Can crop sensing be used for mapping stripe rust resistance loci in wheat?
Jul 12, 2015
Can crop sensing be used for mapping stripe rust resistance loci in wheat?
• Wheat rust research relies heavily on accurate phenotyping
–Pathogen variability
• Virulent or avirulent
–Host reponse
• Resistant or susceptible
• R gene phenotype
Wheat Rust Phenotyping
• Cobb scale (Cobb, 1892)
– Percentages are equal to actual leaf area covered by rust
Diagrammatic Scales for Rust Assessment
1% 5% 10% 20% 50%
• Modified Cobb scale I (Melchers & Parker, 1922)
• 100% disease severity = 37% area covered by pustules
Diagrammatic Scales for Rust Assessment
5% 10% 25% 40% 65% 100%
• Modified Cobb scale II (Peterson et al., 1948)
– Retained 100% disease severity = 37% area covered
– Used a planimeter to measure
area of “pustules”
– Introduced• more equally spaced intervals
• 4 sets of different pustule sizes
– Developed for:• Puccinia graminis
• P. rubigo-vera
• P. hordei
• P. coronata
Diagrammatic Scales for Rust Assessment
“The writers do not consider these diagrams suitable for stripe rust …” Peterson et al. (1948)
Diagrammatic Scales for Rust Assessment
• Is a hand-held crop sensor sensitive enough to phenotype wheat populations in mapping stripe rust resistance QTL?
Research Question
• Population
– Francolin#1 x Avocet-YrA (developed at CIMMYT)
– Francolin#1 is a spring wheat line with pedigree Waxwing2*/Vivitsi
• Locality
– Redgates Research Station, Pannar, Greytown, South Africa
• Plot layout
– 198 F5 RIL entries planted in 1 m rows spaced 75 cm apart
– Two replications of Francolin#1 and Avocet-YrA included
– JIC871 served as susceptible check at regular intervals
• Natural infection by Pst race 6E22A+
– Experiment was part of a larger stripe rust nursery with spreaders and sufficient disease pressure
Materials and Methods
• 4 October 2013:
–Visual disease severity and host response
• Modified Cobb scale (0-100%)
• R > RMR > MR > MRMS > MS > MSS > S – (0.1 - 0.7 transformation)
–NDVI (scan 1)
• 10 October 2013:
–NDVI (scan 2)
Disease Assessment
• NDVI (Pask et al. 2012 – CIMMYT Field Guide)
–Normalized Difference Vegetation Index • Calculated from measurements of light reflectance in
the red and near-infrared regions of the spectrum
• Regularly used in crop canopy characterisation
– Leaf area index, biomass, nutrient status
– Healthy green leaves absorb most of the red light and reflect most of the NIR light
– NDVI = (RNIR – RRed) / (RNIR + RRed)
Disease Assessment
Trimble GreenSeeker™ (model HCS-100) crop sensor
• Relationships between
– Severity and host response
–NDVI and severity
–NDVI and host response
–Used means per response class
• Population reduced to 180 (eliminating mixtures)
Statistical Analysis
• 141 RILs were genotyped with 581 DArT, SSR markers
• Phenotyped in Mexico and China
QTL Mapping
• Uniform and severe stripe epidemic prevailed
• Avocet-YrA = 100S
• Francolin#1 = TR
Results
Entry 1622 Entry 1620
Entry 1586
R
RMR
MR
MRMSMS
MSS
S
Y = 0.0063X+0.0483R² = 0.97
0.00
0.20
0.40
0.60
0.80
0 20 40 60 80 100
Stri
pe
ru
st r
esp
on
se t
ype
Stripe rust severity (%)
• First scan
0.36 to 0.76• Avocet = 0.48
• Francolin#1 = 0.67
• Second scan
0.34 to 0.79• Avocet = 0.45
• Francolin#1 = 0.72
NDVI Range
4 OctY = 439.66-614.36X
R² = 0.95
10 OctY = 259.10-348.84X
R² = 0.99
0
20
40
60
80
100
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80
Stri
pe
ru
st s
eve
rity
(%
)
NDVI
4 OctY =3.11-4.26X
R² = 0.93
10 OctY = 1.87-2.47X
R² = 0.99
0.00
0.20
0.40
0.60
0.80
1.00
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80
Stri
pe
ru
st r
esp
on
se t
ype
NDVI
QTL Mapping
• Lan et al. (2014)
–Mean Final Disease Scores identified QTL on
• 1BL (Francolin#1)
• 2BS (Francolin#1)
• 3BS (Francolin#1)
• 6AL (Avocet)
QTL Mapping with SA Data
Trait name QTL Position Left marker Right marker LOD PVE(%) AddResistance
source
YRDS QYr.cim-1BL 20 wPt-1770 wPt-9028 11.61 18.67 14.37 Francolin#1
YRDS QYr.cim-2BS 77 YrF wmc474 12.87 27.88 17.56 Francolin#1
YRDS QYr.cim-3BS 30 wPt-741331 wPt-741750 3.37 5.19 7.71 Francolin#1
YRDS QYr.CIM-6AL 31 gwm356.1 wPt-743476 3.63 5.52 -8.19 Avocet
SCAN1 QYr.cim-1BL 8 csLV46 gwm140 3.48 13.7 -2.24 Francolin#1
SCAN1 QYr.cim-2BS 70 wPt-6174 wmc344 7.3 14.09 -2.2 Francolin#1
SCAN1 QYr.cim-3BS 29 wPt-0302 wPt-741331 3.1 5.03 -1.35 Francolin#1
SCAN1 QYr.CIM-6AL 35 wPt-743476 wPt-744881 2.88 4.57 1.26 Avocet
SCAN2 QYr.cim-1BL 20 wPt-1770 wPt-9028 5.43 7.85 -2.89 Francolin#1
SCAN2 QYr.cim-2BS 67 barc55 wPt-8548 11.21 22.3 -4.88 Francolin#1
SCAN2 QYr.CIM-6AL 29 wPt-741026 gwm356.1 4.37 6.7 2.83 Avocet
• GreenSeeker™ technology and visual disease severity scores identified the same chromosome regions
• Data were comparable with published mapping studies using multi-location phenotyping of the same population
• Less variation was explained by NDVI data
Conclusions
• Some differences in marker regions occurred for the respective traits
• Timing of assessments is important considering the optimal expression windows of different QTL
• A uniform epidemic is required
• Take measurements at same time of day and during similar weather conditions
Conclusions
• Standardise procedures for distance, angle, trigger time, number of samples per entry
• Non-subjective crop sensing is suitable for detecting stripe rust resistance loci in the field– Works well for Pst where total leaf damage is most
indicative of host response
– More experiments with different populations will be conducted in 2014
Conclusions
– Ravi Singh – FxA population
– Caixia Lan – mapping
– Cornel Bender – disease scores
– Neal McLaren – statistical analyses
– Rikus Kloppers and Vicky Knight – field facilities and NDVI data
Acknowledgements