X-ray computed tomography to study rice (Oryza sativa L.) panicle development
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Journal of Experimental Botany Advance Access published August 11, 2015
Journal of Experimental Botany
doi:10.1093/jxb/erv387
This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)
RESEARCH PAPER
X-ray computed tomography to study rice (Oryza sativa L.)
panicle development
Vibhuti M. Jhala and Vrinda S. Thaker*
Plant Biotechnology and Molecular Biology Laboratory, Department of Biosciences, Saurashtra University, Rajkot, 360 005, Gujarat,
India
* To whom correspondence should be addressed. E-mail: thakervs@gmail.com
Received 20 February 2015; Revised 21 July 2015; Accepted 24 July 2015
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
Editor: Chris Hawes
Abstract
Computational tomography is an important technique for developing digital agricultural models that may help farm-
ers and breeders for increasing crop quality and yield. In the present study an attempt has been made to understand
rice seed development within the panicle at different developmental stages using this technique. During the first
phase of cell division the Hounsfield Unit (HU) value remained low, increased in the dry matter accumulation phase,
and finally reached a maximum at the maturation stage. HU value and seed dry weight showed a linear relationship
in the varieties studied. This relationship was confirmed subsequently using seven other varieties. This is therefore
an easy, simple, and non-invasive technique which may help breeders to select the best varieties. In addition, it may
also help farmers to optimize post-anthesis agronomic practices as well as deciding the crop harvest time for higher
grain yield.
Key words: CT scanning, Hounsfield unit (HU), rice (Oryza sativa L.).
Introduction
Rice (Oryza sativa L.) is an important food crop for over half 1997; Luo et al., 2001). Yield in rice can be calculated as the
the human population. Breeding for better quality and high- number of seeds per panicle, the size of seeds, the number of
yield is the main focus of rice geneticists and breeders. There panicles per plant, and the number of plants per hectare. The
is a great demand for improving grain quality and quantity arrangement of seeds in a panicle is central to the growth of
to meet the requirements of an increasing world population. the seeds and the yield of the crop. Many agronomical and/
Understanding the fundamental genetic factors that con- or physiological studies have been conducted to judge the
trol rice yield, together with supporting the breeders’ ability varietal potential of different plants and the correlation of
to manipulate such factors, would be an added advantage growth parameters in yield determination (Koutroubas and
for ensuring an adequate global food supply. The number Ntanos, 2003; Prasad et al., 2006; Li et al., 2012). These stud-
of seeds per panicle is believed to be the most plastic yield ies give a reasonable understanding of seed development and
component, whereas the panicle number and individual seed yield determination; however, they are time-consuming and
weight are under more rigid genetic control (Zhikang et al., require tedious methodologies. In addition, the separation
Abbreviations: CT, computational tomography; DAF, days after flowering; DMA, dry matter accumulation; HU, Hounsfield unit; MPR, Multi-Planar Reformatting;
W/L, window/level.
© The Author 2015. Published by Oxford University Press on behalf of the Society for Experimental Biology.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.Page 2 of 7 | Jhala and Thaker
of seeds from the panicle, followed by an estimation of seed Growth analysis
size and position within panicle restrict the sample size for an For the measurement of fresh and dry weights, the tagged pani-
evaluation of crop yield. cles were harvested at weekly intervals. The seeds were separated,
High through-put, non-destructive phenotyping methods counted, and weighed before and after oven-drying at 60 °C until a
constant weight was recorded. Seed water content was calculated by
provide an opportunity to analyse the growth and develop- differences in fresh and dry weights. Data were collected in triplicate
ment of plants with accuracy and speed (Yang et al., 2013, and the mean dry weights and water content are reported. The data
Humplík et al., 2015). High-resolution X-ray computed were fitted to polynomial equations for a best-fit curve for each data
tomography (CT) scanning has the possibility of visualizing set. Changes in the value per unit time were calculated as rate for a
the panicle in 3-D space in situ because it is a non-invasive parameter and plotted against days after flowering (DAF). ANOVA
(single factor) was performed for data on dry matter accumulation
and non-destructive procedure (Kalender, 2000). In addition, (DMA) and water content.
continuous sectioning of an object can be skeletonized into
2-D images for detailed analysis. In many plant experiments,
the CT scanned data are used to study root architecture in Computational tomography (CT) scanning
soil that may help researchers to understand root growth and CT scanning technology, routinely used with X-rays to visualize thin
cross-sections of an organ, was used here for rice panicles. X-rays
development in relation to its geological conditions (Grose
are generated from a source located to one side of the gantry, attenu-
et al., 1996; Moradi et al., 2011; Schluter et al., 2010). ated through the panicle, and then registered by a series of detec-
Rice functional genomic research using various genomic tors placed on the opposite side of the panicle. Both X-ray source
tools is a prime focus because of its importance for under- and detectors make a synchronous movement around the panicle so
standing a model monocot genome. In addition, in order to that all angles of measurement are covered. As the X-ray penetrates
through a panicle, some X-rays are absorbed. The higher the density
fill the gaps between genomic analysis and breeding for better
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
of a part of the panicle, the higher is its X-ray absorption.
quality high-yielding rice, automated phenotyping methods In a preliminary experiment, a set of scanning configuration
are required to replace the currently practised manual ones parameters was optimized for standardization (see Supplementary
(Duan et al., 2011). In recent years, the number of tillers per Fig. S1 at JXB online). In the first season, the growth stages were
plant have been calculated using CT scanning in rice (Yang determined and standardized for the CT scanning procedure. In the
second season, the HU value and growth were determined for the
et al., 2011). However, to the best of our knowledge, no such
two varieties and their relationship was developed. This relationship
study for panicle development has been reported. Rice pani- was further confirmed with seven other varieties by measuring CT
cle development over time was studied here using CT scan scan parameters at maturity stage.
technology to understand yield potential in rice together with
exploring its use for the best possible agriculture practices.
Image collection
Two rice varieties were selected that differed in their final
yield in order to understand how the CT scan parameters Panicles from the two rice varieties, collected at different develop-
mental stages, i.e. initial cell division (5 d), milky white (10 d), dry
are related to the yield difference. These data were upgraded matter accumulation (15 d), and maturation stage (25 d), were sub-
and analysed by other optional analytical procedures and the jected to CT scanning.
results for two rice varieties at different developmental stages No special preparation of the samples was required for CT scan-
are reported. The CT scanning configurations were adjusted ning. The collected panicles of both varieties were scanned individu-
to optimize data collection for the different stages from initial ally in a Philips high-resolution medical CT scanner (GE Medical
Merge), installed at a private Medical Centre: Prabhat CT Scanning
seed setting to the mature seed. The procedures to construct Hospital, Rajkot, India. The images collected were visualized in CT
a 3-D image of the panicle from the CT scan data, which scan software eFilmLite. The window/level (W/L) adjustment for
were then skeletonized to 2-D images, are described. Briefly, rice panicles was the first step for proper visualization of the images.
an attempt was made to observe a correlation between the The panicles were best observed in 1500/–600 (lungs Window/level).
HU value and growth parameters using X-ray computation All the scanned images were pre-processed in the Merge Healthcare
CT scanner software utilizing the Multi-Planar Reformatting
tomography of the rice panicle. This correlation was substan- (MPR) parameter (Instrument software). Sagittal (dividing into left
tiated by similar analyses in seven different rice varieties at and right pieces) and coronal (dividing into anterior and posterior
maturity. It is hoped that this approach may act as a guide for pieces) Multi-Planar Reformatting (MPR) provides easier visualiza-
deciding the appropriate harvest time. tion by projecting two orthogonal viewing planes from a 2-D image.
The MPR tool gives an arbitrary perpendicular viewing plane from
a 2-D image. By using the MPR, different views were created and
used for determining Hounsfield unit (HU), for panicle length and
Materials and methods seed size measurement.
Plant sampling
Two hybrid varieties of rice (Oryza sativa), PAC801 and PAC807, Hounsfield unit (HU) and seed size determination
were grown and maintained at the Vikram farm, Vapi, Gujarat
The cross-sectional images constructed from the CT scan data are
(20.3667 °N, 72.9000 °E). The former is higher yielding compared
maps of CT numbers (CTN), which are an average of the relative
with the latter. The normal cultural practices such as irrigation, the
measurement of the pixel density in the image, also known as the
applications of fertilizers, insecticides, pesticides etc. were conducted
corresponding volume or ‘voxel’. The CTN values are calculated
to optimize the yield. On the day of anthesis, uniformly growing
using the X-ray linear attenuation coefficient, and the unit of CTN
panicles were tagged for each variety and harvested at the desired
is the Hounsfield unit (HU), briefly calculated as,
periods. The number of panicles per plant and plant height were
noted for 100 plants and the mean values ± standard deviation (SD)
are shown here. CTN (HU ) = µobject – µwater / µobject – µair × 1000X-ray computed tomography of rice panicle | Page 3 of 7
Where, μobject is the mean value of linear attenuation coefficient for higher up to 18 d and declined thereafter (Fig. 1B, C).
the voxel; μwater is the linear attenuation coefficient of pure water; The rate of dry matter accumulation is one of the param-
and μair is that of a standardized air sample (Kalender, 2000). The eters that determine the final yield (Bewley and Black, 1985).
CTN scale is linear and has a central axis at 0, which corresponds to
water, while air has the calibrated value of –1 000 HU. Hence, for the
CTN values, the intensities received by the detectors are expressed in A
35.0
the linear Hounsfield unit (HU) scale for which typical values are –1
000 for air and 0 for water. 30.0
The CTN values were derived for each processed image scan.
Each panicle was divided into three equal zones; upper, middle, and 25.0
Dry weight (mg)
lower. A total of 100 seeds from each scan were used for HU value
and the calculation of seed size. The measurement of seed size was 20.0
done using the in-built MPR tools (see Supplementary Figs S2 and
15.0
S3 at JXB online).
10.0
Pseudo-colouring of images 5.0
ImageJ ver.1.46 software (http://imagej.nih.gov/ij) was used for the
pseudo-colouring of all the images. The pseudo-colouring option of 0.0
the software (the application of 16 colours) was applied for a better 0 10 20 30 40
visualization of the panicles. A pseudo-colour (or indexed colour) Day after flowering
PAC 801 PAC 807
image is a single channel grey image (8-, 16-, or 32-bit) that has col-
Poly. (PAC 801) Poly. (PAC 807)
our assigned to it via a look-up table in ImageJ. The grey values of
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
ImageJ match red, green, and blue values in such a way that the
shades of grey are displayed as coloured pixels: 8-bit grey-coloured
images were turned into a 16-colour image in order of their devel- PAC 801
opmental stages.
B 1.60 12.00
10.00
Results and discussion 1.20
Dry weight rate
8.00
Water content
Growth analysis
0.80 6.00
The plant heights of PAC801 and PAC807 were almost equal
at 125–130 cm: however, the duration of vegetative growth 4.00
was longer (20 d) in PAC801 than in PAC807. In PAC801, 0.40
2.00
the number of seeds per panicle was also greater (170–215)
than in PAC807 (165–200). In addition, the number of till- 0.00 0.00
ers was greater in PAC801 (30–32) than in PAC807 (25–30) 0 10 20 30
(Table 1). At a given time, if the number of plants per unit Day after flowering
area was equal, the final yield was higher in PAC801. Data of DRY WGT RATE FIT WC
seed dry weight and water content were fitted to polynomial
equations of different degrees, and the best-fit equation was
determined statistically by performing a t test for different r2 PAC 807
values. The data on DMA were adequately explained by a
C 2 12.00
3rd degree polynomial equation (Fig. 1A). Seed dry weight,
10.00
started accumulating after fertilization, and entered the linear 1.6
phase of DMA up to 30 DAF. Maximum DMA was recorded
8.00
Dry weight rate
Water content
in PAC801 (27 mg) and PAC807 (21 mg) on 28 DPF. In sub- 1.2
sequent periods, no significant change in DMA was observed 6.00
in both varieties. 0.8
The rate of DMA per day reached a maximum on day 5 in 4.00
both varieties and declined thereafter. The declining trend was 0.4
very sharp in PAC807, whereas in PAC801 the value remained 2.00
0 0.00
Table 1. Yield component data of two rice varieties 0 10 20 30
Day after flowering
Characters PAC-801 PAC-807
DRY WGT RATE FIT WC
Grain yield (tn/ht) 10.1764 9.9788
Height (cm) n=10 124 ± 4 121.5 ± 3.5 Fig. 1. (A) Change in mg dry weight. (B) Water content (WC) and rate
of dry matter accumulation (DMA) in variety PAC801 and (C) PAC807
No. of seeds/spike n=10 192.5 ± 45 182.5 ± 35
against seed age in days. Each data point represents the mean value of
No. of tillers/plant n=10 31 ± 1 27.5 ± 2.5
25 replicates at a given age; ± represents the standard deviation of the
Seed weight (mg) n=200 33.13 31.9 sample or otherwise within the symbols.Page 4 of 7 | Jhala and Thaker
Data on water content revealed a gradual increase in both CTN value of –1 000 HU. One image is generated from a
the varieties, but initially a higher value was observed in single 512 × 512 matrix of CTNs for a given interval along
PAC801. The maximum value was almost similar (12 mg) the subject. The image represents the cross-sectional volume
in both. However, like the rate of DMA, the duration was
longer in PAC801 than PAC807 (Fig. 1B, C). Water status has PAC801 PAC807
been reported to play an important role in the development A E
of cotton (Rabadia et al., 1999), wheat seeds (Chanda and
Singh, 1998), soybean and other crops (Egli, 1990; Egli and
Tekrony, 1997). In cereal crops, grain-filling is a critical and
dynamic process that determines the final grain yield (Borra
and Westgate, 2006). Seed filling is sensitive to water short-
ages as the water relations of the developing seed plays a fun-
damental role in seed filling (Schnyder and Baum, 1992).
In this experiment, single factor ANOVA for DMA and
WC data of both the varieties were statistically significant;
although that of PAC801 were more significant (P ≤0.001)
than PAC807 (P ≤0. 01). From these data analyses, rice seed
development was divided into (i) a cell division phase (0–5 d), B F
(ii) a cell elongation/milky stage (5–12 d), (iii) DMA (10–25
d), and (iv) a maturation phase (25 d onward).
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
Panicles of rice were scanned separately for PAC801 and
PAC807 at the four stages. The panicle was placed on the
bed of the CT scanner and moved vertically through the
X-ray plane under the conditions described earlier (Table 2).
A total of 177–344 images for PAC801 and 144–287 images
for PAC807 were generated covering 10 cm (initial) to 25cm
(mature) panicle length of the respective varieties. Images
were placed in a 2-D view of the complete panicle and 10
images were selected from the centre (Fig. 2; Table 2).
C G
Calculation of HU value
The CT scanned images were opened in the MRP software
for further analysis. In the first screening, a total of 10 images
was selected from the centre for detailed analysis. In rice
panicles, grain development is asynchronous (Mohapatra
et al., 1993; Ishimaru et al., 2003). The degree and rate of
grain filling among spikelets can be very different accord-
ing to their position on a panicle. Thus from the images, the
seeds were selected from the upper, middle, and lower parts
(see Supplementary Fig. S2 at JXB online). The CTN scale
was linear and centred on 0 (zero) which corresponds to
water. Positive and negative values represent matter denser D H
and less dense than water, respectively. Air has a calibrated
Table 2. Image collection from two rice varieties at stages of
developing seeds
CT scan conditions: voltage at 120 kV, current 50 mA, scanning
thickness 1.5 mm, voxel 0.2621 mm3, Mode Tomo and child head
programme.
Stages Initiation Elongation Dry matter Maturation
accumulation
Scan spike length (cm) 10 12 20 22 Fig. 2. Spikelet images of rice varieties PAC801 (A–D) and PAC807 (E–H),
Image numbers: 144 216 201 287 at the initiation (A, E), elongation (B, F), dry matter accumulation (C, G), and
maturation (D, H) phases, respectively. CT scan image is in the Multi-Planar
PAC-801
Reformatting (MRP) tool to visualize images for calculating various topological
PAC-807 173 173 287 344
parameters (for other details see Supplementary Fig. S1 at JXB online).X-ray computed tomography of rice panicle | Page 5 of 7
of the subject. Accordingly, a CT image is a map of rela- with seed maturation; and when the seed DMA stabilized, the
tive density (Kalender, 2000). There is a direct relationship water content was negligible (Fig. 1B, C). Thus the pattern
between CTN and material density. From these images of of HU value and seed water content also showed a relation-
panicles, 100 seeds were used for the HU value and the deter- ship. A similar decrease in water content with wood density
mination of seed size. Mean HU values and size were calcu- and HU value has been reported by Wu et al. (2011). The
lated in all three parts of both the varieties for the different data analysed for HU value and seed dry weight using sin-
developmental stages (Fig. 3). More negative HU values were gle factor ANOVA also showed statistically significant results
observed during the early stage of seed development and (P ≤0.001) for both varieties. The principal advantage of a
moved towards more positive when seeds reached maturity. direct relationship between HU value and water content is
This pattern was similar in both the varieties. In addition, lit- that it can be used to understand the seed water status dur-
tle difference was observed in HU value with position on the ing the developmental stages which is a guiding point for the
panicle. The value ranged from –200 to –750 in the rice varie- time of harvest and for fertilizer application, which is a tedi-
ties studied. At the time of seed set, water content decreased ous process to estimate manually. It has been demonstrated
UPPER
A PAC-801
MIDDLE
BOTTOM
0.00
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
-200.00
-400.00
HU Value
-600.00
-800.00
-1000.00
Initiation Elongation Dry matter Maturation
accumulation
Growth stages
B UPPER
PAC-807 MIDDLE
BOTTOM
0.00
-200.00
-400.00
HU Value
-600.00
-800.00
-1000.00
Initiation Elongation Dry matter Maturation
accumulation
Growth stages
Fig. 3. Changes in HU value in PAC801 (A) and PAC807 (B) at different growth stages of the seed: (1) initiation, (2) elongation, (3) dry matter
accumulation, and (4) maturation. The upper, middle, and bottom bars indicate the position of the seeds on the panicle from the top to the bottom,
respectively. The vertical bars represent ±std of the mean of 100 replicates.Page 6 of 7 | Jhala and Thaker
A in a number of studies that water content and seed dry mat-
1.5 ter accumulation rate are directly correlated (Rabadia et al.,
1999; Egli and Tekrony, 1997). Since the amount of water is
also related to the rate of dry matter accumulation in the rice
log dryweight
1 varieties in the present study and elsewhere, it should also
provide information for screening better yielding varieties for
the breeders.
0.5
y = 1.3176x - 2.2232
In particular, HU values and seed dry weight data can be
R² = 0.9108 better explained by a linear relationship of log (1 000+HU
0
value) versus log dry weight data. The calculation for dry
2 2.2 2.4 2.6 2.8 3 weight at different growth phases and their respective HU val-
log (1000+HU) ues showed a linear relationship (Fig.4A; r2=0.910, n=800).
The above experiment was repeated with seven other rice
B varieties which also showed the linear relationship (Fig. 4B;
1.45
r2=0.942, n=175), thus the authors believe that it may help
breeders to screen for promising rice varieties.
1.4
Pseudo-colouring simulates the architecture of seed
log dryweight
1.35
development studies as well as their image analysis with the
developmental stages. It is easy to trace more or less densely
packed seeds in the panicle by viewing the contrasting images
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
1.3
(Fig. 5). This pseudo colouring approach could further help
y = 5.239x - 13.72 in evaluating good varieties. The changes in colour pattern,
1.25
R² = 0.942
i.e. initially blue to yellow-red, indicates that water content
1.2 decreases with seed age and this may help to predict harvest
2.85 2.86 2.87 2.88 2.89 time and seed density.
log(1000+HU)
Fig. 4. (A) Pooled linear relationship between mean log dry matter
accumulation versus log (1 000+HU value) at different growth stages of Conclusion
seed development in PAC801 and PAC807, rice varieties (n=200). (B)
Linear relationship between mean log dry matter accumulation versus log In the present study, CT scanning has been used to analyse
(1 000+HU value) in seeds of seven different rice varieties (n=175); other rice panicle development in two rice varieties that differ in
details as in Table 1. their yield potential. The HU value showed a close correlation
PAC801
Initiation Elongation Dry matter Maturation
accumulation
Stages with
colour bar
PAC807
Fig. 5. Pseudo-colouring effect on the 8-bit binary CT scan images of PAC801 and PAC807 with developmental stages by using ImageJ software.
Changes from the blue to the red scales indicate changes in water content and dry matter at the different stages of seed development. (The stronger the
blue colour, the higher the seed water content; the stronger the red colour, the higher the dry matter accumulation in the seed.)X-ray computed tomography of rice panicle | Page 7 of 7
between water content and dry matter accumulation in seeds Duan LF, Yang WN, Huang CL, Liu Q. 2011. A novel machine-vision-
based facility for the automatic evaluation of yield related traits in rice. Plant
at different developmental stages. The correlation between Methods 7, 44.
HU values from CT scans and seed development in the pani- Egli DB, Tekrony DM. 1997. Species differences in seed water status
cle suggests that this technique could be useful for farmers during seed maturation and germination. Seed Science Research 7, 3–11.
and breeders. For example, CT scanning, together with the Egli DB. 1990. Seed water relations and the regulation of seed growth in
pseudo-colour development method, can help to understand soybean. Journal of Experimental Botany 41, 243–248.
plant demand during the post-anthesis period. The HU value Grose MJ, Gilligan CA, Spencer D, Goddard BVD. 1996. Spatial
heterogeneity of soil water around single roots. Use of CT-scanning to
may be taken as a marker to improve agricultural practices predict fungal growth in the rhizosphere. New Phytologist 133, 261–272.
such as irrigation, the application of growth-promoting sub- Humplík JF, Lazár D, Husičková A, Spíchal L. 2015. Automated
stances, and developmental status at harvest time and so may phenotyping of plant shoots using imaging methods for analysis of plant
help farmers to decide when to harvest the standing crop. All stress responses: a review. Plant Methods 11, 29.
this is possible in a predictable and non-invasive manner. In Ishimaru T, Ohsugi R, Matsuda T, Yamagishi T. 2003. Morphological
development of rice caryopses located at the different positions in a
addition, it is also hoped that this technique may help variety panicle from early to middle stage of grain filling. Functional Plant Biology
selection in breeding programmes. The authors are aware of 30, 1139–1149.
the fact that the usefulness of this technique in variety selec- Kalender WA. 2000. Computed tomography. Munich: Publicis MCD
tion requires more work in terms of larger data sets of differ- Verlag.
ent varieties and may, perhaps, require some modification of Koutroubas SD, Ntanos DA. 2003. Genotypic differences for grain yield
and nitrogen utilization in Indica and Japonica rice under Mediterranean
the technique in order to develop a robust system. This will be conditions. Field Crops Research 83, 251–260.
the focus of our future work. Li X, Yan W, Agrama H, Jia L, Jackson A, Moldenhauer K, Yeater K,
Downloaded from http://jxb.oxfordjournals.org/ by guest on October 11, 2015
Maclung A, Wu D. 2012. Unravelling the complex trait of harvest index
with association mapping in rice Oryza sativa L. PLOS One 7, 1–10.
Supplementary data Luo LJ, Li ZK, Mei HW, Shu QY, Tabien R, Zhong DB, Ying CS,
Stansel JW, Khush GS, Paterson AH. 2001. Over dominant epistatic
Supplementary data are available at JXB online. loci are the primary genetic basis of inbreeding depression and heterosis in
rice II Grain yield components. Genetics 158, 1755–1771.
A schematic diagram of CT scan procedure for rice panicle.
Mohapatra P, Patel R, Sahu S. 1993.Time of flowering affects grain
Figure S1. (A) eFilmLite CT scan software, (B) W/L (win- quality and spikelet partitioning within the rice panicle. Functional Plant
dow / level) adjustment, (C) MPR (MultiPlanar Reformatting) Biology 20, 231–241.
tool and (D) Series of sections observed and Export selected Moradi AB, Carminati A, Vetterlein D, Vontobel P, Lehmann E,
image. Weller U, Hopmans JW, Vogel H-J, Oswald SE. 2011. Three-
dimensional visualization and quantification of water content in the
Figure S2. Calculation of HU value from the MPR images, rhizosphere. New Phytologist 192, 653–663.
(A) PAC801 and (B) PAC807. Rice panicle axis divided Prasad PVV, Boote KJ, Allen LH, Sheehy JE, Thomas JMG. 2006.
equally in to three part: upper, middle and lower in both the Species, ecotype and cultivar differences in spikelet fertility and harvest
varieties (individual white portion in Figure indicates seeds). index of rice in response to high temperature stress. Field Crops Research
95, 398–411.
Figure S3. Seed size measurement in eFilmLite CT soft-
Rabadia VS, Thaker VS, Singh YD. 1999. Relationship between water
ware (A) PAC801and (B) PAC 807. content and growth of seed and fiber in three cotton cultivars. Journal of
Table S1. Seven rice varieties used in study. Agronomy and Crop Science 183, 255–261.
Schluter S, Weller U, Vogel H-J. 2010. Segmentation of X-ray
microtomography images of soil using gradient masks. Computers
&Geosciences 36, 1246–1251.
Acknowledgements Schnyder H, Baum U. 1992. Growth of the grain of wheat (Triticum
The authors are very grateful to Dr Navinbhai Patel and Dr Nisha Patel, aestivum L.). The relationship between water content and dry matter
Prabhat CT Scanning Hospital, Rajkot for their technical support and to Dr accumulation. European Journal of Agronomy 1, 51–57.
PP Vaishnav for fruitful discussion. The authors also thank Vikram Farm, Wu H, Jaeger M, Wang M, Li B, Zhang BG. 2011. Three-dimensional
UPL Vapi, Gujarat, for the supply of plant materials. distributions of vessels, passage cells, and lateral roots along the root axis
of winter wheat (Triticum aestivum). Annals of Botany 107, 843–853.
Yang W, Duan L, Chen G, Xiong L, Liu Q. 2013. Plant phenomics
and high-throughput phenotyping: accelerating rice functional genomics
References using multidisciplinary technologies. Current Opinion in Plant Biology 16,
Bewley JD, Black M. 1985. Seeds: physiology of development and 180–187.
germination. New York: Plenum Press. Yang W, Xu X, Duan L, Luo Q, Chen S, Zeng S, Liu Q. 2011. High-
Borra SL, Westgate ME. 2006. Predicting maize kernel sink capacity throughput measurement of rice tillers using a conveyor equipped with
early in development. Field Crops Research 95, 223–233. X-ray computed tomography. The Review of Scientific Instruments 82,
025102.
Chanda SV, Singh YD. 1998. Cell enlargement as an important factor in
controlling grain weight in wheat. Journal of Agronomy and Crop Science Zhikang L, Pinson S, Park WD, Paterson AH, Stansel JW. 1997.
181, 223–228. Epistasis for three grain yield components in rice Oryza sativa L. Genetics
145, 453–465.You can also read
