Effects of root phenotypic changes on the deep rooting of Populus euphratica seedlings under drought stresses

Experimental method

Growth measurements and destructive sampling

Intact root systems were cleaned of soil with a gentle water jet while a sieve was used to collect any root fragments detached from the system during this process. The cleaned root systems were then floated on water in a transparent tray and imaged using an Epson Expression Perfection V850 Pro Scanner with 800 DPI resolution (Fig. 2). Adobe Photoshop CS6 software was used to reduce image noise and black margins. The image analysis software (Win-RhIZO 2013a; Pro Instruments Inc., Québec, Canada) was used to analyze images and to estimate the length of total, lateral, and distal roots, the average diameter of the total roots, the number of lateral roots, as well as the external path length (p_e), magnitude (μ), and altitude (a) (Table 1). The taproot was then detached from the root system with scissors, and its length and dry mass were determined. Lateral roots were defined as the root segments connected to the taproot with a root order >3, so as to guarantee that such roots perform conducting and foraging functions. Three intact lateral root branch segments with over three levels of root order were selected randomly and scanned to estimate the average length and diameter of the distal roots. Using a scalpel, all distal fine roots were dissected and was determined with its dry mass. If it was impossible to meet these criteria, the whole root system was measured. The dry weights of the taproot, sampled distal roots, and the other root parts were obtained by air drying plant matter in an oven at 72 °C for 60 h. Total root biomass, SRL, biomass-related variables, and two topological indices were calculated using these dry weight values (Table 1).

Table 1:

Calculated root variables and its abbreviations, units, description and functional role.

Root variables (Abbrev.) units	Description	Functional role
Relative root depth (RRD) cm taproot (g plant)−1	Taproot length per total plant dry mass	Capacity for deep rooting
Root mass fraction (RMF) g root (g plant)−1	The proportion of total plant mass allocated to roots	Total plant resources allocated to root functions (e.g., absorption, transport, foraging, anchoring) Markesteijn & Poorter (2009)
Taproot mass fraction (TRMF) g taproot (g root)−1	The proportion of total root mass allocated to taproot	Total root resources allocated to the taproot functions (e.g., anchoring, foraging, transport)
Specific taproot length (STRL) cm taproot (g taproot)−1	Taproot length per amount of taproot biomass invested	The efficiency of taproot resources used to deep rooting
Lateral root branching density (LRBD) number lateral root (cm taproot)−1	Lateral roots branching number per unit of taproot length	Potential capacity of exploration and exploitation to horizontal soil resources
Topology Index (TI)	TI = log10(a)/log10(μ); altitude a is number of links in the longest path from base to tips; magnitude μ is number of external links or the number of root tips Glimskär (2000)	TI vary from one close to 0.5, and DBI vary between zero and one, both with large values indicative of a more herringbone-like root system that are thought to be more efficient at intercepting mobile resources, such as water, by extensive soil exploration, contrasted with dichotomous-like systems that are better at acquiring immobile resources, such as phosphorus by intensive soil exploration Fitter (1987)
Dichotomous branching index (DBI)	DBI = (pe − min(pe))/(max(pe) − min(pe)); pe is the sum of the number of links in all paths from each external link to the base link; max(pe) and min(pe), respectively is the theoretical external path length for a system of given magnitude that has a completely herringbone and dichotomous topology, for detailed calculation see Šmilauerová & Šmilauer (2002)

DOI: 10.7717/peerj.6513/table-1

Root phenotypic adjustments and deep-rooting capacity analysis

As in many other studies, root-shoot allocation was quantified as the fraction of plant biomass invested in the roots (RMF; see Table 1 for definition), and taproot morphological change could be expressed as STRL (see Table 1 for definition). These two aspects of root phenotypic change both have independent effects in the adjustment of root length (i.e., root depth in this study).

As it is difficult to fully measure root architecture, previous studies commonly used topology to describe the altered branching patterns indicative of a facet of the root architectural change (Fitter, 1987; Harper, Jones & Sackville, 1991; Lynch, 1995). Here, we not only used the topological index (Fitter, 1987; Glimskär, 2000) and the dichotomous branching index (Šmilauerová & Šmilauer, 2002), both commonly applied to characterize root topology (i.e., TI and DBI, respectively; see Table 1 for definitions), but also used the TRMF (see Table 1 for definition) to characterize root architectural change.

It is logical to characterize root architecture as the biomass proportion of a certain functional root module in relation to the whole root system (e.g., TRMF) because the root branching pattern just refers to the coordinated growing relation among different functional root modules, and this relation could be represented with biomass proportion as a mass proxy. In particular, an extreme herringbone-like branching pattern has been found to be primarily confined to the main axis (Fitter et al., 1991)—that is, possessing the largest proportion of taproot biomass. Therefore, considering taproot as a key root module functioning in deep rooting, in this study TRMF was specially used to trace the facet of root architectural change that contribute to deep rooting. In addition, the correlation between TRMF and the commonly used TI and DBI has been examined to determine the availability of TRMF to represent root architecture (R² = 0.643 and 0.698, respectively, both P < 0.001; Fig. S1), and the results indicated that TRMF could be feasibly used in this study.

Populus euphratica has an obvious taproot that determines root system depth, and thus absolute root depth can be reflected by taproot length. Accordingly, RRD (see Table 1) was calculated as the absolute taproot length to taproot biomass ratio. Finally, RRD can be factored into RMF, TRMF, and STRL, as follows: (1) $\text{RRD}=\text{RMF}\times \text{TRMF}\times \text{STRL}$

Data analyses

Differences observed in biomass allocation as well as in root architecture and morphology among the three drought treatments were tested using one-way analysis of variance (ANOVA, Welch’s F-test). After that, variations of statistical significance were further subjected to post hoc pairwise analysis by applying t-tests with Bonferroni corrections, or Games-Howell tests if the homogeneity of variances was not assumed, considering P < 0.05 as significant. The dependence between TRMF and TI or TRMF and DBI was determined by Pearson’s correlation analysis. Statistical analyses were performed in SPSS (version 19, SPSS Inc., Chicago, IL, USA).

We calculated the relative contributions of the variance in RMF, TRMF, and STRL to RRD, referencing the variance partitioning method of Rees et al. (2010) and Freschet, Swart & Cornelissen (2015). Given that $\text{RRD}=\text{RMF}\times \text{TRMF}\times \text{STRL}$, our calculation can be expressed as: (2) $\text{rrd}=\text{rmf}+\text{trmf}+\text{strl}$ where the lowercase acronyms indicate log_e-transformed variables (e.g., $\text{rrd}=\text{ln}\left(\text{RRD}\right)$). Thus, the variance decomposition of rrd, for instance, can be expressed as follows: (3) $\begin{array}{l}\text{Var}\left(\text{rrd}\right)=\text{Var}\left(\text{rmf}\right)+\text{Cov}\left(\text{rmf},\text{trmf}\right)+\text{Cov}\left(\text{rmf},\text{strl}\right)+\text{Var}\left(\text{trmf}\right)\\ +\text{Cov}\left(\text{rmf},\text{trmf}\right)+\text{Cov}\left(\text{trmf},\text{strl}\right)+\text{Var}\left(\text{strl}\right)+\text{Cov}\left(\text{rmf},\text{strl}\right)\\ +\text{Cov}\left(\text{trmf},\text{strl}\right)\end{array}$

Following Eq. (3), as a sample, the contribution of variation in trmf to the variation in rrd can be written as: (4) $\text{Cont}\left(\text{trmf}\right)=\left[\text{Var}\left(\text{trmf}\right)+\text{Cov}\left(\text{rmf},\text{trmf}\right)+\text{Cov}\left(\text{trmf},\text{strl}\right)\right]/\text{Var}\left(\text{rrd}\right)$

Where Var is the variance and Cov is the covariance. This variance partitioning was only performed when substantial variation (i.e., ≥15%) in rrd was observed across treatments, so as to avoid meaningless results.

Drought-induced phenotypic changes in biomass, root morphology, and root architecture

Drought stress inhibited P. euphratica seedlings’ growth in both biomass and root morphology (Table 2). Drought conditions caused a dramatic decrease in the biomass of different plant parts, total root length, the average length of lateral roots and distal roots, and taproot length. However, when examining responses to differing drought intensities, taproot length, and taproot biomass did not display a consistent declining trend with increasing drought intensity, and there was no difference in taproot length and taproot biomass between MD and SD treatments (Table 2). Moreover, the diameter of different root classifications had different responses to drought (Table 2). Root diameter (RD) (calculated with total roots) did not change significantly under drought conditions, but distal RD increased and lateral RD decreased significantly (Table 2). Taproot diameter decreased with drought, but not significantly. Furthermore, while the SRL (calculated with total roots) did not change by drought, the SRL of the distal roots (i.e., SDRL)—the root components more actively involved in water uptake—showed a significant decrease in SD in relation to OW (Table 2).

Populus euphratica seedlings showed a high plasticity in root branching patterns under different drought stresses. DBI and TI measures the degree to which a root is perfectly herringbone (DBI or TI equal to one) or dichotomous branching (DBI equal to zero, or TI close to 0.5). Values of DBI and TI differed significantly among the three drought treatments, but both of them in MD didn’t differ from in OW or in SD (Table 2). The ranges of value presented by the total samples were 0.650 to 0.844 for TI and 0.032 to 0.392 for DBI. Lateral root branching density, a simple but direct trait reflecting root branching, showed a markedly decreased trend from OW to SD treatments (Table 2).

Different levels of root phenotypic variables related to deep rooting changed differently under drought

Drought had a significant positive effect on RRD, but this increase was not proportional to the differences in RRD between SD and MD, it was almost three times greater than the difference in RRD between MD and OW (Table 3). Regarding the indices representing different levels of root phenotypic changes, RMF and TRMF increased markedly with increasing drought severity, while STRL did not significantly increase under drought conditions. In addition, the TRMF under SD was not significantly different from that under MD, but RMF markedly increased under SD in relation to that under MD (Table 3).

Relative contributions of different root phenotypic changes to deep rooting

These results indicated that drought stress had a positive effect on RRD mainly via significant positive effects on TRMF and RMF. Generally, TRMF and RMF contributed to over 75% of the variation in RRD, while STRL contributed to approximately only 20% (Fig. 3). Moreover, TRMF always had the largest relative contribution to the variation of RRD both under MD (55%) and under SD (41%) in relation to that under OW (Fig. 3). Furthermore, the relative contributions of RMF, TRMF, and STRL to RRD variation changed with drought intensity (Fig. 3). TRMF contributed 55% of RRD variation between MD and OW, which was larger than the 27% contributed by RMF and 18% by STRL. However, under SD, RMF contributed to 37% of the variation in RRD, nearly equal with 41% for TRMF. The contribution of STRL on RRD variation also increased slightly to 22%. In addition, the variation of RRD from MD to SD treatments was even more attributable to RMF changes (41%) than to TRMF changes (36%) (Fig. 3).

Root adaptive responses to drought stress in terms of allocation, morphology, and architecture

One of the basic ways for plants to adapt to a shortage of below-ground resources is to maximize fine root area (e.g., decreased RD and increased SRL under drought stress) (Fitter, 1985; Cortina et al., 2008; Olmo, Lopez-Iglesias & Villar, 2014). However, in the present study, the RD and SRL of whole roots exhibited no significant change, but the RD of distal roots increased and SDRL decreased under drought conditions (Table 2). Commonly, roots with a smaller RD and higher SRL is advantageous under drought stress, as the smaller RD conveys higher resistance to root embolism (Alameda & Villar, 2012) through the smaller xylem vessel diameter (Fitter, 1987), and a higher SRL probably is meant to commit limited carbohydrate supply for extensional growth (Trubat, Cortina & Vilagrosa, 2006). The unusual results (increased RD of distal root and decreased SDRL under drought) in this study were likely due to the smaller SRL, which is positively correlated with root life span and respiration rate (McCormack et al., 2012), allowing P. euphratica seedlings to save considerable energy in dry soil through a low root turnover rate in distal roots. Moreover, thicker fine roots would be able to penetrate into the more compacted soils imposed by soil drying (Bengough et al., 2005). In addition, the thicker distal roots of P. euphratica seedlings show higher drought resistance to preserve their vitality in the topsoil of arid-region floodplains where soil water availability is widely fluctuating with the regime of flood and dry seasons (Leon et al., 2011).

Plants mostly elongate root internode lengths to adapt to drought environments (Nicotra, Babicka & Westoby, 2002) as this also extends the volume of root exploring (Fitter & Stickland, 1992). However, in our results, average root length did not significantly change and lateral root length clearly decreased under drought stress (Table 2). Ecologically, floodplain soil possesses a high infiltration rate, which is due to sandy soil layer (several meters in depth) deposited under the 10–50 cm deep surface clay soil layer formed by floodwater sedimentation (Thevs et al., 2008). Thus, rooting downward into deep soil is more critical for drought-stressed P. euphratica seedlings than rooting in other directions and shortening most of their lateral roots can avoid inefficient investments in the horizontal exploration of soil zone (Padilla & Pugnaire, 2007; Bauerle et al., 2008).

Root architecture plays a major role in determining root resource-uptake efficiency (Fitter, 1987; Lynch, 2005). This multidimensional root feature is generally described by measuring root topology (Harper, Jones & Sackville, 1991) or branching density/intensity (Kong et al., 2014). In the present study, the lateral root branching density was reduced under drought conditions (Table 2). From a functional point of view, the sparse lateral root branching of P. euphratica seedlings seems to conserve the high metabolic cost of root construction and maintenance, which can commonly exceed 50% of daily photosynthesis (Lambers, Atkin & Millenaar, 2002). Besides, the sparse lateral root branching may reduce the competition for water among the roots of an individual plant (Fitter et al., 1991; Taub & Goldberg, 1996), which effectively increases the uptake efficiency per unit of lateral root length (Postma, Dathe & Lynch, 2014).

Our results indicated that P. euphratica seedlings tended to create a herringbone-like branching pattern under drought conditions (Fig. 2), as revealed by the increased TI and DBI values under increased drought treatments (Table 2). This finding is in line with most model-based and empirical studies conducted on plants under drought conditions (Fitter, 1986; Taub & Goldberg, 1996). Herringbone-like root systems possess higher exploration efficiency (Ho et al., 2005; Paula & Pausas, 2011), thereby allowing for P. euphratica seedlings to reach water-rich deep soils quickly. Additionally, it should be noticed that in our experiment there is a vertical soil moisture gradient in the plant pots. Other studies performed with more actual ground water conditions and different soil texture layers would be an alternative strategy to determine and confirm deep-rooting of P. euphratica seedlings.

Changes in absolute and relative root depth under drought stress

The four-month-old P. euphratica seedlings grown in our experiment ultimately presented total biomasses ranging from about 0.2–0.5 g (Table 2), and, surprisingly, they developed taproots with approximately 26 to 35 cm long (Table 2). This indicates that, despite their low biomass accumulation rate, P. euphratica seedlings have a great capability to root deeply at their early stages. The miniscule biomass accumulation by the first-year seedlings might be due to their extremely small and light seeds (0.1–0.2 g per 1,000 seeds). A similar ontogeny was also found in other studies concerning first-year riparian seedlings. An experiment under drought conditions conducted by Wang et al. (2015) showed that P. euphratica seedlings sown in April had about one g dry mass and over 22 cm taproot length by the end of July. Likewise, a study of riparian tree seedlings (family Salicaceae) conducted by Stella et al. (2010) demonstrated that three-month-old cottonwood seedlings (P. fremontii) had 0.3 g dry weight with roots over 20 cm in depth, and that three-month-old Salix exigua and S. gooddingii seedlings developed root depths exceeding 25 and 40 cm, respectively, despite exhibiting dry weights of only 0.22 and 0.4 g, respectively. Clearly, the tiny but deep-rooting seedling phenotype seems common in riparian tree species growing in arid regions.

Changes in plant root depth under drought stress are controlled by two processes (Fig. 1) (Sultan, 2000). On the one hand, drought stress weakens photosynthesis, dwindles the accumulated biomass and body size, and thereby leads to a shortening of the root depth. On the other hand, the root depth is influenced by different drought-induced root phenotypic responses. In this study, the absolute root depth of P. euphratica seedlings decreased significantly (Table 2) while the RRD increased significantly under drought conditions (Table 3), indicating that the limitation in growth caused by drought had an overwhelming effect on root depth, but which was compensated by enhancing deep-rooting capacity of P. euphratica seedlings under drought.

Relative contributions of root phenotypic changes to the increase of relative root depth

Our results indicated that changes to root architecture and root-shoot allocation in P. euphratica seedlings dominated for achieving deep rooting under drought conditions, while the role of morphological changes was minor (Fig. 3). This supports the perspective of Freschet, Swart & Cornelissen (2015) that root-shoot allocation was more important than root morphological changes for plant adaptations to changing environmental conditions. Besides, STRL in this study did not significantly increase under drought stress (table 3) and maintained relatively slight contributions to deep rooting regardless of drought intensity (Fig. 3), although taproot stretching seems to be an efficient way to increase root depth. This implies that phenotypic changes in module-level (morphology) seems not much important for plants to respond to belowground resources, given that the role of SRL of fine roots (absorptive roots) are also marginal for plant phenotypic adaption to nutrient limitation (Kramer-walter & Laughlin, 2017). The potential negative effects of increased STRL on taproot function may account for this result. Because SRL is determined negatively by the RD and root tissue mass density (Nicotra, Babicka & Westoby, 2002), increased STRL means decreased taproot diameter or tissue density. This negatively affects taproot conduction, anchorage, and penetration, which are all essential functions for seedlings that are suffering with drought conditions.

The relative contributions of changes in root-shoot allocation and root architecture were altered under different drought intensities (Fig. 3). Under MD, root architectural changes played a decisive role (contribution over 50%) for increasing RRD. However, under SD, the root-shoot allocation response was stronger than that under MD (Table 3), and its relative contribution to deep rooting became nearly as important as that of root architectural changes (Fig. 3). This is likely because increased root-shoot allocation would have decreased photosynthetic capacity and accumulation of photosynthates (Muller et al., 2011). Root architecture changes seems more carbon economical under drought conditions, by which plant would root deeply only at the cost of weakening exploration capacity to the horizontal and surface soil (Thevs et al., 2008). Moreover, the variation of RRD from SD to MD is about three times that of the variation from OW to MD (Table 3), which indicates that keeping deep root is more important when drought becomes severe. Therefore, adjustment in root architecture seems to be insufficient for P. euphratica seedlings to root deeply under severe drought conditions. From a functional view, the seedlings possibly cannot easily acquire the adequate amount of water to maintain metabolic processes with increased drought stress. Accordingly, the seedlings would rely more on allocating biomass to roots, which have an advantage in minimizing water loss of shoot transpiration and in enhancing deep rooting potential (Brunner et al., 2015). In addition, our study revealed a practical implication for breeding P. euphratica seedlings. To promote the survival of container seedlings transplanted in the field, it is a good idea to breed the seedlings under MD before transplantation, in order to promote deeper and steeper root systems while minimally affecting their sizes.

It has been widely reported for many plants, including P. euphratica seedlings (Bogeat-Triboulot et al., 2007), that root-shoot allocation responds significantly only to severe environmental stresses (Poorter et al., 2012). A general explanation is that plants maintain their above-ground growth for as long as possible under moderate soil environmental stresses to keep aboveground competitiveness (Padilla et al., 2009; Poorter et al., 2012), but how plants tackle moderate belowground resources stress to maintain shoot growth is still unclear. Our results indicate that P. euphratica seedlings are able to adapt to MD mainly via phenotypic adjustments within their root systems. Here, we could raise an assumption that before root-shoot allocation strongly responds to severe drought (Poorter et al., 2012), change in root architecture mainly shoulder the absent role of root-shoot allocation for drought-suffered plant. This hypothesis is based on a premise that plant phenotypic changes at the organ-level are more carbon economical than that at the individual-level, while the latter should be more water economical than the former for drought adaptation. To confirm this hypothesis, more functionally-different species need to be further studied.