345pm_Boote_2_CROPGRO-PFM modelling.pptx

Evaluating Cultivar and Species Traits
with the CROPGRO Perennial Forage
Model for Grasses and Legumes
K. J. Boote1, D. N. L. Pequeno2, G. Hoogenboom1, P.
Alderman3, S. Rymph4, M. Lara5, B. Pedreira6, W. Malik7,
L. Moreno6 , M. Ottman8, J. Torrion9, I. Kisekka10
1Univ. Florida, 2CIMMYT, 3Okla State Univ, 4Purina,
5Fed. Univ. Lavras-Brazil, 6EMBRAPA-Brazil, 7CITA-Spain,
8Univ. Arizona,9Univ. Montana,10Univ. California
Int’l Forage & Turf Breeding Conf, March 2019

• Introduce the perennial forage model, and how common
FORTRAN code, plus read-in species and cultivar files
can describe different species and cultivars (C-4 tropical
grasses, C-3 temperate grasses, C-3 legume/alfalfa).
• Describe physiology, storage organ features, and growth
dynamics needed for a perennial forage that allow multiple
harvests and re-growth.
• Describe required inputs, as well as outputs:
herbage mass, herbage N, crude protein, and
percent leaf of herbage for each harvest.
• Illustrate performance of different grass species and
dormancy type cultivars within alfalfa.
OBJECTIVES OF THIS TALK

CROPGRO-Perennial Forage Model
 Derived from CROPGRO model (1992-2004)
 Coded for perennial and storage organ dynamics by
Stuart Rymph, UF, 2004, & adapted for Paspalum notatum.
(See Rymph Ph.D. dissertation, UF, 2004)
 Adapted for Cynodon dactylon by K. Boote and P.
Alderman (M.S. thesis, UF, 2007)
 Adapted for Brachiaria brizantha (B. Pedreira,
2011 paper)
 Adapted for Panicum maximum by M. Lara (2012
Agron. J.)
 Contrasted traits for Marandu, Mullato, & Tifton-85 at a
single site, by D. Pequeno (2017 paper)
 Recently adapted for alfalfa (Malik et al, 2018).
 Released in DSSAT V4.7 for Brachiaria, Tifton-85,
alfalfa Int’l Forage & Turf Breeding Conf, March 2019

Code Changes for Perennial
 Created: ability to re-grow based on
reserves despite zero LAI. Creates
memory of poor prior management (low
reserves). Winter dormancy.
 Added new state variable (stolon-rhizome-
storage tissue) with TNC and N
concentration
 Added rules for partitioning DM, N, and TNC
to storage tissue as f(daylength, LAI, Ps, etc.)
 Added rules for mobilization of C and N reserves
from storage for re-growth as f(daylength, LAI,
Ps, etc.) Int’l Forage & Turf Breeding Conf, March 2019

Re-growth depends on residual LAI, storage
reserves (mass, TNC & N status), Ps recovery
 Residual LAI is
near zero, &
Canopy Ps of
Tifton 85 is low
after defoliation.
Initial re-growth
depends on TNC
and N reserves,
but Ps recovers
well by 7-14
days (Alderman
et al., 2011)

reserves (mass, TNC & N status), time of year
 Rhizome TNC
reserves of Tifton 85
show cycles of TNC
to minimum at 7-14
days, then recovery
during re-growth
after defoliation
(Alderman et al.,
2011).
 5-15% TNC, and 5-6
mt rhizome
 Rhizome N conc –
no cyclic trends

reserves (mass, TNC & N status), time of year
 Residual stem
TNC reserves of
Tifton 85 show
cyclic behavior to
minimum at 7-14
days and recovery
during re-growth
after defoliation
(Alderman et al.,
2011).
 Modest residual
stem mass &
modest TNC 4-8%.

Additional input requirements:
for “MOW” file
 Dates of harvest events:
 MOW: mass of residual living aboveground
stubble (kg/ha) after each harvest event. If
“missing”, interpolate between dates to set stubble
mass
 Percent leaf of the “living” stubble. Ignores
“dead”
 MVS: leaf # “re-stage” after each harvest
 Not tested, but possible to “graze” a given tissue
mass (kg/ha of leaf, kg/ha of stem) on specified
dates
@TRNO DATE
1 11014
2689
1 11042
2879
19
21
MOW RSPLF MVS
3
3

Additional Forage-Related Outputs
In addition to typical LAI, leaf, stem, storage,
root, and % N of these organs
 Herbage = (shoot – stubble)
 Herbage N = (shoot N – stubble N)
 Herbage N conc (~CP) = herbage N / herbage mass
 Herbage %leaf
 Output herbage, herbage N, CP, and herbage % leaf
in PLANTGRO.OUT and in a FORAGE.OUT file
 Output abscised dead shoot, leaf, and stem since
last harvest in PLANTGRO.OUT
Different from
typical
models

Stubble & Herbage, Panicum maximum
(Lara et al., 2012)
Mow=stubble
Herbage

Simulated vs. observed biomass in 2 seasons at Piracicaba,
Brazil. Brachiaria brizantha Xaraes (Pedreira et al.,
2011)
winter

Simulated vs. observed LAI in 2 seasons at Piracicaba,
Brazil. Brachiaria brizantha Xaraes (Pedreira et al., 2011)

• Harvest frequencies of 28 and 42 days, over 2 years
• 400 kg ha-1 yr-1 of N split-applied after each harvest
Simulation of irrigated Marandu (Brachiaria) at
Piracicaba, Brazil. (Pequeno et al., 2013)

Simulated leaf, stem, shoot of irrigated Marandu
(Brachiaria) at Piracicaba, Brazil. (Pequeno et al., 2013)
Leaf
Shoot
Stem

Simulated leaf mass, storage mass, and storage TNC
for irrigated Marandu (Brachiaria) at Piracicaba, Brazil.
(Pequeno et al., 2013)
Storage
Root
Leaf
TNC %
Store & root started
from seed. Stable
dynamics yr 2 & 3

• Lfmax, light saturated Ps (C-3 or C-4)
• Shape of Ps response to leaf N concentration
• Cardinal temperatures for processes (Ps, leaf area exp,
leaf appearance rate, reproductive, senesence, aging)
• N concentration of tissues (luxury, critical, minimum)
• Rate of leaf appearance
• Specific leaf area
• Partitioning to leaf, stem, root, storage as f(thermal-Vstg)
in both seedling and established regrowth mode
• Daylength sensitivity to enhance partition to storage
versus shoot growth (dormancy)
• Re-fill or rate of mobilization from storage for
regrowth
• Rate of N mobilization from older vegetation
Example – species/cultivar genetic parameters

CP of tissues for
normal critical, luxury,
and senescent tissue
Tifton 85 – higher CP
Cardinal temperatures
(Tb, Top1, Top2, Tfail
for vegetative
development rate,
Tifton 85 – lower Tb

Maximum rate of CHO and N mobilization from storage,
faster for Tifton 85. It allocates more Ps to refill (0.493
maximum of daily). Mobilization is most rapid at low LAI
(.40 to .51), and slowest at LAI of 2.71 to 3.30.

Despite different
species traits,
herbage production
was very similar.
Tifton 85 and
Mullato had
somewhat more
herbage than
Marandu, in Brazil
Tifton 85 – Partitioning to storage (causes less herbage) is more
sensitive to daylength (8, 13 h) than Marandu (7.8, 12 h), and has
higher sensitivity coefficient (0.538 versus 0.475)

MOW input file
www.themegallery.com
 DATES: Harvest dates,
 MOW = stubble mass: the amount of live
forage mass remaining 1000 kg/ha,
 RSPLF: percentage leaf of the stubble 
20%
 MVS: a “re-staged” leaf number  2
Adapting CROPGRO-PFM for Alfalfa
Simulation methods used
 The Penman-Monteith FAO 56
 The CENTURY SOC model
 Leaf photosynthesis mode
Collected data (6 farmer
fields):
• Leaf area index (LICOR-LAI-
2000) weekly
• Herbage DM
• Crude protein
(harvested herbage)
• Crop management (tillage,
fertilization and irrigation
management.
Location: Aragon Spain
Wafa Malik, PhD sandwich
student during visit to UF
Adaption Process
 Set lower Cardinal Temperatures V & R dev,
Ps, LAI expansion, nodule growth, N-fix
rate
 Set tissue composition
 Set Critical N conc for Ps like soybean
 Set partitioning: seedling phase, establ phase

0
1500
3000
4500
Jul-15 Jan-16 Aug-16 Mar-17 Sep-17
Herb.
kg/ha
Obs. 3281
Sim. 3040
RMSE 594
d-Stat. 0.82
20
15
10
5
0
www.themegallery.com
25
30
Sep-15 Apr-16 Oct-16 May-17 Nov-
17
Date (m-yy)
Mean
Herb.
CP %
Obs. 22
Sim. 18
RMSE 5
d-Stat. 0.28
LAI
Obs. 2.83
Sim. 2.49
RMSE 2.13
d-Stat. 0.73
Tops
kg/ha
Obs. 4281
Sim. 3948
RMSE 642
d-Stat. 0.8
Tops (kg ha-1)
Herbage (kg ha-1)
Herbage Crude Protein (%)
LAI
209-A case (1 of 6
fields)

Example – cultivar classes within alfalfa (NIFA grant)
I. Kisseka, M. Ottman, J. Torrion, K. Boote, G. Hoogenboom
M. Ottman,
Tucson,AZ
Fall
dormancy Entry
3 Rugged
4 Magnum 7
5 PGI 557
6 Cisco II
7 SW 7410
8 Pacifico
9 CUF 101
J. Torrion,
Creston, MT
Fall
dormancy Entry
2 Maxi Graze
2 FSG229CR
3 Rugged
3 Big Sky Ladak
6 Cisco II
6 Hi gest 660
Weekly samples on LAI, leaf, stem, total biomass between harvests for
7 cuttings in AZ and 2 cuttings in MT, on Rugged, Cisco II, and CUF101

Dormancy class: Rugged (red, 3), Cisco II (yellow, 6),
CUF101 (green, 9), with default daylength 11.1-12.2,
RDRMT 0.421 of Aragon, simulated at Tucson, AZ (Ottman)
Crop
Mass,
kg/ha
No differentiation

Dormancy class: Rugged (red, 3), Cisco II (yellow, 6),
CUF101 (green, 9), with new daylength FNPTD 9.8 14.2 h,
RDRMT: 0.500 Rugged, 0.320 Cisco II 0.140 CUF101 (AZ)
Crop
Mass,
kg/ha

• Lfmax, light saturated Ps (C-3 or C-4)
• Shape of Ps response to leaf N concentration
• Cardinal temperatures for processes (Ps, leaf area exp,
leaf appearance rate, reproductive, senesence, aging)
• N concentration of tissues (luxury, critical, minimum)
• Rate of leaf appearance
• Specific leaf area
• Partitioning to leaf, stem, root, storage as f(thermal-Vstg)
in both seedling and established regrowth mode
• Daylength sensitivity to enhance partition to storage
versus shoot growth (dormancy)
• Re-fill or rate of mobilization from storage for
regrowth
• Rate of N mobilization from older vegetation
Crop Model potentials – species/cultivar parameters

345pm_Boote_2_CROPGRO-PFM modelling.pptx

More Related Content

Similar to 345pm_Boote_2_CROPGRO-PFM modelling.pptx (20)

More from Fantahun Dugassa (20)

Recently uploaded (20)

345pm_Boote_2_CROPGRO-PFM modelling.pptx