Local root growth and death are mediated by contrasts in nutrient availability and root quantity between soil patches

Abstract

Plants are thought to be able to regulate local root growth according to its overall nutrient status as well as nutrient contents in a local substrate patch. Therefore, root plastic responses to environmental changes are probably co-determined by local responses of root modules and systematic control of the whole plant. Recent studies showed that the contrast in nutrient availability between different patches could significantly influence the growth and death of local roots. In this study, we further explored, beside nutrient contrast, whether root growth and death in a local patch are also affected by relative root quantity in the patch. We conducted a split-root experiment with different splitting ratios of roots of Canada goldenrod (Solidago canadensis) individuals, as well as high- (5× Hoagland solution versus water) or low- (1× Hoagland solution versus water) contrast nutrient conditions for the split roots. The results showed that root growth decreased in nutrient-rich patches but increased in nutrient-poor patches when more roots co-occurred in the same patches, irrespective of nutrient contrast condition. Root mortality depended on contrasts in both root quantity and nutrients: in the high-nutrient-contrast condition, it increased in nutrient-rich patches but decreased in nutrient-poor patches with increasing root proportion; while in the low-nutrient-contrast condition, it showed the opposite trend. These results demonstrated that root growth and death dynamics were affected by the contrast in both nutrient availability and root quantity between patches. Our study provided ecological evidence that local root growth and death are mediated by both the responses of root modules to a nutrient patch and the whole-plant nutrient status, suggesting that future work investigating root production and turnover should take into account the degree of heterogeneity in nutrient and root distribution.

1. Introduction

It has been proposed that a plant comprises different modules, which are repeating and semiautonomous units with the same functions, such as leaves and twigs [1,2]. Plant roots are also thought to be composed of modular units, though there are disputes on the extent of autonomy and the scale of a modular unit [3–6]. It is suggested that the growth and development of plants and their plastic responses to environmental changes are based on responses of modules, and the whole-plant response to an environment is an integrative product of modular responses [3]. Therefore, root modular responses have been highlighted in plant plastic responses to nutrient heterogeneity. For example, root modules in a nutrient-rich patch grow much faster than those in a nutrient-poor patch [7–10].

It has also been found that plant plastic responses are regulated at the whole-plant level by systematic signals such as auxin and cytokinin, and local root responses to nutrient heterogeneity are determined by both the local nutrient patch and the whole-plant nutrient status [11–13]. It was proposed that plant foraging behaviour in heterogeneous environments depends on (i) the plastic responses of individual modules of a plant to local environmental signals and (ii) the systematic control of the whole-plant integrating signals received by other modules of the plant [14]. This interplay of local response and systematic control implies that root responses to a nutrient patch will depend on the contrast in nutrient concentration between the patch and the background soil. It has been shown that root proliferation was greater in patches with higher contrast to their background soil [15–17].

In natural conditions, patches are not necessarily nutrient-rich (positive contrast patch), and it is possible that some root modules are in a nutrient-poor patch while the majority of roots are in a relatively nutrient-rich condition (negative contrast patch). In a split-root experiment with patches of the same nutrient contents but in different contrasts, it was shown that positive contrast facilitated root growth, while negative contrast facilitated root death, and that the facilitation of growth or death was greater when the contrast was higher [18], confirming the co-determination of the growth and death of a root module by local environment and whole-plant control. However, in that study, only one individual root was in each nutrient patch, which was an extreme condition of root quantity contrast between the patch and background soil. If more root modules are in a positive contrast patch, the facilitation effect on growth for each module might decrease. Likewise, if more root modules are in a negative contrast patch, the facilitation effect on death for each module might also decrease.

In this study, we aimed to investigate whether root growth and death in a nutrient patch are affected by relative root quantity in the patch as well as the nutrient contrast between the patch and its background soil. Specifically, we hypothesized that (i) in a nutrient-rich (positive contrast) patch, root growth increases while root death decreases when the proportion of roots in the patch decreases; (ii) in a nutrient-poor (negative contrast) patch, root growth increases, while root death decreases, with increasing proportion of roots in the patch. To test these hypotheses, we performed a greenhouse split-root experiment using the species Solidago canadensis L., which shows high morphological and physiological root foraging plasticity in heterogeneous nutrient conditions [19,20].

2. Material and methods

(a). Experimental set-up

The experiment was conducted from May to June 2014 in a greenhouse in Beijing Normal University, Beijing, China. Seeds of S. canadensis (Ernst Conservation Seed, 9006 Mercer Pike, Meadville, Pennsylvania, USA) were soaked in 1% H₂O₂ solution for 30 min for surface sterilization, washed with distilled water three times and then sowed in germination pans containing moist river sand which had been sterilized at 165°C. The sand was watered every day to keep it moist. When the seeds germinated and seedlings established (five to six true leaves), 200 seedlings with similar size were transplanted into 40 plastic pots (28 × 24 × 15 cm³) with five seedlings in each pot. Five days later, 50 ml 1‰ commercial fertilizer solution (3.94% NH₄-N, 6.05% NO₃-N, 10.01% urea N, 20% P₂O₅ and 20% K₂O; Peters Professional, the Scotts Co., Marysville, OH, USA) was applied to each pot every week to facilitate the growth of the seedlings.

When the rosette leaves were 4 cm long, 60 plants with similar size and abundant roots were chosen. After shaking the attached sand off and weighing the fresh weight of each one, we transplanted the plants into 60 cuboid containers (15.5 × 9.5 × 14.5 cm; figure 1), with one plant in each container. The container consisted of five pieces of Plexiglas and was separated diagonally by another piece of Plexiglas into two compartments. Nylon fabric (200 mesh) was attached on the inside surface of the two 15.5 × 14.5 cm sidewalls for the observation of root growth. On the bottom of each container, 12 holes were drilled to ensure drainage of water. The containers were filled with acid-washed 40-mesh white quartz sand and watered thoroughly before seedlings were planted. When being transplanted, roots were approximately split according to the ratio of 1 : 15, 1 : 3, 1 : 1, 3 : 1 or 15 : 1 such that the proportion of the number of basal roots in the nutrient-rich patch reached 6.25%, 25%, 50%, 75% or 93.75% of the total number (and roots in the nutrient-poor patch were 93.75%, 75%, 50%, 25% or 6.25%, respectively). Twelve plants were assigned at random to each roots quantity contrast. From each compartment, a healthy first-order (most distal) root was carefully separated and placed between the sidewall Plexiglas and nylon fabric. Because of the small mesh size, the quartz sand and roots cannot penetrate the nylon fabric, and the growth and death dynamic of the root can be observed. The container was wrapped with opaque black paper to prevent light exposure.

Figure 1.

Diagram of an experimental unit container. An experimental plant was placed at the centre of a transparent Plexiglas container, which was filled with quartz sand and separated diagonally with a Plexiglas divider for two nutrient treatments and split roots. Roots of the plant were buried into the two separated compartments. In each compartment, the inside of a sidewall was attached with a piece of nylon fabric, and a segment of root from the plant was placed between the sidewall and nylon fabric. (Online version in colour.)

One week later, when the plants established in the container, we started nutrient treatments. We had two types of nutrient contrast: low contrast and high contrast. In the low-nutrient contrast-condition, one compartment was supplied with Hoagland solution (nutrient-rich patch, L_r), and the other compartment was supplied with distilled water (nutrient-poor patch, L_p). In the high-nutrient-contrast condition, one compartment was supplied with five times strength Hoagland solution (nutrient-rich patch, H_r), and the other compartment was supplied with distilled water (nutrient-poor patch, H_p). In total, we had 60 experimental units (five root quantity contrast × two nutrient contrast × six replicates).

Every 8 days, the containers were rearranged and supplied with nutrient solution or distilled water, as differences in nutrient contents between the nutrient-rich and -poor patches diminished after 8 days (electronic supplementary material, figure S1). Before the addition of nutrient solution or water, the compartment was washed thoroughly with adequate distilled water to prevent nutrient accumulation. The containers were watered every day to keep the sand moist. During the experiment, the greenhouse was kept at 20–28°C and 25–35% humidity.

(b). Harvest and measurements

All the plants were harvested 25 days after the first nutrient treatment (i.e. after four rounds of nutrient treatment), when roots were observed to have reached the bottom of some containers. Roots between the nylon fabric and the sidewall Plexiglas were carefully taken out with tweezers. Then, the roots were scanned using the Epson root scanner (G780B, Seiko Epson Corp., Tokyo, Japan) at 400 DPI and scanned pictures were analysed with WinRhizo Tron 2009a (Regent Instruments, Quebec, Canada) for the number, length and branch interval (inter-branch length) of first-order lateral roots.

Each plant was separated into above- and below-ground parts. Roots in each compartment, including the roots for scanning, were separated into live fine roots, dead fine roots and coarse roots. Fine roots were defined as absorptive roots including first-order, second-order and non-lignified third-order roots [21], and a root was considered dead if it was shrunken, hollow, black or the cortex was easily separated from the vascular tissue [18]. All the samples were dried to constant weights at 65°C and weighed. Root mortality was calculated as the ratio of dead fine roots relative to live fine roots.

(c). Statistical analyses

We used linear mixed-effects models to test the relationships between the proportion of roots in a patch and root growth/death in different nutrient contrast conditions, because each container had two compartments (i.e. two treatments), and roots in the two compartments were not independent of each other. In each model, the first-order root number, average branch interval of first-order roots, average length of first-order roots or root mortality of each compartment were included as the dependent variable, with nutrient patch (H_r, H_p, L_r and L_p) and root proportion (the proportion of roots put in a compartment at the beginning) as fixed effects, and container as a random effect. We used a simple linear regression model to test the relationships between root proportion and above- or below-ground biomass in different nutrient contrast conditions. Tukey's HSD tests were used for post hoc pairwise tests when a main effect was significant.

To obtain normal distributions and homogeneity of variances, the dependent variables were ln or cubic root transformed when necessary. Analyses were performed with R (v. 3.4.2; R Foundation for Statistical Computing, Vienna, Austria) in RStudio (v. 1.1.383). Linear mixed effects models were run using package lme4 v. 1.1-12 [22]; p-values were calculated by using package lmerTest v. 2.0-33 [23]; Tukey's HSD tests were performed by using package predictmeans v. 0.99 [24].

3. Results

Root growth of the roots for observation and scanning was facilitated in the nutrient-rich patches. In both the high- and low-nutrient-contrast conditions, the number and average length of first-order lateral roots were higher, and branch intervals were lower in the nutrient-rich patches than in the nutrient-poor patches (figure 2a–c and table 1). Within the nutrient-rich or -poor patches, there were no significant differences between the high- and low-contrast condition for all the three variables (p = 0.239 and 0.352 for root number in the nutrient-rich and nutrient-poor patches, respectively; p = 0.993 and 0.548 for branch interval; p = 0.949 and 0.756 for root length). There were significant interaction effects of root proportion and nutrient patch on the number of first-order lateral roots that had been produced and branch intervals (table 1). The number of first-order lateral roots decreased with root proportion in the high-contrast nutrient-rich (H_r) patches (p = 0.004), while it increased with root proportion in the nutrient-poor patches with either high (H_p; p = 0.016) or low contrast (L_p; p = 0.016; figure 2a). Branch intervals increased with root proportion in nutrient-rich patches (p < 0.001 and p = 0.023 in high and low contrast, respectively), while it showed decreasing trends with increasing root proportion in the nutrient-poor patches (figure 2b). There were no significant relationships between the average length of first-order roots and root proportion (figure 2c and table 1).

Figure 2.

The number (a), average interval (b) and length of first-order lateral roots (c) and root mortality (d) as a function of initial root proportion in respective patch of each nutrient patch (i.e. proportion in nutrient-rich patches for H_r and L_r, and proportion in nutrient-poor patches for H_p and L_p). Red squares denote plants in high-contrast nutrient-rich patches (H_r); blue circles denote plants in high-contrast nutrient-poor patches (H_p); light red rhombuses denote plants low-contrast nutrient-rich patches (L_r); light blue inverted triangles denote plants in low-contrast nutrient-poor patches (L_p). Data in (a–c) were from the scanning of the roots between the nylon fabric and the sidewall plexiglas; data in (d) were from all the roots in each compartment. Significant relationships (p < 0.05) with root proportion are indicated with solid lines. Error bars represent ± s.e. (n = 6).

Table 1.

Summary of analyses of root proportion effects on the number of first-order roots, branch interval, length of first-order roots and root mortality and differences between nutrient patches, using linear mixed-effects models (n = 6 replicates).

		number of first-order rootsa		branch intervala		length of first-order roota		root mortalityb
source	d.f.	MS	F-value	MS	F-value	MS	F-value	MS	F-value
nutrient patch	3	24.99	70.5***	1.68	15.6***	0.7	3.2*	0.40	31.9***
root proportion	1	0.10	0.3	0.14	1.4	0.9	4.5*	0.03	2.3
nutrient patch × root proportion	3	2.83	8.0**	0.61	5.7**	0.1	0.4	0.06	5.0**

^aData were ln transformed.

^bData were cubic root transformed.

*p < 0.05; **p < 0.01; ***p < 0.001.

Root mortality of the whole root system was lower in the nutrient-rich patches than in the nutrient-poor patches, and it was highest in the high-contrast nutrient-poor patches (figure 2d and table 1). Root mortality increased with root proportion when in the high-contrast nutrient-rich patches (p = 0.001) and low-contrast nutrient-poor patches (p = 0.018), while it decreased or showed a decreasing trend with increasing root proportion in the high-contrast nutrient-poor patches (p = 0.013) and the low-contrast nutrient-rich patches (p = 0.382; figure 2d).

Plant above-ground biomass was higher in the high-contrast units than in low-contrast units (figure 3a and table 2). However, below-ground biomass was lower in the high-contrast units than in the low-contrast units (figure 3b and table 2). Root splitting also had effects on plant above- and below-ground biomass (table 2), both of which increased as proportionally more roots were in the nutrient-rich patches, except below-ground biomass in the low-contrast condition (figure 3).

Figure 3.

The final plant (a) above- and (b) below-ground biomass in high- or low-contrast nutrient condition with different proportional root splitting. Black squares and line denote biomass in high-nutrient contrast. Grey circles and lines denote biomass in low-nutrient contrast. Lines denote significant relationships with root proportion. Error bars represent ±s.e.

Table 2.

Summary of the analyses of the effects of nutrient contrast and root splitting on plant above-ground and below-ground biomass, using simple linear regression models (n = 6 replicates).

		above-ground biomass		below-ground biomass
source	d.f.	MS	F-value	MS	F-value
nutrient contrast	1	0.20	12.8***	0.07	6.38*
root proportion	1	0.27	12.5***	0.05	4.50*
nutrient contrast × root proportion	1	0.02	1.19	0.03	2.37

*p < 0.05; **p < 0.01; ***p < 0.001.

4. Discussion

Our results showed that in nutrient-rich patches, the growth of root modules of S. canadensis decreased when more roots were present. On the contrary, in nutrient-poor patches, the growth of root modules increased as more roots were in the same patches. When nutrient contrast was high, root mortality in a nutrient-rich patch increased with increasing roots in the patch, while root mortality in a nutrient-poor patch decreased as more roots were in the same patch. These results support our hypothesis that root growth and death dynamics in a patch were affected not only by the contrast in nutrients between patches but also by the contrast in root quantity between patches (figure 4).

Figure 4.

Root growth and death in a local nutrient patch are co-determined by contrasts in nutrient availability and root quantity between a soil patch and the background soil. Root growth reduces in a nutrient-rich patch but increases in a nutrient-poor patch when more roots are in the patch, with either high or low nutrient contrast between the patch and background soil. Root mortality depends on contrasts in both root quantity and nutrients: it increases when more roots coexist in a high-contrast nutrient-rich patch or low-contrast nutrient-poor patches, and vice versa. (Online version in colour.)

(a). Root growth

In our study, the number and branch interval of first-order lateral roots, both of which represented root growth, were responsive to the contrast in root quantity. The number of first-order lateral roots developed from single root modules in nutrient-rich patches decreased with increasing initial proportion of roots sharing the patches, while it increased in nutrient-poor patches as more roots coexisted in same patches (figure 2a). This phenomenon may reflect two mechanisms in the systematic control of root growth. First, when there are only a few root modules of a plant in a nutrient-rich (relative to its background) patch, the plant proliferates roots in the patch rapidly until the need for nutrients of the whole plant is satisfied or the patch is fully exploited; when there are plenty of root modules in a nutrient-rich patch, nutrient uptake is adequate for plant growth and the plant does not need to stimulate rapid root growth. Second, when there are only a few root modules of a plant in a nutrient-poor (relative to its background) patch, as relatively more roots are in a richer background of nutrients, the requirement for nutrients of the plant can be satisfied even without the root modules in the nutrient-poor patch, and there is no need to proliferate root in the patch for the plant; however, when there are many root modules in a nutrient-poor patch and only a few in a nutrient-rich patch, the plant is limited by nutrients, and thus it proliferates roots in the nutrient-poor patch as well to take up as many nutrients as possible.

The response of branch interval to initial proportion of roots in a patch was opposite to that of root number (figure 2b). It has been reported that in nutrient-rich conditions, the number of lateral roots increases and branch intervals decrease [25–28], suggesting that more lateral roots and shorter branch intervals are two features of higher root proliferation. Thus, the opposite responses of branch interval to root proportion further validate our results about root number. As a result of faster growth when more nutrients were added, the proportion of roots in each compartment relative to total roots of the plant at the end of the experiment deviated from its initial proportion (electronic supplementary material, figure S2). Root proportion in the nutrient-rich patches increased, while that in the nutrient-poor patches decreased.

Inconsistent with our expectation, the relationship between root growth and root quantity was not significantly altered by nutrient contrast, particularly in the nutrient-poor patches (figure 2a,b). The reason could be that proliferating roots and occupying soil volume fast are very important for plant competition, irrespective of nutrient patches [29]. This may be particularly true at the early stage of plant establishment when the root system is small; therefore, root growth was less influenced by nutrient contrast at this stage. However, at the later stage, when plants had grown much larger and some roots started their senescence, the difference in nutrient contrast started to take effect, making root death respond significantly differently between high- and low-contrast patches (see the next section).

(b). Root death

We found that, in general, root mortality was higher in nutrient-poor patches than in nutrient-rich patches (figure 2d), which agrees with findings that, in nutrient-rich patches, root lifespan was longer [30,31]. In our study, root death was influenced by the contrast in both nutrient and root quantity. In the high-contrast nutrient condition, when the initial proportion of roots in a high-contrast nutrient-rich patch was very low (i.e. 6.25%), no roots were found dead; however, when only 6.25% roots were in the opposite nutrient-poor patches at the beginning, nearly 100% of them dead in the end (figure 2d). Conforming to our hypothesis, root death increased with increasing proportion of roots in the high-contrast nutrient-rich patches, and the increase was significantly greater than that in the low-contrast nutrient-rich patches, where root death did not show a significant trend. Root death decreased with increasing proportion of roots in the high-contrast nutrient-poor patches, which also confirmed our hypothesis that when the background nutrients are much richer, roots in a nutrient-poor patch may not be worth maintaining based on cost–benefit balance [32,33]. But when roots in the nutrient-poor patch are in high proportions, their death can have large impacts on the plant, and thus root mortality is low. This can also explain that root mortality in control conditions was lower than in localized nutrient addition conditions in some studies [31,34].

In the low-contrast nutrient-poor patch, contrary to our hypothesis, root death increased with increasing root proportion (figure 2d). The reason might be that in the low-contrast condition, total nutrients added to a plant were lower than that in the high-contrast condition. Therefore, plant growth in the low-contrast condition was more limited by nutrients and could not afford to maintain many roots in the nutrient-poor patches, particularly when most of the roots were in the nutrient-poor patches. It is suggested that root lifespan is dependent on the extent to which nutrients are limiting (i.e. background nutrient level) [35,36]. If nutrients are highly limiting, rhizospheric nutrients can be quickly depleted, making it costly to sustain long-lived roots [35]. Our results implied that local root death might be accelerated by the imbalance between nutrient demand of the whole individual and nutrient availability around individual roots, as root mortality was high either when total added nutrients were high and most roots were in the nutrient-rich patches, or when total added nutrients were low and most roots were in the nutrient-poor patches.

5. Conclusion

It was shown in our study that root modules of S. canadensis responded differently when the level of contrast between high- and low-nutrient patches changed as well as when the amount of roots contained within each patch was altered. Root growth decreased in nutrient-rich patches but increased in nutrient-poor patches with increasing coexisting roots in the same patches. Root mortality increased with increasing coexisting roots in nutrient-rich patches when nutrient contrast was high or in nutrient-poor patches when nutrient contrast was low, and vice versa.

Many studies investigated the systematic regulation of root growth and death in the field of plant physiology and cell biology (e.g. [11,12,37]), but ecological experiments are lacking. Research on fine root production and turnover mainly focuses on interspecific variations and effects of environmental factors such as temperature, precipitation and overall nutrient availability. Little attention has been paid to effects of fine-scale nutrient heterogeneity and heterogeneously distributed root biomass in different soil patches. Our results provided ecological evidence that contrasts in nutrient availability and relative quantity of roots in a certain soil volume can have large effects on the dynamics of root growth and death of individual plants, suggesting that future research investigating fine root production and turnover should also consider the effects of nutrient heterogeneity and root distribution patterns.

Data accessibility

Data are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.cn7mq57 [38].

Authors' contributions

Y.Y., P.M. and Y.L. conceived the study. Y.Y. performed the experiment. P.W., Y.Y. and P.M. analysed the data. P.W., Y.Y., P.M. and Q.Z. wrote the manuscript. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

This research was supported by the National Natural Science Foundation of China (31700453, 30830024 and 30770330) and the Fundamental Research Funds for the Central Universities (KJQN201819).