Schmidt et al. [31] presented a method to determine the contact area of soil touching the root of a lupin plant. A fundamental difference between applying the algorithm to a seed system instead of a root system is that the seed, upon initiating germination, starts to develop increased air space regions within the previously closed environment. In contrast, a root is typically represented as a solid shape in an X-ray scan showing no air spaces within, unless aerenchyma form. The segmentation of the seed surface was, therefore, extended across this air space so that the resulting object has a closed surface area relevant for determining soil contact (Fig. 1a). The extension was performed using the opening function in VG StudioMax ® v2.2 (Volume Graphics GmbH, Heidelberg, Germany) and remaining air space filled manually. In this dataset, the extension after 1 day of growth increased the seed volume by 19.65% (±6.84%) for the untreated seed and 9.07% (±2.10%) for the pelleted and coated seed. The seed surface was thereby decreased by 23.96% (±1.43%) for the untreated seed and 54.92% (±2.10%) for the pelleted and coated seed. The high decrease in seed surface area has no major influence in seed–soil contact as the majority of the filled space is within the seed. The filling of the seed becomes more important throughout the development of the seed as more air space develops over time. Using the surface determination tool in VG StudioMax ® v2.2, the soil was segmented by selecting the air space as background and a selection of different soil aggregates and particles as material. The resulting region of interest (ROI) of the soil was dilated by +1 voxel so that an overlap with the non-dilated seed was created (Fig. 1b). The ROI of the seed surface was then extracted as a volume and both the segmented surface and the dilated soil surface were copied into the new volume (Fig. 1c). By calculating the ratio of the closed surface area of the seed and the surface area of the dilated soil aggregates within the new volume being in contact with the seed surface, a seed–soil contact percentage of the total seed surface was calculated (Fig. 1d).
Fig. 1Schematic representation of the seed–soil contact calculation process. The original image shows contact areas in blue based on the overlap of dilated soil aggregates. a During the process of germination, seeds open to enable radicle penetration which is indicated with the schematic representation of the 2D slice. The standard process of segmentation was conducted with additional segmentation of the air space within the opening seed. The segmented surface was dilated by +1 voxel. b Brown objects indicate a simplified version of soil aggregates around the seed. Using surface determination, the soil aggregates were segmented and finally dilated by +1 voxel. c The dilated surface from step A was extracted as a volume and the segmented surface (3rd step of A) copied into the volume as a ROI. The surface area of the seed is now listed in determine properties as closed surface area. The dilated soil from step B is also copied into the new volume and the surface area determined via determine properties. d The previously determined surface properties (step C) were used to calculate a seed–soil contact percentage
Full size image
Real soil aggregates are not uniform in shape as Zhou et al. [15] described in their modelling approach. Therefore, it is likely that only a section of the soil aggregate is in contact with the seed although the complete soil aggregate has a far greater size. This means that the water uptake of the seed through this aggregate as a whole is higher than estimated by considering only the size of the contact area (Fig. 2a). We refer to this as the iceberg effect which would result in an underestimation of the water uptake rate if only considering the seed–soil contact percentage (Fig. 2b). To avoid this, we propose that the seed–soil contact should be correlated with the air space–soil space ratio of the surrounding volume as well as a contact surface area—soil space surface area ratio. The X-ray CT data was therefore used to generate an artificial ring surrounding the segmented seed surface to calculate the soil volume and the air volume. Two rings were created spanning 5 voxel (equal to 100 µm) (referred to as ‘short range’) and 15 voxel (equal to 300 µm) (referred to as ‘long range’) representing 10 and 30% diameter of the maximum size of the sieved grains of <1 mm (Fig. 2c). The resulting rings were used for surface determination as described earlier for the segmentation of the soil. The air space was used as a background and several locations of the soil were combined for the surface determination of the soil aggregates. A second ROI was created with a threshold using the minimal and maximal greyscale values from which the soil ROI was subtracted to derive soil and air volume. The volumes were then used to generate the ratio between soil and air space to generate a volume effect estimation of the iceberg effect. In addition, a ratio of the contact surface area divided by the soil surface area within each corresponding ring was calculated. This is referred to as the surface effect of the iceberg effect. The change in soil mass was calculated as the ratio between the short range and the long range effect (Fig. 2d).
Fig. 2The underlying “iceberg effect”. a Schematic representation of the iceberg effect based on 2D slices showing in b. b Example 2D slice showing contact are in blue and example underlying soil grain behind a single contact point (red). c Schematic representation of creating rings. d Calculations for the volume and surface effect as well as the change in soil mass based on the short range and long range ring created in step C
Full size image
The image processing method was then applied on naked and pelleted and coated sugar beet seeds to visualise and calculate the seed–soil contact (Fig. 3). Figure 4A displays the segmented surface area which showed a significantly higher value for the pelleted and coated seeds compared to the naked seed, 63.22 mm2 (±2.00 mm2) and 37.67 mm2 (±1.63 mm2), respectively. The amount of soil which is in direct contact with the seed was significantly higher due to the higher size of the seed (p < 0.001) (Fig. 4B). Using both the surface area and the contact area, a contact percentage was calculated showing pelleted and coated seeds had double the contact with the soil than naked seeds (p = 0.003) (Fig. 4C). While naked seeds had a contact area of about 15.19% (±3.84%), the pelleted and coated treatment had 31.85% (±2.39%), with both the surface area and the contact area significantly higher (p < 0.001) (Fig. 4).
Fig. 33D visualisation of seed–soil contact. The images show an extract of the soil core holding a seedling (naked seed) one day after sowing. a Extracted soil core of the column containing the sugar beet seed. Brown = Soil, Yellow = Air, Blue = Seed. b Extracted region in contact with the seed. Pink = Air in contact with seed, Purple = Soil in contact with seed
Full size image
Fig. 4Example contact ratio of naked and pelleted and coated sugar beet seeds. The seeds have grown for 1 day to ensure that the surrounding environment had sufficient time to settle. A Measured surface area of the seed. B Measured seed–soil contact area. C Ratio of contact area divided by surface area as a percentage. Error bars calculated for standard error deviation. N = 4
Full size image
The ‘iceberg effect’ was calculated for both treatments. The short range effect was calculated by creating the ratio of area of contact, and seed to soil volume ratio in the smaller ring showed a significantly higher percentage for pelleted and coated seeds than for naked seeds, regarding the volume effect (p < 0.001) (Fig. 5A). The equivalent long range effect, calculated using the larger ring-shaped region of interest, resulted in a similar significant percentage difference (p < 0.001) (Fig. 5B). A similar behaviour was observed for the surface effect for both the short range and the long range effect (p < 0.001 in both cases). The change in soil mass between both rings showed a significantly larger percentage for pelleted and coated seeds as well in both effect calculations (p < 0.001 in both cases) (Fig. 5C).
Fig. 5Example iceberg effect calculations of naked and pelleted and coated sugar beet seeds based on contact area, a ring of 100 µm and a ring of 300 µm width. The seeds have grown for 1 day to ensure that the surrounding environment had sufficient time to settle. A Contact area divided by soil volume percentage of the 100 µm ring displayed as the volume effect percentage and contact area divided by soil surface within the 100 µm ring displayed as the surface effect percentage. B Contact area divided by soil volume percentage of the 300 µm ring displayed as the volume effect percentage and contact area divided by soil surface within the 300 µm ring displayed as the surface effect percentage. C Soil volume percentage of the 100 µm ring divided by soil volume percentage of the 300 µm ring as well as surface area of the 100 µm ring divided by the soil surface area of the 300 µm ring. Error bars calculated for standard error deviation. N = 4
Full size image
Figure 6 shows a comparison between the previously calculated seed–soil contact ratio and radicle growth (determined using the polyline tool in VG StudioMax ® v2.2) daily up to 4 days after sowing which demonstrates that even with higher seed–soil contact in pelleted and coated treatments, the radicle length had a slower initial growth. In comparison, the naked treatment with the lower percentage displayed a faster growth rate but resulted in similar final lengths of 41.03 mm (±0.89 mm) and 41.80 mm (±1.20 mm), respectively.
Fig. 6Comparison of the seed–soil contact percentage of day 1 to the radicle length for following days. The radicle length is represented as a line graph on the primary axis. The contact percentage is represented as a bar chart on the secondary axis. Error bars calculated for standard error deviation. N = 4
Full size image
In comparison to the soil columns with sieved (<1 mm) soil, field cores were collected directly from the field immediately after planting and processed using the method described above. As the coring was performed without knowledge of the precise location of the seed in the soil, the positioning of the seed in the core could not be controlled to the same level of precision as with the laboratory prepared cores. Figure 7a, c and e show example 2D slices of the field cores. The seed displayed in Fig. 7c was positioned very close to the edge of the column which could have resulted in soil movement around the seed and therefore alteration of the original seed–soil contact percentage. The seed–soil method was applied to the field cores to calculate a seed–soil contact percentage of 4.79% (Fig. 7a), 31.96% (Fig. 7c) and 17.89% (Fig. 7e). This resulted in an average of 18.21% (±7.84%) (Fig. 8a). The contact within the column is visualised in Fig. 7b, d and f showing that the alteration in soil structure around the seed in Fig. 7d resulted in comparably large areas of contact, hence increasing the seed–soil contact. Calculations of the short range iceberg effect percentages showed slightly lower levels (volume effect: 62.82% (±9.44%); surface effect: 11.15% (±3.38%)) compared to the pelleted and coated seeds in soil columns prepared in the lab (volume effect: 70.75% (±2.36%); surface effect: 13.72% (±0.20%)) as well as for the long range effect (volume effect: 40.59% (±13.64%); surface effect: 3.83% (±1.73%) and volume effect: 56.94% (±6.25%); surface effect: 5.35% (±0.18%), respectively) shown in Fig. 8.
Fig. 72D and 3D representation of pelleted and coated commercial sugar beet seeds in field cores sampled directly after sowing by drilling. Sampling was executed blindly which resulted in non-uniform positioning of the seeds in the column. a, c and e represent 2D slices of X-ray scans showing different levels of contact. b, d and f display 3D representations indicating soil (greyscale), air space (yellow), soil in contact with the seed (red)
Full size image
Fig. 8Seed–soil contact calculations based on X-ray scans of field cores with a pelleted and coated sugar beet seed. a Showing the seed–soil contact calculations based on the described method showing the surface area, the contact area and the seed–soil contact percentage. b Displays the iceberg effect calculations showing the contact area divided by soil volume percentage of the 100 µm ring, the soil volume percentage of the 100 µm ring divided by soil volume percentage of the 300 µm ring and the contact area divided by soil volume percentage of the 300 µm ring displayed as a percentage. Error bars calculated for standard error deviation. N = 3
Full size image