Diversity in root growth responses to moisture deficit in young faba bean (Vicia faba L.) plants

Germplasm survey

The germplasm survey was conducted at the University of Helsinki’s Viikki Campus greenhouse facility in a randomized complete block design (RCBD) with four replications. The four blocks were sown at seven-day intervals (owing to space limitation) and allowed to grow for 34 days and each block contained one pot of each accession. Throughout the experiment, the photoperiod was set at 14 h light and 10 h dark, and the temperature maintained at 22 °C during the day and 16 °C in the night.

The original set of 201 wet-adapted and 201 dry-adapted accessions (Khazaei et al., 2013) was reduced to 88 based on differences in canopy temperature depression measured in the glasshouse (Khazaei et al., 2013), country of origin and availability of seeds. Ten other accessions (seven from Ethiopia and three from Europe) were selected from the previous screening experiment for acid-soil and aluminium toxicity tolerance (Belachew & Stoddard, 2017). ILB938/2 and Mélodie/2 were included as they have been well studied previously (Link et al., 1999; Khan et al., 2007; Khazaei et al., 2013). Poor germination of 11 accessions reduced this set of 100 to 89, 44 of which were from the previous dry set, 38 from the wet set and 7 from Ethiopian highlands that conform to the criteria of the wet set (Table 1). Since seed quantities were limited, seed size was evaluated as one-tenth of 10-seed weight rather than hundred- or thousand-seed weight.

The experiment was designed to maximize expression of potential root mass by providing plentiful moisture and nutrients. The pots were 3 L in capacity, 20 cm deep and 15 cm diameter with four drainage holes of 2 cm diameter. The bottom of each pot was covered with a thin membrane sheet and then the pots were filled with 0.2 L of sand at the bottom, 2.6 L of perlite, and 0.2 L of sand on the top. Two seeds per pot were sown and after five days, the weaker seedling was removed, leaving the stronger seedling to grow. Nutrient solution was applied at 200 mL automatically every other day from sowing to harvesting for 34 days to keep the medium at field capacity. Pests (thrips) were controlled biologically with parasitic wasps. The nutrient solution was 1 g/L of Superex Peat (Kekkilä Oy, Vantaa, Finland) supplemented with 2 mmol/L CaCl₂, as previously described (Belachew & Stoddard, 2017).

At BBCH stage 39 (Meier, 2001), when there were approximately 9 visibly extended internodes, 30-34 days after sowing (DAS), the following measurements were taken. Stomatal conductance was measured using a Leaf Porometer (Decagon Devices, Inc, Pullman, WA, USA) once per plant. Leaf surface temperature was measured using a FLUKE Model 574 Precision Infrared Thermometer (Fluke Corporation, Everett, WA, USA), chlorophyll concentration was measured as leaf SPAD values from two leaves per plant and the average of the two was recorded using SPAD-502 (Minolta Camera Co, Ltd, Japan). Measurements were taken between 11:00 and 13:00 local time. Plants were harvested at 34 DAS. Shoots were removed above the collar region and roots were carefully removed from the perlite. Both parts were dried in a drying oven at 70 °C for 48 h. Root and shoot dry weight were measured to the nearest 0.01 g and root to shoot dry weight ratio was calculated by dividing the root weight by the corresponding shoot weight. Root mass fraction was calculated as root dry mass divided by total plant dry biomass.

Root phenotyping experiment

The experiment was conducted at Jülich Plant Phenotyping Center (JPPC) (http://www.jppc.de), Forschungszentrum Jülich GmbH, Germany from 23 January to 20 February 2017.

Eight accessions were chosen (Table 1) from the germplasm survey according to their performance in stomatal conductance, canopy temperature, chlorophyll concentration, root and shoot dry weights and root mass fraction values. Accessions showing high values of stomatal conductance and leaf surface temperature were considered as potentially drought susceptible, whereas those showing high chlorophyll concentration, root dry weight, root to shoot dry weight ratio, and root mass fraction, along with low values of stomatal conductance and leaf surface temperature were considered as potentially drought tolerant.

The experiment was conducted in the automated root and shoot phenotyping platform GROWSCREEN-Rhizo using rhizotrons with a size of 90 × 70 × 5 cm (Nagel et al., 2012). The growth medium used was GRAB-ERDE, a dark peat-based substrate (Plantaflor Humus Verkaufs-GmbH, Germany). A total of 2,400 L peat was first machine broken and then passed through a 0.8 cm sieve. The initial moisture content of the peat-soil was 66.3% measured using Electronic Moisture Analyzer (version 1.1, 03/2013, KERN and Sohn GmbH, Germany). Of this, 1,600 L was air dried to 40% moisture content, when it had a water potential of 0.006 MPa according to the water retention curve analysis conducted by the Institute of Plant Nutrition and Soil Science, University of Kiel, Germany.

Nutrient content and other physical and chemical properties of the growth medium were analyzed by LUFA NRW Laboratory, Germany. Dry matter content was 35%, wet bulk density 450 g/L, dry bulk density 158 g/L, pH 5.8, EC 733 µS/cm, KCl in H₂O 1.76 g/L, KCl in CaSO₄ 0.45 g/L, total nitrogen 27 mg/L in CaCl₂/DPTA-Extract (CAT, where DPTA is diethylenetriamine-pentaacetic acid), NH${}_4^{+}$-N 4 mg/L in CAT, NO${}_3^{-}$-N 23 mg/L in CAT, P₂O₅ 22 mg/L in CAT, K₂O 178 mg/L in CAT, Mg 125 mg/L in CAT, and Mn 11 mg/L in CAT.

Water-limited treatment boxes were filled with air dried peat-soil, whereas well watered treatment boxes were filled without drying. Each rhizotron contained approximately 21 L of growth medium. Filling was done in three steps of 7 L peat-soil each followed by regular pressing, to make the compaction of the medium as uniform as possible among boxes. The boxes were then fixed in the robotic system in the greenhouse and tilted at 43° from vertical.

Research design

The experiment was arranged in a split-plot design, with four replicate blocks, two treatments (well watered and water limited) as the main plots and eight accessions as subplots.

Planting and treatment management

The experiment was conducted for 28 days, from seed soaking to plant harvesting, during the vegetative stage of plant growth. Seeds of uniform size were selected from all 8 accessions, washed three times, surface sterilized with 1% NaClO (sodium hypochlorite) (w/v) for 5 min and rinsed 3 times with running tap water. The seeds were soaked in tap water for 24 h, transferred to three layers of moist filter paper in 14 cm diameter Petri dishes (14 seeds/dish) as described in Belachew & Stoddard (2017), and incubated for 96 h at 22 °C in the dark. The seedlings showing uniform root growth were selected and transferred into the rhizotrons. Initially, for establishment, each seedling in well watered treatment received 200 ml water in the automatic irrigation system and those in water limited treatment received 50 ml of water to their roots manually. Following this, the well watered plants were given 100 ml of water every 12 h until the end of the treatment period. In the water-limited treatment, plants received the second 50 ml of water four days after transplanting and thereafter received no more water. The average peat-soil temperature was 22.6 °C, air humidity 58% and air temperature 20.9 °C. The photoperiod was 15 h light and 9 h dark.

Data collected

Root images were automatically taken every day except on Saturday and Sunday from 30 January to 20 February 2017. Images taken 5 days after treatment (DAT), 12 DAT, and 19 DAT were analyzed using the PaintRHIZO software package and dimensions were converted to SI units using 55.53 pixel = 1 cm. The following root distribution and individual root traits were computed (Nagel et al., 2009; Nagel et al., 2012):

• taproot length (cm);

• first and second order lateral roots length (cm);

• total root length (cm);

• root system depth (cm), which represents the maximum vertical distribution of the root system;

• root system width (cm), which represents maximum horizontal distribution of the root system; and

• convex hull area (cm²), which measures the surface area along the transparent plate of the rhizotrons covered by a root system.

To evaluate how much of the whole root system was visible at the transparent plate of the rhizotrons, we measured total root length destructively, using accession DS70622 as a test-case because it had the largest root system in both irrigation treatments. The roots were carefully removed from the potting medium 19 days after the treatment, washed, and preserved in ethanol solution until analysis. One week later, each root system was thoroughly washed, cut into manageable lengths and spread in water on the WinRhizo scanner.

Data analysis

Root images obtained with GROWSCREEN-Rhizo and manual root scanner EPSON A3 Transparency Unit (Model EU-88, Japan) were analyzed using PaintRHIZO and WinRHIZO, respectively, following the methods developed by Mühlich et al. (2008), Nagel et al. (2009) and Nagel et al. (2012).

Quantitative data from the survey and the phenotyping experiment were subjected to analysis of variance using SPSS version 22.0 (IBM Inc., Chicago, IL, USA). Frequency distributions of the survey data showed that most measures gave acceptable fits to the normal distribution. Stomatal conductance and root-to-shoot dry weight ratio both showed excess kurtosis values above 4, and their skewness values were 0.9 and 1.5, respectively, owing to a single outlying accession in each case. Treatment means were separated by Duncan’s Alpha (5%). The difference between the group means of the dry-adapted and wet-adapted sets in the germplasm survey was tested using an independent-samples t-test. In the phenotyping experiment, the two-way ANOVA tested the main effect of treatment, the main effect of accession, and the treatment by accession interaction effect on each sampling date (5, 12, and 19 DAT) separately. A t-test was used to compare the difference in root visibility of DS70622 between watering treatments.

Germplasm survey

There were significant differences between accessions in stomatal conductance, chlorophyll concentration, root and shoot dry weight and root mass fraction values (P < 0.001), root to shoot dry weight ratio (P < 0.01) and leaf surface temperature (P < 0.05). Stomatal conductance ranged 6-fold, shoot dry weight 12-fold, root dry weight 7-fold, root mass fraction 2-fold, seed size 18-fold and leaf surface temperature by 3.1 °C (Table 2). Accessions originating from dry growing regions of the world showed significantly higher chlorophyll concentration (P < 0.001), shoot (P < 0.01) and root (P < 0.001) dry weights than those from wet regions (Table 2 and Table S1).

Chlorophyll concentration showed a weak but significant negative correlation with stomatal conductance and a similarly weak but positive one with root dry weight (Table 3). Root and shoot dry weight were positively correlated with each other (Fig. 1) and with seed weight (Table 3). Root mass fraction was negatively correlated with seed weight.

The five accessions with the greatest root and shoot weights were from the dry set and the five with the lowest were from the wet set (Fig. 1). The accessions with the two greatest total dry weights were DS70622 (5.1 g) and DS74573 (4.8 g) (Fig. 1). Accession DS11320 was an outlier with the highest value of stomatal conductance, along with a one of the highest values of root dry weight. Accession WS114476 was an outlier with the highest value of root-to-shoot dry weight ratio, but this was combined with very low total dry weight production. Eight accessions (Table 4) were chosen for the root phenotyping experiment.

Root phenotyping

The water-limited treatment was sufficiently strong to reduce the lengths of all three classes of root (taproot, lateral and second order lateral roots) at all three time points (5, 12 and 19 days after treatment started (DAT)) below the values found in the well watered treatment (Fig. 2 and Table S2). The main effect of accession on all three root lengths was also significant at all time points. The treatment × accession effect was significant for taproot and second order lateral root lengths at 12 and 19 DAT, but not for lateral root lengths, in which the standard error was large. Lateral roots made the largest contribution to total root length at 19 DAT (Fig. 2), 76% in well watered and 79% in water limited, a non-significant difference.

At 19 DAT, accession DS70622 had the longest lateral roots in both treatments, the longest second order lateral roots in the well watered treatment, the greatest total root length in both treatments, and the smallest difference in taproot and lateral root growth between treatments (Fig. 2). DS11320 had the longest tap root, the second-longest laterals and the second-longest total root length in the well watered treatment. EH06006-6 had the second-longest taproots in the well watered treatment. Mélodie/2 had the shortest taproot, lateral and second order lateral roots in the well watered treatment, whereas in the water-limited treatment, DS74573 had the shortest tap root, and WS99501 had the shortest laterals, second order laterals and total root length (Fig. 2).

In three of the eight accessions, second order lateral roots were not visible at 5 DAT (Table S2). In the water-limited condition, only two of the accessions showed second order lateral roots at 12 DAT, but at 19 DAT, all of the test materials had these roots.

At 19 DAT, genotypic mean total root length and genotypic mean root system depth were positively correlated (r = 0.86, n = 8, P < 0.01), as were taproot length and total root length (r = 0.82, n = 8, P < 0.05).

At the end of the treatment period, the genotypic mean total root length of DS70622 was 3 times longer than those of Mélodie/2 and WS99501. Accessions DS11320 and DS70622 showed the two deepest root systems consistently at all 3 time points and WS99501 had the shallowest root system (Table 5 and Table S3).

On average, the total root length and root system depth recorded under well watered condition was twice that in the water-limited condition. Droughted roots continued to grow throughout the experiment, but more slowly than in well watered conditions, such that the total root length of the droughted treatment was 50%, 41% and 27% of non-droughted at 5, 12, and 19 DAT, respectively (Fig. 2 and Table S3). Similarly, root system depth was reduced by 40%, 46%, and 50%, respectively, at these three time points (Table 5 and Table S3).

Comparison of total root length records obtained from PaintRHIZO and WinRHIZO image analysis software of accession DS70622 indicated that roots in rhizotrons were 32.4% visible. The difference in visibility between the two treatments, 25.5% in the well watered condition and 39.3% in the water-limited condition, was not statistically significant by a t-test.

Root system width differed between treatments, being 46 cm in the well watered condition and 28 cm in the water-limited treatment, but there was no significant difference between accessions.

Convex hull area showed large differences between treatments and between accessions (Fig. 3), but the interaction was not significant (Table 5). Treatment differences in convex hull area increased across the 3 time points (Table 5 and Table S3). Plants grown in the well watered condition showed about 3 times more convex hull area than plants grown in the water-limited condition. Maximum convex hull area was shown in accession DS70622, closely followed by EH06006-6 and DS11320, while WS99501 and Mélodie/2 had the two minimum values (Table 5). Root system width and convex hull area (genotypic means) were positively correlated (r = 0.97, n = 8, P < 0.01), and both traits were positively correlated with taproot length, total root length, and root system depth (r = 0.82 to 0.89, n = 8, P < 0.01, P < 0.05).