The plastic plant: root responses to heterogeneous supplies of nutrients


Author for correspondence: Angela Hodge Tel: +44 1904 328562 Fax: +44 1904 328505 Email:



  • I. Introduction 000
  • II. Morphological responses  000
  • III. Root demography 000
  • IV. Physiological plasticity 000
  • V. Root plasticity in patches in competition and symbiosis with microorganisms 000
  • VI. Influence of patch attributes 000
  • VII. Control of root proliferation 000
  • VIII. Conclusions 000
  • Acknowledgements 000

  • References 000


When roots encounter a nutrient-rich zone or patch they often proliferate within it. Roots experiencing nutrient-rich patches can also enhance their physiological ion-uptake capacities compared with roots of the same plant outside the patch zone. These plastic responses by the root system have been proposed as the major mechanism by which plants cope with the naturally occurring heterogeneous supplies of nutrients in soil. Various attempts to predict how contrasting species will respond to patches have been made based on specific root length (SRL), root demography and biomass allocation within the patch zone. No one criterion has proved definitive. Actually demonstrating that root proliferation is beneficial to the plant, especially in terms of nitrogen capture from patches, has also proved troublesome. Yet by growing plants under more realistic conditions, such as in interspecific plant competition, and with a complex organic patch, a direct benefit can be demonstrated. Thus, as highlighted in this review, the environmental context in which the root response is expressed is as important as the magnitude of the response itself.

I. Introduction

In soil, nutrients are distributed in a heterogeneous or patchy manner. In order to enhance resource capture, plant roots have to respond to, and exploit, these nutrient patches. This exploitation is aided by the modular structure of the root system which enables exceptional flexibility in architectural patterns and allows root deployment in nutrient-rich zones. Physiological plasticity is also important for successful patch exploitation. These mechanisms ultimately determine the below-ground success of the plant, therefore it is hardly surprising they have been the focus of much research interest (see reviews by Hutchings & de Kroon, 1994; Robinson, 1994; Casper & Jackson, 1997; Robinson & van Vuuren, 1998; Schenk et al., 1999). This review focuses on what these responses actually contribute to the nutrient status of the plant in the context of competition with both other plants and soil microorganisms. The attributes of the patch itself are also considered.

Heterogeneity in resource distribution arises as a result of organic inputs and their subsequent microbial decomposition. These inputs vary widely in their chemical and physical quality, as widely as the material from which they are derived (e.g. leaf litter, dead roots, animal corpses and dead microorganisms). Microbial decomposition of both simple and complex organic materials will release inorganic nutrients for plant capture. However, the spatial and temporal release of these inorganic nutrients will be more complex than simply applying a patch of inorganic nutrients directly. The diffusion of these released inorganic ions in soil also varies. Nitrate (NO3) is universally soluble in water and has a diffusion coefficient (D) of approx. 10−10 m2 s−1 which is close to the diffusion coefficient in free solution. By contrast, phosphate ions absorb strongly to surfaces dominated by Al3+, Fe3+ and Ca2+, forming insoluble complexes, thus the mobility of phosphate ions is much more restricted at approx. 10−13−10−15 m2 s−1 (Tinker & Nye, 2000). Collectively, these processes of organic material input and their subsequent decomposition contribute to the heterogeneous or ‘patchy’ distribution of nutrient resources in soil.

This variation in nutrient distribution can be considerable. For example, geostatistical analysis has demonstrated that within the rooting zone of an individual plant there can be as much variation in nutrient availability as within an entire 120 m2 plot (Jackson & Caldwell, 1993a, 1993b). Similarly, Farley & Fitter (1999a) found nitrate (NO3) and ammonium (NH4+) concentrations in the soil solution from a deciduous woodland varied two- to fivefold at scales of only 20 cm and, although there was a degree of seasonality in nutrient variation, the overall temporal pattern was largely unpredictable.

II. Morphological responses

When roots encounter a nutrient-rich patch they often proliferate roots within it. But what does root proliferation actually mean? It has been used to cover a range of morphological responses by roots to nutrient patches, including increases in elongation of individual roots (Jackson & Caldwell, 1989; Bilbrough & Caldwell, 1995; Zhang & Forde, 1998; Zhang et al., 1999); total root length (van Vuuren et al., 1996; Hodge et al., 1999a, 2000c); root production (Gross et al., 1993; Pregitzer et al., 1993; Hodge et al., 1999b, 2000d); and extent of lateral branching (Larigauderie & Richards, 1994; Farley & Fitter, 1999b). With such a wide definition it is not surprising that proliferation responses are commonly reported. However, root proliferation strictly involves the initiation of new laterals rather than their elongation; although both can increase root length, lateral initiation is more likely to do so locally. Quantifying lateral initiation is more easily done in some species than in others, which is why total root length in the patch is commonly measured to indicate the total absorptive area in the patch zone. Although less often reported, some roots show no proliferation response to nutrient-rich patches, even when a response may be expected. Poa pratensis produced greater root lengths in a control patch containing just the background growth medium than in an organic patch (Hodge et al., 1998), but in another study, when it was in competition with another grass species for the same type of patch, it did proliferate by an increase in root length (Hodge et al., 1999a). Moreover, proliferation responses are far from uniform across species (Campbell et al., 1991; Farley & Fitter, 1999b); even within species the response may be context-dependent. But can any generalizations be made to the extent to which allocation of roots to nutrient-rich zones occurs? First I will consider root biomass allocation.

1. Biomass allocation within the root system

It is often assumed that fast-growing species from fertile habitats show more morphological plasticity to nutrient patches than slow-growing species from infertile habitats, as the higher resource availability in fertile sites offsets the expense of new organ construction (Grime, 1979; Chapin, 1980; Grime et al., 1986; Grime et al., 1991; but see Reynolds & D’Antonio, 1996). Robinson & van Vuuren's (1998) survey on plants grown with a heterogeneous nutrient supply of N, P and/or K found some support for this hypothesis: fast-growing species had a larger ‘root size’ (root number, length or dry mass) in the nutrient-rich zone than slow-growing species when compared with a uniformly deficient control. This type of control assesses the plant's ability to capture nutrients supplied as a local discrete zone against an otherwise nutrient-poor background, and thus is more relevant to seminatural unfertilized habitats. However, fast-growing species are usually associated with higher background fertility, yet no significant differences occurred between fast- or slow-growing species in comparison to the uniformly high nutrient control. In addition, life form was as important a factor as maximum relative growth rate (RGRmax), with forbs proliferating more than grasses. Thus there was only weak support for the general assumption that fast-growing species exhibit greater morphological plasticity. This may be caused by the grouping of root length, mass and number into the general term ‘root size’ used by Robinson & van Vuuren (1998). What if only root biomass allocation within the root system was considered?

Campbell et al. (1991) studied the root response of competitively dominant and subordinate plant species. Plants were grown in microcosm units onto the surface of which nutrients were continuously dripped in four quadrants. By obtaining the correct drip rate, Campbell et al. (1991) were able to present the plant with nutrient-rich or nutrient-poor patches without the use of barriers, thus allowing free root growth within the unit. The competitively dominant plants had the greatest absolute amount of root in the nutrient-rich zone (and also in the nutrient-poor zone). However, the competitively inferior or subordinate plants located more of their new root growth into the nutrient-rich zone (i.e. they located their roots with more precision). Campbell et al. (1991) referred to this as a contrast between scale and precision of the response. Although the results of some studies support the findings of Campbell et al. (1991): (Grime, 1994; Robinson & van Vuuren, 1998; Wijesinghe et al., 2001), others do not (Einsmann et al., 1999; Farley & Fitter, 1999b; Bliss et al., 2002) and have even reported a tendency for foraging precision to be greater in species with larger root systems (Einsmann et al., 1999; Farley & Fitter, 1999b). These contrasting findings may be caused by differences among the various studies in measuring scale and precision. In addition, Wijesinghe et al. (2001) have recently demonstrated that precision of foraging is not a fixed property, but varies within species among different treatments. Thus the environmental context in which the response by the root system is expressed is as important as the response itself.

2. Root diameter and specific root length

Root biomass measurements are not necessarily indicative of the total absorptive area of the root system, and alterations of the root system architecture can occur without a change in the total root biomass. This led Eissenstat & Caldwell (1988) to suggest that the ability to proliferate roots in patches may be related to specific root length (SRL). SRL, or root length per unit mass (cm mg−1), varies with root diameter and is often used as a surrogate for diameter, but is also influenced by tissue density. Eissenstat (1991) found sweet orange (Citrus sinensis) rootstocks with the highest SRL had a more rapid rate of root proliferation than those of smaller SRL. The variation in SRL was not entirely explained by differences in root diameter, however, as differences in tissue density also occurred. Measuring root diameter directly, Fitter (1994) found a close relationship between the ratio of relative extension rates in a nutrient-rich patch to that in a nutrient-poor patch and the mean root diameter of terminal roots among four related species in the family Caryophyllaceae (Silene vulgaris, Silene noctiflora, Silene alba and Agrostemma githago). As expected, the more coarse-rooted (larger diameter) species were much less responsive to the nutrient-rich zone. However, in another study proliferation of grass roots in an organic patch was not as predicted by SRL (Hodge et al., 1998). In addition to the SRL of the species, the SRL of the roots experiencing the patch can also alter, and if it does, generally an increase is observed caused by a greater length of thinner roots (Robinson & Rorison, 1983; Fitter, 1985; Farley & Fitter, 1999b).

3. The importance of root proliferation

Drew and coworkers in the 1970s conducted a series of classic experiments on the response of barley (Hordeum vulgare) plants with only part of their root system exposed to high concentrations of phosphate, ammonium (NH4+), nitrate (NO3) or potassium (K). Roots responded by increasing the length and number of primary laterals and secondary laterals to phosphate, NH4+ and NO3 but not K (Drew, 1975; Drew et al., 1973; Drew & Saker, 1975, 1978; Fig. 1). The figures produced by Drew demonstrated that root proliferation can be quite spectacular when it does occur. Intuitively it must therefore be important for enhanced nutrient capture, but does the experimental evidence support this view?

Figure 1.

Proliferation of primary and secondary laterals by barley (Hordeum vulgare) grown in solution culture with the middle root section exposed to a 100-fold greater concentration of phosphate, nitrate, ammonium or potassium ions compared with roots above or below (LHL). Controls were supplied with high concentration of nutrients in all zones (HHH). Abbreviations: H, high; L, low, referring to nutrient concentrations experienced by the top, middle and bottom sections of the root system. From Drew (1975) with kind permission from the trustees of the New Phytologist.

Both as predicted by theoretical models (Baldwin, 1975; Barber & Cushman, 1981; Silberbush & Barber, 1983) and demonstrated experimentally (Fitter et al., 2002), an increase in root-length density is more important for enhanced capture of immobile than mobile ions. Hence the ‘benefit’ of root proliferation in phosphate-rich patches is relatively easy to interpret. But what about N-rich patches? Barley roots in Drew's (1975) experiment proliferated as much to inorganic N sources as to phosphate, thus it was similarly expected that this proliferation would be beneficial to N capture. Demonstrating that this was the case, however, proved more troublesome. Three different studies, one on wheat (van Vuuren et al., 1996), the other two using a range of different grass species (Fransen et al., 1998; Hodge et al., 1998), failed to demonstrate a relationship between root proliferation in, and N capture from, N-rich organic patches. Hodge et al. (1998) and van Vuuren et al. (1996) used Lolium perenne shoot material as patches, while Fransen et al. (1998) used soil patches against a sand background. In the study by Hodge et al. (1998) the L. perenne patch material was dual-labelled with 15N and 13C. As only 15N and not 13C enrichment of the plant material grown with these patches occurred, the plants must have been taking up N from the patch in inorganic form after microbial decomposition. As NO3 is highly mobile in soil it should not be necessary for roots to proliferate as much to this ion as to immobile ions such as phosphate, but this is what they do (Fig. 1). Moreover, molecular evidence confirmed a direct role for NO3 supplied locally in lateral root elongation rates in Arabidopsis thaliana (Zhang & Forde, 1998; Zhang et al., 1999). The problem of why roots respond to NO3, however, remained unresolved, or as Robinson (1996) asked, ‘why do plants bother?’.

Plants have evolved as members of a community in which both above- and below-ground competitive interactions are of importance in determining resource capture. Surprisingly, below-ground competitive interactions have seldom been studied directly, and most of the literature is based on individual plants or monocultures with competitive interactions inferred from these data. Maina et al. (2002) recently demonstrated that when two bean (Phaseolus varigaris) plants were grown with their root systems split evenly between two pots, overproduction of roots occurred, shoot : root ratios decreased, and reproductive yields declined compared with bean plants grown individually in a single pot. Soybean (Glycine max) plants responded similarly (Gersani et al., 2001). The identity of neighbours can have a large impact on root morphological responses (Huber-Sannwald et al., 1997) and competitive success in patch exploitation (Huber-Sannwald et al., 1996). In the studies by Maina et al. (2002) and Gersani et al. (2001), however, not only the same species but identical varieties were grown together. Thus it is puzzling how the roots could distinguish between roots produced by the same plant and those produced by the other in order to result in the observed overproduction of roots.

In experiments where two different plant species were grown together for a common organic patch, root proliferation did confer an advantage: there was a direct relationship between root proliferation in, and N capture from, the organic material (Fig. 2; Hodge et al., 1999a; Robinson et al., 1999). As in the study by Hodge et al. (1998), only 15N enrichments and not 13C were detected in the plant tissue, suggesting that N from the organic material was captured in inorganic N form. The organic material added (L. perenne shoots) would have contained a mixture of N sources. Release of N therefore would have occurred both spatially and temporally throughout the patch until microbial decomposition was complete. Thus root proliferation is important for N capture when plants are in interspecific competition for organic patches containing a finite local supply of mixed N sources. Remove one of these factors and the importance for N capture is obscured. This explains why, when plants are grown as individuals or in monoculture (Wiesler & Horst, 1994; van Vuuren et al., 1996; Hodge et al., 2000c) or in competition for simple inorganic N forms such as NO3 (Fitter et al., 2002), no relationship between root proliferation and N capture could be demonstrated. These results suggest that, in agricultural situations, root proliferation may benefit plant N capture only when in mixed cropping systems and when N additions are in slow-release forms. Although Maina et al. (2002) concluded from their work that bean plants with a reduced ability to proliferate roots should be selected for cultivation purposes to enhance yields, their variety of bean is commonly mix-cropped with maize. If the maize proliferates more than the cultivars of bean selected, bean yields may decline further.

Figure 2.

Nitrogen capture by Lolium perenne (squares) and Poa pratensis (circles) from a patch of decomposing L. perenne shoot material is a simple function of root length density in the patch when the plants are grown in competition. Points shown are means (n= 4) ± SE of plants harvested with time. From Hodge et al. (1999a) with kind permission from Blackwell Science Ltd..

Whether a plant proliferates roots in a nutrient patch probably depends on the plant's demand for the nutrient, the mobility of the nutrient within the plant and the strength of the patch relative to background fertility (Table 1). Proliferation responses to K are perhaps less commonly reported because plants are better able to redistribute K internally to satisfy demand (de Jager, 1982; Scott & Robson, 1991). In the case of N, the data in Table 1 suggest that when patches supply > 10% of the total plant N content, root proliferation occurs. There are exceptions. Poa pratensis did not proliferate roots in an N-rich patch, even when the N captured from the patch represented 13% of its total N content, but in another study where patch N represented a larger proportion of its total N (18%), proliferation did occur. By contrast, both L. perenne and Plantago lanceolata proliferated roots when the N captured from the patch represented only 2% of their total N content (Table 1). Different species may have different thresholds for different nutrients before proliferation occurs, and where values representative of natural soils and patches occur on this scale is currently unknown. There also is a need for more studies to measure the benefit of the proliferation response in terms of nutrient capture rather than simply the size of the proliferation response.

Table 1.  Root proliferation responses in organic patches of Lolium perenne shoots of contrasting C : N ratio, percentage of N originally added in the patch captured by the plant and N captured from the patch as a percentage of total plant N at harvest
C : N ratioTime (d)Plant speciesRoot proliferation in patch?% N originally added in patch captured by plantN captured from patch as a % of total plant NCalculated from
  1. (Yes), proliferation response did occur compared with controls without a patch, but only at final sampling date. *Studies that involved growing plant species together in interspecific competition.

31 : 139Festuca arundinaceaYes 4 7Hodge et al. (1998)
39Phleum pratenseYes 416 
39Poa pratensisNo 313 
39Dactylis glomerata(Yes) 511 
39Lolium perenneYes 510 
31 : 156Poa pratensis*Yes 518Hodge et al. (1999a)
56L. perenne*Yes 916 
22 : 122Plantago lanceolata and L. perenne*Yes 8 2Hodge (2003b)
12 : 170Poa pratensis and L. perenne*No26 1Hodge et al. (2000c)

4. Broader implications

The limited evidence from studies at the plant population level suggests that, even when individuals within the population respond to, and were affected by, nutrient heterogeneity (Casper & Cahill, 1996), there was a negligible impact at the population level (Casper & Cahill, 1996, 1998; Morris, 1996). However, these studies used plant monocultures and nutrient capture was not followed directly. Nutrient heterogeneity may have a more subtle impact on plant communities. For example, although capture of N from added organic material was similar regardless of whether L. perenne or P. pratensis was grown in monoculture or interspecific competition, within the interspecific treatment P. pratensis (which in this study also produced the greater root lengths) captured more of the N than L. perenne. Moreover, N capture from the added organic material was independent of the scale of heterogeneity, suggesting that roots were sufficiently plastic in their response to track the scale of heterogeneity and maintain performance (Hodge et al., 2000c). Nutrient supply also varies heterogeneously down the soil profile and plants differ in their ability to acquire nutrient from different depths (Berendse, 1982, 1983; Veresoglou & Fitter, 1984; Genney et al., 2002). This ability to acquire nutrients at depth can vary depending on which species are present and therefore competing for resources (Berendse, 1981, 1982; D’Antonio & Mahall, 1991; Jumpponen et al., 2002). For example, when in monoculture Achillea millefolium acquired more N (as 15NH4Cl) when injected at 5 cm rather than 20 cm depth. When grown with Festuca ovina, however, Achillea increased its acquisition of N added at the lower injection depth (Jumpponen et al., 2002). In addition, Fitter (1982) demonstrated that heterogeneous arrangement of acidic and calcareous soil resulted in higher plant diversity than a homogenous mixture of the two soils. Collectively, the differences in nutrient-acquisition patterns and the response of different plant species to patches suggests that nutrient heterogeneity promotes species coexistence. However, in addition to differences between species, considerable variation within species can occur depending on the scale of heterogeneity (Wijesinghe et al., 2001).

Plants may also avoid resource competition by displaying root system plasticity. Pea (Pisum sativum) plants with their root system split evenly between two pots (called ‘fence-sitters’; Gersani et al., 1998) hardly altered their total root biomass and fitness (measured as fruit d. wt) in response to an increasing number of competing pea plants grown in one of the pots, but the allocation of the fence-sitters’ roots between the two pots changed dramatically. For every additional competitor plant added, the fence-sitter shifted 0.12 g of its biomass from the pot containing the competitor plants into the pot with no competitors, so avoiding any detrimental effects of competition (Gersani et al., 1998). In the field, such diversion of resources into patches free of other competitor roots may not always be possible. Under natural conditions, nutrient concentrations are generally higher in the uppermost layers of the soil profile caused by litter inputs and their subsequent decomposition. Roots are also generally more concentrated in the top layers of the soil profile (McGonigle & Fitter, 1988; Mamolos et al., 1995; Jackson et al., 1996; Leuschner et al., 2001). Because of the natural development of the root system such distributions are to be expected, but in some cases root distributions do not relate to nutrient distributions (Caldwell et al., 1996). Additional constraints on root development and distribution in the field include soil physical and chemical properties, microorganisms both beneficial and detrimental to root growth, interactions with other roots and, at larger scales, the terrestrial biome in which the plant is growing (Jackson et al., 1996; Schenk et al., 1999; Robinson et al., 2003).

III. Root demography

Root biomass measurements are often insufficient to determine successful patch exploitation, and give no information about the processes of root production and death that occurred before measurement (Hendrick & Pregitzer, 1992). To follow these processes requires an understanding of root demography. For example, although there was little net change in numbers of Ambrosia artemisiifolia roots in either fertile or H2O control patches, both root birth and death rates increased in the fertile patch, thus root turnover was higher (Gross et al., 1993). Had only root numbers, length or biomass been measured, no response would have been detected. These fine roots associated with patch exploitation are expensive to construct particularly in terms of N, a key resource (Pregitzer et al., 1997). Fransen & de Kroon (2001) studied whether the proliferation of roots in nutrient-rich zones was disadvantageous in the long term by comparing shoot biomass of Holcus lanatus (characteristic of nutrient-rich habitats) and Nardus stricta (characteristic of nutrient-poor habitats) grown over two growing seasons in a heterogeneous nutrient treatment where 50% of the area was nutrient-rich and the other 50% nutrient-poor compared with homogeneous nutrient-rich or nutrient-poor treatments. Holcus proliferated roots of short lifespan preferentially in the nutrient-rich side of the heterogeneous treatment whereas Nardus did not. Nardus roots tended to live longer than Holcus, but this difference was only weakly significant (P = 0.072). Initially Holcus produced more shoot biomass in the heterogeneous treatment, but by the end of the first growing season this advantage had disappeared, and by the end of experiment Holcus biomass in the heterogeneous treatment was similar to that produced in the homogeneous nutrient-poor treatment. By contrast, shoot biomass of Nardus in the heterogeneous treatment was similar to that expected based on shoot biomass production from the homogeneous nutrient arrangements over both growing seasons. Thus Fransen & de Kroon (2001) concluded that selective placement of short-lived roots in nutrient-rich zones was a disadvantage in the long term. However, these two species tend to grow in different habitats and their different strategies to nutrient patches may be optimal for their particular environment.

Root demography is influenced by a wide range of abiotic and biotic factors including differences among species (Eissenstat & Yanai, 1997); root order and diameter (Eissenstat et al., 2000; Wells & Eissenstat, 2001); mean radiation flux (Fitter et al., 1998, 1999); temperature (Forbes et al., 1997; Edwards et al., 2004); pathogen populations (Kosola et al., 1995); mycorrhizal colonization (Hooker et al., 1995; Majdi & Nylund, 1996; Hodge, 2001, 2003a; King et al., 2002); and nutrient and water availability (Eissenstat & Yanai, 1997; King et al., 2002). Yet it is still unclear how root death is controlled. The link between plasticity in root branch structure and function (resource acquistion) is also unclear (Pregitzer, 2002). Following demography in trees is particularly difficult because of their widespread root system, and often it is unclear exactly where on the root system the roots being measured occur in relation to other root orders. Although fine root turnover in trees has been extensively studied, there is increasing evidence that all fine roots are not equal among different species, and to classify differences among species simply based on size attributes is unacceptable (Pregitzer et al., 1998, 2002; Pregitzer, 2002).

Therefore it is hardly surprising that no one strategy appears optimal for successful patch exploitation, with increased (Pregitzer et al., 1993), decreased (Hodge et al., 1999c), or no effect (Hodge et al., 2000d) on root longevity in fertile patches being reported. Although some of these differences may be caused by the different plant species studied, a single plant species (L. perenne) displayed a range of demographic responses to l-lysine patches of varying concentration and size, whereas N capture from the l-lysine was a simple function of the amount of N added (Hodge et al., 1999b). Thus the attributes of the patch itself may also influence the longevity of the roots exploiting it.

IV. Physiological plasticity

In addition to root morphological responses, physiological alterations in uptake capacity can play an important role in nutrient acquisition. Increased uptake may be caused by a higher uptake capacity (Vmax) or a higher ion affinity (low Km) of the roots or, more simply, an increase in nutrient (substrate) concentrations in the nutrient-rich patch. It is often assumed, based largely on the predictions of Grime (1979), that slow-growing species exhibit greater physiological plasticity than fast-growing species, as physiological plasticity is generally viewed as a less expensive alternative for the plant. Robinson & van Vuuren (1998) in their survey of the literature found support for Grime's predictions, but qualified their support by emphasizing it is equally important to resolve what these morphological or physiological responses actually achieve in terms of nutrient capture for different species. In other words, the plant response must be measured in the context in which the response is expressed.

When roots of nutrient-deprived plants are supplied with nutrients locally, uptake per unit of root usually increases (in 75% of the cases reviewed by Robinson, 1994). This increase is generally two- to threefold at the maximum (Robinson, 1994), although even greater increases have been reported (fivefold, Drew & Saker, 1978; sevenfold, van Vuuren et al., 1996, 11-fold, Robinson et al., 1994). However, such increases are transient (Burns, 1991; Robinson, 1994; van Vuuren et al., 1996). Physiological responses usually occur before morphological ones (Drew & Saker, 1978; Burns, 1991; Caldwell, 1994; van Vuuren et al., 1996), thus increased ion uptake may act as a signal of locally improved soil nutrient status and a trigger for new root investment in the patch zone. It also implies that while root physiological changes may be important in the short term, morphological responses are required for the longer-term exploitation of patches. By contrast, Fransen et al. (2001) concluded from their 2 yr study on competition between Festuca rubra and Anthoxanthum odoratum for nutrients supplied either homogenously or at two levels of heterogeneity, that higher physiological, rather than morphological, plasticity was critical for long-term competitive advantage because in the heterogeneous treatment both F. rubra and A. odoratum contained equivalent amounts of strontium (an analogue for Ca and a measure of root activity) injected at the end of the experiment, despite A. odoratum having a sparser root system. So, why do morphological responses occur at all if, as hypothesized by Grime (1979), they are so costly that only plants from fertile sites can ‘afford’ root proliferation, and if plants maximize nutrient acquisition at minimal cost (Bloom et al., 1985). Could it be the type of resource which is limiting that determines the response?

Physiological plasticity is generally assumed to be more important in enhancing capture of mobile ions, such as NO3, as uptake is limited not by diffusion in the soil but by uptake at the root surface. By contrast, immobile ions such as phosphate are limited more by diffusion to the root surface, so enhancing ion uptake at the root surface will not greatly enhance phosphate capture. The plant may need to respond rapidly to increase capture of mobile ions before they diffuse to other roots, while a rapid response is less crucial for immobile ions, allowing time for new root construction. Therefore it may be tempting to speculate that physiological responses occur in response to NO3 while morphological responses are more important in enhancing phosphate capture. However, physiological responses may also enhance phosphate capture provided the phosphate patch is exceptionally fertile (Caldwell et al., 1992), and morphological rather than physiological responses were found to be more important for plant N capture in the study by Hodge et al. (1999a). Furthermore, maintenance of high ion-uptake rates is very energy-consuming (van der Werf et al., 1988), thus they are not necessarily a less expensive alternative to new root growth, particularly in young plants on which most of the literature is based. Similarly, Robinson (2001) estimated that root proliferation costs may be as little as 0.2% of the plant's daily carbon gain, although this cost will differ according to the conditions the plant experiences, such as photosynthetic supply, rooting volume and patch size. Burns (1991) suggested that if the cost of maintaining high absorption rates in the part of the root system experiencing the patch becomes too high, the plant may switch to new root production instead. This could explain why alterations in kinetic rates are often reported to occur before root proliferation (Drew & Saker, 1978; van Vuuren et al., 1996) but not why slow-growing species generally exhibit greater physiological plasticity than fast-growing species, unless the balance of cost is seldom in favour of new root growth or the nutrient inputs are so transient their duration is insufficient to trigger new root production.

Caldwell and coworkers, from their field investigations on cold desert species of the Great Basin, USA, have demonstrated that physiological responses of roots extracted from N-P-K solution patches can be both rapid, occurring within a 3 d period, and large, showing as much as 80 and 88% higher uptake rates for phosphate and N, respectively, than control roots extracted from the same plant but which had received an H2O patch (Jackson et al., 1990; Jackson & Caldwell, 1991). Although these values are certainly impressive, their ecological significance may be questionable as they were obtained under laboratory assay conditions. Under natural conditions the ions must diffuse to the root surface at sufficient rates to satisfy this increase in uptake capacity. Expansion of the microbial population in these nutrient patches may effectively block ion diffusion to the root, or block it sufficiently so that the increased uptake rate is ineffectual. However, as the greatest relative increases in NH4+ uptake occurred at the lowest assay concentration (50 µm), which reflected the typical soil solution concentration (Jackson & Caldwell, 1991), this would suggest that such increased uptake kinetics could represent an important mechanism for enhanced nutrient uptake under field conditions. Increased root-uptake kinetics for both NH4+ and NO3 in these cold desert species have also been reported following pulses of simulated rain (BassiriRad & Caldwell, 1992a, 1992b; Cui & Caldwell, 1997a; Ivans et al., 2003), but not always (BassiriRad et al., 1999), and there is considerable variation among these species in their ability to alter uptake kinetics (BassiriRad & Caldwell, 1992a, 1992b; Ivans et al., 2003). A physiological response to these small summer rain events makes more economic sense than producing new roots because of the transient nature of the N pulse, but even this response may play only a minor role. More important were the increases in N diffusion to the root surface because of the elevated soil moisture and the increased microbial turnover of soil N pools releasing N for root capture (Ivans et al., 2003). To understand root responses to nutrient patches (or pulses) therefore requires an understanding of the interaction between soil microorganisms and the attributes of the nutrient patch itself.

V. Root plasticity in patches in competition and symbiosis with microorganisms

Root proliferation, which involves the production of new roots, is generally slow to occur (approx. 35 d for a range of grasses, Hodge et al., 1998, 1999a), thus it is unlikely that root proliferation allows the plant to compete directly with the microbial community. Experimental evidence also suggests that plant nutrient capture from added organic material starts to increase when microbial and soil fauna populations decline (Griffiths et al., 1994; Hodge et al., 2000e). Plants outcompete microorganisms in the long term, probably because of their longer turnover times (reviewed by Kaye & Hart, 1997; Hodge et al., 2000a).

Although often ignored in root-proliferation studies, most plant species have an additional mechanism to enhance nutrient acquisition, namely mycorrhizal symbiosis. It is well established that fungi in the ericoid and ectomycorrhizal association possess saprophytic capabilities which enable nutrient (predominantly N) capture from complex organic sources to which the plant would not normally have direct access (Smith & Read, 1997; Chalot & Brun, 1998). It is also recognized that different fungal species, and even strains of the same species, differ widely in their ability to capture nutrients from such complex sources (Smith & Read, 1997).

In common with the response of roots, both free-living (Dowson et al., 1989; Ritz et al., 1996) and mycorrhizal symbiont fungi of both the ectomycorrhizal (Bending & Read, 1995) and AM symbiosis (Mosse, 1959; Nicolson, 1959; St John et al., 1983a, 1983b; Hodge et al., 2001) can proliferate hyphae in nutrient-rich zones, often at the expense of growth elsewhere. For the purpose of this review, only the response of the colonized root to nutrient patches is of interest, and information from the ectomycorrhizal and ericoid associations is lacking in this respect. A limited amount of data on the impact of AM colonization is available. As fungal hyphae also proliferate in patches, and as root proliferation responses are generally slow to occur, Tibbett (2000) suggested that root proliferation will become obsolete when roots are colonized by mycorrhizal fungi. Because of their smaller size, AM hyphae should: (i) be less expensive to construct than fine roots in terms of C (Fitter, 1991) and presumably N (Pregitzer et al., 1997); (ii) be able to detect and respond to short-lived nutrient patches or pulses that the roots cannot detect; and (iii) be able to penetrate to the sites of patch decomposition and therefore be able to compete directly with other soil microorganisms for the nutrients released. Also, as AM fungi (AMF) obtain a C supply from their host plant, this should give the AMF a competitive advantage over other soil microorganisms should C become limiting in the patch zone. Thus proliferation of AMF hyphae instead of roots should, in theory, be more beneficial to the host plant, but does the experimental evidence support this view?

The evidence available shows that, generally, AMF have little impact on nutrient capture from patches other than that of P, although there is one report of enhanced N capture, but only when plants were grown in interspecific competition (Hodge, 2003b; Table 2a). Furthermore, root responses are demonstrably more important than those of AMF when both are present in the nutrient patch. Three different AMF (Glomus mosseae, Glomus hoi or Scutellospora dipurpurescens) showed different levels of internal colonization and production of external mycelium but, whereas none of the fungal parameters responded to a glycine patch, the roots did. In all cases, including the uncolonized control, net numbers of roots were higher in glycine compared with H2O control patches (Hodge, 2001). Although in the presence of roots AMF may be relatively unresponsive to nutrient patches, they may exert their influence through modifications of the root's response. AMF colonization has been shown to increase root production and alter root length densities in patches (Cui & Caldwell, 1996a; Hodge et al., 2000b). Moreover, in the absence of roots AMF have been demonstrated to capture and transfer both P and N (although not when added as NO3) from patches to their host plant (Table 2b). In the field, plants can be linked into a common mycelial network. This has led to the suggestion that plants linked by this common mycelial network may benefit by the redistribution of nutrients along the network, thus reducing the consequences of heterogeneity for individual plants (Ozinga et al., 1997). However, the little experimental evidence available suggests nutrient transfer along the common mycelial network occurs only over very short distances (Chiariello et al., 1982; Newman, 1988), and we are only just beginning to address how mycorrhizal fungi may influence, either directly or indirectly, nutrient capture from heterogeneous nutrient supplies.

Table 2.  Influence of arbuscular mycorrhizal (AM) colonization on nutrient capture when (a) both roots and AM hyphae are present in the patch, and (b) only AM hyphae have access to the patch
(a) Roots and AM hyphae together in patch
Patch addedInfluence of AM colonization on nutrient capture compared with control*Reference
GlycineNo difference in N captureHodge (2001)
Lolium perenne shootsNo difference in N captureHodge et al. (2000b)
NO3, H2PO4No difference in N capture; enhanced P captureCui & Caldwell (1996a)
SoilOnly one species (of seven) showed enhanced P uptakeFarley & Fitter (1999b)
L. perenne shootsIncreased N capture but only when in interspecific plant competitionHodge (2003b)
(b) AM hyphae only in patch
Patch addedInfluence of AM hyphae on nutrient capture compared with control*Reference
  • *

    Controls: studies marked

  • †used an indigenous mycorrhizal control, those marked

  • ‡used a nonmycorrhizal control;

  • §

    Hodge et al. (2001) used an AM-colonized control with hyphae denied access to the patch. Data are from microcosm studies.

NO3, H2PO4No difference in N capture; only AM colonized plants captured PCui & Caldwell (1996b)
L. perenne shoots§Enhanced N captureHodge et al. (2001)

VI. Influence of patch attributes

Nutrient patches are both spatially and temporally variable. Fitter (1994) further categorized these patches based on various attributes (Fig. 3). The intensity or strength of the patch encountered is also important. For example, although not measuring root proliferation directly, Jingguo & Bakken (1997) found more root biomass was allocated to N-rich clover patches than to N-poor straw patches, while Jackson & Caldwell (1989) observed that the timing of the response by roots after application of half-strength N-P-K patches was the same as to full strength patches, but the magnitude of the response was less. The combination of differing responses by plants to patches and intrinsic variation in patches themselves (Fig. 3) anticipates a wide range of responses in the natural environment. Farley & Fitter (1999b) examined how seven co-occurring woodland herbaceous perennials responded to varying patch size (40, 70 or 160 cm3) and quality (100% soil or 50% soil, 50% sand in a background of 100% sand). The probability of encountering the patch was the same in all treatments. Four (Silene dioica, Veronica montana, Ajuga reptans, Glechoma hederacea) of the seven species proliferated roots by increasing root length in the patch, but only two (S. dioica and V. montana) responded to the quality of the patch, and only one (G. hederacea) was sensitive to patch size. Such subtle differences in species response to different patches may enable coexistence but, as nutrient capture from the patches was not directly followed, this cannot be stated with any certainty. The results of the few studies that have investigated some aspect of patch variation and its impact on plant nutrient capture are shown in Table 3.

Figure 3.

Basic attributes of patches (distribution, extent and number) on a temporal and spatial scale (after Fitter, 1994) and some predictions of root responses to these patch attributes, all of which will be determined by the strength or intensity of the patch above background fertility.

Table 3.  Influence of various patch attributes on nutrient capture by different plant species
ScalePatch typePatch attributePlant speciesNutrient capture from patchReference
  • *

    These studies used 13C, 15N dual-labelled organic material as patches. At no time were 13C enrichments detected in the plant tissue.

TemporalNutrient solutionNutrient pulses of varying durationArrhenatherum elatius spp. bulbosum Festuca ovinaN capture by Arrhenatherum greater from long-duration pulses and by Festuca from short pulsesCampbell & Grime (1989)
Temporall-lysineSmall, frequent pulse or a large, single patch*Lolium perenne swardNo difference in rate or total amount of N capturedHodge et al. (1999c)
Spatiall-lysineDiffering concentration and volume*L. perenne swardAbsolute N capture simple function of amount added; relative N capture depended on patch concentrationHodge et al. (1999b)
SpatialPhosphateDiffering in number, concentration and distanceArtemisia tridentataAgropyron desertorum Pseudoroegneria spicataAll species captured most P from closest patch. From distant patches, P capture by Artemisia and Agropyron increased with increased patch numbervan Auken et al. (1992)
SpatialNitrate and phosphateTwo patches of varying quality (both containing NO3 + H2PO4− or one NO3 only and one H2PO4 only)Artemisia tridentataAgropyron desertorumNo difference in N capture. P capture by Artemisia higher from phosphate-only patchCui & Caldwell (1998)
SpatialUrea l-lysine Algal amino acids Algal lyophilized cellsL. perenne shootsOrganic patches of varying chemical and physical complexity*L. perenne swardN capture highest from patches of low C : N ratioHodge et al. (2000d)
SpatialEarthworm L. perenne shootsOrganic patches of contrasting C : N ratio*L. perenne swardN capture highest from lowest C : N ratio (earthworm) patchHodge et al. (2000e)
SpatialPhosphateFollowing addition of a ‘primary’ patch, additional patches varying in number and concentration appliedArtemisia tridentataUptake capacity of roots in the primary patch reduced only by highest level of additional patchesDuke & Caldwell (2000)

Nutrient patches can vary as much on a temporal scale as they can spatially, yet this aspect has been less well studied (Table 3). The duration of the nutrient pulse had a large impact on N capture by two grass species in the study by Campbell & Grime (1989). By contrast, neither the rate nor the total N captured by another grass species (L. perenne) differed if the patch had been added as a single large addition or as a series of smaller pulses at regular time intervals (Hodge et al., 1999c; Table 3). Similarly, on a spatial scale, varying patch size and concentration had little impact on absolute N and P capture, with only P capture by Artemisia being affected (Cui & Caldwell, 1998; Hodge et al., 1999b; Table 3). Under natural conditions plant root systems will be exposed not to a single patch but to several, all of which may be of varying nutrient concentration, size and duration. Capture of phosphate by Agropyron and Artemisia from patches added at a 5 cm distance increased as the number of patches increased, even though these patches were of lower nutrient concentration, simply because roots were more likely to encounter them. By contrast, Pseudoroegneria could not detect these patches and thus hardly captured any phosphate from them (Table 3). Studies on patches of varying durability and quality, however, have demonstrated that plants can capture large amounts of the N originally available, particularly from simple patches, despite microbial decomposition of the material occurring first (Hodge et al., 2000d, 2000e; Table 3).

Roots are also likely to encounter different patches at different times. When part of the root system locates a new patch, root proliferation in the original patch may be subject to higher turnover if nutrients in the original patch are depleted, and if the new patch contains sufficient nutrient concentrations to divert resources away. Thus, a plant root system should be capable of exploiting a number of patches at the same time, although root proliferation and/or enhanced nutrient uptake of roots in these patches may be expected to be reduced or ‘damped down’. By contrast, Duke & Caldwell (2000) found that, although roots exposed to an initial or ‘primary patch’ containing phosphate had greater uptake capacities than controls receiving no soil phosphate amendment, the addition of further patches to other areas of the root system did not reduce the uptake capacity of roots in the primary patch except at the highest level of additional phosphate amendment (Table 3). Thus there was no clear evidence that root-uptake capacities in the primary patch were regulated by the overall nutrient enrichment of the root system. Also, there was no apparent increase in uptake capacities of roots exposed to the additional patches of low and medium concentrations compared with controls (Duke & Caldwell, 2000). Plant tissue P concentration did not differ among treatments. This may have been because the increased soil phosphate concentrations resulted in P acquisition which met the demands of the plant, even though generally uptake capacities did not alter.

Compared with root responses to patches, the attributes of the patch itself have largely been ignored, with most studies applying a single heterogeneous treatment in comparison to a uniformly nutrient-rich or deficient control. However the dynamics of organic patch decomposition can be as important as the root response to these patches (Hodge et al., 2000d; 2000e). Only by following both together can we understand why different plant species respond to varying patches in certain ways. Moreover, plant capture from nutrient patches, whether in organic or inorganic form, will depend not only on the response of the plant's root system itself, but also on complex interactions with other plant roots, soil microorganisms and fauna, and physical/chemical interactions in the soil (Schenk et al., 1999; Hodge et al., 2000d; Bengough, 2003; de Kroon et al., 2003). If these factors are ignored, the relevance of the plant response to patches may be open to misinterpretation. For example, in the competition study by Hodge et al. (1999a), if the patch had been added as an NO3-rich solution it is questionable whether the ‘benefit’ of enhanced root length production by L. perenne would have been observed because of the rapid mobility of the NO3 ion in soil. More studies using relevant material, such as dead roots, leaves, soil animals, etc., as patches added to soil systems, and close monitoring of the decomposition dynamics of such material, are required to place root responses and the subsequent decomposition and nutrient capture from patches in a more appropriate ecological context.

VII. Control of root proliferation

Both root physiological and morphological plasticity comes at the price of ATP expenditure to maintain high ion-uptake capacities, construction of new roots or increased extension rates of existing roots. Often, reduced growth outside the patch zone is observed (Drew, 1975; Granato & Raper, 1989; Gersani & Sachs, 1992; Linkohr et al., 2002), which may offset some of the cost of new root production inside the patch. Moreover, it seems reasonable to assume these costs must provide a net benefit (and, as discussed above, under certain conditions have been demonstrated to do so); but what happens when photosynthetic supply becomes restricted?

A rather severe (but ecologically realistic) method of reducing photosynthetic supply to the roots is by clipping or grazing; a less invasive method is achieved by shading. Following clipping of 80% of the shoot material of three semiarid range grasses the differences between the species in response to nutrient treatment did not alter but were further exaggerated (Arredondo & Johnson, 1999). Studies comparing shaded and unshaded plants have demonstrated that exploitation of phosphate patches is particularly demanding of photosynthetic C supply (Jackson & Caldwell, 1992; Cui & Caldwell, 1997b). Shading can also reduce the physiological uptake of N (Cui & Caldwell, 1997b), but not always (Jackson & Caldwell, 1992), probably depending on the N demand by the plant at the time. Relative growth rate (RGR) of roots in nutrient-rich patches can also be adversely affected by shading. The RGR of Agropyron desertorum roots in N-rich patches was reduced by more than 50%, while root RGR in H2O control patches was unaffected (Bilbrough & Caldwell, 1995). Shading may exert its influence through a reduction of the total nonstructural carbohydrate concentration, particularly in shoots (Jackson & Caldwell, 1992).

Shading may also alter root plasticity differently and thus influence competitive interactions. While unshaded sweetgum (Liquidambar styraciflua) was more plastic than loblolly pine (Pinus taeda), when shaded both were equally plastic because of shaded loblolly pine allocating its roots with more precision (Mou et al., 1997). Both Heliocarpus pallidus and Caesalpinia eriostachys monocultures showed enhanced root proliferation in fertile patches when light levels increased, but when grown together in interspecific competition, root proliferation by the fast-growing Heliocarpus was reduced under both high and low light levels whereas Caesalpinia was affected only by low light (Huante et al., 1998). Collectively, these studies indicate that capture of heterogeneous nutrient supplies requires considerable C investment by the plant. If shading simply reduced the nutrient demand by the plant, then capture of nutrients would be lower regardless of their spatial placement. Clearly this is not the case. Moreover, shaded (30% of full sunlight) sweetgum and loblolly pine seedlings had reduced total biomass when grown in a patchy environment than when grown in a uniform nutrient environment (Mou et al., 1997), again highlighting that growth in patches is more expensive than under uniform conditions.

The identification of the first component of the NO3 signalling system, the gene ANR1, by Zhang & Forde (1998) represented a major step forward in understanding how root proliferation is controlled. In addition, an Arabidopsis mutant (nia1 nia2) with very low nitrate reductase activity (0.5% of that found in the wild type) showed a similar response to NO3-rich zones, implying that the signal for increased meristematic activity comes from the NO3 ion itself, not a product of NO3 metabolism (Zhang & Forde, 1998). Subsequent work by Forde and coworkers (Zhang et al., 1999; Zhang & Forde, 2000; Forde, 2002) has suggested that the auxin-sensitive gene axr4 may also be involved in the NO3 signal transduction pathway because of the failure of axr4 mutants to respond to localized NO3 additions. By contrast, Linkohr et al. (2002) found that several auxin-resistant mutants including axr4 responded to localized NO3-rich zones in the same manner as the wild type. Although auxin plays an important role in many plant growth, development and physiological processes, including being involved in lateral root development (Evans et al., 1994; Boerjan et al., 1995), the results of Linkohr et al. (2002) argue against a role for auxin in the NO3 signal transduction pathway. Similarly, using phosphate-rich patches, Williamson et al. (2001) reported wild-type Arabidopsis root-system architecture changes in the auxin mutants axr1, aux1 and axr4, suggesting that auxin does not play a role in root responses to phosphate-rich patches either. More research into the mechanistic control of root proliferation responses is still clearly required, but with the recent sequencing of the Arabidopsis genome research advances in this area should be forthcoming.

VIII. Conclusions

Root morphological and physiological responses are far from uniform across species, and even within species the size of the response depends on the scale of heterogeneity. Future research should focus on the importance of root plasticity for nutrient capture rather than simply measuring the size of the response. This also involves more accurate measurements of the costs associated with physiological and morphological responses and growing plants under more ecologically relevant conditions with more realistic patch materials. Moreover, the proliferation by roots observed under controlled conditions, which are generally manipulated so as to maximize such responses, is likely to be significantly reduced under natural conditions where other environmental factors will influence plant growth responses. Most of the data reviewed in this paper come from studies using only one or two plants grown at a time. More studies at the plant community level are required if we are to determine whether heterogeneity is important for plant community dynamics or simply exerts its influence locally on individual members in the community. Such studies are technically challenging, but are essential and should benefit from the application of molecular approaches. For example, by the use of molecular markers, Jackson and coworkers (Jackson et al., 1999; Linder et al., 2000) were able to trace the distribution of roots in soil. If such approaches could be combined with microbial and nutrient distributions, our understanding of plant competitive interactions in patchy soils would be greatly advanced.

At the other end of the spectrum, molecular research is also needed on the signal pathways involved in root proliferation responses. Some progress has been made with the finding that nitrate is involved directly in at least part of the signal pathway, and the identification of the nitrate-inducible gene ANR1, in root proliferation responses of Arabidopsis, but it remains to be determined whether other ions (notably phosphate) are equally specific. The use of model plants such as Arabidopsis lend themselves to such molecular investigations, but ultimately if more ecologically important issues relating to plant responses to soil heterogeneity are to be addressed then a wider range of plant species from different habitats must be examined.


A. H. is funded by a BBSRC David Phillips Research Fellowship. I thank Alastair Fitter, Richard Norby, Rob Jackson and two anonymous referees for their extremely helpful comments on the manuscript.