Why plants bother: root proliferation results in increased nitrogen capture from an organic patch when two grasses compete


Dr A. Hodge Fax No. + 44 11904–432860; e-mail: ah29@york.ac.uk


We investigated whether the capacities of Lolium perenne L. and Poa pratensis L. roots to proliferate locally and to alter local nitrogen (N) inflows in a decomposing organic matter patch were important in their capture of N when grown together. In the presence of a patch, plants of both species were significantly heavier and contained more N. Root length and weight densities increased in the patch, but specific root length was unaltered. Although both species proliferated roots in the patch, L. perenne produced greater root length densities than P. pratensis, and also captured more N from the patch. Indeed, total N uptake from the patch was related to root length density within the patch. N inflows (rate of N uptake per unit root length) in the patch were no faster than in the whole root system for both species. Under the conditions of this study, root proliferation in an organic patch was more important for N capture from the patch than alterations in N inflows. Local proliferation of roots may be a key factor in interspecific competition for non-uniformly distributed supplies of N in natural habitats, so resolving the previous uncertainty as to the ‘adaptive’ nature of root proliferation.


Root proliferation in nutrient-rich zones or patches has been widely reported for many species (reviewed by Robinson 1994; Robinson & van Vuuren 1998), as has the stimulation of nutrient uptake per unit of root in zones enriched in N, P or K (Jackson, Manwaring & Caldwell 1990; Robinson, Linehan & Gordon 1994). Both responses may be adaptations to the spatial and temporal heterogeneity of nutrient supply in soil (Hutchings & de Kroon 1994).

However, Hodge et al. (1998) found, for five grass species grown in monoculture, that their capture of N from a patch of decomposing organic matter was not related strongly to root proliferation in the patch. This agrees with solute transport theory (Nye & Tinker 1977) which predicts only a weak association between root geometry (e.g. length per unit soil volume) and the uptake of relatively mobile ions such as NO3. The demonstrable proliferation of roots in response to locally available NO3 is paradoxical on the basis of simple resource capture arguments (Robinson 1996a). One possible resolution to this paradox may be that proliferation has less to do with nutrient capture per se, and more to do with advantages gained when root systems of different species compete. Then, root proliferation may be ‘adaptive’ if it allows faster uptake of nutrients from soil relative to the uptake possible by the roots of a competing plant. Even so, proliferation will be advantageous only if nutrient capture exceeds the cost of constructing new roots, and this may not occur if the patch is short-lived compared with the life-span of the root (see Fitter 1994; Eissenstat & Yanai 1997).

In contrast to the many studies of root proliferation in response to nutrient patches, only a few have examined its importance during interspecific competition for those nutrients (e.g. Fitter 1976; Caldwell et al. 1985; Caldwell, Manwaring & Jackson 1991; Reynolds et al. 1997). In soil, any nutrient-rich patch will also be a zone of increased microbial activity (Christensen et al. 1992; Griffiths et al. 1993; Griffiths & Caul 1993). Roots in such a patch must also compete with micro-organisms for nutrients. Again, only a few studies have explicitly included plant–microbial competition in nutrient patches (Griffiths, van Vuuren & Robinson 1994; van Vuuren, Robinson & Griffiths 1996; Hodge et al. 1998). Even then, plants were grown as monocultures rather than potential competitors.

The aims of this investigation were: (i) to quantify spatial and temporal responses of root growth of two grass species, Lolium perenne L. and Poa pratensis L., when grown together in an organic patch and, potentially, in competition for nutrients; and (ii) to compare those responses with the activity of microbial decomposers in the patch by monitoring changes of microbial-feeding protozoa, which are indicators of microbial activity (Sohlenius 1990) and important in N mineralization (Griffiths 1989). In addition, we used destructive harvests and dual-labelled (13C and 15N) organic matter to follow patch decomposition and plant uptake of N with time, to assess the contribution made to each species’ N uptake from the patch by root proliferation and N inflow.


General design

In order to test the hypotheses of this experiment that root proliferation results in increased N capture from an organic patch when two species are grown together within that patch, two grass species, L. perenne and P. pratensis, were grown together in the presence or absence (controls) of an organic patch. Monoculture controls were not used because this has been done previously (Hodge et al. 1998). To follow the dynamics of the organic patch decomposition and the uptake of N by the plants, four replicate experimental microcosm units were harvested after 0, 7, 14, 21, 28, 35, 42, 49 and 56 d. In addition, four replicate control plates (with a sand : soil patch only) were harvested after 0, 14, 28, 42 and 56 d.

Lolium perenne and P. pratensis were selected as they have previously been shown to differ in root proliferation within an organic patch when grown in monoculture (Hodge et al. 1998), L. perenne being a stronger proliferator than P. pratensis. In addition, these two species occur together in pastures and thus are natural competitors.

Dual-labelled organic material

15N labelled plant material was generated by growing L. perenne L. cv. Aurora from seed on a 1 : 1 peat/sand mix, watered with Hewitt's nutrient solution (Hewitt 1966) containing 15NH415NO3 (99 atom percentage 15N; Europa Scientific, Crewe, UK) for 12 weeks. The plants were then grown in an atmosphere containing 13C-enriched CO2 for 2 d prior to harvest as described in Hodge et al. (1998) to obtain dual-labelled material. Shoots were then harvested, dried and analysed by continuous flow isotope ratio mass spectrometry (CF-IRMS; Tracermass, Europa Scientific, Crewe, UK). The shoot material contained 1·60% N (28·2 atom percentage 15N) and 50% C (1·82 atom percentage 13C), with a C : N mass ratio of approximately 31 : 1.

Plant culture

Seeds of L. perenne (perennial rye-grass) and P. pratensis (smooth meadow-grass) were supplied by Johnson seeds, Lincolnshire. Seeds were soaked for 24 h in distilled H2O, then germinated on filter paper. Seedlings were transferred after 4 d to microcosm units (see below).

A medium loam soil, collected from a site at the University of York, was sieved through a 2 mm mesh. To reduce its nutrient status, it was mixed with an equal mass of fine sand (Redhill 65, particle size approximately 0·23 mm, Hepworth Minerals & Chemicals Ltd, Cheshire, UK).

Plants were grown in glass microcosms in which labelled organic matter was confined into a patch in only part of the rooting zone (Fig. 1). The base plate of each unit had strips of perspex attached to three of its sides. Two perspex bands (7·5 cm × 2·0 cm × 0·3 cm, Fig. 1) were attached 12 cm down from the top edge of the base plate. A patch was created in the area between these two bands (Fig. 1). The patch comprised 0·5 g of the labelled organic matter (see above) mixed with 10·5 g of sand : soil mix. Controls were units filled completely with sand : soil mix, that is, they contained no labelled organic matter; the ‘patch’ in them contained 10·5 g of sand : soil mix. The bulk density was 1·41 g cm–3 within the organic patches and 1·35 g cm–3 within the control patches. Root growth into the patch zone (in both treatment and control units) was prevented, initially, by a removable perspex strip directly above the patch. A 3-mm deep layer of sand : soil mix was placed over the rest of the area of each base plate. Drainage gaps were incorporated to prevent waterlogging at the base of the unit.

Figure 1.

. Microcosm unit base plate layout. In experimental plates, patch = soil : sand mix with13C/15N labelled organic matter; in the controls patch = soil : sand mix only.

Two seedlings, one of each species, were placed on the upper surface of the sand : soil mix, planted 11 cm in from either side in each microcosm. Because of the different relative growth rates of the two species (Grime, Hodgson & Hunt 1988) P. pratensis seedlings were planted 11 d before L. perenne to ensure that roots of both would enter the patch simultaneously. After a further 7 d, the perspex strip, preventing growth into the patch was removed; this was designated as day 0 of the experiment. The gap created by removal of the strip was filled with sand : soil mix.

The seedlings and sand : soil mix were sprayed twice weekly with distilled H2O over the plate's entire area (i.e. 28 cm × 40 cm) to a known target weight for each microcosm in order to maintain a moisture content between 0·17 and 0·20 g water g–1 dry medium. The second glass plate was taped into position over the base plate so that the root systems were sandwiched into the microcosm unit. Black polythene sheets were wrapped and clipped around the microcosm unit to prevent illumination of the root system.

The microcosm units were transferred to a Conviron® (model E15; Conviron, Winnipeg, Manitoba, Canada) growth cabinet where a combination of fluorescent tubes and incandescent bulbs provided a photon flux density of approximately 450 μmol m–2 s–1 at plant height. The relative humidity was set at 80% with an 16 h, 25 °C day and 8 h, 15 °C night regime.

Root length measurements

At harvest, the bands 2 cm directly above and below the patch were lightly marked by gently scoring the perimeter of the zones with a scalpel without actually cutting the roots within these zones. This procedure was repeated until nine zones had been marked (i.e. 2 cm directly above and below the patch, 2 cm zones to the right and left hand side above and below the patch, the rest of the plate above and below these bands, the patch zone itself; Fig. 1). The root systems of both plant species were then carefully separated from each other and removed from the appropriate band in which they occurred using tweezers until all roots from both species in all zones had been recovered. Total root length in each zone for each species was measured by using a Magiscan (Applied Imaging Ltd, Gateshead, UK) Image Analysis System, using the program FIBRE v4·4. No attempt was made to assess the number of lateral root primordia that were initiated. Roots in the patch had a tendency to bind particles of the labelled organic matter and great care was taken to remove these fragments before chemical analysis.

N inflows

Rates of N uptake per unit root length (inflows) were derived separately for L. perenne and P. pratensis using Hunt & Parsons's (1974) stepwise regression program. Inflows were calculated as instantaneous values (I, pmol m–1 s–1) as I = (1/ L) dN/dt, where L is root length (m), N is plant N content (pmol) and t is the time (s, to the nearest whole day) at which L and N were measured. For inflows into the whole root system (patch + soil), L was total root length and N the amount of N per plant; t was from 0 to 56 d. For inflows into roots within the patch, L was root length in the patch and N the amount of plant N originating from the patch; t was from 7 to 56 d as roots entered the patch between 0 and 7 d.

Plant and soil analysis

The roots extracted from the different zones were washed thoroughly, oven-dried at 60 °C, weighed separately and then bulked together for each species before milling. The shoot material was also oven-dried at 60 °C and weighed. A subsample of the root and shoot material was analysed for total N and 15N by CF-IRMS. The total N (i.e. 14N and 15N) derived from the patch was calculated using the equation: [(15N in experimental sample – background (i.e. 15N in the appropriate control sample)/total (N) × (100/28·2 atom percentage 15N)]. The rate of 15N uptake per day was calculated using the equation [(individual plant 15N content at harvest – mean 15N content of the previous harvest)/time between harvests].

At harvest, subsamples of the control and experimental patches, the sand : soil mix in bands 2 cm above and below the patches, and the remaining sand : soil mix above and below these bands were used for moisture content determinations (105 °C) and for total N and 15N analysis (by CF-IRMS). Protozoan biomass and soil NH4+-N and NO3-N concentrations in experimental and control patches were estimated as described in Hodge et al. (1998). In addition, four labelled patches and four control patches were prepared identically to those in the microcosm units. These were subsampled immediately to determine initial NH4+-N and NO3-N concentrations.

Statistical analysis

The data were analysed as a split-plot experiment using Genstat (v. 5 release 3·2; Lawes Agricultural Trust, Rothamsted Experiment Station, Harpenden, Herts, UK) where comparisons between species were to be made. Analysis of patch properties were performed treating each plate separately.


Inorganic N and microfauna

The labelled organic matter initially added as a patch contained approximately equal amounts of NH4+-N and NO3-N. No NH4+-N was recovered from control patches in the absence of roots. When roots were allowed access to the patch, 18 d later, concentrations of both NO3-N and NH4+-N had declined (Fig. 2).

Figure 2.

. Concentrations (μg g–1) of NH4+-N in the control (□) or organic (□) patch of NO3-N in the control (○) or organic (○) patch initially added at –18 d and during the experimental period (0–56 d). Data are means (n = 4) ± SE. For both NH4+-N and NO3-N concentrations there was a significant (P < 0·05) effect of patch. Statistics are given in Table 1.

NO3-N and NH4+-N varied with time and were significantly more concentrated in the experimental patch compared with controls (Table 1). After the roots were allowed access to the patch (between 0 and 56 d), NH4+-N was always < 0·6 μg g–1 and declined significantly with time (Table 1). Concentrations of NO3-N were generally greater than for NH4+-N (Fig. 2). NO3-N concentrations in the experimental patches significantly (P < 0·001) increased over the duration of the experiment (regression equation: [NO3] (μg cm–3) 0·87 + 0·0379 d), but when data for control patches were included, time was not a significant covariate showing that no increase occurred in the control patches (Table 1).

Table 1.  . Analysis of covariance of soil biological and chemical properties in experimental and control patches using patch type as the factor. For the analysis of (a) soil mineral N, time (d) was the only covariate and for (b) protozoan biomass (ng g–1), time (d) or soil mineral N (NO3-N and NH4+-N) was the covariate. A significant covariate means that there was a linear trend of the variate with respect to that covariate; a significant patch effect means that experimental and control treatments differed for the variate after the influence of the covariate had been eliminated Thumbnail image of

Protozoan biomass was greater in the organic patches than in controls (Table 1) but regression analysis showed biomass decreased with time (data not shown). The significant relationship between log protozoan biomass and soil NH4+ concentration is probably artefactual because biomass and NH4+ concentrations both declined with time. Further, when data for the start (0 d) and the end of the experiment (56 d) were analysed separately there was no significant relationship between NH4+ and protozoan biomass in these two subsets of the data. There was no significant relationship between soil NO3 concentrations and log protozoan biomass and NO3 was not a significant covariate (Table 1). The effect of the organic patch could not be explained by increased N availability, since patch remained a significant factor in the analysis of covariance (Table 1).

Patch decomposition

When the roots were allowed access to the organic patch at day 0 (18 d after the patch had first been added), the patch still contained 88% of the original 15N and 58% of the original 13C (data not shown). The amount of 13C and 15N remaining in the patch declined progressively with time (Fig. 3a & b), demonstrating that decomposition continued after the roots entered the patch. The sigmoidal relationship between 15N and 13C in the patch (Fig. 3c) suggested that rates of N and C release during decomposition, and in the presence of roots, varied. The original patch material had a log mass ratio (15N/13C) of 0·48 (2·4 mg 15N and 6·1 mg 13C), but the slope of a linear regression of Fig. 3(c) was 1·27 (data not shown). C was therefore lost more rapidly from the patch, at least initially, presumably as CO2 produced by microbial respiration.

Figure 3.

. The relationships between the amounts (mg) of 13C and 15N in the decomposing organic patch with time (d) and between 15N and 13C in the organic patch. (a) 13C (mg) against time (y = 3·10 – 0·0253x, P < 0·001, R2 = 46·8). (b) 15N (mg) against time (y = 2·17 – 0·0234x, P < 0·001, R2 = 63·3). (c) log mg 15N against log mg 13C (y = –0·712 + 1·27x, P < 0·001, R2 = 71·6). Points shown are individual differences in isotopic content between experimental and control patches.

Slight 13C and 15N enrichment (relative to background) was detected in the zones 2 cm either side of the patch. This indicated movement of some labelled material out of the patch. Amounts detected, averaged over the entire experiment, were small: 4·1 (above), 4·5 (below) for 15N and 3·6 (above), 1·4 (below) for 13C as a percentage of the original amount added in the patch. 15N movement into the zone 2 cm above the patch was detected at all harvests except at 0 d when the perspex barrier preventing root growth into the patch was removed. No 13C was detected outside the patch after 28 d. This lack of synchrony between 13C and 15N movement indicates that mobile products of decomposition, rather than the patch material itself, were moving.

Plant biomass

The presence of an organic patch did not affect root weight, but it increased shoot and total dry weight (Fig. 4) of both species equally (P = 0·022). In both, the total dry weights were 1·8-fold greater than controls after 56 d.

Figure 4.

. Total dry weights (DW) (mg per plant) against time of P. pratensis grown with (□) or without (□) an organic patch and of L. perenne grown with (○) or without (○) an organic patch. Data shown are means (n = 4) ± SE. Lolium perenne produced significantly (P < 0·05) more total biomass than P. pratensis. Total DW in both species was significantly (P = 0·022) increased by the presence of an organic patch.

Lolium perenne produced significantly (P < 0·05) more total (Fig. 4) and root biomass than P. pratensis, total weights being 2·6-fold greater than those of P. pratensis at 56 d in both experimental and control plants. However, there was never a significant interaction between species and patch: both species responded similarly to the patch. The interaction between time and patch was significant (P < 0·05) for shoots only. This suggests that the effect of the patch on shoot dry weight was progressive.

Root growth in the patch

Root length density increased significantly within and 2 cm below the patch (Fig. 5a & b; Table 2). Lolium perenne produced greater root length densities in the patch than P. pratensis, but the species×patch interaction was never significant (Table 2). There was, however, a significant time×patch interaction. Root length densities (RLD) responded markedly in both species after 28 d (Fig. 5a & b). Lolium perenne appeared to allocate its roots more precisely than P. pratensis, its root length densities being > 60 cm cm–3 in the patch but < 10 cm cm–3 in the zones either side of the patch at 56 d. There was no significant species effect or species×patch interaction in the zones 2 cm either side of the patch (Table 2). Root weight density followed trends similar to those for RLD (Table 2).

Table 2.  . Summary of the analyses of variance of root lengths, weights and specific root length (root length/root dry weight) for P. pratensis and L. perenne grown with or without an organic patch Thumbnail image of
Figure 5.

. Mean differences between experiment and control root length densities (mm root cm–3 soil) for (a) P. pratensis and (b) L. perenne in the bands 2 cm above □, within □ and 2 cm below the patch (shaded columns). Bars show SE.

The root weight in the patch, expressed as a fraction of the total root weight, was significantly (P = 0·002) greater than that in controls (Fig. 6). There was no significant difference between species nor was there a species×patch interaction. Root weight ratio (RWR; total root dry weight/total plant dry weight) did not differ among treatments (data not shown).

Figure 6.

. Root weight in the patch expressed as a fraction of the total root weight of (a) P. pratensis grown with (□) or without (□) an organic patch and of (b) L. perenne grown with (○) or without (○) an organic patch. Data shown are means (n = 4) ± SE. The fraction of roots in the patch was significantly (P = 0·002) larger in the presence of an organic patch. There were no significant differences between species.

An organic patch had no effect on specific root lengths (SRL; Table 2), indicating no change in root morphology in response to the patch. Poa pratensis had a larger (P < 0·05) SRL than L. perenne in the zones 2 cm above and below the patch. There was no difference in SRL in the patch between the two species and there was never a significant species×patch interaction.

N capture from the patch

Plant N contents were greater in the presence of an organic patch (Fig. 7). Poa pratensis contained less N than L. perenne. Tissue concentrations of N (mg N g–1) were greater (P < 0·001) in the presence of an organic patch, but there was no significant species effect nor any species×patch interaction (data not shown).

Figure 7.

. Total plant nitrogen content (mg N per plant) of P. pratensis grown with (□) or without (□) an organic patch and of L. perenne grown with (○) or without (○) an organic patch. Data shown are means (n = 4) ± SE. Plant N contents increased significantly (P < 0·001) in the presence of an organic patch. Lolium perenne contained significantly (P < 0·05) more N than P. pratensis.

Shoots and roots were enriched in 15N compared with controls (P < 0·001), indicating uptake of N from the organic patch. In contrast, tissues were not enriched in 13C and there was no consistent relationship between the 13C and 15N enrichments in either roots or shoots, which would have implied that the plants where taking up unmineralized organic compounds (e.g. as amino acids or other such soluble organic N compounds) from the patch (data not shown).

The amount of N (15N + 14N) captured from the patch increased throughout the experiment for both species, but L. perenne had captured more N by 56 d (Fig. 8). Ultimately, however, both species captured only small fractions of the N in the patch, namely, 8·7% by L. perenne and 4·6% by P. pratensis. Total N uptake from the patch was related to the root length density in the patch (Fig. 9), and the latter was a significant covariate of total N content in the plants (data not shown).

Figure 8.

. Uptake of N (14N + 15N) from the organic patch by P. pratensis (□) and L. perenne (□) expressed as N content (μg) in the total plant tissue. Species were significantly (P < 0·05) different. Data shown are means (n = 4) with ± SE.

Figure 9.

. Plant N derived from the organic patch in L. perenne (○) and P. pratensis (○) against root length density (mm cm–3) within the patch. Data shown are means (n = 4) ± SE. There were no significant differences between species.

The daily rate of N uptake from the patch was the same in the two species, and again root length and weight in the patch were both significant covariates of N uptake rate (Table 3). Patch root weight as a fraction of total root weight, and SRL, were not significant covariates. Species was never a significant factor (Table 3).

Table 3.  . Analysis of covariance using rate of 15N uptake per day for the period 7–56 d, as the variate, species as the factor and root length in the patch, root dry weight (DW), the fraction of patch root DW/total root DW or SRL as the covariates Thumbnail image of

There were no significant interspecific differences in total N inflow among control and experimental plants (Fig. 10). The same was true of N inflow from the patch. Maximum N inflows in the patch occurred at 28 d in P. pratensis (11·5 pmol m–1 s–1) and in L. perenne (14·5 pmol m–1 s–1), but these were not significantly faster than N inflows in the whole root system.

Figure 10.

. Instantaneous N inflows (bold curves) ± 95% confidence limits (fine curves), in P. pratensis (a–c) and L. perenne (d–f). Plots a, b, d and e are inflows for the whole root system in control (a and d) and organic -amended (b, e) treatments. Plots c and f are inflows for roots in the experimental patches.


Patch decomposition

The presence of an organic patch significantly improved the growth of both L. perenne and P. pratensis. This contrasts with our previous study using a patch of similar C : N ratio (Hodge et al. 1998), and with other studies (e.g. Elliott, Cochran & Papendick 1981; Seligman et al. 1986), where growth suppressions were reported. Presumably this difference was caused by the timing of patch addition. Hodge et al. (1998) added the patch after plants had grown for 11 d. In this study, by contrast, the patch was added at the start of the experiment. By the time that the roots gained access to it, decomposition was advanced beyond the initial phase where organic matter of a C : N ratio greater than 25 : 1 (Bartholomew 1965) may immobilize rather than mineralize N.

Sequential analysis of the patches showed that N and C were being released by microbial activity. Protozoan biomass is an effective indicator of preceding microbial activity (Sohlenius 1990; Christensen, Griffiths & Christensen 1992), more so than actual microbial biomass (Andren, Paustian & Rosswall 1988). The progressive decline that we observed in protozoan biomass suggests that microbial activity had already peaked in the 18 d prior to the roots accessing the patch. A large, rapid increase in protozoan biomass in response to the organic patch followed by a subsequent decline (i.e. after 10 d) has been reported previously (Griffiths et al. 1994). The patch material lost approximately 40% of its original 13C, and 12% of its original 15N in the 18 d before roots gained access to it, confirming that the patch contained active decomposers. It was during that period when the NH4+ and NO3 mineralized from the patch were most concentrated.

Root proliferation

Roots of L. perenne and P. pratensis were slow to proliferate in the patch (≈ 35 d). This agrees with our previous study (Hodge et al. 1998) when these species were grown in isolation with patches similar to that used here. This is in sharp contrast to the rapid root proliferation (i.e. after 1 d) of Agropyron desertorum and Artemisia tridentata reported by Jackson & Caldwell (1989) after the localized addition of a readily available N-rich solution.

Root length density was also greater in the zone 2 cm below the patch compared with controls, although, root proliferation in this zone was much smaller than in the patch itself, for both L. perenne and P. pratensis. It is probable that some inorganic N diffused downwards from the patch. There was no relationship between 13C and 15N movement from the patch, which suggests that any soluble organic N derived from the patch was relatively immobile. Zhang & Forde (1998) suggested that localized increases in NO3 concentration trigger root proliferation in N-deprived plants. If that is true, NO3 diffusion below the patch may account for the small increase in root proliferation in this zone.

There was no significant effect of patch treatment on SRL. This contrasts with other studies where rapid root proliferation was related to SRL (Eissenstat & Caldwell 1988; Eissenstat 1991), and increases in SRL in a nutrient-rich zone have often been reported (e.g. Eissenstat & Caldwell 1988; Robinson & Rorison 1983; Hodge et al. 1998). Further, the proliferation responses by L. perenne and P. pratensis were not as predicated by SRL. Poa pratensis had a greater SRL than L. perenne in all zones except the patch, yet its roots proliferated less.

Allocation responses

There was no difference in RWR of L. perenne and P. pratensis. Generally, RWR decreases when plants are grown in uniformly N-rich media compared with their RWR in uniformly N-poor media (Reynolds & D’Antonio 1996). However, Robinson & van Vuuren (1998) showed that when N, P or K are locally available, RWR is generally similar to that of controls in grasses and forbs. Lolium perenne and P. pratensis did respond to the organic patch by changing biomass allocation, but within their root systems. This was shown by: (i) increased root weight in the patch as a fraction of total weight; (ii) increased root weight density in the zones 2 cm either side of the patch and in the patch itself; and (iii) increased RLD, especially in the patch, but also in the 2 cm zone below the patch. These results imply some form of co-ordinated response within the root system of each (Gersani & Sachs 1992; Robinson 1996b). Although Robinson & van Vuuren (1998) recently concluded that such co-ordination generally occurs most strongly in slow-growing species, in this study the increases in root length and weight density occurred more precisely in the faster growing L. perenne. An increase in biomass allocation to roots in the patch itself does not ensure effective exploitation of a patch unless the architecture of the root system (i.e. (iii) above) also changes (Fitter 1994).

Patch N capture

In this study, the species which did proliferate its roots most strongly in the patch (i.e. L. perenne) also captured more N from it (i.e. approximately 8·7% compared with 4·6% in P. pratensis). This contrasts with our previous study (Hodge et al. 1998) in which the species that, when grown in monoculture, proliferated roots most strongly in an organic patch did not necessarily capture the most N from it. The ‘received wisdom’ (see Fitter 1994) that species which proliferate roots most strongly in a patch also capture most N from it, holds true if the patches are of N mineralized microbially from organic matter and if the species are competing for that N.

We observed no significant increase in N inflow in the patch, relative to those attained by the whole root system. van Vuuren et al. (1996) reported a approximately 5-fold increase in N inflow in wheat roots growing in an N-rich patch, and was the main mechanism by which wheat captured N from the patch. Clearly, this was less important for P. pratensis and L. perenne in our experiment. One explanation for this difference may be that in our study we used a patch with a C : N ratio more likely to be encountered in the natural environment, compared with the N-rich patch used by van Vuuren et al. (1996) which generated NO3 and NH4+ concentrations up to 50 μg g–1, approximately 25 to 50-fold greater than in our experiment. It may be that, in the present experiment, the NO3 and NH4+ concentrations produced by the mineralization of organic N in the patch were too small to stimulate inflow in the roots which had grown into it.

Ecological implications

The results of this study show RLD produced in an organic patch is related to N capture from that patch (Fig. 9). Lolium perenne, the species with the larger relative growth rate, and consequently higher N demand, proliferated the most in the patch and consequently captured more N. Localized root proliferation could therefore confer an advantage when plants compete for N, one that is masked in monocultures. This may answer Robinson's (1996a) question ‘Why do plants bother’ to proliferate in an N rich patch? We obtained this result by using an organic patch residue where NO3 and NH4 were continuously produced both spatially and temporally over the duration of the experiment. Thus, the spatial placement of root length under these circumstances ensured effective N capture. The results of this study therefore underline the importance of investigating plant responses to soil heterogenity using realistic types of patch material and realistic conditions, that is, not hydroponics and in the presence of a microbial component.


Our experiment has shown that the importance of various mechanisms of N capture that plants express when challenged with nutrient patches are very context-dependent. A plant's response to locally available nutrients does not, in itself, say anything about the importance of the response for nutrient capture; the context in which a response is expressed and the nutrient that is being captured are equally important (Robinson & van Vuuren 1998). Stimulation of N inflow was important in the van Vuuren et al. (1996) experiment, but localized root proliferation (even though it occurred) less so. N inflow was not important in our experiment, even though both Poa and Lolium species can vary inflow with N supply (Robinson & Rorison 1987). Conversely, localized root proliferation by P. pratensis and L. perenne was important in capturing N from the patch in our study but only because (a) N was continually available and (b) interspecific competition for that N was likely.


Robinson et al. (1999) demonstrated, using data from this experiment and a simulation model, that the capture of patch N by L. perenne and P. pratensis was in proportion to their respective root length densities and could be explained by inter-specific competition for a common N pool.


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 Office Agriculture, Environment and Fisheries Department. We thank C. Scrimgeour, W. Stein, D. Gordon, L. Williamson, P. Wilson and in particular, J. Stewart, for their invaluable technical assistance.