Morphological and topological responses of roots to defoliation and nitrogen supply in Lolium perenne and Festuca ovina


Author for correspondence: L. A. Dawson Tel: +44 (0)1224 498200 Fax: +44(0)1224 311556 Email:


  • • This study examined morphological and topological responses of nodal root axes to defoliation in a fast- and a slow-growing grass species.
  • • Vegetative tillers of both Lolium perenne and Festuca ovina were grown on slant boards and either left intact or subjected to repeated defoliation, under both a high nitrogen (N) and a low N supply. Root length, diameter and branching characteristics were measured on individual nodal root axes.
  • • The total axis root length of F. ovina was less when plants had been defoliated. Root axis weight, primary root axis length and primary root diameter were also less with defoliation than an undefoliated control, under high N. Under low N conditions the root axes of F. ovina had a more randomly branched topology without defoliation. For L. perenne under low N conditions, the length of the primary root axis was longer with defoliation than in an undefoliated control, while the primary root axis diameter decreased. By contrast to F. ovina, the root axes of L. perenne had a more randomly branched topology without defoliation only when supplied with high N.
  • • The greatest plasticity in branching caused by defoliation was observed under high N for L. perenne and under low N for F. ovina. Although grass root axis topology has, in general, a herringbone in structure, the nodal root system can alter root axis structure in response to defoliation.


Root systems of graminaceous plants consist of two main components: seminal roots arising from the primordia laid down in the embryo, and nodal or adventitious roots, which arise from the basal nodes of the main shoot and tillers, and research has shown that these root types can behave very differently (Wahbi & Gregory, 1995). Some authors have suggested that the term ‘shoot-born roots’ better distinguishes these nodal roots (Harper et al., 1991). Much of the previous work on root response to defoliation and nutrient supply has used seedling plants, whereas this study examines the response of the nodal root system of the vegetative tillers, being more ecologically relevant to the field situation. In addition, much of the previous research on root response to defoliation and nutrient supply have been as single factors, while our study examines these together and aims to identify common responses of two grasses.

Research on defoliation effects on plant growth and competition has concentrated on above-ground responses, however, below-ground responses can be equally important (Dawson et al., 2000). For grasses, partial removal of above-ground tissue results in less assimilate allocated below ground and causes remobilization of root and crown storage compounds to support growth (Briske, 1996; Donaghy & Fulkerson, 1998). Such responses aim to optimize the balance between acquisition of nutrients and carbon. Root mass (Ennik & Hofman, 1983; Holland & Detling, 1990; Matthew et al., 1991), length and elongation rate (Evans, 1971; Richards, 1984; Jarvis & Macduff, 1989) have all been shown to be reduced with defoliation, compared with an intact plant, but few studies (with the exception of Arredondo & Johnson, 1998, 1999), have examined the distribution of root branching as influenced by defoliation. In particular, no previous study has looked at common responses within the nodal root system to defoliation and nitrogen (N) supply.

Plant responses to multiple factor stresses may differ from those seen as a result of single-factor stresses. For example, response to defoliation may depend upon the availability of soil nutrients, in particular N. Plant response to a low N supply is, in general, reduced accumulation of dry matter and a greater proportion of biomass partitioned below ground (Marschner et al., 1996). However, grass growth is often limited to a greater extent than photosynthesis under stress conditions, such as those of low nutrient status and low temperature. Although defoliated plants can show increased nutrient absorption per unit root (Rosenthal & Kotanen, 1994), this might not fully compensate for a defoliation induced reduction in root mass or length when nutrient availability is low (Chapin & McNaughton, 1989). It has been suggested that species from poor habitats or growing in infertile soil will exhibit simple branching patterns (herringbone), whereas species from fertile habitats will exhibit a more complex topology (random, moving to dichotomous) (Fitter et al., 1991). Hypotheses about the branching patterns of root systems indicate that herringbone root systems (i.e. consisting of only a main axis and first-order laterals), are most effective at acquisition of resources because they minimize intraplant competition (Fitter et al., 1991). However, associated with herringbone topology is a large tissue volume and high construction costs (Fitter et al., 1991). Arredondo & Johnson (1998) found that root topology in three range grasses was little affected by defoliation, while Taub & Goldberg (1996) found altered resource availability did not affect root topology in seedling grasses. However, Arredondo & Johnson (1999), also using seedling plants, found that root topology changed with nutrient and defoliation treatments in three range-grass cultivars.

To test whether defoliation had common effects on root morphology and topology in vegetative pasture grass, we chose two pasture grasses, Lolium perenne and Festuca ovina. Both species are important grasses and are commonly found in the UK; Lolium perenne is a relatively fast-growing grass, associated with improved and fertile pastures, whereas F. ovina is a relatively slow-growing grass, which is found in less productive, infertile, more extensive upland grazing lands. In addition, the chosen cultivars reflect the most widely used seed source for reseeding. In addition, to represent the more favoured environment for these two grasses, we supplied both high N and low N nutrient treatments. In this study, vegetative tillers, with nodal root growth only, were used to best represent root responses in an established plant, rather than seedlings, which have been used in studies on this topic previously (Arredondo & Johnson, 1998, 1999; Taub & Goldberg, 1996). The working hypothesis was that defoliation would result in a smaller root axis size and branching relative to the intact plant, in both species, but that N would have a contrasting effect on the root response in the two grasses.

Materials and Methods

Plants were grown using a modified slant board plant culture method (Fig. 1) (Kendall & Leath, 1991). The angled slant boards were set up within a controlled environment growth room (Conviron, Winnipeg, Canada); relative humidity was constant at 60%, irradiance was provided for a 16-h photoperiod with 600 µmol m−2 s−1 photosynthetically active radiation at plant height, temperature was 20°C in the light and 12°C in the dark. Individual vegetatively propagated tillers of either L. perenne L. cv. Magella or F. ovina L. cv. Bornito were used. Two to four main nodal root axes were placed directly between two cloth sheets, with the stem base 5 cm below the top edge of the tray. Although the root system grown in a slant board will not reflect natural soil-grown root systems, it allows the complete removal of the whole root system, including all orders of laterals on the nodal roots, while permitting well-supported two-dimensional growth to take place. All plants were continually irrigated with a nutrient solution, applied at a rate of 100 cm3 h−1, which trickled through a Perlite bag (Fig. 1), to retain moisture around the roots. Excess nutrient solution ran freely from the board.

Figure 1.

Diagram of slant board design, adapted from Kendall & Leath (1991).

The nutrient solution contained 2.0 mol m−3 NH4+NO3, was fully balanced for other nutrients (Thornton et al., 1993) and was supplied for a period of 3 wk to establish plant growth. Defoliated plants were cut to a height of 4 cm from the stem base three times each week. The plants were laid out in a randomised block design consisting of four blocks, each block containing one replicate of each treatment combination. After the initial preconditioning period, plants were subjected to one of four treatments: undefoliated low N supply (ULN), undefoliated high N supply (UHN), defoliated low N supply (DLN) and defoliated high N supply (DHN). Low N supply was 0.02 mol m−3 NH4+NO3 and high N supply was 2.0 mol m−3 NH4+NO3, with all other nutrients held constant. These nutrient rates were supplied to provide tissue concentrations comparable to those found in the field, determined in a previous study (Pratt, 1997).

Plants were destructively harvested when the root system approached the bottom of the slant board, 14 d after imposition of treatments for L. perenne and at 28 d for F. ovina. A random subsample of five individual root axes was taken and the fresh weight of each primary root axis was measured. The root axes were then immersed for 1 h at 4°C in a 1 : 1 (v : v) mixture of 0.01% methyl violet and glycerol. The stained roots were then rinsed using deionized water with all laterals spread carefully onto glass plates, 0.2 × 0.2 m and 0.003 m thick. A thin layer of glycerol was added using a Pasteur pipette and a sheet of transparent acetate was placed on top. The glass plate was placed in a DeVere 504 photographic enlarger (Godstone, Surrey, UK), projected onto a calibrated digitizing tablet and a stylus was used to measure digitally the diameter of the projected root image. The primary root axes were measured at 10 randomly chosen locations along their length to obtain a mean diameter for each primary root. No diameter measurements were made on any other order of root.

The individual spread axes were then scanned into an AppleMac computer, edited to remove loops or holes using NIH image (Freeware, and analysed for topology using branching software (Berntson, 1992). Root systems can have the same degree of branching (i.e. number or length of branches per unit root), but can have a different spatial arrangement. An analysis of the branching arrangement can be obtained by using topological methods such as those developed by Fitter (1985, 1987). Root topology refers to how individual root segments are connected to each other through branching. The root system is divided into individual root segments and to provide information on the number of external root segments that end in a meristem, the concept of magnitude (µ) is used. Magnitude is the number of exterior links in the system (an exterior link is an internode ending in a meristem). Trees of equal magnitude can differ in the total exterior pathlength (Pe), which is the sum of numbers of all path lengths from all exterior links to the base, and altitude (a) that is the number of links in the longest individual path. As well as these measured variables, the topology of the root system are characterized by other indices, such as the slope of altitude to magnitude and the slope of total pathlength to magnitude, which are independent of root size (Werner & Smart, 1973; Fitter, 1986). A ‘herringbone’ topology is one where branching is restricted to the main axis, while a ‘dichotomous’ is one where branch initiation occurs with equal probability on all external links.

An analysis of variance and slope analysis were performed on each species separately, using genstat for Windows, version 5 (VSN International Ltd., Oxford, UK). The slope of the linear regression of log altitude on log magnitude and log pathlength on log magnitude were compared, with larger values (steeper positive slopes) indicating a more herringbone topology (maximum = 1 for log altitude ratio and maximum = 1.92 for log pathlength ratio), referred to as topological indices. The mean value of the five replicate axes from each single plant was used as a single value in subsequent analyses.


The effects of defoliation on the structure of a randomly chosen root axis for (1) F. ovina under low N and (2) L. perenne under high N supply, where significant effects of defoliation in reducing root axis spread can be seen in Fig. 2.

Figure 2.

The effect of repeated defoliation (three times weekly) under contrasting levels of nitrogen supply (high N, 2.0 mM NH4+NO3; low N, 0.02 mM NH4+NO3) on branching of a randomly chosen root axis of (a) Festuca ovina under low N supply and (b) Lolium perenne under high N supply. Arrows show the point of attachment to the shoot.

Festuca ovina

The fresh weight of the root axis (F1,9 = 18.93, P = 0.002) was less under defoliation than in the undefoliated control at both N levels (Fig. 3a). The mean diameter of the primary axis was also less under defoliation than in the undefoliated plants, although only under high N conditions, as shown by the significant interaction term (F1,9 = 5.95, P = 0.037, Fig. 3). With undefoliated plants, the mean diameter of the primary root was smaller under low N conditions than under high N conditions (Fig. 3b).

Figure 3.

The effect of repeated defoliation (three times weekly; D) and undefoliated control (U) under contrasting levels of nitrogen supply (high N (HN), 2.0 mM NH4+NO3; low N (LN), 0.02 mM NH4+NO3) on the morphology of root axes of Lolium perenne (shaded bars) and Festuca ovina (open bars). (a) Mean root axis fresh weight (mg); (b) mean primary root axis diameter (µm); (c) mean length of primary root axis (mm); (d) mean length of first-order laterals attached to a single primary axis (mm); (e) mean length of first-order lateral roots per length of primary root (mm mm−1); (f) mean length total root (mm). Columns within a species not sharing a common letter are significantly different at P < 0.05 (df = 9) and bars represent the LSD value for the interaction (shaded bars for Lolium perenne and open bars for Festuca ovina).

Total root axis length in F. ovina was less with shoot defoliation (F1,9 = 23.48, P < 0.001; Fig. 3f), mainly as a result of a reduction in length of first-order laterals at both N levels (F1,9 = 18.49, P = 0.002) (Fig. 3d). Primary axis root length was less under defoliation only with high N supply, as shown by the significant interaction term (F1,9 = 6.75, P = 0.03) (Fig. 3c). There were fewer total numbers of links in the root system of F. ovina when plants were defoliated (F1,9 = 38.81, P < 0.001) at both N levels (Table 1). The length of first-order laterals per unit primary root length was less with defoliation (F1,9 = 10.45, P < 0.01) (Fig. 3e). The topological measures – log magnitude and log altitude – were also significantly altered with defoliation (F1,9 = 34.51, P < 0.001 and, F1,9 = 29.13, P < 0.001, respectively) (Table 1). The root topology index (Pe/µ) indicated that the roots of F. ovina were significantly more randomly branched when plants were undefoliated than when defoliated under low N conditions (Table 2).

Table 1.  Topological analysis of root axes of Festuca ovina on day 28 of the experimental period and of Lolium perenne on day 14 of the experimental period
  1. UHN, undefoliated high N supply; DHN, defoliated high N supply; ULN, undefoliated low N supply; DLN, defoliated low N supply. Low N supply was 0.02 mol m−3 NH4+NO3 and high N supply was 2.0 mol m−3 NH4+NO3, with all other nutrients held constant. df = 9. Superscripts a and b denote a significant difference at P < 0.05.

Festuca ovina
Total number of links in axis557b218a638b304a122
Log altitude2.35b2.00a2.29b2.05a0.12
Log magnitude2.42b2.03a2.49b2.09a0.15
Lolium perenne
Total number of links in axis248a199a193a312b121
Log altitude1.92a1.89a1.95a2.15a0.18
Log magnitude2.04a1.91a1.98a2.19a0.24
Table 2.  Slopes of log altitude (alt) and a log path length (Pe) on log magnitude (µ)
 Alt/µ Without defoliationWith defoliationPe/µ Without defoliationWith defoliation
  1. The critical topological benchmarks derived from simulated root systems (Werner & Smart, 1973) are 0.59 (random) and 1.00 (herringbone) for alt/µ, and 1.52 (random) and 1.92 (herringbone) for Pe/µ. Data are means ±SE; superscripts a and b denote a significant difference at P < 0.05.

Festuca ovina
LNa0.702 ± 0.397a0.850 ± 0.131a1.308 ± 0.395b1.821 ± 0.130
HNa0.720 ± 0.242a0.856 ± 0.460ab1.595 ± 0.241b1.831 ± 0.459
Lolium perenne
LNb0.880 ± 0.153b0.948 ± 0.191b1.936 ± 0.241b1.777 ± 0.302
HNa0.498 ± 0.058b0.894 ± 0.045a1.479 ± 0.091b1.889 ± 0.072

Lolium perenne

In general, treatments had fewer significant effects on parameters measured (Table 1b) on L. perenne than on F. ovina. Lolium perenne plants subject to defoliation had thinner primary root axes, irrespective of N supply (F1,9 = 12.23, P = 0.007) (Fig. 3b). Under high N conditions, root axis fresh weight was significantly less with defoliation compared with undefoliated plants (F1,9 = 6.59, P = 0.05) (Fig. 3a). Although defoliation had no effect on the total root axis length, length of laterals or density of laterals (Fig. 3f), the length of the primary root axis was significantly greater with defoliation under the low N supply than in undefoliated control plants (F1,9 = 14.36, P = 0.004) (Fig. 3c). There was a significant difference in root topology with defoliation, the topological indices reflecting a more randomly branched structure without defoliation than with defoliation, under the high N treatment only (Table 2).


The effect of defoliation on root morphology and topology of vegetatively propagated grass tillers was marked, but depended upon the N supply. Root system size and diameter was less under defoliation in both species, allowing allocation of carbon preferentially to the shoot system for regrowth (Richards, 1984). However, under low N supply, L. perenne was unaffected. These effects may reflect a relatively greater N limitation under low N conditions for the faster-growing L. perenne compared with F. ovina. In addition, for L. perenne, defoliation led to the production of a greater length of the primary root axis under low N conditions.

When L. perenne was grown under higher N conditions, as well as reducing overall root axis size, defoliation resulted in a more herringbone structure (i.e. where branching is largely confined to the main axis). This herringbone structure has been suggested to be the best structure for acquisition of readily diffusible resources, such as nitrate ions (Fitter, 1987), which would consequently improve nutrient acquisition following defoliation in this faster-growing species. However, reduction in nitrate uptake in L. perenne resulting from defoliation has been shown to be greater in N replete compared with N limited plants (Macduff et al., 1989).

In both species studied, the length of lateral root per unit length of primary nodal root was less under defoliation. This contrasts with a study of the effects of clipping on grass seedlings (Arredondo & Johnson, 1998), where it was found that defoliation significantly increased the number of secondary order laterals per unit of seminal root length in ‘Whitmar’, a grazing sensitive cultivar of wheatgrass (Agropyron spp.). For ‘Hycrest’, a grazing tolerant cultivar of wheatgrass, seminal root branching characteristics were unaffected by defoliation. It was later found that ‘Whitmar’ exhibited the largest plasticity in root topology, and this plasticity was suggested as the main mechanism allowing it to forage in heterogeneous soil conditions (Arredondo & Johnson, 1999). Neither of the species in our study showed an increase in lateral production or lateral length under defoliation, perhaps suggesting a contrasting response in the seminal and nodal root systems to defoliation. The work by Arredondo & Johnson (1998, 1999) reflected root response in seedlings and, consequently, predominantly that of the seminal root system to defoliation, the result of which may contrast to that of the adventitious or nodal roots of vegetative tillers. The seminal root system may respond in a manner less dependent upon plant nutrient reserves than the nodal root system.

Undefoliated plants of L. perenne had a more randomly branched structure under high N than under low N, agreeing with previous predictions about seminal root systems (Fitter, 1986; Berntson & Woodward, 1992), that roots of a given species or genotype should become more herringbone as the availability of soil resource decreases. However, the opposite response to N availability was found in F. ovina, where the evidence for herringbone structure was least under low N conditions, allowing the plant roots to ramify through the soil and potentially increase the root surface area. This more randomly branched topology can offer a greater potential for exploring the soil, and can contribute to the potential competitiveness of a plant growing in nutrient limiting conditions (Bouma et al., 2001). In previous studies, topological indices have been found to change little with alteration in nutrient supply in a range of species, while external link lengths have been shown to reduce at high N supply (Fitter et al., 1988). These results, carried out on seedling plants, compare well with results from our study, on F. ovina (the species common to both studies), where a shorter length of laterals was found on root axes grown under a high N supply. These common responses to N supply, found in Festuca spp. grown either from seed or from vegetative tillers, is encouraging for the extension of results from seedling plants to nutrient responses to conditions in the field situation. Linkohr et al. (2002) found that, in Arabidopsis, primary root length decreased with increasing nitrate availability, while lateral root elongation was also suppressed by high nitrate supply.

Festuca ovina appears the most plastic in response to defoliation in terms of morphological and topological responses at low N supply whereas L. perenne shows greatest plasticity under high N supply, reflecting the contrasting environments in which these species are most commonly found. In a study of architectural plasticity to high nutrient patches in a low N background, Fitter (1994) found that L. perenne was the least plastic, whereas F. ovina was intermediate in response. Our study shows that morphology and topology of the nodal roots of both L. perenne and F. ovina can alter with defoliation, but that is dependent upon the N supply. In addition to measuring changes in topological indices, root link length responses should be considered in studies examining response to multiple stresses. When considering response to defoliation in particular, it appears that results depend upon whether a seedling or a vegetatively propagated plant is studied. Future work should examine a wider range of species, and consider the response of the root system of the mature plant.


The authors thank the Scottish Executive Environment and Rural Affairs Department for funding this research.