Variation in global traits
Among the global traits, the CVs of nitrogen uptake, phosphorus uptake, total root length and root tips number were ≥ 0.25 (Table 2). Nitrogen uptake, phosphorus uptake, root angle, maximal root depth, root width and branch intensity were significantly different among the tested genotypes (P < 0.05). The CVs of three shoot traits including trefoil number, shoot height and shoot dry mass were all < 0.25, and there was no significant difference among tested genotypes. Total root length ranged from 116 to 371 cm, with an average of 217 cm. There was a positive correlation between root dry mass and total root length (Figure. 2). The largest genotype of total root length and root dry mass were 3.19- and 2.80-fold times than the smallest genotype, respectively. Twelve genotypes (LT, DY, YS, BJX, PL, KS, B416, DYH, WL354, Z2, Z1 and L801) had large root systems, and 36 genotypes had medium root systems, and 5 genotypes (XB, B218, WL168, MF and LB) had small root systems. Chinese breeding lines had the highest total root length with an average of 229 cm, followed by introduced cultivars and Chinese native cultivars, but there was no significant difference among them. The Chinese breeding lines LM801 had the longest total root length and greatest root dry mass among all alfalfa genotypes. There were significant differences in root spatial distribution (root angle, maximal root depth and root width) among tested alfalfa genotypes (P < 0.05). Root angles ranged from 106° to 168°, with an average of 140°; root width ranged from 8.47 to 21.2 cm, with an average of 17.0 cm; the maximal root depth ranged from 20.2 to 51.8 cm, with an average of 39.1 cm. There were 29 alfalfa genotypes with a maximum root depth of more than 40.0 cm. The introduced cultivars WL354 and QTZ had the maximum and minimum root depth, respectively.
Table 2
Minimum, Maximum, Mean, Medium, standard deviation (SD), coefficient of variation (CV) and P value of 44 global traits, local traits and anatomical traits of 53alfalfa genotypes
Traits | Minimum | Maximum | Mean | Median | SD | CV | P value |
Global traits | | | | | | |
TN | 3.00 | 6.33 | 4.28 | 4.33 | 0.64 | 0.15 | 0.202 |
SH | 5.97 | 12.7 | 8.77 | 8.50 | 1.58 | 0.18 | 0.099 |
NU | 1.93 | 9.77 | 5.51 | 5.14 | 1.89 | 0.34 | 0.000 |
PU | 0.22 | 1.23 | 0.68 | 0.70 | 0.22 | 0.33 | 0.000 |
SDM | 15.8 | 50.7 | 28.8 | 28.1 | 6.84 | 0.24 | 0.571 |
RDM | 9.00 | 25.2 | 15.5 | 15.4 | 3.65 | 0.24 | 0.276 |
TDM | 25.1 | 75.9 | 44.4 | 45.2 | 9.93 | 0.22 | 0.495 |
RA | 106 | 168 | 140 | 141 | 13.0 | 0.09 | 0.020 |
MRD | 20.2 | 51.8 | 39.1 | 40.5 | 7.60 | 0.19 | 0.007 |
RW | 8.47 | 21.2 | 17.0 | 17.2 | 2.92 | 0.17 | 0.037 |
RD | 0.29 | 0.36 | 0.33 | 0.33 | 0.02 | 0.05 | 0.264 |
RL | 116 | 371 | 217 | 208 | 57.0 | 0.26 | 0.073 |
RSA | 12.8 | 35.3 | 21.1 | 20.9 | 5.09 | 0.24 | 0.131 |
RV | 0.10 | 0.27 | 0.16 | 0.17 | 0.04 | 0.23 | 0.231 |
RTN | 34.3 | 108 | 70.1 | 70.7 | 18.9 | 0.27 | 0.281 |
SRL | 8.54 | 19.5 | 14.4 | 14.3 | 1.86 | 0.13 | 0.559 |
SRA | 0.89 | 1.95 | 1.40 | 1.40 | 0.16 | 0.11 | 0.747 |
RSM | 0.41 | 1.23 | 0.56 | 0.55 | 0.12 | 0.22 | 0.151 |
RTD | 77.6 | 240 | 96.9 | 94.1 | 22.4 | 0.23 | 0.450 |
RLI | 3.43 | 8.60 | 5.83 | 5.75 | 1.30 | 0.22 | 0.836 |
BI | 0.21 | 0.51 | 0.33 | 0.31 | 0.06 | 0.19 | 0.010 |
Local traits | | | | | | | |
RL-thin | 25.3 | 157 | 77.6 | 73.5 | 30.4 | 0.39 | 0.015 |
RL-thick | 71.9 | 214.3 | 140 | 144 | 32.8 | 0.23 | 0.204 |
RTN-20 | 22.7 | 72.7 | 49.6 | 50.7 | 13.7 | 0.28 | 0.773 |
RTN-40 | 3.00 | 27.0 | 16.2 | 15.7 | 6.65 | 0.41 | 0.271 |
RTN-60 | 0.00 | 17.3 | 4.32 | 3.00 | 4.41 | 1.02 | 0.021 |
RL-20 | 98.2 | 302 | 176 | 173 | 46.5 | 0.26 | 0.373 |
RL-40 | 6.75 | 69.9 | 32.0 | 29.8 | 13.2 | 0.41 | 0.107 |
Traits | Minimum | Maximum | Mean | Median | SD | CV | P value |
RL-60 | 0.00 | 37.8 | 9.16 | 7.37 | 9.16 | 1.00 | 0.035 |
RA-20 | 8.92 | 28.5 | 16.8 | 16.7 | 4.21 | 0.25 | 0.466 |
RA-40 | 0.86 | 6.84 | 3.31 | 3.24 | 1.22 | 0.37 | 0.160 |
RA-60 | 0.00 | 3.29 | 0.95 | 0.84 | 0.87 | 0.92 | 0.011 |
RV-20 | 0.06 | 0.21 | 0.13 | 0.13 | 0.03 | 0.25 | 0.531 |
RV-40 | 0.01 | 0.05 | 0.03 | 0.03 | 0.01 | 0.34 | 0.249 |
RV-60 | 0.00 | 0.03 | 0.01 | 0.01 | 0.01 | 0.86 | 0.003 |
Anatomical traits | | | | | | |
TCA | 0.11 | 0.58 | 0.31 | 0.30 | 0.14 | 0.44 | 0.000 |
CCA | 0.03 | 0.33 | 0.16 | 0.13 | 0.09 | 0.54 | 0.000 |
CCF | 4.17 | 22.0 | 10.7 | 8.17 | 5.27 | 0.49 | 0.000 |
CCS | 96.3 | 570 | 240 | 233 | 106 | 0.44 | 0.000 |
CCC | 133 | 2898 | 1055 | 568 | 893 | 0.85 | 0.000 |
AA | 0.03 | 0.24 | 0.10 | 0.09 | 0.06 | 0.57 | 0.000 |
TSA | 0.05 | 0.47 | 0.23 | 0.23 | 0.10 | 0.42 | 0.000 |
VN | 13.2 | 44.5 | 27.3 | 28.2 | 6.53 | 0.24 | 0.000 |
XVA | 0.01 | 0.07 | 0.03 | 0.03 | 0.01 | 0.44 | 0.000 |
Variation in local traits
The local root traits had larger variation than the global root traits (Table 2). The CVs of all local root traits were greater than 0.25, except for root length in diameter-thick (diameter class ≥ 0.2 mm) which was 0.23. There were significant differences among the tested genotypes in root length in diameter-thin, root tips number-60, root length-60, root area-60 and root volume-60 (P < 0.05). The average root diameter of all genotypes was 0.33 mm, the root length in diameter-thin (diameter class < 0.2 mm) and thick accounted for 35.7% and 64.3% of the total root length, respectively. On average, the distribution of root tips number, root length, root area and root volume decreased with the increase of soil depth. About 79.0%, 16.0% and 5.00% of the root length across all genotypes was distributed in the 0–20 cm, 20–40 cm and below the 40 cm soil layer, respectively. Genotypes with large root systems had the maximum root tips number, root length, root area and root volume followed by medium and small root systems in each soil layer (Figure. 3). The differences in root distribution among genotypes with different root system sizes were mainly reflected in the soil layer below 20 cm.
Variation in anatomical traits
Except for vessel number, the CVs of other anatomical traits were all higher than 0.25 (Table 2). Significant difference was detected in all tested root anatomical traits across genotypes (P < 0.001). Example root anatomical features of genotypes DYH and BJX (large root systems), LD and CY3 (medium root systems), and LB and XB (small root systems) were shown in Fig. 4. Cortical cell count of large root system was higher than that of small root system, but xylem vessel area was contrary (Fig. 5). Aerenchyma area of small root system was higher than that of medium root system followed by large root system.
Correlation among traits
Among 44 measured traits, 4 global traits, 13 local traits and 8 anatomical traits with larger coefficients of variation (CVs ≥ 0.25, Table 2) were used for Pearson’s correlation analysis. Nitrogen uptake, phosphorus uptake both showed positive correlation with total cortical area, total stele area and xylem vessel area (Table 3). Total root length and root tips number exhibited positive correlation and these two traits were positively correlated with all local root traits (P < 0.01). Most local traits had significant correlation with each other (P < 0.05), such as root length in diameter-thin and root tips number-40 were positively correlation with all local traits (P < 0.01).
Root tips number-20 was positively correlated with total stele area, and negatively correlated with total cortical area, cortical cell area, cortical cell files and cortical cell count (P < 0.05). Root length-40 and root area-40 were both positively correlated with cortical cell files and cortical cell count (P < 0.05). The rest of anatomical traits did not exhibit significant correlation with global and local traits. But most anatomical traits were correlated with each other (P < 0.01).
Moreover, total root length, root length in diameter thin, root length in each section, root tips number, root tips number in each section and maximal root depth were all positively correlated with shoot dry mass and root dry mass (P < 0.05; Fig. 6). Total stele area and xylem vessel area were both positively correlated with phosphorus uptake and nitrogen uptake (P < 0.05; Fig. 7).
Determination of trait variation
Principal component analysis (PCA) was performed for the selected 25 traits with CVs ≥ 0.25 (Table 4). Six principal components (PCs) were identified with eigenvalues > 1, capturing 88.9% of the total variation in these traits across the tested genotypes. Those six PCs could represent most of the information provided by the raw data. PC1 represented 29.4% of the variability and consisted of all the root morphology except for root tips number-20. PC2 represented 20.9% of the variability and consisted of root tips number-20 and all the cortical traits, such as total cortical area, cortical cell area, cortical cell files, cortical cell size and cortical cell count. PC4 represented 11.4% of the variability and consisted of nitrogen uptake, phosphorus uptake, aerenchyma area, total stele area and xylem vessel area.
Table 4
Principal component (PC) analysis of 25 selected traits with CVs ≥ 0.25 and the proportion of variation in each principal component
Trait | PC1 | PC2 | PC3 | PC4 | PC5 | PC6 |
NU | 0.06 | -0.01 | 0.24 | 0.73 | -0.16 | -0.57 |
PU | -0.04 | -0.02 | 0.26 | 0.68 | -0.23 | -0.59 |
RL | 0.91 | -0.28 | 0.22 | -0.06 | 0.07 | -0.05 |
RTN | 0.82 | -0.34 | 0.18 | 0.02 | 0.04 | 0.09 |
RL-thin | 0.82 | -0.14 | 0.18 | 0.00 | 0.04 | -0.13 |
RTN-20 | 0.59 | -0.60 | 0.33 | -0.07 | 0.14 | 0.01 |
RTN-40 | 0.64 | 0.17 | 0.04 | 0.19 | -0.42 | 0.32 |
RTN-60 | 0.58 | 0.41 | -0.48 | 0.04 | 0.36 | -0.14 |
RL-20 | 0.73 | -0.51 | 0.38 | -0.11 | 0.11 | -0.06 |
RL-40 | 0.74 | 0.42 | -0.13 | 0.16 | -0.37 | 0.13 |
RL-60 | 0.57 | 0.47 | -0.53 | 0.01 | 0.33 | -0.17 |
RA-20 | 0.70 | -0.54 | 0.37 | -0.14 | 0.13 | -0.03 |
RA-40 | 0.74 | 0.41 | -0.14 | 0.16 | -0.42 | 0.17 |
RA-60 | 0.58 | 0.47 | -0.54 | 0.01 | 0.32 | -0.16 |
RV-20 | 0.65 | -0.54 | 0.35 | -0.16 | 0.15 | 0.01 |
RV-40 | 0.70 | 0.39 | -0.15 | 0.15 | -0.46 | 0.21 |
RV-60 | 0.57 | 0.47 | -0.53 | 0.03 | 0.29 | -0.14 |
TCA | -0.04 | 0.59 | 0.58 | 0.33 | 0.31 | 0.27 |
CCA | -0.02 | 0.66 | 0.65 | 0.19 | 0.23 | 0.16 |
CCF | 0.01 | 0.75 | 0.62 | -0.05 | 0.16 | 0.05 |
CCS | -0.04 | -0.54 | -0.49 | 0.40 | 0.04 | 0.30 |
CCC | 0.04 | 0.73 | 0.65 | -0.02 | 0.10 | 0.03 |
Trait | PC1 | PC2 | PC3 | PC4 | PC5 | PC6 |
AA | -0.04 | -0.31 | -0.20 | 0.56 | 0.37 | 0.41 |
TSA | -0.03 | -0.38 | -0.06 | 0.60 | 0.07 | 0.07 |
XVA | -0.11 | -0.21 | 0.00 | 0.82 | 0.22 | 0.20 |
Eigenvalue | 7.34 | 5.23 | 3.76 | 2.84 | 1.67 | 1.39 |
Contributive ratio (%) | 29.4 | 20.9 | 15.0 | 11.4 | 6.69 | 5.56 |
Cumulative contributive ratio (%) | 29.4 | 50.3 | 65.3 | 76.7 | 83.3 | 88.9 |
For each trait, the largest absolutely variable loading score crossing the six components appears in bold. Principal components with eigenvalues > 1 are presented and considered significant (Qiao et al. 2019)
Based on the data of PCA, the distribution of 53 alfalfa genotypes and 23 root traits with CVs ≥ 0.25 in PC1 and PC2 were shown in Fig. 8. Genotypes with three types of root system size showed a clear separation. Except for total cortical area, cortical cell area, cortical cell size, aerenchyma area, total stele area, vessel number, xylem vessel area, the remaining root traits all had positive contribution to root system size. Of the 23 selected root traits, total root length, root length in diameter thin, cortical cell files and cortical cell count contributed the most to root system size.
Composite score and genotype distribution
Based on the composite score of PCA, 12 genotypes with large root systems, 36 genotypes with medium root systems, and 5 genotypes with small root systems were ranked from the least to the most, respectively (Fig. S1). The average composite score of large root system was the highest, followed by medium root system and small root system, but there was no significant difference in composite score among the three. Among the large root system, medium root system and small root system, genotypes DY, AH and MF had the highest composite score, respectively.
The dendrogram of agglomerative hierarchical clustering (AHC) separated the 53 genotypes into five major groups at a rescaled distance of 15 using the average linkage method with squared Euclidean distance as the interval measurement on the same set of 25 traits with CVs ≥ 0.25 (Fig. S2). Group 1 to 5 had 26, 12, 4, 4 and 7 genotypes, respectively. Genotypes with medium root systems distributed in every group, genotypes with large root systems distributed in group 1, 2, 3 and 5, and genotypes with small root systems distributed in group 1, 2 and 4.
Fifty-three genotypes arranged in 5 groups determined by AHC according to the composite score from the least to the most (Fig. S3). Twelve genotypes with the highest composite score in each group presented by root system size (genotypes YS, WL363, LB, LT, G3, B218, Z1, JG, MTW, MF, DY and AH) were selected for further study. The root traits of 12 alfalfa genotypes selected were significantly different, and the ranking of 25 traits (CVs ≥ 0.25) of each genotype was significantly different among 53 genotypes (Table. S2). For example, LT and LB had obvious differences in the ranking of all traits. YS and MTW had similar root length-40 (ranking 15th and 13th, respectively), aerenchyma area (ranking 34th and 38th, respectively) and xylem vessel area (ranking 16th and 13th, respectively), but there were significant differences in other root traits between them.