (Fig. 1b).

At day 1 instantaneous root births were greater in the control than in all other treatments except urea (data not shown). At day 17 instantaneous births in the algal amino acids treatment were significantly higher than those in the algal cells, but neither of these treatments differed from any of the remaining treatments. There was a peak in instantaneous root births at day 25 in all treatments, and differences among treatments were not significant. In contrast to the net numbers, cumulative and instantaneous birth data, no significant differences were observed in either the cumulative deaths or the instantaneous deaths of roots at any time (data not shown).

Plant biomass and total n capture

At final harvest, shoot biomass was significantly greater than the control in all treatments except the L. perenne patch (Fig. 2a). There was no difference among treatments in root biomass (Fig. 2a), nor in the root biomass in the middle (patch addition) section expressed as a fraction of total root biomass (data not shown).

Figure 2.

Shoot (□) and root (▪) (a) biomass and (b) N concentration (mg N g–1) at final harvest (day 49). Different letters show significant (P < 0.05) differences between treatments. In (a) significant differences occurred between the shoots only, while in (b) the pattern of significant differences was the same for the root and shoot material. Data shown are means (n = 6) with standard errors.

N concentrations (Fig. 2b) and contents (data not shown) of roots and shoots showed a similar response across treatments. Nitrogen concentration and content were least in the H2O control and L. perenne shoot treatment and did not differ among the other four treatments. Nitrogen was more concentrated in shoot than root tissue (Fig. 2b).

N uptake from the organic patch, and patch decomposition

Plants progressively captured N from the patch and transferred it to the shoots, as shown by shoot 15N enrichment relative to the patch (Fig. 3). However, shoot 13C enrichment was never greater than background.

Figure 3.

Concentration of 15N in the individual shoot material harvested with time. Different letters show significant (P < 0.05) differences between treatments at each time interval. Data are means (n = 6) with SE bars.

Initially (days 1 and 6), N acquisition rate measured by 15N appearance in shoots (Fig. 3) was greatest from the urea patch, but from day 13 onwards (except at day 37) similar amounts of N were captured by L. perenne from urea (lowest C : N ratio) and algal amino acids. The other two N treatments with moderate C : N ratios, l-lysine and algal cells, produced slower rates of 15N capture, although after day 29 there were few differences between these four treatments. In contrast, N capture from the L. perenne patch, with a much higher C : N ratio, proceeded even more slowly and always caused the least enrichment of 15N in shoots, mostly significantly less than all other treatments.

Overall, total amounts of patch N recovered in the roots were small (i.e. 2–6% of the N originally added). Roots of L. perenne seedlings grown in the presence of a L. perenne shoot patch contained the least patch N, while those grown in the urea and both algal patches contained the most. The roots recovered at harvest from the top and middle (including the patch zone) contained similar amounts of patch N, while those from the bottom section had generally two to three times less N (although c. six times in the case of the L. perenne shoot patch). Roots, like shoots, were never 13C enriched.

The total amount of patch N captured by the L. perenne shoots was calculated at day 49 by adding the patch N in the shoot material removed over the course of the experiment to that recovered in the remaining shoots at harvest. The shoots generally contained c. 8–11 times more patch N than the roots, although shoots from L. perenne patches contained only four times the amounts in the roots. Shoots of L. perenne grown in the presence of a L. perenne shoot patch captured only 9% of the total patch N, compared with 40–49% for the other four treatments. Thus, by day 49 the L. perenne swards (roots and shoots) grown in the presence of a L. perenne shoot patch had captured c. 11% of the total N originally available. In contrast, swards in the other four treatments captured 45–54% of the original patch N.

Amounts of 13C recovered (as 13CO2) from the soil gas, as a percentage of the amount of 13C originally added, peaked for most treatments at days 0.5–3 (Fig. 4). The algal cell patches and planted L. perenne shoot patches were generally slower to show a peak of 13C release (i.e. day 9 and 13, respectively) than all the other treatments (Fig. 4). Consequently, at day 13 more patch 13C was recovered as 13CO2 from the planted L. perenne shoot patches than from all the others, which by then were declining. Statistical analysis of just the two treatments that contained both planted and unplanted microcosm units (i.e. urea and L. perenne shoot material; Fig. 4) showed that the presence of plants had no significant (P = 0.763) main effect on 13C release from the patch.

Figure 4.

Amount of 13C (expressed as a percentage of µg 13C initially added in the patch) recovered over time from the soil gas. Data are means (n = 6) with SE bars. Note the time axis is plotted as square-root time.

At harvest, 13C was above background only in the two algal treatments and both L. perenne shoot patches. Soil from the unplanted L. perenne shoots treatment retained the largest fraction of added 15N. Recoveries from the planted L. perenne shoot patches were smaller but not significantly so. In the other treatments, 15N recoveries from the patch were much smaller (Table 2). The presence of plants decreased recoveries of patch N from the urea tubes by day 49. Less patch 13C than 15N was recovered from the L. perenne shoot patches. There was a positive linear relationship between mg 15N and 13C remaining from the L. perenne shoot patches in both the planted and unplanted microcosms, suggesting that release of N and C from the patches was coupled. However, in the presence of plant roots the slope of the regression was 12.8, compared with 5.13 in the unplanted microcosms (data not shown). Thus, in these patches, for a given amount of patch decomposition (as indicated by 13C release) N was released more readily when plant roots were present. There was no significant relationship between the mass of 15N and that of 13C in the algal amino acid or algal cell patches, even though more 13C (i.e. 16–19%) remained in the algal amino acid and cells patches than in L. perenne shoot patches, and levels were comparable to amounts of 15N recovered.

Table 2.  Nematode numbers, protozoan biomass and fractions (%) of substrate (Table 1) N present in plants, microbes and soil at the end of the experiment. Values followed by different letters are significantly different. Data shown are means (n = 6) with SE in parentheses
%
Patch treatmentNematode
numbers (g–1)
Protozoan biomass
(ng g–1)
Plant N derived
from substrate
15N recovered
in patch soil
Total substrate N
left in soil
Substrate N
lost from system*
Microbial N derived
from substrate
Residual
substrate N
  • *

    Calculated as: 100 – (%plant N derived from substrate + %total substrate N left in soil).

  • † 

    Calculated as 100 × (mean protozoan biomass × microbial N derived from urea/protozoan biomass with urea). This assumes that protozoan biomass is a reliable indicator of microbial biomass (Christensen et al. 1996).

  • ‡ 

    Calculated as %total substrate N left in soil –%microbial N derived from substrate.

  • § 

    It was assumed that all urea-derived N remaining in the soil was in microbial biomass.

L. perenne shoots (with plants)189 (71) a38 500 (16700) b11 (1.6) a69 (9.2) a77 (8.9) a121265
L. perenne shoots (without plants)89 (8) ab102 000 (41100) a88 (7.9) a95 (8.5) a51382
Algal amino acids55 (13) bc7790 (4790) c54 (1.7) b16 (0.5) b20 (1.3) c25911
Algal cells44 (14) c1690 (383) cd45 (2.1) b27 (1.5) b27 (1.5) c28918
l-Lysine54 (9) bc4040 (1650) c46 (4.6) b10 (1.8) c10 (1.8) d4590
Urea (with plants)62 (7) abc789 (238) de53 (1.4) b5 (0.4) d7 (0.6) d407§0
Urea (without plants)6 (1) d280 (34) e11 (2.1) c46 (6.6) b54640
Deionized H2O (control)45 (7) bc346 (64) e

Total recoveries of 15N from the soil in the entire tube were in the order: L. perenne (unplanted), L. perenne (planted) > urea (unplanted) > algal cells, algal amino acids > l-lysine, urea (planted) (Table 2). No 13C above background was detected in either the top or bottom soil sections.

Soil inorganic n and microfauna

At harvest, there was no significant difference in the concentration of NH4+-N in the soil from the various treatments, and amounts recorded were small (treatment mean = 0.33 µg g–1; P = 0.888). Concentrations of NO3-N were greater than for NH4+-N. The concentration of NO3-N (14.7 µg g–1) in the unplanted L. perenne patch was greater than in all other treatments. The NO3-N concentration in the planted L. perenne patch soil, although less than half of that of the unplanted tubes, was greater than in all other treatments except in those containing algal cells. NO3-N concentrations in the other treatments were no greater than for the H2O control treatment.

Nematode numbers were only significantly different from the H2O control in the planted L. perenne shoot material patch and the unplanted urea treatment (Table 2). Of the planted treatments, patches with L. perenne shoot material contained more nematodes than patches with algal amino acids, algal cells and l-lysine, but not patches with urea. There were no significant differences between the patch materials apart from L. perenne. Unplanted soil amended with L. perenne shoot material contained more nematodes than unplanted soil amended with urea (6 g–1), which contained fewer nematodes than all other treatments.

Protozoan biomass was significantly greater in the L. perenne shoot patch treatments (Table 2) than in all other treatments. The algal amino acid and l-lysine patches had greater protozoan biomass than both (i.e. planted and unplanted) urea patches and the control patch, but were no different from those in the algal cell patch. Protozoan biomass in the planted and unplanted urea patches was similar to the H2O control patches, although protozoan biomass from the urea-planted tubes was not different from the algal cell patch. Between 7% and 13% of the patch N was estimated to be within the microbial biomass, with the higher values associated with the most complex substrates (Table 2). There was an inverse relationship between percentage patch N lost and the patch C : N ratio (i.e. regression equation, % patch N lost from the system = 40.1 – 9.92 log initial patch C : N ratio, P = 0.003, F1,6 = 27.02, R2 = 81.3%), with the greater loss of 15N occurring in the two lowest C : N ratio patches (l-lysine and urea) and the least from the most complex patch (L. perenne shoot material), which also contained the most unaltered patch material (Table 2).

There was a direct relationship between soil NO3-N concentration and log protozoan biomass (Fig. 5a). In an analysis of covariance, both soil NO3-N concentration as the covariate and the treatment term were significant. This shows that the relationship was independent of the overall differences in log protozoan biomass caused by the type of patch added. The apparent relationship between protozoan biomass and numbers of nematodes (Fig. 5b) was not confirmed by an analysis of covariance, as log protozoan biomass was not a significant covariate. However, the significant (P = 0.001) patch term showed that the two taxa were independently enhanced by the different patch treatments.

Figure 5.

The relationship between log protozoan biomass in the patch zone at harvest and (a) NO3-N concentration and (b) log nematode numbers in the patch zone at harvest. Data shown are mean values with SE bars for each patch treatment. Treatment codes as in Fig. 4 and H2O control (star symbol). For (a) NO3-N concentration was a significant covariate (F1,7 = 5.59, P = 0.023); in (b) log protozoan biomass was not a significant covariate (F1,39 = 0.08, P = 0.778) but the patch term was significant (F1,7 = 4.31, P = 0.001) in an analysis of covariance using patch treatment as the main factor.

In the CLPP, the number of substrates utilized, out of a possible maximum of 95, was significantly fewer in the unplanted urea treatment (73.3, SD 6.22). The only other significant differences from the H2O control treatment (85.5, SD 4.55) were the unplanted L. perenne shoot material patch (91.7, SD 2.34) and the l-lysine treatment (80.5, SD 1.52). Of the planted patch treatments, algal cells (90.0, SD 1.79) and L. perenne shoot material (87.7, SD 4.69) had significantly more positive wells than algal amino acids (81.0, SD 6.29), l-lysine (80.5, SD 1.52), and urea (82.5, SD 3.62).

Lolium perenne shoot material, both planted and unplanted, resulted in a clearly distinct CLPP, as determined by multivariate analysis of the 95 substrates (Fig. 6), whereas the profile following urea amendment was significantly affected by the presence of the plant. Of the other planted patch treatments, algal cells and l-lysine gave statistically similar profiles, while algal amino acids and L. perenne gave statistically distinct profiles (Fig. 6). Separation of the treatments along principal component 1 (PC1) was largely due to the increased utilization of cellobiose, d-melibiose, d-raffinose, α-methyl d-glucoside, maltose and l-rhamnose in treatments with L. perenne shoot material, and along PC2 due to increased utilization of d-saccharic acid and α-amino butyric acid in the urea and H2O (control) planted patches.

Figure 6.

Results of principal component analysis on Biolog data for Lolium perenne shoot material (treatment symbols as Fig. 4; H2O control, star symbol) showing the pattern of carbon source utilization after 120 h incubation. Data represent the ordination of PC1 vs. PC2. Bars represent LSD (P = 0.05).

Discussion

Root proliferation and n capture from the patches

Our first hypothesis was that both root production and mortality rates would increase in the least complex organic patches, as these would decompose most rapidly resulting in a flush of available nutrients. Root birth rates were greater in all patches with C : N < 4 than with higher ratios. However, there were no differences among treatments in either instantaneous or cumulative root deaths over the period of study. This contrasts with previous observations in both nutrient-rich soils (Aber et al. 1985; Pregitzer et al. 1995) and microsites (Hodge et al. 1999c), where shorter root life spans are generally reported. Aber et al. (1985) suggested that root life spans may be influenced by the form of the available N. This was not the case in the present study, as although the same total amount of N was applied, patches varied widely in their chemical and physical properties and C : N ratio.

Plants captured 45–54% of the initially added N in all treatments except the L. perenne shoot patch, where values were much lower (11%). Thus, N capture from the patch was greatest in the treatments with fastest root production rates. Hodge et al. (1999a) found that absolute N capture from a simple organic patch was a function of the amount of N added. Here we added the same amount of N in all patches and showed that N capture was only qualitatively related to the C : N ratio of the patch: only in the patch with the highest C : N ratio (c. 21 : 1) was N capture significantly reduced. Part of our second hypothesis, which predicted that N capture from the patch would decrease as the C : N ratio of the patch increased, therefore received some support at high ratios. However, there was no difference in plant N capture from those patches with similar C : N ratios but differing in their physical and chemical complexity (i.e. algal cells vs. both amino acid patches).

Net immobilization by the microbial biomass is generally thought to reduce plant N capture when the C : N ratio is > 20–30 : 1 (Bartholomew 1965). In this study, 13CO2 evolution and the analysis of the patch soil at the end of the experiment indicated that the L. perenne patch (C : N ratio 20.7 : 1) was still actively decomposing and releasing N. As expected, the proportion of N captured by plants from the L. perenne patch in this study was higher than that captured by a range of different grass species (including L. perenne) when grown in the presence of a similar type of patch with a C : N ratio of 31 : 1 (Hodge et al. 1998). The amount of patch N estimated here to be in the microbial biomass in the L. perenne patch was similar to the amount of N captured by the plants. This implies that the net N capture by plants from this particular patch was low because supplies were restricted by continuing decomposition (and net N mineralization) rather than because patch N was unavailable due to immobilization within the microbial biomass. Other values for plant N capture from organic residue patches vary widely (c. 5–55%) (Yaacob & Blair 1980; Azam et al. 1985; Müller & Sundman 1988; van Vuuren et al. 1996). Nutrient capture from patches may be influenced by factors such as crop history (Yaacob & Blair 1980), placement of the patch in relation to other patches (Jingguo & Bakken 1997), field conditions (Ta & Faris 1990), type of plant species present (Einsmann et al. 1999; Farley & Fitter 1999b) and origin of the patch material in relation to the plant species grown (Reynolds et al. 1997), as well as the time scales of the studies.

Plant growth

Our third hypothesis was that plant growth would be depressed in the most complex patch due to immobilization of nutrients by the microbial biomass. At final harvest, shoot biomass was greater than the control treatment in all N-containing patches except L. perenne shoots, but neither the total root biomass nor the proportion of root dry weight that was located in the patch differed among treatments. Growth depressions after addition of complex organic residues are often reported (Seligman et al. 1986; Azam et al. 1990) and may be overcome if additional N is applied (Elliott et al. 1981). It was surprising that no difference in root biomass occurred between treatments in this study. We (Hodge et al. 1998) have previously observed that plant, particularly root, growth of several grass species is repressed after addition of an organic residue, although of a higher C : N ratio than used in this study. Furthermore, increases in root biomass in nutrient-rich zones at the expense of growth in nutrient-poor areas are frequently reported, implying some co-ordination of total root growth (reviewed by Robinson 1994; Robinson & van Vuuren 1998). However, addition of l-lysine either as a patch of varying concentration (Hodge et al. 1999a) or differing temporal availability (Hodge et al. 1999c) did not alter the fraction of roots within the patch. Moreover, an increase in root dry weight in the patch does not itself necessarily translate to increased patch exploitation; changes in the architecture of the root system must also occur (Fitter 1994).

Patch decomposition

Our fourth hypothesis was that protozoan biomass would be greater in unplanted than planted microcosm tubes because of a lack of competition from roots. However, at final harvest, microfaunal populations more closely mirrored the original substrate C : N ratio, being highest in the L. perenne shoot patches and lowest in the urea patches. They did not reflect whether the microcosm units were planted or not. The presence of plant roots also had little influence on 13C release from the soil, suggesting that decomposition was predominantly microbially driven. Plant roots similarly had no effect on the 13CO2 release from a l-lysine patch but did influence decomposition when the l-lysine was added as a series of smaller pulses over time (Hodge et al. 1999c). As no 13C excess was detected in the plant tissues, this also implies that N captured from the added patches was taken up in inorganic form after microbial transformation, rather than as simple intact organic compounds as has been shown in some studies (cf. Jones & Darrah 1992; Lipson & Monson 1998; Näsholm et al. 1998). Thus, regardless of the physical, chemical or C : N properties of the patches used in this study, initial decomposition was microbially driven. As shown by the pattern of 13CO2 release (Fig. 4) this decomposition was rapid except in the most complex patch (L. perenne shoots).

Inorganic N concentrations were low except in the most complex (L. perenne shoots) patch. This indicates that the N mineralized from the patches of lower C : N ratios was rapidly utilized either by the plants and/or by the microbial biomass. The 15N and 13C enrichments of the soil, the estimated amounts of residual substrate, and the large protozoan populations (Table 2) confirm that, unlike the other patches, degradation of the L. perenne shoots was still continuing at final harvest. In addition, NO3-N concentrations were greater in the unplanted than the planted L. perenne shoots patch, indicating that the rate of N utilization and/or mineralization was slower in the absence of roots. This conclusion is further supported by the regression analysis of 15N and 13C from these tubes, which shows that C and N release from the L. perenne shoots patch was coupled and that N was more likely to be lost from the soil–microbial system, presumably due to plant N uptake, when plant roots were present. This confirms the results of Hodge et al. (1998), also using L. perenne shoot material but of a higher C : N ratio (31 : 1), where the presence of roots enhanced N loss from the patch soil irrespective of the plant species tested. In contrast, the presence of plants reduced decomposition of added residues in other studies (Sparling et al. 1982; Moorhead et al. 1998).

Protozoan biomass was related to soil NO3-N concentration and the relationship was independent of the overall changes caused by the type of patch added. A relationship between soil NO3-N and protozoa has been reported previously (Alphei et al. 1996; Hodge et al. 1998) and may be a result of protozoa feeding on nitrifying and denitrifying bacteria, or protozoa enhancing nitrification activity (Griffiths 1989; Verhagen et al. 1994, 1995). This, together with the small 15N recoveries from the patch soil at final harvest, suggests that nitrification and denitrification probably occurred in the patches, leading to substantial losses of gaseous N from the system.

In addition to differences in microfaunal biomass within the patches, CLPP revealed differences in the structure of the microbial communities, thus confirming our fifth hypothesis. Adding substrates of greater complexity (L. perenne and algal cells) increased the metabolic diversity of the microbial community (i.e. number of positive wells). Also, the pattern of substrate utilization differed as substrate complexity changed, with urea causing no detectable change in the profile when planted, small changes apparent with l-lysine, algal amino acids and algal cells, and large changes brought about by the addition of L. perenne. Thus, while Grayston et al. (1998) had already shown characteristic changes in the rhizosphere CLPP as a result of planting different plant species, here we show that significant changes can also be due to the C : N ratio of the patches added.

Plants versus microbial n capture

We measured protozoan biomass, which is a better indicator of microbial activity than microbial biomass per se (Christensen et al. 1996), to estimate the amount of patch N present in the microbial component (Table 2). In the urea patches with and without plants the fraction of 15N retained in the microbial biomass was the same, but in the unplanted treatment a further 40% of the added N remained in the soil. In the absence of roots as a major sink for N, the microbial biomass was apparently unable to use the large amounts of N released; the surplus was presumably adsorbed to soil exchange sites as NH4+. Estimated N immobilization by the microbial biomass increased with increasing C : N ratio of the patch, but was much lower than N capture by the plants, except in the L. perenne patches where N capture by microbial and plant biomass was similar. These results contrast with the widely held view that micro-organisms are better competitors for nutrients than roots. For example, Schimel & Chapin (1996) recovered c. 40–70% of the 15N from a glycine patch in the microbial biomass, while Jackson et al. (1989) and Schimel et al. (1989) found micro-organisms were better competitors than plants for inorganic N added to an annual grassland at contrasting times of the year. Although Lipson & Monson (1998) reported almost equal partitioning of N capture from an amino acid patch between plant and micro-organisms, amounts were much smaller (c. 3–5% N captured, respectively) than those we report here. All of these previous studies were conducted over short time periods (i.e. 1–5 days). Jackson et al. (1989) suggested that roots may be better competitors for nutrients over longer time scales due to rapid microbial turnover releasing nutrients back into the soil–plant system. Our results confirm that this is true: even by day 6, 20–35% of the added 15N from the urea and algal amino acid patches was detected in the sampled shoots (Fig. 3). Therefore, our results show that plants can be highly effective competitors for N, particularly in patches of low C : N ratios, even when the microbial biomass is first decomposing the substrate.

Conclusions

Previously, we (Hodge et al. 1998) suggested that the microbial community was more important than plant attributes in controlling patch exploitation per se. The results of the present study, using a range of organic patch materials differing in their physical, chemical and C : N characteristics, also suggest that micro-organisms initially out-compete plants for the added N, as plants only capture N mineralized by the microbial community. Although it is often difficult to factor out the influence of physical and chemical complexity from that of the C : N ratio, we attempted to do this by comparison of plant N uptake from simple amino acid patches with that from a more physically and chemically complex source but with a similar C : N ratio (i.e. algal cells patch). Plant responses (i.e. biomass production, tissue N concentrations and N capture from the patch) were similar, as were the microbial community metabolic profiles and protozoan biomass. Indeed, none of the plant responses differed between treatments except in the most complex (physical, chemical and highest C : N ratio) patch compared with all other treatments. In contrast, there was some evidence that the C : N ratio of the patch did alter the physiological profile of the microbial community, as shown by the Biolog data, but this did not affect the amount of N immobilized in the microbial biomass (c. 7–13%). Despite microbial decomposition being the first step in patch decomposition, N mineralized from the patches quickly became available for plant uptake (i.e. in a matter of days), presumably due to rapid microbial turnover releasing N back into the soil–plant system. Plants were effective at competing for this released N, and their eventual capture of patch-derived N was greater than that immobilized in the microbial biomass, particularly in the patches with low C : N ratios (Table 2). In the patch with the highest C : N ratios, N loss was slower and more N was retained in the patch. In this relatively long-term experiment, therefore, plant attributes were the main determinants of N capture from patches, in contrast to previous views. Finally, we measured total N capture by a L. perenne sward. In natural ecosystems, roots of one plant species have to compete with those of others for the N available due to release from a patch by microbes. Under these conditions both the speed and the extent of root proliferation will be critical for effective patch N capture (Hodge et al. 1999b; Robinson et al. 1999) and potential competitive superiority.

Acknowledgements

This work is funded by the Biotechnology and Biological Sciences Research Council (BBSRC). The Scottish Crop Research Institute receives grant-in-aid from the Scottish Executive Rural Affairs Department. We thank Charles Scrimgeour and Winnie Stein for their invaluable technical assistance. The authors are grateful to two anonymous referees and Drs Ian Sanders and Lindsay Haddon for their valuable comments on the manuscript.

Received 27 April 1999revision accepted 23 September 1999

Ancillary