|
ON-FARM TESTING OF THE PNUTGRO
CROP MODEL IN FLORIDA
By
ROBERT A. GILBERT
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
1992
ON-FARM TESTING OF THE PNUTGRO
CROP MODEL IN FLORIDA
By
ROBERT A. GILBERT
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
1992
ACKNOWLEDGEMENTS
I would like to give my heartfelt thanks to Dr. Kenneth
J. Boote, my advisor and chairman, and Dr. Jerry M. Bennett
for their help both in the lab and in the field. I doubt
that I would have completed this study without their
knowledge and encouragement.
I am deeply grateful to Dr. Clif Hiebsch and Dr. Peter
Hildebrand for their constructive suggestions as my
committee members and for the knowledge I acquired in their
classes.
Special thanks to Dr. Ken Buhr for his friendship.
I am indebted to Anthony Drew and Ed Jowers, who took
time out of their busy extension schedules to liaise with
farmers and drive me around their counties.
Diane Hornby, Ed Blazey and Niel Hill provided
invaluable technical help (and sweat) in the field for two
years.
Finally, to my wife Ami, I give my deepest gratitude.
Without her inspiration, I would not have the good fortune
to be a graduating Gator.
TABLE OF CONTENTS
page
ACKNOWLEDGEMENTS . . . .. ii
LIST OF TABLES . . . . ... v
LIST OF FIGURES . . . . ... vii
ABSTRACT . . . . . xiv
INTRODUCTION . . . 1
REVIEW OF LITERATURE . . . . 4
Peanut Growth Dynamics .... . . 4
Environmental Influences on Peanut Growth and Yield 6
Abiotic Factors. .. . . . 6
Water, solar radiation and temperature 6
Soil fertility .. . . 11
Biotic Factors . . . . 13
Diseases . . . 13
Nematodes . . . . 17
Insects . . .. 19
Crop Growth Modelling. . ... .. .20
Purpose of Crop Modelling . . 20
Peanut Model Development . . 20
Model Calibration . . . 23
Model Validation . . .. ... 24
MATERIALS AND METHODS .. . . . .. 26
Site Selection . ... . . 26
Soil Fertility and Nematode Tests. . . 28
Soil Water Sampling . . . . 29
Weather Data Collection. . . .. .30
Plant Growth Sampling . . . .. 30
Final Harvest Sampling ............ .32
Data Entry . . . . 33
RESULTS AND DISCUSSION . . . . 36
1990 On-Farm Experiments . . . 36
Levy County ............. 36
Jackson County .. . . . .. 58
iii
1991 On-Farm Experiments . . . .. .83
Levy County . . . ... .83
Jackson County . . 102
Accounting for Pest and Disease Effects
in PNUTGRO . . . . 125
SUMMARY AND CONCLUSIONS . . . .155
APPENDIX . . . . 158
BIBLIOGRAPHY . . . ... .. 191
BIOGRAPHICAL SKETCH . . . ... .199
LIST OF TABLES
Table page
1. Growth stage descriptions for peanut
(from Boote, 1982) . . . 5
2. Field sites selected for sampling in
Levy and Jackson Counties in 1990 and
1991 . . . . . 27
3. Seasonal water received, and average daily
solar radiation and temperature for all
field sites in 1990 and 1991 . ... .39
4. Field-observed (OBS.) vs. PNUTGRO-simulated
(SIM.) growth and yield data from Levy County
in 1990 . . . . .. .47
5. Field-observed (OBS.) vs. PNUTGRO-simulated
(SIM.) growth and yield data from Jackson County
in 1990 . . . . . 68
6. Field-observed (OBS.) vs. PNUTGRO-simulated
(SIM.) growth and yield data from Levy County
in 1991 . . . ... 93
7. Field-observed (OBS.) vs. PNUTGRO-simulated
(SIM.) growth and yield data from Jackson County
in 1991 . .. . . . 113
8. Observed and simulated pod yields, root-knot
nematode levels, and other pest and disease
incidence in Levy and Jackson County in
1991 . . . .... .. 114
9. The 1991 observed (OBS.) and PNUTGRO-simulated
(SIM.) root-knot nematode levels, and adjusted
pod yields based on the 1979, 1980 and 1981
Rodriguez-Kabana (1982c) regression equations.
The loss functions were either normalized
(NORM.), or the loss function slope was
assumed to be correct (SLOPE) . ... 130
10. PNUTGRO-simulated pod yields, observed pod
yields, root-knot nematode levels, and
adjusted pod yields obtained by covariate
analysis, for all sites in 1991 . .. ..134
11. PNUTGRO-simulated (SIM.) and observed (OBS.) pod
yields, root-knot nematode levels, and adjusted
Rodriguez-Kabana regression equations and
PNUTGRO pod yields for all 1991 sites with
high root-knot nematode levels. Estimated
assimilate removed by nematodes was used to
calculate cumulative carbohydrate consumed 140
A-1. Soil fertility test results for all field sites
in Jackson and Levy Counties in 1991 . 159
A-2. Genetic coefficients used for peanut genotypes
at all field sites in Levy and Jackson Counties
in 1990 and 1991. Coefficients are defined in
Boote et al. (1989a) . . .. 163
A-3. Soil profile properties at all field sites in
Levy and Jackson Counties in 1990 and 1991.
Soil profile properties are defined in
IBSNAT (1990) . . . . 164
A-4. Daily water received at all operating raingauge
sites in Levy and Jackson Counties in 1990 166
A-5. Daily solar radiation and temperature data
from the Graham farm weather station in Levy
County in 1990 . . . .. 170
A-6. Daily solar radiation and temperature data from
the Morgan rainfed farm weather station in
Jackson County in 1990 . . .. 174
A-7. Daily water received at all operating rain-
gauge sites in Levy and Jackson Counties
in 1991 . . . . .. 178
A-8. Daily solar radiation and temperature data
from the Sandlin and Graham farm weather
stations in Levy County in 1991 ...... 182
A-9. Daily solar radiation and temperature data
from the Morgan rainfed farm weather
station in Jackson County in 1991 . .. 187
LIST OF FIGURES
Figure
page
1. Cumulative water received at all
Levy County field sites in 1990 . ... .38
2. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers at
the Brookins farm in Levy County in 1990 . 40
3. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers at the
Graham farm in Levy County in 1990 ...... 41
4. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Lowman farm in Levy County in 1990 42
5. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Sandlin farm in Levy County in 1990
6. Maximum and minimum daily temperature
recorded at the weather station on the
Graham farm in Levy County in 1990 .
7. Daily solar radiation recorded at
the weather station on the Graham farm
in Levy County in 1990 . . .
* 43
. 45
. 46
8. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the
Sandlin farm in Levy County in 1990 . .
9. Measured (points) and PNUTGRO-simulated (line)
pod harvest index at the Sandlin
farm in Levy County in 1990 . . .
10. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the
Sandlin farm in Levy County in 1990 . .
vii
11. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the
Lowman farm in Levy County in 1990 . ... .52
12. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the
Lowman farm in Levy County in 1990 . ... .53
13. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the
Brookins farm in Levy County in 1990 . 54
14. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the
Graham farm in Levy County in 1990 . ... 55
15. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the
Brookins farm in Levy County in 1990 . .. .56
16. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the
Graham farm in Levy County in 1990 . .. 57
17. Cumulative water received at the Crawford
and Morgan rainfed sites in Jackson County
in 1990 . . . .... .. 60
18. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Crawford rainfed farm in
Jackson County in 1990 . . ... 61
19. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Morgan rainfed farm in
Jackson County in 1990 . . . 62
20. Simulated total extractable soil water
for the Brookins (BR), Crawford rainfed (CR R)
and Morgan rainfed (MO R) farms in 1990 . 64
21. Maximum and minimum daily temperature
recorded at the weather station on the
Morgan farm in Jackson County in 1990 . 65
22. Daily solar radiation recorded at
the weather station on the Morgan rainfed
farm in Jackson County in 1990 . ... .66
viii
23. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Crawford
irrigated farm in Jackson County in 1990 . 69
24. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Crawford
irrigated farm in Jackson County in 1990 . 70
25. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Crawford
rainfed farm in Jackson County in 1990 ... .71
26. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Crawford
rainfed farm in Jackson County in 1990 ... .72
27. Measured (points) and PNUTGRO-simulated (line)
pod harvest index at the Crawford
rainfed farm in Jackson County in 1990 ... .74
28. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Morgan
irrigated farm in Jackson County in 1990 . 75
29. Measured (points) and PNUTGRO-simulated (line)
pod harvest index at the Morgan
irrigated farm in Jackson County in 1990 . 76
30. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Morgan
irrigated farm in Jackson County in 1990 . 77
31. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Morgan
rainfed farm in Jackson County in 1990 ... .78
32. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Morgan
rainfed farm in Jackson County in 1990 . 79
33. Relationship between observed and simulated pod
yield for all sites in 1990 . ... .81
34. Relationship between observed and simulated
biomass at harvest maturity (R8)
for all sites in 1990 . . ... .82
35. Cumulative water received (rainfall plus
irrigation) at all Levy County field sites
in 1991 . . . .... .. 84
36. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from the 5-15 and 15-30 cm soil layers
at the Graham farm in Levy County in 1991 85
37. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from the 5-15 and 15-30 cm soil layers
at the Lowman farm in Levy County in 1991 86
38. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from the 5-15 and 15-30 cm soil layers
at the Sandlin farm in Levy County in 1991 87
39. Daily solar radiation recorded at
the weather station on the Sandlin farm
in Levy County in 1991 . . ... .89
40. Daily solar radiation recorded at
the weather station on the Graham farm
in Levy County in 1991 . . ... .90
41. Maximum and minimum daily temperature
recorded at the weather station on the
Sandlin farm in Levy County in 1991 . .. .91
42. Maximum and minimum daily temperature
recorded at the weather station on the
Graham farm in Levy County in 1991 ...... 92
43. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the
Graham farm in Levy County in 1991 ...... 95
44. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Graham
farm in Levy County in 1991 . . 96
45. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Lowman
farm in Levy County in 1991 . ... .97
46. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Lowman
farm in Levy County in 1991 . ... .98
47. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Sandlin
farm in Levy County in 1991 . . .. 100
48. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Sandlin
farm in Levy County in 1991 . . .
. 101
49. Cumulative water received at the Crawford
rainfed, Crawford irrigated, and Morgan
rainfed sites in Jackson County in 1991 . 103
50. Cumulative water received at the Graham
and Morgan rainfed sites in 1990 and 1991 104
51. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Crawford rainfed farm in
Jackson County in 1991. . . ... 106
52. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Crawford irrigated farm in
Jackson County in 1991 . . .. 107
53. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Morgan rainfed farm in
Jackson County in 1991 ... . . 108
54. Observed (points) and PNUTGRO-simulated
(lines) volumetric soil water content
from 5-15 and 15-30 cm soil layers
at the Morgan irrigated farm in
Jackson County in 1991 . . ... .109
55. Daily solar radiation recorded at
the weather station on the Morgan
rainfed farm in Jackson County in 1991 . 110
56. Maximum and minimum daily temperature
recorded at the weather station on the
Morgan rainfed farm in Jackson County in
1991 . . . ... . 111
57. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Crawford
irrigated farm in Jackson County in 1991 . 115
58. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Crawford
irrigated farm in Jackson County in 1991 . 116
59. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Crawford
rainfed farm in Jackson County in 1991 .... .118
60. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Crawford
rainfed farm in Jackson County in 1991 .... .119
61. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Morgan
irrigated farm in Jackson County in 1991 . 120
62. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Morgan
irrigated farm in Jackson County in 1991 . 122
63. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Morgan
rainfed farm in Jackson County in 1991 .... .123
64. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Morgan
rainfed farm in Jackson County in 1991 .... .124
65. Relationship between observed and simulated pod
yield for all sites in 1991 . ... 126
66. Relationship between observed and simulated
biomass at harvest (R8) for all sites
in 1991 . . . 127
67. Measured (points) and PNUTGRO-simulated (lines)
leaf area index (LAI) with and without
simulated leaf senescence at the Graham farm
in Levy County in 1991 . . ... .136
68. Measured (points) and PNUTGRO-simulated (lines)
leaf weight with and without simulated leaf
senescence at the Graham farm in Levy
County in 1991 . . . ... 137
69. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights with and without simulated
nematode damage at the Lowman farm in Levy
County in 1991 . . . ... 142
70. Measured (points) and PNUTGRO-simulated (lines)
leaf area index (LAI) with and without simulated
nematode damage at the Lowman farm in Levy
County in 1991 . . . ... 143
xii
71. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights with and without simulated
nematode damage at the Morgan rainfed farm in
Jackson County in 1991 . . .. 144
72. Measured (points) and PNUTGRO-simulated (lines)
leaf area index (LAI) with and without simulated
nematode damage at the Morgan rainfed farm in
Jackson County in 1991 . . ... .145
73. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights with and without simulated
nematode damage at the Morgan irrigated farm in
Jackson County in 1991 . . ... .146
74. Measured (points) and PNUTGRO-simulated (lines)
leaf area index (LAI) with and without simulated
nematode damage at the Morgan irrigated farm in
Jackson County in 1991 . . ... .147
75. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights with and without simulated
nematode damage at the Crawford rainfed farm in
Jackson County in 1991 . . ... .148
76. Measured (points) and PNUTGRO-simulated (lines)
leaf area index (LAI) with and without simulated
nematode damage at the Crawford rainfed farm in
Jackson County in 1991 . . .. 149
77. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights with and without simulated
nematode damage at the Crawford irrigated farm in
Jackson County in 1991 . . ... .150
78. Measured (points) and PNUTGRO-simulated (lines)
leaf area index (LAI) with and without simulated
nematode damage at the Crawford irrigated farm in
Jackson County in 1991 . . ... .151
79. Relationship between observed and simulated pod
yield adjusted for pest and disease damage for
all sites in 1991 . . . .. .154
xiii
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
ON-FARM TESTING OF THE PNUTGRO
CROP MODEL IN FLORIDA
By
ROBERT A. GILBERT
August, 1992
Chairman: Kenneth J. Boote
Major Department: Agronomy
The PNUTGRO crop growth model simulates peanut growth
and yield using weather, soil and genetic inputs. PNUTGRO
was developed from growth measurements taken in research
plots. This thesis research was conducted to evaluate the
PNUTGRO model under actual farm conditions, in order to
determine PNUTGRO's utility for farmers and extension
agents, and to identify areas for model improvement.
Field testing of the PNUTGRO model was conducted at a
total of 15 farm sites in Jackson and Levy Counties in 1990
and 1991. Rainfall and irrigation were recorded in each
field daily, as were solar radiation and air temperatures in
each county. Plant growth and soil water samples were
collected every 3 to 4 weeks, and used with final harvest
samples to evaluate PNUTGRO simulations.
The accuracy of the PNUTGRO model simulations varied
xiv
between years and locations. The production system of Levy
County growers included long-term bahiagrass rotations of 2-
8 years between peanut crops. The farms studied in Levy
County achieved the yield potential predicted by PNUTGRO in
1990 (8.9% average difference from observed). The farming
system of Jackson County included row-crop rotations between
peanut crops. The yields on these farms did not reach the
yield potential predicted by PNUTGRO (29.5% difference in
1990, 43.8% in 1991).
PNUTGRO model simulations were more accurate in 1990
than in 1991 in both counties. Greater rainfall in 1991 led
to higher pest and disease incidence (especially root-knot
nematode, stem rot and leafspot) which the PNUTGRO model
does not account for.
It appears that the present version of PNUTGRO is most
useful as a predictor of potential peanut yield under given
weather conditions. The greater the level of pest and
disease pressure on peanut growth, the less accurate the
PNUTGRO predictions became (e.g., Jackson County, 1991).
The use of regression equations or pest-coupling models
to account for pest and disease effects on peanut growth
improved PNUTGRO model fit. Further research on the effects
of pests and diseases on peanut growth and yield is needed.
Incorporation of pest and disease damage into the PNUTGRO
model would increase its predictive capability on farms in
which potential peanut yield is not being attained.
INTRODUCTION
Peanut (Arachis hvpocaea L.) is a native South American
legume, which was probably introduced to the United States
via Africa (Hammons, 1982). Commercialization of peanut
production did not begin until the onset of European oil
shortages in the 1800s, with U.S. production increasing with
oil needs during the Civil War. Increasing boll weevil
pressure on cotton (Gossypium hirsutum) production in the
early 1900s, and the introduction of a fixed production area
and the peanut combine in 1948, helped contribute to
intensification of peanut production and increasing yields
per acre in the U.S. (Hammons, 1982).
However, peanut yields have fluctuated greatly in the
southeastern U.S. during the last decade due to drought.
Average pod yields from Florida, Georgia and Alabama have
ranged from 2110 kg ha'1 in 1990 to 3560 kg ha1' in 1984
(USDA, 1988; USDA, 1991). Computer growth simulation models
may be able to help account for yield fluctuations due to
weather variation (Boote and Jones, 1988). These models may
also be useful to the grower in making management decisions
such as irrigation scheduling or the acquisition of
irrigation equipment.
Peanut growth is influenced by many complex
environmental factors. Abiotic influences include weather
(water, solar radiation and temperature) and soil fertility.
Biotic factors which influence peanut growth include
diseases, insects and nematodes.
The PNUTGRO crop growth simulation model simulates
peanut growth and yield in response to weather, soil and
genetic inputs (Boote et al., 1986). It does not presently
respond to soil fertility or the biotic factors mentioned
above.
Crop simulation models such as PNUTGRO can serve as
aids in agricultural research or crop management (Whisler et
al., 1986). However, they must be evaluated using data sets
independent of their development or calibration (Burk,
1986). This testing requires growth analysis throughout the
crop growing season; final yield data alone are inadequate
for model validation (Acock and Acock, 1991).
The PNUTGRO model was first evaluated on peanut farms
in North Florida in 1988 (Boote et al., 1988). The model
tended to overpredict peanut growth and yield on farms with
disease and nematode problems. A version of the PNUTGRO
model calibrated to the 1988 North Florida data predicted
growth and yield reasonably well in the 1989 growing season.
The objectives of this study were to: 1) evaluate the
PNUTGRO model under actual farm conditions in two Florida
counties, 2) determine the utility of PNUTGRO to farmers and
3
extension agents, and 3) identify areas for model
improvement. Independent crop and soil data sets from
Florida peanut fields were gathered to compare with PNUTGRO
simulated values.
REVIEW OF LITERATURE
Peanut Growth Dynamics
Boote (1982) developed a uniform growth stage
designation for peanut based on visually observed vegetative
(V) and reproductive (R) events (see Table 1). Peanut dry
matter accumulation follows a sigmoidal pattern with maximum
crop growth rate (CGR) occurring at 60-90 days after
planting (DAP). Maximum CGR from 25 experiments was 20 + 4
g m'2 day'1 (Ketring et al., 1982). Leaf and stem weights
increase until 90-100 DAP, when leaf weights may begin to
decline while stem weight remains constant. Pod number and
weight become measurable 60-70 DAP. After a lag phase, pod
weight increases linearly during pod filling until harvest.
Pod growth rates for 24 experiments averaged 8 + 2 g m-2
day '1 (Ketring et al., 1982).
Peanut genetic yield potential has doubled over the
last 40 years due to an increase in the partitioning
coefficient, greater length of pod filling period and more
rapid fruit establishment (Duncan et al., 1978). Crop
growth rate has not changed significantly, but partitioning
of assimilate to reproductive parts increased from 41% in
old cultivars to 98% for more recent cultivars. Duncan et
al. (1978) presented data on four cultivars introduced
REVIEW OF LITERATURE
Peanut Growth Dynamics
Boote (1982) developed a uniform growth stage
designation for peanut based on visually observed vegetative
(V) and reproductive (R) events (see Table 1). Peanut dry
matter accumulation follows a sigmoidal pattern with maximum
crop growth rate (CGR) occurring at 60-90 days after
planting (DAP). Maximum CGR from 25 experiments was 20 + 4
g m'2 day'1 (Ketring et al., 1982). Leaf and stem weights
increase until 90-100 DAP, when leaf weights may begin to
decline while stem weight remains constant. Pod number and
weight become measurable 60-70 DAP. After a lag phase, pod
weight increases linearly during pod filling until harvest.
Pod growth rates for 24 experiments averaged 8 + 2 g m-2
day '1 (Ketring et al., 1982).
Peanut genetic yield potential has doubled over the
last 40 years due to an increase in the partitioning
coefficient, greater length of pod filling period and more
rapid fruit establishment (Duncan et al., 1978). Crop
growth rate has not changed significantly, but partitioning
of assimilate to reproductive parts increased from 41% in
old cultivars to 98% for more recent cultivars. Duncan et
al. (1978) presented data on four cultivars introduced
Table 1. Growth stage descriptions for peanut (from Boote,
1982).
Typical
Stage Stage DAPi for
Number Name Florunner#
VE Emergence
V1 First tetrafoliate to
Vn nth tetrafoliate -
R1 Beginning bloom 31
R2 Beginning peg 42
R3 Beginning pod 51
R4 Full pod 60
R5 Beginning seed 62
R6 Full seed 74
R7 Beginning maturity 93
R8 Harvest maturity 129
R9 Over-mature pod 123
:Days after planting
#In southeastern United States
MATERIALS AND METHODS
Site Selection
Field testing of the PNUTGRO model was conducted in two
Florida counties in 1990 and 1991. Jackson County sites
were chosen near Marianna and Greenwood, and Levy County
sites were chosen near Bronson. The agricultural extension
agents in both counties were asked to recommend peanut
farmers with representative management practices willing to
collaborate with University of Florida researchers.
Extension agents Mr. Anthony Drew of Levy County and Mr. Ed
Jowers of Jackson County initiated contact and liaised with
the peanut farmers whose fields were used.
A total of 15 field sites were used during the two year
study. In 1990, four sites were selected in Jackson County
and four in Levy County, while in 1991 there were four field
sites in Jackson County and three in Levy County. Table 2
identifies the field sites, and shows the cultivar, soil
type, planting date and irrigation use at each site. The
Crawford farm in Jackson County in 1990 and 1991 used a
center pivot irrigation system. Sites were selected under
the pivot and in a dry corner of the field, providing
rainfed and irrigated treatments in one field. The Morgan
MATERIALS AND METHODS
Site Selection
Field testing of the PNUTGRO model was conducted in two
Florida counties in 1990 and 1991. Jackson County sites
were chosen near Marianna and Greenwood, and Levy County
sites were chosen near Bronson. The agricultural extension
agents in both counties were asked to recommend peanut
farmers with representative management practices willing to
collaborate with University of Florida researchers.
Extension agents Mr. Anthony Drew of Levy County and Mr. Ed
Jowers of Jackson County initiated contact and liaised with
the peanut farmers whose fields were used.
A total of 15 field sites were used during the two year
study. In 1990, four sites were selected in Jackson County
and four in Levy County, while in 1991 there were four field
sites in Jackson County and three in Levy County. Table 2
identifies the field sites, and shows the cultivar, soil
type, planting date and irrigation use at each site. The
Crawford farm in Jackson County in 1990 and 1991 used a
center pivot irrigation system. Sites were selected under
the pivot and in a dry corner of the field, providing
rainfed and irrigated treatments in one field. The Morgan
Table 2. Field sites selected for study in Levy
YEAR
1990
1991
COUNTY FARM
LEVY Brookins
Graham
Lowman
Sandlin
JACKSON Crawford
Crawford
Morgan
Morgan
LEVY Graham
Lowman
Sandlin
JACKSON Crawford
Crawford
Morgan
Morgan
I.D.
BR 90
GR 90
LO 90
SA 90
CR R 90
CR I 90
MO R 90
MO I 90
GR 91
LO 91
SA 91
CR R 91
CR I 91
MO R 91
MO I 91
CULTIVAR
Florunner
Florunner
Sunrunner
Sunrunner
Agritech-127
Agritech-127
Florunner
Florunner
Marc I
Florunner
Sunrunner
Agritech-127
Agritech-127
Florunner
Florunner
and Jackson Counties in 1990 and 1991.
PLANTING
IRRIGATION SOIL SUBGROUP DATE
Yes Grossarenic May 5
Paleudalfs
Yes Grossarenic May 7
Paleudalfs
No Arenic April 18
Hapludalfs
Yes Typic April 14
Quartzipsamments
No Grossarenic April 27
Paleudults
Yes Grossarenic April 27
Paleudults
No Typic April 26
Paleudults
Yes Typic April 25
Paleudults
No Grossarenic May 26
Paleudalfs
No Arenic May 3
Hapludalfs
No Typic May 1
Quartzipsamments
No Grossarenic April 20
Paleudults
Yes Grossarenic April 20
Paleudults
No Typic April 20
Paleudults
Yes Typic April 30
Paleudults
6
sequentially in the Southeast. Partitioning percent and pod
yield (kg ha'1) were: Dixie Runner (41%/2472), Early Runner
(76%/3843), Florunner (85%/4642) and Early Bunch (98%/5378).
Increasing partitioning to reproductive parts in newer
cultivars has led to higher yields, but also to a smaller
vegetative canopy at harvest, requiring increased attention
to insect and disease control (Duncan et al., 1978; McGraw,
1979).
Environmental Influences on Peanut Growth and Yield
Peanut growth is influenced by many complex
environmental factors. Abiotic influences include weather
(water, solar radiation and temperature) and soil fertility.
Biotic factors which influence peanut growth include
diseases, nematodes and insects.
Abiotic Factors
Water, solar radiation and temperature
The drought tolerance of peanut is well documented. A
comparison of four grain legumes under season-long moisture
stress at the International Rice Research Institute (IRRI)
in the Phillipines showed that peanut had greater water use
and smaller seed yield reduction (46%) than cowpea (Vicna
unquiculata, 65%), soybean (Glycine max, 66%) or mungbean
(Vigna radiata, 83%). Water use was positively correlated
to yield (Pandey et al., 1984a). Peanut also had the
highest leaf water potential at 60 DAP (-0.67 MPa), and
maintained a lower canopy temperature (Pandey et al.,
1984b). Higher peanut root densities at 0.4 and 0.8 m soil
depth were found to be an adaptive mechanism leading to
greater water extraction and leaf water potential (Pandey et
al., 1984c). Maliro (1987) also found that peanut yield was
greater than that of soybean, pigeonpea (Caianus caian) and
cowpea in both rainfed and irrigated conditions. Peanut had
the highest crop growth rate and longest effective seed
filling period of the four legumes.
Further experiments conducted at IRRI with peanut,
cowpea, soybean, mungbean and pigeon pea compared responses
under soil water deficit imposed during the reproductive
growth phase (Senthong and Pandey, 1989). Peanut had the
highest overall yield in both irrigated and rainfed
conditions. Seed yield increased linearly with water
applied (3.5-5.7 kg ha'' mm'1 for peanut) for all legumes
except pigeon pea. Peanut again had the greatest water
extraction. Differences in drought susceptibility were
thought to be primarily associated with the ability of
peanut to sustain root growth into deeper soil layers under
drying soil conditions.
Maximum daily water use for peanut ranges from 0.5 to
0.6 cm day"' during maximum LAI. Sixty cm of well
distributed water is generally sufficient for optimal peanut
production. Evapotranspiration will increase with
increasing solar radiation, temperature and wind speed, and
decreasing relative humidity and soil water deficit (Boote
et al., 1982).
The effect of timing of drought on peanut has been
studied extensively. Peanut appears more sensitive to
drought after pegging than during early vegetative growth
(Boote et al., 1982; Nagaswara-Rao et al., 1985; Nageswara-
Rao et al., 1988; Pallas et al., 1979; Stirling et al.,
1989a, 1989b; Stirling and Black, 1991). Florunner yields
were decreased 65% with the imposition of a 70 day late
season drought (Pallas et al., 1979). Early season drought
decreased yields only 16-33%.
Peanut pod yields were four times greater under late
irrigation (after pod initiation) versus early irrigation
only (sowing to pod initiation) (Stirling et al., 1989a).
Peanut's "plasticity of development" allowed growth rates to
resume to non-stressed levels in late irrigated stands.
Apical sink activity, leaf expansion and peg turgor were all
maintained during early drought (Stirling et al., 1989b).
Late season moisture deficits: 1) induced flower abortion,
2) reduced peg turgor and ability to penetrate soil, and 3)
restricted pod growth with inadequate assimilate production.
Timing of peg initiation was found to be sensitive to air
temperature, and thermal time was suggested as an
appropriate means to schedule initial irrigations (Stirling
et al., 1989a; Stirling and Black, 1991).
Pod initiation was delayed 16 to 33 days under severe
water stress in India. This delay between peg initiation
and pod growth was the cause of differences in harvest index
and pod yield (Stirling and Black, 1991). Peanut leaf area,
flower number, total biomass, pod yield, root weight and
leaf osmotic potential decreased as soil water potential
decreased (Patel et al., 1983).
Stansell and Pallas (1985) reported lower Florunner
yields when drought was imposed 35-105 DAP (1387 kg ha'1),
than when drought was imposed at 70-140 DAP (2592 kg ha'").
They hypothesized that the peanut roots were able to reach
deeper soil profiles after 70 days to withstand drought.
Peanut drought tolerance differs among cultivars.
Comet, a spanish type cultivar, showed a greater tolerance
to dehydration and higher pod yield under rainfed conditions
than Florunner, while Florunner had greater yield potential
under irrigated conditions (Erickson and Ketring, 1985).
Percent yield loss from drought was associated with yield
potential for 22 varieties when water deficit was applied
during seed fill in India (Nageswara-Rao et al., 1989),
implying a conflict between yield potential and earliness to
escape drought.
Robut 33 grown on an Alfisol in India showed a yield
increase of 12-18% when drought was applied from emergence
to pegging. "Greater synchrony of pod set in moderately
stressed plots resulted in a greater proportion of mature
10
pods at final harvest" (Nageswara-Rao et al., 1988, p. 431).
Soil water deficits during pod fill reduced yield 28 to 68%,
due to decreased initiation and development of pods caused
by increased soil temperature and decreased soil moisture
(Nageswara-Rao et al., 1985).
In summary, soil water deficits have been shown to:
increase water extraction from lower rooting depths,
decrease dry matter and crop growth rate, decrease flower
number and pod number, and inhibit pegging and pod
development through decreased peg turgor, high soil
temperatures and decreased Ca uptake. Water stress reduces
photosynthesis and transpiration while increasing diffusive
resistance (stomatal closure), and reducing leaf water
potential (Boote et al., 1982; Ketring et al., 1982).
Seedling growth is optimal when photosynthetically
active radiation (PAR) reaches 25 mol m-2 day'1. Shading
decreases peanut shoot weight, flower number, peg number,
pod weight, and pod number (Hang et al., 1984; Ketring et
al., 1982). Hang et al. (1984) found that Florunner peanut
was most sensitive to shading during pod filling. A 75%
reduction in light intensity from 83 to 104 DAP reduced pod
yield 30%. Twenty-one days of shade at flowering, however,
did not significantly reduce final yield.
Optimal temperature for peanut dry matter production is
approximately 30 OC, but flowering has a lower optimal
temperature at 20-25 OC. Photosynthetic rate decreases 25%
at 40 oC and 65% at 10 oC as air temperatures vary from the
30 oC optimum (Ketring et al, 1982). Ong (1982) compared 14
genotypes for thermal time requirements and reported base,
optimal and maximum temperatures to be 8.0-11.5, 29.0-36.5
and 41.0-47.0 oC, respectively. Peanut fruit development is
also sensitive to temperature. An increase in soil
temperature from 23 to 37 OC decreased fruit number, yield
and filling period (Dreyer, 1980; Dreyer et al., 1981).
Fruiting zone temperature is influenced by canopy and air
temperature as well as soil temperature (Sanders et al.,
1985).
Soil fertility
Peanut response to liming of soils has been found to be
caused mainly by increased calcium levels. Peanut was
tolerant of acid soils of pH 4.3 in Malawi (Cox et al.,
1982). Fertilization with K, Mg and N is not usually
necessary, although K might be needed following bahiagrass
(Paspalum notatum) (Cox et al., 1982). Phosphorous
deficiency is probably the most common nutrient deficiency
worldwide, correctable with P fertilization to levels
greater than 7 ppm (Cox et al., 1982).
Calcium is the most yield limiting nutrient in the
southeastern U.S. (Cox et al., 1982). Calcium is necessary
in the fruiting zone for peanut pod development. Three
separate "critical" levels of solution Ca levels, expressed
as the activity ratio of Ca/total cations, were 0.10 for
12
vegetative growth, 0.15 for fruit load and 0.25 for pod fill
(Wolt and Adams, 1979). Recommended levels of extractable
soil calcium vary from 335 kg ha'1 in Alabama to 540-720 kg
ha-1 in North Carolina. Ca in excess of 600 kg ha'1 showed
no yield response in Florida, nor did additions of Mo or B
(Nagel, 1981). Increase in calcium sulfate applications
from 0 to 2240 kg ha'1 did not decrease Pythium or
Rhizoctonia pod rot (Filonow et al., 1988).
Twenty-four peanut cultivars were grown under low soil
fertility on a Tifton loamy sand, pH 5.3, with 10, 25, 69,
and 19 mg kg'1 of P, K, Ca and Mg respectively (Branch and
Gascho, 1985). Significant differences were found among
cultivars for peanut grade and yield, although no
significant relationship was found for a single nutrient and
yield. University of Florida breeders should be
congratulated, because Florunner had the highest average pod
yield (3215 kg ha"1) of the 24 cultivars, while Early Runner
had the highest total SMK yield (1700 kg ha''). It should
come as no surprise that 98% of all southeastern peanut
acreage was planted to Florunner in 1982 (Henning et al.,
1982), although the proportion planted to Florunner has been
decreasing in recent years.
A study of large-seeded virginia type cultivars showed
significant differences among cultivars and nutrient
treatments for nutrient content of petioles, leaflets and
seeds (Coffelt and Hallock, 1986). Seed yield, market grade
and germination percentage were also affected. Results
indicate cultivars differed in nutrient requirement,
nutrient uptake and redistribution of nutrients.
A five year field trial in northern Ghana established
that the inclusion of peanuts in crop rotations led to
significantly higher levels of soil organic N (Tiessen,
1988).
Biotic Factors
Diseases
During its early cultivation, peanut was regarded as
relatively free of disease. However, rust infection was
identified in 1884, leafspot in 1914, and stem rot in 1917
(Porter et al., 1982).
Early and late leafspots (Cercospora arachidicola and
Cercospora personatum). Peanut yield losses up to 50% are
common in fields where fungicides are not used for leafspot
control (Porter et al., 1982). Early leafspot infection on
the adaxial leaf surface, or late leafspot infection on the
abaxial surface, results in small necrotic flecks which
enlarge and become light brown to black circular spots
ranging from 1-10 mm in diameter (Porter et al., 1982;
Smith, 1984). Optimal temperature for leafspot infection is
25-32 C, and infection is more severe when crop rotation is
not practiced (Porter et al., 1982; Smith, 1984). Severe
leafspot infection decreases leaf area index (LAI) and
canopy carbon exchange rate (Porter et al., 1982).
14
Bourgeois et al. (1991) found significant differences in LAI
and canopy weight between fungicide treated and leafspot-
infected plots, beginning at 80-90 DAP.
A screening of 16 genotypes showed similar V-stage
measurements but reduced vegetative weights and partitioning
coefficients under early and late leafspot pressure (Knauft
and Gorbet, 1990). Both Florunner and Sunrunner had high
percentage necrotic leaf area (9.3-10.9%) and disease
ratings (8.5-8.8 on a scale of 1.0-9.0). Pod yield of
Florunner and Sunrunner was high until 120 DAP, then
declined rapidly without disease control. Florunner was
also found highly susceptible to early leafspot in Oklahoma
(Melouk and Banks, 1984). Florunner's high assimilate
partitioning to pod growth (92%) was found to limit
Florunner's leaf production during pod fill and preclude
replacement of diseased leaves (Pixley et al., 1990).
Partially resistant genotypes compensated for leafspot
induced defoliation by lower partitioning to pods, thus
allowing greater leaf area growth during pod fill.
A survey of 35 peanut fields in seven Florida counties
revealed that late leafspot predominated, accounting for 88%
of the leafspots counted and 66% of the leafspot area
(Jackson, 1981). Leafspot infection levels were low in
Levy and Marion County fields following bahiagrass (Paspalum
notatum) rotations.
Rust (Puccinia arachidis). Before 1969, rust was
confined to South and Central America, but it is now spread
worldwide. Orange colored rust pustules first appear on the
abaxial leaf surface. Infection is favored by optimal
temperatures of 20-30 oC, and available water on the leaf
surface (Mallaiah and Rao, 1979; Porter et al., 1982;
Subrahmanyam et al., 1984a).
Rust infection early in the peanut life cycle leads to
reduced leaf size and premature leaf senescence. Shoot
growth is slowed, and the crop life cycle is reduced by 15
to 20 days. Seed size and oil content are also reduced.
Rust uredial pustules rupture the leaf epidermis, and are
more concentrated on the lower leaf. Rupture leads to
increased transpiration, increased water stress and leaf
senescence. Leaf fall may provide organic matter for
secondary white mold (Sclerotia) attack (Bromfield, 1971).
Rust can cause yield losses up to 70% for genotypes
without disease resistance. Pod yield is correlated to
green leaf area remaining at maturity, indicating that the
disease does not affect the photosynthetic efficiency of the
remaining green leaf area (Subrahmanyam et al., 1984b). Rust
infection in Africa is favored under "optimal temperature
and rainfall" conditions, while weather requirements for
leafspot infection are more flexible. Frequent fungal
applications are not "a realistic alternative" for most
African farmers, and more rust resistant varieties are
needed (Savary et al., 1988).
Stem rot (Sclerotium rolfsii). Sclerotium rolfsii
infection is known as stem rot, white mold, southern stem
rot, southern blight or Sclerotium rot. The disease is
found worldwide, with yield losses up to 80% in severe
cases. Symptoms start with yellowing and wilting of a lower
branch followed by the development of sheaths of white
mycelium near or at the base of the plant (Backman, 1984;
Porter et al., 1982). The disease is favored by moist, warm
conditions which usually coincide with increased canopy
cover and relative humidity at pegging and pod formation.
The use of transparent plastic mulch which increased soil
temperature beyond the stem rot optimum of 30-35 oC, caused
lower disease incidence (Porter et al., 1982). Pod yield
has been found to be negatively correlated to number of
disease loci, decreasing an average of 49 kg ha'' per locus
(Backman, 1984).
"Accumulation of dried leaves around the base of peanut
plants partially defoliated by any stress factor (e.g.
leafspot, insects or drought) creates optimal conditions for
initiating an epidemic if sclerotia are present" (Beute and
Rodriguez-Kabana, 1979, p. 872). High moisture and
temperature conditions then influence the severity of the
epidemic (Beute and Rodriguez-Kabana, 1979; Bromfield,
1971). Biological control using Pseudomonas fluorescens
bacterial strains decreased Sclerotium mycelial growth in
vitro and protected 99% of inoculated greenhouse peanut
plants from infection. Non-inoculated greenhouse plants
died within 10 days (Ganesan and Gnanamanickam, 1987).
Nematodes
All nematode pests decrease root growth, leading to
increased nutrient deficiencies and susceptibility to
drought. However, root-knot nematodes (Meloidoqyne arenaria
Chitwood) are the most important nematode pest of peanut,
causing yield losses of 20-90% (Porter et al., 1982;
Rodriquez-Kabana, 1984). Root-knot nematodes are
distributed worldwide between 350 N and 350 S latitude.
Infected plants develop galls of various sizes on the roots.
The galls are internal to peanut tissue, not appended to the
root like Rhizobium nodules. Infected tissue spread by
farming operations or running water greatly expand the range
of nematode infection (Rodriguez-Kabana, 1984; Porter et
al., 1982).
At present, there are no peanut cultivars resistant to
root-knot nematode. Control of root-knot consists of the
use of systemic nematicides along with crop rotations. A
study of four systemic root-knot nematicides (aldicarb,
phenamiphos, oxamyl and carbofuran) on Florunner revealed
they were most effective when applied during the first 2
weeks of crop growth. Yield enhancement was significantly
less if nematicides were applied at blooming, 8 weeks after
planting. Early applications delayed the exponential
population growth characteristic of nematodes and prevented
early damage to the peanut main tap root (Rodriguez-Kabana
et al., 1982a). Banded placement of aldicarb or phenamiphos
was superior to in-furrow or combined placement (Rodriguez-
Kabana et al., 1982b). Early bunch-type genotype UF77118
inoculated at seeding with root-knot nematodes had only 50%
pod yield of the control, and had lower LAI, stem weight,
leaf weight, peg and pod number (Senthong, 1979). Plants
inoculated at pegging and complete ground cover were not
significantly different from control plants.
Florunner yields from 16 experiments over 3 years were
negatively correlated to root-knot nematode larval levels
(number of larvae per 100 cm soil), with a statistically
significant quadratic regression fit (Rodriguez-Kabana et
al., 1982c). However, there were differences in the
equations among seasons. At average larval levels, average
yield loss was 427-539 kg ha"'. Yield was reduced even
with low (<50 larvae 100 cm-3 soil) nematode numbers.
Unfortunately, larval levels from pre-season assays are not
reliable for prediction.
Root-knot nematodes do not infect the forage crop
bahiagrass (Paspalum notatum). Crop rotation of peanut with
bahiagrass was found to be as effective as using aldicarb in
reducing population densities of M. arenaria and increasing
peanut yields (Rodriguez-Kabana et al., 1988). Peanut
19
yields were 27% higher, and M. arenaria population densities
were 41% lower, in plots following one year of bahiagrass
than in plots under peanut monoculture.
Insects
Smith and Barfield (1982) identified 360 species of
peanut insect pests worldwide. Peanut was ranked 10th out
of 77 plant species for number of pests. Smith and Barfield
elucidated a relationship between peanut growth stage and
tolerance to insect attack. For example, as peanuts age,
the crop becomes less tolerant to nut feeders, while
tolerance to leaf feeders is lowest at maximum LAI and rises
thereafter. They proposed an emphasis on Integrated Pest
Management (IPM) as opposed to chemical pesticides, but
admitted that economic threshold data and sampling
methodology were lacking.
Wilkerson (1980) found that early season defoliated
peanut plants suffered a decline in all plant part weights.
Late season defoliated plants lost stem and pod weight, but
used stem reserves to grow new leaves.
Peanut insect pests in the semi-arid tropics include
both above ground foliage feeders such as leaf miners
(Aproaerema modicella) and armyworms (Spodoptera spp.), and
below ground root and pod attackers such as termites
(Termitadae) and white grubs (Holotrichia spp.). Aphids
(Aphis craccivora) and thrips (Thripidae) are vectors of
groundnut rosette and tomato spotted wilt virus. More
research needs to be done on plant resistance to these
pests, and the interaction of microclimate, genotype and
natural enemies within traditional cropping systems of the
tropics (Wightman and Amin, 1988).
Crop Growth Modelling
Purpose of Crop Modelling
Whisler et al. (1986) catalogued use of crop simulation
models into three broad categories: (1) aids in interpreting
experimental results, (2) agronomic research tools, or (3)
agronomic management tools. The level of detail used in the
model is related to these objectives and to the available
data needed to build and run the model.
Crop growth models have the potential to simulate
crops, crop varieties and cropping practices over a wide
range of climatic conditions. The model selected should be
evaluated previously, be sensitive to the factors of
interest, use available inputs, and be easy to use (Boote
and Jones, 1988). Boote and Jones (1988) used PNUTGRO
simulations to choose optimal planting dates for 21 years of
weather data for Gainesville, FL. and for four years for
Niamey, Niger. The best planting dates for Florida
coincided with recommended practices, while simulated peanut
yield in semi-arid Niger declined as planting date was
delayed after the start of the rainy season.
Peanut Model Development
The PNUTGRO crop growth simulation model is a
physiologically based model which considers crop carbon
balance, nitrogen balance and water balance at the process
level. Daily canopy photosynthesis, respiration, growth,
phenology, and partitioning are predicted given weather,
soil and varietal inputs (Boote et al., 1986; Boote et al.,
1989a). Photosynthesis is a function of solar radiation,
temperature, available soil water, LAI and leaf nitrogen.
Maintenance respiration depends on temperature, crop
photosynthesis rate and current crop biomass. Growth
respiration is calculated using conversion equations (in
glucose equivalents) for synthesis costs of different plant
tissues. Rates of crop development and phenology are based
on heat unit accumulation with an optimal temperature range
of 28-320C, declining to zero at 11 and 450C (Boote et al.,
1989a).
Partitioning varies with growth stage, drought stress
and among cultivars. As pods and seeds develop, increasing
assimilate is partitioned to reproductive parts until a
varietal limit is reached, e.g. 0.85 for Florunner (Boote et
al., 1989b). Crop maturity occurs when a given heat unit
accumulation occurs, maximum shelling percentage is reached,
or photosynthetic capacity is lost through crop aging or
drought.
The PNUTGRO model uses the Ritchie soil water balance
model (Ritchie, 1985), which divides the soil into up to 10
layers. Each soil layer contains soil water and root
densities which change over time due to weather effects,
plant growth dynamics and soil characteristics.
In addition to PNUTGRO, other peanut models have been
developed. AUNUTS, an expert system for managing pests in
peanuts, is designed to be a "planting to harvest decision
aid" (Davis et al., 1989). The expert system questions
users on their changing problems and provides expert
knowledge from a stored database. AUNUTS, for example, will
predict yield loss from root-knot nematode larval numbers
and advise growers whether to grow peanuts on a given field.
Bourgeois (1989) created a simulation model of the
progression of late leafspot (LATESPOT) and coupled it to
PNUTGRO. LATESPOT predicts conidia dissemination, disease
development and plant damage based on daily environmental
conditions. Canopy photosynthesis is reduced by disease-
induced defoliation. The model predicted dry weight losses
adequately on Florunner field data in Florida from 1983,
1985 and 1987. Knudsen et al. (1987) also generated a
computer simulation model that predicts Cercospora leafspot
infection rate as a function of relative humidity, minimum
air temperature and amount of infected tissue. The model
was validated using independent weather and disease data
from the Florigiant cultivar.
A regression model for predicting yield loss as a
function of Cvlindrocladium black rot incidence has been
developed (Pataky et al., 1983). However, multiple
regression analysis models do not adequately describe
interactions between plant growth factors (Acock and Acock,
1991). Crop simulation models have greater potential than
empirical models for making extrapolations. Models
developed from growth chamber data, however, need to be
calibrated and validated with field data. Long term field
research can be especially useful because: 1) some
environmental factors would be stable over time, and 2)
effects of previous unknown treatments would be minimized
(Acock and Acock, 1991).
Model Calibration
Crop growth model development includes the process of
calibration of model parameters to improve model performance
over a range of environments. An alfalfa (Medicago sativa)
growth simulation model, ALF2LP was evaluated under Quebec
conditions by Bourgeois (1990). He calibrated the specific
leaf area and maximum growth rate, after which the model
accurately simulated dry matter yield and crude fiber
percentage when validated with independent data sets.
However, the ALF2LP model "can be considered as a potential
yield estimator because it does not include functions for
soil fertility, insect damage and disease effects"
(Bourgeois, 1990, p. 10).
Modellers must be careful not to make changes that only
serve to fit data from one site. A potato (Solanum
24
tuberosum) model developed in Idaho was used in New York and
underpredicted biomass by 17-40% (Ewing et al., 1990). When
the model was calibrated to fit New York data, it then
overestimated Idaho yields by 51%.
A limited geographical model for corn (Zea mays L.)
yield response to water stress and heat units was developed
and calibrated for the deep loessial lands of the upper
Mississippi valley (Swan et al., 1990). Limited range of
soil types and climate, no soil nutrient deficiencies or
serious pest damage, and continuous corn cropping led to
ease of model development.
Model Validation
As Burk (1986, p. 35) has noted, "modelers often spend
too little time on validation or provide results of limited
significance to the user." However, caution must be
exercised when discussing validation experiments.
Validation is not testing the truth of a model; every model
fails at some level of resolution. Neither is validation a
statistical analysis of an individual model component.
Validation is "an evaluation of the usefulness of a model,
as a whole, in providing reliable information for a specific
problem" (Burk, 1986, p. 35). Data sets independent of
those used in model development and calibration are used to
compare model predicted and observed values. Both model
error and data variability are sources of error in this
process (Whisler et al., 1986).
Final yield data alone are inadequate for model
validation. Model error in predicting final yield can not
be corrected without knowledge of crop growth dynamics
during the season. Alternatively, model results may match
final yields without correctly modelling crop growth
patterns. Acock and Acock (1991) show a wide range of
weather, soil and plant parameters needed for crop
simulation. Regression analysis may be used to compare
simulated results to observed values at different points
during the growing season (Huda, 1988).
Computer models are being tested in a wide array of
agronomic fields today, from planning planting dates in
single-truss tomato (Lycopersicon esculentum) cropping
systems (McAvoy et al., 1989) to predicting ammonia
volatilization in flooded soil systems (Jayaweera et al.,
1990) or for predicting early and late blight outbreaks on
potato (Solonum tuberosum) (Shtienberg et al., 1989).
This thesis research was conducted to evaluate the
PNUTGRO peanut crop growth model under actual farm
conditions in order to determine PNUTGRO's utility for
farmers and extension agents, and to identify areas for
model improvement. Independent crop and soil data sets from
Florida peanut fields were gathered to compare with
simulated PNUTGRO values.
29
replicates to the IFAS Nematode Assay Laboratory, where the
soil and roots were analyzed for Meloidoqyne root-knot and
other nematode populations. Another similar nematode assay
was taken in 1991 after approximately three months (August
14 and August 19) at each site to obtain population
estimates on mature plants. Late-season nematode assays
were not collected in 1990.
Soil Water Sampling
Gravimetric soil water content was sampled in two
replications per site every 3 weeks in Levy County farms and
every 4 weeks in Jackson County farms. Soil water and plant
growth samples were taken at 4 week intervals in Jackson
County due to its greater distance from Gainesville and
corresponding logistical requirements. Aluminum tubing
marked at 15, 30, 45, 60, 90 and 120 cm was manually driven
into the soil and cores from these six depths were placed in
ziplockTM freezer storage bags.
The ziplockTM bags and soil were weighed immediately
upon return to the field laboratory in Gainesville. The
bags were opened and oven dried at 90 OC for 48 hours, then
reweighed. The gravimetric water content was obtained using
the equation:
% Gravimetric soil water =
(Wet wt. (g) Dry wt. (g))/(Dry wt. (g) Ziplock wt. (g)).
Bulk density (BD) values for the six soil layers at
each site were obtained using Soil Conservation Service
(SCS) maps. Volumetric water content was calculated as:
% Volumetric soil water = % Gravimetric soil water x BD.
Weather Data Collection
In order to simulate peanut growth and yield the
PNUTGRO model requires several weather inputs: solar
radiation, minimum and maximum daily air temperature, and
daily rainfall or irrigation. Automated tipping bucket
raingauges were installed at each field site to record
rainfall and/or irrigation in each field. LICOR LI-1200
weather stations were placed in each county to obtain daily
solar radiation and maximum and minimum air temperatures.
In 1990, weather stations were placed on the Morgan farm (MO
R) in Jackson County and the Graham farm in Levy County. In
1991, LI-1200 weather stations were placed at the Morgan
farm (MO R) in Jackson County and the Sandlin farm in Levy
County. In addition, a LI-1000 data logger was installed to
record weather data at the Graham farm in Levy County.
Solar radiation and air temperatures from weather
stations in each county were used to create weather files
for each site in that county, combined with raingauge data
obtained at each field site.
Plant Growth Sampling
Plant growth samples were taken every 3 weeks at Levy
County sites and every 4 weeks at Jackson County sites
throughout the growing season. One meter of row (0.914 m2
area) was sampled from each of the four replications. Care
(SCS) maps. Volumetric water content was calculated as:
% Volumetric soil water = % Gravimetric soil water x BD.
Weather Data Collection
In order to simulate peanut growth and yield the
PNUTGRO model requires several weather inputs: solar
radiation, minimum and maximum daily air temperature, and
daily rainfall or irrigation. Automated tipping bucket
raingauges were installed at each field site to record
rainfall and/or irrigation in each field. LICOR LI-1200
weather stations were placed in each county to obtain daily
solar radiation and maximum and minimum air temperatures.
In 1990, weather stations were placed on the Morgan farm (MO
R) in Jackson County and the Graham farm in Levy County. In
1991, LI-1200 weather stations were placed at the Morgan
farm (MO R) in Jackson County and the Sandlin farm in Levy
County. In addition, a LI-1000 data logger was installed to
record weather data at the Graham farm in Levy County.
Solar radiation and air temperatures from weather
stations in each county were used to create weather files
for each site in that county, combined with raingauge data
obtained at each field site.
Plant Growth Sampling
Plant growth samples were taken every 3 weeks at Levy
County sites and every 4 weeks at Jackson County sites
throughout the growing season. One meter of row (0.914 m2
area) was sampled from each of the four replications. Care
was taken to ensure that the meter-row sample was measured
from peanut plant midgap to midgap, and that border plants
were adequate on all sides of the sample.
Plant height, width and number were measured in the
field and the plants uprooted (root samples were not taken)
and transported in plastic garbage bags to the field
laboratory in Gainesville. Reproductive stages were
determined for all the plants in the sample, and total
biomass dry weight was determined after drying at 70 oC for
3 days. A three plant subsample of representative plants
within each replication was used to calculate the following
growth data: vegetative stage, specific leaf area, leaf
area, stem dry weight, leaf dry weight, number of pegging
sites, number of pods less than and greater than R4, pod dry
weight less than and greater than R4, number of seeds, and
seed dry weight.
The principle of allometry was used to multiply total
biomass by subsample fraction leaf, stem, and pod values to
obtain total plant component dry matter per unit land area.
For example, leaf area index (LAI) was calculated by
computing a leaf area to leaf mass ratio (specific leaf
area) and a leaf to total plant mass ratio (fraction leaf)
from the three plant subsample, and multiplying these ratios
by the total biomass dry weight (g m2):
LAI = m2 leaf/g leaf x g leaf/g plant x g biomass/m2 soil.
The same approach was used to compute leaf, stem, seed and
pod biomass; i.e. fraction leaf, stem, pod, etc. was
multiplied by total biomass (including the three plant
subsample) to express dry matter partitioning to plant parts
on a land area basis. This technique assumes that the ratio
of leaf (or stem, pod, etc.) weight to total biomass is
similar for neighboring plants of the same age and genotype
(Pixley et al., 1990).
Final Harvest Sampling
A final harvest sample of 8.94 m2/replication was taken
as close as possible to the grower harvest date. Two rows,
each 4.9 m long, were dug manually in each plot with
pitchforks and the peanut plants placed on a tobacco sheet.
Dropped pods were then recovered from the soil. Pods with
healthy peg attachments that had been accidently pulled off
by the pitchfork were combined with the harvested pod
sample. Dropped pods with rotting peg attachments due to
disease and/or age were dried and weighed separately.
Harvested plants were transported to Gainesville and
dried outdoors on the tobacco sheets for 3 to 4 days and
then threshed mechanically. The remaining pegs and pods
were dried at 35-40 oC for 1 week in field dryers, then the
pegs were removed mechanically. Total pod dry weight was
measured, and a 500 g subsample was removed. Loose shelled
kernels (LSK) were extracted from the subsample and weighed
separately. The remaining pods were sized and shelled
farms in Jackson County and all the Levy County sites were
in different fields.
Farmers managed the research plots exactly the same as
the rest of the field. Survey forms on management practices
were completed by each farmer to establish crop history and
agrichemical use.
Soil Fertility and Nematode Tests
Four plots of eight rows were marked at each site to
create four replications 66.8 m2 in area. The four plots
were adjacent to one another in one block selected within a
uniform area of the field. The plots were staked just after
seedling emergence to allow evaluation of adequate plant
population. Care was taken to ensure that the research
plots were not close to the field borders and that the soil
type within the plots was as uniform as possible.
Initial soil fertility tests and nematode assays were
collected soon after plant emergence. The top 20 cm of soil
was sampled randomly and mixed in a plastic bucket.
Chemical analyses of Mehlich I extractable P, K, Ca, Mg, Zn,
Mn, Cu, Na and Fe, and water extractable pH, NH4 and NO3
were conducted at the IFAS Soils Testing Laboratory (see
Table A-l). Although PNUTGRO does not presently respond to
soil fertility factors, this information was collected to
allow comparisons and for future reference.
Nematode assays were conducted by submitting seedling
tap roots and surrounding soil from each of the four
RESULTS AND DISCUSSION
1990 On-Farm Experiments
Levy County
The Levy County farmers used peanut crop rotations with
livestock-grazed bahiagrass (Paspalum notatum) in their
farming system. Bahiagrass was grown for 2 to 8 years
between peanut plantings. Bahiagrass rotations have been
shown to reduce nematode populations (Rodriguez-Kabana et
al., 1988) and leafspot infection rates (Jackson, 1981) in
subsequent peanut crops.
The soils in Levy County were all Arenic or Grossarenic
paleudalfs except for the Typic quartzipsamment on the
Sandlin farm. Table A-1 shows soil fertility test results.
Soil pH ranged from 5.30-6.83 on the Levy County farms. At
the beginning of the growing season (after N-P-K
fertilization), the Sandlin farm had the lowest measured
Mehlich I extractable levels of Ca, Mg, K, and P of 78.2,
11.0, 13.8 and 12.5 mg kg'1, respectively. These levels are
comparable to the low values in the Branch and Gascho (1985)
study on low soil fertility effects on peanut growth.
However, nutrient deficiencies did not limit peanut growth
at the Sandlin farm. Despite having the lowest soil
fertility, the peanut pod yield was greater at the SA farm
RESULTS AND DISCUSSION
1990 On-Farm Experiments
Levy County
The Levy County farmers used peanut crop rotations with
livestock-grazed bahiagrass (Paspalum notatum) in their
farming system. Bahiagrass was grown for 2 to 8 years
between peanut plantings. Bahiagrass rotations have been
shown to reduce nematode populations (Rodriguez-Kabana et
al., 1988) and leafspot infection rates (Jackson, 1981) in
subsequent peanut crops.
The soils in Levy County were all Arenic or Grossarenic
paleudalfs except for the Typic quartzipsamment on the
Sandlin farm. Table A-1 shows soil fertility test results.
Soil pH ranged from 5.30-6.83 on the Levy County farms. At
the beginning of the growing season (after N-P-K
fertilization), the Sandlin farm had the lowest measured
Mehlich I extractable levels of Ca, Mg, K, and P of 78.2,
11.0, 13.8 and 12.5 mg kg'1, respectively. These levels are
comparable to the low values in the Branch and Gascho (1985)
study on low soil fertility effects on peanut growth.
However, nutrient deficiencies did not limit peanut growth
at the Sandlin farm. Despite having the lowest soil
fertility, the peanut pod yield was greater at the SA farm
than any other farm in this study. PNUTGRO does not
presently account for soil fertility factors, and soil
fertility was not considered a limiting growth factor in
this experiment.
Levy County farms received enough combined rainfall and
irrigation (see Figure 1) to exceed the 500-600 mm of well-
distributed water needed for adequate peanut growth and
development (Boote et al., 1982). Table A-4 lists water
received at all raingauges in 1990. Table 3 shows
cumulative water received at all sites, and average seasonal
solar radiation and temperature values.
Volumetric soil water percentages were measured for 15,
30, 45, 60, 90, and 120 cm soil layers at 3-week intervals
throughout the growing season for all farms. The PNUTGRO
model tended to overestimate volumetric soil water content,
but followed the general trend of wetting and drying cycles
observed in the individual fields (Figures 2-5). The
Brookins soil water percentages varied from 2.9 to 14.7% at
the 15 and 30 cm soil layers throughout the growing season
(Figure 2). Similarly, soil water at the Graham farm varied
from 6.1 to 13.0% at the 30 cm layer (Figure 3). The
PNUTGRO model tended to overestimate volumetric soil water
when observed values dropped below 5% on the Lowman farm
(Figure 4). Also, the measurements on the Sandlin farm were
extremely low (usually between 2.0 and 5.0% for all soil
LEVY COUNTY, 1990
E 800
w 600
0
w
w 400
I-
w
> 200
D 0
0 100
125 150 175 200 225 250
DAY OF YEAR
Figure 1. Cumulative water received at all Levy County field
sites in 1990.
Table 3. Seasonal water received, and average daily solar
radiation and temperature for all field
1991.
sites in 1990 and
Cum. Average Cum. Average
Water Solar Solar Daily
County Farm Year Received Rad. Rad. Temp.
--mm-- -MJ m"2 d'-- -MJ m-?- -C-
Levy GR 90 583 21.04 2632 25.9
BR 773 2726 -
SA 783 3214
LO 645 2859 -
Jackson MO R 90 421 22.29 3028 26.1
MO I @ 3051 -
CR R 407 2656
CR I @ 2656 -
Levy GR 91 1184 19.18 2419 27.0
SA 875 19.26 2603 26.2
LO 626 2374 -
Jackson MO R 91 808 19.07 2651 26.6
MO I @ 2758 -
CR R 675 2354
CR I 732 2354
@ Indicates missing data due to raingauge malfunction.
.240
.160
S \
0 0
24 14 81 158 15
DAYS AFTER PLANTING
-.- a SMC 15-38 :BR 9
I SC 5-15 c:BR 90
Figure 2. Observed (points) and PNUTGRO-simulated (lines)
volumetric soil water content from 5-15 and 15-30
cm soil layers at the Brookins farm in Levy County
in 1990.
.240
.200
S1690
S..
9,848
i 216 2 7? 104 13
DAYS AFTER PLANTING
-*- SWC 15-30 :GR 90
SMC 5-15 c:GR 99
Figure 3. Observed (points) and PNUTGRO-simulated (lines)
volumetric soil water content from 5-15 and 15-30
cm soil layers at the Graham farm in Levy County
in 1990.
.300
.250 t
.200
.050
6 29 58 81 116 141
DAYS AFTER PLANTING
--- o SUC 15-30 :LO 90
SC 5-15 c:LO 98
Figure 4. Observed (points) and PNUTGRO-simulated (lines)
volumetric soil water content from 5-15 and 15-30
cm soil layers at the Lowman farm in Levy County
in 1990.
.180
,1508
,098 i"'"* "
S 32 64 96 18
DAYS AFTER PLANTING
--o SMC 15-38 :SA 9
SMC 5-15 c:SA 98
Figure 5. Observed (points) and PNUTGRO-simulated (lines)
volumetric soil water content from 5-15 and 15-30
cm soil layers at the Sandlin farm in Levy County
in 1990.
layers), despite receiving 783 mm of rainfall, indicating
the coarse texture of the soil. PNUTGRO overestimated
volumetric percentages throughout the season on the Sandlin
farm (Figure 5), except for the sampling date 140 DAP when a
seasonal high of 11.1% was measured.
Figure 6 shows daily maximum and minimum temperatures
from the 1990 Graham farm weather station in Levy County.
Early season minimum temperatures were unusually low (7 days
below 10 oC), but average seasonal temperatures were
adequate for peanut growth. Average maximum and minimum
temperatures were 33.1 and 18.6 oC, respectively. Daily
mean temperature for the entire growing season was 25.9 oC.
Solar radiation at the Graham weather station is shown
in Figure 7 and Table A-5. Intermittent cloud cover and
Florida's convectional thunderstorms caused the fluctuating
daily values. The average seasonal value was 21.04 MJ m'2
day"1.
Levy County growers in 1990 generally reached the
growth and yield potential predicted by PNUTGRO. Table 4
shows observed versus simulated growth and yield parameters
for the Sandlin (SA), Lowman (LO), Brookins (BR), and Graham
(GR) farms in 1990.
Peanut growth on the Sandlin farm was vigorous despite
low native soil fertility of the Typic quartzipsamments.
Figure 8 shows PNUTGRO simulated growth (lines) versus field
measured values (points) for the SA farm. Both canopy
LEVY COUNTY, 1990
125 150 175 200 225 250
DAY OF YEAR
Figure 6. Maximum and minimum daily temperature recorded at
the weather station on the Graham farm in Levy
County in 1990.
40
S30
0
LU
120
UIl
a-
OL-
100
30
E 25
20
z
S15
CO
0
100
LEVY COUNTY, 1990
125 150 175 200 225 250
DAY OF YEAR
Figure 7. Daily solar radiation recorded at the weather
station on the Graham farm in Levy County in 1990.
Table 4. Field-observed (OBS.) vs. PNUTGRO simulated (SIM.) growth and yield data
from Levy County in 1990.
Maximum LAI Biomass at R8 Pod Yield Pod Harvest Index
Farm OBS. SIM. OBS. SIM. OBS. SIM. OBS. SIM.
----------------kg ha'--------------- -----fraction----
SA 5.5 5.0 11851 10894 5681 5470 0.48 0.50
LO 3.7 5.9 8525 11622 5009 5478 0.59 0.47
BR 5.0 5.5 12070 11273 5367 4661 0.45 0.41
GR 5.1 5.2 12190 10974 4981 4522 0.41 0.41
15900
12900
9088
6000
I ,,e *e9 **
3000 T....
3 64 96 18 160
DAYS AFTER PLANTING
--a POD-kg/ha :SA 90
CANOPY HT :SA 98
Figure 8. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Sandlin farm in Levy
County in 1990.
weight and pod weight (kg ha'1) showed close agreement to
model simulations throughout the season, as did partitioning
of assimilate to pods (Figure 9). However, the model ended
growth simulations earlier than plant harvest, indicating an
underestimate of the length of the Sunrunner growth cycle.
Leaf area index (LAI) on the Sandlin farm (Figure 10) peaked
at approximately 90 DAP and declined thereafter, while
PNUTGRO simulated a less rapid decline. PNUTGRO pod yield
estimates were within 3.7% of observed (Table 4).
Peanut growth at the Lowman farm was less vigorous than
at Sandlin's. PNUTGRO overestimated canopy weight, pod
weight and LAI (see Figures 11 and 12) during seasonal
growth. Final pod yield and biomass at maturity were
overestimated by 9.4 and 36.3% as well (Table 4).
Peanut growth and yield at the Brookins and Graham
farms were both underpredicted by PNUTGRO. Figures 13 and
14 show that measured canopy and pod weights were
consistently greater than simulated. PNUTGRO predicted that
drought stress at the Graham field would reduce
photosynthetic rates by 9.2% from July 3 to July 25, but no
drought stress was observed in the field or in the growth
measurements. LAI (Figures 15 and 16) on both farms was
higher than PNUTGRO predictions early in the season but
tended to decline faster than PNUTGRO predicted.
Overall, peanut growth and yield predictions were very
good for Levy County in 1990. PNUTGRO-simulated pod yields
.600
.500
.400 9
.300
.200
.199
.100
132 64 9 1h8 110
DAYS AFTER PLANTING
-aFRAC POD :SA 90
Figure 9. Measured (points) and PNUTGRO-simulated (line) pod
harvest index at the Sandlin farm in Levy County
in 1990.
12.9
10.0
8.00
6.00
4.00
2.00
400
S32 64 95 128 168
DAYS AFTER PLANTING
LAI :SA 90
Figure 10. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Sandlin farm in Levy
County in 1990.
12008
12000 ~ ~ --- --------------------
8000
6888
6000
82 8 8 16 11i
DAYS AFTER PLANTING
-.- a POD-kg/ha : LO 9
I CANOPY WT :LO 90
Figure 11. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Lowman farm in
Levy County in 1990.
6.00 -
5.00
4.00
3.00
2.00
1.00
.000 -
LAI
Figure 12.
DAYS AFTER PLANTING
:LO 90
Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Lowman farm in Levy
County in 1990.
18000
15000
12000
6000
6999
S2 54 8f 18
DAYS AFTER PLANTING
..- POD-kg/ha :BR 99
CANOPY HT :BR 90
Figure 13. Measured (points) and PNUTGRO-simulated
(lines) pod and canopy weights at the Brookins
farm in Levy County in 1990.
15000
12500
7500
2500
t 26
-*-o POD-kg/ha :GR 90
CANOPY MT :GR 98
Figure 14.
DAYS AFTER PLANTING
Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Graham farm in
Levy County in 1990.
12.8
10.0
8.00
4,00
4.2,00
2.00
1 2 4 81 18
DAYS AFTER PLANTING
LAI :BR 90
Figure 15. Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Brookins farm in
Levy County in 1990.
12.09
10.9
8.09
6.00
4.00
2,00
.000 -
LAI
Figure 16.
DAYS AFTER PLANTING
:GR 90
Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Graham farm in Levy
County in 1990.
differed by an average of only 8.9% from those measured.
Similarly, canopy weights at R8 differed by 15.3%. PNUTGRO
seemed to predict peanut growth and yield reasonably well in
peanut growers' fields that followed long-term bahiagrass
crop rotations that reduce pest and disease pressures.
Jackson County
Crop rotations practiced by the growers in Jackson
County were different from those in Levy County. Bahiagrass
rotations with livestock grazing were not practiced on the
four field sites in Jackson County. Crawford (CR R and CR I
sites) planted peanut in 1990 following soybean (Glycine
max) in 1989 and a soybean-wheat (Triticum aestivum)
rotation in 1988. Morgan's irrigated peanut field (MO I)
followed wheat, corn (Zea mays) and cotton (Gossypium
hirsutum) planted the previous two years. Morgan's rainfed
field (MO R), however, had been in bermuda grass (Cynodon
dactylon) since 1982.
The soils on the fields studied in Jackson County were
both Paleudults, but the Morgan fields were red clayey Typic
paleudults, while the Crawford fields were sandy Grossarenic
paleudults. Soil pH was higher than in Levy County,
averaging 6.57 on the Morgan fields and 7.18 on the Crawford
fields. Soil fertility results are shown in Table A-1.
Jackson County farmers applied gypsum (CaS04) and N-P-K
fertilizers to ensure adequate plant nutrition, and nutrient
deficiencies were not observed in the field.
Rainfall, however, was a serious constraint to peanut
growth in rainfed Jackson County fields in 1990. Figure 17
and Table A-4 show cumulative rainfall in the Crawford and
Morgan fields of only 407 and 421 mm, respectively (Table
3). This value is below the minimum 500-600 mm of water
required for adequate peanut growth (Boote et al., 1982).
Peanut plants were wilted at 11 A.M. on July 17 at the MO R
and CR R farms, and on the August 15 visit, the peanut
leaves were so dry as to feel crispy on the MO R farm.
Unfortunately, raingauge malfunction at the MO I and CR I
fields occurred in mid-season. Thus, it was necessary to
assume for the PNUTGRO simulation that the plants in these
fields were not under water stress.
The drought at the CR R site caused volumetric soil
water content to gradually decrease throughout the season at
all soil layers, as did the simulations from the PNUTGRO
soil water balance model (see Figure 18).
Similarly, volumetric soil water percent declined
throughout the season at the MO R field (Figure 19),
although the absolute percentage was always higher than at
the CR R field due to the higher clay content and water
holding capacity of the MO R field. The clayey MO R and MO
I fields caused some difficulty in driving the aluminum
tubes into the dry soil to obtain soil samples. Three tubes
were broken in attempting to sample soil water on August 15
when the soil was exceptionally dry (the upper 15 cm of the
E
E 800
0
LUI
I 600
O
CC
w 400
> 200
0 1
0 1
125
150
175 200
DAY OF YEAR
225
250
Figure 17. Cumulative water received at the Crawford and
Morgan rainfed sites in Jackson County in 1990.
JACKSON COUNTY, 1990
Crawford
-"""
............
- .- *l
- 4..-*. ,,*-** **
4********
----
/ i i iii _
I
00
.249
.200
.160
o129
,0* .4t ,
.884
.000 ----------i- -
S26 52 73 1A4 13
DAYS AFTER PLANTING
-.- SWC 15-30 :CR R 90
-i SWC 5-15 c:CR R 90
Figure 18. Observed (points) and PNUTGRO-simulated (lines)
volumetric soil water content from 5-15 and 15-
30 cm soil layers at the Crawford rainfed farm
in Jackson County in 1990.
.390
.259
.150
.100 |
.000----
-- oSWC 15-30 :O R
SIC 5-15 c:0O R
Figure 19.
DAYS AFTER PLANTING
Observed (points) and PNUTGRO-simulated (lines)
volumetric soil water content from 5-15 and 15-
30 cm soil layers at the Morgan rainfed farm in
Jackson County in 1990.
63
MO R and CR R fields were only 3.3% and 1.4% volumetric soil
water), causing missing data for the 30, 45, 60, 90 and 120
cm soil layers. PNUTGRO consistently overestimated
volumetric soil water percent at MO R, although the model
did follow the general seasonal drying trend (Figure 19).
This discrepancy may indicate deficiencies in the Decision
Support System for Agrotechnology Transfer (DSSAT) method of
computing water holding capacities of the soil from percent
sand, silt clay and organic carbon. The latter percentages
were taken from SCS soil descriptions, which may have not
been accurate for individual fields. Furthermore, assuming
a full initial soil water profile may also have
overestimated soil water available.
Total extractable soil water, computed by considering
the 120 cm rooting zone, also shows the general drying trend
throughout the season for the MO R and CR R farms (Figure
20), contrasted with the adequate soil water during the
latter part of the season at the Brookins farm in Levy
County.
Figure 21 shows daily maximum and minimum temperatures
at the Morgan rainfed farm in 1990. The seasonal average
was 26.1 oC (Table 3). Solar radiation (Figure 22 and Table
3) averaged 22.3 MJ m-2 day-' over the growing season. The
complete weather files for the 1990 Morgan weather station
can be found in Table A-6.
The 1990 drought in Jackson County caused large growth
289.
249.
169,.
120.
80.0
40,0 -
...... PES
u- PES
Figure 20.S
Figure 20.
S- MM :MO R J#YS AFTER PLANTING
H nM CR R 90
; n :BR 90
Simulated total extractable soil water for the
Brookins (BR), Crawford rainfed (CR R) and
Morgan rainfed (MO R) farms in 1990.
JACKSON COUNTY, 1990
40
o30
! 20
W
C--
1io
0O
100
125 150 175 200 225
DAY OF YEAR
Figure 21. Maximum and minimum daily temperature recorded
at the weather station on the Morgan farm in
Jackson County in 1990.
250
30
25 -
Z 20
0
15 -
S10 -
0 5
100 125 150 175 200 225 250
DAY OF YEAR
Figure 22. Daily solar radiation recorded at the weather
station on the Morgan rainfed farm in Jackson
County in 1990.
and yield differences between the rainfed and irrigated
fields on both the Crawford and Morgan farms. PNUTGRO also
responded to these weather effects by creating drought
stress on the rainfed crops: simulated photosynthesis was
decreased by 45% on the MO R field and 55% on the CR R site
from end of pod addition to physiological maturity. Table 5
shows observed versus simulated growth and yield parameters
for the Jackson County farms in 1990.
The cultivar Agritech-127 was grown on the Crawford
farm. This is an earlier maturing variety than Florunner,
with lower yield potential, primarily because of its shorter
life cycle (Table A-2). Figure 23 shows that seasonal pod
and canopy growth were consistent with PNUTGRO predictions
on the CR I field. However, measured LAI for the CR I field
was significantly lower than predicted after the R4 stage
(Figure 24). Simulated maximum LAI was two units higher
than observed (5.49 to 3.49). Crawford's peanuts had been
planted following two years of soybeans, and several
diseases (leafspot and white mold) were noted in his field.
PNUTGRO does not account for pest or disease pressures and
overestimated total biomass at R8 and pod yield by 14.2 and
40.5% for the CR I field.
The Crawford rainfed corner (CR R) showed the effects
of drought as biomass was reduced and pod yield declined
(Table 5). PNUTGRO underestimated canopy weight and LAI
during the season (see Figures 25 and 26), while simulating
Table 5. Field-observed (OBS.) vs. PNUTGRO simulated (SIM.) growth and yield data
from Jackson County in 1990.
Maximum LAI Biomass at R8 Pod Yield Pod Harvest Index
Farm OBS. SIM. OBS. SIM. OBS. SIM. OBS. SIM.
----------------kg ha-1-------------- -----fraction----
CR R 3.8 3.1 7214 5385 1486 2266 0.21 0.42
CR I 3.5 5.5 9744 11127 3916 5502 0.40 0.49
MO R 3.1 5.0 8365 7744 3611 2822 0.43 0.36
MO I 5.1 5.9 12240 12206 5034 5195 0.41 0.43
15000
12500
10000
7500
2500 .
.000 "---- *"
i 26 2 7 104 1
DAYS AFTER PLANTING
-*- POD-kg/ha :CR I 90
CANOPY WT :CR I 90
Figure 23. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Crawford
irrigated farm in Jackson County in 1990.
6.00
5.00
4.00
3.00
2.00
1.00
.000
DAYS AFTER PLANTING
:CR I 90
Figure 24.
Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Crawford irrigated
farm in Jackson County in 1990.
128000
61000
8800
logo
S26 2 7 104 13(
DAYS AFTER PLANTING
--o POD-kg/ha :CR R 90
CANOPY MT :CRR 98
Figure 25. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Crawford
rainfed farm in Jackson County in 1990.
DAYS AFTER PLANTING
:CR R 90
Figure 26.
Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Crawford rainfed
farm in Jackson County in 1990.
4.80
4.00
3.20
2.46
1.60
.000
- LAI
I-
104
130
the drought effects. Contrast the standard sigmoidal LAI
curve of CR I in Figure 24 with the drought-affected LAI
curve of CR R in Figure 26. However, PNUTGRO overestimated
fractional partitioning to pods (Figure 27). Final pod
yield estimates were 52.5% higher than observed (Table 5).
The measured final harvest index was only 0.206, compared to
0.402 in the CR I field. This value is very low compared to
typical Florunner plants, indicating a problem with drought
or pests in the CR R field.
Florunner was grown on both the irrigated and rainfed
fields on the Morgan farm. Canopy and pod growth were both
simulated well by PNUTGRO on the MO I site (Figure 28), as
was seasonal partitioning of assimilate to pods (Figure 29).
PNUTGRO estimated final biomass and pod yield within 0.3 and
3.2% of observed (Table 5). LAI was again overestimated
after the R3 stage (Figure 30). The growth and yield of
peanut on the Morgan rainfed (MO R) farm was remarkable
considering the amount of drought stress (crispy leaves on
August 15). PNUTGRO underestimated pod and canopy growth
around the R6 stage, as predicted drought stress began to
decrease photosynthesis (Figure 31). However, PNUTGRO
overestimated LAI again (Figure 32). Final biomass and
yield were underestimated by 7.4 and 21.8% (Table 5).
Overall, PNUTGRO's yield predictions were not as
accurate in Jackson County (29.5% difference from observed)
as they were in Levy County (8.9% difference) in 1990. The
.480
.400
.320
.240
.160
.080
.000
S26 52 7 154 1
DAYS AFTER PLANTING
FMC POD :CR R 90
Figure 27. Measured (points) and PNUTGRO-simulated (line)
pod harvest index at the Crawford rainfed farm
in Jackson County in 1990.
15000
12500
10000
7500
2500 .8
1 2 4 8 1 8 13
DAYS AFTER PLANTING
--o POD-kg/ha :O I 90
CANOPY WT :MO I 90
Figure 28. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Morgan irrigated
farm in Jackson County in 1990.
.488
.40099
.320
.240
.160
.080
1 2 54 81 1i8 L
DAYS AFTER PLANTING
FRAC POD :MO I 99
Figure 29. Measured (points) and PNUTGRO-simulated (line)
pod harvest index at the Morgan irrigated farm
in Jackson County in 1990.
12.9
10.0
8.00
6.00
2.00
LAI
Figure 30.
DAYS AFTER PLANTING
:NO I 90
Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Morgan irrigated
farm in Jackson County in 1990.
12000
10000
8000 -
8909
6000 1
40690
2 0 9 0 -"-' '- -
.,-..-.-.-,
S2 54 81 108 1
DAYS AFTER PLANTING
-.- POD-kg/ha :MO R 90
CANOPY MT :NO R 99
Figure 31. Measured (points) and PNUTGRO-simulated (lines)
pod and canopy weights at the Morgan rainfed
farm in Jackson County in 1990.
5.40
4.50
3,60
2.70
1.80
,000
-I LAI
Figure 32.
DAYS AFTER PLANTING
:MO R 90
Measured (points) and PNUTGRO-simulated (line)
leaf area index (LAI) at the Morgan rainfed farm
in Jackson County in 1990.
80
row crop rotations followed in three of the Jackson County
fields most likely led to higher pest levels than the
bahiagrass rotations of Levy County, causing growth limiting
factors unaccounted for in PNUTGRO. However, PNUTGRO did
predict the relative decrease in pod yield due to drought in
Jackson County. The 1990 observed yield decrease was 62.1
and 28.3% compared to irrigated fields at the Crawford and
Morgan farms. PNUTGRO predicted declines of 58.8 and 45.7%
respectively on the CR R and MO R farms. The assumption of
full irrigation compared with uncertainty on actual
irrigation may have caused some of the overestimation of the
irrigated fields.
In 1990, pod yields and crop biomass at maturity (R8)
were reasonably close to PNUTGRO simulated values in both
Jackson and Levy County combined, as shown by the proximity
of the data points to the 1:1 line on Figures 33 and 34.
Points above the 1:1 line indicate PNUTGRO underestimates
(simulated less than observed) while points below the line
are overestimates (simulated greater than observed).
Weather (especially rainfall) and cultivar effects were the
primary factors influencing growth and yield in 1990, and
PNUTGRO accounted for these factors well.
1990 POD YIELD
LEVY AND JACKSON
COUNTIES
0
O
O
1J
CO
0)
Uj
0
CL
0
W
rCr
SIMULATED POD YIELD (kg/ha x 1000)
Figure 33. Relationship between observed and simulated pod
yield for all sites in 1990.
0 2 4 6
LU
C)
0
O
O
, 16
S14
- 12
*4 A
Cl)
C/)
0
O
m
CO
0
1990 BIOMASS AT HARVEST
LEVY AND JACKSON COUNTIES
0 2 4 6 8 10 12 14
SIMULATED BIOMASS (kg/ha x 1000)
Figure 34. Relationship between observed and simulated
biomass at harvest maturity (R8) for all sites
in 1990.
mechanically. Peanut seeds were shaken over two different
size screens. Seeds riding a 21.5/64" screen were
classified as extra large kernels (ELK), and seeds riding a
16/64" screen were classified as sound mature kernels (SMK).
The seeds falling through both screens were separated into
sound seeds split in half (SS) and other kernels (OK).
Shelling percentage and SMK percentage were calculated as:
Shelling % = Total seed wt. (g)/Total pod wt. (g)
SMK % = (ELK + SMK + SS (g))/(Total pod wt. LSK (g)).
Data Entry
The PNUTGRO crop growth model predicts peanut growth
and yield in response to soil, weather and genetic inputs.
These input files were created for each site to allow
PNUTGRO to run properly. PNUTGRO's predictive capability
under actual farm conditions was then evaluated by comparing
field-observed growth and yield data to PNUTGRO-simulated
values.
PNUTGRO version 1.02 was used; this version had been
tested in North Florida in 1988 (Boote et al., 1989b), and
found to overestimate peanut growth on farms with disease
and nematode problems. The genetics file used in this study
was calibrated to 1988 on-farm studies in Jackson County,
and shown to work reasonably well in the 1989 season in
Jackson County. Table A-2 shows the genetic coefficients
used for the standard Florunner file in PNUTGRO V1.02, and
the 1988-calibrated version of the Florunner genetics file
|
RSS
HIDE
