Plasticity in relative growth rate after a reduction in nitrogen availability is related to root morphological and physiological responses

Abstract

Background and Aims

To maximize growth and fitness a plant must adjust its phenotype by an amount and speed that matches changes in nitrogen availability. To determine how plastic ontogenetic changes in root physiological and morphological traits interact and whether or not these responses are likely to maximize growth, ontogenetic changes in relative growth rate (RGR, proportional rate of change of plant dry mass), unit root rate (URR, rate of change of plant dry mass per unit root length or area), specific root length (SRL, root length per dry root mass), specific root area (SRA, root area per dry root mass), and other root traits before and after a decrease in nitrogen supply, were studied in ten herbaceous species.

Methods

Plants of each species were grown in hydroponic culture under controlled conditions in a control treatment where the supply of nitrogen remained constant at 1 mm, and in a stress treatment where the nitrogen supply was abruptly reduced from 1 to 0·01 mm during the growth period.

Key Results and Conclusions

In the treatment series the number of bifurcations per root area and per root length, specific root area (SRA) and length (SRL), areal (URR_area) and length-based (URR_mass) unit root rate and RGR decreased, and root tissue density increased relative to the control. Species having greater plasticity in the percentage decrease in SRA at the end of the experiment also had smaller reductions in RGR; plasticity in SRA is therefore adaptive. In contrast, species which showed a greater reduction in URR_area and in the number of bifurcations per root area and per root length, showed stronger reductions in RGR; plasticity in URR_area and in the number of bifurcations per root area and per root length is therefore not adaptive. The plastic responses observed in SRA, SRL and in root tissue density constitute a set of plastic adjustments that would lead to resource conservation in response nutrient stress.

INTRODUCTION

Because of the sedentary nature of plants, many aspects of plant ecology are strongly influenced by spatial and temporal variation in above- and below-ground resources. Indeed, positive fluctuations (pulses) of below-ground resources affect species interactions (James and Richards, 2007) and population dynamics (Angert et al., 2007) and are expected to influence species diversity within plant communities (Chesson et al., 2004). Soil nitrogen availability is particularly critical because it frequently limits plant growth (Aerts and Chapin, 2000), and fluctuates strongly in many tropical and temperate ecosystems (Lodge et al., 1994; Farley and Fitter, 1999), even at scales as short as 1 d (Cui and Caldwell, 1997).

Due to this temporal variation in soil nitrogen availability, a well-adapted plant must be able to adjust its root system with respect to timing, rate and amount (Levins, 1968). When soil resources are limiting, the ability to alter root systems so as to minimize the loss of performance (maintain viability and growth) is considered to be a key aspect of adaptive plasticity (Bell and Sultan, 1999). However, little is known about the ecological consequences of temporal decreases in nitrogen availability (but see Useche and Shipley, 2010).

Research in trait plasticity can be conceptually divided into comparisons between different environments at a single point in time (‘snapshot approach’) and comparisons of ontogenetic trajectories that the trait follows in each environment (‘ontogenetic approach’) (Pigliucci et al., 1997; Useche and Shipley, 2010). Since the variation in the value that a trait expresses at the end of different treatments or environmental conditions is the result of plasticity in its developmental trajectory between the different environments, an ontogenetic approach is more informative. More importantly, to assess whether a species or genotype is able to adaptively adjust its phenotype in response to environmental heterogeneity, the timing, degree and rate of development of the target trait must be measured and compared between the different environments. Since growth rate has adaptive value (Walters and Reich, 1996; Lambers et al., 1998; Hastwell and Facelli, 2003; Sletvold, 2005; Portsmuth and Niinemets, 2007), and since phenotypic plasticity is only adaptive if it appropriately tracks the trajectory of the environmental fluctuation (Sultan, 2004), the amount, timing and rate of plastic changes was studied in some morphological and physiological root traits that may influence relative growth rate following a decrease in nitrogen supply, and the degree to which plasticity in such root traits could buffer the expected decrease in relative growth rate.

Relative growth rate (RGR, g g⁻¹ d⁻¹), a measure of the growth efficiency of the plant, is the rate of production of new dry mass per unit of existing dry mass. It is well known that RGR can be broken down into three leaf-based properties as follows: RGR = NAR × SLA × LWR (net assimilation rate × specific leaf area × leaf weight ratio). However RGR can also be broken down into root-based properties. The rate of nutrient absorption at a given soil supply rate is generally related to the number and specific activity of transport proteins and the number of these in turn will be related to either the length or surface area of roots (Eissenstat, 1992, Lambers et al., 1998). Therefore RGR is broken down with respect to root mass, length and area as follows:

where M is plant mass, A_root is the root surface area, L_root is root length and M_root is root mass. Root weight ratio (RWR) is the root-based equivalent to LWR. Area-based (URR_area, g d⁻¹ cm⁻²), length-based (URR_length, g d⁻¹ cm⁻¹) and mass-based (URR_mass, g d⁻¹ g⁻¹) unit root rate are the root-based equivalents of NAR and measure the rate of increase in plant biomass per unit of root area, length or mass. The specific root length (SRL, cm g⁻¹; i.e total root length divided by total root dry mass) and specific root area (SRA, cm² g⁻¹; i.e. total surface root area divided by total root dry mass) are the root-based equivalents of SLA. Also measured were five other root traits: root tissue density (total root dry mass per total root volume, g cm⁻³), root area per volume (cm² cm⁻³), root branching density (estimated both as number of root bifurcations per root length and area) and average root diameter (mm). Each of these root traits can potentially present adaptive plastic adjustments that would buffer decreases in RGR following a reduction in nitrogen supply.

Although there are some studies describing the plasticity patterns of some of these root traits in response to soil resources (Berntson et al., 1995; Fransen et al., 1998; Bell and Sultan, 1999; Fransen et al., 1999), evidence for adaptive value in such plastic responses is scarce and some root physiological and morphological traits have been neglected. Useche and Shipley (2010) found that species having greater increases in the maximum rate of change in biomass allocated to roots also had smaller reductions in RGR following a reduction in nitrogen supply. Plasticity in biomass allocation to roots was therefore adaptive. However, instead of increasing the relative allocation of biomass to roots, plants can also enhance uptake of soil resources by changing the geometrical and architectural characteristics of their root systems (Berntson et al., 1995), for example, by increasing the surface available for nutrient uptake per unit of biomass invested in roots. The available evidence shows that species with high potential RGR (RGR_max) are more sensitive to variation in nutrient supply (Shipley, 1988; Meziane and Shipley, 1999; Useche and Shipley, 2010). However, little is known (Fransen et al., 1999) about the relative importance of plasticity in physiological versus morphological root traits between species from nutrient-poor (i.e. generally low RGR_max) and nutrient-rich habitats (i.e. generally high RGR_max) or whether this plasticity is adaptive. There is no consistent pattern in the response of specific root length to variation in nutrient supply (Paponov et al., 1999).

This study focuses on interspecific plastic responses to a reduction in nitrogen supply. The timing, degree and rate of development of RGR and the above-mentioned root traits were estimated during a control treatment where species were exposed to a constant nitrogen supply, and during a stress treatment where the same species were first grown in identical conditions and then exposed to a sudden reduction in the nitrogen supply for the rest of their growth trajectory. Specifically the following questions were addressed: (a) Do plants modify the timing, degree and rate of development in the physiology and morphology of their root systems in response to a reduction in nitrogen availability? (b) Does plasticity in root physiology and morphology minimize the expected decrease in RGR following a reduction in nitrogen availability? (c) Are species with lower RGR_max more able to develop plastic changes in the selected root traits in such a way that they minimize the reduction in RGR, thus minimizing the loss of performance?

METHODS AND MATERIALS

Forty-four plants from each of ten species were grown under each of two experimental conditions. Individuals in a control treatment were exposed to a constant hydroponic solution (see below) for 4 weeks, in which the nitrogen concentration was maintained at 1 mm. In the nutrient stress treatment individuals from the same species were grown in this same solution for the first 12 d, after which the nitrogen concentration was reduced 100 times to 0·01 mm and maintained at this level until the end of the 4-week period. All plants remained in the vegetative phase during the experiment. Data of the present study were collected from the same experiments as reported in Useche and Shipley (2010) based on leaf traits. These ten species are herbaceous perennial species typical of open habitats and representing a wide range of growth rates: Aquilequia canadensis Munz, Carex crinita Lam., Epilobium glandulosum Lehm., Hypericum pyramidatum Aiton, Lycopus americanus Muhl., Mentha arvensis L., Penthorum sedoides L., Scutellaria lateriflora L., Silene niveae (Nutt.) Muhl. ex Otth and Solidago nemoralis Aiton.

Growth conditions and harvests

Growth conditions and harvest schedules are the same as those used in Useche and Shipley (2010). Plants were initially started in hydroponic sand culture and germination was timed in order to provide sufficient individuals per species having a similar size and large enough to allow transplantation into the fully liquid hydroponic system in a growth chamber; irradiance was 500 µmol m⁻² s⁻¹ PAR with a photoperiod of 16/8 (light/dark) h and a temperature of 22/18 °C for both the treatment and control series. The hydroponic system consisted of a 148 cm × 66 cm × 26 cm container placed within the growth chamber and filled with 279 L of nutrient solution. This container was covered with a lid having a grid of 220 equidistant holes (6 cm apart). The plants were placed in these holes and randomly positioned on the cover. The distance between plants increased over time as harvests began and the plants did not overtop one another. When this was likely to occur, the location of plants was changed in order to avoid overtopping. The nutrient solution was in constant movement and was continuously aerated by two water and two air pumps. The solution used during the control consisted of 1/3 mm KNO₃, 1/4 mm Ca(NO₃)₂·4H₂O, 1/12 mm (NH₄)₂SO₄, 5/3 mm K₂SO₄, 5/4 CaCl₂, 2 mm MgSO₄·7H₂O, 1 mm KH₂PO₄, 10 µm MnSO₄·H₂O, 1 µm Na₂MoO₄·4H₂O, 46 µm H₃BO₃, 1 µm ZnSO₄·7H₂O, 1 µm CuSO₄ and 68·1 µm Fe-EDTA, which made a total nitrogen concentration of 1 mm. The same hydroponic solution was used during the first 12 d of the treatment. Midway through day 12 the nitrogen concentration in the treatment series was reduced 100 times and maintained at this level until the end of the experiment. To obtain this nutrient solution with 0·01 mm of total nitrogen, the concentrations of some components were modified as follows: 1/3 × 10⁻² mm KNO₃, 1/4 × 10⁻² mm Ca(NO₃)₂·4H₂O, 1/12 × 10⁻² mm (NH₄)₂SO₄. To maintain the potassium and calcium concentrations as in the control series, the concentrations of K₂SO₄ and CaCl₂ were adjusted to 1·996 mm and 1·497 mm, respectively. The pH was monitored daily and adjusted to 5·85. The solution concentration was monitored daily and was completely renewed every 5 d for both the control and the treatment. Because of these culture conditions, and because plants did not overtop one another due to the increasing distances between them as harvesting proceeded, competition was minimal.

The harvest schedule began on day 8 following transplantation into the growth chamber. To be able to detect sudden plastic responses, for both experimental conditions, the harvest programme was characterized by a more intensive harvest frequency bracketing day 12 (the day when the reduction in nitrogen availability was imposed in the treatment) when generally the number of plants harvested per species per day was as follows: from day 12 to day 15, three; from day 16 to day 20, two; from day 21 until the end of the experiment, one. Plants were randomly chosen for harvest. The actual number of plants per species per day that were harvested sometimes differed slightly from these values in some species due to seedling mortality during the initial establishment phase before transplantation. The maximum number of consecutive days with no harvested plants for a particular species was two.

At each harvest the individual was separated into leaves, roots and support tissues. Water was removed from the surface of each plant part with absorbent paper, and its fresh weight was measured. After digitizing the root system, plant parts were then oven-dried for at least 48 h at 80 °C before obtaining dry weights. Root systems grown in hydroponics differ from soil-grown roots in the mechanical resistance that roots encounter in the soil and the availability of nutrients at the root surface. However, working with different wheat genotypes, Mian et al. (1994) found that root length of soil-grown plants was predicted by root length of plants grown in hydroponics. In any case, the objective of the present study is not to recreate field conditions but rather to compare plasticity in RGR under the experimental conditions and plasticity in these root properties.

Growth analysis and estimation of plasticity parameters

The root system of each plant was digitized. Total root length, total root area, total root volume, the number of root bifurcations and the average root diameter were measured by image analysis with WinRhyzo2001a for Windows (Regent Instruments Inc.). Pixel resolution was approx. 63 µm. These primary measurements were used to derive further root traits: specific root length (SRL, total root length divided by total root dry mass), the specific root area (SRA, total surface root area divided by total root mass), root tissue density (total root mass per total root volume), root area per root volume and root branching density, both as number of root bifurcations per total root length, and as the number of root bifurcations per total root area.

The predicted values through time of RGR, URR and the morphological root traits correspond to a curve describing the ontogenetic trajectory followed by each trait. Because these traits could change continuously in these experiments, the predicted values and 95 % confidence intervals of these measurements were estimated using cubic-spline smoothers. Cubic-spline smoothers do not impose any functional relationships on the data and can detect even subtle changes in complicated nonlinear growth trajectories (Shipley and Hunt, 1996). The predicted values of RGR at each harvest were estimated as the first derivative with respect to the time of the predicted natural logarithm of plant dry mass. Predicted values of URR were obtained from the product of RGR and whole-plant dry mass divided by total root area (URR_area) or by total root length (URR_length).

To test whether plants show plasticity in the ontogenetic trajectory of the growth components after a reduction in nitrogen availability, measures of the strength and rapidity of change that the growth component would follow during ontogeny in each environment were needed. For this purpose, the following parameters were calculated from the curve of each growth component of each species in each treatment series: (a) the maximum rate of change (either positive or negative); (b) the time at which this maximum rate occurs with reference to day 12·7, i.e. the time when the nitrogen stress began during the treatment; (c) the percentage change in the growth component at the time of maximum rate of change compared with the mean value before day 12·7; (d) the percentage change at the end of the experiment compared with the mean value before day 12·7; (e) the predicted value at the time when maximum rate of change occurred; and (f) the predicted value at the end of the experiment (Fig. 1). Cubic-spline smoothers were fitted using the smooth.spline function in R (R Foundation for Statistical Computing, http://www.R-project.org). All other statistical analyses were done using SPSS 13 for Windows computer package. t-Tests were conducted without assuming equality of variances between groups, resulting in Welch (or Satterthwaite) approximations to the degrees of freedom. The degree of plasticity for each ontogenetic parameter of each trait was calculated as the absolute value of the percentage of the difference between the two environments in the value of the ontogenetic parameter. The arcsin of the square root of these percentages was calculated before running correlation analyses to stabilize the variance and to obtain approximate linearity. Bivariate correlations between the plasticity of different traits were tested using the Pearson correlation coefficient. The maximum potential RGR (RGR_max) was estimated as the maximum value from the curve of predicted values of RGR during the control series.

Fig. 1.

Parameters of ontogenetic change (text in rectangles) calculated from the predicted values (hypothetical curved line) of RGR and root traits. Figure from Useche and Shipley (2010).

Ideally, one should randomly allocate the treatment/control conditions independently to each plant. For obvious reasons it was not possible to do this and so the experimental design is ‘pseudo-replicated’ at the level of the growth chamber. Because it was possible to accurately control the environmental conditions of the growth chamber and the nutrient solutions over time, and therefore insure that these conditions did not vary except with respect to the target treatment, it was assumed that any covariation caused by growth chamber effects is minimal and can be ignored.

Describing ontogenetic trajectories with respect to time instead of mass or size was not an arbitrary decision. Since plants that are exposed to different environments can grow at different rates, when plants are compared at a common age (as in the snapshot approach) it is possible that their traits may differ between environments only because plants in the poorer environment are growing more slowly and are at an earlier point along a fixed developmental trajectory. Comparing plants at a common size in a snapshot approach might correct for this, but only if size and ontogenetic stage is tightly correlated. This is probably a reasonable assumption when comparing individuals of the same species but not when comparing across species. By comparing plants at the same age, but quantifying the growth trajectory (thus changes in size) over time, age and size can be separated.

RESULTS

The values of the basic growth components (RGR, URR_length, URR_area, SRL, SRA, RWR) of each species were calculated for each time period in both the control and treatment series (see Supplementary data, available online). In the control series, in which the nitrogen concentration of the nutrient solution remained constant and high, the instantaneous RGR_max varied from 0·14 g g⁻¹d⁻¹ (Hypericum pyramidatum) to 0·32 (Lycopus americanus) with an average interspecific value of 0·26. At the end of the control series (mean = 29·6 d) RGR values decreased from their maxima for the majority of species (9 of 10) being, on average, 14·3 % ± 26·8 (standard deviation) lower than the mean value before day 12·7 (i.e. the time at which the nitrogen concentration of the solution was reduced during the treatment). In the treatment series RGR generally started to decrease within the first 3 d after reducing the nitrogen concentration of the solution at day 12·7. The interspecific RGR at the end of the treatment series (mean = 28·6 d) was, on average, 59·6 % (±26·7) lower than the mean value before day 12·7. Table 1 summarizes the changes in the measured root traits relative to those measured at day 12·7 in both the control and treatment environments.

Table 1.

Comparison between a control environment and a treatment environment regarding the ontogenetic changes in a series of root attributes over a period of growth in a set of ten herbaceous species

Variable	Control series	Nutrient stress series	t	d.f.	P
Part A
RGR at its maximum rate of change (g g−1 d−1)	0·22 (0·06)	0·12 (0·04)	4·56	16·23	3 × 10−4
URRarea at its maximum rate of change (g d−1 cm−2)	8 × 10−4 (3 × 10−4)	3 × 10−4 (1 × 10−4)	3·58	12·31	0·004
URRmass at its maximum rate of change (g d−1 g−1)	1·50 (0·63)	0·67 (0·28)	3·80	12·48	0·002
SRA at its maximum rate of change (cm2 g−1)	1990 (437)	1870 (568)	0·53	16·89	0·604
SRL at its maximum rate of change (cm g−1)	34370 (14655)	33250 (10843)	0·19	16·58	0·848
Root area per root volume at its maximum rate of change (cm2 cm−3)	213 (39)	226 (39)	–0·71	18·00	0·487
Root tissue density at its maximum rate of change (g cm−3)	0·11 (0·03)	0·15 (0·03)	–2·85	17·28	0·011
Forks per root length at their maximum rate of change (forks cm−1)	5·41 (1·97)	5·51 (2·90)	–0·09	15·84	0·932
Forks per root area at their maximum rate of change (forks cm−2)	90·3 (35·1)	97·6 (50·3)	–0·37	16·07	0·712
Average root diameter at its maximum rate of change (mm)	0·20 (0·04)	0·18 (0·03)	1·07	17·15	0·300
Part B
RGR at the end of the experiment (g g−1 d−1)	0·19 (0·05)	0·07 (0·04)	6·45	16·14	8 × 10−6
URRarea at the end of the experiment (g d−1 cm−2)	8 × 10−4 (4 × 10−4)	3 × 10−4 (8 × 10−5)	3·80	9·71	0·004
URRmass at the end of the experiment (g d−1 g−1)	1·38 (0·74)	0·40 (0·16)	4·09	9·86	0·002
SRA at the end of the experiment (cm2 g−1)	1821 (453)	1502 (368)	1·73	17·28	0·102
SRL at the end of the experiment (cm g−1)	32113 (16757)	27616 (9716)	0·73	14·44	0·475
Root area per root volume at the end of the experiment (cm2 cm−3)	213 (56)	230 (43)	–0·77	16·83	0·454
Root tissue density at the end of the experiment (g cm−3)	0·11 (0·02)	0·16 (0·04)	–3·12	14·57	0·007
Forks per root length at the end of the experiment (forks cm−1)	5·92 (2·54)	5·38 (3·84)	0·37	15·60	0·716
Forks per root area at the end of the experiment (forks cm−2)	99·7 (41·7)	97·3 (62·7)	0·10	15·66	0·921
Average root diameter at the end of the experiment (mm)	0·20 (0·04)	0·18 (0·03)	1·19	15·52	0·253
Part C
Time of maximum rate of change in RGR (d)	18·1 (4·7)	19·1 (2·7)	–0·56	14·28	0·581
Time of maximum rate of change in URRarea (d)	23·0 (7·8)	18·2 (5·3)	1·63	15·92	0·122
Time of maximum rate of change in URRmass (d)	21·7 (7·6)	17·6 (4·9)	1·43	15·44	0·173
Time of maximum rate of change in SRA (d)	20·4 (7·1)	18·9 (5·6)	0·53	17·14	0·600
Time of maximum rate of change in SRL (d)	21·0 (6·5)	18·1 (6·2)	1·01	17·94	0·328
Time of maximum rate of change in root area per root volume (d)	17·0 (6·0)	21·2 (8·0)	–1·35	16·73	0·195
Time of maximum rate of change in root tissue density (d)	17·8 (5·5)	25·5 (4·4)	–3·46	17·17	0·003
Time of maximum rate of change in forks per root length (d)	24·8 (6·0)	17·1 (3·8)	3·43	15·22	0·004
Time of maximum rate of change in forks per root area (d)	22·0 (5·9)	20·4 (3·4)	0·77	14·41	0·454
Time of maximum rate of change in average root diameter (d)	19·7 (8·1)	21·5 (7·7)	–0·51	17·96	0·619
Part D
Maximum rate of change in RGR	–0·005 (0·011)	–0·015 (0·007)	2·47	15·62	0·026
Maximum rate of change in URRarea	–3 × 10−6 (3 × 10−5)	–1 × 10−5 (2 × 10−5)	0·83	15·17	0·418
Maximum rate of change in URRmass	–0·007 (0·058)	–0·037 (0·034)	1·38	14·50	0·189
Maximum rate of change in SRA	–38·3 (74·5)	–62·2 (31·5)	0·94	12·11	0·367
Maximum rate of change in SRL	–436 (1045)	–1461 (926)	2·32	17·74	0·032
Maximum rate of change in root area per root volume	–0·22 (5·68)	0·66 (5·45)	–0·35	17·97	0·728
Maximum rate of change in root tissue density	–2 × 10−4 (0·008)	5 × 10−3 (0·001)	–1·95	9·64	0·080
Maximum rate of change in forks per root length	0·55 (1·27)	0·02 (0·25)	1·29	9·68	0·226
Maximum rate of change in forks per root area	2·02 (1·72)	0·72 (2·67)	1·29	15·38	0·217
Maximum rate of change in average root diameter	0·007 (0·004)	–1 × 10−4 (0·005)	0·72	17·91	0·480
Part E
Percentage decrease in RGR at its maximum rate of change	5·4 (13·5)	30·4 (15·7)	–3·81	17·60	0·001
Percentage decrease in URRarea at its maximum rate of change	–0·3 (15·9)	5·2 (29·7)	–0·52	13·78	0·611
Percentage decrease in URRmass at its maximum rate of change	1·6 (21·4)	14·0 (19·8)	–1·34	17·89	0·196
Percentage decrease in SRA at its maximum rate of change	1·2 (25·1)	16·5 (8·6)	–1·83	11·11	0·094
Percentage decrease in SRL at its maximum rate of change	7·7 (16·1)	17·6 (10·7)	–1·62	15·65	0·124
Percentage decrease in root area per root volume at its maximum rate of change	1·4 (12·5)	–2·0 (10·9)	0·65	17·70	0·527
Percentage decrease in root tissue density at its maximum rate of change	3·7 (23·4)	–40·9 (20·5)	4·53	17·69	3 × 10−4
Percentage decrease in forks per root length at its maximum rate of change	–17·9 (21·3)	–1·9 (13·7)	–2·00	15·37	0·064
Percentage decrease in forks per root area at its maximum rate of change	–12·1 (13·6)	–0·7 (15·9)	–1·76	17·75	0·096
Percentage decrease in average root diameter at its maximum rate of change	–6·9 (18·5)	0·6 (10·3)	–1·13	14·12	0·276
Part F
Percentage decrease in RGR at the end of the experiment	14·2 (26·8)	59·6 (26·7)	–3·78	18·00	0·001
Percentage decrease in URRarea at the end of the experiment	–0·9 (33·5)	20·3 (48·2)	–1·15	16·05	0·269
Percentage decrease in URRmass at the end of the experiment	11·2 (33·6)	40·2 (46·6)	–1·60	16·37	0·129
Percentage decrease in SRA at the end of the experiment	8·1 (31·7)	32·1 (9·7)	–2·29	10·67	0·043
Percentage decrease in SRL at the end of the experiment	14·4 (25·1)	31·3 (14·1)	–1·85	14·18	0·085
Percentage decrease in root area per root volume at the end of the experiment	1·3 (21·5)	–3·9 (11·6)	0·68	13·82	0·509
Percentage decrease in root tissue density at the end of the experiment	–6·9 (38·9)	–50·5 (16·4)	3·27	12·12	0·007
Percentage decrease in forks per root length at the end of the experiment	–27·3 (29·1)	3·7 (28·7)	–2·40	18·00	0·027
Percentage decrease in forks per root area at the end of the experiment	–23·0 (22·5)	1·7 (28·3)	–2·16	17·15	0·045
Percentage decrease in average root diameter at the end of the experiment	–6·0 (23·4)	3·1 (10·0)	–1·13	12·19	0·281

The control environment consisted in a constant and high nitrogen supply rate (1 mm nitrogen), and the treatment environment consisted in an initial high nitrogen supply rate that was abruptly decreased 100 times (from 1 to 0·01 mm nitrogen) after 12 d of growth.

Significantly different values are shown by bold type for the variable.

Reported values are means and standard deviations in parentheses.

Ontogenetic plasticity in RGR and in root morphology and physiology

Plasticity in the ontogenetic trajectory of a trait corresponds to significant differences between the different environments in any of the measures of change calculated from the ontogenetic curve followed by that trait in each environment (Fig. 1). In comparison with the control series, in the treatment series the predicted value at the time of maximum rate of change was significantly lower for RGR, for URR_area, and for URR_mass but significantly higher for root tissue density. This predicted value at the time of maximum rate of change did not differ between the treatment and the control series for the rest of the variables (Table 1, part A). Similarly, in comparison with the control series, in the treatment series the predicted value at the end of the experiment was significantly lower for RGR, for URR_area and for URR_mass but significantly higher for root tissue density. This predicted value at the end of the experiment did not differ between the treatment and the control series for the rest of the variables (Table 1, part B).

After day 12·7 plant species in the treatment series took longer to reach the time at which root tissue density was increasing at its maximum rate of change in comparison with the control, but less time to reach the point at which the number of bifurcations per root length was changing at a maximum rate. The time to attain this maximum rate of change did not differ significantly between the control and treatment series for the rest of the variables (Table 1, part C). The maximum rates of change of RGR and SRL after day 12·7 were significantly more negative in the treatment than in the control series. Maximum rate of change did not differ significantly between the control and treatment series for the rest of the variables (Table 1, part D).

The percentage decrease in RGR at the time of its maximum rate of change was greater in the treatment than in the control series, while the percentage increase in root tissue density at the time of its maximum rate of change was greater in the treatment that in the control series. In contrast, the percentage change at the time of maximum rate of change did not differ significantly between the control and treatment series for the rest of the variables (Table 1, part E).

In comparison with the control series the percentage decrease at the end of the experiment was greater in the treatment series for RGR, SRA, the number of bifurcations per root length and the number of bifurcations per root area. In contrast, the percentage increase in root tissue density at the end of the experiment was greater in the treatment series. The percentage change at the end of the experiment did not differ significantly between the control and treatment series for the rest of the variables (Table 1, part F).

Plasticity in RGR vs. plasticity in root morphological and physiological traits

Plasticity in the percentage decrease in RGR at the time of its maximum rate of change was negatively related with plasticity in the percentage decrease in SRA at the end of the experiment, and was positively related both with plasticity in the predicted value in URR_area at the time of its maximum rate of change and with plasticity in the percentage decrease in the number of bifurcations per root area at the end of the experiment (Fig. 2). Similarly, plasticity in the percentage decrease in RGR at the end of the experiment was positively related with plasticity in the predicted value in URR_area at the time of its maximum rate of change (Fig. 3A). Plasticity in the predicted value in RGR at the end of the experiment was positively related with plasticity in the predicted value in URR_area at the time of its maximum rate of change (Fig. 3B). Plasticity in the maximum rate of change in RGR was positively related with plasticity in the time at which the number of bifurcations per root length reached their maximum rate of change (Fig. 3C).

Fig. 2.

(A) Relationship between plasticity in the percentage decrease in RGR at the time of its maximum rate of change and plasticity in the percentage decrease in SRA at the end of the experiment (r² = 0·50, P = 0·023), y = –0·335x + 0·221. (B) Relationship between plasticity in the percentage decrease in RGR at the time of its maximum rate of change and plasticity in the percentage decrease in the number of bifurcations per root area at the end of the experiment (r² = 0·64, P = 0·005), y = 0·217x + 0·064. (C) Relationship between plasticity in the percentage decrease in RGR at the time of its maximum rate of change and plasticity in the predicted value in URR_area at the time of its maximum rate of change (r² = 0·57, P = 0·012), y = 0·363x – 0·127. Plasticity was estimated as the absolute value of the percentage of the difference between the control and the treatment in the value of the ontogenetic parameter. Percentages were transformed with the arcsin of their square root.

Fig. 3.

(A) Relationship between plasticity in the percentage decrease in RGR at the end of the experiment and plasticity in the predicted value in URR_area at the time of its maximum rate of change (r² = 0·43, P = 0·038), y = 0·424x – 0·176. (B) Relationship between plasticity in the predicted value in RGR at the end of the experiment and plasticity in the predicted value in URR_area at the time of its maximum rate of change (r² = 0·74, P = 0·001), y = 0·267x + 0·036. (C) Relationship between plasticity in the maximum rate of change in RGR and plasticity in the time at which the number of bifurcations per root length reached their maximum rate of change (r² = 0·42, P = 0·041), y = 0·680x – 0·068. Plasticity was estimated as the absolute value of the percentage of the difference between the control and the treatment in the value of the ontogenetic parameter. Percentages were transformed with the arcsin of their square root.

RGR_max vs. plasticity

RGR_max was not related to plasticity in RGR or to any of the measured root traits. However, when plasticity was estimated as the absolute value of the difference between treatment and control (instead of as the absolute value of the percentage of the difference), RGR_max was positively related with the predicted value in URR_mass at the end of the experiment (Fig. 4).

Fig. 4.

Relationship between RGR_max and plasticity in the predicted value in URR_mass at the end of the experiment (r² = 0·48, P = 0·027), y = 7·874x – 1·053. Plasticity was estimated as the absolute value of the difference between treatment and control in the value of the ontogenetic parameter.

DISCUSSION

Plasticity in the physiology and morphology of root systems after a reduction in nitrogen availability

Since the uptake of soil resources occurs across the root surface it is assumed that plants with high SRA invest their root biomass more efficiently than species with low SRA (Eissenstat, 1992). One might therefore expect that plants would compensate for a reduction in nitrogen supply with an increase in SRA and SRL, a pattern that was observed by Paterson and Sim (2000) in the case of SRL. In the present study the opposite was observed: SRA and SRL actually decreased while root tissue density increased. The pattern shown by each of these traits has been found to be related to increased root longevity (Ryser, 1996; Eissenstat et al., 2000; Craine et al., 2002; but see Espeleta and Donovan, 2002). These patterns of plasticity after the reduction in nitrogen availability are comparable to the suite of traits shown by species typical of chronically unproductive environments (Grime, 1977; Chapin, 1980), i.e. a resource conservation strategy. Similarly, a low SRL is related to low root respiration rates and reduced carbon loss (Reich et al., 1998). The consequence of the plastic changes in these root traits following the reduction in nitrogen availability is therefore more likely to be related to the conservation of nitrogen already captured rather than to increasing the uptake rate of additional nitrogen.

Does plasticity in root physiology and morphology buffer decreases in RGR following a reduction in nitrogen availability?

Species that showed higher plasticity in the ontogenetic trajectory in root branching density and species that presented greater plasticity in URR_area also showed larger reductions in RGR after the reduction in nitrogen supply. This plasticity in root branching density and in URR_area therefore cannot be interpreted as being adaptive. This mirrors the response of the equivalent leaf trait (unit leaf or net assimilation rate) reported in Useche and Shipley (2010). The decrease in URR_area, which presumably reflects the decreased ability of the roots to absorb nitrogen, is likely to be due to the reduced supply rate of nitrogen to the root surface and is a consequence of the environment rather than an adaptive response to it. It is not obvious why the greater plasticity in root branching was associated with greater reductions in RGR following the nitrogen stress but may simply be an allometric consequence of changing root size or relative size.

In contrast, after the reduction in nitrogen supply, plasticity in RGR was negatively related with plasticity in SRA; in other words, increased plasticity in SRA was adaptive because it buffered the decrease in growth. A possible mechanism by which this buffering effect occurs might be because a reduction in SRL reduces root respiration rate (Reich et al., 1998). If so then, although a reduction in SRA and SRL may reduce the amount of nutrient taken up per unit root mass (Eissenstat, 1992), this would be overcompensated by minimization of carbon lost in respiration. Unlike the similar plastic responses of unit root and leaf rates, described above, the plastic responses of SRA and specific leaf area (SLA, the equivalent leaf trait) were not the same. In the companion paper to this one, Useche and Shipley (2010) did not observe any clear change in SLA following the reduction in nitrogen supply. However, it is perhaps more biologically relevant to compare plastic changes in SRA following nitrogen reduction with the plastic changes in SLA following a reduction in photon flux density rather than in nutrient supply. Certainly reducing the photon flux density, the areal equivalent of nutrient supply rate, causes specific leaf area to increase when comparing within species (Meziane and Shipley, 1999; Shipley, 2000 ), although the experiment providing the most appropriate interspecific comparison has not yet been done.

Useche and Shipley (2010) found that plants adaptively respond to reductions in nitrogen availability by adjusting the ontogenetic trajectory of biomass allocation to roots, which contributes to buffer the reduction in RGR. Here it was found that plants also show plastic adjustments in the developmental trajectory of some root traits which would constitute a resource conservation response under nutrient stress, and which could form part of the competitive response of plants competing for nitrogen. In the short term, minimization of carbon losses could occur by the reduction in SRL and SRA, which should be associated with a reduction in root respiration rates, and in the long term, the observed increase in root tissue density and the reduction in SRL and SRA can be expected to increase root longevity.

Understanding the functional significance of phenotypic plasticity in root systems requires an interplay between mechanistic investigations of physiology and more synthetic investigations comparing the consequences of such plasticity in terms of growth or other fitness components. In this study we have concentrated on the latter approach so that sufficient numbers of plant species could be screened to provide some generality; it is not yet possible to provide convincing mechanistic explanations for the observed patterns. Clearly, however, more detailed physiological studies are required to provide a more complete understanding of why some plastic responses are adaptive and others are not.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org and give the values of the basic growth components (RGR, URR_length, URR_area, SRL, SRA and RWR) of each species for each time period in both the control and treatment series.