<%BANNER%>

Comparison of Measured and Simulated Responses of Maize to Phosphorus Levels in Ghana

Permanent Link: http://ufdc.ufl.edu/UFE0021774/00001

Material Information

Title: Comparison of Measured and Simulated Responses of Maize to Phosphorus Levels in Ghana
Physical Description: 1 online resource (161 p.)
Language: english
Creator: Dzotsi, Kofikuma A
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: active, agriculture, analysis, available, biomass, dssat, evaluation, fertilizer, ghana, grain, inorganic, labile, management, model, modeling, nutrient, organic, phosphorus, plant, pools, prediction, response, sensitivity, simulation, soil, stable, systems, yield
Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Efficient nutrient management in agricultural systems requires the availability of tools that can help in meeting research objectives of understanding the transformations that nutrients undergo to become available to plants and predicting how these transformations are related to economic outputs from the systems. The crop models in the Decision Support System for Agrotechnology Transfer (DSSAT) have been recognized worldwide for meeting these objectives for nitrogen. However, without a phosphorus model, the applicability of the DSSAT crop models in phosphorus deficient environments will remain questionable. In this study, a soil-plant phosphorus model linked to DSSAT was described, analyzed and tested. The sensitivity of the model to six key input factors was studied based on a global sensitivity analysis approach. The model was tested on two P-deficient soils from Ghana (Kpeve and Wa) with maize as the test plant. Processes accounted for by the model include phosphorus movements between inorganic (labile, active and stable), organic (active and stable) pools and plants. Results of the sensitivity analysis showed the greatest effects of initial inorganic labile P (initial PiLabile) and fertilizer P on biomass, grain yield and total P uptake (sensitivity index of 0.11 for initial PiLabile and 0.30-0.43 for P fertilizer). Smaller effects were found for the fraction of root labile P that is soluble (sensitivity index of 0.03-0.04), the shoot P (sensitivity index of 0.03-0.09) and seed P (sensitivity index of 0.15) on total P uptake. Statistical analysis of grain yield and biomass did not reveal any significant differences at the 0.05 probability level at Kpeve because the phosphorus content of this soil was at the limit between deficiency and sufficiency and the organic matter content of the soil was relatively high (close to 2.0%). Grain yield and final biomass responded at Wa with 100% increases in the 60 kg P2O5/ha treatments over the nil-P treatments. Biomass and yield were stable between the two treatments of 60 and 90 kg P2O5/ha at Wa. Evaluation of the model indicated that the model was able to achieve good predictability skill at Kpeve with a grain yield RRMSE of 8% and a final biomass RRMSE of 5%. The congruence between simulation and measurement was fair at Wa. The RRMSE was 14% for grain yield and 30% for final biomass. At Wa however, the model gave a reasonable prediction of the pattern of variability among measurements with an LCS averaged over the five sampling dates of 17%. Because the complex soil P chemistry makes the availability of phosphorus to plants extremely variable in general, further testing of this model in other agro-ecological conditions should precede its application.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kofikuma A Dzotsi.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Jones, James W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021774:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021774/00001

Material Information

Title: Comparison of Measured and Simulated Responses of Maize to Phosphorus Levels in Ghana
Physical Description: 1 online resource (161 p.)
Language: english
Creator: Dzotsi, Kofikuma A
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: active, agriculture, analysis, available, biomass, dssat, evaluation, fertilizer, ghana, grain, inorganic, labile, management, model, modeling, nutrient, organic, phosphorus, plant, pools, prediction, response, sensitivity, simulation, soil, stable, systems, yield
Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Efficient nutrient management in agricultural systems requires the availability of tools that can help in meeting research objectives of understanding the transformations that nutrients undergo to become available to plants and predicting how these transformations are related to economic outputs from the systems. The crop models in the Decision Support System for Agrotechnology Transfer (DSSAT) have been recognized worldwide for meeting these objectives for nitrogen. However, without a phosphorus model, the applicability of the DSSAT crop models in phosphorus deficient environments will remain questionable. In this study, a soil-plant phosphorus model linked to DSSAT was described, analyzed and tested. The sensitivity of the model to six key input factors was studied based on a global sensitivity analysis approach. The model was tested on two P-deficient soils from Ghana (Kpeve and Wa) with maize as the test plant. Processes accounted for by the model include phosphorus movements between inorganic (labile, active and stable), organic (active and stable) pools and plants. Results of the sensitivity analysis showed the greatest effects of initial inorganic labile P (initial PiLabile) and fertilizer P on biomass, grain yield and total P uptake (sensitivity index of 0.11 for initial PiLabile and 0.30-0.43 for P fertilizer). Smaller effects were found for the fraction of root labile P that is soluble (sensitivity index of 0.03-0.04), the shoot P (sensitivity index of 0.03-0.09) and seed P (sensitivity index of 0.15) on total P uptake. Statistical analysis of grain yield and biomass did not reveal any significant differences at the 0.05 probability level at Kpeve because the phosphorus content of this soil was at the limit between deficiency and sufficiency and the organic matter content of the soil was relatively high (close to 2.0%). Grain yield and final biomass responded at Wa with 100% increases in the 60 kg P2O5/ha treatments over the nil-P treatments. Biomass and yield were stable between the two treatments of 60 and 90 kg P2O5/ha at Wa. Evaluation of the model indicated that the model was able to achieve good predictability skill at Kpeve with a grain yield RRMSE of 8% and a final biomass RRMSE of 5%. The congruence between simulation and measurement was fair at Wa. The RRMSE was 14% for grain yield and 30% for final biomass. At Wa however, the model gave a reasonable prediction of the pattern of variability among measurements with an LCS averaged over the five sampling dates of 17%. Because the complex soil P chemistry makes the availability of phosphorus to plants extremely variable in general, further testing of this model in other agro-ecological conditions should precede its application.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kofikuma A Dzotsi.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Jones, James W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021774:00001


This item has the following downloads:


Full Text





COMPARISON OF MEASURED AND SIMULATED RESPONSES OF MAIZE TO
PHOSPHORUS LEVELS IN GHANA




















By

KOFIKUMA ADZEWODA DZOTSI


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

2007



































2007 Kofikuma Adzewoda Dzotsi









ACKNOWLEDGMENTS

Although the final presentation of this thesis is my responsibility, several persons assisted

with the planning and the implementation of the study and helped with a better distillation of the

ideas presented here. Dr. James W. Jones (my supervisory committee chair) provided the

fundamental guidance that I needed to complete this study. He represents for me at this particular

time and in this particular situation the ideal advisor I could imagine. I have been particularly

impressed by his appropriate and prompt interventions related to my questions and needs even if

I drop by his office without an appointment or if he is on a trip.

Dr. Dorota Z. Haman discussed with me on almost every occasion that we met in the

department or on the campus, progress in my study. I was impressed by her suggestions during

our first committee meeting as it relates to the contents of the chapters I should write even

though she is from the irrigation area. Congratulations on her recent appointment as Chair of the

Agricultural and Biological Engineering Department.

Dr. Samira H. Daroub was never tired of my emails and was always in touch from the

Everglades. I was enriched by her experience with the initial version of the soil-plant phosphorus

model.

Ms. Cheryl H. Porter (coordinator of computer applications in the McNair Bostick

Simulation Laboratory) endured with me multiple trips through the phosphorus model computer

code. She was always available to assist and facilitated tremendously my understanding of the

phosphorus model.

Dr. Samuel K. Adiku, Professor of Soil Science at the University of Ghana made the initial

suggestion of letting me conduct the phosphorus field experiment in Ghana. I benefited a lot

from his experience and those of his collaborators at the Kpeve agricultural research station, in

soil science when I was in Ghana. The field experiment in Kpeve would not have been a success









without him. Although he is now in the ABE department working on his own experiments, he

continues to ensure that I obtain the soil, plant and growth analysis data that are accurate.

Drs Upendra Singh, Ken Boote, Arj an Gij sman, Shrinkant Jagtap, and Jon Lizaso directly

assisted me in various ways at critical moments during the fine-tuning of the phosphorus model.

I would like to thank Dr Jesse Naab from the Savannah Agricultural Research Institute in

Ghana for the data he made available for the purpose of this study.

I would like to express my appreciation to my colleagues in the McNair Bostick

Simulation Laboratory; at the International Center for Soil Fertility and Agricultural

Development (IFDC) in Lome, Togo; my friends at Maguire Village, and all my other friends

not specifically listed here for their various support and encouragement during this study.

My family in the United States, Pascaline Akitani-Bob and Robert E. Dzotsi, have been my

true continuous, human, and spiritual support throughout this study.









TABLE OF CONTENTS

page

A CK N O W LED G M EN TS ................................................................. ........... ............. 3

LIST O F TA BLE S .............. ......... ........................................................... 9

LIST OF FIGURES ................................. .. .... ..... ................. 13

A B S T R A C T ............ ................... ............................................................ 16

CHAPTER

1 INTRODUCTION TO MODELING PHOSPHORUS LIMITATIONS TO CROP
P R O D U C T IO N ...................................... .................................................... 18

In tro d u ctio n .................. ...................................... ................... ................ 18
Phosphorus Problem in Agricultural System s .............. ....................... ............... .... 18
Understanding Excess and Deficiency of Phosphorus in Agricultural Systems ..................20
Coping with Excess and Deficiency of Phosphorus in Agricultural Systems........................21
Modeling as a Phosphorus Management Tool in Soils and Plants.................................21
Soil-Plant Phosphorus Simulation Model in DSSAT................ ............ ...............24
O objective and H hypothesis ................ ..................................................... .. ...... 25

2 STATISTICAL ANALYSIS OF FIELD EXPERIMENT FOR TESTING THE MODEL...28

Introdu action ................... .......................................................... ................. 2 8
M materials and M methods ..................................... ... .. .......... ....... ...... 29
Field Experim ents in Ghana .............. .............. ........ ...................... ............... 30
Experim ent in K peve, G hana ................................................ .............................. 30
S ite d description ......... ...... ..................... ......... .. .......................................3 0
Experiment design and management............... .................................. 31
S o il sam p lin g .................................... ................ ... ......................................3 1
Soil m moisture m easurem ents................................ ......................................... 32
Plant sampling and growth measurements........................ ....... ............ 33
E xperim ent in W a, G hana ....................................................................... ..................35
Site and experim ent set up .............................................. ............................. 35
Field and laboratory measurements ......... ................... ......... ............ ............... 35
Statistical A analysis ................................................... ..................... 36
Regression analysis (soil m oisture)...................................... ................................. 36
Analysis of variance at individual tim e points .................................... .................. 37
Analysis of variance considering the effect of time on the repeated
m easu rem en ts .................................................... ................ 3 8
Results and Discussion ...................................... .. ......... ....... ..... 39
Calibration of the TDR M eter .................................................................... ................ 40
Crop Response Results at Kpeve Using Individual Time Points Analysis ...................41
Phenology .................................. .................................. ........... 41









Grain yield and yield com ponents................................ ......................... ........ 41
A boveground biom ass...................................................................... ..................42
P lant height.........................................................................................42
G re e n le a f a re a ................................................................................................... 4 2
Soil m moisture .................. ......... ............. ................. ..... ...............43
Crop Response Results at Kpeve Using Repeated Measures Analysis Techniques .......43
Selection of a correlation structure using the AIC ........................ ..................43
Effect of time on repeated measurements of crop response variables ...................44
Phosphorus treatments by time interactions effects on repeated measurements......44
Effect on repeated measurements of phosphorus treatments averaged over time....44
Discussion of Results Obtained at Kpeve .......................................................... 45
Results and Discussion for the Wa Experiment .................................... ....47
C onclu sions.......... ..........................................................49

3 THE SOIL-PLANT PHOSPHORUS MODEL IN DSSAT ...................................... 64

In tro d u ctio n .................. .................................................................................................... 6 4
Soil and Plant Phosphorus Modeling in DSSAT ..........................................65
Description of the Soil Phosphorus M odel .................................................................. 67
Soil Inorganic M odule ............................................................................. 67
Inorganic phosphorus pools ....................... .............. .......... ...............68
Phosphorus transformations between the inorganic pools ..................................68
Phosphorus availability for uptake by plants .................................................69
Soil O organic M odule ........................................................70
Organic phosphorus pools..................................................... 70
Phosphorus flows between the organic pools ................. ................. ..........71
Phosphorus mineralization and immobilization ...................... .......... ........72
The net phosphorus mineralized .................................................. ....73
Description of the Plant Phosphorus Model.............. ...... ........ ..................73
Phosphorus in the Plant ................................. .......................... .................... 74
Uptake ................ ........................................ 75
Soil Supply .......................................................................75
Plant Dem and and P M mobilization Pools......................................... ............... 75
Partitioning and Translocation .................................. .......................................76
Stress Factors ............................... ...... ..... ........ .... .. .... .... ... ..... 77
M odel Inputs and O utputs ................................................................................... 78
Sensitivity A analysis ...................................................... ...... 78
In tro d u ctio n .............................................................................................7 8
M materials and M methods ............................................................ ............ 81
Com puter experim ent ................ ........ .. ............................ ............. ............... 81
Settings for the computer experiment ..........................................82
Input factors, scenarios and m odel outputs .............. .......................... ...............83
M ethod and design of the sensitivity analysis............................ .....................87
Sensitivity index ..................................................... ................ .................87
R results and D iscu ssion ................... ......................... .................... ........................88
Soil inputs effects ............. ................... ............. 88
Plant param eters effects .....................................................................................91









In te ra ctio n s ......................................................................................................... 9 2
Special case of zero P fertilizer.............................. ....................... ............... 92
C o n c lu sio n ....................................................................................................................... 9 3
Sum m ary and C conclusion ..................................................................... .......................... 94

4 FIELD TESTING OF THE DSSAT PHOSPHORUS MODEL................................ 110

In tro d u ctio n ................... .............................. ..................................... 1 10
M materials and M methods .......... ........................................ .. ...... ....... .. ...... .. 112
The Soil-Plant Phosphorus M odel ........................... ................... ................. .... 112
D atasets for T testing the M odel...................... .. .. ......... .... ................... ...............113
Parameters and Inputs for the Model Tests .....................................................113
W weather conditions ................ ............. ................................. ..... 113
Soil conditions ....................... ....... ........... .... ............... ......... 114
G genetic coefficients ............. ................ .. ......... ........ .................... .. 114
Phosphorus param eters........................ ................... ........... ............... 115
In itial co n d itio n s ................................................... ........... ................ 1 16
M odel E valuation ............ ...................... ......................... .... ........... .. 116
M odel evaluation tools ................ ...... .... ... .. .................... .. 117
R results and D discussion .................................................. .... ...... .. ..........119
W e ath e r .................. ................................. ....................... ................1 1 9
G en etic C o efficient ts ........ ........................................................ .......... .. ................. ... 12 0
P h o sph oru s P aram eters............................................... .................................. .......... .... 12 0
Initial Conditions ........................................................................... ......... ................... 121
M odel Evaluation at K peve ......... ................... ......... .................................. 121
In-season growth ........................................ .... .. ... .. ............122
F final grain yield .....................................................................122
W a............... ....................................... ........ 123
In -season grow th ............... ............................................................ 12 3
In-season shoot P concentration........................................................ 124
F in al g rain y field ...............................................................12 5
C o n clu sio n ......... .... ................................................. ...........................12 6

5 SUMMARY AND CONCLUSIONS ............. ......... .............138

APPENDIX

A MEASURED GROWTH DATA AT KPEVE ........................................ ............... 141

B MAPS OF THE EXPERIMENT SITES LOCATIONS .............. ......... ...................1..44

C INITIALIZATION OF SOIL INORGANIC AND ORGANIC PHOSPHORUS POOLS
IN THE SOIL-PLANT PHOSPHORUS MODEL ........................................................146

Initialization of Inorganic Phosphorus Pools ..................................................... ................. 147
From P Fractionation D ata ...................................... .. ................... ...............147
From Measured Available P Using the Anion Exchange Resin Method ......................147
From Other M methods ....................................... .. ...... ............. 148









Initialization of Soil Organic Phosphorus Pools..................... ...............149
Initialization from P Fractionation Data............................................... .................. 149
Initialization from Measured Organic P ................................................................150
Initialization from Organic C and soil pH .......................................... ............... 150

L IST O F R E F E R E N C E S ..................................................................................... ..................153

B IO G R A PH IC A L SK E T C H ......................................................................... ... ..................... 161









LIST OF TABLES


Table page

1-1 Response of maize to phosphorus application on a phosphorus-deficient soil in a
fertilizer experiment carried out in Ghana in 1999.................... .....................26

1-2 Partitioning of total soil phosphorus in pools specified on Figure 1-1 in a soil from
Carimagua, Colombia .................... ......................... .......... 26

2-1 Growth and development genetic coefficients for the Obatanpa cultivar used at both
sites, K peve and W a (G hana).................................................. ............................... 51

2-2 Summary of fertilizer application methods used in the experiment in Kpeve, Ghana......51

2-3 Specifications of two different covariance structures used for modeling the effect of
time on repeated measures in PROC MIXED for the Kpeve dataset..............................51

2-4 Analysis of Variance for simple linear regression between soil moisture
measurements using TDR and gravimetric methods at Kpeve, Ghana ...........................51

2-5 Test of parameter estimates used to fit the linear regression model in the Kpeve
ex p erim en t ......... ......... ................ ....... ..................................................... 5 2

2-6 Analysis of variance of phenological events, tasseling, anthesis and silking in the
K peve experim ent .............. .................. ..... ......... ...... .............. 52

2-7 Analysis of variance for grain yield (measured in kg ha-1) at Kpeve, Ghana....................52

2-8 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual
time points for crop aboveground biomass measured in kg ha-1 in the Kpeve
ex p e rim e n t ...................................... ..................................... ................ 5 2

2-9 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual
time points for plant height measured in cm in the Kpeve experiment .............................52

2-10 Treatment means at each day for plant height measured in cm, with least significant
difference (LSD) in the Kpeve experiment (a = 0.05)............................... ............... 53

2-11 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual
time points for green leaf area measured in cm2 per plant at Kpeve ..............................53

2-12 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual
time points for soil moisture readings (in %) using TDR at Kpeve................................53

2-13 Akaike Information Criterion (AIC) test for two covariance structures in PROC
MIXED for repeated measures analysis for the Kpeve experiment ..................................53









2-14 F-values and significance probabilities using univariate ANOVA, and for test of
fixed effect using two covariance structures in PROC MIXED for the Kpeve
ex p e rim e n t ...................................... ..................................... ................ 5 4

2-15 Physical and chemical characteristics of the soil at the experimental site in Kpeve,
G h an a ...................... ....... ......... .......... ................................................ 5 4

2-16 Classes of phosphorus availability according to the Bray 1 extraction method ...............55

2-17 Characterization of the different forms of soil phosphorus at Kpeve, Ghana. Data are
reported in m g/kg .................. ..................... ... ........ ...... .............. .. 55

2-18 Physical and chemical characteristics of the soil at the experimental site in Wa,
G h an a ...................... ....... ......... ... ................................................. 5 5

2-19 Main effects of nitrogen and phosphorus on phenological development in maize at
W a, G h an a ......... .. ......... ................ ....... .................................................... 5 6

2-20 Main effects of nitrogen and phosphorus on leaf area indices of maize at Wa, Ghana.....56

2-21 Main effects of nitrogen and phosphorus on cumulative aboveground biomass (in kg
ha-) of m aize at W a, G hana .............. .. ......... .. ......... ........ ..................... 56

2-22 Main effects of nitrogen and phosphorus fruit yield components of maize at Wa,
G h an a ...................... ....... ......... .......... ................................................ 5 7

3-2 Soil category-dependent calculation of P Fertilizer Availability Index...........................96

3-3 Summary of decomposition rates for the soil organic pools and C:P ratios at which
phosphorus is allowed to enter the specific pools................................... ............... 96

3-4 Optimum and minimum phosphorus content (%) in different plant parts and
maximum and minimum plant N:P ratio at three growth stages, as used in the model
fo r m aize .................................................................................9 7

3-5 Summary of parameters in the soil-plant phosphorus model ............ ........................98

3-6 Summary of additional inputs required to run the soil-plant phosphorus model in
D S S A T ...... ......... ................ ....... ........................................................9 9

3-7 Selected physical and chemical properties of the Kpeve soil used in the sensitivity
analysis, as estimated from pedo-transfer functions in DSSAT .....................................100

3-8 Summary of inputs factors and outputs for the sensitivity analysis of the P model ........100

3-9 Specification of the different levels of the input factors "Shoot P" and "Seed P" for
the sensitivity analysis of the P model ............. .............. ....... .. .................. 101









3-10 Main, interactions, and total sensitivity indices (unitless) of biomass for factors used
in the sensitivity analysis ...................... .... ................ .......................... 101

3-11 Main, interactions, and total sensitivity indices (unitless) of grain yield for factors
used in the sensitivity analysis............................................... .............................. 102

3-12 Main, interactions, and total sensitivity indices (unitless) of plant uptake of P for
factors used in the sensitivity analysis...................................... ........................... 102

3-13 Main, interactions, and total sensitivity indices (unitless) of biomass for a special
case of zero P fertilizer. ...................... ............. ...... ............ .. ...... 102

3-14 Mean aboveground biomass, grain yield and total P uptake corresponding to each
level of the input factors used in the sensitivity analysis...... ..... ......... ..................... 103

4-1 Growth and development genetic coefficients for the Obatanpa cultivar used at both
sites, Kpeve and Wa, for testing the phosphorus model..............................127

4-2 Plant parameters used for testing the phosphorus model at Kpeve and Wa .................... 128

4-3. Soil parameters used for testing the phosphorus model at Kpeve and Wa ......................129

4-4 Values of additional inputs required to run the soil-plant phosphorus model for the
K peve and W a experim ents ...................................... ........... .................................... 130

4-5 Estimated initial condition soil parameters for Kpeve.....................................................130

4-6 Estim ated initial condition soil param eters for W a ........................................................ 130

4-7 Summary of aboveground biomass error statistics for the Kpeve and Wa experiments .131

A-i Monthly total rainfall in 2006 (one standard deviation of rainfall), mean daily solar
radiation, and mean daily temperature collected during the Kpeve experiment in
2 006 .............. ...................... ...................................... ......... ..... 14 1

A-2 Days to tasseling (one standard deviation of four replications), days to anthesis (one
standard deviation of four replications), and days to silking (one standard deviation
of four replications) for the experiment in Kpeve, Ghana................ ...................141

A-3 Measured mean aboveground biomass (one standard deviation of four replications)
for four phosphorus treatments, sampled four times during the growing season in the
Kpeve experiment .............. ... ........... ............. .......... 141

A-4 Mean green leaf area (one standard deviation of four replications) for four
phosphorus treatments, measured seven times during the growing season in the
K peve experim ent ............. .. .... ....................................... .......... ......... 142









A-5 Mean maize height (one standard deviation of four replications) for four phosphorus
treatments, measured seven times during the growing season in the Kpeve
ex p erim en t .............................................................................................14 2

A-6 Mean soil moisture (one standard deviation of four replications) in four phosphorus
treatments plots, measured using TDR eight times during the growing season in the
K pev e experim ent .............................................................................. .. ................ .. 142

A-7 Measured mean grain yield (one standard deviation of four replications), unit grain
weight (one standard deviation of four replications), and grain number (one standard
deviation of four replications) for four phosphorus levels in the Kpeve experiment......143

C-1 Relationship between inorganic P pools and P extracted using the Hedley procedure...151

C -2 Specification of soil categories ........................................................................... ..... 151

C-3 Equations for calculating initial inorganic P labile from different extraction methods
for different soil categories ........................................................................ .. 151

C-4 Relationship between organic P pools and P extracted using the Hedley procedure ......152

C-5 Equations for calculating initial total organic P from soil organic carbon (OrgC) and
pH for different soil categories ............................................... ............................. 152









LIST OF FIGURES


Figure page

1-1 Pools of phosphorus in the soil and relationships between soil and plant phosphorus...... 27

2-1 Monthly total rainfall in 2006 (mm) with error bars corresponding to one standard
deviation of rainfall, monthly total rainfall average from 2003 to 2005, and monthly
average daily temperature (C) in 2006 at Kpeve............... ........................................... 58

2-2 Sequential fractionation steps used for extracting the different forms of phosphorus
from soil samples taken before planting of the Kpeve experiment ...................................59

2-3 Monthly total rainfall (mm) with error bars corresponding to one standard deviation
of rainfall and monthly average daily temperature (C) at Wa in 2004.............................59

2-4 Simple linear regression of volumetric soil moisture (%) determined by using Time
Domain Reflectrometry and gravimetric sampling at Kpeve ........................................60

2-5 Phenology of maize as affected by phosphorus application at Kpeve. Error bars
represent standard deviations of measurements taken from four replications .................60

2-6 Stover and grain yield of maize as affected by phosphorus fertilizer at Kpeve. Error
bars represent one standard deviation of measurements taken from four replications......61

2-7 Grain number per m2 and unit grain weight as affected by phosphorus fertilizer at
K p ev e ............. .... ..............................................................6 1

2-8 Aboveground biomass of maize as affected by phosphorus fertilizer application at
Kpeve. Error bars represent one standard deviation of measurements taken from four
replications .................................................................................62

2-9 Height of maize as affected by phosphorus fertilizer in the Kpeve experiment ...............62

2-10 Green Leaf Area of maize in the phosphorus fertilizer experiment at Kpeve ...................63

2-11 Variation in soil moisture measured using Time domain reflectometry in the Kpeve
phosphorus experiment. ........................................ ............ ............. 63

3-1 Processes in the integrated soil-plant phosphorus model in DSSAT............................. 104

3-2 Optimum and minimum P concentration in maize shoots used in the plant P model .....105

3-3 Maximum and minimum N:P ratios used in the plant module to limit uptake of P ........105

3-4 Relationship between maize shoot P concentrations and P stresses affecting
vegetative partitioning and photosynthesis.................... .... ........................ 106









3-5 Simulated plant total aboveground biomass, grain yield and plant uptake of P at
different levels of initial PiLabile, initial organic P and P fertilizer. Error bars shown
represent one standard deviation...................................... ......... .. ............... 106

3-6 Simulated total plant aboveground biomass, grain yield and plant uptake of P at
different levels of maximum uptake fraction, optimum shoot and seed P
concentrations. Error bars shown represent one standard deviation ............................107

3-7 Sensitivity indices for the six input factors and their interactions............................. 108

3-8 Simulated response of total plant aboveground biomass to phosphorus fertilizer at
different levels of P iL abile.............................................................................. ....... 109

3-9 Simulated response of total plant aboveground biomass to PiLabile at different levels
of fraction of labile P ............................................... .................................. ....... .... 109

4-1 Comparison of simulated and measured grain for different phosphorus levels in the
Kpeve experiment ............................... ............ .. ....... .. 132

4-2 Decomposition of the grain yield MSE for the Kpeve experiment ..............................132

4-3 Comparison of simulated and measured biomass on four samples taken during the
season for the four treatments tested in Kpeve. ..................................... ............... 133

4-4 Decomposition of the in-season biomass MSE for the Kpeve experiment .....................133

4-5 Comparison of measured and simulated maturity grain yield obtained in the Wa
experim ent using the 1:1 line ..................................................................... 134

4-6 Measured and simulated responses of maturity grain yield to different combinations
of nitrogen and phosphorus levels in the W a experiment............................................... 134

4-7 Decomposition of the grain yield MSE for the Wa experiment .................................... 135

4-8 Components of the biomass MSE for the Wa experiment at five sampling times ..........135

4-9 Measured and simulated responses of cumulative biomass to different combinations
of nitrogen and phosphorus levels in the W a experiment .............................................. 136

4-10 Measured and simulated responses of shoot P concentration to different
combinations of nitrogen and phosphorus levels in the Wa experiment......................... 137

4-11 Variation of the shoot P concentration during plant growth as affected by three
phosphorus levels in the W a experiment. ............................................. ............... 137

B-l Map of the African continent showing Ghana, the country where the field
experim ents w ere carried out.......... ......... ................. .... ................. ............... 144









B-2 Map of Ghana showing the location of the two study sites, Kpeve in the South and
W a in the N north ....................................................................... .... .... .. ...... .... 145









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

COMPARISON OF MEASURED AND SIMULATED RESPONSES OF MAIZE TO
PHOSPHORUS LEVELS IN GHANA

By

Kofikuma Adzewoda Dzotsi

December 2007

Chair: James W. Jones
Major: Agricultural and Biological Engineering

Efficient nutrient management in agricultural systems requires the availability of tools that

can help in meeting research objectives of understanding the transformations that nutrients

undergo to become available to plants and predicting how these transformations are related to

economic outputs from the systems. The crop models in the Decision Support System for

Agrotechnology Transfer (DSSAT) have been recognized worldwide for meeting these

objectives for nitrogen. However, without a phosphorus model, the applicability of the DSSAT

crop models in phosphorus deficient environments will remain questionable. In this study, a soil-

plant phosphorus model linked to DSSAT was described, analyzed and tested.

The sensitivity of the model to six key input factors was studied based on a global

sensitivity analysis approach. The model was tested on two P-deficient soils from Ghana (Kpeve

and Wa) with maize as the test plant. Processes accounted for by the model include phosphorus

movements between inorganic (labile, active and stable), organic (active and stable) pools and

plants. Results of the sensitivity analysis showed the greatest effects of initial inorganic labile P

(initial PiLabile) and fertilizer P on biomass, grain yield and total P uptake (sensitivity index of

0.11 for initial PiLabile and 0.30-0.43 for P fertilizer). Smaller effects were found for the fraction









of root labile P that is soluble (sensitivity index of 0.03-0.04), the shoot P (sensitivity index of

0.03-0.09) and seed P (sensitivity index of 0.15) on total P uptake.

Statistical analysis of grain yield and biomass did not reveal any significant differences at

the 0.05 probability level at Kpeve because the phosphorus content of this soil was at the limit

between deficiency and sufficiency and the organic matter content of the soil was relatively high

(close to 2.0%). Grain yield and final biomass responded at Wa with 100% increases in the 60 kg

[P205] ha-1 treatments over the nil-P treatments. Biomass and yield were stable between the two

treatments of 60 and 90 kg [P205] ha-1 at Wa.

Evaluation of the model indicated that the model was able to achieve good predictability

skill at Kpeve with a grain yield RRMSE of 8% and a final biomass RRMSE of 5%. The

congruence between simulation and measurement was fair at Wa. The RRMSE was 14% for

grain yield and 30% for final biomass. At Wa however, the model gave a reasonable prediction

of the pattern of variability among measurements with an LCS averaged over the five sampling

dates of 17%. Because the complex soil P chemistry makes the availability of phosphorus to

plants extremely variable in general, further testing of this model in other agro-ecological

conditions should precede its application.









CHAPTER 1
INTRODUCTION TO MODELING PHOSPHORUS LIMITATIONS TO CROP
PRODUCTION

Introduction

Phosphorus (P) is recognized as a major nutrient that must be present in living organisms

to enable them to maintain a continuous life cycle. It is an essential component of adenosine

triphosphate (ATP), the energy currency of the living cell. The energy-consuming biochemical

processes that continuously take place in the cell are driven by the energy-rich phosphate group

contained in ATP. For example, in crop nutrition, the uptake and assimilation of nutrients use

energy in the form of ATP. The synthesis of new molecules responsible for mass accumulation

in living organisms and perpetuation of species on earth all involve P either as ATP or

deoxyribonucleic acid (DNA) or ribonucleic acid (RNA). Phosphorus is transferred from one

organism to another through the various food chains. For terrestrial organisms, soils satisfy

most of their P need mainly through plants (Johnston, 2000). The P content of healthy plant

leaf tissue is low however, ranging from 0.2 to 0.4% of the dry matter (Brady and Weil, 2002).

Although plant uptake of P is constrained by the low quantity of this element present in

soil and the very low solubility of P compounds found in soils, most natural ecosystems have

developed relatively well without any P management programs (Brady and Weil, 2002). These

systems are naturally organized to recycle the nutrient and maintain an overall non

thermodynamic equilibrium.

Phosphorus Problem in Agricultural Systems

In agricultural systems where most nutrients' balances have been displaced by human

intervention, the relatively low mobility of P in many soils has led to the appearance of areas of

soil accumulation and depletion of this nutrient.









Most soils in sub-Saharan Africa have very little capacity to supply P for plant growth,

which allows plants grown on those soils to be responsive to P fertilizer applications (Table 1-

1). Phosphorus deficiency is thought to be one of the reasons why sub-Saharan Africa is the

only major region in the world where per capital food production has actually declined in the

past three decades (Brady and Weil, 2002). The phosphorus problem in most sub-Saharan

African soils has five facets:

* The soils have developed under conditions conducive to advanced weathering. During
these relatively long periods of intensive weathering, extensive losses of P occurred
resulting in low P soils. Most soils' solution P ranges from 0.03 to 0.50 ppm with 0.25
ppm considered adequate. For a crop requirement of 40 kg [P] ha-1 for example, the soil
solution containing 0.25 ppm must be replenished 80 times in a hectare furrow slice (15
cm deep x 1 ha area), which does not happen naturally;

* The P compounds commonly present in soils are highly insoluble and have a very low
diffusion rate in many soils posing a problem for plant uptake. They do not readily
release P to the soil solution in a useable form by plants. For example, P fixed by
reaction with aluminum in acid soils is insoluble for plant uptake. The readily available
pool of P that is in equilibrium with the soil solution P (Figure 1-1) can be as small as
10% in some soils (Table 1-2). The rate of diffusion of P in some soils can be as low as
10-12 to 10-15 m2 s-1 and high plant uptake rates create a zone around the root that is
depleted ofP (Schachtman et al., 1998);

* When soils are supplied with external P in the form of fertilizers, the nutrient is fixed,
adsorbed or absorbed and with time tends to return to stable forms, strengite, variscite (in
acid soils) and apatite (in alkaline soils) (Figure 1-1). As a consequence P fertilizer
recovery is low in most agricultural systems relative to the other major nutrients
(nitrogen and potassium);

* Crop harvest exports significant amounts of P from the soil with limited amounts of
residues returned to the cropping system;

* Use of external P inputs in the form of mineral fertilizers or manure, especially for food
crops, is not common practice. Farmers do not have access to the appropriate P
fertilizers, or the cost of their being transported and applied is prohibitive. In addition,
the fertilizer requirement for improved yield can be high on soils with high P sorption
capacity (Table 1-1).

In industrialized areas, P fertilizer use has increased drastically during the past few

decades (FAO, 2003). The relatively low plant uptake of P coupled with the low mobility of









the nutrient in some soils mean that much of the P fertilizer applied is not removed with the

harvested crop or lost from the soil (Schmidt et al., 1996). In fact, soils in these areas have

developed rather high P levels resulting from many years of over-fertilization with P.

Understanding Excess and Deficiency of Phosphorus in Agricultural Systems

The challenge of dealing with both the excess and deficiency of P in agricultural

ecosystems is crucial to attempt to restore the P balance in these systems and make P

management programs sustainable and environmentally sound. A primary step towards

tackling the challenge of P management in agricultural systems is understanding the P behavior

in soils in relation to plants' needs and their environment.

Excess P can be detrimental to the aquatic ecology. Although plant proliferation

stimulated by supply of limiting nutrients is considered beneficial in terrestrial ecosystems,

aquatic systems like lakes, streams and ponds can become unsatisfactory environments when

enriched with excessive P through runoff, erosion and, in some cases, leaching. The unwanted

growth of algae and of aquatic weeds (termed eutrophication) resulting from this P enrichment

can seriously perturb the aquatic ecosystem. When this community of opportunistic algae and

weeds die, they sink to the bottom of the water where their decomposition by microorganisms

uses much of the oxygen in the water creating anoxic conditions. This process leads to fish

kills, displaced nutrients balance, and can make the water unsuitable for drinking (Brady and

Weil, 2002; Sturgul and Bundy, 2004).

Phosphorus deficiency can constitute a serious problem for crop production because it

has a negative effect on leaf area index (Pellerin et al., 2000) limiting the interception of

photosynthetically active radiation by the plant and resulting in low biomass accumulation

(Colomb et al., 2000). The rate of leaf appearance is slowed down and the final leaf number is

reduced in P stressed plants (Singh and al., 1999). Colomb et al. (2000) showed that in P-









deficient maize plants in their study the rate of leaf appearance and the final area of leaves

located below the main ear were reduced by 18 to 27%. The ultimate economic effect of P

deficiency is yield reduction (Table 1-1).

Coping with Excess and Deficiency of Phosphorus in Agricultural Systems

Extensive and long term agronomic experiments have been conducted to attain a greater

understanding of the behavior of soil P and propose options for P management in agricultural

systems. The mechanism motivating and the time factor associated with P draw-down or build-

up in soils have been examined as steps towards assessing opportunities for reducing P

loadings in waters (Kelling et al., 1998; Sartain, 1980). Management strategies proposed for

decreasing the P content of high P soils include mining soil P (i.e., harvesting P taken up from

the soil by a crop grown without external P addition) (Koopmans et al., 2004), growing

appropriate corn varieties as P removal agents (Eghball et al., 2003). With the twofold concern

of replenishing soil phosphorus in P deficient agricultural systems while avoiding losses to

aquatic ecosystems, continuous applications of small rates of P have been proposed as adapted

management strategy for smallscale farming systems (Nziguheba et al., 2002; Schmidt et al.,

1996).

Modeling as a Phosphorus Management Tool in Soils and Plants

The necessity of developing P management strategies requires the availability of

appropriate tools that empower managers and decision-makers with the ability to control the

human-modified P cycle in agricultural systems. Statistical summaries have been routinely

used to produce in an integrated way a logic interpretation of agronomic experimental results.

However, these parameters and other classical mathematical methods used to study and explain

the behavior of nonliving physical or chemical processes may be insufficient (Jones and

Luyten, 1998). Agricultural ecosystems involving biological processes are highly complex and









have many components that interact in non linear ways. The non linear interaction means that

many disciplines working, for example, on the same P management problem may be looking at

the facets that are only meaningful for their study while interacting components that may

provide clues to the solution are not given enough attention. An interdisciplinary approach that

places the P management problem at the center of the soil-plant-atmosphere system and

recognizes the effects on the problem of interactions between disciplines concerned is useful

for developing efficient management tools. An ideal P management program is at the minimum

concerned with i) understanding the behavior of P in agricultural systems; ii) synthesizing the

knowledge and information obtained in an integrated way so that interactions occurring in the

system are not lost but harnessed to enhance understanding; iii) producing user-friendly

management tools that depict the best understanding of the system. Well-tested simulation

models that represent the cropping system with mathematical relationships provide a sound

scientific approximation of physical, chemical and biological processes governing complex

ecological systems and represent such tools. When appropriately validated, those simulation

models provide the opportunity to understand simplifications of the universe (Odum and

Odum, 2000); study ecosystems without having to experiment on actual systems (Uehara,

1998) especially when experimentation is impossible or ethically unacceptable; make

predictions; support decision making and communicate more efficiently research findings by

integrating information into a more useable form (Newman, 2000).

Phosphorus modeling has given attention to soil and plant processes that affect the P

cycle. Residual effects of soil P have been of interest because in many soils, the P that is

applied to the soil but not taken up by the plant is not lost, and the build-up in the soil toplayer

can be reused for crop production (Pheav et al., 2003) and thought of as a capital investment.









Models now exist that address the issues of long term recovery of applied P in the form of

fertilizers (Wolf et al., 1987; Janssen et al., 1987; Schmidt et al., 1997), long term changes in

soil P extracted using conventional methods (Karpinets et al., 2004), long term P leaching from

the soil profile (Del Campillo et al., 1999), and long term effects of erosion-induced soil

nutrient loss, including P, on crop productivity (Jones et al., 1984). Lewis and McGechan

(2002) compared four soil P models, AMINO from the Netherlands, GLEAMS and

DAYCENT from the USA and MACRO from Sweden, in order to ascertain their limitations

and evaluate their capability to simulate the transport of soluble and particulate P, surface

application, mineralization / immobilization, absorption / desorption, leaching, runoff and

uptake by plants. The P module of GLEAMS is actually an essentially unaltered version of the

model developed by Jones et al. (1984). Lewis and McGechan concluded from their analyses

that all the models only have a partial representation of the soil processes examined. They

suggested that more accurate dynamic simulations of soil processes will necessitate a hybrid

model that incorporates the different aspects of soil P dynamics that the models studied have

failed to critically account for.

The most relevant processes for crop production in P deficient systems include the

quantification of development, growth and yield as limited by P. These processes can only be

handled by comprehensive simulation models of crop growth and development. For the

purpose of enhancing the applicability of a model of this kind, soil P gains by the plant

resulting from the mineralization of organic matter, particularly in low input cropping systems

must be recognized in addition to P obtained from chemical fertilizers (Probert, 2004). In fact,

organic materials are used in many competing ways in smallscale cropping systems including

soil fertility replenishment (Fofana et al., 2005). Factors controlling the decomposition of









organic materials in these systems often favor faster mineralization rates of nutrients. The

addition of limited amounts of fertilizer tends to offset the negative effect of low-quality

organic materials and accelerate their decomposition. Nutrients that would normally cycle over

extended periods could therefore be released over a relatively short time, increasing total

available nutrients for plant uptake (Kaboneka et al., 2004).

Soil-Plant Phosphorus Simulation Model in DSSAT

To meet these needs, a capability is needed in the Decision Support System for

Agrotechnology Transfer (DSSAT) Cropping System Models (CSM) (Jones et al., 2003) to

model i) P limited crop growth, development and yield and ii) P released by organic resources.

The soil-plant P module that has been linked to the DSSAT CSM and still operates as an

experimental version is based on studies by Daroub et al. (2003). A new Soil Organic Matter-

Residue module called CENTURY that accounts for nutrient mineralization from organic

resources was recently implemented in DSSAT. Gijsman et al. (2002) reported that the

CENTURY module simulated with high accuracy the development of SOM content in a long

term bare fallow experiment in Rothamsted, UK and gave a fair congruence between simulated

and measured data for a 1-year experiment in Brazil.

The Decision Support System for Agrotechnology Transfer is comprised of a suite of

models that simulate on a daily basis the development, growth and yield of more than 16 crops.

The models are organized under different groups. Among these groups, the CERES family for

cereals and the CROPGRO family for legumes are important components that have proven to

be successful in their applications worldwide. The system also contains programs that allow

the analysis of effects of multiyear variations of factors like weather on crop production. Other

programs permit the analysis of rotational and spatial experiments. The DSSAT models use

quantitative climate, soil, genetic and management information as inputs to simulate, among









many other outputs, grain yield and its components, crop biomass, anthesis, silking and

physiological maturity dates. The Decision Support System for Agrotechnology Transfer has

been used in several studies that include residues and inorganic nitrogen management in

Nigeria (Jagtap and Abamu, 2003), crop management strategies in extension systems in Kenya

(Wafula, 1995), investigating variety and sowing time technologies in Nigeria (Jagtap, 1999)

and Togo (Dzotsi et al., 2003), studying the effect of water and nitrogen deficiency on crop

duration and yield in Florida, Hawaii, Nigeria and Togo (Singh and al., 1999), analyzing soil

fertility research and development options in Malawi (Singh et al., 1993).

Objective and Hypothesis

The overall objective of this study is to present the soil-plant phosphorus model

implemented in the DSSAT CSM for maize and results of field testing. The main hypothesis

was that P fertilization increases soil inorganic P availability for plant uptake, which promotes

higher grain yield and biomass production and shortens the time required for ripening in maize.









Table 1-1. Response of maize to phosphorus application on a phosphorus-deficient soil in a
fertilizer experiment carried out in Ghana in 1999
Source P P205 Grain Yield
of phosphate (kg ha-1) (kg ha-1) (kg ha-1)
Control 3949
Triple super phosphate 40 92 5135
Togo rock phosphate 63 144 6252
Source: Adapted from FAO, 2005

Table 1-2. Partitioning of total soil phosphorus in pools specified on Figure 1-1 in a soil from
Carimagua, Colombia
Soil Parts per million Per cent of kg ha- in an
phosphorus (mg kg-1) total soil P (%) 1ha-15cm deep soil
Readily available 18 10 36
Stable forms 78 43 156
Organic 86 47 172
Total soil P 182 100 364









Residue


Uptake


Fertilizer

Leaching (minor)


dissolution I. .. I

Figure 1-1. Pools of phosphorus in the soil and relationships between soil and plant phosphorus









CHAPTER 2
STATISTICAL ANALYSIS OF FIELD EXPERIMENT FOR TESTING THE MODEL

Introduction

Assessing plant response to phosphorus (P) is an important step towards understanding its

behavior in agricultural systems. Plant response to P can be evaluated using different soils with P

levels ranging from low to high (Colomb et al., 2000) or by testing different applications of a P

fertilizer on the same P deficient soil (Fofana et al., 2005). In both situations initial soil testing

for P is essential to determine the P status of the soil of interest.

The diagnosis of P level in soils is complicated by the complex chemistry of the nutrient.

This complexity is the basis for assessing soil P content using extractants for their effectiveness

to solubilize P tied up in different forms. The quantity of P extracted will vary with the reagent

used. However, many P extraction methods are widely accepted and used because they

adequately distinguish between soils on the basis of the responsiveness of crops to P supply. Soil

P test does not provide information about the available P that can be actually taken up by crops

but relates to that quantity of P which is correlated with plant response. This implies that soil

analytical data allow the classification of soils descriptively in terms of P availability (e.g.

deficient, sufficient, high) but these classes are only related to the probable response of a crop to

an appropriate supply of P (Johnston, 2000). The relationship between soil P test and plant

response cannot be established as necessarily deterministic. For example, if a soil tests very low

in P, a 75% probability exists that plant response will be observed. If the soil test is low in P,

there is a 50% chance that plants will respond to P applications. If the soil P level is medium, the

response probability of plants is only 25%. Plants will not be responsive to P if the soil test

indicates a high P level (Havlin et al., 1999).









Inorganic and organic fertilizers can be used as P sources in experimenting with P-

deficient soils. The amount of P provided by the decomposition of organic materials over a

cropping season can be small and relatively slow however, but continuous over several years.

Inorganic fertilizers on the contrary can supply more P to the plants at a higher rate and over a

relatively short period of time, a cropping season for instance. However, a big portion of the

fertilizer applied to plants in cropping systems is retained in the soil, which constitutes the cause

of long term accumulation of P in soils with a history of continuous P fertilization. The fraction

of P applied in the form of inorganic fertilizer that is actually taken up by plants is about 0.2.

This fraction is termed apparent recovery of the fertilizer. An appropriate art of managing P in

cropping systems would be to integrate the use of organic resources with inorganic fertilizers so

that soil organic matter can play its role of improving physical and biological properties of soils

while immediate nutrients needs are satisfied by inorganic fertilizers (Janssen, 1993).

Plant response to P has been described as increasing the leaf area therefore allowing a

higher interception of photosynthetically active radiation and resulting in a higher biomass

accumulation and grain yield (Pellerin et al., 2000; Colomb et al., 2000; Plenet et al., 2000b).

Other studies reported that the rate of leaf appearance was slowed down and the final leaf

number was reduced in P-stressed plants (Plenet et al, 2000a; Singh and al., 1999).

The present chapter describes two field experiments conducted in Ghana in 2004 and 2006

on soils that tested low in available P and therefore considered P-deficient. Statistical procedures

are used to summarize the data collected and understand the plant response observed.

Materials and Methods

Field experiments were conducted in Kpeve and Wa, Ghana in order to measure growth

and development of maize as affected by P fertilizer. These experiments are described next with

the statistical methods used to analyze them.









Field Experiments in Ghana

The experiments were carried out in two different agro-ecological zones of maize

production in Ghana. The Kpeve site is located in the south in the Transitional zone with two

distinct rainy seasons and annual rainfall ranging from 1100 to 1400 mm. The Wa site is located

in the North in the Guinea Savannah with one rainy season and annual rainfall ranging from 800

to 1200 mm (FAO, 2005).

Experiment in Kpeve, Ghana

The experiment in Kpeve measured the effect of different levels of phosphorus fertilizer on

the growth and development of the maize cultivar Obatanpa (Table 2-1).

Site description

The experiment was conducted in 2006 during the major rainy season (March to July) on a

sandy loam soil at the experimental site of the Ministry of Food and Agriculture research station

at Kpeve in southern Ghana (6 40.80' N, 0 19.20' E). The site is characterized by an altitude of

67 m above sea level, an average annual temperature of 28 degrees C and an average annual

rainfall of 1300 mm falling in two rainy seasons, March to July and September to October (FAO,

2005) (Figure 2-1). The landscape is highly uneven with chains of hills surrounding the

experimental station. Although the topography of the experimental site was almost flat, small

micro-topography differences may have important water and nutrient management consequences

in this type of terrain. This topography reinforced the need for treating the experimental plots

individually regarding all data collected. The soil is classified as Haplic Lixisol which has a dark

grayish brown topsoil and grayish brown to brown subsoil (Adiku, 2006). An automatic weather

station to monitor maximum temperature, minimum temperature, and solar radiation and an

automatic rainfall datalogger to record daily rainfall were located respectively at 1 km and 200 m

from the experimental field.









Experiment design and management

Maize (Zea mays L. cultivar Obatanpa "a good nursing mother", Table 2-1), was planted

on May 27. Seven days prior to planting, 5206 kg ha-1 of vegetation of a two-year natural bush

fallow dominated by elephant grass (Pennisetum purpureum) and guinea grass (Panicum

maximum) was plowed into the soil to a depth of 30 cm. The field was hand-harrowed to a depth

of 10 cm and leveled three days before planting. Three levels of P and a control treatment were

explored: low (10 kg P ha-1), medium (30 kg P ha-1) and high (80 kg P ha-1).

A total of 30 kg K ha-1 was applied as Potassium Nitrate two times during the growing

season, at planting and two weeks after planting. A total of 150 kg N ha-1 was applied at the rate

of 50 kg ha-1, at planting, four weeks after planting and six weeks after planting as Ammonium

Sulfate. The Potassium Nitrate provided 10 kg ha-1 of the total 150 kg ha-1 of the nitrogen

applied. The application methods varied with the growth stage (Table 2-2).

The existence of a slope gradient on the experimental field motivated the arrangement of

the treatments in a Randomized Complete Blocks design with four replications. Each plot or

experimental unit was composed of 14 rows 80 cm apart. Each row contained 30 hills 40 cm

apart making up a total of 420 hills per plot. The total area of a plot was 134.4 m2. At planting

and emergence, the plant population was 9.38 plants m-2. The field was thinned 10 days after

planting to reduce the plant population to 6.25 plants m-2. The application of sufficient rates of

nitrogen (150 kg ha-1) and potassium (30 kg ha-1) and the control of soil variability through

blocking were expected to highlight the effect of P deficiency in the crop.

Soil sampling

Soil samples were taken at three depths, 0-10, 10-20 and 20-30 cm on all 16 plots before

planting, at silking and at final harvest. Texture, organic matter content, phosphorus content,

exchange complex and acidity were determined on the samples in the soil testing laboratory at









the University of Ghana. A modified Hedley approach for P fractionation was used to quantify

inorganic labile, microbial and stable, and organic P in the samples. The P fractions (Table 2-17)

were performed in the Wetland Biogeochemistry Laboratory at the University of Florida

following a four-step sequential extraction described by Reddy et al. (1998) (Figure 2-2). The

extractants in order were: 1, 1.0 M KC1; 2, 0.1 M NaOH; 3, 0.5 M HC1; 4, Residual P. The

extraction with reagent 1, potassium chloride removed that portion of P readily available to

plants. The alkaline reagent (NaOH) extracted P associated with iron and aluminum while that

extracted by reagent 3 (HC1) is probably associated with calcium (Figure 2-2). In the solution

extracted by the alkaline reagent, both organic (Po) and inorganic P (Pi) were determined. The

residual P in the soil was recovered after combustion at 550 C for 4 h and dissolution in 6.0 M

HC1 (Reddy et al., 1998).

Soil moisture measurements

Soil moisture is generally determined from oven-dried soil samples at 105 C until constant

weight. The moisture difference between the fresh and the dried soil relative to the dried soil is

established as the gravimetric soil water content. The gravimetric soil water content can be

further converted into volumetric soil water content by multiplying it by the bulk density of the

soil.

Monitoring soil water using gravimetric sampling can be tedious and not practical

especially when the desired frequency is two to three-day intervals. Time domain reflectometry

(TDR) technology is one of the best methods to quickly and accurately measure soil moisture.

The technique is based on generating and remotely sensing a return energy signal that travels

down and back through the soil. The travel time measured is dependent on the quantity of water

present in the soil. This information is then converted into volumetric water content. Because

soils have different properties that can influence the way TDRs "capture and read" the moisture









status of the soil, the calibration of the meter to field conditions is an important step towards its

use for extensive applications.

In the present study, soil moisture was monitored during the entire course of the

experiment at four GPS-referenced prelocated points on each plot in the top 12 and 20 cm using

a portable soil moisture meter (FieldScout TDR 300) manufactured by Spectrum Technologies,

Inc. The TDR readings were taken at 2 to 6-day intervals.

In order to calibrate the TDR, 67 pairs of TDR readings and gravimetric soil samples were

separately taken at anthesis at three soil moisture conditions, low, medium and high. Fresh

weight and soil volume were determined on the samples prior to drying. After oven-drying at

105 C until constant weight, the fraction of the gravel was determined on each sample. The data

were used to correct the bulk density of the soil using the following formula (Vincent and

Chadwick, 1994):


Corrected Bulk Density (g cm-3) = Gf )BD (2-1)
S (1 Gj)
Where Gf is the fraction of gravel in the soil sample, on a mass basis;
-3
BD, is the uncorrected bulk density in g cm-3
G, is the volume of gravel in the sample, expressed as a fraction of the total volume and
(BDm Gf)
calculated as follow: (BD
2.65
-3
2.65 is the density of solid particles in the soil expressed in g cm-3

Plant sampling and growth measurements

Detailed measurements of leaf area, plant height, dates of tasseling, anthesis, and silking

were made throughout the growing season.

Aboveground biomass samples were taken four times during the season, 17 days after

planting (dap), 31 dap, 52 dap anthesiss), and 108 dap (final harvest).

After emergence 6 plants were randomly selected and tagged within three rows on each

individual plot. Length and width of each expanding leaf were measured at 7-day intervals until









maximum values were reached or 50% leaf senescence observed (Colomb et al., 2000). At any

time during the season, total area of leaves produced by the plant was computed as the sum of

individual leaves' areas. The area of a single growing leaf was calculated as the product of length

and width multiplied by 0.75 (Colomb et al., 2000). Visible leaf numbers, plant height and

phenology events tasselingg, anthesis and silking) were recorded from emergence to silking for

the six tagged plants on each plot.

Plant height was taken from the base of the plant to the tip of the most recent leaf.

Tasseling, anthesis, and silking dates were established as the date when 50% (three out of the

six) of the tagged plants tasseled paniclee visible, tasseling), showed some pollens or anthers

anthesiss), or showed some silks silkingg).

The samples taken for aboveground biomass determination consisted of randomly

prelocated 12 plants from two continuous rows corresponding to a sampling area of 1.92 m2. The

dry weights of each plant part, stem plus petiole, leaves, fruit and grains, were determined on

each sample by oven drying at 60 degrees C for 48 hours or until constant weight was reached.

Biomass accumulation from week six to maturity was so high that for ease of handling, only

fresh weights were determined on the 12 plants. A subsample of 6 plants (17 dap, 31 dap and

anthesis) and 5 plants (final harvest) was used for dry weight measurements of each plant

component. The samples were analyzed at the University of Ghana for total N, P and K content

in each plant part.

An area of 6 mx 1.6 m = 9.6 m2 was used for final harvest on each plot, 108 days after

planting (September 12th). The two innermost rows harvested were bordered by two rows on

each side. Grain yield, total aboveground biomass, stover biomass, grain number per m2, unit

grain weight, and N, P and K content of grain and stalk were determined.









Experiment in Wa, Ghana

The experiment in Wa measured the effect of combinations of nitrogen and phosphorus

fertilizers on the growth and development of the maize cultivar Obatanpa (Table 2-1).

Site and experiment set up

The experiment was carried out in 2004 at the Savannah Agricultural Research Institute

(SARI)'s experimental station in Wa, Ghana (10o3' N, 2o30' W, altitude 320 m above sea level)

by Naab (2005). The average annual rainfall is 1100 mm falling mainly between April and

September (Figure 2-3). The mean annual temperature is 27 C.

The experimental field was cleared of native vegetation that was plowed in. The field was

harrowed and laid out in a Randomized Complete Block Design with four replications.

Treatment plots measured 6 m x 8 m. The factors tested were: nitrogen at three levels, 0, 60 and

120 kg N ha-1 (NO, N60, and N120 respectively); and P also at three levels, 0, 60 and 90 kg P205

ha-1 (PO, P60, and P120 respectively). Initial P content of the soil was measured on soil samples

taken at planting.

Maize was sown on June 17th at a spacing of 70 cm x 40 cm. A pre-emergence herbicide

(Roundup or glyphosphate) was applied a few days after sowing.

All of the phosphorus was broadcast as Single Super Phosphate and incorporated by hand

hoeing to a depth of 5 cm in all treatments, two days before sowing. Nitrogen was split-applied

as urea at the bottom of 5-cm holes near the maize stands, 2 and 6 weeks after planting.

Field and laboratory measurements

Phenological observations on the number of days to emergence, tasseling, silking, blister

stage, milk stage, dough stage, dent stage and physiological maturity were made on the middle

four rows of each plot. The dates were established as corresponding to the time when 50% of the

four rows sampled on each plot reached the different stages of interest.









Plant samples were taken five times randomly on the plots during the growing season.

Each sample was obtained from a 0.8 m x 1.7 m = 1.12 m2 area (two rows) that yielded about 8

plants. A sub-sample of 2-3 plants was taken from each sample for an oven-drying dry matter

determination at 700C for 48 hours.

Leaf area index was directly measured on randomly selected plants using a Delta-T

SunScan Canopy Analyzer.

Maturity total biomass and number of cobs were determined on the four middle rows (12

m2) used for phenological observations. Sub-samples were taken for dry matter estimation of

stover, grain yield and components, and 100-seed weight.

Statistical Analysis

Statistical analysis of the data obtained from the experiments in Kpeve and Wa was

conducted using the SAS (SAS, 2002). A regression analysis was used to analyze soil moisture

data collected at Kpeve and analysis of variance was used to summarize and understand the

growth and development data from both experiments.

Regression analysis (soil moisture)

A regression analysis was carried out using the SAS software (SAS, 2002) to verify how

well a linear regression model can be fitted to the dataset composed of the 67 pairs of TDR

readings and soil moisture values obtained from the gravimetric samples. The purpose of the

regression analysis was to derive an equation that can be used to convert TDR readings into

volumetric water content of the soil. Tests of slope and intercept were carried out and a

regression equation was set up that allowed the estimation of volumetric soil moisture values

from specific TDR readings.









Analysis of variance at individual time points

Analysis of Variance (ANOVA) F-test was used to test the effect of P fertilizer application

on crop phenology, grain yield, aboveground biomass, height, green leaf area, and soil moisture

readings using TDR. This analysis was carried out at specific time points when the data were

obtained. Hypotheses that were tested are described here.

Crop phenology. Inadequate supply of P has been reported to delay silking in maize

resulting in an increase in anthesis to silking interval. It was hypothesized that increasing P

fertilizer levels will result in early tasseling, anthesis and silking. These events should be delayed

in no and low P treatments.

Green Leaf Area. Maize grown on low P soils has been reported to have access to a

limited amount of Photosynthetically Absorbed Radiation (PAR), which reduces the area of

expanding leaves (Pellerin et al., 2000). The green leaf area in the Kpeve experiment and the leaf

area index in the Wa experiment were hypothesized to increase with higher levels of P fertilizer.

Aboveground biomass. The accumulation of aboveground biomass is directly related to

the amount of PAR intercepted by the crop canopy and should be affected by P deficiency in the

same way as green leaf area or leaf area index.

Grain yield. An increase of anthesis to silking interval or a delayed silking will result in a

low grain number per square meter and a low grain yield (Plenet et al., 2000b). Growth deficit

due to insufficient biomass accumulation can also affect negatively grain formation.

Height. Lack of adequate biomass accumulation and energy for physiological processes

can result in stunted plants. Higher values of plant height are expected in P fertilized plots.

Aboveground biomass, height, green leaf area, leaf area index and soil moisture data

were collected on the same plots, plants, and locations on the plots over time and were

considered repeated measurements. Preliminary analyses of these data were first carried out at









individual time points. The individual time point analysis was used to examine treatment effects

at specific sampling dates only. However, these analyses were one way ANOVAs considering

each time point separately, as independent from each other, and did not make comparisons

among different sampling dates.

Analysis of variance considering the effect of time on the repeated measurements

The individual time point analysis was extended using a repeated measures technique to

account for the effects of time on the response variables taken in sequence over time during the

growing season. More information can be derived from repeated measures than revealed by

individual time point analysis ANOVAs: comparisons of treatments averaged over time and

comparisons of times within a treatment are also informative.

When measurements (of height e.g.) are repeated on the same subject (e.g. plant) at

specific time intervals (e.g. every 2 weeks, during the growth of the plant), the data are generally

viewed as coming from a factorial experiment with treatments and time as the factors, and

analyzed as if they came from a split-plot design because most statistical packages do not

provide users with the capability of accounting properly for the effects of time. In this example,

the plant would be considered as the whole-plot unit, and plants at specific times as the sub-plot

unit. This method is known as the split-plot in time approach to analyzing repeated

measurements (Littell et al., 1998). The assumptions supporting this split-plot in time approach

are that variances of measurements taken at different times are equal and that pairs of measures

coming from the same plant are equally correlated. This means that the correlation pattern

among the measurements taken on the same plant is not affected by time. The split-plot in time

analysis would have been optimal if the assumptions could be fully met in all circumstances. The

peculiar property of repeating the measurement on the same plant means that sets of data from

the same plant, though taken at different time points, are not independent. They include a









covariance structure resulting from differences between plants (between plants variation) and

differences between times on the same plant (within plant variation). The covariance structure

refers to two things: 1) variances in the data collected on the same plant at individual time points

and 2) correlation between measurements taken on the same plant at different times. Littell et al.

(1998) underlined the two aspects that are important to the correlation. First, two measures taken

on the plant are correlated simply because they share common contributions from the same plant.

Second, measures on the same plant close in time are often more correlated than measures far

apart in time.

This covariance structure is not captured by the common ANOVA implemented in SAS

with the general linear model PROC GLM.

In order to model the covariance structure related to the effect of time in this study, we

used the PROC MIXED procedure now available in SAS since 1992 (Littell et al., 1998).

For comparison purposes we present results of the split-plot in time method (also called

univariate ANOVA) and two covariance structures, summarized with their specifications in

Table 2-3. The Akaike Information Criterion (AIC) was used as goodness of fit criterion to select

the appropriate covariance structure for this study. The AIC is presented with the SAS output

when PROC MIXED is run. The smaller the value of AIC, the better the structure.

The repeated measures analysis technique was applied only to green leaf area and height

measurements taken from day 17 to day 52 after planting. Biomass and soil moisture were not

analyzed using this technique because their measurements were taken at variable time intervals.

Results and Discussion

Response of maize biomass and grain yield to P fertilizer was not observed at Kpeve. At

Wa, the plant response was measured not only on biomass and grain yield but also on phenology.









The analysis of the soil moisture data at Kpeve yielded an equation for prediction volumetric soil

moisture from TDR readings.

Calibration of the TDR Meter

The soil moisture readings taken with the TDR meter appeared to be linearly related to

the gravimetric sampling moisture determinations (Figure 2-4).

There is sufficient evidence to suggest that the relationship between the two methods of

soil moisture determination is linear: an analysis of variance of the simple linear regression

model was highly significant and the mean squared error was very low (Table 2-4). Seventy

three percent of the variation in the meter readings was accounted for by the gravimetric

samplings (coefficient of determination R2 = 0.73). The regression equation model relating the

volumetric soil moisture to the TDR reading is:

volumetric = 4.90 + 0.72 x tdr (2-2)
Where volumetric is the volumetric soil moisture measured using gravimetric sampling;
tdr is the volumetric soil moisture read by the TDR meter;
The coefficient 0.72 (slope in the regression equation) is an estimate of the rate of increase
in gravimetric soil moisture for each unit increase in TDR readings. Gravimetric soil
moisture increases by 0.72% for each 1% reading by the TDR.

This equation can serve the purpose of predicting the volumetric soil moisture using the

gravimetric method (considered as the true measurement) from any single or population of TDR

readings. Tests of the slope and the intercept in this equation (Ho: slope = intercept = 0) lead to

highly significant p values for rejecting Ho (Table 2-5).

The regression line plotted on Figure 2-4 would be parallel to the 1:1 line ideally, but it is

slightly more horizontal, thus crossing the perfect agreement line. This illustrates a tendency of

the TDR meter to overestimate soil moisture at high soil moisture status. A possible explanation

is that the meter would continue to read high soil moisture values as long as the rods are inserted

into the soil with a steady, non wiggling downward pressure even if a high percentage of gravel









is present in the profile. The gravel content in the profile (0-30 cm) as measured for correction of

the soil bulk density was between 25 and 45%. Gravel (soil solid particles with size greater than

2 mm) cannot hold water and would reduce the available water for the plant when they are

present in relatively high quantities. The corrected bulk density used in this experiment to control

the effect of the presence of stones on the soil moisture status helped to obtain lower values of

soil moisture using the gravimetric sampling method. The equation established from this

regression analysis is only valid for the type of soil used in the experiment.

Crop Response Results at Kpeve Using Individual Time Points Analysis

The individual time point analysis revealed no significant difference in phenology and

growth at Kpeve except for the height (during the mid-season).

Phenology

The expected trend of P effect on tasseling, anthesis and silking was not observed. There

was no consistent trend depicting the phenological response of the plant to P (Figure 2-5). The

data collected in this experiment did not provide enough evidence to suggest an effect of P

fertilizer on the phenology of maize (Table 2-6). On average, tasseling and anthesis dates

differed among the P treatments by only one day. At silking however, the treatment receiving 80

kg P ha-1 was delayed by 4-6 days compared to the other treatments. The anthesis to silking

interval (ASI) was 9 days on average and higher for the 80P treatment (13 days).

Grain yield and yield components

Both grain yield and stover weight were not affected by the P levels. Grain yield of about

3000 kg ha-1 was attained in all treatments and the average stover yield was 6500 kg ha-1 (Figure

2-6). The grain and stover yields were stable but the grain yield was more variable than the

stover yield (overall coefficient of variation of 27% for grain yield and 15% for stover yield).

This lack of response to P resulted in no statistical significance (Table 2-7).









The grain yield components were also stable between treatments (Figure 2-7). The average

unit grain weight was 0.24 g and the average grain number per m2 was 1500.

Aboveground biomass

Differences between P treatments were not significant at the individual time points

analyzed, but the p-values decreased consistently over time (Table 2-8). The treatment mean

squares increased with time corresponding to increased biomass accumulation with plant growth.

The aboveground biomass as a combination of stover and grain yield, likewise did not respond to

the P fertilizer. The coefficient of variation between treatments varied from 8 to 27% at 17 dap, 7

to 22% at 31 dap, 9 to 15% at 52 dap, and 10 to 19% at 108 dap. However, this variability in

biomass and standard deviations (Figure 2-8) did not result in any statistical significance.

Plant height

Significant differences in plant height were observed mostly during mid-season (Pr = 0.03

at 31 dap, and Pr = 0.01 at 45 dap, Table 2-9). A least significant difference discrimination test

showed that the treatment receiving 30 kg P ha-1 produced the tallest plants, not only at 31, 38

and 45 dap but also throughout the season (Table 2-10 and Figure 2-9). A maximum height of

250 cm was reached after anthesis.

Green leaf area

No statistical significant difference was generally observed among the treatments (Table 2-

11) at the 0.05 level. At anthesis however, the treatments receiving 10 and 30 kg P ha- produced

the highest green leaf area (6000 cm2 per plant or a leaf area index of 3.75 using a plant

population of 6.25 plants m-2) (Figure 2-10) and were statistically different from the treatments

receiving 0 and 80 kg P ha1 at the threshold of a = 0.07.









Soil moisture

The mean differences in soil moisture across the 4 treatment plots as shown on Figure 2-11

were not significant until after the drought spell (i.e. after 58 dap) when the soil started to be

rewetted. There were significant soil moisture variations between blocks at the commencement

of the trial, which justified blocking (Table 2-12). These interblock moisture differences

disappeared however, from 48 dap onwards, at the start of the droughtspell and were not detected

again until final harvest (Figure 2-11). It is noteworthy that when the soil moisture started to go

up again, the significance probabilities for differences between blocks maintained a decreasing

trend.

Crop Response Results at Kpeve Using Repeated Measures Analysis Techniques

The analysis revealed that the autoregressive structure was suitable for the datasets

analyzed. Days after planting had a significant effect on growth but interacted significantly only

with height. Averaged over time, height was the only measured variable that was significantly

affected by P.

Selection of a correlation structure using the AIC

The values of AIC were consistently smaller for the autoregressive structure regardless of

the variable of concern (Table 2-13). This statistic essentially confirmed that the correlation

between pairs of height and green leaf area measurements taken on the same plant and at the

same location on the field decreased with the age of the crop. For example, the autoregressive

structure means that measurements of heights taken at days 17 and 24 after planting are more

correlated than heights obtained at 17 and 52 days after planting on the same plant and at the

same location on the field. Thus, the autoregressive structure was used in this application.









Effect of time on repeated measurements of crop response variables

Days after planting had a highly significant effect on all the repeated measurements

regardless of the covariance structure. (Table 2-14). For plant height for instance, this means that

the height values reached by the plant averaged over the four treatments were statistically

different for days 17, 24, 31, 38, 45, and 52 after planting. This is expected because of the plant

growth and development processes that notably increased the height between the measurement

times.

Phosphorus treatments by time interactions effects on repeated measurements

The interactions involving time and P treatments were not significant for green leaf area

regardless of the structure. The green leaf area curves for treatment 10P and 30P crossed each

other at dap 45 but generally the shapes of the curves were essentially the same for each

treatment (Figure 2-10). There was not enough evidence to suggest that the change over time in

the responses of maize green leaf area was affected by P application.

The interaction between day after planting and P treatment was significant for maize height

based on the autoregressive structure (Pr = 0.0354, Table 2-14). This statistical significance

suggested that the height response curves that could be derived from Figure 2-9 were not the

same. Differences between these responses curves came from the quantity of P applied in each

treatment.

Effect on repeated measurements of phosphorus treatments averaged over time

The effect of P treatments on green leaf area averaged over sampling dates 17 through 52

days after planting was not significant (Pr = 0.3, Table 2-14) based on the autoregressive

structure. This means that the overall effect of P treatments on maize green leaf area as tested in

this study was not important at a = 0.05. This finding is a confirmation of the results obtained









when the ANOVA was performed at individual time points: no statistical significance at a = 0.05

was found at all sampling dates (Table 2-11).

Significant effects of P treatments on maize height averaged over sampling dates 17

through 52 days after planting were found using the autoregressive structure (Pr = 0.0228, Table

2-14). This suggested that plant height measurements taken at weekly intervals over the period

17 to 52 days after planting were significantly different between P treatments. This general

conclusion on maize height response to P applications in this experiment was expected because

the individual time point analysis showed significant differences between the P treatments at

days 31 and 45 after planting and low probabilities for these differences at days 17, 24, 38 and 52

after planting (Table 2-9).

The importance of the choice of an appropriate correlation structure for the analysis of

repeated measurements is highlighted by the contradictory results obtained with the univariate

ANOVA and the compound symmetry structure (Table 2-14). For example, significance

probability values produced by the univariate ANOVA were generally low for the effects of P

(Table 2-14), suggesting strong evidence of differences in green leaf area between the four P

treatments. We already knew that this was not true because most of the p-values obtained from

analyses at individual time points were relatively high (Tables 2-9 and 2-11). The choice of a

different correlation structure would lead to different conclusions about the effect of P on the

green leaf area and height of maize averaged over six measurement dates.

Discussion of Results Obtained at Kpeve

The lack of response of phenology, biomass, yield and yield components of maize to P

fertilizer in the Kpeve experiment was because P did not limit plant growth and development in

the experiment. Since all other major nutrients and water (at least until anthesis) were supplied in

sufficient amounts, it is reasonable to suggest that the plants had access to and were able to take









up P from an adequate supply of indigenous P throughout the growing season. Although the soil

P level (Bray 1) was apparently low (Tables 2-15 and 2-16), there are at least five problems

associated with relying on the classification in Table 2-16 alone to draw conclusions about the P

status of the soil:

* Problem 1: the Bray-1 P method does not measure P available for plant uptake but only
that amount of P that would probably correlate with plant growth (Johnston, 2000);

* Problem 2: the P level in the soil top 20 cm used in this experiment is close to the
sufficiency level of 16 ppm as defined by Shapiro et al. (2003) (Table 2-16). Since P exists
in the soil in many forms that exchange P between each other, it is not clear how P would
behave in the soil at the boundary between sufficiency and deficiency. Other studies (for
example Adeoye and Agbola (1985)) found a critical range of Bray 1 P availability of 10-
16 ppm for tropical soils. Measured Bray 1 P in or below this range would be considered
low;

* Problem 3: Other forms of P that were not measured by the Bray-1 method could have
become soluble. The inorganic active P (represented by NaOH-Pi, Table 2-17) is an
important source of P that can become directly soluble during the growing season. The
organic carbon content of the soil that was nearly 2% in the topsoil could have contributed
P through mineralization especially under tropical conditions (Osiname et al., 2000).
Current thinking envisages the different forms of P in the soil as existing in equilibrium.
Field preparation disturbs the equilibrium and subsequent decomposition of soil organic
may release additional P. Also, plant uptake can displace this equilibrium in such a way
that replenishment of soluble P from other forms of P is continuous. Johnston (2000)
showed that in addition to providing P through mineralization, soil organic matter provides
sites with low bonding energy for P;

* Problem 4: The experimental field was left to natural bush fallow for about 2 years and P
fertilizer at 7.5 kg P20O ha-1 was applied to maize grown on the field in 2004 prior to the 2-
year fallow. Plant residues from the two-year natural bush fallow that preceded the
experiment was mixed with the soil during plowing, seven days before maize planting. The
2004 P fertilizer and the decomposition of the maize and fallow residues could contribute
significant amount to soil P build up that could have been made available when the soil
was brought out of the fallow for the experiment;

* Problem 5: The soil had an ideal pH (6.5) for P transformations and availability (Table 2-
15).

The delay in silking was due to the drought spell that occurred right after anthesis and

lasted until after silking (Figure 2-11). Studies have shown that water stress delays silking in

maize and results in an increase in the anthesis to silking interval (Balanos and Edmeades, 1993).









Similar findings on the positive effects of P on plant height were reported by Khan et al.,

2005. In the Kpeve experiment, the significant differences in heights between the treatments did

not result, however, in differences in biomass production. The LSD for height differences at 31,

38 and 45 days after planting were respectively 6, 11, and 16 cm. Compared to the respective

height ranges of 75-84, 124-138, and 188-215 cm (Table 2-9), those height differences (LSD)

corresponded to a part of the upper canopy that did not contribute much to the weight of the plant

and was mostly leaf blade.

The significance of the effect of days after planting on measurements repeated over time

has an agronomic meaning. It corresponds to an active period of growth for plant height and

green leaf area as shown on Figures 2-9 and 2-10.

Results and Discussion for the Wa Experiment

The soil in Wa contained very little available P and organic matter (Table 2-18). Maize

responded to nitrogen and phosphorus to the expected extent, confirming past and current

findings by other researchers (Singh and al., 1999; Colomb et al, 2000; Khan et al., 2005). The

results obtained in the Wa experiment are fully reported and discussed in Naab (2005). A

summary of the responses observed are presented here.

Nitrogen and phosphorus had similar effects on the phenological development of maize.

Tasseling was not affected by nutrient management. On the contrary, silking was delayed by

about 1-3 days in treatments that did not receive nitrogen or phosphorus (Table 2-19). Statistical

differences at silking were observed only between the no and medium or high nutrient

application. Differences were not found between the medium (60 kg [N] ha-1 and 60 kg [P20s]

ha-1) and high (120 kg [N] ha-1 and 90 kg [P205] ha-1) nutrient applications. The overall effect of

the nutrient deficiency on physiological maturity of the crop was small and not significant (Table

2-19). Grain filling duration was shortened in the no nutrient treatments in such a way that









physiological maturity did not differ among treatments. Similar results were obtained in Hawaii

by Singh et al. (1999).

Significant leaf area index (LAI) differences due to nitrogen and phosphorus applications

were observed throughout the season (Table 2-20). The effect of P on LAI disappeared at 90 dap

and thereafter. These LAI differences were observed only between the no nutrient treatments and

the 60 kg [N] or [P] ha-1 treatments, and LAI did not increase beyond the level of 60 kg [N] or

[P] ha-1. The maximum LAI advantage over PO, 50% in P60, was observed at 40 dap. Plenet et

al. (2000a) found an LAI reduction of the same magnitude (60%) between the 7- and 14-visible

leaves in a P response experiment in France.

Aboveground biomass responded consistently both to nitrogen and phosphorus fertilization

at all sampling dates (28, 46, 61, 81, and 125 days after planting). Nitrogen applied at 120 kg ha-1

did not result in any significant biomass accumulation over the 60 kg ha-1 N level at days 28, 46,

and 61 dap. At 81 and 125 dap however, the difference between N60 and N120 were amplified

and were significant (Table 2-21). The N60 treatment resulted in biomass differences of 75 to

4500 kg ha-l over NO, which represented 19 and 67% of the biomass obtained in N60. Response

to P was less drastic but also significant. Differences were not found between P60 and P90 at all

sampling dates. The biomass gain of P60 over PO ranged from 130 to 3300 kg ha-l corresponding

to 31 to 56% of the biomass measured in P60. The highest biomass and LAI gains (over NO or

PO) were obtained at the same sampling periods (40-46 dap for nitrogen and 61-68 dap for

phosphorus). This could be a confirmation of findings by Plenet et al. (2000b) according to

which poor biomass accumulation in P deficient plants was mainly due to reduced

photosynthetically active radiation absorbed by the canopy caused by reduced leaf area.









Fruit weight increased significantly between NO, N60, and N120. For phosphorus,

differences were found only between PO and P60, and PO and P90. No significant differences in

fruit weight were revealed between P60 and P90 (Table 2-22). Fruit growth was affected more

severely by nitrogen than phosphorus (Table 2-22). For example, grain yield gains were 77% in

N60 over NO and only 42% in P60 over PO. The effect of P applications on seed weight was

relatively small compared to the effect of nitrogen (Table 2-22). These ultimate effects on grain

yield and yield components were probably associated with the consequences of nitrogen and

phosphorus stress on photosynthesis (Singh et al., 1999).

Conclusions

The study at Kpeve did not result in the expected response to P fertilizer applications. No

significant differences in plant phenology, aboveground biomass, green leaf area and grain yield

were found between fertilized and unfertilized treatments. Significant differences in plant height

observed at 31, 38, and 45 days after planting were not reflected in biomass accumulation or

grain yield. Although the initial available P (Bray 1) was relatively low in all layers, other

important P sources such as chemical contributions of organic matter not accounted for by the

Bray 1 extraction could have been responsible for high indigenous P supply in the soil.

At Wa, soil P levels were sufficiently low to cause a P fertilizer response in the crop. Delay

in silking of about 1 day was observed in the treatment that did not receive any P input. The

delay in silking was 2 days in the no nitrogen treatments and 1 day in the no P treatments. Leaf

area index and aboveground biomass were reduced in no nitrogen and no phosphorus treatments

throughout the season. The highest reduction in leaf area index and biomass occurred at the same

time period, which strengthens the idea that poor biomass accumulation in P deficient conditions

is associated with reduced photosynthetically absorbed radiation by the plant, which is a









consequence of reduced leaf area. The reduction in grain yield could have been a result of

nutrient stress on photosynthesis.

The contrasting results obtained at the two sites in terms of response to P would be useful

in testing the ability of computer simulation models of soils and plants to capture and mimic the

effect of variability in P management on crops. In Chapter 4, this attempt is made using the soil-

plant phosphorus model in the Decision Support System for Agrotechnology Transfer.









Table 2-1. Growth and development genetic coefficients for the Obatanpa cultivar used at both
sites, Kpeve and Wa (Ghana)
Definition DSSAT ID Obatanpa
Degree-days (base 8C) from emergence to end of juvenile phase P1 300
Photoperiod sensitivity P2 0.00
Degree-days (base 8C) from silking to physiological maturity P5 830
Potential kernel number (/plant) G2 900
Potential kernel growth rate (mg/day) G3 6.50
Phyllochron interval (degree-days) PHINT 38.90

Table 2-2. Summary of fertilizer application methods used in the experiment in Kpeve, Ghana
Days after Ammonium Triple Potassium
planting Sulfate Superphosphate Nitrate
0 Broadcast without Side placement, bottom Broadcast without
incorporation of hole incorporation
13 Side placement, without Side placement, bottom Side placement, without
incorporation of hole incorporation
0 Side placement, without No No
30
incorporation Application application
44 Side placement, without No No
incorporation Application application

Table 2-3. Specifications of two different covariance structures used for modeling the effect of
time on repeated measures in PROC MIXED for the Kpeve dataset
Covariance structure Specifications
1. Measures at all times have the same variance;
Compound Symmetry 2. Pairs of measures from the same subject have the same correlation;

1. Measures at all times have the same variance;
Autoregressive 2. Correlations between pairs of measures from the same subject
decrease as the time lag between measures increases;
Source: Adapted from Littell et al. (1998).

Table 2-4. Analysis of Variance for simple linear regression between soil moisture
measurements using TDR and gravimetric methods at Kpeve, Ghana
Source of variation Degree of Sum of Mean F Value Pr > F
Freedom Squares Square
Model 1 805.45874 805.45874 178.19 <.0001
Error 65 293.82273 4.52035
Total 66









Table 2-5. Test of parameter estimates used to fit the linear regression model in the Kpeve
experiment


Variable

Intercept


Degree of
Freedom


Parameter
Estimate
4.90191
0.71900


Standard
Error
0.68030
0.05386


T Value


7.21
13.35


Pr> |t

<.0001
<.0001


Table 2-6. Analysis of variance of phenological events, tasseling, anthesis and
Kpeve experiment
Source of variation Degree of Sum of Mean F Value
Freedom Squares Square


Tasseling
Block
Phosphorus
Anthesis
Block
Phosphorus
Silking
Block
Phosphorus


0.6875
6.1875

1.2500
2.2500

31.1875
100.1875


0.2292
2.0625

0.4167
0.7500

10.3958
33.3958


0.11
1.03

0.88
1.59

0.70
2.26


silking in the

Pr > F


0.9496
0.4254

0.4861
0.2594

0.5737
0.1506


Table 2-7. Analysis of variance for grain yield (measured in kg ha-1) at Kpeve, Ghana
Source of variation Degree of Sum of Mean F Value Pr > F
Freedom Squares Square
Block 3 2447042 815681 1.12 0.39
Phosphorus 3 427698 142566 0.20 0.90
Error 9 6557203 728578

Table 2-8. Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time
points for crop aboveground biomass measured in kg ha-1 in the Kpeve experiment
DAP 17 31 52 108
Block 394 (0.45) 9342 (0.78) 462057 (0.36) 3781140 (0.11)
Phosphorus 55 (0.94) 11257 (0.73) 366238 (0.45) 1189617(0.50)
Error 410 25489 381818 1382973

Table 2-9. Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time
points for plant height measured in cm in the Kpeve experiment


DAP 17 24 31 38 45 5
Block 45.82 61.69 90.81 727.11 794.51 7
(0.20) (0.46) (0.54) (0.15) (0.40) (C
Phosphorus 54.87 106.03 388.72 978.82 3517.83 1
(0.14) (0.22) (0.03) (0.07) (0.01) (C
Error 29.21 70.63 124.47 397.13 805.88 1


2
75.17
).52)
882.73
).14)
020.16


68
1079.05
(0.45)
2401.73
(0.13)
1224.96









Table 2-10. Treatment means at each day for plant height measured in cm, with least significant
difference (LSD) in the Kpeve experiment (a = 0.05)
DAP Treatment LSD
OP 10P 30P 80P
17 25.91 ab 28.52 a 25.65 ab 25.14 b 3.10
24 42.43 b 44.85 ab 47.40 a 43.81 ab 4.82
31 76.06 b 75.858 b 84.05 a 81.22 ab 6.40
38 123.70 b 128.94 ab 138.15 a 125.75 b 11.43
45 194.80 bc 207.78 ab 214.59 a 187.93 c 16.28
52 235.90 a 250.15 a 249.94 a 233.67 a 18.32
68 240.29 a 258.60 a 256.45 a 240.24 a 20.08
Means with the same letter are not statistically different at a = 0.05.

Table 2-11. Summary of results from ANOVA (mean squares (p-values), n =4) at individual
time points for green leaf area measured in cm2 per plant at Kpeve
DAP 17 24 31 38 45 52 68
Block 33194 302147 1014051 2427809 5623424 2411939 5006509
(0.03) (0.03) (0.08) (0.07) (0.002) (0.12) (0.004)
Phosphorus 13567 151693 647931 1742106 1022362 3016789 1860147
(0.27) (0.19) (0.22) (0.17) (0.40) (0.07) (0.16)
Error 10192 94055 430784 1010465 1040870 1220628 1052365

Table 2-12. Summary of results from ANOVA (mean squares (p-values), n = 4) at individual
time points for soil moisture readings (in %) using TDR at Kpeve
DAP 30 37 45 48 53 55 58 64 69 74
19.62 11.01 12.25 3.02 1.15 1.67 4.90 14.96 34.92 17.08
Block
(0.07) (0.04) (0.02) (0.39) (0.76) (0.53) (0.11) (0.37) (0.18) (0.16)
9.01 4.72 6.72 2.37 2.02 0.49 3.31 39.14 54.51 23.48
Phosphorus
(0.34) (0.31) (0.14) (0.49) (0.56) (0.88) (0.25) (0.05) (0.06) (0.07)
Error 7.93 3.83 3.51 2.93 2.90 2.23 2.35 14.05 20.94 9.63

Table 2-13. Akaike Information Criterion (AIC) test for two covariance structures in PROC
MIXED for repeated measures analysis for the Kpeve experiment
Variable AIC value
Compound Symmetric Autoregressive + random
Height 4909.8 4846.6
G. Leaf Area 9015.5 8865.1









Table 2-14. F-values and significance probabilities using univariate ANOVA, and for test of
fixed effect using two covariance structures in PROC MIXED for the Kpeve
experiment
Source of df Univariate Compound AR(1)
variation ANOVA Symmetric plus random effect
Height
P 3 10.63 (< .0001) 1.79(0.1822) 3.21 (0.0228)
DAP 5 2096.7 (<.0001) 2096.7 (<.0001) 1770.88 (<.0001)
PxDAP 15 1.88 (0.023) 1.88 (0.0343) 1.77 (0.0354)
G. Leaf area
P 3 6.80 (0.0002) 1.35 (0.3) 1.35 (0.3)
DAP 5 5.03 (<.0001) 888.8 (< .0001) 823.53 (< .0001)
PxDAP 15 0.87(0.6) 0.87 (0.6) 1.15(0.3066)

Table 2-15. Physical and chemical characteristics of the soil at the experimental site in Kpeve,
Ghana
Parameter 0-10 cm 10-20 cm 20-30 cm
Texture
Clay (%) 18 20 18
Silt (%) 28 29 27
Sand (%) 54 51 55
Gravel (%) 40 40 35
Organic Matter
Organic carbon, Walkley-Black (%) 1.84 1.8 1.55
Total nitrogen (%) 0.26 0.25 0.22
Phosphorus
Total phosphorus (mg/kg) 294 299 229
Bray 1 (mg/kg) 11.69 10.4 7.43
Mehlich 1* (mg/kg) 90.44 46.12 50.19
Exchange Complex
Potassium K (cmol/kg) 0.11 0.08 0.06
Calcium Ca (cmol/kg) 7.39 7.31 7.65
Magnesium Mg (cmol/kg) 2.61 2.40 2.38
Acidity
pH-H20 6.45 6.56 6.48
*The Mehlich 1 analysis was done by the Wetland Biogeochemisty Laboratory at the University
of Florida. All other tests were done in the soil testing laboratory at the University of Ghana.









Table 2-16. Classes of phosphorus availability according to the Bray 1 extraction method
Soil test, P Bray 1 (ppm) Relative P level
0-5 Very Low
6-15 Low
16-24 Medium
25-30 High
> 30 Very High
Source: Shapiro et al. (2003).

Table 2-17. Characterization of the different forms of soil phosphorus at Kpeve, Ghana. Data are
reported in mg/kg
P fraction 0-10 cm 10-20 cm 20-30 cm
Inorganic
KC1 Pi 3.3 1.7 3.6
NaOH Pi 56.4 40.8 40.9
HC1 Pi 76.4 28.3 27.6
Organic
NaOH Po 65.8 65.0 64.7

Residual P 153.6 120.2 140.1

Table 2-18. Physical and chemical characteristics of the soil at the experimental site in Wa,
Ghana
Parameter 0-20 cm 20-40 cm 40-60 cm 60-90 cm
Texture
Clay (%) 7.50 14.50 40.90 52.90
Silt (%) 8.30 8.20 10.70 16.90
Sand (%) 84.20 77.30 48.40 30.20
Gravel (%) 4.30 6.40 49.20 80.70
Organic Matter
Organic carbon (%) 0.49 0.48 0.51 0.43
Total nitrogen (%) 0.06 0.06 0.04 0.04
Phosphorus
Bray 1 (mg/kg) 2.50 Not measured
Exchange Complex
Potassium K (cmol/kg) 0.06 0.08 0.11 0.13
Sodium Na (cmol/kg) 0.49 0.45 0.52 0.45
Calcium Ca (cmol/kg) 1.54 1.23 1.62 1.86
Magnesium Mg (cmol/kg) 0.32 0.51 0.74 0.86
Acidity
pH-H20 6.34 6.25 5.94 6.02









Table 2-19. Main effects of nitrogen and phosphorus on phenological development in maize at
Wa, Ghana
Treatment Days to Phenological Stage (days)
Nitrogen Tasseling Silking Grain filling duration Physiological maturity
NO 49 57 a 39 96 a
N60 48 55 b 40 95 ab
N120 48 54b 41 95 b
Phosphorus
PO 49 56 a 40 96 a
P60 48 55 b 40 95 a
P90 48 55 ab 40 95 b
Source: Adapted from Naab (2005). Means with the same letter are not statistically different at a
= 0.05 in each column.

Table 2-20. Main effects of nitrogen and phosphorus on leaf area indices of maize at Wa, Ghana
Treatment Days After Planting (days)
28 40 68 81 90
Nitrogen
NO 0.60 a 0.52 a 0.57 a 0.83 a 0.54 a
N60 0.88 b 1.03 b 1.32 b 1.46 b 1.03 b
N120 0.88 b 0.92 b 1.55 b 1.74 c 1.13 b
Phosphorus
PO 0.52 a 0.51 a 0.83 a 1.01 a 0.75 a
P60 0.90 b 1.02b 1.38b 1.55 b 1.01 a
P90 0.93 b 0.95 b 1.23 b 1.47 b 0.94 a
Source: Naab (2005). Means with the same letter are not statistically different at a = 0.05 in each
column.

Table 2-21. Main effects of nitrogen and phosphorus on cumulative aboveground biomass (in kg
ha-) of maize at Wa, Ghana
Treatment Days After Planting
28 46 61 81 125
Nitrogen
NO 334 a 1088 a 2289 a 2155 a 1740 a
N60 410 b 2435 b 6961 b 6473 b 5621 b
N120 403 b 2436 b 7811 b 8287 c 7502 c
Phosphorus
PO 292 a 1069 a 3223 a 3520 a 3222 a
P60 426 b 2403 b 6549 b 6816b 6025 b
P90 429 b 2486 b 7289 b 6580 b 5615 b
Source: Naab (2005). Means with the same letter are not statistically different at a = 0.05 in each
column.









Table 2-22. Main effects of nitrogen and phosphorus fruit yield components of maize at Wa,
Ghana
Treatment Fruit Yield
Cob weight (kg ha-1) Grain yield (kg ha-1) 1000-seed weight (g)
Nitrogen
NO 632 a 479 a 190 a
N60 2556 b 2063 b 226 b
N120 4042 c 3340 c 250 c
Phosphorus
PO 1646 a 1320 a 203 a
P60 2833 b 2292 b 235 b
P90 2750 b 2271 b 228 b
Source: Adapted from Naab (2005). Means with the same letter are not statistically different at a
= 0.05 in each column.













250 35
30 -
200 25
E
150 20 2
S15 C
100 E



0 0
Mar Apr May Jun Jul Aug
month

Monthly total rainfall in 2006
S Monthly total rainfall average (2003-2005)
---Maximum temperature in 2006
Minimum temperature in 2006


Figure 2-1. Monthly total rainfall in 2006 (mm) with error bars corresponding to one standard
deviation of rainfall, monthly total rainfall average from 2003 to 2005, and monthly
average daily temperature (C) in 2006 at Kpeve. 2006 data were collected during the
experiment and 2003-2005 data taken from Adiku (2006)










Initial
soil


.O0 M KCl extraction N Readily available
inorganic P (KC1 Pi)


0. 1 M NaOH extraction Alkaline extractable
organic P (NaOHPo)


0.5 M HC


Fe/Al bound P
(NaOH Pi)

I extraction Ca/Mg bound P
(HC1 Pi)


Ashing


Residual P


Figure 2-2. Sequential fractionation steps used for extracting the different forms of phosphorus
from soil samples taken before planting of the Kpeve experiment. Samples analyzed
by the Wetland Biogeochemistry Laboratory, University of Florida


Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
month


SRain -- Maximum temperature


Minimum temperature


Figure 2-3. Monthly total rainfall (mm) with error bars corresponding to one standard deviation
of rainfall and monthly average daily temperature (C) at Wa in 2004. Data from
Naab (2005).


7













S25


S20
0
o
E
S15


S10
E

> 5


volumetric = 4.90 +


0 5 10 15 20 25 30
TDR reading (%)

S0-20 cm o0-12 cm


Figure 2-4. Simple linear regression of volumetric soil moisture (%) determined by using Time
Domain Reflectrometry and gravimetric sampling at Kpeve


Silking





Anthesis





Tasseling


10 20 30 40 50 60 70
days

D OP 10P U 30P o 80P


Figure 2-5. Phenology of maize as affected by phosphorus application at Kpeve. Error bars
represent standard deviations of measurements taken from four replications











10000

8000

6000

4000


2000

0


phosphorus level (kg ha"1)

o grain yield c stover


Figure 2-6. Stover and grain yield of maize as affected by phosphorus fertilizer at Kpeve. Error
bars represent one standard deviation of measurements taken from four replications


1800
1600
1400
1200
1000
800
600
400
200
0


0.30

0.25 .

0.20
o'-
0.15 ,

0.10 ,

0.05 -

0.00


phosphorus level (kg ha"1)


grain number -- unit grain weight


Figure 2-7. Grain number per m2 and unit grain weight as affected by phosphorus fertilizer at
Kpeve.


S
S
S


T _

_L r__ _











12000


( 10000




0
if 8000


6000


o 4000

2000
o 2000


0 1 1 1 1 M M JI 1
17 31 52 anthesiss) 108 (final harvest)
days after planting

O OP O 10P 30P 80P


Figure 2-8. Aboveground biomass of maize as affected by phosphorus fertilizer application at
Kpeve. Error bars represent one standard deviation of measurements taken from four
replications.

300


S200


150

S100
100


anthesis


38
days after planting


OP 10P 30P 80P



Figure 2-9. Height of maize as affected by phosphorus fertilizer in the Kpeve experiment












7000


6000


'" 5000
E

S4000


S3000
c
2000
2000


anthesis


1000


38
days after planting


*OP 10P 30P 80P


Figure 2-10. Green Leaf Area of maize in the phosphorus fertilizer experiment at Kpeve


20


_


emergence


E+


U


anthesis silking


0 4 10 13 16 18 20 24 27 30 32 37 40 45 48 53 55 58 62 64 69 74
days after planting

*OP 10P 30P 80P


Figure 2-11. Variation in soil moisture measured using Time domain reflectometry in the Kpeve
phosphorus experiment. The possible effects of the four phosphorus treatments on the
soil moisture are analyzed and reported at individual time points in Table 2-12.









CHAPTER 3
THE SOIL-PLANT PHOSPHORUS MODEL IN DSSAT

Introduction

Biological, physical and chemical processes affecting phosphorus (P) transformations in

soils and plants create dynamic soil P pools that interact with plants in a complex way.

Phosphorus is present in the soil in two main forms: inorganic and organic. The inorganic forms

represent mineral P in the soil solution, P bound to calcium, P retained by iron and aluminum

oxides and by clay, and P occluded in iron and aluminum minerals. The organic forms

correspond to P in fresh organic residues, P in soil organisms' biomass, and P in slowly

decomposing soil organic matter. The forms of P in the soil are dependent on many soil

properties but the most important are soil pH, organic carbon content, the quantity and type of

clay, and the cation exchangeable capacity of the soil. For example, plant P nutrition is optimal

at pH 6.5 because the available form of P for uptake by plants predominates at that pH.

Plants take up phosphorus from the soil solution that is replenished by the other forms of

inorganic P and through mineralization of organic P. Several transformations between the

different forms of P occur in the soil, but the contribution to the soil solution is very low. The

amount of phosphorus available to plants from the soil solution at any one time would seldom

exceed about 0.01% of the total phosphorus in most soils (Brady and Weil, 2002).

Description of phosphorus processes taking place in a soil-plant system therefore would

deal with i) the different forms present in the soil; ii) the transformations that make insoluble

forms soluble for plant uptake; iii) plant uptake mechanisms; and iv) conditions and mechanisms

of P supply to plants.

The soil-plant phosphorus model in the Decision Support System for Agrotechnology

Transfer (DSSAT) attempts to consider these aspects to simulate phosphorus transformations in









soils and plants and their effect on crop production. The soil inorganic P module of the model

simulates phosphorus transformations between a labile, active and stable pool. The soil organic P

module simulates phosphorus transformations between a surface litter, a microbial pool, and a

stable pool. The model accounts for the mineralization of organic P to inorganic pools and the

immobilization ofP to organic pools. In phosphorus deficient soils where organic matter play an

important role in the supply of nutrients, simulation of the release of phosphorus from organic

matter accounts for an important source of P. Available phosphorus for uptake by plants is

described as being provided by the labile pool within a short distance of plants' roots (2 mm).

Phosphorus taken up by the plant is partitioned to seeds, shells and vegetative tissues.

During the reproductive phase, phosphorus accumulated in the vegetative tissues can be

remobilized and translocated to seeds. Plant growth is limited by phosphorus between two

thresholds that are species-specific optimum and minimum concentrations ofP defined at

different stages of plant growth. Phosphorus stress factors are computed to reduce

photosynthesis, dry matter accumulation and partitioning.

The present chapter summarizes procedures used in the simulation of the phosphorus

balance in the DSSAT cropping system models. The objectives of the chapter are to: i) present

the phosphorus modeling framework in the DSSAT cropping system model; ii) present a

description of the soil and plant phosphorus model in DSSAT; iii) present a sensitivity analysis

of the model to selected key phosphorus-related parameters. The main question of interest in the

sensitivity analysis was: how does the variability of six major phosphorus-related factors affect

biomass and grain yield as simulated by the model?

Soil and Plant Phosphorus Modeling in DSSAT

The need for implementing a phosphorus model in DSSAT was already recognized as a

limitation of the software at the release of its first version (Jones et al., 1998). It was clear that









the integration of a phosphorus component into DSSAT would considerably increase and extend

its applicability not only to P deficient environments, but also to low input cropping systems

receiving significant amounts of phosphorus from the decomposition of organic matter. The

development of a phosphorus model in DSSAT presented at least two challenges: i) scientists

need to improve their understanding of phosphorus behavior in soils and plants because of the

complexity of P chemistry in soils and its interaction with other major nutrients that limit plant

growth; ii) the design of the initial version of DSSAT was not suitable for maintenance of the

software as new models were included and modifications were made. The software was a

collection of independent models operating in the same framework to integrate information about

soil, climate, crop and management. The models in the decision support system were operating

in their own original programming settings.

For many years the models in DSSAT have been used to simulate potential, water and

nitrogen limited production only, even in areas where phosphorus deficiency is widespread.

Advances in modular programming techniques have enabled DSSAT developers to completely

redesign the software (Jones et al., 2003); the crop models can now operate using the same soil

module, the same climate module, and the same management module. Other modules could

therefore be more easily included and connected to existing crop models with minimal

modification to the system. The new programming technique also facilitates documentation and

updating of the individual modules that were developed by specialists from different disciplines

working together as a team.

The first version of a soil-plant phosphorus model linked to the DSSAT cropping system

model (CSM) was developed and evaluated by Daroub et al. (2003) for calcareous and highly

weathered soils. The model could have been plugged into any crop models within the DSSAT









CSM to simulate phosphorus-limited production, but was tested only for maize and soybean.

Although the model predicted grain yield and plant uptake with a reasonable degree of accuracy

(Daroub et al., 2003), two important modifications were necessary to make it satisfy the concern

of extending the applicability of the CSM to low inputs cropping systems and to users having

access to limited soil information: i) linking the model to the DSSAT-CENTURY model

(Gijsman et al., 2002a) for simulation of organic P transformations; and ii) integrating a soil

expert system that can allow the estimation of initial amounts of inorganic and organic soil

phosphorus pools as influenced by major soil categories and using different methods of P

extraction, pH and organic carbon.

The initial P model developed by Daroub et al. was thus updated with these two major

modifications and is described here.

Description of the Soil Phosphorus Model

The soil phosphorus model is comprised of the soil inorganic module and the soil organic

module. The two modules are linked in a way that soil phosphorus mineralized from organic

matter is transferred to the inorganic module and phosphorus immobilized in the inorganic

module is moved to the organic module. The initial sizes of the inorganic and organic pools can

be derived directly from P fractionation data or indirectly from other P extraction methods. The

initialization procedures, developed based on studies by Jones (1984), Singh (1985) and Sharpley

(1984, 1989), are described in Appendix C.

Soil Inorganic Module

The soil inorganic module describes transformations that occur between the inorganic P

pools to make P available for plant uptake.









Inorganic phosphorus pools

The soil inorganic module distinguishes three pools: labile (PiLabile), active (PiActive)

and stable (PiStable). The three pools exist in two soil zones: the zone that is in direct contact

with roots (within 2 mm) and the zone that is not in direct contact with roots (Figure 3-1). The

labile inorganic P pool includes the P in the soil solution. Because roots do not develop at

planting, the initial soil volume in direct contact with roots is assumed to be zero at the beginning

of the simulation and the total amount of inorganic phosphorus available at that time is assigned

to the no-root zone. If transplants are used, an initial soil root volume is estimated to initialize the

simulation. As the roots develop, a proportional mass of P is added to the root zone pools and

subtracted from the no-root pools in proportion to the soil volume adjacent to the new root

growth.

Phosphorus transformations between the inorganic pools

Per day phosphorus transformations between the three pools occur according to the

following first-order relationships:

P flow from the labile pool to the active pool = K x PiLabile (3-1)
P flow from the active pool to the labile pool = KAL x PiActive (3-2)
P flow from the active pool to the stable pool = K A x PiActive (3-3)
P flow from the stable pool to the active pool = Ks x PiStable (3-4)
Where the P flows between the different pools are in units of mg [P] kg-1 [Soil] day-'
The coefficients KLA, KAL, KAS, and KSA are the respective transformation rate constants,
in unit of day-1. The values of KLA, KAL and KAS depend on the phosphorus availability
index (Table 3-1) (Jones et al., 2005a; Jones et al., 1984a; Sharpley et al., 1984, 1989).
KLA, KAL and KAS are calculated as follow:
(1 PAvailhdex) ,
K = 0.03x F(1 (3-5)
L PAvailIndex
K,
K AL = K x PAvailndex (3-6)
3
K A = e(- 77 PAvaillndex)7 05 (3-7)
KA = 0.0001
Where PAvaillndex = P availability index defined in Table 3-1.









KAS = rate constant for transformation from active P to stable P.
KSA = rate constant for transformation from stable P to active P.

Mineralization and immobilization of phosphorus from the organic matter are handled in

the soil organic module. A net mineralized P amount is calculated and is added proportionally to

the root and no-root zones when its value is positive and subtracted from the labile P pool if its

value is negative.

Phosphorus uptake calculated in the plant model is subtracted directly from the PiLabile

pool in the root zone.

P fertilizer applied is directly added to the labile and active pools. The amount of applied

P that enters those pools depends on the soil category and the application method. Fertilizer

applied in bands or hills is used more efficiently by the plant. When these application methods

are used, all of the P is applied directly into the root soil volume. When broadcast or other

application methods are used, the fertilizer is proportioned to the root and no-root zones to the

depth of incorporation.

A P fertilizer availability function is computed using soil composition (Table 3-2) based

on studies by Jones (1984) and Sharpley et al. (1984, 1989). The P fertilizer availability index is

expressed as a fraction of fertilizer which enters the labile pool. The remaining P fertilizer is

added to the active pool.

Phosphorus availability for uptake by plants

The available phosphorus for plant uptake is the soluble phosphorus that is in the root

zone. A fraction of the root labile Pi is assumed to be soluble, and that fraction defines the

portion of the root labile Pi in the soil solution that is "sensed" and is available for extraction by

plant roots on a daily basis.

SoilPiAvail = FracPSol x SoilPiLabileRoots (3-8)
Where SoilPiAvail is the plant available inorganic phosphorus;









FracPSol is the fraction of root labile Pi that is soluble (i.e. enters the soil solution).
SoilPiLabileRoots is the inorganic labile P that is in the root zone.

Soil Organic Module

The soil organic module describes transformations of organic materials that eventually

contribute P to or extract P from the inorganic P pools through mineralization or immobilization.

Organic phosphorus pools

The soil organic P module has of four litter pools and two soil organic matter (SOM) pools

(Gijsman et al., 2002a):

* Organic residues added to the surface of the soil become either surface litter or soil litter.
The residue materials themselves are divided into easily decomposable or metabolic
materials (i.e. sugars and proteins) and recalcitrant or structural materials (i.e. lignin and
other fibers) (Figure 3-1). As a consequence, four litter pools can be defined: a surface
structural litter pool, a surface metabolic litter pool, a soil structural litter pool, and a soil
metabolic litter pool.

* Microbial activity creates two active pools, one on the surface and another in the soil
(SOM1 pools).

* A stable pool exists in the soil only (SOM23) and is the combination of the slow SOM
(SOM2) and passive SOM (SOM3) pools for carbon (Gijsman et al., 2002a).

The soil SOM1 and SOM23 are the main pools that control inorganic phosphorus

dynamics in the soil. The surface litter pools generate flows of carbon and nutrients into the

surface SOM1 and the soil litter pools through tillage (Figure 3-1), and the surface and soil litter

pools eventually become part of the soil SOM1 pool.

Phosphorus movements between the different organic P pools follow carbon flows

according to a carbon to phosphorus ratio at which phosphorus is allowed to enter a specific pool

(Table 3-3). The different flows and their directions are summarized in the soil organic

phosphorus processes section of Figure 3-1.









Phosphorus flows between the organic pools


Phosphorus flow from any organic pool A to any organic pool B (PflowAB) is

proportional to the carbon flow between the same pools. The terms "pool A" and "pool B" as

used in this section refer to any combination of the four litter pools and two soil organic

phosphorus pools between which P flow can occur (Figure 3-1). A typical flow can be described

by the following equation:

CFlowAB
P flow (from pool A to pool B in a specific layer) in kg [P] ha-1 = P, x Co (3-9)
CA
Where PA is the amount of phosphorus in pool A of that layer (kg [P] ha-1). The amount of
phosphorus in each of the five pools (3 inorganic and 2 organic) is defined at initialization.
CA is the amount of carbon in pool A of that layer (kg [C] ha-1). The partitioning of the
measured total organic carbon defines the amount of carbon that belongs to the three
different soil organic matter pools. The fractions of carbon in SOM1 (active), SOM2
(slow) and SOM3 (passive) are defined by the model user according the cultivation history
of the soil. Typical partitioning of the total SOM respectively into SOM1, SOM2 and
SOM3 for a previously cultivated, irrigated and highly fertilizer loamy soil is 2%, 39% and
59% (Parton et al., 1988, 1994), but this can be varied but the user.
CFlowAB is the carbon flow from pool A to pool B (kg [C] ha-1) for that layer.

The flow of carbon out of a pool is calculated as follow (Gijsman et al., 2002a):

C flow (out of pool A) in kg [C] ha-1 d-1 = CAx DECA x CULA x DEFAC x OTHER (3-10)
Where CA is the carbon content of pool A (kg [C] ha-1);
DECA is the maximum decomposition rate of pool A under optimal conditions and without
increased decomposition due to soil disturbance (day-l). The maximum decomposition
rates of various pools are listed in Table 3-3.
CULA is the effect of cultivation on the decomposition rate of pool A. CULTA functions as
a multiplier on the maximum decomposition rate (0 to 1);
DEFAC is the decomposition factor that represents the effect of temperature and low soil
water conditions on the decomposition rate parameter. DEFAC functions as a multiplier on
the maximum decomposition rate (0 to 1);
OTHER represents the effect of other factors on the maximum decomposition rate. These
factors include the lignin content of the structural material and the clay content of the soil
which are used to reduce the decomposition rate of the structural litter and the soil SOM1.
The lignin concentration of the structural litter is used to partition its total carbon flow to

SOM1 and SOM23. The non-lignin portion enters SOM1 and the lignin portion flows into

SOM23 with carbon and phosphorus.









Phosphorus mineralization and immobilization

The material flowing out of pool A is allowed to enter pool B only under a certain C:P

ratio that is computed assuming a potential immobilization rate of phosphorus. This constraint is

depicted by the following equation:

CFIowAB
CPB = CowAB (3-11)
(PFlowAB + IMMOB)
Where CPB is the C:P ratio of the material allowed to enter the receiving pool B;
IMMOB is the immobilization ofP (kg [P] ha-1) from the inorganic labile pool;
CFlowAB and PFlowAB are respectively the C flow and the P flow from pool A to pool B
in (kg [C or P] ha-1).

When the material flowing from pool A to pool B has a C:P ratio that is larger than the C:P

ratio of the material that is allowed to enter pool B (CPB), an immobilization of phosphorus from

the inorganic labile pool occurs to compensate for the deficit of P in the material flowing. The

amount of phosphorus immobilized is derived from equation (3-11):

CFlowAB
IMMOB = AB PFlowAB (3-12)
CPB

Mineralization (MINERAB) occurs only when the actual flow (PFlowAB) exceeds the

f low(CFlowAB
expected flow .CoAB
CPB

CFIowAB
MINERAB = PFlowAB CFIAB (3-13)


Each carbon flow is accompanied by respiration losses in the form of carbon dioxide

(C02), which is a flow of carbon that does not enter the receiving pool. Phosphorus

mineralization is also concomitant with this loss of carbon to CO2. The amount of phosphorus

that is mineralized during the respiration process (PFlowCO2) is calculated as:


P flow C02 (from A to B) in kg [P] ha-1 = PA x C2FwA (3-14)
CA









Where CO2FlowA is the CO2 flow out of pool A (kg [C] ha-1);
PA and CA are respectively the amount of phosphorus and carbon in pool A
(kg[C or P] ha-1).

The total P mineralization (MINERTOT) resulting from the carbon flow and the respiration

losses to CO2 is therefore:

MINERTOT = MINERAL + PFlowCO2 (3-15)

The net phosphorus mineralized

The flow of carbon or phosphorus from pool A to pool B generates an immobilization of

phosphorus in the material flowing and a mineralization of phosphorus that does not enter pool

B. Immobilization holds phosphorus and depletes the soil inorganic P but mineralization releases

phosphorus that can be made available for plant uptake. Total P mineralized (SUMPMIN) and

total P immobilized (SUMPIMM) are computed by summing up the mineralization and

immobilization P from all the different flows. A net P mineralized corresponding to the

difference (SUMPMIN SUMPIMM) is calculated and added to the inorganic labile pool for

plant uptake (Figure 3-1). However, if the total P immobilization and other P takeoff are greater

than the amount of P available in the soil, the SOM and litter decomposition are reduced by a

reduction factor, so that the amount of P needed for immobilization equals the amount of P

available in the soil.

Description of the Plant Phosphorus Model

The plant phosphorus component models P taken up from the soil and stored in four

different plant parts: roots, shoots (leaves plus stems), shells and seeds. Phosphorus supplied by

the soil to meet the plant's demand and phosphorus exported by the crop at harvest are external

to the plant P component.









Phosphorus in the Plant

The P accumulated in the whole plant is the sum of the P taken up into the different plant

parts (Jones et al., 2005b; Daroub et al., 2003).

PPlant PRoot + PShoot + PShell + PSeed (3-16)

The P in the different parts of the plant is computed as P concentrations (g [P] g [shoot]-'

for instance). The mass of P (in kg P ha-1) is further calculated after its combination with growth

data provided by the appropriate crop growth model.

Minimum and optimum concentrations of P for maize defined at three growth stages are

derived from literature and stored in a species file (Table 3-4). The optimum shoot P

concentration was calculated using growth stage dependent equations developed by Jones

(1983). The minimum shoot P concentration was taken as 60% of the optimum values (Daroub et

al., 2003). Initial P concentration values for the different plant parts are set to the optimum when

the plant emerges from the soil. Because the model uses a daily time step to compute phosphorus

in the plant and the optimum and minimum concentrations of P are available at discrete growth

stages only, linear interpolation between the growth stages is used to determine the optimum and

minimum P concentrations every day (Figure 3-2). These interpolations depend on the actual

plant growth as influenced by cultivar characteristics, soil and weather conditions. The N:P ratio

is handled in a similar way to constrain the uptake of P when nitrogen is limiting (Figure 3-3).

Actual phosphorus accumulated in the plant on any day is increased by uptake. Amount

of P mobilized (from roots, shells and shoots only) and lost due to senescence, pest and disease is

furthermore subtracted from the P in the plant part considered.









Uptake

The amount of phosphorus available for uptake by the whole plant is the minimum of

demand and soil supply.

PTotal Uptake = MIN(PTota Demand ,PSoil _Supply) (3-17)

The maximum and minimum N:P ratio computed daily by linear interpolation from

Figure 3-3 is used to limit P uptake if on any day the actual N:P ratio is below the minimum.

PTotal Uptake Nlimited = PTotal Uptake x PUptake Re auction Factor (3-18)

The P uptake reduction factor utilized when the actual N:P ratio falls below the minimum

value is calculated as follows:


PUptakeReductionFactor MIN N: Pctual 1.0 (3-19)
N: *Mznzmum
Where N:PActual is the actual N:P ratio and
N:PMinimum is the minimum N:P ratio.

Soil Supply

Soil P supply is the amount of root zone labile P computed in the soil inorganic module.

Only a fraction of the soil supply is considered available (soluble) to meet the plant demand on

any day. That fraction is a parameter changeable by the user. The value currently used is 0.2

meaning that 20% of the labile inorganic P in the soil zone adjacent to roots on any one day can

be taken up by the plant during that day. This value was obtained based on a best-fit compromise

between simulated and measured biomass from the Wa experiment described in Chapter 2.

Plant Demand and P Mobilization Pools

Plant concentration of phosphorus at any one time during the growing cycle dropping

below the optimum concentration (specified in the species file) is considered a deficit and

induces stress (Figure 3-4). Demand is calculated for each plant part based on the amount of P









required to bring the P concentration in each of the plant parts up to the optimum, plus P required

for new growth.

PDemand = POptmum PActual +PNew growth (3-20)
Where PDemand is the amount of P in kg ha-' required to bring the actual concentration of P
to the optimum;
Poptimum is the computed optimum P in kg ha-1 using linear interpolation;
PActual is the amount of P in kg ha-1 present in the plant part concerned;
PNew growth is the amount of P in kg ha-1 needed for new growth.

High biomass accumulation can cause PActual to be greater than POptimum resulting in a

negative PDemand for any plant part at any moment during the growth. The excess P accumulated

is therefore stored in a mobilization pool for each plant part. There is no P mobilization pool for

seeds.

Demand for each plant part is first met by P stored in the mobilization pools. P moves from

root and shoot mobilization pools to satisfy P demand in shells and seeds. The P leftover stays in

the respective mobilization pools. Total P demand is recalculated and is possibly met by the soil

supply. If the soil supply is insufficient to meet this demand, uptake and subsequently P

concentration in plant's parts are reduced and will increase P stress.

Partitioning and Translocation

During the reproductive phase, total phosphorus taken up by the plant (PTotal Uptake) from

the soil is first used to meet seed demand. If the seed demand is greater than the amount of

phosphorus available to meet this demand, phosphorus translocation from roots, shoots and

shells to the seeds occurs. The available P for translocation is calculated as follows:

PTranslocation (PRoots Actual Roots n (Poo ctua PShoots Min + (Shells Actual PShells Min

(3-21)

The maximum amount of P that can be mined from the shells and the vegetative tissue in

one day (PTranslocation Max) is a fraction of the available P for translocation.









PTranslocation Max = MAX(0.0, FracPMobil x Pnslocaon ) (3-22)
Where FracPMobil is the fraction of the translocated P that can be used by the seeds in one
day. FracPMobil is defined as a parameter in the species file.

The P translocated is used to meet the seed demand if it is still positive after total P uptake

from the soil has been used up. The remaining P after seed demand is fully met is used by shells.

Vegetative tissues (roots and shoots) are supplied with P after reproductive organs' (seeds and

shells) demand is met.

Stress Factors

Two stress factors are computed based on a P stress ratio when the actual shoot phosphorus

concentration falls below the optimum. The stress ratio is computed as follows:

PStress- Ratio MIN 1.0, PShoots Actual Concentration PShoots Mmn Concentration (3-23)
PstressRatio .MINr .0, ..
SPShoots Optimum Concentration PShoots Mn Concentration
Pstress Ratio = 0 means maximum stress and
Pstress Ratio = 1 means no stress.

P stress effects on photosynthesis and P partitioning are modeled differently. Thresholds

values are defined in the species file and are used to compute stress factors for photosynthesis

and P partitioning:


Pstress Factor Photosynthesis = MIN tes R 1.0 (3-24)
SRATPHOTO

PStressFactor Partitioning = MIN StPre1ss Rato 1.(3-25)

Where SRATPHOTO is the minimum value of the ratio of P in vegetative tissue to the
optimum P, below which reduced photosynthesis will occur.
SRATPART is the minimum value of the ratio of P in vegetative tissue to the optimum P,
below which vegetative partitioning will be affected.

The two P stress factors, which are given different weights, are used to reduce

photosynthesis and P partitioning on any day during the growth of the plant when the actual

shoot P concentration falls below the computed optimum shoot P concentration. Values of









SRATPHOTO and SRATPART as read from the species file are respectively 0.80 and 1.00

meaning that P deficits in shoot tissue will first affect root-shoot partitioning before it affects

photosynthesis (Figure 3-4).

Model Inputs and Outputs

The soil-plant phosphorus model does not run as a standalone application but is

intrinsically linked to DSSAT crop growth models. As a consequence the soil-plant P model also

uses the basic inputs required to run the crop growth models. Additional inputs and parameters

required to run the model are summarized in Tables 3-5 and 3-6. The model essentially modifies

the crop growth model's outputs to allow them to be phosphorus-limited in P-limiting

environments.

Sensitivity Analysis

Sensitivity analysis is an important assessment tool that assists with evaluating the

uncertainty and variability associated with model structure and inputs during model

development, calibration and validation.

Introduction

A simulation model of crop growth and development is the result of several cycles of fine

tuning of model theory and structure, parameter estimations and adjustment of number of

required input variables. The ultimate objective of the continuing model refinement is to obtain a

model that is as close to the ideal model as possible, predicting measurable outputs with

maximum accuracy. Scientists admit, however, that even the most carefully-built simulation

model is not expected to give simulations that exactly equal observations. Uncertainty associated

with model equations, measured model input variables and estimated model parameters will

always remain an integrated part of the model and will contribute a great deal to simulation

biases. The uncertainty is not an accident; it may be the substance of the scientific method itself









(Saltelli, 2002). More specifically, the role of sensitivity analyses is to help apportion the

uncertainty in the model output to the different sources of uncertainty and variability in inputs

(Saltelli, 2005).

Uncertainty and variability justifying the usefulness of sensitivity analysis can stem from

various sources:

Choice of an appropriate complexity. Modeling agricultural and biological systems

requires an appropriate choice of components that are meaningful for the system and will

eventually form the structure of the model. The same real world system can be approached

differently by various scientific communities although they may set the same objectives and have

similar technical backgrounds. The main classical theories may be the same but the way

scientists "see" the system can be influential in the way they "model" it. For example, modelers

of phosphorus-limited production have used different P pools (Probert, 2004; Daroub et al.,

2003). Because the components modeled and the structure used set the mathematical

representation, the choice of the model complexity can be subjective and introduce some

uncertainty with respect to the processes involved.

Parameter estimation. Uncertainty can also come from parameters estimated based on

weak evidence or not-so-well established experimental results, especially during model

development.

Measured model inputs. Another important source of variability is model input variables

that may have been measured from field experiments or obtained from various data sources.

Field measurements (even replicated) can include important precision errors, sometimes due to

variability in natural processes. Errors of this kind can propagate to model outputs more than

proportionally.









Expert systems. Parameters or input variable that cannot be easily measured are

sometimes indirectly estimated using expert opinions or empirical relationships such as pedo-

transfer functions (Gijsman et al., 2002b). When confidence limits are not provided for these

methods, reliability on the indirect estimates may be questionable.

Modeling language and untrained model users. When the modeling language is not

English-like, model users, sometimes performing calibration with no capability to read the

programming language, may have to use it as a black box. It is not evident that members of

interdisciplinary teams in which a model was developed are aware of the way inputs are mapped

to outputs in the total model (Oberkampf et al., 2004). With complex models requiring hundreds

of parameters and input variables to operate, it may become unclear how the model behaves

independently of any evaluation with real world datasets.

Sensitivity analysis has become an ingredient of modeling (Saltelli et al., 2000) and been

used in many studies at various stages of model development (Makowski et al., 2005; Ratto et

al., 2001; Rahn et al., 2001). General objectives of a sensitivity analysis are (Monod et al., 2006):

i) to verify that the model behaves as expected when inputs are changed; ii) to quantify the

magnitude of the influence of parameters on outputs; iii) to identify the model parameters that

require maximum accuracy in their estimation; iv) to identify input variables to the model that

need to be measured accurately for the simulations to be correct; and v) to isolate possible

parameter interactions effects on outputs.

This section describes and presents results of a sensitivity analysis performed on some

major P-related parameters of the soil-plant phosphorus model in DSSAT. The overall objective

of this sensitivity analysis was to assess the effect on maize biomass, grain yield and P uptake of









major inputs and parameters that are directly related to plant response to P and P stress. Specific

questions for this sensitivity analysis were

* Question 1: Does the model respond to P fertilizer as expected?

* Question 2: How does biomass and grain yield react to changes in initial PiLabile where

plant uptake occurs, initial organic P that add P to this uptake pool, and the fraction of this

uptake pool that is soluble?

* Question 3: What is the magnitude of the effect of variability in initial PiLabile and initial

organic P, two P pools that are estimated in the model with uncertainty, on biomass and grain

yield in the model?

* Question 4: How does the model react to variability (as found in the literature), in

parameters that are used to compute P stress, optimum and minimum shoot P concentration?

* Question 5: Would variability in optimum and minimum seed P concentration have any

effect on biomass accumulation by the plant or crop grain yield?

Materials and Methods

The sensitivity analysis requires the specification of a computer experiment, a sensitivity

analysis method and inputs factors. These three components of the sensitivity are described next.

Computer experiment

A sensitivity analysis can be regarded as a highly controlled experiment carried out using

specific treatments applied to a specific crop growing in a specific environment under specific

management conditions. The mention of "highly controlled" carries an important meaning for a

sensitivity analysis because only the effects of the treatments are investigated and all other

growing factors are fixed at constant values. While this kind of experiment can bear a high

resemblance to a field station trial, some peculiar characteristics must be pointed out:









* The aim of a sensitivity analysis is to study the behavior of a model whereas the aim of a
field station trial is to examine the behavior of nature. The sensitivity analysis of a crop
model can be thought of as an experiment where nature is replaced by the simulated crop
model (Monod et al., 2006);

* The experiment described through a sensitivity analysis can be hypothetical. Some
conditions relevant to the experiment cannot be met in a station trial due to limits to
control nature or ethical considerations. In the sensitivity analysis experiment, there is no
limit to the achievement of conditions required to isolate treatments effects.

* A station trial examines the behavior of nature through independent factors usually
external to the field but that are known or suspected to influence dependent variables. In a
sensitivity analysis, the input factors are necessarily sampled or chosen from the input
variables to the model or the model parameters.

* In a station trial, replications are key components to statistical analysis of the results of the
experiment due to field variability that result in measurement error. When an analysis of
variance is performed, the measurement error is used as a proxy to assess the amount of
variation accounted for by each factor under study. In sensitivity analysis experiments,
there is no need for repeating the same treatments as long as the model is deterministic. As
a result, measurement error (actually simulation error) cannot be computed and formal
hypothesis testing has no scientific meaning and cannot even be performed (Monod et al.,
2006).

To differentiate the sensitivity analysis experiment from real world station trials, we will

call it a "computer experiment". The treatments will be called "scenarios". Each scenario is the

combination of levels of factors that will be named "input factors" (Monod et al., 2006).

Settings for the computer experiment

The computer experiment was performed using agro-ecological and modified

management information from the phosphorus experiment conducted in Kpeve, Ghana (6 40.80'

N, 0 19.20' E, altitude 67 m above sea level), in 2006 and described in Chapter 2.

The experiment was conducted during the main rainy season (April to August) on a

loamy soil (Tables 3-7 and 2-15). Daily rainfall, solar radiation, and maximum and minimum

temperatures were monitored using an automatic weather station located within the research

station. A medium-duration cultivar, Obatampa (Table 2-1) was sown on May 27, 2006. The

plant population at emergence was 6.25 plants m-2. To remove any nitrogen stress, a total of 500









kg N ha-1 was applied in the form of urea and split as follow: 100 at planting; 150 at 14 days

after planting; 150 at 27 days after planting; and 100 at 41 days after planting. Water stress was

controlled by automatic irrigation when the available water in the first 50 cm dropped below

70% of the drained upper limit.

Input factors, scenarios and model outputs

Three input variables to the soil modules and three plant module parameters were

selected and constituted two categories of input factors for the sensitivity analysis (Table 3-8).

Each input factor had 3 levels. The factor levels were specified in a way that a low, a medium

and a high setting of the factor were investigated through the sensitivity analysis. Wherever

applicable the medium level reflected the default (or the nominal) values initially specified in the

model and defined the control scenarios (Tables 3-8 and 3-9).

The input factor values were selected based on the knowledge of their uncertainty around

a nominal value or the medium settings. To answer specific questions addressed in this

sensitivity analysis, six input factors were selected. The input factors were selected from model

parameters (Table 3-5) and inputs (Table 3-6). The six inputs factors for the sensitivity analysis

are discussed next.

Initial Inorganic labile P and organic P. Although accurate knowledge of labile

inorganic phosphorus and total organic phosphorus present in the soil at planting is crucial for

good quality simulations of growth and development, it was anticipated that most model users

will not have access to the data measured as needed. If an alternate method is not provided for

indirect estimation, potential model users could possibly resort to indicative values found in

literature or eventually conclude that the model is not of any practical use because required input

data are not readily available. Access to heavy data requirement by the model user is a known

constraint of model use in research and development in large parts of the world (Struif-Bontkes









and Wopereis, 2003; Matthews and Stephens, 2002; Walker, 2000). Since organic carbon, pH

and available phosphorus are routinely measured in most traditional agronomic experiments,

developing relationships that can make use of those data and provide reasonable estimates of

inorganic labile P and organic P was thought to be helpful. Some soil properties are related to

each other and may be estimated from selected measurements; however, indirect estimation of

inorganic and organic phosphorus can pose serious uncertainty problems due to the complex

chemistry of P in soils. In this specific situation, the uncertainty comes from two main sources:

1) measurement errors of the soil parameters; 2) regression errors associated with developing the

equations for estimating indirectly the soil phosphorus. This was a strong motivation for studying

the effect on variable initial inorganic labile P and total organic P on some key model outputs.

Studies by Sharpley et al. (1984 and 1989) who developed linear relationships to predict

inorganic labile P and organic P for different categories of soils provided a basis for the ranges of

values used in the sensitivity analysis. Most soils considered in Sharpley (1984) have measured

PiLabile in the range 0-15 ppm and total organic P in the range 50-200 ppm. The PiLabile in the

first 20 cm was 6.5 ppm at Wa and 16 ppm at Kpeve. The estimated organic P (from Table C-5,

Appendix C, Slightly weathered soils) in the top 20 cm was 37 ppm at Wa and 133 ppm at

Kpeve. These two soils clearly have different P-supply capabilities in the ranges explored by

Sharpley, and they present indicative starting points for the specific levels of the input factors

initial PiLabile and organic P. Since some soils in Sharpley's survey would be more PiLabile-

depleted than the Wa soil, the "low" level of the input factor PiLabile was set at 2.0 ppm. The

medium level of the input factor Initial PiLabile was set at 8 ppm, around the middle of the range

0-15 ppm found in Sharpley (1984). The high level of the same input factor was set at the upper

boundary of that range.









The organic P level in the Wa soil (37 ppm) seems representative of a very low level so the

low level of the input factor Initial Organic P was set at 40 ppm. The medium level of the input

factor Initial Organic P was set at 100 ppm, around the middle of the range 50-200 ppm found in

Sharpley (1984). The upper limit of the range 50-200 ppm (that is 200 ppm) was considered as

the high level of the input factor Initial Organic P.

The low, medium and high levels of the two input factors Initial PiLabile and Initial

organic P were therefore respectively 2, 8 and 15 ppm for Initial PiLabile and 40, 100 and 200

ppm for Initial organic P. The original soil ratio between organic C and P (Table 3-7) was

maintained for the soil used in the analysis meaning that the soil carbon input value was changed

along with the organic P.

P fertilizer. Fertilizer application is one of the most important tactical management

strategies used to balance nutrient requirements by crops. Studying and understanding a crop

growth model's behavior to varying fertilizer levels is key to checking on the model's ability to

simulate variable fertilizer input feasibilities. Small applications of phosphorus (20-40 kg [P] ha-

1) to degraded soils have been recommended to restore progressively the soil P status (Shapiro et

al., 2003) and reported to increase inorganic soil labile P (Nziguheba et al., 2002) and improve

grain yield in maize (Fofana et al., 2005). The levels of phosphorus fertilizer were set to 0, 30

and 60 kg P ha-1. The P fertilizer was managed in the same way as in the Kpeve experiment to

ensure efficient use by the plant, split-applied in bands, 50% at planting and 50% 14 days after

planting.

Fraction of root labile P that is soluble. The maximum fraction of available phosphorus

which can be taken up in a day has a value of 0.20 in the phosphorus model and is not allowed to

vary regardless of growth stage, cultivar variation or differences in phosphorus uptake efficiency.









However, the uptake rate of phosphorus by cereal plants varies with plant age and intrinsic

cultivar differences (Johnston, 2000). If there is a large supply of phosphorus in the soil,

restricting the soluble P to 20% throughout the season could result in P shortage at the maximum

uptake period. Phosphorus uptake can also be greatly increased in mychorizae-colonized

environments, which will critically modify any parameter setting for P availability under normal

conditions. These reasons may have motivated uptake of nutrients including phosphorus by

biological systems to be modeled as a substrate saturation process with end-product inhibition

using the Michaelis-Menten function (Wilson and Botkin, 1990; Lehman et al., 1975a). To

quantify the effects on some model outputs of possible variations of the maximum fraction of

available phosphorus which can be taken up in a day around the approximated average of 0.20,

three uptake fraction levels 0.10, 0.20 and 0.80 were tested in the present sensitivity analysis.

Shoot P and seed P. Optimum and minimum values of shoot and seed phosphorus

concentration used in the plant module were derived from literature (Daroub et al., 2003; Jones,

1983). Although those concentration limits are essentially associated with crop physiology, they

can be subject to cultivar variability. In addition, P modelers have used different optimum shoot

P concentrations. For example, Daroub et al. (2003) used an initial shoot P concentration of 0.7%

whereas Probert (2004) used 0.5%, which is lower than the shoot P level of maize found by

Jones (1983) in his survey. Other studies reported even lower initial shoot P concentration under

non-limiting P conditions (e.g. 0.45, Ziadi et al., 2007). This sensitivity analysis was designed to

cover the range of optimum shoot P variability found in literature. The optimum shoot P

concentration was increased or decreased by 50% around the nominal to obtain the high and low

levels of the input factor "shoot P concentration" (Table 3-9). The minimum shoot P

concentration was taken as 60% of the optimum (Daroub et al., 2003). The optimum seed









concentration was also increased or decreased by 50% around the nominal value of 0.35% to

obtain the high and low levels of the input factor "seed P concentration". The ratio of 2:1

between optimum and minimum seed P concentration was kept for all the input factor levels. The

six input factors and their levels are summarized in Tables 3-8 and 3-9.

Model outputs. The effects of the total input space constituted of the six input factors

were assessed on three model outputs: aboveground biomass, grain yield and total plant uptake.

Method and design of the sensitivity analysis

The analysis was conducted using a global approach where the input factors and their

interactions were explored simultaneously (Saltelli, 2004). The six factors, each having 3 levels

were combined in a complete factorial design yielding 36 = 729 observations or simulation runs.

Sensitivity index

An analysis of variance was carried out on the model outputs using the SAS software

(SAS, 2002). Since formal hypothesis testing cannot be performed due to the lack of a valid error

term, the most useful assessment procedure was to compare the sum of squares contributions

from each factor and interactions to the total sum of squares. This is denoted by the name

"sensitivity index" and is calculated as follow:

SS(FactorF)
SI (Main effect of F) = S c (3-26)
Total SS
SS(InteractionsF)
SI (Interactions involving F) =SS(InteractionsF) (3-27)
Total SS
SI (Total effect ofF) = SI (Main effect ofF) + SI (Interactions involving F) (3-28)
Where SI represents sensitivity index;
F represents factor F;
SS represents ANOVA Sum of Squares;
InteractionsF represents interactions involving factor F.









The sensitivity index was used to measure the effect of factors and their interactions on

outputs. It has values between 0 and 1 with 0 indicating that the model is not sensitive at all to

the factor for the particular output, and 1 indicative of maximum sensitivity.

Results and Discussion

Phosphorus fertilizer and initial PiLabile had the most influential effects on the output

variables. In the absence of P fertilizer, the effect of the fraction of labile P in solution and the

shoot P became also important.

Soil inputs effects

Figures 3-5A to C show the response of crop total aboveground biomass, grain yield and

crop total phosphorus uptake to the three levels of the soil input factors (the levels low, medium

and high along the abscissa have different meaning depending on the input factor considered and

are explained in Table 3-9).

Among the soil input factors tested, initial PiLabile and P fertilizer had the most influential

effects on the three output variables. Aboveground biomass, grain yield and total plant P uptake

responded clearly to variable levels of initial PiLabile and P fertilizer application with the same

pattern. Biomass, grain yield and uptake increased and their corresponding standard deviations

decreased with increasing levels of initial PiLabile and P fertilizer (Figures 3-5A, 3-5B, and 3-

5C).

The sensitivity indices of the response of initial PiLabile and P fertilizer to the three output

variables were relatively high compared to the other input factors (Figure 3-7). These two input

factors alone explained 54%, 47% and 41% of the total sum of squares respectively for

aboveground biomass, grain yield and total plant P uptake (Tables 3-10 to 3-12).

The total sensitivity indices of P fertilizer and initial PiLabile were about two times their

main effects showing that these two input factors were also influential in terms of interaction









with the other input factors. The total sensitivity index of P fertilizer was 0.73 for biomass (Table

3-10), 0.71 for grain yield (Table 3-11) and 0.52 for plant P uptake (Table 3-12). The total

sensitivity index of initial PiLabile was 0.27 for biomass (Table 3-10), 0.31 for grain yield (Table

3-11) and 0.24 for P uptake (Table 3-12).

Organic phosphorus did not contribute much to the variation in any of the output variables

over the range tested. This range that was derived from Sharpley's studies (1984) reflected

approximately the variability in organic P in most P-depleted soils (Brady and Weil, 2002). The

sensitivity indices for this factor were smaller than 0.01 (Tables 3-10 to 3-12).

The response to the input factors initial PiLabile and P fertilizer varied with the output

variable and the level of the input factor. For example, when the six input factors are considered

together, the response was more pronounced for the output variable total P uptake than for

biomass and grain yield. Concerning the individual levels low, medium and high, of the input

factors, the response of the output variables was especially marked at the low levels.

Variability in the input factors initial PiLabile and P fertilizer did not result in a

proportional variability in the output variables. For example, a decrease in initial PiLabile of

75% relative to the nominal value resulted in 16% decrease in biomass and grain yield relative to

the nominal values, and 24% decrease in total P uptake on average (Table 3-14). An increase in

initial PiLabile of 88% resulted in only 8% increase in biomass and grain yield and 11% increase

in total P uptake on average. Biomass increased by 37%, grain yield by 33% and total P uptake

by 42% on average when P fertilizer was increased from 0 to 30 kg [P] ha-1 (Table 3-14). The

increase in the output variables barely exceeded 5% when P fertilizer increased from 30 to 60 kg

[P] ha-.









The reason initial PiLabile, P fertilizer and their interactions have so much effect on the

output variables is that in the model, plants take up phosphorus directly from the labile pool. For

slightly weathered soils with no base saturation measured, as the one used in this sensitivity

analysis, 85% of the P fertilizer applied enters directly the labile pool making this pool the most

important for plant uptake and productivity. The higher sensitivity of total P uptake is associated

with the fact that the uptake is the primary process directly connected to phosphorus availability.

The behavior of the model in terms of response of crop productivity and uptake to initial

PiLabile and fertilizer is supported by numerous fertilizer studies (Colomb et al., 2000; Fofana et

al., 2005; Nziguheba et al., 2002; Pellerin et al., 2000) and does not per se raise new issues.

Initial organic P had virtually no effect on the output variables suggesting that over a

growing season, the organic matter contribution to phosphorus uptake may be more dependent

on the rate constants controlling the mineralization and immobilization of organic P than the

amount of organic P available at the beginning of the growing season. In fact in this sensitivity

analysis, the total P mineralized from organic matter in the soil profile varied from 45 to 55 kg

ha-1 at the end of the season, but the fraction that is actually contributing to plant uptake was

small. This is because 1) the roots do not explore the whole soil profile and therefore cannot

access all the P mineralized; 2) the portion of organic P mineralized that is soluble for plant

uptake is only that amount that enters the volume of the soil where roots are present at the time

of the mineralization. This means that mineralized P not "seen" by the plant at any one time in

the season because the volume of roots is small is transformed over time into insoluble forms

that cannot be used by the plant. The synchronization between the availability of the mineralized

organic P and the accessibility of the plant to it was an influential factor of the small sensitivity

observed of the input factor Initial organic P.









Plant parameters effects

The variability in plant parameters studied had less influence on the output variables than

the soil parameters (Figures 3-6A to C). The graphs were scaled uniformly to Figures 3-5A to C

to highlight the differences in response among the two groups of input factors (soil and plant).

Aboveground biomass, grain yield and plant uptake varied over a smaller range, which resulted

in main sensitivity indices between 0.02 and 0.15 (Tables 3-10 to 3-12). Shoot P had a higher

influence on biomass (main SI of 0.04, Table 3-10), Fraction of labile P had a higher influence

on grain yield (main SI of 0.03) and seed P had a higher influence on total P uptake (main SI of

0.15).

Increase in fraction of labile P from 0.2 to 0.8 resulted in about 7% increase on average in

the output variables. Decrease in this fraction from 0.2 to 0.1 caused an 8% decrease on average

in the output variables (Table 3-14).

Variation of the shoot P concentration from the medium to the low level (Table 3-9)

caused the biomass to decrease by 13%, the grain by 3% on average. However, this decrease in

shoot P concentration increased the total P uptake by 39%. When the shoot P levels were

increased from medium to high, the biomass and the grain yield decreased by 10% but the total P

uptake increased by 18% (Table 3-14).

Variation in seed P from medium to low decreased biomass and grain yield by about 4%

on average. Increase in seed P from medium to high increased biomass and grain yield by only

1% on average. The total P uptake was more responsive to seed P variation. The uptake was

reduced by 39% between the medium and the low seed P level and increased by 18% on average

between the medium and high seed P levels (Table 3-14).

It is possible that higher effects could be detected at wider ranges of the three plant

parameters. However, realistic motivations must support such analyses because the variability in









optimum phosphorus concentration in maize shoots and seed for instance is not of any huge

magnitude (Jones, 1983).

Interactions

As might be expected from their high contributions to the total sum of squares Initial

PiLabile and P fertilizer produced relatively high interactions with other factors. The interactions

between Initial PiLabile and P fertilizer and P fertilizer and shoot P were the strongest (Tables 3-

10 to 3-12).

The interaction plot (Figure 3-8) shows how the strong response to P fertilizer at low initial

PiLabile disappeared very quickly as initial PiLabile increases. This has important implications

when simulating a fertilizer trial: accurate measurements of initial PiLabile must be obtained in

order to achieve good simulation of crop production. On the contrary, the response to Initial

PiLabile did not seem to be affected by the fraction of labile P in solution. Increasing Initial

PiLabile resulted in an increase of biomass at all fraction of labile P in solution levels. However,

the rate of biomass increase did not depend on the level of fraction of labile P in solution (Figure

3-9).

Special case of zero P fertilizer

When the sensitivity indices were recalculated considering the OP fertilizer level only, the

effect on the relative order of importance of the input factors was small (Table 3-13). Initial

PiLabile was the most influential factor (main SI of 0.47, Table 3-13). Shoot P had a much

higher effect on the biomass (main SI of 0.25) than the case when P fertilizer was included. The

Fraction of labile P in solution had also a much higher effect on the biomass (main SI of 0.15)

than the case when P fertilizer was included. The effects of Initial organic P and Seed P remained

relatively small (main and total SIs less 0.01, Table 3-13).









The sensitivity indices of Initial PiLabile, Shoot P and Fraction of labile P in solution

increased because the dominant factor P fertilizer was removed from the input space. The main

sensitivity index of Initial PiLabile and Fraction of labile P in solution increased by a factor of 4

and the main sensitivity index of Shoot P increased by a factor of 6 (Tables 3-10 and 3-13). The

persistence of main sensitivity indices less than 0.01 for Initial organic P and Seed P suggested

that the effects of these two factors on biomass were small per se and were not masked by the

dominant effect of P fertilizer.

However, interactions between factors were weak (highest interaction SI for biomass was

0.04, Table 3-13, compared to highest interaction of 0.15 with P fertilizer, Table 3-10) showing

that most of the interactions between factors were due to the presence of P fertilizer. In fact, the

presence of P fertilizer as a factor had two effects: i) making all interactions involving P fertilizer

relatively stronger than all other interactions between factors (Table 3-13); ii) decreasing

interaction SIs between other factors by accounting for more variability. For example, the

interaction SI between Initial PiLabile and Shoot P was 0.01 with P fertilizer (Table 3-10) and

0.04 without P fertilizer (Table 3-13).

Conclusion

While errors in observations tend to be the direct and most evident target when

discrepancies between simulations and measurements are recorded, inaccurately-measured input

variables and model parameters estimated based on weak evidences or regression analyses can

contribute a great deal to simulation errors. The sensitivity analysis of some key model

parameters and input variables help to learn the behavior of the model and to diagnose in

advance sources of possible simulation errors. A sensitivity analysis on three of the model's

parameters and three of the model's input variables revealed that i) initial PiLabile and P

fertilizer had the greatest impact on grain yield, total plant biomass and total plant uptake of P; ii)









initial organic P had little effect on plant production in the range tested and over a single growing

season; iii) the fraction of labile P in solution, the optimum shoot and seed P concentrations had

smaller effect on grain yield, total plant biomass and uptake in the range of sensitivity used in

this analysis; iv) the relative order of importance of the input factors was not affected by P

fertilizer application but generally, interaction between the factors tested are strengthened in the

presence of P fertilizer.

Accurate estimation of initial PiLabile present in the soil is crucial to simulating crop

productivity in phosphorus deficient cropping systems especially when phosphorus fertilizer is

simulated. Failure to estimate accurately cultivars' optimum shoot and seed P concentration by

50% around the nominal value used in the DSSAT soil-plant phosphorus model can result in

biomass, grain yield and total P uptake variation of up to 39%. Inaccurate estimation of the

fraction of labile P in solution can also become a cause of poor simulation results.

Summary and Conclusion

The soil-plant phosphorus model described in this chapter integrates soil and plant

phosphorus processes that are linked using modular programming techniques to crop growth

models in the DSSAT CSM. The model simulates phosphorus in plants and soils based on

integrated processes between i) inorganic phosphorus present in the soil in three pools, labile,

active and stable; ii) organic phosphorus present in two pools, active and stable; iii) plant

phosphorus present in roots, shoots, shells and seeds. Phosphorus-limited production including

plant biomass, grain yield and plant uptake can also be calculated thanks to the linkage to crop

growth models in DSSAT.

A sensitivity analysis of the model limited to six input factors showed that initial PiLabile

and P fertilizer were the most important forces driving simulations of plant production. The

fraction of root labile P, the shoot and seed P appeared to have less impact on biomass, grain









yield and P uptake of maize. Accurate predictions require therefore that at least initial PiLabile

be measured or estimated correctly. If PiLabile cannot be measured directly and has to be

indirectly estimated based on available P like P-Brayl or Olsen, careful attention needs to be

paid to the relationships derived and their agronomic validity. Diagnosing causes of poor model

predictions should not only focus on checking measurements compared to simulations but also

verifying the validity of input data, initial PiLabile in this instance.









Table 3-1. Soil category-dependent calculation of P availability index
Soil category P availability index
Calcareous (- 0.0058 x CaC03)+ 0.60
Slightly weathered [(0.0043 x TotalBaseSaturation) + (0.0034 x PiLabile)+ (0.11 x pH)]- 0.70
(CLAY +.
Highly weathered 0.30 x log +0.68

Other soils 0.40 + 0.00023 PLab~le
Source: Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in
the tropics. PhD thesis, University of Hawaii, Honolulu.

Table 3-2. Soil category-dependent calculation of P Fertilizer Availability Index
Soil category P Fertilizer Availability Index
Calcareous (- 0.0042 x CaCO3)+ 0.72
Slightly weathered (0.0043 x TotalBaseSaturation)+ (0.0034 x PiLabile)+ (0.11 x pH)- 0.50


Highly weathered


-0.19 xlog ) +0.70
S100


Other soils 0.60 + 0.0002PLabile
Source: Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in
the tropics. PhD thesis, University of Hawaii, Honolulu.

Table 3-3. Summary of decomposition rates for the soil organic pools and C:P ratios at which
phosphorus is allowed to enter the specific pools
Pool generating the flow Pool location Maximum rate at which flow occurs (d-1) Pool C:P
Metabolic litter Surface 0.040550
Soil 0.050680
Structural litter Surface 0.010680
Soil 0.013420
SOM1 Surface 0.016440 50
Soil 0.020000 50
SOM2 Soil 0.000548
SOM3 Soil 0.000012
SOM23 Soil 100
Source: Parton, W.J., Ojima, D.S., Cole, C.V., Schimel, D.S., 1994. A general model for soil
organic matter dynamics: Sensitivity to litter chemistry, texture and management. In:
Bryant, R.B., Arnold, R.W. (Eds), Quantitative modeling of soil forming processes.
SSSA Spec. Publ. 39. SSSA, Madison, WI, pp 147-167.
Parton, W.J., Stewart, J.W.B., Cole, C.V., 1988. Dynamics of C, N, P and S in grassland soils: A
model. Biogeochemistry 5:109-131.
Gijsman, A.J., Hoogenboom, G., Parton, W.J., Kerridge, P.C., 2002. Modifying DSSAT crop
models for low-input agricultural systems using a soil organic matter-residue module
from CENTURY. Agronomy Journal 94, 462-474.









Table 3-4. Optimum and minimum phosphorus content (%) in different plant parts and maximum
and minimum plant N:P ratio at three growth stages, as used in the model for maize
Effective grain filling/
Plant part Emergence f e grn f / Physiological maturity
End of leaf growth*
Root Optimum 0.041 0.041 0.041
Minimum 0.020 0.020 0.020
Shoot Optimum 0.700 0.250 0.200
Minimum 0.400 0.150 0.100
Shell Optimum 0.500 0.500 0.050
Minimum 0.250 0.250 0.025
Seed Optimum 0.350 0.350 0.350
Minimum 0.175 0.175 0.175
Plant N:P ratio Maximum 25.000 15.000 9.300
Minimum 4.200 2.700 2.100
Source: Jones, C.A. 1983. A survey of the variability in tissue nitrogen and phosphorus
concentrations in maize and grain sorghum. Field Crops Research 6, 133-147.
Daroub, S.H., Gerakis, A., Ritchie, J.T., Friesen, D.K., Ryan, J., 2003. Development of a soil-
plant phosphorus simulation model for calcareous and weathered tropical soils.
Agricultural Systems 76, 1157-1181.
*The end of leaf growth applies to shoots only.









Table 3-5. Summary of parameters in the soil-plant phosphorus model
Parameter Unit
P transformations between pools and P availability
Rate constant for transformation from labile P to active P d-1
Rate constant for transformation from active P to labile P d-1
Rate constant for transformation from active P to stable P d-1
Rate constant for transformation from stable P to active P d-1
P availability index unitless
Fraction of root labile inorganic P that is soluble unitless
Shoot P concentrations
Optimum shoot P concentration at emergence g g-1
Optimum shoot P concentration at tasseling g g-1
Optimum shoot P concentration at physiological maturity g g-1
Minimum shoot P concentration at emergence g g-1
Minimum shoot P concentration at tasseling g g-1
Minimum shoot P concentration at physiological maturity g g-1
Root P concentrations
Optimum root P concentration at emergence g g-1
Optimum root P concentration at effective grain filling g g-1
Optimum root P concentration at physiological maturity g g-1
Minimum root P concentration at emergence g g-1
Minimum root P concentration at effective grain filling g g-1
Minimum root P concentration at physiological maturity g g-1
Shell P concentrations
Optimum shell P concentration at emergence g g-1
Optimum shell P concentration at effective grain filling g g-1
Optimum shell P concentration at physiological maturity g g-1
Minimum shell P concentration at emergence g g-1
Minimum shell P concentration at effective grain filling g g-1
Minimum shell P concentration at physiological maturity g g-1
Seed P concentrations
Optimum seed P concentration at emergence g g-1
Optimum seed P concentration at effective grain filling g g-1
Optimum seed P concentration at physiological maturity g g-1
Minimum seed P concentration at emergence g g-1
Minimum seed P concentration at effective grain filling g g-1









Table 3-5. continued
Minimum seed P concentration at physiological maturity g g-
N to P ratios
Maximum vegetative N:P ratio at emergence unitless
Maximum vegetative N:P ratio at effective grain filling unitless
Maximum vegetative N:P ratio at physiological maturity unitless
Minimum vegetative N:P ratio at emergence unitless
Minimum vegetative N:P ratio at effective grain filling unitless
Minimum vegetative N:P ratio at physiological maturity unitless
P mobilization and stress
Maximum fraction of P which can be mobilized from shoot per
day unitless
Minimum value of the ratio of P in vegetative tissue to the
optimum P below which reduced photosynthesis will occur unitless
Minimum value of the ratio of P in vegetative tissue to the
optimum P below which vegetative partitioning will be affected unitless

Table 3-6. Summary of additional inputs required to run the soil-plant phosphorus model in
DSSAT
Input Unit
Initial labile inorganic P ppm
Initial active inorganic P ppm
Initial stable inorganic P ppm
Initial active organic P ppm
Initial stable organic P ppm
P in residue (if applied) %
P fertilizer (if applied) kg ha-1
Soil CEC cmolc kg-1
Soil texture %
Soil CaCO3 content %









Table 3-7. Selected physical and chemical properties of the Kpeve soil used in the sensitivity


analysis,
SLLL
0.180
0.070
0.040
0.060
0.040
0.050
0.080
0.060
0.090


as estimated from pedo-transfer functions in DSSAT


SDUL
0.260
0.140
0.080
0.120
0.060
0.090
0.150
0.110
0.160


SSAT
0.460
0.280
0.160
0.240
0.120
0.180
0.300
0.220
0.320


SRGF
1.000
1.000
0.607
0.497
0.407
0.333
0.273
0.223
0.183


SBDM
0.83
1.08
1.47
0.74
0.47
0.56
0.97
0.77
1.04


SLB, depth, base of soil layer (cm); SLLL, soil lower limit (cm3 cm-3);


SLCF
40.0
40.0
35.0
74.5
88.1
82.3
61.9
72.9
57.9
SDUL, soil


C:P
138
136
130
138
123
218
200
127
124
upper limit,


drained (cm3 cm-3); SSAT, soil upper limit, saturated (cm3 cm-3); SRGF, soil root growth factor
(unitless); SBDM, soil bulk density, moist (g cm3), corrected for gravel content; SLCF, soil
coarse fraction or gravel content (%); C:P, ratio of organic carbon to organic phosphorus
(unitless).

Table 3-8. Summary of inputs factors and outputs for the sensitivity analysis of the P model
Input and output Nominal value Variability limits Unit
variable or parameter (medium value) Lower Upper


Inputs
Soil
Initial inorganic labile P
Initial organic P
P Fertilizer

Plant
Maximum P uptake fraction
Shoot P concentration
Seed P concentration

Outputs
Total plant aboveground biomass
Grain yield
Total plant uptake of P


8
100


0 60


0.2
Medium
Medium


0.1
low
low


0.8
high
high


ppm
ppm
kg P ha-1



unitless
g/g
g/g


kg ha-
kg ha-
kg ha-


SLB
10
20
30
40
50
60
70
80
90









Table 3-9. Specification of the different levels of the input factors "Shoot P" and "Seed P" for
the sensitivity analysis of the P model
End of leaf growth / Physiological
Shoot P concentration (g/g) Emergence Effective grain maturity
filling*


Low

Medium

High


Optimum
Minimum
Optimum
Minimum
Optimum
Minimum


Seed P concentration (g/g)
Low Optimum
Minimum
Medium Optimum
Minimum
High Optimum
Minimum
*End of leaf growth for shoot P


0.0035
0.0021
0.0070
0.0040
0.0105
0.0063


0.0013
0.0008
0.0025
0.0015
0.0038
0.0023


0.0018 0.0018
0.0009 0.0009
0.0035 0.0035
0.0018 0.0018
0.0053 0.0053
0.0026 0.0026
and effective grain filling for seed P.


Table 3-10. Main, interactions, and total sensitivity indices (unitless) of biomass for factors used
in the sensitivity analysis


Main SI


Interactions SI


Total SI


PiLabile


PiLabile
Organic P
P fertilizer
FracLabileP
Shoot P
Seed P


0.11
0.00
0.43
0.04
0.04
0.00


0.00
0.15
0.00
0.01
0.00


Organic P
P fertilizer
0.00 0.15
0.00


0.00
0.00
0.00
0.00


0.05
0.11
0.00


FracLabileP: Fraction of Labile P that is soluble


0.0010
0.0006
0.0020
0.0010
0.0030
0.0018

0.0018
0.0009
0.0035
0.0018
0.0053
0.0026


Shoot
P
0.01
0.00
0.11
0.00


Frac
LabileP
0.00
0.00
0.05

0.00
0.00


Seed
P
0.00
0.00
0.00
0.00
0.00


0.27
0.00
0.73
0.09
0.15
0.00


0.00









Table 3-11. Main, interactions, and total sensitivity indices (unitless) of grain yield for factors
used in the sensitivity analysis
Main SI Interactions SI Total SI


PiLabile Organic
P f
PiLabile 0.11 0.00
Organic P 0.00 0.00
P fertilizer 0.36 0.19 0.00
FracLabileP 0.03 0.00 0.00
Shoot P 0.02 0.01 0.00
Seed P 0.00 0.00 0.00
FracLabileP: Fraction of Labile P that is soluble


P Frac
'ertilizer LabileP
).19 0.00
).00 0.00
0.04


).04
).12
).00


0.01
0.00


Shoot
P
0.01
0.00
0.12
0.01


Seed
P
0.00
0.00
0.00
0.00
0.00


0.00


0.31
0.00
0.71
0.08
0.15
0.01


Table 3-12. Main, interactions, and total sensitivity indices (unitless) of plant uptake of P for
factors used in the sensitivity analysis
Main SI Interactions SI Total SI


PiLabile Organic
P
PiLabile 0.11 0.00
Organic P 0.00 0.00
P fertilizer 0.30 0.09 0.00
FracLabileP 0.03 0.00 0.00
Shoot P 0.09 0.01 0.00
Seed P 0.15 0.02 0.00
FracLabileP: Fraction of Labile P that is soluble


P Frac
fertilizer LabileP
0.09 0.00
0.00 0.00
0.02


0.02
0.10
0.01


0.00
0.00


Shoot
P
0.01
0.00
0.10
0.00


Seed
P
0.02
0.00
0.01
0.00
0.00


0.00


0.24
0.00
0.52
0.05
0.20
0.18


Table 3-13. Main, interactions, and total sensitivity indices (unitless) of biomass for a special
case of zero P fertilizer. The P fertilizer was also removed as a factor.
Main SI Interactions SI Total SI
PiLabile Organic Frac Shoot Seed
P LabileP P P
PiLabile 0.47 0.00 0.01 0.04 0.00 0.52
Organic P 0.00 0.00 0.00 0.00 0.00 0.00
FracLabileP 0.15 0.01 0.00 0.01 0.00 0.17
Shoot P 0.25 0.04 0.00 0.01 0.00 0.30
Seed P 0.00 0.00 0.00 0.00 0.00 0.00
FracLabileP: Fraction of Labile P that is soluble









Table 3-14. Mean aboveground biomass, grain yield and total P uptake corresponding to each
level of the input factors used in the sensitivity analysis. Each mean contains 243 =
729/3 observations.
Input Factors Output Variables (kg ha-1)
Biomass Grain yield Total P uptake
PiLabile (ppm)
2 8282 3007 18
8 9885 3567 23
15 10704 3839 26
Organic P (ppm)
40 9634 3474 22
100 9608 3465 22
200 9629 3474 22
P fertilizer (kg ha-1)
0 6872 2604 15
30 10867 3885 25
60 11132 3925 27
Fraction of solution P (unitless)
0.1 8886 3228 20
0.2 9643 3493 22
0.8 10342 3693 24
Shoot P (See Table 3-12)
Low 9084 3494 26
Medium 10401 3621 19
High 9387 3299 22
Seed P (See Table 3-12)
Low 9488 3403 27
Medium 9738 3530 17
High 9646 3481 23









Plant Phosphorus Processes


Uptake


Net P
IMineralized PSurface PSoil
Litter Litter

Root NoRt Structural Metabolic Metabolic Structural
Root NoRoot
Labile Pi Labile Pi

Fertilizer P PSurface PSoil
Root NoRoot SOM1 SOM1
Active Pi Active Pi


Root NoRoot PSoil
Stable Pi Stable Pi SOM23

Soil Inorganic Phosphorus Processes Soil Organic Phosphorus Processes


Figure 3-1. Processes in the integrated soil-plant phosphorus model in DSSAT












0.008
-0.007 Optimum
0.007
\ *-Minimum
= 0.006

2 0.005
a.
0.004

0.003

| 0.002 -
0
o 0.001 -
a.
0
Emergence End of leaf growth Physiological Harvest
Maturity
growth Stage


Figure 3-2. Optimum and minimum P concentration in maize shoots used in the plant P model



30.000
--Maximum
25.000 --Minimum


o 20.000


15.000


.. 10.000 -


5.000 -


0.000
Emergence End of leaf growth Physiological Harvest
Maturity

growth Stage


Figure 3-3. Maximum and minimum N:P ratios used in the plant module to limit uptake of P















1.00


0.80


0.60


0.40


0.20


0.00


0 10 20 30 40 50 60 70 80 90 100

days after planting


-Optimum shoot P
P stress partitioning


-Minimum shoot P
-P stress photosynthesis


Figure 3-4. Relationship between maize shoot P concentrations and P stresses affecting
vegetative partitioning and photosynthesis


16000

B 12000 T
10000



2000
Low Medium High
input fadcor level
SPiLabile Organic P P fertizer




Lo Medium High
input factor level
I PILabile DOrganic PP P ertilier
0[


125
20
15



L- Medium High
input factor level
[ PLablle O Organi P 0 P Fertilizer]


I I.B


inputfactorlevel
ie Oraanic P P F


Figure 3-5. Simulated plant total aboveground biomass, grain yield and plant uptake of P at
different levels of initial PiLabile, initial organic P and P fertilizer. Error bars shown
represent one standard deviation. A) Aboveground biomass. B) Grain yield. C) Plant
uptake of P.


Actual shoot P















16000
















o Uptake fraction a Shoot P *Seed P
14000
i. 12000
10000
8000
0 6000
4000
A 2000


Low Medium High
input factor level
o Uptake fraction o Shoot P Seed P


40











C o
LOW Medium High
input factor level
D Uptake fraction O Shoot P Sooeed P


5000
4500
4000
3500
3000
2500
2000
S1500
1000
500
0 B
Low Medium High
input factor level
O Uptake fraction O Shoot P m Seed P


Figure 3-6. Simulated total plant aboveground biomass, grain yield and plant uptake of P at

different levels of maximum uptake fraction, optimum shoot and seed P

concentrations. Error bars shown represent one standard deviation. A) Aboveground

biomass. B) Grain yield. C) Plant uptake of P.












Shoot P*Seed P

Fraction of solution P*Seed P

Fraction of solution P*Shoot P

P fertilizer*Seed P

P fertilizer*Shoot P

P fertilizer*Fraction of solution P

Organic P*Seed P

Organic P*Shoot P

Organic P*Fraction of solution P

Organic P*P fertilizer

PiLabile*Seed P

PiLabile*Shoot P

PiLabile*Fraction of solution P

PiLabile*P fertilizer

PiLabile*Organic P

Seed P

Shoot P

Fraction of solution P

P fertilizer

Organic P

PiLabile

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Sensitivity index for total aboveground biomass

Figure 3-7. Sensitivity indices for the six input factors and their interactions












12000


10000


S8000
E
o0
S6000


S4000
0
C 2000

0
0 -
0 30 60

P fertilizer levels (kg ha"1)

-- PiLabile=2 -m- PiLabile=8 PiLabile=15

Figure 3-8. Simulated response of total plant aboveground biomass to phosphorus fertilizer at
different levels of PiLabile


12000



8000

co 8000
E
o
V 6000

0
o 4000
0
.0
l 2000


0
2 8 15
Inorganic labile P levels (ppm)



--Fraction of labile P=0.10 ---Fraction of labile P=0.20 Fraction of labile P=0.80

Figure 3-9. Simulated response of total plant aboveground biomass to PiLabile at different levels
of fraction of labile P









CHAPT ER 4
FIELD TESTING OF THE DSSAT PHOSPHORUS MODEL

Introduction

The complex nature of relationships between components present in agricultural systems

suggests the use of simulation techniques to study those systems rather than directly

experimenting continuously on the systems themselves. Simulation models can assist with

assessing alternatives and making decisions that would consume the entire career of an

agronomist (Struif Bontkes and Wopereis, 2003; Matthews et al., 2000).

Because models developed in a certain environment can be adapted and applied in

different agroecological conditions, the suitability of a model to simulate processes of interest is

a major criterion in order to achieve meaningful inferences. Models have become so complex

and been described with so many variables and parameters that their degrees of freedom have

increased drastically. With an appropriate choice of input variable and parameter values they can

be made to produce realistic outputs that can agree erroneously with real world measurements.

Therefore, in addition to the suitability criterion, model testing or evaluation using the right

combination of inputs and parameters is an important step that diagnoses the ability of the model

to capture appropriately the essence of crop-environment interactions and their variability at the

meso and micro scales. Annino and Russell (1981) underlined the risks associated with the

application of simulation models that have not passed the test of a sound scientific assessment.

They cited the use of an untested or invalid model as one of the seven most frequent causes of

failure in many simulation modeling studies.

To enable crop models in DSSAT to simulate phosphorus dynamics in cropping systems, a

soil-plant phosphorus model was modified from initial studies by Daroub et al. (2003) and

implemented in the software. Modifications applied to the Daroub et al. version of the model









include: 1) linkage of the model to the CENTURY module to allow simulations of organic P

transformations in soils; 2) implementation of a generic, modular crop P module that is usable by

all crops in DSSAT, and 3) addition of algorithms for initialization of the different phosphorus

pools using measured soil phosphorus data. Daroub et al. (2003) reported that the initial soil-

plant phosphorus model simulated with good accuracy P uptake for maize grown under acidic

conditions when linked to the DSSAT CSM. The redesigned soil-plant phosphorus model still

operates as an experimental version and has not been tested yet. This chapter is centered on

evaluating the soil-plant phosphorus model described in chapter 3. The datasets used for the

evaluation of the model are two phosphorus experiments conducted in Ghana and described in

Chapter 2. Results from these experiments are also discussed in Chapter 2.

The main question addressed in this chapter was: can the soil-plant phosphorus model

simulate the responses of maize biomass and grain yield to different levels of phosphorus as

observed in the field in Ghana?

The objectives of the present chapter are: 1) to describe selected methods for evaluating the

performance of the soil-plant phosphorus model; 2) to present an assessment of the ability of the

soil-plant phosphorus model to simulate soil and crop conditions in two locations in Ghana

(Kpeve and Wa) using those tools.

To meet these objectives, some model parameters (genetic coefficients describing the

cultivar used and the fraction of labile P in solution) were first calibrated using essentially the

dataset from Wa and partially the dataset from Kpeve. The evaluation reported in this chapter

focused on the following key outputs: grain yield, aboveground biomass and shoot P

concentrations.









Materials and Methods

The Kpeve and Wa datasets described in Chapter 2 were used to evaluate the performance

of the model.

The Soil-Plant Phosphorus Model

The model simulates phosphorus transformations between 1) three inorganic pools:

labile, active and stable; 2) two organic pools: active and stable, and 3) four plant parts: roots,

shoots, shells, and seeds. The model was implemented in DSSAT for CERES-Maize to enable

the maize model to predict nitrogen and phosphorus-limited maize production as affected by

cultivar, soil, weather, and management information.

The soil inorganic P module of the model simulates phosphorus transformations between a

labile, active and stable pool. The soil organic P module simulates phosphorus transformations

between a surface litter, a microbial pool, and a stable pool. The model accounts for the

mineralization of organic P to inorganic pools and the immobilization of P to organic pools.

Available phosphorus for uptake by plants is described as being provided by the labile pool

within 2 mm of plants' roots.

Phosphorus taken up by the plant is partitioned to seeds, shells and vegetative tissues.

During the reproductive phase, phosphorus accumulated in the vegetative tissues can be

remobilized and translocated to seeds. Plant growth is limited by phosphorus between two

thresholds that are species-specific optimum and minimum concentrations ofP defined at three

stages in the growth of the plant. Phosphorus stress factors are computed to reduce

photosynthesis, dry matter accumulation and partitioning.

A sensitivity analysis of the model to some key phosphorus-related parameters established

that the model responds well to phosphorus fertilizer applications on soils with low initial

available phosphorus (Chapter 3). The analysis also isolated the initial soil inorganic labile









phosphorus and the fraction of P that is available for uptake per day as two important soil

parameters that have significant influence on major model outputs.

Datasets for Testing the Model

The datasets used to evaluate the model came from two phosphorus experiments carried

out in Ghana in 2004 and 2006. A description of the experiments is provided in Chapter 2. The

treatments at Kpeve, OP, 10P, 30P, 80P, received respectively 0, 10, 30, and 80 kg [P] ha-1. At

Wa, the treatments were combinations of levels of 2 factors: nitrogen fertilizer, 3 levels, 0, 60,

and 120 kg [N] ha-l; phosphorus fertilizer, 3 levels, 0, 60, and 90 kg [P] ha-1. The experiment

implemented in Kpeve in 2006 did not respond to phosphorus although available P measured as

Bray-1 was low. It was found that the soil in Kpeve had relatively high organic matter content

(1.8%) and other phosphorus forms that could have been made available to the plant during the

growing season. This observation may help explain the lack of response to phosphorus observed

at Kpeve. The second experiment conducted in Wa responded well to phosphorus and nitrogen

fertilizer applications.

Parameters and Inputs for the Model Tests

The parameters for the model tests included genetic coefficients and phosphorus-related

parameters. Inputs included soil and weather conditions. The parameters and inputs used are

described next.

Weather conditions

The experiment in Kpeve, Southern Ghana (6 40.80' N, 0 19.20' E, altitude 67 m above

sea level, Figures B-l and B-2) was conducted in 2006 during the primary rainy season (March

to July). The site has a bimodal rainfall pattern with an average annual rainfall of 1300 mm

falling in two rainy seasons, March to July and September to October (Figure 2-1). The average

annual temperature is 28 degrees C.









The experiment in Wa, Northern Ghana (10o3' N, 2o30' W, altitude 320 m above sea level,

Figures B-l and B-2) was carried out during the only rainy season in 2004. The rainfall pattern in

Wa is unimodal. The average annual rainfall is 1100 mm falling mainly between April and

September (Figure 2-3). The mean annual temperature in Wa is 27 C.

Soil conditions

The soil in Kpeve has a sandy loam texture and is classified as Haplic Lixisol which has a

dark grayish brown topsoil and grayish brown to brown subsoil (Adiku, 2006). Soil analysis

(Table 2-15) showed that the soil has good organic carbon content and available phosphorus

(Brayl) that is at the limit between sufficiency and deficiency (11.69 ppm). The relatively high

Mehlichl P (90.44 ppm) value obtained from a different soil testing laboratory suggested that

this soil may not be severely P-deficient.

The soil in Wa has a loamy sand texture with very low levels of organic carbon, organic

nitrogen, available P (Brayl) and exchangeable K (Table 2-18).

Genetic coefficients

The genetic coefficients for the cultivar used, Obatanpa were calibrated based on the

growth and development data obtained essentially from Wa for the high N and P treatments.

Because the experiment in Kpeve was affected by a drought spell starting at silking, the dataset

from this location was not quantitatively involved in the calibration of the genetic coefficients for

Obatanpa. However, qualitative comparisons were used to ensure that the model predicted well

the anthesis date (that was not affected by the drought) at Kpeve as well.

Since the cultivar Obatanpa was described as a medium-maturing variety with a maturity

period of 105-110 days (Anonymous, 1996), the calibration starting values of the genetic

coefficients (Table 4-1) were from a medium-duration cultivar taken from the DSSAT database

of cultivars. The coefficients were manually adjusted until an agreement between simulated and









measured days to anthesis and to physiological maturity, biomass and grain yield was obtained.

The development coefficients (P1 and P5) were adjusted first using measured days to silking and

physiological maturity from the Wa experiment. The growth coefficients G2 and G3 were

calibrated next, using measured end-of-season biomass and grain yield from Wa. The coefficient

PHINT for thermal time between the appearances of two successive leaf tips was not changed

because leaf number data was not available. Since water stress affected the phenology and

probably the biomass and grain yield in the Kpeve experiment, the data from this experiment was

not used in any genetic coefficient calibration. The calibrated coefficients for Wa was used with

no further altering for model evaluation purposes at Kpeve.

Phosphorus parameters

Optimum and minimum P concentrations in roots, shells, seeds as well as maximum and

minimum N:P ratios were taken from the literature (Jones, 1983; Probert and Okalebo, 1992;

Daroub et al., 2003; Probert, 2004). Optimum shoot P concentration at different stages of growth

was estimated using the following equations (Jones, 1983):

At emergence and end of leaf growth: Optimum shoot P concentration (%) = 0.684 0.108X

At physiological maturity: Optimum shoot P concentration (%) = 0.238 0.0056X

Where:

Xis the growth stage.

Emergence was defined as growth stage 0 (X= 0), end of leaf growth as growth stage 4, and

physiological maturity as growth stage 10 (Jones, 1983). Minimum shoot P concentration was

taken as 60% of the estimated optimum (Daroub et al., 2003).

For the soil parameters, the rate constants for inorganic P transformation from labile to

active pools (KLA), active to labile pools (KAL), and active to stable pools (KAS) were estimated

from the value of the P availability index respectively using equations 3-5, 3-6 and 3-7. The P









availability index was approximated as 0.40 (Table 3-1, Other soils). The fraction of soluble P

was adjusted until the lowest error between simulated and measured grain yield was obtained

using the Wa dataset.

Initial conditions

Initial PiLabile was calculated from measured P Brayl and exchangeable K (Table C-l) as

(1.09*PBrayl) + (10.59*ExchangeableK) + 2.71 for both sites, Kpeve and Wa. Initial PiActive

and PiStable were calculated using equations 3-4 and 3-7.

Initial total organic P was calculated from the values of organic carbon and pH (Tables 2-

-1 5x pH 102
15 and 2-18 as 900 x e 12 x (l e 010xorgcC ) (Singh, 1985). This total organic P was

partitioned as 6% active and 94% stable (Parton et al., 1988, 1994; Gijsman et al., 2002).

Measured soil parameters that included soil organic carbon and nitrogen (Tables 2-15 and

2-18) were used as input to the crop model. Other soil parameters not measured but necessary to

run all DSSAT models were estimated using pedotransfer functions in DSSAT. Soil's water

lower limit (SLLL), drained upper (SDUL) and upper limit saturated (SSAT) for Kpeve were

taken from Adiku (2006). The bulk density used at Kpeve was corrected for gravel content using

equation 2-1. At Wa, the SLLL, SDUL, SSAT and bulk density values were estimated using

pedotransfer functions in DSSAT.

Initial soil water at planting was set at the SDUL level for both sites. Initial nitrate-N and

ammonium-N were not measured and assumed to be 0.01 ppm.

Model Evaluation

Simulation of exact real world values by models would not generally be expected because

of the many simplifications with which the model approximates reality. The primary concern of

model evaluation is comparing simulations and measurements and explaining possible









deviations. In this study, attention will be given to the analysis of these deviations to assess the

model performance and gradually introduce modifications to get more understanding of the

causes of simulation error. The evaluation presented here was a first step in testing the ability of

the model to mimic the wide differences in responses to P.

Model evaluation tools

Simple scatter plots were used wherever appropriate to stimulate intuitive and preliminary

evidence of model performance. The simulations and measurements were compared globally

using a standardized mean deviation, the root mean square error (RMSE):

[I ) \2 -05
RMSE (Simulation -Measurement)2 (4-1)
N
Where N is the number of pairs of measurements and simulations.

The RMSE estimates the dispersion between simulated and measured data (Du Toit et al.,

1997).

The RMSE can be expressed relative to the mean of measurements to visualize how the

deviation compares to the average observation:

RMASE
RRMSE = x 100 (4-2)
M
Where RRMSE is the relative RMSE (in percent) and
M is the mean of measurements.

Deviations between simulations and measurements can be furthermore explored by

partitioning the overall RMSE into components that relate to specific types of discrepancies

(Kobayashi and Salam, 2000; Gauch et al., 2003). If the simulations and the measurements

agreed perfectly, the (simulation, measurement) pairs of points would be aligned along the 1:1

line in a scatter plot and the RMSE would be equal to zero. This perfect agreement situation

would mean the following: 1) the mean of simulations, S, equals the mean of measurements, M;









2) if a regression analysis was performed, the slope of the equation would be equal to 1; and 3)

the coefficient of determination R2 resulting from a simple linear regression analysis would be

equal to 1. An RMSE different from zero can therefore be envisioned as the result of three

potential problems:

* Problem 1: The model failed to simulate the mean of measurements, introducing a

simulation bias: there is shift in the fitted regression line from the original perfect agreement

line. This situation can be quantified by the Squared Bias: SB = (S -M)2 (Kobayashi and

Salam, 2000). SB reveals a possible trend of the model to overestimate or underestimate the

measurements.

* Problem 2: The deviation is the result of the model failing to simulate correctly the

magnitude of fluctuation among the measurements: there is rotation of the fitted regression

line around the perfect agreement line with the axis of rotation passing through the origin.

This condition can be measured by the Squared Difference between the Standard Deviations,

SDSD,

SDSD = (SD, SDm)2 (4-3)

Where SDs is the standard deviation of simulations and

SDm is the standard deviation of measurements.

* Problem 3: The deviation is attributable to the failure of the model to simulate the pattern of

the fluctuation across the measurements: the pairs of points would appear in a random pattern

in a scatter plot. This situation can be quantified by the Lack of positive Correlation weighted

by the standard deviations,

LCS = 2SD,SD(1- r) (4-4)
Where r is the Pearson coefficient of correlation.









The LCS can also be interpreted as the residual error sum of squares after removing SB

and SDSD. Kobayashi and Salam (2000) found that the three components SB, SDSD and LCS

add up to the Mean Squared Error, MSE

MSE = SB + SDSD + LCS (4-5)

Since MSE = (RMSE)2, equation (4-5) can be rewritten to relate the RMSE to the

components of the model error:

(RAISE)2 = SB + SDSD + LCS (4-6)

The advantage of using the partitioned MSE resides in the possibility to investigate what

components of the overall model deviation were most important.

Results and Discussion

The soil-plant P model was able to capture the response of maize to P fertilizer as observed

at both sites. Results of genetic coefficients calibration and P-related parameters estimation are

also described next.

Weather

The crop in Kpeve experienced a drier than average July (2006) (Figure 2-1) that affected

maize phenology and growth. The major season, which normally ends in late July, ended earlier

(in June) at a critical stage during crop growth. The total rainfall in July was only 40.40 mm,

which was below the calculated 2003-2005 average for that month (Figure 2-1). Although the

total rainfall received in 2006 from planting to harvest was higher than the amount received

during the same period in 2003-2005 (713 mm in 2006, 464 mm in 2005, 612 mm in 2004, and

526 mm in 2003), the rainfall received in the month of July 2006 was low: the July total rainfall

was 60 mm in 2003, 105 mm in 2004 and 26 mm in 2005 compared to 40.40 mm during the year

of the experiment.









As a response to this unexpected drought, the field was sprinkler-watered for three days

from July 26th to July 28th (60 to 62 days after planting). Because irrigation equipment was not

set up on the field at the commencement of the trial, the sprinkler-watering was improvised,

which delayed the water application for about 10 days after drought symptoms were first

observed, and provided only about 15 mm of water. The adverse effects of rainfall variability

and unreliability at Kpeve in recent years were also pointed out by Adiku (2004 and 2006).

Genetic Coefficients

Thermal time related to days to anthesis (P1) was set (Table 4-1) so that the model could

simulate correctly the measured anthesis dates at Kpeve and Wa. Thermal time related to days to

physiological maturity (P5) was set (Table 4-1) so that the model could simulate correctly the

measured physiological maturity dates at Wa. The potential kernel number per plant (G2) was

increased from 700 to 900 and the potential kernel growth rate (G3) decreased from 8.50 to 6.50

mg/day to obtain the best fit to the measured grain yield at Wa.

Phosphorus Parameters

Optimum and minimum P concentrations in roots, shells, seeds and maximum and

minimum N:P ratios taken from the literature (Jones, 1983; Probert and Okalebo, 1992; Daroub

et al., 2003; Probert, 2004) are presented in Table 4-2. Estimated optimum and minimum shoot P

concentrations using the equations developed by Jones (1983) are presented in Table 4-2. To

reflect the fact that phosphorus stress should affect vegetative partitioning before photosynthesis,

the minimum value of the ratio of P in vegetative tissue to the optimum P below which reduced

photosynthesis occurs was set to 1.0, and the minimum value of the ratio of P in vegetative tissue

to the optimum P below which vegetative partitioning will be affected was set to 0.8. It was

assumed that the maximum fraction of P which can be mobilized from shoot per day cannot

exceed 0.10.









The estimated soil P parameters were identical for both Kpeve and Wa (Table 4-3). This

was because all of the soil P parameters (except the fraction of labile P in solution) depend on the

value of P availability index. The dependency of P availability index on Initial PiLabile as shown

in the equation 0.40 + 0.00023Initial PiLabile is weak and the calculation essentially yields 0.40 for

Initial PiLabile values > 1 ppm (Initial PiLabile was 16.52 ppm at Kpeve and 6.49 ppm at Wa).

The adjustment to the fraction of labile P in solution was challenging. Specific studies

were not conducted on this fraction in the way it is used in the soil-plant phosphorus model

discussed here. Studies that suggested a value of 0.015-0.020 related the fraction directly to the

total PiLabile pool (Daroub et al., 2003). Since in the present phosphorus model, the fraction

applies to the part of PiLabile in the root zone only, it was certain that the calibrated value of this

fraction would be higher than 0.020. Through calibration using the Wa dataset a value of 0.20

was obtained. This value was also used to evaluate the model at Kpeve. A sensitivity analysis on

this fraction of labile P in solution showed that this parameter did not have as much influence on

model outputs as the size of the initial PiLabile pool itself and the optimum shoot P

concentration (Tables 3-10 and 3-13).

Initial Conditions

Initial sizes of the different phosphorus pools for both sites are given in Table 4-4. The

initial PiLabile in the soil at Kpeve was nearly three times that of Wa (Table 4-4). Organic P was

relatively high at Kpeve (Table 4-4). Other soil parameters are summarized for Kpeve in Table

4-5 and for Wa in Table 4-6.

Model Evaluation at Kpeve

The soil-plant P model was able to capture the lack of response to P as observed in the

experiment at Kpeve.









In-season growth

Simulation of accumulated biomass over time was in good agreement with measurements

(Figure 4-3). The RMSE of 87 kg ha-1 at 17 dap increased with biomass over time but remained

at about 470 kg ha-1 between 31 dap and final harvest (108 dap). The RMSE increase during the

season was due to increasing biomass values. In relative terms, the simulation actually improved

over time. The RRMSE was only 5% at final harvest (Table 4-7).

The early season error was mostly due to an overprediction by the model that resulted in a

high squared bias (Figures 4-4 and 4-3). At anthesis (52 dap) and final harvest (108 dap), the SB

was less and the error due to the pattern of variation among the measurements (LCS) became the

important component of MSE (Figure 4-4), but the overall errors were actually small (Table 4-7).

The negative correlation coefficients observed at 17 dap and anthesis (52 dap) were caused by

two opposite trends in the variation of the biomass: measured biomass decreased while simulated

biomass increased at those periods with increasing P applications. The observed decreasing

biomass with P additions was not significant.

Final grain yield

Predicted grain yields were in good agreement with measurements (Figure 4-1). The

RMSE was 255 kg ha-1 representing 8% of the mean of measurements. The model captured well

the lack of response to P at Kpeve as shown by the non significant differences among the

measured grain yields (Figure 4-2) even though the growth data from Kpeve was not used in

calibrating the genetic coefficients of the cultivar Obatanpa. The simulation error was mainly due

to the pattern of variation of grain yield among the four treatments (Figure 4-2). This was also

reflected in the low correlation coefficient observed between measurements and simulations

(0.36). The statistical non significance of the grain yield means that the slight differences

observed in grain yield among the four treatments were not determined by the phosphorus









applied but rather to other causes such as measurement error or field variability. Since a

deterministic model does not account for such fluctuations, the low RMSE suggested good

performance for final yield.

Wa

At Wa, the model predicted the response of maize to both nitrogen and phosphorus with

higher error than Kpeve.

In-season growth

The aboveground biomass was underpredicted by the model at most planting dates in all

treatments (Figure 4-9). The RMSE varied with sampling date between 216 and 2574 kg ha-1,

which corresponded to 19-57% RRMSE values (Table 4-7). However, the different components

of the MSE showed that the general tendency of the model to underpredict the biomass did not

actually affect its ability to effectively capture most of the responses of biomass to nitrogen or

phosphorus fertilizer. The LCS or the SDSD, which correspond to the failure of the model to

simulate correctly the pattern or the magnitude of fluctuation among the measurements,

generally represented the smallest portion except for days after planting 46 (Figure 4-8).

The correlation coefficient between simulated and measured biomass at each sampling date

can lead to misleading interpretations when used alone. For example, the correlation coefficient

had the same value of 0.97 at days 46 and 61 after planting. However, the RRMSE doubled from

46 to 61 dap (from 24 to 45%) (Table 4-7). The increase in the RRMSE is an indicator of a

progression towards a poorer performance of the model, but at the same time the persistence of a

high correlation coefficient suggests that the strong linear association between simulations and

measurements has not been lost. The problem with the use of correlation and linear regression

alone for model evaluation is that when simulations and measurements are treated as dependent

and explanatory variables, many assumptions of the analysis are violated (Mitchell, 1997).









In-season shoot P concentration

The performance of the simulation of shoot P concentration at Wa depended on N and P

fertilization of the crop: 1) When neither nitrogen or phosphorus were applied (treatment ON OP,

Figure 4-10), simulation of shoot P concentration followed a similar pattern as in phosphorus-

fertilized treatments (Figure 4-10); 2) In the no phosphorus treatments that received nitrogen

(treatments 60N OP, 120N OP, Figure 4-10), simulated shoot P concentration was less variable

and remained closer to the minimum shoot P concentration than the measurements (Figure 4-10);

In treatments that received both nitrogen and phosphorus fertilizer, simulated and observed

shoot P concentration were similar during the vegetative phase (Figure 4-10). After this phase,

simulated shoot P concentration remained stable at a higher level than measured (Figure 4-10).

These response patterns are summarized in Figure 4-11. At least three problems of

incompatibility between simulations and measurements are highlighted in this figure: 1)

Simulated shoot P concentration in the OP treatment was lower than in the 60 and 90P

treatments, which logically reflects low available soil P in the OP treatment, low P uptake and

low P status in the plant due to high P stress on biomass growth. This is the expected relationship

between shoot P concentrations in plants grown on P-limiting and non P-limiting soils as found

in several studies surveyed in Jones, 1983. In the experiment reported here, the shoot P

concentrations did not show much variation between P-limiting and non P-limiting conditions

(Figure 4-11). Shoot P concentration varied in the experiment between 0.58% and 0.05% on

average regardless of the treatment considered. 2) The measured decrease in shoot P

concentration from 0.57 at 28 dap to 0.05% at 125 dap was also in contrast to the simulated 0.50

to 0.20% during the same period in the well-supplied phosphorus treatments (Figure 4-11). In-

season variability in shoot P concentration found in other studies under non P-limiting conditions

is 0.50% 0.25% (Plenet et al., 2000a) and 0.45% 0.15% (Ziadi et al., 2007). The shoot P









concentration measured at maturity (125 dap) was lower than any of these values in the

experiment reported here (0.05%).

Final grain yield

The grain yield in Wa was simulated with a low error. The RMSE of 266 kg ha-1

represented 13.6% of the mean measurement. The mean difference between simulations and

observations (bias) was only 3 kg ha-1. The simulations agreed well with the data (Figure 4-5).

The model was able to capture the three yield ranges that are dependent on the amount of

nitrogen applied with a correlation coefficient of 0.99 (Figure 4-5). The differential responses to

nitrogen and phosphorus fertilizer were equally well simulated (Figure 4-6).

The decomposition of the error revealed that much of the overall MSE could essentially be

partitioned among the SDSD and the LCS (Figure 4-7). The SB was very small because the

mean measurement was well predicted (mean of measurements = 1961 kg ha-1; mean of

simulations = 1958 kg ha-1).

The prediction of the measured variance of yield was also good (standard deviation of

measurements = 1336 kg ha-1; standard deviation of simulations = 1472 kg ha-1). The variances

of the grain yield themselves were generally high because of the wide range in fertilizer inputs

(0, 60 and 120 kg ha-1 for nitrogen and 0, 60 and 90 kg ha-1 for phosphorus). These high grain

yield variances reflected the low nitrogen and phosphorus status of the soil prior to the start of

the experiment, which made the soil responsive to the application of either nutrient. The

relatively high LCS (compared to the other model error components) does not mean, in this

particular situation, that the model failed to simulate correctly the pattern of variation among the

measurements because 1) the overall simulation error was low and 2) the LCS is a weighted

product of the standard deviations which are inherently high in this experiment.









Conclusion

The assessment presented in this paper showed that the soil-plant phosphorus model

simulated maize grain yield and biomass with a good degree of accuracy both under phosphorus-

limiting (Wa) and non phosphorus-limiting (Kpeve) conditions in Ghana.

Grain yield was simulated with an RRMSE of 8% at Kpeve and 14% at Wa. Final biomass

was simulated with an RRMSE of 5% at Kpeve and 30% at Wa. The higher errors at Wa were

mostly due to more bias in biomass simulations, but the model actually simulated well the

response to P fertilizer.

Simulation of shoot P concentration at Wa was generally good and in agreement with in-

season shoot P variability found in the literature. However, the shoot P concentrations measured

in the Wa experiment at 81 dap and harvest maturity (125 dap) were exceptionally low (0.05%).

The soil-plant P model captured the observed response to P fertilizer at Wa, and lack of

response to P fertilizer at Kpeve. These results are promising because this is a first evaluation of

the model across two contrasting conditions of P availability to plants.

However, the calibration of the fraction of labile P in solution was challenging mostly

because it is difficult to measure directly and sufficient information was not available in the

literature. Different values of this parameter might work better on other soil types. This

parameter was found to be influential on plant P uptake, biomass and grain yield when P

fertilizer was not applied.

Methods for indirect estimation the initial sizes of the inorganic and organic P pools that

play an important role in the response of the model to P have uncertainties associated with them.

Model performance can be expected to improve as refinements are introduced in these methods.









Table 4-1. Growth and development genetic coefficients for the Obatanpa cultivar used at both
sites, Kpeve and Wa, for testing the phosphorus model


Definition
Degree days (base 8C) from
emergence to end of juvenile phase
Photoperiod sensitivity
Degree days (base 8C) from silking to
physiological maturity
Potential kernel number (/plant)
Potential kernel growth rate (mg/day)
Phyllochron


DSSAT ID Starting Value Obatanpa


200
0.00

800


G2
G3
PHINT


700
8.50
38.90


300

0.00
830

900
6.50
38.90









Table 4-2. Plant parameters used for testing the phosphorus model at Kpeve and Wa
Parameter Unit Value
Shoot P concentrations
Optimum shoot P concentration at emergence % 0.70
Optimum shoot P concentration at tasseling % 0.25
Optimum shoot P concentration at physiological maturity % 0.20
Minimum shoot P concentration at emergence % 0.40
Minimum shoot P concentration at tasseling % 0.15
Minimum shoot P concentration at physiological maturity % 0.10
Root P concentrations
Optimum root P concentration at emergence % 0.041
Optimum root P concentration at effective grain filling % 0.041
Optimum root P concentration at physiological maturity % 0.041
Minimum root P concentration at emergence % 0.020
Minimum root P concentration at effective grain filling % 0.020
Minimum root P concentration at physiological maturity % 0.020
Shell P concentrations
Optimum shell P concentration at emergence % 0.50
Optimum shell P concentration at effective grain filling % 0.50
Optimum shell P concentration at physiological maturity % 0.050
Minimum shell P concentration at emergence % 0.25
Minimum shell P concentration at effective grain filling % 0.25
Minimum shell P concentration at physiological maturity % 0.025
Seed P concentrations
Optimum seed P concentration at emergence % 0.35
Optimum seed P concentration at effective grain filling % 0.35
Optimum seed P concentration at physiological maturity % 0.35
Minimum seed P concentration at emergence % 0.175
Minimum seed P concentration at effective grain filling % 0.175
Minimum seed P concentration at physiological maturity % 0.175
N to P ratios
Maximum vegetative N:P ratio at emergence unitless 28.0
Maximum vegetative N:P ratio at effective grain filling unitless 15.0
Maximum vegetative N:P ratio at physiological maturity unitless 9.3
Minimum vegetative N:P ratio at emergence unitless 4.2
Minimum vegetative N:P ratio at effective grain filling unitless 2.7









Table 4-2 continued
Minimum vegetative N:P ratio at physiological maturity unitless 2.1
P mobilization and stress
Maximum fraction of P which can be mobilized from shoot per day unitless 0.10
Minimum value of the ratio of P in vegetative tissue to the optimum P unitless 0.80
S. unitless 0.80
below which reduced photosynthesis will occur
Minimum value of the ratio of P in vegetative tissue to the optimum P unitless 1.00
below which vegetative partitioning will be affected
Source: Jones, C.A. 1983. A survey of the variability in tissue nitrogen and phosphorus
concentrations in maize and grain sorghum. Field Crops Research 6, 133-147.
Jones, C.A., Cole, C.V., Sharpley, A.N. Williams, J.R., 1984a. A simplified soil and plant
phosphorus model: I. Documentation. Soil Science Society of America Journal 48, 800-
805.
Daroub, S.H., Gerakis, A., Ritchie, J.T., Friesen, D.K., Ryan, J., 2003. Development of a soil-
plant phosphorus simulation model for calcareous and weathered tropical soils.
Agricultural Systems 76, 1157-1181.
Probert, M.E., 2004. A capability in APSIM to model phosphorus responses in crops. In: Delve,
R.J. Probert, M.E. (Eds), Modelling Nutrient Management in Tropical Cropping Systems.
ACIAR Proceedings No. 114. pp. 92-100.

Table 4-3. Soil parameters used for testing the phosphorus model at Kpeve and Wa. The values
correspond to the top layer of the soils (0-10 cm for Kpeve and 0-20 cm for Wa)
Parameter Unit Kpeve Wa
Rate constant for transformation from labile P to active P d-1 0.03674 0.03674
Rate constant for transformation from active P to labile P d-1 0.00490 0.00490
Rate constant for transformation from active P to stable P d-1 0.00043 0.00043
Rate constant for transformation from stable P to active P d-1 0.00010 0.00010
P availability index unitless 0.40 0.40
Fraction of root labile inorganic P that is soluble unitless 0.20 0.20









Table 4-4. Values of additional inputs required to run the soil-plant phosphorus model for the
Kpeve and Wa experiments. Values correspond to the top layer of the soils (0-10 cm
for Kpeve and 0-20 cm for Wa)
Input Unit Kpeve Wa
Initial labile inorganic P ppm 16.52 6.49
Initial active inorganic P ppm 123.91 48.70
Initial stable inorganic P ppm 495.66 194.82
Initial active organic P ppm 7.99 2.22
Initial stable organic P ppm 125.15 34.78
P in residue (if applied) % 1.1 Not measured
P fertilizer (if applied) kg ha-1 0, 10, 30, and 80 0, 60, and 90
Soil CEC cmol kg-1 17.8 10.0
Soil Clay % 18.3 7.5
Source: estimated from soil composition data from the experiments.

Table 4-5. Estimated initial condition soil parameters for Kpeve
SLB SLLL SDUL SSAT SRGF SBDM C:P SLTX
10 0.180 0.260 0.460 1.000 0.83 138 Sandy Loam
20 0.070 0.140 0.280 1.000 1.08 136 Loam
30 0.040 0.080 0.160 0.607 1.47 130 Sandy Loam
40 0.060 0.120 0.240 0.497 0.74 138 Clay
50 0.040 0.060 0.120 0.407 0.47 123 Sandy Clay
60 0.050 0.090 0.180 0.333 0.56 218 Sandy Clay
70 0.080 0.150 0.300 0.273 0.97 200 Sandy Clay
80 0.060 0.110 0.220 0.223 0.77 127 Sandy Clay Loam
90 0.090 0.160 0.320 0.183 1.04 124 Sandy Clay
SLB, depth, base of soil layer (cm); SLLL, soil lower limit (cm3 cm-3); SDUL, soil upper limit,
drained (cm3 cm-3); SSAT, soil upper limit, saturated (cm3 cm-3); SRGF, soil root growth
factor (unitless); SBDM, soil bulk density, moist (g cm3), corrected for gravel content; C:P, ratio
of organic carbon to organic phosphorus (unitless); SLTX, soil texture (unitless).

Table 4-6. Estimated initial condition soil parameters for Wa
SLB SLLL SDUL SSAT SRGF SBDM SSKS SLTX
20 0.085 0.155 0.383 1.000 1.54 2.59 Loamy Sand
40 0.122 0.190 0.362 0.549 1.57 2.59 Sandy Loam
60 0.124 0.170 0.204 0.368 1.52 0.12 Sandy Clay
90 0.059 0.079 0.088 0.223 1.38 0.06 Clay
SLB, depth, base of soil layer (cm); SLLL, soil lower limit (cm3 cm-3); SDUL, soil upper limit,
drained (cm3 cm-3); SSAT, soil upper limit, saturated (cm3 cm-3); SRGF, soil root growth factor
(unitless); SBDM, soil bulk density, moist (g cm3); SSKS, saturation hydraulic conductivity (cm
h-1); SLTX, soil texture (unitless).









Table 4-7. Summary of aboveground biomass error statistics for the Kpeve and Wa experiments
Kpeve


Days after planting
RMSE (kg ha-1)
RRMSE (%)
Correlation Coefficient
Wa
Days after planting
RMSE (kg ha-1)
RRMSE (%)
Correlation Coefficient


17
87
83
-0.63

28
216
57
0.74


31
475
51
0.88

46
481
24
0.97


52
470
9
-0.88

61
2574
45
0.97


108 (harvest)
470
5
0.53


81
1048
19
0.99


125 (harvest)
1479
30
0.99











7 4000
c0
I 3000

S2000

S1000

* 0
0)


P level (kg ha-1)

0 measured ---simulated


Figure 4-1. Comparison of simulated and measured grain for different phosphorus levels in the
Kpeve experiment


100

80

60

40

20 -


1 .0,066


11 40
M


- 94


Bias Squared


SDSD


Figure 4-2. Decomposition of the grain yield MSE for the Kpeve experiment, using the method
developed by Kobayashi and Salam (2000)


III


1 ~t~--


























17 31 52
days after planting

measured -*--simulated


17 31 52
days after planting


9000

S6000
E
0
' 3000

n


1201

S901
go


60
E
30


m-


17 31 52 108
days after planting

S measured -*-simulated




00

00- 0

30 -

30
JOt


17 31 52 108
days after planting


C measured -*-simulated


a measured -*-simulated D


Figure 4-3. Comparison of simulated and measured biomass on four samples taken during the

season for the four treatments tested in Kpeve. A) Treatment OP. B) Treatment 10P.

C) Treatment 30P. D) Treatment 80P.


100 --


17 31 52 108

days after planting


m Squared Bias m SDSD O LCS




Figure 4-4. Decomposition of the in-season biomass MSE for the Kpeve experiment, using the

method developed by Kobayashi and Salam (2000)


S 9000

g 6000
E
.
S3000
o;


12000


9000

S6000
E
3000
0o









5000


4000

" 3000

S2000

1000


1000 2000 3000 4000


5000


measured grain yield (kg ha-1)


Figure 4-5. Comparison of measured and simulated maturity grain yield obtained in the Wa
experiment using the 1:1 line


a 5000
4000
i 3000

2000

> 1000


E OP 60P 90P OP 60P 90P OP 60P 90P

ON 60N 120N
treatments (kg ha-1)

measured ---simulated


Figure 4-6. Measured and simulated responses of maturity grain yield to different combinations
of nitrogen and phosphorus levels in the Wa experiment
























Bias Squared
Bias Squared


SDSD


Figure 4-7. Decomposition of the grain yield MSE for the Wa experiment, using the method
developed by Kobayashi and Salam (2000)


28 46 61 81 125


days after planting

M Squared Bias U SDSD O LCS


Figure 4-8. Components of the biomass MSE for the Wa experiment at five sampling times


2G ?2
















12000 ......

8000 E
4 4000
0 S
28 46 61 81 125 28 46 61 81 125
days after planting days after planting

measured-a-simulated measured-a-simulated




8. .... .





28 46 61 81 125 28 46 61 81 125
days after planting days after planting

measured --simulated measured --simulated


E.



28 46 61 81 125
days after planting

measured --simulated







E-5


28 46 61 81 125
days after planting

measured --simulated


E -



28 46 61 81 125 28 46 61 81 125 28 46 61 81 125
days after planting days after planting days after planting


* measured --simulated


* measured --simulated


H measured --simulated


Figure 4-9. Measured and simulated responses of cumulative biomass to different combinations

of nitrogen and phosphorus levels in the Wa experiment. A) Treatment ON OP. B)

Treatment ON 60P. C) Treatment ON 90P. D) Treatment 60N OP. E) Treatment 60N

60P. F) Treatment 60N 90P. G) Treatment 120N OP. H) Treatment 120N 60P. I)

Treatment 120N 90P.











































136



















I .... -


28 46 61 81 125
days after planting

measured --simulated


..


28 46 61 81 125
days after planting

S measured --- simulated


28 46 61 81 125
days after planting

a rmasured --sirmulated


* I
I .... I ....2 0 ..




28 46 61 81 125 28 46 61 81 125 28 46 61 81 125
days after planting days after planting days after planting

S measured---simulated a measured --simulated E measured ---simulated F


I ... .

L .....


28 46 61 81 125 28 46 61 81 125
days after planting days after planting

measured --- simulated measured ---simulated H


I "

4-

28 46 61 81 125
days after planting

measured --simulated


Figure 4-10. Measured and simulated responses of shoot P concentration to different

combinations of nitrogen and phosphorus levels in the Wa experiment. A) Treatment

ON OP. B) Treatment ON 60P. C) Treatment ON 90P. D) Treatment 60N OP. E)

Treatment 60N 60P. F) Treatment 60N 90P. G) Treatment 120N OP. H) Treatment

120N 60P. I) Treatment 120N 90P.


28 46 61
days after planting


-o-0 P -e-60 P 90 P


81 125


28 46 61 81 125
days after planting

P O -e-60 P 90 P


070
0 60
S050
0 0
5 040
,I 030
E I 020
u 010


F.\


Figure 4-11. Variation of the shoot P concentration during plant growth as affected by three

phosphorus levels in the Wa experiment. A) Measured. B) Simulated.


















137


070
060
o 050
o 0
o 040
T 1 030
S020
010
E 0 10
000









CHAPTER 5
SUMMARY AND CONCLUSIONS

The soil-plant phosphorus model in the DSSAT CSM integrates information on

phosphorus in soils and plants to simulate phosphorus transformations in soils and their effects

on plant production. Information on soil phosphorus includes the quantity of inorganic, readily-

available phosphorus (labile P), slowly available phosphorus (active P), very slowly available

phosphorus (stable P) and organic phosphorus. Transformation constants control the way

phosphorus is moved among these pools. The model differentiates between soils with different P

sorption capacities to partition fertilizer applied to the inorganic P pools. A fraction of the

phosphorus in the readily-available pool becomes soluble and may be taken up on any day that a

plant is growing on the soil. Information on the plant includes optimum and minimum

phosphorus concentration in different plant parts (roots, shoots, shells and seeds). The plant's

demand for phosphorus is estimated as the P deficit relative to a seasonally-varying optimum P

concentration. This demand is satisfied by P uptake from the readily available inorganic P pool.

If this uptake is not sufficient to meet the demand of seeds present, phosphorus can be removed

from other vegetative organs. Phosphorus not removed with harvest constitutes a capital

investment in the soil in the organic form.

A sensitivity analysis of the model, limited to six key factors, showed that P fertilizer

application and the initial value of the readily available P were the most important P-related

inputs affecting the predictability of plant biomass, yield and P uptake. The fraction of readily-

available P that is soluble, the shoot and seed P were also influential but to a smaller extent.

However, these parameters have more influence on the model outputs in the absence of P

fertilizer. Accurate predictions require therefore that at least initial readily available P be

measured or estimated correctly. In this regard, different names of readily available P have been









used in the literature and can be the source of model input error. In the DSSAT soil-plant

phosphorus model, the readily available P that provides soluble P for plant uptake is

approximately the inorganic labile P extracted with resin. If the resin P measurement is not

available but any of the following extractants were used to measure available P, Bray 1, Colwell,

Mehlichl, Morgan, Olsen, Truog, and water, the model will use empirical relationships from an

expert system to indirectly estimate the readily available inorganic P. The correct specification of

the quantity of fertilizer applied can become another source of error. Although in many

agronomic experiments, P fertilizer application is expressed as phosphate (P205), the amount of

phosphorus applied is expressed as pure P in the model. There is a 2.29 factor for converting

between the two.

The contrasting results obtained from the two experiments used to evaluate the phosphorus

model provided an ideal situation for testing the robustness of the model under opposite

conditions. The available phosphorus (Brayl) was relatively low at Kpeve (southern Ghana) but

other important phosphorus sources such as chemical contributions of organic matter (organic

matter content in the soil top 20 cm at Kpeve was 1.8%) not accounted for by the Bray

extraction could have been responsible for high indigenous phosphorus supply in the soil. No

significant difference in measured plant phenology, aboveground biomass, green leaf area and

grain yield was found between fertilized and unfertilized treatments at this site.

The soil at Wa (Northern Ghana) was relatively low both in available P (2.5 ppm Bray-P in

the top 20 cm) and organic carbon (0.49% in the top 20 cm). Maize responded well to

phosphorus fertilizer application on this soil. Leaf area index and aboveground biomass were low

in no nitrogen and no phosphorus treatments throughout the season. The highest reduction in leaf

area index and biomass occurred at the same time, which supports the reported finding that poor









biomass accumulation in P deficient conditions is associated with reduced photosynthetically

absorbed radiation by the plant, due to reduced leaf area. The reduction in grain yield could have

been a result of indirect and direct effects of N and P stress on photosynthesis.

Testing of the phosphorus model under both P-limitation (Wa) and no P limitation

(Kpeve) conditions showed that plant biomass and grain yield were quite predictable. Grain yield

was simulated with an RRMSE of 8% at Kpeve and 14% at Wa. Final biomass was simulated

with an RRMSE of 5% at Kpeve and 30% at Wa. Although the simulation skill was lower at Wa,

the model reasonably captured the response of biomass and grain yield to P fertilizer at both

sites.

The soil-plant phosphorus model described, analyzed and tested with field data performed

acceptably well over specific and known soil phosphorus conditions. The potential exists for

using the model as an application tool or in decision-support because model simulation of crop

response to P fertilizer is promising. However, the current level of confidence in the model must

be enhanced through further testing and validation studies. Some P model parameters are highly

uncertain and must be estimated from other, more easily measurable variables. For example, the

initial inorganic labile P that has a major influence on crop response need greater precision in its

estimation. This confidence raising process includes: 1) verification or re-verification of the

model; 2) more accurate estimation of the inorganic labile P from measured available P when

new data become available for calibration of the expert system; 3) special study on the

estimation of the fraction of inorganic labile P that is soluble for a specific soil and how this

fraction changes with soil properties like the P-sorbing capacity.









APPENDIX A
MEASURED GROWTH DATA AT KPEVE

Table A-1. Monthly total rainfall in 2006 (one standard deviation of rainfall), mean daily solar
radiation, and mean daily temperature collected during the Kpeve experiment in 2006
Rain Solar Radiation Maximum Minimum
Month (mm) (MJ m-2 day-1) Temperature (C) Temperature (C)
March 107.6(7.4) 14.8 35.2 23.4
April 84.0 (8.8) 14.0 35.0 24.3
May 257.4 (12.3) 14.4 32.7 22.8
June 202.4 (18.3) 14.5 31.5 22.7
July 40.4 (3.6) 11.6 30.2 22.9
August 21.0(1.8) 10.5 30.0 22.7

Table A-2. Days to tasseling (one standard deviation of four replications), days to anthesis (one
standard deviation of four replications), and days to silking (one standard deviation of
four replications) for the experiment in Kpeve, Ghana
P level (kg ha-1) Tasseling (day) Anthesis (day) Silking (day)
0 48(1.0) 51(0.6) 59(3.9)
10 49(0.8) 51(1.0) 57 (3.3)
30 48(0.6) 50(0.5) 57(5.1)
80 48(2.1) 51(0.6) 63 (1.9)

Table A-3. Measured mean aboveground biomass (one standard deviation of four replications)
for four phosphorus treatments, sampled four times during the growing season in the
Kpeve experiment
P level (kg ha-1) 17 dap 31 dap 52 dap 108 dap
0 110(22) 913 (145) 5681 (739) 9802(1572)
10 102(18) 897(91) 5436(805) 10002(949)
30 105 (13) 914 (106) 5047 (883) 9816 (706)
80 101 (9) 1013 (81) 5083 (680) 8926 (914)
dap = days after planting. Data are reported in kg ha-1.









Table A-4. Mean green leaf area (one standard deviation of four replications) for four
phosphorus treatments, measured seven times during the growing season in the Kpeve
experiment
P level (kg ha-1) 17 dap 24 dap 31 dap 38 dap 45 dap 52 dap 68 dap
327 1010 2409 4495 5269 5554 4258
0 (47) (44) (176) (309) (481) (187) (470)
332 1078 2634 4842 5699 6235 4543
10 (77) (318) (656) (881) (1034) (1067) (1309)
373 1175 2807 5102 5612 5946 4330
30 (47) (189) (235) (454) (411) (454) (464)
320 1004 2576 4605 5345 5473 4090
80 (25) (78) (235) (315) (202) (278) (217)
dap = days after planting. Data are reported in cm2 plant1.

Table A-5. Mean maize height (one standard deviation of four replications) for four phosphorus
treatments, measured seven times during the growing season in the Kpeve experiment
P level (kg ha-) 17 dap 24 dap 31 dap 38 dap 45 dap 52 dap 68 dap
26 42 76 124 195 236 240
0 (5) (10) (10) (22) (32) (36) (42)
29 45 76 129 208 250 259
10 (5) (8) (14) (22) (22) (27) (33)
26 47 84 138 215 250 256
30 (5) (6) (8) (15) (19) (29) (31)
25 44 81 126 188 234 240
80 (7) (9) (12) (21) (37) (35) (32)
dap = days after planting. Data are reported in cm plant-.

Table A-6. Mean soil moisture (one standard deviation of four replications) in four phosphorus
treatments plots, measured using TDR eight times during the growing season in the
Kpeve experiment
P level (kg ha-) 0 dap 16 dap 24 dap 30 dap 37 dap 45 dap 53 dap 69 dap
12.1 8.8 17.1 13.7 11.5 9.4 6.4 13.7
0 (3.1) (2.2) (4.8) (3.1) (2.6) (2.6) (2.1) (7.3)
11.8 8.3 15.3 12.1 10.2 8.0 6.0 9.4
10 (2.1) (1.4) (2.9) (1.4) (1.5) (1.9) (1.9) (1.6)
12.7 8.4 17.3 13.3 10.8 8.3 5.6 10.6
30 (2.5) (1.1) (4.0) (3.1) (1.6) (1.6) (1.2) (3.3)
13.2 8.6 17.9 13.7 11.1 9.0 5.7 12.0
80 (3.2) (1.7) (5.4) (3.6) (1.7) (1.7) (1.4) (4.5)
dap = days after planting. Data are reported in %.








Table A-7. Measured mean grain yield (one standard deviation of four replications), unit grain
weight (one standard deviation of four replications), and grain number (one standard
deviation of four replications) for four phosphorus levels in the Kpeve experiment
P level Grain yield Unit grain weight Grain number
(kg ha-1) (kg ha-1) (g grain1) (# m-2)
0 3286 (683) 0.23 (0.04) 1655 (193)
10 2859 (384) 0.25 (0.03) 1344 (405)
30 3025 (358) 0.24 (0.02) 1492 (471)
80 2918 (411) 0.23 (0.04) 1467(334)











APPENDIX B
MAPS OF THE EXPERIMENT SITES LOCATIONS


L -
C- ,NW "


.oiao~.rxo
y .


------I~ Lla;'lj
Tntis%~


6 1 --~ -A
i-~~~~qeif I SpT ~yvaj
LsIi.- ray \.
- r-Qt~lai*W ,


.. ` rj 1"I
.- Al gena E gy
-i I Ly Eg5y 3L
.'. LT arnara I ',

Maurnani a I '
I 'r I -] r ,.
a pep4 rle i M li 'Jg
CapeAtrde J-- -' C had ntrel
"u ui" 3 1 .." -" -' -'
G ,, AL -. F n J" n ..,
'-t irie Lf- ,-' ,'

!L._ r1 merlor _. "

r E~ua:' 3lg:fa l gia IanJ
E ua. p._a .,

j ,r. I UZair e i .

11ai 1 a-- Tanara Uniled Ri epliI,.: .:.re


- t H ;I rl


I rar 1I31Jnr


.__ --

remen .-


so "-iala C


.a3 I a .asar

I|J 3m 3 --i

I ,a "ll r
'.Soui. Ahi-i.i~eaIy


Figure B-1. Map of the African continent showing Ghana, the country where the field
experiments were carried out











































Figure B-2. Map of Ghana showing the location of the two study sites, Kpeve in the South and
Wa in the North









APPENDIX C
INITIALIZATION OF SOIL INORGANIC AND ORGANIC PHOSPHORUS POOLS IN THE
SOIL-PLANT PHOSPHORUS MODEL

Initial values of the three soil inorganic P pools (labile, active and stable) and two soil

organic P pools (active and stable) described in Chapter 3 are needed to simulate phosphorus in

soils and plants. These values would ideally come from P fractionation studies on the soil of

interest following the procedure developed by Hedley et al. (1982), Tiessen et al. (1984), and

Tiessen and Moir (1993). The Hedley/Tiessen fractionation procedure is a thorough phosphorus

extraction method that treats a soil sample with increasingly aggressive chemicals. The soil is

first shaken with water plus resin (to extract the most labile part of P), then treated with

NaHCO3, NaOH, (and sometimes NaOH with sonication), diluted HC1 and hot concentrated

HC1. Each chemical extracts a more resistant form of phosphorus that escaped the previous

extractant. The residual phosphorus still remaining in the soil is measured after digestion of the

soil sample with perchloric or sulfuric acid. This procedure sequentially extracts both inorganic

and organic forms of P. The data obtained from the P fractionation are used to determine directly

the sizes of the inorganic and organic pools.

However, few researchers make use of the P fractionation method probably because it is

expensive and also because simple and inexpensive extraction methods, such as resin and Bray 1

methods, would generally answer phosphorus availability questions that arise in most agronomic

experiments.

If an alternate method is not provided for indirect estimation of the pools sizes, potential

model users could possibly resort to indicative values found in the literature or eventually

conclude that the model is not of any practical use because required input data are not readily

available. Since organic carbon, pH and available phosphorus are routinely measured in most









traditional agronomic experiments, developing relationships that can make use of those data and

provide reasonable estimates of inorganic labile P and organic P was thought to be helpful.

The aim of this appendix was to present direct and indirect methods of estimation of initial

inorganic labile, active and stable P, and initial organic active and stable P for use by the soil-

plant phosphorus model in DSSAT. The relationships discussed in this appendix are based on

studies by Singh (1985), Sharpley (1984, 1989).

Initialization of Inorganic Phosphorus Pools

From P Fractionation Data

The quantities of inorganic labile, active, and stable P (Table C-l) initially present in the

soil can then be derived, in mg kg1, from the fractionation data for each soil layer as (Jones et

al., 2005a):

Initial PiLabile = Pi Re sin+ PiNaHC03 (C-1)
Initial PiActive = 0.5 x PiNaOH (C-2)
Initial PiStable =
(0.5 x PiNaOH) + PiNaOHSonic + PiHC1 + PiHClHot + (0.5 x P Re sidual) (C-3)
Where PiResin is inorganic P extracted with water and resin.
PiNaHC03 is inorganic P extracted with bicarbonate of sodium.
PiNaOH is inorganic P extracted with sodium hydroxide.
PiNaOHSonic is inorganic P extracted with sodium hydroxide plus sonication.
PiHCI is inorganic P extracted with diluted HC1.
PiHClHot is inorganic P extracted with hot concentrated HC1.
PResidual is residual P measured after digestion of the remaining sample with perchloric
or sulfuric acid.

From Measured Available P Using the Anion Exchange Resin Method

P extraction using this method can be approximated as a direct measurement of PiLabile in

the soil. The anion exchange resin technique extracts phosphorus from the soil in the same

manner as plants and has been reported as a reliable method for measuring plant available

phosphorus (Myers, 2005; Abdu, 2006). PiActive and PiStable are assumed to be in equilibrium

initially and are calculated based on the value of PiLabile as follow (Jones et al., 1984a):









KL
PiActive = PiLabile x (C-4)
KAL
Where KL = rate constant for transformation from labile P to active P
KAL = rate constant for transformation from active P to labile P.


K = 0.03[(1- PAvailndex)] (C-5)
PAvailhndex

K
K AL= x PAvaillndex (C-6)
3
Where PAvaillndex is used as a measure of the activity level of P in the soil. The
calculation of the PAvaillndex depends on soil category (Sharpley et al., 1984, Table C-2) and is
provided in Table 3-1.

According to Jones et al. (1984a), PiStable is four times as large as PiActive:

PiStable = 4 x PiActive (C-7).

From Other Methods

If fractionation data are not available and resin measurements were not made, the initial

PiLabile can be estimated from other P extraction methods based on regression equations

between resin P and extractable P (such as Brayl and Olsen P) that were used to build an expert

system.

The equations that appear in the expert system are based on studies conducted by Sharpley

et al. (1984, 1989) and Singh (1985). The expert system in its current version has not been tested

independently for its ability to estimate accurately labile P from different available P extraction

methods. It is used in the soil modules of the P model as an experimental version to estimate the

initial inorganic and organic phosphorus pool sizes based on the method used for measuring

available P and the soil category concerned. The criteria used for assigning soil categories are

presented in Table C-2.









The inorganic labile P is computed first and the active Pi is derived from the labile Pi in

such a way that the two pools remain in equilibrium (Equation C-4). The stable Pi is calculated

using the size of PiActive (Equation C-7).

The following P extraction methods are used in the expert system proposed by Singh

(1985) to compute initial PiLabile in the soil: water, Bray 1, Olsen, Mehlich 1, Truog, Morgan's

solution and Colwell (Table C-3). In slightly weathered soils, measured exchangeable potassium

is used in combination with Bray 1, Olsen, Mehlich 1 and Truog for a more accurate estimation

of PiLabile (Table C-3). The soil-plant phosphorus model cannot be initialized if none of these

measured P data is available.

Initialization of Soil Organic Phosphorus Pools

The division of the organic residues added to the surface of the soil into metabolic and

structural components (Figure 3-1) is governed by the lignin to N ratio (lignin:N) of the residues.

The metabolic fraction is estimated as equal to 0.85 0.013*(lignin:N ratio) (Gijsman et al.,

2002a).

The procedures for estimating the initial sizes of the active and stable soil SOM (SOM1

and SOM23) from P fractionation data or from the expert system (using organic C and pH) are

described next.

Initialization from P Fractionation Data

The initial values of the SOM1 and SOM23 pools can be obtained from Hedley/Tiessen

soil P fractionation data (Table C-4) (Gijsman and Porter, unpublished):

Initial PoActive (SOM 1) = PoNaHC03 + PoNaOH (C-8)

Initial PoStable (SOM23) = PoNaOHSonic + PoHCl + PoHClHot + (0.5 x PResidual)
(C-9)
Where PoNaHC03 is organic P extracted with bicarbonate of sodium.
PoNaOH is organic P extracted with sodium hydroxide.









PoNaOHsonic is organic P extracted with sodium hydroxide plus sonication.
PoHCl is organic P extracted with diluted HC1.
PoHClhot is organic P extracted with hot concentrated HC1.
PResidual is P recovered after digestion with perchloric or sulfuric acid.

Initialization from Measured Organic P

If only a pooled total organic P value is known, the partitioning between active and stable

P depends on the land and crop use history of the soil (previous crop in DSSAT). The initial

active organic P is set to 3% and the initial stable P to 97% of the total organic P if the previous

crop is bahia grass or grass weeds. For all other previous crops, the initial active and the initial

stable organic P represent respectively 6% and 94% of the measured total organic P.

Initialization from Organic C and soil pH

Indirect estimation of the active and stable organic P (if total organic P was not

measured) through the expert system uses measured soil organic C and pH (Table C-5). These

soil properties are known to be correlated with soil organic P; equations relating them to total

organic P in the soil have been developed based on studies by Sharpley et al. (1984, 1989) and

Singh (1985). The distribution between active and stable P is exactly the same as if the total

organic P was directly measured.









Table C-1. Relationship between inorganic P pools and P extracted using the Hedley procedure
Inorganic P fractionation methods
P pools Resin NaHCO3 NaOH NaOH and Sonication HC1 Hot HCl Residual


Labile
Active


+ +


+1/2


Stable +1/2 +
Source: Gijsman and Porter (2005)


+ +


+1/2


Table C-2. Specification of soil categories
Soil Category Criteria
Andl Soil description or taxonomy includes the terms "ANDOSOL"
Andisol
or "ANDISOL" or "VOLCAN" or "ANDEPT"
Calcareous CaCO3 content > 15%
CEC
Slightly Weathered Ratio C > 16
(CLAY/100)
CEC
Highly Weathered Ratio CA <16
(CLAY/100)
--Soil description or taxonomy does not include the terms
Other Soils "ANDOSOL" or "ANDISOL" or "VOLCAN" or "ANDEPT";
--CaCO3 and CEC are not measured.
Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in the
tropics. PhD thesis, University of Hawaii, Honolulu.

Table C-3. Equations for calculating initial inorganic P labile from different extraction methods
for different soil categories
Soil Category P or K data available PiLabile (mg kg-1)
Calcareous Olsen (1.17 x POlsen) + 0.18


Bray 1
Mehlich 1 (double acid 1:5)
Water

Olsen
Olsen and Exchangeable K
Bray 1
Bray 1 and Exchangeable K
Mehlich 1 (double acid 1:5)
Mehlich 1 & Exchangeable K
Truog
Truog and Exchangeable K
Morgan's Solution


(1.81 x PBrayl)+ 1.88
(0.10 x PMehlichl)+ 10.20
(5.92 x PWater)+ 0.09


(0.76 x POlsen)+ 6.53
(0.62 x POlsen)+ (10.09 x ExchK)+ 2.62
(1.37 x PBrayl)+ 6.77
(1.09 x PBrayl)+(10.59 x ExchK)+ 2.71
(2.71 x PMehlichl)+ 5.82
(2.16 x PMehlichl 1)+ (9.58 x ExchK)+ 2.42
(0.34 x PTruog)+ 3.35
(0.30 x PTruog)+ (5.85 x ExchK)+1.48
187.30 x PMorgan x 11.87


Slightly
Weathered









Table C-3 continued
Highly Olsen
Weathered Bray 1
Mehlich 1 (double acid 1:5)
Truog
Mehlich 1 (double acid 1:10)
Colwell


Andisol


Olsen
Bray 1
Mehlich 1 (double acid 1:5)
Truog


(2.50 xPOlsen)- 2.19
(2.88 xPBrayl)- 0.30
(5.97 x PMehlichl)- 0.21
(1.07 xPTruog)-1.49
(0.64 x PMehlichl) +5.72
(0.43 x PColwell)+ 4.21

(1.41x POlsen)- 2.56
(2.88xPBrayl)- 2.11
(4.52 xPMehlichl) +6.67
(0.27 x PTruog) 0.73


Unknown Olsen (0.74 x POlsen)- 11.39
Or Bray 1 (1.35 xPBrayl)-10.24
Not specified Mehlich 1 (double acid 1:5) (2.65 x PMehlichl)+ 9.39
Truog (0.28 x PTruog)- 6.15
Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in the
tropics. PhD thesis, University of Hawaii, Honolulu.

Table C-4. Relationship between organic P pools and P extracted using the Hedley procedure
Inorganic P fractionation method
P pools NaHCO3 NaOH NaOH and Sonication HC1 Hot HCl Residual
Active + +
Stable + + + +1/2
Source: Gijsman and Porter (2005)

Table C-5. Equations for calculating initial total organic P from soil organic carbon (OrgC) and
pH for different soil categories
Soil Category Organic P (mg kg-1)
Calcareous pH -3 2
Calcareous -1 H 6 -0 55xOrgC


Highly Weathered

Slightly Weathered


200 x e
200xe


S-e
S85 pH-3 2
1 6 85x _e 035xOrgC


-15xe PH10 2
900xe 1 x( 12 e -0 10x)rgC


pH-7 2
Other Soils 520 xe 15 x x1 e-0135xOrgC
Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in the
tropics. PhD thesis, University of Hawaii, Honolulu.









LIST OF REFERENCES


Abdu, N., 2006. Soil-phosphorus extraction methodologies: a review. African Journal of
Agricultural Research 1 (5), 159-161.

Adiku, S.G.K., 2004. Exploring options for improving crop productivity within some farming
zones of Ghana using coupled climatic-crop models, Final Report for the period November
2002 to August 2004. Department of Soil Science, University of Ghana, Legon, Ghana.

Adiku, S.G.K., 2006. Measuring and Assessing Soil Carbon Sequestration by Agricultural
Systems in Developing Countries, Final Report for the period March to July 2006.
Department of Soil Science, University of Ghana, Legon, Ghana.

Annino, J.S., Russell, E.C., 1981. The most frequent causes of simulation analysis failure and
how to avoid them. Interfaces 11 (3), 59-63.

Anonymous, 1996. Obatanpa, a new maize variety with high quality protein for farmers. Crops
Research Institute Factsheets, Animal Science Department, Ministry of Food and
Agriculture, Accra, Ghana.

Bolanos, J., Edmeades, G.O., 1993. Eight cycles of selection for drought tolerance in lowland
tropical maize. II. Response in reproductive behaviour. Field Crops Research 31, 253-268.

Brady, N.C., Weil, R.R., 2002. The Nature and Properties of Soils. Pearson Education, NJ.

Colomb, B., Kiniry, J.R. Debaeke, P., 2000. Effect of soil phosphorus on leaf development and
senescence dynamics of field-grown maize. Agronomy Journal 92, 428-435.

Daroub, S.H., Gerakis, A., Ritchie, J.T., Friesen, D.K., Ryan, J., 2003. Development of a soil-
plant phosphorus simulation model for calcareous and weathered tropical soils.
Agricultural Systems 76, 1157-1181.

Del Campillo, M.C., Van Der Zee, S.E.A.T.M., Torrent, J., 1999. Modelling long-term
phosphorus leaching and changes in phosphorus fertility in excessively fertilized acid
sandy soils. European Journal of Soil Science 50, 391-399.

Du Toit, A.S, Booysen, J., Human, J.J., 1997. Use of linear regression and correlation matrix in
the evaluation of CERES3 (Maize). South African Journal of Plant and Soil 14, 177-182.

Dzotsi, K.A., Agboh-Noameshie, A., Struif Bontkes, T.E., Singh, U., Dejean, P., 2003. Using
DSSAT to derive optimum combinations of cultivar and sowing date for maize in Southern
Togo. In: Struif-Bontkes, T.E., Wopereis, M.C.S. (Eds), Decision-Support Tools for
Smallholder Agriculture in Sub-Saharan Africa: A Practical Guide. IFDC and CTA, pp.
100-113.

Eghball, B., Shanahan, J.F., Varvel, G.E., Gilley, J.E., 2003. Reduction of high soil test
phosphorus by corn and soybean varieties. Agronomy Journal 95, 1233-1239.









FAO, 2003. Fertilizers and their use. Food and Agriculture Organization of the United Nations,
Rome, Italy.

FAO, 2005. Fertilizer Use by Crop in Ghana. Land and Plant Nutrition Management Service,
Land and Water Development Division, Food and Agriculture Organization of the United
Nations, Rome, Italy.

Fofana, B., Tamelokpo, A., Wopereis, M.C.S., Breman, H., Dzotsi, K.A., Carsky, R.J., 2005.
Nitrogen use efficiency by maize as affected by a mucuna short fallow and P application in
the coastal savanna of West Africa. Nutrient Cycling in Agroecosystems, 71 (3), 227-237.

Gauch, G.H., Jr., Hwang, J.T.G., Fick, G.W., 2003. Model evaluation by comparison of model-
based predictions and measured values. Agronomy Journal 95, 1442-1446.

Gijsman, A.J., Hoogenboom, G., Parton, W.J., Kerridge, P.C., 2002. Modifying DSSAT crop
models for low-input agricultural systems using a soil organic matter-residue module from
CENTURY. Agronomy Journal 94, 462-474.

Gijsman, A.J, Jagtap, S.S., Jones, J.W., 2002b. Wading through a swamp of complete confusion:
how to choose a method for estimating soil water retention parameters for crop models.
European Journal of Agronomy 18(1-2), 75-105.

Gijsman, A.J., Porter, C.H., 2005. The Soil Organic Phosphorus Module of the DSSAT Cropping
Systems Model. University of Florida, Gainesville.

Havlin, J.L., Beaton, J.D., Tisdale, S.L., Nelson, W.L., 1999. Soil fertility and fertilizers: an
introduction to nutrient management. Prentice Hall, Upper Saddle River, NJ.

Hedley M.J., Stewart, J.W.B., Chauhan, B.S., 1982. Changes in inorganic and organic soil
phosphorus fractions induced by cultivation practices and by laboratory incubations. Soil
Sci. Soc. Am. J. 46, 970-976.

Hunt, L.A., Boote, K.J., 1998. Data for model operation, calibration, and evaluation. In: Tsuji,
G.Y. et al. (Eds), Understanding Options for Agricultural Production. Kluwer Academic
Publisher, Boston, pp. 9-39.

Jagtap, S.S., Abamu, F.J., 2003. Matching improved maize production technologies to the
resource base of farmers in a moist savanna. Agricultural Systems 76, 1067-1084.

Jagtap, S.S., Abamu, F.J., Kling, J.G., 1999. Long-term assessment of nitrogen and variety
technologies on attainable maize yields in Nigeria using CERES-Maize. Agricultural
Systems 60, 77-86.

Janssen, B.H., 1993. Integrated nutrient management: the use of organic and mineral fertilizers.
In: Van Reuler, H. Prins, W.H. (Eds.). The Role of Plant Nutrients for Sustainable Food
Crop Production in Sub-Saharan Africa. Dutch Association of Fertilizer Producers (VKP),
Leidschendam, The Netherlands, pp. 89-105.









Janssen, B.H., Lathwell, D.J., Wolf, J., 1987. Modeling long-term crop response to fertilizer
phosphorus. II. Comparison with field results. Agronomy Journal 79, 452-458.

Johnston, A.E., 2000. Soil and Plant Phosphate. International Fertilizer Industry Association.
Paris, France.

Jones, C.A. 1983. A survey of the variability in tissue nitrogen and phosphorus concentrations in
maize and grain sorghum. Field Crops Research 6, 133-147.

Jones, C.A., Cole, C.V., Sharpley, A.N. Williams, J.R., 1984a. A simplified soil and plant
phosphorus model: I. Documentation. Soil Science Society of America Journal 48, 800-
805.

Jones, J.W., Hoogenboom, G., Porter, C.H., Boote, K.J., Batchelor, W.D., Hunt, L.A., Wilkens,
P.W., Singh, U., Gijsman, A.J., Ritchie, J.T., 2003. The DSSAT cropping system model.
European Journal of Agronomy 18, 235-265.

Jones, J.W., Luyten, J.C., 1998. Simulation of biological processes. In: Peart, R.M., Curry, R.B.
(Eds), Agricultural Systems Modeling and Simulation. Marcel Dekker, New York, pp. 19-
62.

Jones, J.W., Singh, U., Porter, C.H., 2005a. Soil Inorganic Phosphorus Module. Research Report.
University of Florida, Gainesville.

Jones, J.W., Singh, U., Porter, C.H., Bostick, W.M., 2005b. Generic Plant Phosphorus Module.
Research Report. University of Florida, Gainesville.

Jones, J.W., Tsuji, G.Y., Hoogenboom, G., Hunt, L.A., Thornton, P.K., Wilkens, P.W., Imamura,
D.T., Bowen, W.T., Singh, U., 1998. Decision support system for agrotechnology transfer:
DSSAT v3. In: Tsuji, G.Y. et al. (Eds), Understanding Options for Agricultural
Production. Kluwer Academic Publisher, Boston, pp. 157-177.

Kaboneka, S., Nivyiza, J.C., Sibomana, L., 2004. Effects of nitrogen and phosphorus fertilizer
addition on wheat straw carbon decomposition in a Burundi acidic soil. In: Bationo, A.
(Ed.), Managing Nutrient Cycles to Sustain Soil Fertility in Sub-Saharan Africa. Academy
Science Publisher, Nairobi, Kenya, pp. 151-161.

Karpinets, T.V., Greenwood, D.J., Ammons, J.T., 2004. Predictive mechanistic model of soil
phosphorus dynamics with readily available inputs. Soil Science Soc. Am. J. 68, 644-653.

Kelling, K.A., Bundy, L.G., Combs, S.M., Peters, J.B., 1998. Soil test recommendations for field
vegetable, and fruit crops. Univ. Wis. Coop. Extn. Ser. Bull. A2809.

Khan, M.A., Hussain, N., Abid, M., Masood, M.U., 2005. Effect of phosphorous levels on
growth and yield of maize (Zea mays L.) cultivars under saline conditions. International
Journal of Agriculture and Biology 7 (3), 511-514.









Kobayashi, K. Us Salam, M., 2000. Comparing simulated and measured values using mean
squared deviation and its components. Agronomy Journal 92, 345-352.

Koopmans, G.F., Chardon, W.J., Ehlert, P.A.I., Dolfing, J., Suurs, R.A.A., Oenema, O., van
Riemsdijk, W.H., 2004. Phosphorus availability for plant uptake in a phosphorus-enriched
noncalcareous sandy soil. Journal of Environment Quality 33, 965-975.

Lehman, J.T., Botkin, D.B., Likens, G.E., 1975a. The assumptions and rationales of a computer
model of phytoplankton population dynamics. Limnology and Oceanography 20, 343-364.

Lewis, D.R., McGechan, M.B., 2002. A review of field scale phosphorus dynamics models.
Biosystems Engineering 82 (4), 359-380.

Littell, R.C., Henry, P.R., Ammerman, C.B., 1998. Statistical analysis of repeated measures data
using SAS procedures. Journal of Animal Science 76, 1216-1231.

Makowski, D., Naud C., Monod, H., Jeuffroy, M-H., Barbottin, A., 2005. Global sensitivity
analysis for calculating the contribution of genetic parameters to the variance of crop
model prediction. In: Hanson, K.M., Hemez, F.M. (Eds), Sensitivity Analysis of Model
Output. Proceedings of the 4th International Conference on Sensitivity Analysis of Model
Output held in Santa Fe, NM, 8-11 March 2004. pp. 307-316.

Matthews, R., Stephens, W., Hess, T., Masson, T., Graves, 2000. Application of Crop/Soil
Simulation Models in Developing Countries, Final Report. Institute of Water and
Environment, Cranfield University, Silsoe, Bedfordshire, U.K.

Miao, Y., Mulla, D.J., Batchelor, W.D., Paz, J.O., Robert, P.C., Wiebers, M., 2006. Evaluating
management zone optimal nitrogen rates with a crop growth model. Agronomy Journal 98
(3), 545-553.

Mitchell, P.L. 1997. Misuse of regression for empirical validation of models. Agricultural
Systems 54, 313-326.

Monod, H., Naud, C., Makowski, D., 2006. Uncertainty and sensitivity analysis for crop models.
In: Wallach, D., Makowski, D., Jones, J.W. (Eds), Working with Dynamic Crop Models:
Evaluation, Analysis, Parameterization and Applications. Elsevier, Amsterdam, The
Netherlands. pp. 55-96.

Myers, R.G., Sharpley, A.N., Thien, S.J., Pierzynski, G.M., 2005. Ion-sink phosphorus extraction
methods applied on 24 soils from the continental USA. Soil Sci. Soc. Am. J. 69, 511-521.

Naab, J.B., 2005. Measuring and Assessing Soil Carbon Sequestration by Agricultural Systems
in Developing Countries, 2004 Annual Report. Savanna Agricultural Research Institute,
Wa, Ghana.









Newman, S., Lynch, T., Plummer, A.A., 2000. Aspects of success and failure of decision support
systems in agriculture. In: Abbass, H.A., Towsey, M. (Eds), Application of Artificial
Intelligence, Optimization and Bayesian Methods in Agriculture. Queensland University of
Technology Press, Brisbane, Australia, pp. 17-30.

Nziguheba, G., Merckx, R., Palm, C.A., 2002. Soil phosphorus dynamics and maize response to
different rates of phosphorus fertilizer applied to an Acrisol in western Kenya. Plant and
Soil 243, 1-10.

Oberkampf, W.L., Helton, J.C., Joslyn, C.A., Wojtkiewicz, S.F., Ferson, S., 2004. Challenge
problems: uncertainty in system response given uncertain parameters. Reliability
Engineering and System Safety 85, 11-19.

Odum H.T., Odum, E.C., 2000. Modeling for All Scales: An Introduction to System Simulation.
Academic Press, San Diego.

Pellerin, S., Mollier, A., Plenet, D., 2000. Phosphorus deficiency affects the rate of emergence
and number of maize adventitious nodal roots. Agronomy Journal 92, 690-697.

Osiname, O.A., Meppe, F., Everett, L., 2000. Response of maize (Zea mays) to phosphorus
application on basaltic soils in Northwestern Cameroon. Nutrient Cycling in
Agroecosystems 56, 209-217.

Parton, W.J., Ojima, D.S., Cole, C.V., Schimel, D.S., 1994. A general model for soil organic
matter dynamics: Sensitivity to litter chemistry, texture and management. In: Bryant, R.B.,
Arnold, R.W. (Eds), Quantitative modeling of soil forming processes. SSSA Spec. Publ.
39. SSSA, Madison, WI, pp 147-167.

Parton, W.J., Stewart, J.W.B., Cole, C.V., 1988. Dynamics of C, N, P and S in grassland soils: A
model. Biogeochemistry 5:109-131.

Pellerin, S., Mollier, A., Plenet, D., 2000. Phosphorus deficiency affects the rate of emergence
and number of maize adventitious nodal roots. Agronomy Journal 92, 690-697.

Pheav, S., Bell, R.W., White, P.F., Kirk, G.J.D., 2003. Fate of applied phosphorus in a highly
weathered sandy soil under lowland rice cropping, and its residual effect. Field Crops
Research 81, 1-16.

Plenet, D., Etchebest, S., Mollier, A., Pellerin, S., 2000a. Growth analysis of maize field crops
under phosphorus deficiency. I. Leaf growth. Plant Soil 223, 117-130.

Plenet, D., Mollier, A., Pellerin, S., 2000b. Growth analysis of maize field crops under
phosphorus deficiency. II. Radiation-use efficiency, biomass accumulation and yield
components. Plant Soil 224, 259-272.

Probert, M.E., 2004. A capability in APSIM to model phosphorus responses in crops. In: Delve,
R.J. Probert, M.E. (Eds), Modelling Nutrient Management in Tropical Cropping Systems.
ACIAR Proceedings No. 114. pp. 92-100.









Probert, M.E., Okalebo, J.R., 1992. Effect of phosphorus on the growth and development of
maize. In: Probert, M.E. (Ed), Semi-Arid Eastern Kenya. ACIAR, Australia, pp. 55-62.

Rahn, C., Mead, A., Draycott, A., Lillywhite, R., Salo, T., 2001. A sensitivity analysis of the
prediction of the nitrogen fertilizer requirement of cauliflower crops using the HRI WELL-
N computer model. Journal of Agricultural Science 137, 55-69.

Ratto, M., Tarantola, S., Saltelli, A., 2001. Sensitivity analysis in model calibration: GSA-GLUE
approach. Computer Physics Communications 136, 212-224.

Reddy, K.R., Wang, Y., DeBusk, W.F., Fisher, M.M., Newman, S., 1998. Forms of soil
phosphorus in selected hydrologic units of the Florida Everglades. Soil Sci. Soc. Am. J.
62, 1134-1147.

Saltelli, A., 2002. Sensitivity analysis for importance assessment. Risk Analysis 22 (3), 579-588.

Saltelli, A., 2004. Global sensitivity analysis: an introduction. In: Hanson, K.M., Hemez, F.M.
(Eds), Sensitivity Analysis of Model Output. Proceedings of the 4th International
Conference on Sensitivity Analysis of Model Output held in Santa Fe, NM, 8-11 March
2004. pp 27-43.

Saltelli, A., Tarantola, S., Campolongo, F., 2000. Sensitivity analysis as an ingredient of
modeling. Statistical Science 15 (4), 377-395.

Sartain, J.B. 1980. Mobility and extractability of phosphorus applied to the surface of tifway
bermudagrass turf. Soil and Crop Science Society of Florida Proc. 39, 47-50.

SAS, 2002. SAS version 9.1 for Windows. SAS Institute Inc., Cary, NC.

Schachtman, D.P., Reid, R.J., Ayling, S.M., 1998. Phosphorus uptake by plants: from soil to cell.
Plant Physiology 116, 447-453.

Schmidt, J.P., Buol, S.W., Kamprath, E.J., 1996. Soil phosphorus dynamics during 17 years of
continuous cultivation: a method to estimate long-term P availability. Geoderma 78, 59-70.

Shapiro, C.A., Fergusson, R.B., Hergert, G.W., Dobermann, A.R., Wortmann, C.S., 2003.
Fertilizer Suggestions for Corn. NebGuide G74-174-A. Cooperate Extension, Institute of
Agriculture and Natural Resources, University of Nebraska-Lincoln, Lincoln.

Sharpley, A.N., Jones, C.A., Gray, C. Cole, C.V., 1984. A simplified soil and plant phosphorus
model II. Prediction of labile, organic, and sorbed phosphorus. Soil Sci. Soc. Am. J. 48,
805-809.

Sharpley, A.N., Singh, U., Uehara, G., Kimble, J., 1989. Modeling soil and plant phosphorus
dynamics in calcareous and highly weathered soils. Soil Sci. Soc. Am. J. 53, 153-158.

Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in the
tropics. PhD thesis, University of Hawaii, Honolulu.









Singh, U., Thornton, P.K., Saka, A.R., Dent, J.B., 1993. Maize modeling in Malawi: a tool for
soil fertility research and development. In: De Vries, F.P. et al. (Eds). Systems Approach
for Agriculture Development. Proceedings of the International Symposium on Systems
Approach for Agriculture Development held in Bangkok, Thailand, 2-6 December 1991.
pp. 253-273.

Singh, U., Wilkens, P.W., Chude, V., Oikeh, S., 1999. Predicting the effect of nitrogen
deficiency on crop growth duration and yield. In : Robert, P.C. et al. (Eds.), Proceedings of
the Fourth International Conference on Precision Agriculture held in St. Paul, Minnesota,
19-22 July 1998. American Society of Agronomy, pp. 1379-1393.

Struif Bontkes, T.E., Wopereis, M.C.S., (Eds) 2003. Decision Support Tools for Small-Holder
Agriculture in SubSaharan Africa, A Pratical Guide. IFDC and CTA.

Sturgul, S.J., Bundy, L.G., 2004. Understanding soil phosphorus: an overview of phosphorus,
water quality and agricultural management practices. Nutrient and Pest Management
Program, University of Wisconsin-Madison, Madison.

Tiessen, H., Moir, J.O., 1993. Characterisation of available P by sequential extraction. In: Carter,
M.R. (Ed.), Soil sampling and methods of analysis. CRC Press, Boca Raton. pp. 75-86.

Tiessen H., Stewart, J.W.B., Cole, C.V., 1984. Pathways of phosphorus transformations in soils
of differing pedogenesis. Soil Sci. Soc. Am. J. 48, 853-858.

Tsuji, G.Y., Hoogenboom, G., Thornton, P.K., (Eds) 1998. Understanding Options for
Agricultural Production. Kluwer Academic Publisher, Boston.

Uehara, G. 1998. Synthesis. In: Tsuji, G.Y. et al. (Eds), Understanding Options for Agricultural
Production. Kluwer Academic Publisher, Boston, pp. 389-392.

Vincent, K.R., Chadwick, O.A., 1994. Synthesizing bulk density for soils with abundant rock
fragments. Soil Sc. Soc. Am. J. 58, 455-464.

Wafula, B.M. 1995. Applications of crop simulation in agricultural extension and research in
Kenya. Agricultural Systems 49, 399-412.

Walker, D.H., 2000. Decision support, learning and rural resource management. Agricultural
Systems 73, 113-127.

Wallach, D., Makowski, D., Jones, J.W., (Eds) 2006. Working with Dynamic Crop Models:
Evaluation, Analysis, Parameterization, and Applications. Elsevier, Amsterdam, The
Netherlands.

Wilson, M.V., Botkin, D.B., 1990. Models of simple microcosms: emergent properties and the
effect of complexity and stability. American Naturalist 135 (3), 414-434.

Wolf, J., De Wit, C.T., Janssen, B.H., Lathwell, D.J., 1987. Modeling long-term crop response to
fertilizer phosphorus. I. The model. Agronomy Journal 79, 445 451.









Ziadi, N., Belanger, G., Cambouris, A.N., Tremblay, N., Nolin, M.C., Claessens, A.. 2007.
Relationship between P and N concentrations in corn. Agronomy Journal 99, 833-841.









BIOGRAPHICAL SKETCH

Kofikuma Adzewoda Dzotsi was born in Lome, Togo (West Africa). He obtained his

GED in 1996 and in fall 1996 entered the School of Agronomy at the University of Lome (UL),

Togo. During his fifth and last year at the School of Agronomy (2001), Kofikuma participated in

a training workshop on systems analysis and modeling organized by the African division of the

International Center for Soil Fertility and Agricultural Development (IFDC). This program was

his first exposure to systems analysis and simulation modeling applied to soils and crops and he

decided to do his thesis research in this field. In March 2001, Kofikuma joined IFDC to carry out

his thesis research on "Long-term assessment of variety and sowing time effects on grain yield of

maize in southern Togo" that was defended in November 2002 and Kofikuma graduated as an

"agronomy engineer" from the Department of Plant Productions, School of Agronomy, UL.

Between September and December 2002, Kofikuma worked as a research assistant at IFDC's

Systems Approach Unit in Lome. In January 2003, he took up the position of agronomist in

IFDC's Natural Resource Management Program in Lome and worked on developing integrated

soil fertility management (ISFM) options for basil. During his tenure in the program, he was

responsible for evaluating and fine-tuning some ISFM-oriented decision support tools like

DSSAT, QUEFTS, SIMFIS. He also supervised two "agronomy engineer" theses in 2004 and

2005. After spending almost 3 years in the program, he decided to pursue a graduate program at

the University of Florida (UF). In fall 2005, Kofikuma joined the McNair Bostick Simulation

Laboratory in the Agricultural and Biological Engineering department at UF as a master's

student. Kofikuma married Pascaline Akitani-Bob in April 2005. They have one son, Eyram R.

Dzotsi.





PAGE 1

1 COMPARISON OF MEASURED AND SIMULATED RESPONSES OF MAIZE TO PHOSPHORUS LEVELS IN GHANA By KOFIKUMA ADZEWODA DZOTSI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Kofikuma Adzewoda Dzotsi

PAGE 3

3 ACKNOWLEDGMENTS Although the final presentation of this thesis is my responsibility, seve ral persons assisted with the planning and the implemen tation of the study and helped w ith a better distillation of the ideas presented here. Dr. James W. Jones (my supervisory committee chair) provided the fundamental guidance that I needed to complete this study. He represents for me at this particular time and in this particular situation the ideal a dvisor I could imagine. I have been particularly impressed by his appropriate and prompt interventi ons related to my questions and needs even if I drop by his office without an appoint ment or if he is on a trip. Dr. Dorota Z. Haman discussed with me on almost every occasion that we met in the department or on the campus, progress in my study. I was impressed by her suggestions during our first committee meeting as it relates to the conten ts of the chapters I should write even though she is from the irrigation area. Congratula tions on her recent appointment as Chair of the Agricultural and Biological Engineering Department. Dr. Samira H. Daroub was never tired of my emails and was always in touch from the Everglades. I was enriched by her experience with the in itial version of the soil-plant phosphorus model. Ms. Cheryl H. Porter (coordi nator of computer applicati ons in the McNair Bostick Simulation Laboratory) endured with me multiple trips through the phosphorus model computer code. She was always available to assist and facilitated tremendously my understanding of the phosphorus model. Dr. Samuel K. Adiku, Professor of Soil Science at the University of Ghana made the initial suggestion of letting me conduct the phosphorus fiel d experiment in Ghana. I benefited a lot from his experience and those of his collaborators at the Kpeve ag ricultural research station, in soil science when I was in Ghana. The field expe riment in Kpeve would not have been a success

PAGE 4

4 without him. Although he is now in the ABE de partment working on his own experiments, he continues to ensure that I obtai n the soil, plant and growth an alysis data that are accurate. Drs Upendra Singh, Ken Boote, Arjan Gijsman, Sh rinkant Jagtap, and Jon Lizaso directly assisted me in various ways at critical mome nts during the fine-tuning of the phosphorus model. I would like to thank Dr Jesse Naab from th e Savannah Agricultural Research Institute in Ghana for the data he made available for the purpose of this study. I would like to express my appreciation to my colleagues in the McNair Bostick Simulation Laboratory; at the International Center for Soil Fertility and Agricultural Development (IFDC) in Lome, Togo; my friends at Maguire Village, and all my other friends not specifically listed here for their various support and encouragement during this study. My family in the United States, Pascaline Akit ani-Bob and Robert E. Dzotsi, have been my true continuous, human, and spir itual support throughout this study.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES................................................................................................................ .......13 ABSTRACT....................................................................................................................... ............16 CHAPTER 1 INTRODUCTION TO MODELING PHO SPHORUS LIMITATIONS TO CROP PRODUCTION..................................................................................................................... ..18 Introduction................................................................................................................... ..........18 Phosphorus Problem in Agricultural Systems........................................................................18 Understanding Excess and Deficiency of Phosphorus in Agricultural Systems....................20 Coping with Excess and Deficiency of Phosphorus in Agricultural Systems........................21 Modeling as a Phosphorus Manageme nt Tool in Soils and Plants.........................................21 Soil-Plant Phosphorus Simulation Model in DSSAT.............................................................24 Objective and Hypothesis.......................................................................................................25 2 STATISTICAL ANALYSIS OF FIELD EXPERIMENT FOR TESTING THE MODEL...28 Introduction................................................................................................................... ..........28 Materials and Methods.......................................................................................................... .29 Field Experiments in Ghana............................................................................................30 Experiment in Kpeve, Ghana..........................................................................................30 Site description.........................................................................................................30 Experiment design and management........................................................................31 Soil sampling............................................................................................................31 Soil moisture measurements.....................................................................................32 Plant sampling and growth measurements...............................................................33 Experiment in Wa, Ghana...............................................................................................35 Site and experiment set up.......................................................................................35 Field and laboratory measurements..........................................................................35 Statistical Analysis..........................................................................................................36 Regression analysis (soil moisture)..........................................................................36 Analysis of variance at individual time points.........................................................37 Analysis of variance considering th e effect of time on the repeated measurements.......................................................................................................38 Results and Discussion......................................................................................................... ..39 Calibration of the TDR Meter.........................................................................................40 Crop Response Results at Kpeve Using Individual Time Points Analysis.....................41 Phenology.................................................................................................................41

PAGE 6

6 Grain yield and yi eld components............................................................................41 Aboveground biomass..............................................................................................42 Plant height...............................................................................................................42 Green leaf area.........................................................................................................42 Soil moisture............................................................................................................43 Crop Response Results at Kpeve Using Rep eated Measures Analysis Techniques.......43 Selection of a correlation structure using the AIC...................................................43 Effect of time on repeated measur ements of crop response variables.....................44 Phosphorus treatments by time interacti ons effects on repeated measurements......44 Effect on repeated measurements of phos phorus treatments averaged over time....44 Discussion of Results Obtained at Kpeve.......................................................................45 Results and Discussion for the Wa Experiment..............................................................47 Conclusions.................................................................................................................... .........49 3 THE SOIL-PLANT PHOSPHORUS MODEL IN DSSAT...................................................64 Introduction................................................................................................................... ..........64 Soil and Plant Phosphorus Modeling in DSSAT....................................................................65 Description of the Soil Phosphorus Model.............................................................................67 Soil Inorganic Module.....................................................................................................67 Inorganic phosphorus pools.....................................................................................68 Phosphorus transformations be tween the inorganic pools.......................................68 Phosphorus availability for uptake by plants...........................................................69 Soil Organic Module.......................................................................................................70 Organic phosphorus pools........................................................................................70 Phosphorus flows between the organic pools..........................................................71 Phosphorus mineralization and immobilization.......................................................72 The net phosphorus mineralized..............................................................................73 Description of the Plant Phosphorus Model...........................................................................73 Phosphorus in the Plant...................................................................................................74 Uptake......................................................................................................................... .....75 Soil Supply.................................................................................................................... ..75 Plant Demand and P Mobilization Pools.........................................................................75 Partitioning and Translocation........................................................................................76 Stress Factors................................................................................................................. ..77 Model Inputs and Outputs......................................................................................................78 Sensitivity Analysis........................................................................................................... .....78 Introduction................................................................................................................... ..78 Materials and Methods....................................................................................................81 Computer experiment...............................................................................................81 Settings for the computer experiment......................................................................82 Input factors, scenarios and model outputs..............................................................83 Method and design of the sensitivity analysis..........................................................87 Sensitivity index.......................................................................................................87 Results and Discussion....................................................................................................88 Soil inputs effects.....................................................................................................88 Plant parameters effects...........................................................................................91

PAGE 7

7 Interactions...............................................................................................................92 Special case of zero P fertilizer................................................................................92 Conclusion..................................................................................................................... ..93 Summary and Conclusion.......................................................................................................94 4 FIELD TESTING OF THE DSSAT PHOSPHORUS MODEL...........................................110 Introduction................................................................................................................... ........110 Materials and Methods.........................................................................................................112 The Soil-Plant Phosphorus Model.................................................................................112 Datasets for Testing the Model......................................................................................113 Parameters and Inputs for the Model Tests...................................................................113 Weather conditions.................................................................................................113 Soil conditions........................................................................................................114 Genetic coefficients................................................................................................114 Phosphorus parameters...........................................................................................115 Initial conditions.....................................................................................................116 Model Evaluation..........................................................................................................116 Model evaluation tools...........................................................................................117 Results and Discussion.........................................................................................................119 Weather........................................................................................................................ ..119 Genetic Coefficients......................................................................................................120 Phosphorus Parameters..................................................................................................120 Initial Conditions...........................................................................................................121 Model Evaluation at Kpeve...........................................................................................121 In-season growth....................................................................................................122 Final grain yield.....................................................................................................122 Wa............................................................................................................................. .....123 In-season growth....................................................................................................123 In-season shoot P concentration.............................................................................124 Final grain yield.....................................................................................................125 Conclusion..................................................................................................................... .......126 5 SUMMARY AND CONCLUSIONS...................................................................................138 APPENDIX A MEASURED GROWTH DATA AT KPEVE.....................................................................141 B MAPS OF THE EXPERIMENT SITES LOCATIONS.......................................................144 C INITIALIZATION OF SOIL INORGANIC AND ORGANIC PHO SPHORUS POOLS IN THE SOIL-PLANT PHOSPHORUS MODEL...............................................................146 Initialization of Inorganic Phosphorus Pools........................................................................147 From P Fractionation Data............................................................................................147 From Measured Available P Using the Anion Exchange Resin Method......................147 From Other Methods.....................................................................................................148

PAGE 8

8 Initialization of Soil Organic Phosphorus Pools...................................................................149 Initialization from P Fractionation Data........................................................................149 Initialization from Measured Organic P........................................................................150 Initialization from Organic C and soil pH.....................................................................150 LIST OF REFERENCES.............................................................................................................153 BIOGRAPHICAL SKETCH.......................................................................................................161

PAGE 9

9 LIST OF TABLES Table page 1-1 Response of maize to phosphorus app lication on a phosphorus-deficient soil in a fertilizer experiment carri ed out in Ghana in 1999............................................................26 1-2 Partitioning of total so il phosphorus in pools specified on Figure 1-1 in a soil from Carimagua, Colombia........................................................................................................26 2-1 Growth and development genetic coefficien ts for the Obatanpa cultivar used at both sites, Kpeve and Wa (Ghana).............................................................................................51 2-2 Summary of fertilizer application methods used in the experiment in Kpeve, Ghana......51 2-3 Specifications of two diff erent covariance structures used for modeling the effect of time on repeated measures in PROC MIXED for the Kpeve dataset................................51 2-4 Analysis of Variance for simple linear regression between soil moisture measurements using TDR and gravim etric methods at Kpeve, Ghana.............................51 2-5 Test of parameter estimates used to fit the linear regression model in the Kpeve experiment..................................................................................................................... .....52 2-6 Analysis of variance of phenological events tasseling, anthesis and silking in the Kpeve experiment..............................................................................................................52 2-7 Analysis of variance for gr ain yield (measured in kg ha-1) at Kpeve, Ghana....................52 2-8 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time points for crop aboveground biomass measured in kg ha-1 in the Kpeve experiment..................................................................................................................... .....52 2-9 Summary of results fr om ANOVA (mean squares (p-val ues), n = 4) at individual time points for plant height measured in cm in the Kpeve experiment.............................52 2-10 Treatment means at each day for plant height measured in cm, with least significant difference (LSD) in the Kpeve experiment ( = 0.05).......................................................53 2-11 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time points for green leaf area measured in cm2 per plant at Kpeve.................................53 2-12 Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time points for soil moisture readings (in %) using TDR at Kpeve...................................53 2-13 Akaike Information Criterion (AIC) te st for two covariance structures in PROC MIXED for repeated measures analysis for the Kpeve experiment..................................53

PAGE 10

10 2-14 F -values and significance probabilities us ing univariate ANOVA, and for test of fixed effect using two covariance stru ctures in PROC MIXED for the Kpeve experiment..................................................................................................................... .....54 2-15 Physical and chemical char acteristics of the soil at the experimental site in Kpeve, Ghana.......................................................................................................................... .......54 2-16 Classes of phosphorus availability a ccording to the Bray 1 extraction method................55 2-17 Characterization of the different forms of soil phosphorus at Kpeve, Ghana. Data are reported in mg/kg.............................................................................................................. .55 2-18 Physical and chemical characteristics of the soil at the experimental site in Wa, Ghana.......................................................................................................................... .......55 2-19 Main effects of nitrogen and phosphorus on phenological development in maize at Wa, Ghana...................................................................................................................... ....56 2-20 Main effects of nitrogen and phosphorus on leaf area indices of maize at Wa, Ghana.....56 2-21 Main effects of nitrogen and phosphor us on cumulative aboveground biomass (in kg ha-1) of maize at Wa, Ghana..............................................................................................56 2-22 Main effects of nitrogen and phosphorus fruit yield components of maize at Wa, Ghana.......................................................................................................................... .......57 3-2 Soil category-dependent calculation of P Fertilizer Availability Index.............................96 3-3 Summary of decomposition rates for the soil organic pools and C:P ratios at which phosphorus is allowed to enter the specific pools..............................................................96 3-4 Optimum and minimum phosphorus cont ent (%) in different plant parts and maximum and minimum plant N:P ratio at thre e growth stages, as used in the model for maize...................................................................................................................... ......97 3-5 Summary of parameters in the soil-plant phosphorus model.............................................98 3-6 Summary of additional i nputs required to run the so il-plant phosphorus model in DSSAT.......................................................................................................................... .....99 3-7 Selected physical and chemical propertie s of the Kpeve soil used in the sensitivity analysis, as estimated from pedotransfer functions in DSSAT......................................100 3-8 Summary of inputs factors and outputs fo r the sensitivity analysis of the P model........100 3-9 Specification of the differ ent levels of the input factor s Shoot P and Seed P for the sensitivity analysis of the P model.............................................................................101

PAGE 11

11 3-10 Main, interactions, and total sensitivity i ndices (unitless) of bi omass for factors used in the sensitivity analysis.................................................................................................101 3-11 Main, interactions, and total sensitivity indices (unitless) of grain yield for factors used in the sensitivity analysis.........................................................................................102 3-12 Main, interactions, and total sensitivity indices (unitless) of plant uptake of P for factors used in the sensitivity analysis.............................................................................102 3-13 Main, interactions, and total sensitivity indices (unitless) of biomass for a special case of zero P fertilizer....................................................................................................102 3-14 Mean aboveground biomass, grain yiel d and total P uptake corresponding to each level of the input factors used in the sensitivity analysis.................................................103 4-1 Growth and development genetic coefficien ts for the Obatanpa cultivar used at both sites, Kpeve and Wa, for testing the phosphorus model..................................................127 4-2 Plant parameters used for testi ng the phosphorus model at Kpeve and Wa....................128 4-3. Soil parameters used for testing the phosphorus model at Kpeve and Wa......................129 4-4 Values of additional inputs required to run the soil-plant phosphorus model for the Kpeve and Wa experiments.............................................................................................130 4-5 Estimated initial conditio n soil parameters for Kpeve.....................................................130 4-6 Estimated initial condit ion soil parameters for Wa..........................................................130 4-7 Summary of aboveground biomass error st atistics for the Kpeve and Wa experiments.131 A-1 Monthly total rainfall in 2006 (one standa rd deviation of rainfa ll), mean daily solar radiation, and mean daily te mperature collected during the Kpeve experiment in 2006........................................................................................................................... .......141 A-2 Days to tasseling (one standard deviation of four replications), days to anthesis (one standard deviation of four replications), a nd days to silking (one standard deviation of four replications) for the experiment in Kpeve, Ghana...............................................141 A-3 Measured mean aboveground biomass (one standard deviation of four replications) for four phosphorus treatments, sampled four times during the growing season in the Kpeve experiment............................................................................................................141 A-4 Mean green leaf area (one standard deviation of four replications) for four phosphorus treatments, measured seven tim es during the growing season in the Kpeve experiment............................................................................................................142

PAGE 12

12 A-5 Mean maize height (one standard deviation of four replications) for four phosphorus treatments, measured seven times duri ng the growing season in the Kpeve experiment..................................................................................................................... ...142 A-6 Mean soil moisture (one standard deviation of four re plications) in four phosphorus treatments plots, measured using TDR ei ght times during the growing season in the Kpeve experiment............................................................................................................142 A-7 Measured mean grain yield (one standard deviation of four re plications), unit grain weight (one standard deviati on of four replications), and grain number (one standard deviation of four replications) for four phosphorus levels in the Kpeve experiment......143 C-1 Relationship between inorganic P pools a nd P extracted using the Hedley procedure...151 C-2 Specification of soil categories........................................................................................151 C-3 Equations for calculating initial inorgani c P labile from differ ent extraction methods for different soil categories..............................................................................................151 C-4 Relationship between organic P pools and P extracted using the Hedley procedure......152 C-5 Equations for calculating initial total organic P from so il organic carbon (OrgC) and pH for different soil categories........................................................................................152

PAGE 13

13 LIST OF FIGURES Figure page 1-1 Pools of phosphorus in the soil and rela tionships between soil and plant phosphorus......27 2-1 Monthly total rainfall in 2006 (mm) with error bars co rresponding to one standard deviation of rainfall, mont hly total rainfall average fr om 2003 to 2005, and monthly average daily temperature (oC) in 2006 at Kpeve..............................................................58 2-2 Sequential fractionation steps used for extracting the different forms of phosphorus from soil samples taken before planting of the Kpeve experiment...................................59 2-3 Monthly total rainfall (mm) with error bars corresponding to one standard deviation of rainfall and monthly average daily temperature (oC) at Wa in 2004.............................59 2-4 Simple linear regression of volumetric soil moisture (% ) determined by using Time Domain Reflectrometry and gravimetric sampling at Kpeve............................................60 2-5 Phenology of maize as affected by phos phorus application at Kpeve. Error bars represent standard deviati ons of measurements taken from four replications...................60 2-6 Stover and grain yield of maize as affect ed by phosphorus fertilizer at Kpeve. Error bars represent one standard deviation of m easurements taken from four replications......61 2-7 Grain number per m2 and unit grain weight as affect ed by phosphorus fertilizer at Kpeve.......................................................................................................................... .......61 2-8 Aboveground biomass of maize as affected by phosphorus fertilizer application at Kpeve. Error bars represent one standard deviation of m easurements taken from four replications................................................................................................................... ......62 2-9 Height of maize as affected by phosp horus fertilizer in the Kpeve experiment................62 2-10 Green Leaf Area of mai ze in the phosphorus fertilizer experiment at Kpeve...................63 2-11 Variation in soil moisture measured us ing Time domain reflectometry in the Kpeve phosphorus experiment......................................................................................................63 3-1 Processes in the integrated so il-plant phosphorus model in DSSAT...............................104 3-2 Optimum and minimum P concentration in maize shoots used in the plant P model.....105 3-3 Maximum and minimum N:P ratios used in the plant module to limit uptake of P........105 3-4 Relationship between maize shoot P c oncentrations and P stresses affecting vegetative partitioning and photosynthesis......................................................................106

PAGE 14

14 3-5 Simulated plant total aboveground bioma ss, grain yield and plant uptake of P at different levels of initial PiLabile, initial organic P and P fertiliz er. Error bars shown represent one standard deviation......................................................................................106 3-6 Simulated total plant aboveground bioma ss, grain yield and plant uptake of P at different levels of maximum uptake fraction, optimum shoot and seed P concentrations. Error bars shown re present one standard deviation................................107 3-7 Sensitivity indices for the six i nput factors and their interactions...................................108 3-8 Simulated response of total plant abov eground biomass to phosphorus fertilizer at different levels of PiLabile...............................................................................................109 3-9 Simulated response of total plant abovegr ound biomass to PiLabile at different levels of fraction of labile P.......................................................................................................109 4-1 Comparison of simulated and measured gr ain for different phos phorus levels in the Kpeve experiment............................................................................................................132 4-2 Decomposition of the grain yield MSE for the Kpeve experiment.................................132 4-3 Comparison of simulated and measured biomass on four samples taken during the season for the four treatments tested in Kpeve................................................................133 4-4 Decomposition of the in-season biomass MSE for the Kpeve experiment.....................133 4-5 Comparison of measured and simulated maturity grain yield obtained in the Wa experiment using the 1:1 line...........................................................................................134 4-6 Measured and simulated responses of matu rity grain yield to di fferent combinations of nitrogen and phosphorus levels in the Wa experiment................................................134 4-7 Decomposition of the grain yield MSE for the Wa experiment......................................135 4-8 Components of the biomass MSE for the Wa experiment at five sampling times..........135 4-9 Measured and simulated responses of cu mulative biomass to different combinations of nitrogen and phosphorus levels in the Wa experiment................................................136 4-10 Measured and simulated responses of shoot P concentration to different combinations of nitrogen and phosphorus levels in the Wa experiment.........................137 4-11 Variation of the shoot P concentratio n during plant growth as affected by three phosphorus levels in th e Wa experiment.........................................................................137 B-1 Map of the African continent show ing Ghana, the country where the field experiments were carried out...........................................................................................144

PAGE 15

15 B-2 Map of Ghana showing the location of the two study sites, Kpeve in the South and Wa in the North................................................................................................................145

PAGE 16

16 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science COMPARISON OF MEASURED AND SIMULATED RESPONSES OF MAIZE TO PHOSPHORUS LEVELS IN GHANA By Kofikuma Adzewoda Dzotsi December 2007 Chair: James W. Jones Major: Agricultural and Biological Engineering Efficient nutrient management in agricultural sy stems requires the availability of tools that can help in meeting research objectives of understanding the transforma tions that nutrients undergo to become available to plants and pred icting how these transformations are related to economic outputs from the systems. The crop models in the Decisi on Support System for Agrotechnology Transfer (DSSAT) have been recognized worldwid e for meeting these objectives for nitrogen. However, without a phos phorus model, the applicability of the DSSAT crop models in phosphorus deficient environments will remain questionable. In this study, a soilplant phosphorus model linked to DSSAT was described, analyzed and tested. The sensitivity of the model to six key i nput factors was studied based on a global sensitivity analysis approach. The model was test ed on two P-deficient soils from Ghana (Kpeve and Wa) with maize as the test plant. Proce sses accounted for by the model include phosphorus movements between inorganic (lab ile, active and stable), organi c (active and stable) pools and plants. Results of the sensitivity analysis showed the greatest effects of initial inorganic labile P (initial PiLabile) and fertilizer P on biomass, grain yield and tota l P uptake (sensitivity index of 0.11 for initial PiLabile and 0.30-0.43 for P fertilize r). Smaller effects were found for the fraction

PAGE 17

17 of root labile P that is sol uble (sensitivity index of 0.03-0.04), the shoot P (sensitivity index of 0.03-0.09) and seed P (sensitivity in dex of 0.15) on total P uptake. Statistical analysis of grain yield and biomass did not reveal any signi ficant differences at the 0.05 probability level at Kpeve because the p hosphorus content of this soil was at the limit between deficiency and sufficiency and the orga nic matter content of the soil was relatively high (close to 2.0%). Grain yield and final biomass responded at Wa with 100% increases in the 60 kg [P2O5] ha-1 treatments over the nil-P treatments. Biom ass and yield were stable between the two treatments of 60 and 90 kg [P2O5] ha-1 at Wa. Evaluation of the model indicated that the m odel was able to achieve good predictability skill at Kpeve with a grain yield RRMSE of 8% and a final biomass RRMSE of 5%. The congruence between simulation and measuremen t was fair at Wa. The RRMSE was 14% for grain yield and 30% for final bi omass. At Wa however, the model gave a reasonable prediction of the pattern of variability among measurements with an LCS averaged over the five sampling dates of 17%. Because the complex soil P chemis try makes the availability of phosphorus to plants extremely variable in general, further testing of this model in other agro-ecological conditions should preced e its application.

PAGE 18

18 CHAPTER 1 INTRODUCTION TO MODELING PHO SPHORUS LIMITATIONS TO CROP PRODUCTION Introduction Phosphorus (P) is recognized as a major nutrient that must be present in living organisms to enable them to maintain a continuous life cycle. It is an essential component of adenosine triphosphate (ATP), the energy currency of the living cell. Th e energy-consuming biochemical processes that continuously take place in the cell are driven by the energy-rich phosphate group contained in ATP. For example, in crop nutriti on, the uptake and assim ilation of nutrients use energy in the form of ATP. The synthesis of new molecules responsible for mass accumulation in living organisms and perpetuation of species on earth all involve P either as ATP or deoxyribonucleic acid (DNA) or ribonucleic acid (RNA). Phosphor us is transferred from one organism to another through the various food chai ns. For terrestrial organisms, soils satisfy most of their P need mainly through plants (Johnston, 2000). The P content of healthy plant leaf tissue is low however, ranging from 0.2 to 0.4% of the dry matter (Brady and Weil, 2002). Although plant uptake of P is constrained by th e low quantity of this element present in soil and the very low solubility of P compounds found in soils, most natural ecosystems have developed relatively well wit hout any P management program s (Brady and Weil, 2002). These systems are naturally organized to recycle the nutrient and maintain an overall non thermodynamic equilibrium. Phosphorus Problem in Agricultural Systems In agricultural systems where most nutrien ts balances have been displaced by human intervention, the relatively low mobility of P in many soils has led to the appearance of areas of soil accumulation and deplet ion of this nutrient.

PAGE 19

19 Most soils in sub-Saharan Africa have very little capacity to supply P for plant growth, which allows plants grown on those soils to be re sponsive to P fertilizer applications (Table 11). Phosphorus deficiency is thought to be one of the reasons why subSaharan Africa is the only major region in the world where per capita food production has actually declined in the past three decades (Brady and Weil, 2002). The phosphorus problem in most sub-Saharan African soils has five facets: The soils have developed under conditions c onducive to advanced weathering. During these relatively long periods of intensive weathering, exte nsive losses of P occurred resulting in low P soils. Mo st soils solution P ranges from 0.03 to 0.50 ppm with 0.25 ppm considered adequate. For a crop requireme nt of 40 kg [P] ha-1 for example, the soil solution containing 0.25 ppm must be replenished 80 times in a hectare furrow slice (15 cm deep x 1 ha area), whic h does not happen naturally; The P compounds commonly present in soils ar e highly insoluble and have a very low diffusion rate in many soils posing a problem for plant uptake. They do not readily release P to the soil solution in a useabl e form by plants. For example, P fixed by reaction with aluminum in acid soils is inso luble for plant uptake. The readily available pool of P that is in equilibrium with the soil solution P (Figure 1-1) can be as small as 10% in some soils (Table 1-2). The rate of di ffusion of P in some soils can be as low as 10-12 to 10-15 m2 s-1 and high plant uptake rates create a zone around the root that is depleted of P (Schachtman et al., 1998); When soils are supplied with external P in the form of fertilizers, the nutrient is fixed, adsorbed or absorbed and with time tends to re turn to stable forms, strengite, variscite (in acid soils) and apatite (in alka line soils) (Figure 1-1). As a consequence P fertilizer recovery is low in most agricultural syst ems relative to the other major nutrients (nitrogen and potassium); Crop harvest exports significant amounts of P from the soil with limited amounts of residues returned to the cropping system; Use of external P inputs in th e form of mineral fertilizers or manure, especially for food crops, is not common practice. Farmers do not have access to the appropriate P fertilizers, or the cost of their being tran sported and applied is prohibitive. In addition, the fertilizer requirement for improved yield can be high on soils with high P sorption capacity (Table 1-1). In industrialized areas, P fertilizer use ha s increased drasticall y during the past few decades (FAO, 2003). The relatively low plant upta ke of P coupled with the low mobility of

PAGE 20

20 the nutrient in some soils mean that much of th e P fertilizer applied is not removed with the harvested crop or lost from the soil (Schmidt et al., 1996). In fact, soils in these areas have developed rather high P levels resulting from many years of over-fertilization with P. Understanding Excess and Deficiency of Phosphorus in Agricultural Systems The challenge of dealing with both the ex cess and deficiency of P in agricultural ecosystems is crucial to attempt to restore the P balance in these systems and make P management programs sustainable and envir onmentally sound. A primary step towards tackling the challenge of P management in agri cultural systems is understanding the P behavior in soils in relation to plants needs and their environment. Excess P can be detrimental to the a quatic ecology. Although plant proliferation stimulated by supply of limiting nutrients is cons idered beneficial in terrestrial ecosystems, aquatic systems like lakes, streams and ponds ca n become unsatisfactory environments when enriched with excessive P through runoff, eros ion and, in some cases, leaching. The unwanted growth of algae and of aquatic weeds (termed eu trophication) resulting fr om this P enrichment can seriously perturb the aquatic ecosystem. Wh en this community of opportunistic algae and weeds die, they sink to the bottom of the wa ter where their decom position by microorganisms uses much of the oxygen in the water creating anoxic conditions. This process leads to fish kills, displaced nutrients balance, and can make the water unsuitable for drinking (Brady and Weil, 2002; Sturgul and Bundy, 2004). Phosphorus deficiency can constitute a se rious problem for crop production because it has a negative effect on leaf area index (Pellerin et al., 2000) limiting the interception of photosynthetically active radiation by the plan t and resulting in low biomass accumulation (Colomb et al., 2000). The rate of leaf appearance is slowed down and the final leaf number is reduced in P stressed plants (Singh and al., 1999). Colomb et al. (2000) showed that in P-

PAGE 21

21 deficient maize plants in their study the rate of leaf appearance and the final area of leaves located below the main ear were reduced by 18 to 27%. The ultimate economic effect of P deficiency is yield reduction (Table 1-1). Coping with Excess and Deficiency of Phosphorus in Agricultural Systems Extensive and long term agronomic experiment s have been conducted to attain a greater understanding of the behavior of soil P and pr opose options for P management in agricultural systems. The mechanism motivating and the time f actor associated with P draw-down or buildup in soils have been examined as steps towards assessing oppor tunities for reducing P loadings in waters (Kelling et al., 1998; Sartain, 1980). Manage ment strategies proposed for decreasing the P content of high P soils include mining soil P (i.e., harvesting P taken up from the soil by a crop grown wit hout external P addition) (K oopmans et al., 2004), growing appropriate corn varieties as P removal agents (Eghball et al., 2003). With the twofold concern of replenishing soil phosphorus in P deficient ag ricultural systems while avoiding losses to aquatic ecosystems, continuous applications of sm all rates of P have been proposed as adapted management strategy for smallscale farming syst ems (Nziguheba et al., 2002; Schmidt et al., 1996). Modeling as a Phosphorus Management Tool in Soils and Plants The necessity of developing P management strategies requires the availability of appropriate tools that empower managers and d ecision-makers with the ability to control the human-modified P cycle in agricultural system s. Statistical summaries have been routinely used to produce in an integrated way a logic in terpretation of agronomic experimental results. However, these parameters and other classical mathematical methods used to study and explain the behavior of nonliving physical or chemi cal processes may be insufficient (Jones and Luyten, 1998). Agricultural ecosystems involving biological processes ar e highly complex and

PAGE 22

22 have many components that interact in non linea r ways. The non linear interaction means that many disciplines working, for example, on the sa me P management problem may be looking at the facets that are only meani ngful for their study while interacting components that may provide clues to the solution are not given enough attention. An interdisci plinary approach that places the P management problem at the cente r of the soil-plant-atmosphere system and recognizes the effects on the problem of interac tions between disciplines concerned is useful for developing efficient management tools. An ideal P management program is at the minimum concerned with i) understanding the behavior of P in agricultural systems; ii) synthesizing the knowledge and information obtained in an integrated way so that interactions occurring in the system are not lost but harnessed to enhan ce understanding; iii) producing user-friendly management tools that depict the best unders tanding of the system. Well-tested simulation models that represent the cr opping system with mathematical relationships provide a sound scientific approximation of physical, chemical and biological processes governing complex ecological systems and represent such tools. When appropriately validated, those simulation models provide the opportunity to understand simplifications of the universe (Odum and Odum, 2000); study ecosystems w ithout having to experiment on actual systems (Uehara, 1998) especially when experimentation is im possible or ethically unacceptable; make predictions; support decision ma king and communicate more effi ciently research findings by integrating information into a more useable form (Newman, 2000). Phosphorus modeling has given a ttention to soil and plant pr ocesses that affect the P cycle. Residual effects of soil P have been of interest because in many soils, the P that is applied to the soil but not taken up by the plant is not lost, and the build-u p in the soil toplayer can be reused for crop production (Pheav et al., 2003) and thought of as a capital investment.

PAGE 23

23 Models now exist that address th e issues of long term recovery of applied P in the form of fertilizers (Wolf et al., 1987; Janssen et al., 198 7; Schmidt et al., 1997), long term changes in soil P extracted using conventiona l methods (Karpinets et al., 2004) long term P leaching from the soil profile (Del Campillo et al., 1999), and long term effects of erosion-induced soil nutrient loss, including P, on crop productivity (Jones et al., 1984). Lewis and McGechan (2002) compared four soil P models, AMINO from the Netherlands, GLEAMS and DAYCENT from the USA and MACRO from Swede n, in order to ascertain their limitations and evaluate their capability to simulate the tr ansport of soluble and particulate P, surface application, mineralization / immobilization, absorption / desorption, leaching, runoff and uptake by plants. The P module of GLEAMS is actua lly an essentially una ltered version of the model developed by Jones et al. (1984). Lewis and McGechan concluded from their analyses that all the models only have a partial repres entation of the soil pro cesses examined. They suggested that more accurate dynamic simulatio ns of soil processes wi ll necessitate a hybrid model that incorporates the diffe rent aspects of soil P dynamics that the models studied have failed to critically account for. The most relevant processes for crop produc tion in P deficient systems include the quantification of development, growth and yiel d as limited by P. These processes can only be handled by comprehensive simulation models of crop growth and development. For the purpose of enhancing the applicability of a m odel of this kind, soil P gains by the plant resulting from the mineralization of organic matt er, particularly in low input cropping systems must be recognized in addition to P obtained from chemical fer tilizers (Probert, 2004). In fact, organic materials are used in many competing wa ys in smallscale cropping systems including soil fertility replenishment (F ofana et al., 2005). Factors c ontrolling the decomposition of

PAGE 24

24 organic materials in these systems often favor faster mineralization ra tes of nutrients. The addition of limited amounts of fertilizer tends to offset the negative effect of low-quality organic materials and accelerate their decompos ition. Nutrients that would normally cycle over extended periods could therefore be released over a relatively short time, increasing total available nutrients for plant uptake (Kaboneka et al., 2004). Soil-Plant Phosphorus Simulation Model in DSSAT To meet these needs, a capability is needed in the Decision Support System for Agrotechnology Transfer (DSSAT) Cropping System Models (CSM) (Jones et al., 2003) to model i) P limited crop growth, development and yi eld and ii) P released by organic resources. The soil-plant P module that has been linked to the DSSAT CSM and still operates as an experimental version is based on studies by Daroub et al. (20 03). A new Soil Organic MatterResidue module called CENTURY that accounts for nutrient mineralization from organic resources was recently impleme nted in DSSAT. Gijsman et al. (2002) reported that the CENTURY module simulated with high accuracy the development of SOM content in a long term bare fallow experiment in Rothamsted, UK and gave a fair congruence between simulated and measured data for a 1-y ear experiment in Brazil. The Decision Support System for Agrotechnolo gy Transfer is comprised of a suite of models that simulate on a daily basis the develo pment, growth and yield of more than 16 crops. The models are organized under different gr oups. Among these groups, the CERES family for cereals and the CROPGRO family for legumes ar e important components that have proven to be successful in their applicati ons worldwide. The system also contains programs that allow the analysis of effects of multiy ear variations of f actors like weather on crop production. Other programs permit the analysis of rotational and spatial experiments. The DSSAT models use quantitative climate, soil, genetic and manage ment information as inputs to simulate, among

PAGE 25

25 many other outputs, grain yield and its compone nts, crop biomass, anthesis, silking and physiological maturity dates. The Decision S upport System for Agrotechnology Transfer has been used in several studies that include residues and inor ganic nitrogen management in Nigeria (Jagtap and Abamu, 2003), crop management strategies in extension systems in Kenya (Wafula, 1995), investigating variety and sowi ng time technologies in Nigeria (Jagtap, 1999) and Togo (Dzotsi et al., 2003), studying the effect of water and nitrogen deficiency on crop duration and yield in Florida, Hawaii, Nigeria and Togo (Si ngh and al., 1999), analyzing soil fertility research and development op tions in Malawi (Singh et al., 1993). Objective and Hypothesis The overall objective of this study is to present the soil-p lant phosphorus model implemented in the DSSAT CSM for maize and re sults of field testing. The main hypothesis was that P fertilization increases soil inorganic P availability for plant uptake, which promotes higher grain yield and biomass pr oduction and shortens the time requi red for ripening in maize.

PAGE 26

26 Table 1-1. Response of maize to phosphorus ap plication on a phosphorus-deficient soil in a fertilizer experiment carri ed out in Ghana in 1999 Source of phosphate P (kg ha-1) P2O5 (kg ha-1) Grain Yield (kg ha-1) Control 3949 Triple super phosphate 40 92 5135 Togo rock phosphate 63 144 6252 Source: Adapted from FAO, 2005 Table 1-2. Partitioning of total soil phosphorus in pools specified on Figure 1-1 in a soil from Carimagua, Colombia Soil phosphorus Parts per million (mg kg-1) Per cent of total soil P (%) kg ha-1 in an 1ha-15cm deep soil Readily available 18 10 36 Stable forms 78 43 156 Organic 86 47 172 Total soil P 182 100 364

PAGE 27

27 Figure 1-1. Pools of phosphorus in the soil and relationships be tween soil and plant phosphorus weatherin g dissolution p reci p itation adsorption desorption mineralization immobilization Soil Solution Phosphorus Readily Available Phosphorus Stableforms --Apatite --Variscite --Strengite Organic Phosphorus Leaching (minor) Fertilizer Plant Uptake Plant Residue

PAGE 28

28 CHAPTER 2 STATISTICAL ANALYSIS OF FIELD EXPERIMENT FOR TESTING THE MODEL Introduction Assessing plant response to phos phorus (P) is an important st ep towards understanding its behavior in agricultural systems. Plant response to P can be evalua ted using different soils with P levels ranging from low to high (Colomb et al., 2 000) or by testing different applications of a P fertilizer on the same P deficien t soil (Fofana et al., 2005). In both situations initia l soil testing for P is essential to determine the P status of the soil of interest. The diagnosis of P level in so ils is complicated by the comp lex chemistry of the nutrient. This complexity is the basis for assessing soil P content using extractants for their effectiveness to solubilize P tied up in different forms. The qua ntity of P extracted will vary with the reagent used. However, many P extraction methods ar e widely accepted and used because they adequately distinguish between so ils on the basis of the responsi veness of crops to P supply. Soil P test does not provide information about the ava ilable P that can be actually taken up by crops but relates to that quantity of P which is correl ated with plant response. This implies that soil analytical data allow the classi fication of soils descriptively in terms of P availability (e.g. deficient, sufficient, high) but these classes are on ly related to the probabl e response of a crop to an appropriate supply of P (Johnston, 2000). Th e relationship between soil P test and plant response cannot be established as necessarily deterministic. For ex ample, if a soil tests very low in P, a 75% probability exists that plant response will be observed. If the soil test is low in P, there is a 50% chance that plants will respond to P applications. If the soil P level is medium, the response probability of plants is only 25%. Plants will not be responsive to P if the soil test indicates a high P level (Havlin et al., 1999).

PAGE 29

29 Inorganic and organic fertilizers can be us ed as P sources in experimenting with Pdeficient soils. The amount of P provided by the decompositi on of organic materials over a cropping season can be small and relatively slow however, but continuous over several years. Inorganic fertilizers on the contra ry can supply more P to the plan ts at a higher rate and over a relatively short period of time, a cropping season for instance. However, a big portion of the fertilizer applied to plants in cropping systems is retained in th e soil, which constitutes the cause of long term accumulation of P in soils with a history of conti nuous P fertilization. The fraction of P applied in the form of inor ganic fertilizer that is actua lly taken up by plants is about 0.2. This fraction is termed apparent recovery of the fertilizer. An appropriate art of managing P in cropping systems would be to integrate the use of organic resources with i norganic fertilizers so that soil organic matter can play its role of improving physical a nd biological properties of soils while immediate nutrients need s are satisfied by inorganic fertilizers (Janssen, 1993). Plant response to P has been described as increasing the leaf area therefore allowing a higher interception of ph otosynthetically active radiation an d resulting in a higher biomass accumulation and grain yield (Pel lerin et al., 2000; Colomb et al., 2000; Plenet et al., 2000b). Other studies reported that the rate of leaf appearance was sl owed down and the final leaf number was reduced in P-stressed plants (Plenet et al, 2000a; Singh and al., 1999). The present chapter describe s two field experiments conducted in Ghana in 2004 and 2006 on soils that tested low in availa ble P and therefore considered P-de ficient. Statistical procedures are used to summarize the data collected and understand the plant response observed. Materials and Methods Field experiments were conducted in Kpeve a nd Wa, Ghana in order to measure growth and development of maize as affected by P ferti lizer. These experiments are described next with the statistical methods used to analyze them.

PAGE 30

30 Field Experiments in Ghana The experiments were carried out in two different agro-ecological zones of maize production in Ghana. The Kpeve site is located in the south in the Transitional zone with two distinct rainy seasons and annua l rainfall ranging from 1100 to 1400 mm. The Wa site is located in the North in the Guinea Savannah with one rainy season and annual ra infall ranging from 800 to 1200 mm (FAO, 2005). Experiment in Kpeve, Ghana The experiment in Kpeve measured the effect of different levels of phosphorus fertilizer on the growth and development of the maize cultivar Obatanpa (Table 2-1). Site description The experiment was conducted in 2006 during the major rainy season (March to July) on a sandy loam soil at the experimental site of the Mi nistry of Food and Agricu lture research station at Kpeve in southern Ghana (6o 40.80 N, 0o 19.20 E). The site is charac terized by an altitude of 67 m above sea level, an average annual temp erature of 28 degrees C and an average annual rainfall of 1300 mm falling in two rainy seasons, March to July and September to October (FAO, 2005) (Figure 2-1). The landscape is highly une ven with chains of hills surrounding the experimental station. Although the topography of th e experimental site was almost flat, small micro-topography differences may have important water and nutrient management consequences in this type of terrain. This topography reinforced the need for treating the experimental plots individually regarding all data co llected. The soil is classified as Haplic Lixisol which has a dark grayish brown topsoil and grayish brown to br own subsoil (Adiku, 2006). An automatic weather station to monitor maximum temperature, mini mum temperature, and solar radiation and an automatic rainfall datalogger to r ecord daily rainfall were locate d respectively at 1 km and 200 m from the experimental field.

PAGE 31

31 Experiment design and management Maize ( Zea mays L. cultivar Obatanpa a good nursi ng mother, Table 2-1), was planted on May 27. Seven days prior to planting, 5206 kg ha-1 of vegetation of a two-year natural bush fallow dominated by elephant grass ( Pennisetum purpureum ) and guinea grass ( Panicum maximum ) was plowed into the soil to a depth of 30 cm. The field was hand-harrowed to a depth of 10 cm and leveled three days before planting. Three levels of P and a control treatment were explored: low (10 kg P ha-1), medium (30 kg P ha-1) and high (80 kg P ha-1). A total of 30 kg K ha-1 was applied as Potassium Nitrate two times during the growing season, at planting and two weeks af ter planting. A total of 150 kg N ha-1 was applied at the rate of 50 kg ha-1, at planting, four weeks after planting and six weeks after planting as Ammonium Sulfate. The Potassium Nitrate provided 10 kg ha-1 of the total 150 kg ha-1 of the nitrogen applied. The application methods varied with the growth stage (Table 2-2). The existence of a slope gradient on the expe rimental field motivated the arrangement of the treatments in a Randomized Complete Blocks design with four replications. Each plot or experimental unit was composed of 14 rows 80 cm apart. Each row contained 30 hills 40 cm apart making up a total of 420 hills per pl ot. The total area of a plot was 134.4 m2. At planting and emergence, the plant population was 9.38 plants m-2. The field was thinned 10 days after planting to reduce the plant population to 6.25 plants m-2. The application of sufficient rates of nitrogen (150 kg ha-1) and potassium (30 kg ha-1) and the control of soil variabili ty through blocking were expected to highlight the effect of P deficiency in the crop. Soil sampling Soil samples were taken at three depths, 010, 10-20 and 20-30 cm on all 16 plots before planting, at silking and at final harvest. Texture, organic matter content, phosphorus content, exchange complex and acidity were determined on the samples in the soil testing laboratory at

PAGE 32

32 the University of Ghana. A modified Hedley a pproach for P fractionation was used to quantify inorganic labile, microbial and stable, and organic P in the samples. The P fractions (Table 2-17) were performed in the Wetland Biogeochemistr y Laboratory at the University of Florida following a four-step sequential extraction descri bed by Reddy et al. (1998) (Figure 2-2). The extractants in order were: 1, 1.0 M KCl; 2, 0.1 M NaOH; 3, 0.5 M HCl; 4, Residual P. The extraction with reagent 1, potassium chloride re moved that portion of P readily available to plants. The alkaline reagent (NaOH) extracted P associated with iron and aluminum while that extracted by reagent 3 (HCl) is probably associ ated with calcium (Figure 2-2). In the solution extracted by the alkaline reagent, both organic (Po) and inorgani c P (Pi) were determined. The residual P in the soil was recovered after combustion at 550 oC for 4 h and dissolution in 6.0 M HCl (Reddy et al., 1998). Soil moisture measurements Soil moisture is generally determined from oven-dried soil samples at 105 oC until constant weight. The moisture difference between the fresh and the dried soil relative to the dried soil is established as the gravimetric soil water conten t. The gravimetric soil water content can be further converted into volumetric soil water content by multiplying it by the bulk density of the soil. Monitoring soil water using gravimetric sa mpling can be tedious and not practical especially when the desired frequency is two to three-day intervals. Time domain reflectometry (TDR) technology is one of the best methods to quickly and accurately measure soil moisture. The technique is based on generating and remotely sensing a return ener gy signal that travels down and back through the soil. The travel time m easured is dependent on the quantity of water present in the soil. This information is then converted into volumetric water content. Because soils have different properties that can influe nce the way TDRs capture and read the moisture

PAGE 33

33 status of the soil, the ca libration of the meter to field conditio ns is an important step towards its use for extensive applications. In the present study, soil moisture was m onitored during the enti re course of the experiment at four GPS-referenced prelocated po ints on each plot in the top 12 and 20 cm using a portable soil moisture meter (FieldScout TD R 300) manufactured by Spectrum Technologies, Inc. The TDR readings were ta ken at 2 to 6-day intervals. In order to calibrate the TDR, 67 pairs of TDR readings and gravimetric soil samples were separately taken at anthesis at three soil moisture conditions low, medium and high. Fresh weight and soil volume were determined on the samples prior to drying. After oven-drying at 105 oC until constant weight, the fraction of the gravel was determined on each sample. The data were used to correct the bulk density of the soil using the following formula (Vincent and Chadwick, 1994): Corrected Bulk Density (g cm-3) v m fG BD G 1 1 (2-1) Where Gf is the fraction of gravel in the soil sample, on a mass basis; BDm is the uncorrected bulk density in g cm-3; Gv is the volume of gravel in the sample, e xpressed as a fraction of the total volume and calculated as follow: 65 2f mG BD 2.65 is the density of solid particle s in the soil expressed in g cm-3. Plant sampling and growth measurements Detailed measurements of leaf area, plant he ight, dates of tasseling, anthesis, and silking were made throughout the growing season. Aboveground biomass samples were taken four times during the season, 17 days after planting (dap), 31 dap, 52 dap (anthe sis), and 108 dap (final harvest). After emergence 6 plants were randomly sel ected and tagged within three rows on each individual plot. Length and width of each expandi ng leaf were measured at 7-day intervals until

PAGE 34

34 maximum values were reached or 50% leaf se nescence observed (Colomb et al., 2000). At any time during the season, total area of leaves produ ced by the plant was computed as the sum of individual leaves areas. The area of a single growing leaf was calc ulated as the product of length and width multiplied by 0.75 (Colomb et al., 2000). Visible leaf numbers, plant height and phenology events (tasseling, anthesis and silking) were recorded from emergence to silking for the six tagged plants on each plot. Plant height was taken from the base of the plant to the ti p of the most recent leaf. Tasseling, anthesis, and silking date s were established as the date when 50% (three out of the six) of the tagged plants tassele d (panicle visible, tasseling), s howed some pollens or anthers (anthesis), or showed some silks (silking). The samples taken for aboveground biomass determination consisted of randomly prelocated 12 plants from two continuous rows corresponding to a sampling area of 1.92 m2. The dry weights of each plant part, stem plus petiol e, leaves, fruit and grains, were determined on each sample by oven drying at 60 degrees C for 48 hours or until constant weight was reached. Biomass accumulation from week six to maturity was so high that for ea se of handling, only fresh weights were determined on the 12 plants A subsample of 6 plants (17 dap, 31 dap and anthesis) and 5 plants (final harvest) was us ed for dry weight measurements of each plant component. The samples were analyzed at the Univ ersity of Ghana for total N, P and K content in each plant part. An area of 6 m x 1.6 m = 9.6 m2 was used for final harvest on each plot, 108 days after planting (September 12th). The two innermost rows harvested were bordered by two rows on each side. Grain yield, total aboveground biom ass, stover biomass, grain number per m2, unit grain weight, and N, P and K content of grain and stalk were determined.

PAGE 35

35 Experiment in Wa, Ghana The experiment in Wa measured the effect of combinations of nitrogen and phosphorus fertilizers on the growth and development of the maize cultivar Obatanpa (Table 2-1). Site and experiment set up The experiment was carried out in 2004 at the Savannah Agricultural Research Institute (SARI)s experimental station in Wa, Ghana (10o3 N, 2o30 W, altitude 320 m above sea level) by Naab (2005). The average annual rainfall is 1100 mm falling mainly between April and September (Figure 2-3). The mean annual temperature is 27 oC. The experimental field was cleared of native vegetation that was plowed in. The field was harrowed and laid out in a Randomized Comple te Block Design with four replications. Treatment plots measured 6 m x 8 m. The factors tested were: nitr ogen at three levels, 0, 60 and 120 kg N ha-1 (N0, N60, and N120 respectively); and P al so at three levels, 0, 60 and 90 kg P2O5 ha-1 (P0, P60, and P120 respectively). Initial P cont ent of the soil was measured on soil samples taken at planting. Maize was sown on June 17th at a spacing of 70 cm x 40 cm. A pre-emergence herbicide (Roundup or glyphosphate) was applied a few days after sowing. All of the phosphorus was broa dcast as Single Super Phospha te and incorporated by hand hoeing to a depth of 5 cm in all treatments, two days before sowing. Nitrogen was split-applied as urea at the bottom of 5-cm holes near th e maize stands, 2 and 6 weeks after planting. Field and laboratory measurements Phenological observations on the nu mber of days to emergence, tasseling, silking, blister stage, milk stage, dough stage, dent stage and physiological maturity were made on the middle four rows of each plot. The dates were established as corresponding to the time when 50% of the four rows sampled on each plot reache d the different stages of interest.

PAGE 36

36 Plant samples were taken five times random ly on the plots during the growing season. Each sample was obtained from a 0.8 m x 1.7 m = 1.12 m2 area (two rows) that yielded about 8 plants. A sub-sample of 2-3 plants was taken from each sample for an oven-drying dry matter determination at 70oC for 48 hours. Leaf area index was directly measured on randomly selected plants using a Delta-T SunScan Canopy Analyzer. Maturity total biomass and number of cobs were determined on the four middle rows (12 m2) used for phenological observations. Sub-sample s were taken for dry matter estimation of stover, grain yield and components, a nd 100-seed weight. Statistical Analysis Statistical analysis of the data obtained from the experiments in Kpeve and Wa was conducted using the SAS (SAS, 2002). A regression analysis was used to analyze soil moisture data collected at Kpeve and analysis of va riance was used to summarize and understand the growth and development data from both experiments. Regression analysis (soil moisture) A regression analysis was carried out using the SAS software (SAS, 2002) to verify how well a linear regression model can be fitted to the dataset composed of the 67 pairs of TDR readings and soil moisture values obtained fr om the gravimetric samples. The purpose of the regression analysis was to derive an equation that can be used to convert TDR readings into volumetric water content of the soil. Tests of slope and intercept were carried out and a regression equation was set up that allowed the estimation of volumetric soil moisture values from specific TDR readings.

PAGE 37

37 Analysis of variance at individual time points Analysis of Variance (ANOVA) F-te st was used to test the effect of P fertilizer application on crop phenology, grain yield, aboveground biomass, height, green leaf area, and soil moisture readings using TDR. This analysis was carried out at specific time points when the data were obtained. Hypotheses that were tested are described here. Crop phenology. Inadequate supply of P has been reported to delay silking in maize resulting in an increase in anth esis to silking interval. It was hypothesized that increasing P fertilizer levels will result in early tasseling, anthesis and silk ing. These events should be delayed in no and low P treatments. Green Leaf Area. Maize grown on low P soils has been reported to have access to a limited amount of Photosynthetically Absorbed Radiation (PAR), whic h reduces the area of expanding leaves (Pellerin et al., 2000). The green l eaf area in the Kpeve experiment and the leaf area index in the Wa experiment were hypothesized to increase with higher levels of P fertilizer. Aboveground biomass. The accumulation of aboveground biomass is directly related to the amount of PAR intercepted by the crop canopy and should be affected by P deficiency in the same way as green leaf area or leaf area index. Grain yield. An increase of anthesis to silking interv al or a delayed silking will result in a low grain number per square meter and a low gr ain yield (Plenet et al ., 2000b). Growth deficit due to insufficient biomass accumulation can also affect negatively grain formation. Height. Lack of adequate biomass accumulation and energy for physiological processes can result in stunted plants. Highe r values of plant height are e xpected in P fertilized plots. Aboveground biomass, height, green leaf area, leaf area index and soil moisture data were collected on the same plots, plants, and locations on the pl ots over time and were considered repeated measurements. Preliminary anal yses of these data were first carried out at

PAGE 38

38 individual time points. The indivi dual time point analysis was used to examine treatment effects at specific sampling dates only. However, thes e analyses were one way ANOVAs considering each time point separately, as independent from each other, and did not make comparisons among different sampling dates. Analysis of variance considering the effect of time on the repeated measurements The individual time point analysis was exte nded using a repeated measures technique to account for the effects of time on the response va riables taken in sequenc e over time during the growing season. More information can be derive d from repeated measures than revealed by individual time point analysis ANOVAs: comparis ons of treatments averaged over time and comparisons of times within a tr eatment are also informative. When measurements (of height e.g.) are re peated on the same subject (e.g. plant) at specific time intervals (e.g. every 2 weeks, during the growth of the plant), the data are generally viewed as coming from a factorial experiment with treatments and time as the factors, and analyzed as if they came from a split-plot design because most statistical packages do not provide users with the capability of accounting properly for the eff ects of time. In this example, the plant would be considered as the whole-plot unit, and plants at specific times as the sub-plot unit. This method is known as the split-plo t in time approach to analyzing repeated measurements (Littell et al., 1998). The assumptions supporting this split-plot in time approach are that variances of measurements taken at diffe rent times are equal and that pairs of measures coming from the same plant are equally correla ted. This means that the correlation pattern among the measurements taken on the same plant is not affected by time. The split-plot in time analysis would have been optimal if the assumpti ons could be fully met in all circumstances. The peculiar property of repeating the measurement on the same plant means that sets of data from the same plant, though taken at different time points, are not independent. They include a

PAGE 39

39 covariance structure resulting from differences be tween plants (between plants variation) and differences between times on the same plant (with in plant variation). The covariance structure refers to two things: 1) variances in the data co llected on the same plant at individual time points and 2) correlation between measurements taken on the same plant at different times. Littell et al. (1998) underlined the two as pects that are important to the corr elation. First, two measures taken on the plant are correlated simply because they sh are common contributions from the same plant. Second, measures on the same plant close in time are often more correlated than measures far apart in time. This covariance structure is not capture d by the common ANOVA implemented in SAS with the general linea r model PROC GLM. In order to model the covarian ce structure related to the eff ect of time in this study, we used the PROC MIXED procedure now availabl e in SAS since 1992 (Littell et al., 1998). For comparison purposes we present results of the split-plot in time method (also called univariate ANOVA) and two covariance structures summarized with their specifications in Table 2-3. The Akaike Information Criterion (AIC) was used as goodness of fit criterion to select the appropriate covariance structure for this st udy. The AIC is presented with the SAS output when PROC MIXED is run. The smaller the va lue of AIC, the better the structure. The repeated measures analysis technique was applied only to green leaf area and height measurements taken from day 17 to day 52 afte r planting. Biomass and soil moisture were not analyzed using this technique because their measur ements were taken at variable time intervals. Results and Discussion Response of maize biomass and grain yield to P fertilizer was not observed at Kpeve. At Wa, the plant response was measured not only on biomass and grain yi eld but also on phenology.

PAGE 40

40 The analysis of the soil moisture data at Kpev e yielded an equation for prediction volumetric soil moisture from TDR readings. Calibration of the TDR Meter The soil moisture readings taken with the TD R meter appeared to be linearly related to the gravimetric sampling moisture determinations (Figure 2-4). There is sufficient evidence to suggest that the relationship between the two methods of soil moisture determination is linear: an analys is of variance of the simple linear regression model was highly significant and the mean squared error was very low (Table 2-4). Seventy three percent of the variation in the meter readings was accounted for by the gravimetric samplings (coefficient of determination R2 = 0.73). The regression equatio n model relating the volumetric soil moisture to the TDR reading is: tdr volumetric 72 0 90 4 (2-2) Where volumetric is the volumetric soil moisture m easured using gravimetric sampling; tdr is the volumetric soil moistu re read by the TDR meter; The coefficient 0.72 (slope in the regression equation) is an estim ate of the rate of increase in gravimetric soil moisture for each unit increase in TDR readings. Gravimetric soil moisture increases by 0.72% fo r each 1% reading by the TDR. This equation can serve the purpose of predic ting the volumetric soil moisture using the gravimetric method (considered as the true meas urement) from any single or population of TDR readings. Tests of the slope and the intercept in this equation (H0: slope = intercept = 0) lead to highly significant p values for rejecting H0 (Table 2-5). The regression line plotted on Figure 2-4 would be parallel to the 1:1 line ideally, but it is slightly more horizontal, thus crossing the perfect agreement line. This illustrates a tendency of the TDR meter to overestimate soil moisture at high soil moisture status. A possible explanation is that the meter would continue to read high soil moisture values as long as the rods are inserted into the soil with a steady, non wi ggling downward pressure even if a high percentage of gravel

PAGE 41

41 is present in the profile. The gr avel content in the pr ofile (0-30 cm) as meas ured for correction of the soil bulk density was between 25 and 45%. Gravel (soil solid particles with size greater than 2 mm) cannot hold water and would reduce the av ailable water for the plant when they are present in relatively high quanti ties. The corrected bulk density used in this experiment to control the effect of the presence of stones on the soil mo isture status helped to obtain lower values of soil moisture using the gravimetric sampling method. The equation established from this regression analysis is only valid for th e type of soil used in the experiment. Crop Response Results at Kpeve Usin g Individual Time Points Analysis The individual time point an alysis revealed no significan t difference in phenology and growth at Kpeve except for the he ight (during the mid-season). Phenology The expected trend of P effect on tasseling, anthesis and silking was not observed. There was no consistent trend depicti ng the phenological response of the plant to P (Figure 2-5). The data collected in this experi ment did not provide enough eviden ce to suggest an effect of P fertilizer on the phenology of maize (Table 26). On average, tasseling and anthesis dates differed among the P treatments by only one day. At silking however, the treatment receiving 80 kg P ha-1 was delayed by 4-6 days compared to the other treatments. The anthesis to silking interval (ASI) was 9 days on average and higher for the 80P treatment (13 days). Grain yield and yield components Both grain yield and stover we ight were not affected by the P levels. Grain yield of about 3000 kg ha-1 was attained in all treatments a nd the average stover yield was 6500 kg ha-1 (Figure 2-6). The grain and stover yields were stable but the grain yiel d was more variable than the stover yield (overall coefficient of variation of 27% for grain yield and 15% for stover yield). This lack of response to P resulted in no statistical significance (Table 2-7).

PAGE 42

42 The grain yield components were also stable between treatments (Figure 2-7). The average unit grain weight was 0.24 g and the average grain number per m2 was 1500. Aboveground biomass Differences between P treatments were not significant at the individual time points analyzed, but the p-values decreased consistent ly over time (Table 2-8). The treatment mean squares increased with time co rresponding to increased biomass accumulation with plant growth. The aboveground biomass as a combination of stove r and grain yield, likewise did not respond to the P fertilizer. The coefficient of variation between tr eatments varied from 8 to 27% at 17 dap, 7 to 22% at 31 dap, 9 to 15% at 52 dap, and 10 to 19% at 108 dap. However, this variability in biomass and standard deviations (Figure 2-8) did not result in any statistical significance. Plant height Significant differences in plant height were observed mostly during mid-season (Pr = 0.03 at 31 dap, and Pr = 0.01 at 45 dap, Table 2-9). A l east significant differenc e discrimination test showed that the treatment receiving 30 kg P ha-1 produced the tallest pl ants, not only at 31, 38 and 45 dap but also throughout the season (Table 2-10 and Figure 2-9). A maximum height of 250 cm was reached after anthesis. Green leaf area No statistical significant difference was gene rally observed among the treatments (Table 211) at the 0.05 level. At an thesis however, the treatme nts receiving 10 and 30 kg P ha-1 produced the highest green leaf area (6000 cm2 per plant or a leaf area index of 3.75 using a plant population of 6.25 plants m-2) (Figure 2-10) and were statistica lly different from the treatments receiving 0 and 80 kg P ha-1 at the threshold of = 0.07.

PAGE 43

43 Soil moisture The mean differences in soil moisture across the 4 treatment plots as shown on Figure 2-11 were not significant until after the drought spell (i.e. after 58 dap) when the soil started to be rewetted. There were significant soil moisture variations between blocks at the commencement of the trial, which justified blocking (Table 2-12). These interblock moisture differences disappeared however, from 48 dap onwards, at the start of the dr oughtspell and were not detected again until final harvest (Figure 211). It is noteworthy that when the soil moisture started to go up again, the significance probabilities for differe nces between blocks ma intained a decreasing trend. Crop Response Results at Kpeve Using Repeated Measures Analysis Techniques The analysis revealed that the autoregressi ve structure was suitable for the datasets analyzed. Days after planting had a significant effect on growth but interacted significantly only with height. Averaged over time, height was th e only measured variable that was significantly affected by P. Selection of a correlation structure using the AIC The values of AIC were consistently smaller fo r the autoregressive structure regardless of the variable of concern (Table 2-13). This sta tistic essentially confirmed that the correlation between pairs of height and gr een leaf area measurements take n on the same plant and at the same location on the field decrea sed with the age of the crop. Fo r example, the autoregressive structure means that measurements of heights ta ken at days 17 and 24 after planting are more correlated than heights obtained at 17 and 52 da ys after planting on the same plant and at the same location on the field. Thus, the autoregressi ve structure was used in this application.

PAGE 44

44 Effect of time on repeated measurements of crop response variables Days after planting had a highly significant effect on all the repeated measurements regardless of the covariance structure. (Table 2-14). For plant height for instance, this means that the height values reached by the plant averaged over the four treatments were statistically different for days 17, 24, 31, 38, 45, and 52 after plan ting. This is expected because of the plant growth and development processes that notably increased the height between the measurement times. Phosphorus treatments by time interact ions effects on repeated measurements The interactions involving time and P treatmen ts were not significant for green leaf area regardless of the structure. The green leaf ar ea curves for treatment 10P and 30P crossed each other at dap 45 but generally the shapes of th e curves were essentially the same for each treatment (Figure 2-10). There was not enough eviden ce to suggest that the change over time in the responses of maize gr een leaf area was affected by P application. The interaction between day after planting and P treatment was significa nt for maize height based on the autoregressive structure (Pr = 0.0354 Table 2-14). This st atistical significance suggested that the height response curves that could be derived from Figure 2-9 were not the same. Differences between these responses curves came from the quantity of P applied in each treatment. Effect on repeated measurements of pho sphorus treatments averaged over time The effect of P treatments on green leaf ar ea averaged over sampling dates 17 through 52 days after planting was not significant (Pr = 0.3, Table 2-14) based on the autoregressive structure. This means that the overall effect of P treatments on maize green leaf area as tested in this study was not important at = 0.05. This finding is a confir mation of the results obtained

PAGE 45

45 when the ANOVA was performed at individual ti me points: no statisti cal significance at = 0.05 was found at all sampling dates (Table 2-11). Significant effects of P treatments on maize height averaged over sampling dates 17 through 52 days after planting were found using th e autoregressive structure (Pr = 0.0228, Table 2-14). This suggested that plant height measurem ents taken at weekly intervals over the period 17 to 52 days after planting were significantly different between P treatments. This general conclusion on maize height response to P applicat ions in this experiment was expected because the individual time point analysis showed signi ficant differences between the P treatments at days 31 and 45 after planting and low probabilities for these differences at days 17, 24, 38 and 52 after planting (Table 2-9). The importance of the choice of an appropriate correlation structure for the analysis of repeated measurements is highlighted by the contradictory results obtained with the univariate ANOVA and the compound symmetr y structure (Table 2-14). For example, significance probability values produced by the univariate A NOVA were generally low for the effects of P (Table 2-14), suggesting strong evidence of differe nces in green leaf area between the four P treatments. We already knew that this was not true because most of the p-values obtained from analyses at individual time poi nts were relatively high (Tables 2-9 and 2-11). The choice of a different correlation structure w ould lead to different conclusions about the effect of P on the green leaf area and height of maize av eraged over six measurement dates. Discussion of Result s Obtained at Kpeve The lack of response of phenology, biomass, yield and yield components of maize to P fertilizer in the Kpeve experiment was because P did not limit plant growth and development in the experiment. Since all other ma jor nutrients and water (at least until anthesis) were supplied in sufficient amounts, it is reasonable to suggest that the plants had access to and were able to take

PAGE 46

46 up P from an adequate supply of indigenous P throughout the growing season. Although the soil P level (Bray 1) was apparently low (Tables 215 and 2-16), there are at least five problems associated with relying on the classification in Table 2-16 alone to draw conclusions about the P status of the soil: Problem 1: the Bray-1 P method does not measure P available for plant uptake but only that amount of P that would probably corre late with plant growth (Johnston, 2000); Problem 2: the P level in the soil top 20 cm used in this experiment is close to the sufficiency level of 16 ppm as defined by Shapir o et al. (2003) (Table 2-16). Since P exists in the soil in many forms that exchange P be tween each other, it is not clear how P would behave in the soil at the bounda ry between sufficiency and deficiency. Other studies (for example Adeoye and Agbola (1985)) found a criti cal range of Bray 1 P availability of 1016 ppm for tropical soils. Measured Bray 1 P in or below this range would be considered low; Problem 3: Other forms of P that were not measured by the Bray-1 method could have become soluble. The inorganic active P (re presented by NaOH-Pi, Table 2-17) is an important source of P that can become dire ctly soluble during th e growing season. The organic carbon content of the soil that was near ly 2% in the topsoil could have contributed P through mineralization especially under tr opical conditions (Osiname et al., 2000). Current thinking envisages the different forms of P in the soil as ex isting in equilibrium. Field preparation disturbs the equilibrium and subsequent decomposition of soil organic may release additional P. Also, plant uptake can displace this equilibrium in such a way that replenishment of solubl e P from other forms of P is continuous. Johnston (2000) showed that in addition to providing P through mineralization, soil organic matter provides sites with low bonding energy for P; Problem 4: The experimental field wa s left to natural bush fallow for about 2 years and P fertilizer at 7.5 kg P2O5 ha-1 was applied to maize grown on th e field in 2004 prior to the 2year fallow. Plant residues from the two-y ear natural bush fallow that preceded the experiment was mixed with the soil during pl owing, seven days before maize planting. The 2004 P fertilizer and the decomposition of the maize and fallow residues could contribute significant amount to soil P build up that coul d have been made available when the soil was brought out of the fallow for the experiment; Problem 5: The soil had an ideal pH (6.5) for P tr ansformations and availability (Table 215). The delay in silking was due to the drought sp ell that occurred right after anthesis and lasted until after silking (Figure 2-11). Studies have shown that water st ress delays silking in maize and results in an increase in the anthesis to silking interv al (Balanos and Edmeades, 1993).

PAGE 47

47 Similar findings on the positive effects of P on plant height were reported by Khan et al., 2005. In the Kpeve experiment, the significant differe nces in heights between the treatments did not result, however, in differences in bioma ss production. The LSD for height differences at 31, 38 and 45 days after planting were respectivel y 6, 11, and 16 cm. Compared to the respective height ranges of 75-84, 124-138, a nd 188-215 cm (Table 2-9), thos e height differences (LSD) corresponded to a part of the upper canopy that did not contribute much to the weight of the plant and was mostly leaf blade. The significance of the effect of days afte r planting on measurements repeated over time has an agronomic meaning. It corresponds to an active period of growth for plant height and green leaf area as shown on Figures 2-9 and 2-10. Results and Discussion for the Wa Experiment The soil in Wa contained very little availa ble P and organic matter (Table 2-18). Maize responded to nitrogen and phosphorus to the ex pected extent, confir ming past and current findings by other researchers (Singh and al., 1999; Colomb et al, 2000; Khan et al., 2005). The results obtained in the Wa experiment are fu lly reported and discussed in Naab (2005). A summary of the responses obs erved are presented here. Nitrogen and phosphorus had similar effects on the phenological development of maize. Tasseling was not affected by nutrient manageme nt. On the contrary, silking was delayed by about 1-3 days in treatments that did not receiv e nitrogen or phosphorus (T able 2-19). Statistical differences at silking were observed only between the no and medium or high nutrient application. Differences were not f ound between the medium (60 kg [N] ha-1 and 60 kg [P2O5] ha-1) and high (120 kg [N] ha-1 and 90 kg [P2O5] ha-1) nutrient applications. The overall effect of the nutrient deficiency on physiological maturity of the crop was small and not significant (Table 2-19). Grain filling duration was shortened in th e no nutrient treatments in such a way that

PAGE 48

48 physiological maturity did not differ among treatme nts. Similar results were obtained in Hawaii by Singh et al. (1999). Significant leaf area index (LAI) differences due to nitrogen and phos phorus applications were observed throughout the season (Table 2-20). The effect of P on LAI disappeared at 90 dap and thereafter. These LAI differences were obser ved only between the no nutrient treatments and the 60 kg [N] or [P] ha-1 treatments, and LAI did not increase beyond the level of 60 kg [N] or [P] ha-1. The maximum LAI advantage over P0, 50% in P60, was observed at 40 dap. Plenet et al. (2000a) found an LAI reduction of the same ma gnitude (60%) between th e 7and 14-visible leaves in a P response experiment in France. Aboveground biomass responded consistently both to nitrogen and phosphorus fertilization at all sampling dates (28, 46, 61, 81, and 125 days after planting). Nitr ogen applied at 120 kg ha-1 did not result in any significant biomass accumulation over the 60 kg ha-1 N level at days 28, 46, and 61 dap. At 81 and 125 dap however, the difference between N60 and N120 were amplified and were significant (Table 2-21). The N60 trea tment resulted in biomass differences of 75 to 4500 kg ha-1 over N0, which represented 19 and 67% of the biomass obtained in N60. Response to P was less drastic but also significant. Differences were no t found between P60 and P90 at all sampling dates. The biomass gain of P60 over P0 ranged from 130 to 3300 kg ha-1 corresponding to 31 to 56% of the biomass measured in P60. The highest biomass and LAI gains (over N0 or P0) were obtained at the same sampling peri ods (40-46 dap for nitrogen and 61-68 dap for phosphorus). This could be a confirmation of fi ndings by Plenet et al. (2000b) according to which poor biomass accumulation in P defici ent plants was mainly due to reduced photosynthetically active radiation absorbed by the canopy caused by reduced leaf area.

PAGE 49

49 Fruit weight increased significantly between N0, N60, and N120. For phosphorus, differences were found only between P0 and P60, and P0 and P90. No significant differences in fruit weight were revealed between P60 and P90 (Table 2-22). Fruit growth was affected more severely by nitrogen than phosphorus (Table 2-22). For example, grain yi eld gains were 77% in N60 over N0 and only 42% in P60 over P0. The e ffect of P applications on seed weight was relatively small compared to the effect of nitrogen (Table 2-22). These ultimate effects on grain yield and yield components were probably associ ated with the consequences of nitrogen and phosphorus stress on photosynt hesis (Singh et al., 1999). Conclusions The study at Kpeve did not result in the expected response to P fertilizer applications. No significant differences in plant phenology, abovegr ound biomass, green leaf area and grain yield were found between fertilized and un fertilized treatments. Significan t differences in plant height observed at 31, 38, and 45 days after planting we re not reflected in biomass accumulation or grain yield. Although the initial av ailable P (Bray 1) was relativ ely low in all layers, other important P sources such as chemical contribu tions of organic matter not accounted for by the Bray 1 extraction could have been responsible for high indigenous P supply in the soil. At Wa, soil P levels were sufficiently low to cause a P fertilizer re sponse in the crop. Delay in silking of about 1 day was obs erved in the treatment that di d not receive any P input. The delay in silking was 2 days in th e no nitrogen treatments and 1 da y in the no P treatments. Leaf area index and aboveground biomass were reduced in no nitrogen and no phosphorus treatments throughout the season. The highest reduction in leaf area index and biomass occurred at the same time period, which strengthens the idea that poor biomass accumulation in P deficient conditions is associated with reduced photosynthetically absorbed radiation by the plant, which is a

PAGE 50

50 consequence of reduced leaf area. The reduction in grain yield could have been a result of nutrient stress on photosynthesis. The contrasting results obtained at the two sites in terms of response to P would be useful in testing the ability of computer simulation mode ls of soils and plants to capture and mimic the effect of variability in P manage ment on crops. In Chapter 4, this attempt is made using the soilplant phosphorus model in the Decision S upport System for Agrotechnology Transfer.

PAGE 51

51 Table 2-1. Growth and development genetic coeffici ents for the Obatanpa cultivar used at both sites, Kpeve and Wa (Ghana) Definition DSSAT ID Obatanpa Degree-days (base 8oC) from emergence to end of juvenile phase P1 300 Photoperiod sensitivity P2 0.00 Degree-days (base 8oC) from silking to physiological maturity P5 830 Potential kernel number (/plant) G2 900 Potential kernel growth rate (mg/day) G3 6.50 Phyllochron interval (degree-days) PHINT 38.90 Table 2-2. Summary of fertilizer application met hods used in the experiment in Kpeve, Ghana Days after planting Ammonium Sulfate Triple Superphosphate Potassium Nitrate 0 Broadcast without incorporation Side placement, bottom of hole Broadcast without incorporation 13 Side placement, without incorporation Side placement, bottom of hole Side placement, without incorporation 30 Side placement, without incorporation No Application No application 44 Side placement, without incorporation No Application No application Table 2-3. Specifications of two different covarian ce structures used for modeling the effect of time on repeated measures in PR OC MIXED for the Kpeve dataset Covariance structure Specifications Compound Symmetry 1. Measures at all times have the same variance; 2. Pairs of measures from the same subject have the same correlation; Autoregressive 1. Measures at all times have the same variance; 2. Correlations between pairs of m easures from the same subject decrease as the time lag be tween measures increases; Source: Adapted from Littell et al. (1998). Table 2-4. Analysis of Variance for simple linear regression between soil moisture measurements using TDR and gravimetric methods at Kpeve, Ghana Source of variation Degree of Freedom Sum of Squares Mean Square F Value Pr > F Model 1 805.45874 805.45874 178.19 <.0001 Error 65 293.82273 4.52035 Total 66

PAGE 52

52 Table 2-5. Test of parameter estimates used to fit the linear regression model in the Kpeve experiment Variable Degree of Freedom Parameter Estimate Standard Error T Value Pr > | t | Intercept 1 4.90191 0.68030 7.21 <.0001 tdr 1 0.71900 0.05386 13.35 <.0001 Table 2-6. Analysis of variance of phenological even ts, tasseling, anthesis and silking in the Kpeve experiment Source of variation Degree of Freedom Sum of Squares Mean Square F Value Pr > F Tasseling Block 3 0.6875 0.2292 0.11 0.9496 Phosphorus 3 6.1875 2.0625 1.03 0.4254 Anthesis Block 3 1.2500 0.4167 0.88 0.4861 Phosphorus 3 2.2500 0.7500 1.59 0.2594 Silking Block 3 31.1875 10.3958 0.70 0.5737 Phosphorus 3 100.1875 33.3958 2.26 0.1506 Table 2-7. Analysis of variance for grain yield (measured in kg ha-1) at Kpeve, Ghana Source of variation Degree of Freedom Sum of Squares Mean Square F Value Pr > F Block 3 2447042 815681 1.12 0.39 Phosphorus 3 427698 142566 0.20 0.90 Error 9 6557203 728578 Table 2-8. Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time points for crop aboveground biomass measured in kg ha-1 in the Kpeve experiment DAP 17 31 52 108 Block 394 (0.45) 9342 (0.78) 462057 (0.36) 3781140 (0.11) Phosphorus 55 (0.94) 11257 (0.73) 366238 (0.45) 1189617 (0.50) Error 410 25489 381818 1382973 Table 2-9. Summary of results from ANOVA (mean squares (p-values), n = 4) at individual time points for plant height measured in cm in the Kpeve experiment DAP 17 24 31 38 45 52 68 Block 45.82 (0.20) 61.69 (0.46) 90.81 (0.54) 727.11 (0.15) 794.51 (0.40) 775.17 (0.52) 1079.05 (0.45) Phosphorus 54.87 (0.14) 106.03 (0.22) 388.72 (0.03) 978.82 (0.07) 3517.83 (0.01) 1882.73 (0.14) 2401.73 (0.13) Error 29.21 70.63 124.47 397.13 805.88 1020.16 1224.96

PAGE 53

53 Table 2-10. Treatment means at each day for plant height measured in cm, with least significant difference (LSD) in the Kpeve experiment ( = 0.05) DAP Treatment LSD 0P 10P 30P 80P 17 25.91 ab 28.52 a 25.65 ab 25.14 b 3.10 24 42.43 b 44.85 ab 47.40 a 43.81 ab 4.82 31 76.06 b 75.858 b 84.05 a 81.22 ab 6.40 38 123.70 b 128.94 ab 138.15 a 125.75 b 11.43 45 194.80 bc 207.78 ab 214.59 a 187.93 c 16.28 52 235.90 a 250.15 a 249.94 a 233.67 a 18.32 68 240.29 a 258.60 a 256.45 a 240.24 a 20.08 Means with the same letter are not statistically different at = 0.05. Table 2-11. Summary of results from ANOVA (mean squares (p-val ues), n = 4) at individual time points for green leaf area measured in cm2 per plant at Kpeve DAP 17 24 31 38 45 52 68 Block 33194 (0.03) 302147 (0.03) 1014051 (0.08) 2427809 (0.07) 5623424 (0.002) 2411939 (0.12) 5006509 (0.004) Phosphorus 13567 (0.27) 151693 (0.19) 647931 (0.22) 1742106 (0.17) 1022362 (0.40) 3016789 (0.07) 1860147 (0.16) Error 10192 94055 430784 1010465 10408701220628 1052365 Table 2-12. Summary of results from ANOVA (mean squares (p-val ues), n = 4) at individual time points for soil moisture readings (in %) using TDR at Kpeve DAP 30 37 45 48 53 55 58 64 69 74 Block 19.62 (0.07) 11.01 (0.04) 12.25 (0.02) 3.02 (0.39) 1.15 (0.76) 1.67 (0.53) 4.90 (0.11) 14.96 (0.37) 34.92 (0.18) 17.08 (0.16) Phosphorus 9.01 (0.34) 4.72 (0.31) 6.72 (0.14) 2.37 (0.49) 2.02 (0.56) 0.49 (0.88) 3.31 (0.25) 39.14 (0.05) 54.51 (0.06) 23.48 (0.07) Error 7.93 3.83 3.51 2.93 2.90 2.23 2.35 14.05 20.94 9.63 Table 2-13. Akaike Information Criterion (AIC) test for two covariance structures in PROC MIXED for repeated measures analysis for the Kpeve experiment Variable AIC value Compound Symmetric Auto regressive + random Height 4909.8 4846.6 G. Leaf Area 9015.5 8865.1

PAGE 54

54 Table 2-14. F -values and significance probabilities us ing univariate ANOVA, and for test of fixed effect using two covariance stru ctures in PROC MIXED for the Kpeve experiment Source of variation df Univariate ANOVA Compound Symmetric AR(1) plus random effect Height P 3 10.63 (< .0001) 1.79 (0.1822) 3.21 (0.0228) DAP 5 2096.7 (< .0001) 2096.7 (< .0001) 1770.88 (< .0001) P x DAP 15 1.88 (0.023) 1.88 (0.0343) 1.77 (0.0354) G. Leaf area P 3 6.80 (0.0002) 1.35 (0.3) 1.35 (0.3) DAP 5 5.03 (< .0001) 888.8 (< .0001) 823.53 (< .0001) P x DAP 15 0.87 (0.6) 0.87 (0.6) 1.15 (0.3066) Table 2-15. Physical and chemical characteristics of the soil at th e experimental site in Kpeve, Ghana Parameter 0-10 cm 10-20 cm 20-30 cm Texture Clay (%) 18 20 18 Silt (%) 28 29 27 Sand (%) 54 51 55 Gravel (%) 40 40 35 Organic Matter Organic carbon, Walkley-Black (%) 1.84 1.8 1.55 Total nitrogen (%) 0.26 0.25 0.22 Phosphorus Total phosphorus (mg/kg) 294 299 229 Bray 1 (mg/kg) 11.69 10.4 7.43 Mehlich 1* (mg/kg) 90.44 46.12 50.19 Exchange Complex Potassium K (cmol/kg) 0.11 0.08 0.06 Calcium Ca (cmol/kg) 7.39 7.31 7.65 Magnesium Mg (cmol/kg) 2.61 2.40 2.38 Acidity pH-H20 6.45 6.56 6.48 *The Mehlich 1 analysis was done by the Wetland Biogeochemisty Laboratory at the University of Florida. All other tests were done in the soil testi ng laboratory at the University of Ghana.

PAGE 55

55 Table 2-16. Classes of phosphorus availability according to the Bray 1 extraction method Soil test, P Bray 1 (ppm) Relative P level 0-5 Very Low 6-15 Low 16-24 Medium 25-30 High > 30 Very High Source: Shapiro et al. (2003). Table 2-17. Characterization of the different forms of soil phosphorus at Kpeve, Ghana. Data are reported in mg/kg P fraction 0-10 cm 10-20 cm 20-30 cm Inorganic KCl Pi 3.3 1.7 3.6 NaOH Pi 56.4 40.8 40.9 HCl Pi 76.4 28.3 27.6 Organic NaOH Po 65.8 65.0 64.7 Residual P 153.6 120.2 140.1 Table 2-18. Physical and chemical characteristics of the soil at the experimental site in Wa, Ghana Parameter 0-20 cm 20-40 cm 40-60 cm 60-90 cm Texture Clay (%) 7.50 14.50 40.90 52.90 Silt (%) 8.30 8.20 10.70 16.90 Sand (%) 84.20 77.30 48.40 30.20 Gravel (%) 4.30 6.40 49.20 80.70 Organic Matter Organic carbon (%) 0.49 0.48 0.51 0.43 Total nitrogen (%) 0.06 0.06 0.04 0.04 Phosphorus Bray 1 (mg/kg) 2.50 Not measured Exchange Complex Potassium K (cmol/kg) 0.06 0.08 0.11 0.13 Sodium Na (cmol/kg) 0.49 0.45 0.52 0.45 Calcium Ca (cmol/kg) 1.54 1.23 1.62 1.86 Magnesium Mg (cmol/kg) 0.32 0.51 0.74 0.86 Acidity pH-H20 6.34 6.25 5.94 6.02

PAGE 56

56 Table 2-19. Main effects of n itrogen and phosphorus on phenological development in maize at Wa, Ghana Treatment Days to Phenological Stage (days) Nitrogen Tasseling Silking Grain fill ing duration Physiological maturity N0 49 57 a 39 96 a N60 48 55 b 40 95 ab N120 48 54 b 41 95 b Phosphorus P0 49 56 a 40 96 a P60 48 55 b 40 95 a P90 48 55 ab 40 95 b Source: Adapted from Naab (2005). Means with the same letter are not sta tistically different at = 0.05 in each column. Table 2-20. Main effects of nitr ogen and phosphorus on leaf area in dices of maize at Wa, Ghana Treatment Days After Planting (days) 28 40 68 81 90 Nitrogen N0 0.60 a 0.52 a 0.57 a 0.83 a 0.54 a N60 0.88 b 1.03 b 1.32 b 1.46 b 1.03 b N120 0.88 b 0.92 b 1.55 b 1.74 c 1.13 b Phosphorus P0 0.52 a 0.51 a 0.83 a 1.01 a 0.75 a P60 0.90 b 1.02 b 1.38 b 1.55 b 1.01 a P90 0.93 b 0.95 b 1.23 b 1.47 b 0.94 a Source: Naab (2005). Means with the same letter are not statistically different at = 0.05 in each column. Table 2-21. Main effects of nitrogen and phos phorus on cumulative aboveground biomass (in kg ha-1) of maize at Wa, Ghana Treatment Days After Planting 28 46 61 81 125 Nitrogen N0 334 a 1088 a 2289 a 2155 a 1740 a N60 410 b 2435 b 6961 b 6473 b 5621 b N120 403 b 2436 b 7811 b 8287 c 7502 c Phosphorus P0 292 a 1069 a 3223 a 3520 a 3222 a P60 426 b 2403 b 6549 b 6816 b 6025 b P90 429 b 2486 b 7289 b 6580 b 5615 b Source: Naab (2005). Means with the same letter are not statistically different at = 0.05 in each column.

PAGE 57

57 Table 2-22. Main effects of n itrogen and phosphorus fruit yield components of maize at Wa, Ghana Treatment Fruit Yield Cob weight (kg ha-1) Grain yield (kg ha-1) 1000-seed weight (g) Nitrogen N0 632 a 479 a 190 a N60 2556 b 2063 b 226 b N120 4042 c 3340 c 250 c Phosphorus P0 1646 a 1320 a 203 a P60 2833 b 2292 b 235 b P90 2750 b 2271 b 228 b Source: Adapted from Naab (2005). Means with the same letter are not sta tistically different at = 0.05 in each column.

PAGE 58

58 0 50 100 150 200 250 300 MarAprMayJunJulAug monthrainfall (mm)0 5 10 15 20 25 30 35 40temperature (o C) Monthly total rainfall in 2006 Monthly total rainfall average (2003-2005) Maximum temperature in 2006 Minimum temperature in 2006 Figure 2-1. Monthly total rainfall in 2006 (mm) with e rror bars corresponding to one standard deviation of rainfall, mont hly total rainfall average fr om 2003 to 2005, and monthly average daily temperature (oC) in 2006 at Kpeve. 2006 data were collected during the experiment and 2003-2005 data taken from Adiku (2006)

PAGE 59

59 Figure 2-2. Sequential fractionation steps used for extracting the different forms of phosphorus from soil samples taken before planting of the Kpeve experiment. Samples analyzed by the Wetland Biogeochemistry Laboratory, University of Florida 0 50 100 150 200 250 300 350 JanFebMarAprMayJunJulAugSepOctNovDec monthrainfall (mm)0 5 10 15 20 25 30 35 40temperature (o C) Rain Maximum temperature Minimum temperature Figure 2-3. Monthly total rainfall (mm) with erro r bars corresponding to one standard deviation of rainfall and monthly average daily temperature (oC) at Wa in 2004. Data from Naab (2005). Initial soil Readily available inorganic P (KCl Pi) 1.0MKClextraction Alkaline extractable organic P (NaOHPo) 0.1 M NaOH extraction Fe/Al bound P (NaOH Pi) Ca/Mg bound P (HCl Pi) 0 5 M H Clextraction Ashing Residual P

PAGE 60

60 Figure 2-4. Simple linear regression of volumetri c soil moisture (%) determined by using Time Domain Reflectrometry and gravimetric sampling at Kpeve 010203040506070 Tasseling Anthesis Silking days 0P 10P 30P 80P Figure 2-5. Phenology of maize as affected by ph osphorus application at Kpeve. Error bars represent standard deviati ons of measurements taken from four replications 0 5 10 15 20 25 30 051015202530 TDR reading (%)volumetric soil moisture (%) 0-20 cm 0-12 cm volumetric = 4.90 + 0.72*tdr

PAGE 61

61 0 2000 4000 6000 8000 10000 0103080 phosphorus level (kg ha-1)grain yield or stover weight (kg ha-1) grain yield stover Figure 2-6. Stover and grain yield of maize as affected by phosphorus fertilizer at Kpeve. Error bars represent one standard deviation of m easurements taken from four replications 0 200 400 600 800 1000 1200 1400 1600 1800 0103080 phosphorus level (kg ha-1)grain number (# m-2)0.00 0.05 0.10 0.15 0.20 0.25 0.30unit grain weight (g grain-1) grain number unit grain weight Figure 2-7. Grain number per m2 and unit grain weight as affect ed by phosphorus fertilizer at Kpeve.

PAGE 62

62 0 2000 4000 6000 8000 10000 12000 173152 (anthesis)108 (final harvest) days after plantingaboveground biomass (kg ha-1) 0P 10P 30P 80P Figure 2-8. Aboveground biomass of maize as affect ed by phosphorus fertilizer application at Kpeve. Error bars represent one standard deviation of m easurements taken from four replications. Figure 2-9. Height of maize as affected by phosphorus fertilizer in the Kpeve experiment 0 50 100 150 200 250 300 17243138455268 days after plantingplant height (cm) 0P 10P 30P 80P anthesis

PAGE 63

63 Figure 2-10. Green Leaf Area of maize in the phosphorus fer tilizer experiment at Kpeve Figure 2-11. Variation in soil moisture measured using Time domain reflectometry in the Kpeve phosphorus experiment. The possible effects of the four phosphorus treatments on the soil moisture are analyzed and reported at individual time poin ts in Table 2-12. 0 2 4 6 8 10 12 14 16 18 20 041013161820242730323740454853555862646974 days after plantingvolumetric Soil Moisture (%) 0P 10P 30P 80P anthesis silking emergence 0 1000 2000 3000 4000 5000 6000 7000 17243138455268 days after plantinggreen Leaf Area (cm2) 0P 10P 30P 80P anthesis

PAGE 64

64 CHAPTER 3 THE SOIL-PLANT PHOSPHORUS MODEL IN DSSAT Introduction Biological, physical and chemical processes affecting phosphorus (P) transformations in soils and plants create dynamic soil P pools th at interact with plan ts in a complex way. Phosphorus is present in the soil in two main fo rms: inorganic and organic. The inorganic forms represent mineral P in the so il solution, P bound to calcium, P retained by iron and aluminum oxides and by clay, and P occluded in iron and aluminum minerals. The organic forms correspond to P in fresh organic residues, P in soil organisms biomass, and P in slowly decomposing soil organic matter. The forms of P in the soil are dependent on many soil properties but the most important are soil pH, or ganic carbon content, th e quantity and type of clay, and the cation exchangeable capacity of th e soil. For example, plant P nutrition is optimal at pH 6.5 because the available form of P for uptake by plants predominates at that pH. Plants take up phosphorus from the soil soluti on that is replenished by the other forms of inorganic P and through mineraliz ation of organic P. Several transformations between the different forms of P occur in the soil, but the co ntribution to the soil solution is very low. The amount of phosphorus available to plants from the soil solution at any one time would seldom exceed about 0.01% of the total phosphorus in most soils (Brady and Weil, 2002). Description of phosphorus proces ses taking place in a soil-pla nt system therefore would deal with i) the different forms present in the soil; ii) the transformations that make insoluble forms soluble for plant uptake; iii) plant uptak e mechanisms; and iv) conditions and mechanisms of P supply to plants. The soil-plant phosphorus model in the D ecision Support System for Agrotechnology Transfer (DSSAT) attempts to consider these aspects to simulate phosphorus transformations in

PAGE 65

65 soils and plants and their effect on crop produc tion. The soil inorganic P module of the model simulates phosphorus transformations between a la bile, active and stable pool. The soil organic P module simulates phosphorus transformations betw een a surface litter, a microbial pool, and a stable pool. The model accounts for the mineraliz ation of organic P to inorganic pools and the immobilization of P to organic pools. In phosphorus deficient soils where organic matter play an important role in the supply of nutrients, simu lation of the release of phosphorus from organic matter accounts for an important source of P. Available phosphorus for uptake by plants is described as being provided by the labile pool w ithin a short distance of plants roots (2 mm). Phosphorus taken up by the plant is partitioned to seeds, shells and vegetative tissues. During the reproductive phase, phosphorus accumu lated in the vegetative tissues can be remobilized and translocated to seeds. Plan t growth is limited by phosphorus between two thresholds that are species-specific optimum and minimum concentrations of P defined at different stages of plant growth. Phosphor us stress factors are computed to reduce photosynthesis, dry matter accumulation and partitioning. The present chapter summari zes procedures used in the simulation of the phosphorus balance in the DSSAT cropping system models. The objectives of the chapte r are to: i) present the phosphorus modeling framework in the D SSAT cropping system model; ii) present a description of the soil and plant phosphorus model in DSSAT; iii) present a sensitivity analysis of the model to selected key phosphorus-related pa rameters. The main question of interest in the sensitivity analysis was: how does the variability of six major phosphorus-r elated factors affect biomass and grain yield as simulated by the model? Soil and Plant Phosphorus Modeling in DSSAT The need for implementing a phosphorus model in DSSAT was already recognized as a limitation of the software at the release of its firs t version (Jones et al., 1998). It was clear that

PAGE 66

66 the integration of a phosphorus component into DSSAT would considerably increase and extend its applicability not only to P deficient envir onments, but also to low input cropping systems receiving significant amounts of phosphorus from the decomposition of organic matter. The development of a phosphorus model in DSSAT presen ted at least two challe nges: i) scientists need to improve their understanding of phosphorus behavior in soils and plants because of the complexity of P chemistry in soils and its intera ction with other major nut rients that limit plant growth; ii) the design of the initi al version of DSSAT was not su itable for maintenance of the software as new models were included and modifications were made. The software was a collection of independent models operating in the same framework to inte grate information about soil, climate, crop and management. The models in the decision support system were operating in their own original programming settings. For many years the models in DSSAT have been used to simulate potential, water and nitrogen limited production only, even in areas where phosphorus deficiency is widespread. Advances in modular programming techniques have enabled DSSA T developers to completely redesign the software (Jones et al., 2003); the cr op models can now opera te using the same soil module, the same climate module, and the same management module. Other modules could therefore be more easily included and connect ed to existing crop models with minimal modification to the system. The new programming technique also facilit ates documentation and updating of the individual modules that were deve loped by specialists from different disciplines working together as a team. The first version of a soil-plant phosphorus model linked to the DSSAT cropping system model (CSM) was developed and evaluated by Da roub et al. (2003) for calcareous and highly weathered soils. The model could have been pl ugged into any crop models within the DSSAT

PAGE 67

67 CSM to simulate phosphorus-limited production, but was tested only for maize and soybean. Although the model predicted grain yield and plan t uptake with a reasonabl e degree of accuracy (Daroub et al., 2003), two important modifications we re necessary to make it satisfy the concern of extending the appli cability of the CSM to low inputs cr opping systems and to users having access to limited soil information: i) linki ng the model to the DSSAT-CENTURY model (Gijsman et al., 2002a) for simulation of organic P transformations; and ii) integrating a soil expert system that can allow the estimation of initial amounts of inorganic and organic soil phosphorus pools as influenced by major soil ca tegories and using different methods of P extraction, pH and organic carbon. The initial P model developed by Daroub et al. was thus updated with these two major modifications and is described here. Description of the Soil Phosphorus Model The soil phosphorus model is comprised of th e soil inorganic module and the soil organic module. The two modules are linked in a way th at soil phosphorus mineralized from organic matter is transferred to the inorganic modul e and phosphorus immobilized in the inorganic module is moved to the organic module. The init ial sizes of the inorganic and organic pools can be derived directly from P frac tionation data or indirectly from other P extraction methods. The initialization procedures, developed based on st udies by Jones (1984), Sing h (1985) and Sharpley (1984, 1989), are described in Appendix C. Soil Inorganic Module The soil inorganic module describes transforma tions that occur between the inorganic P pools to make P available for plant uptake.

PAGE 68

68 Inorganic phosphorus pools The soil inorganic module disti nguishes three pools: labile (PiLabile), active (PiActive) and stable (PiStable). The three pools exist in two soil zones: the zone that is in direct contact with roots (within 2 mm) and the zo ne that is not in direct cont act with roots (Figure 3-1). The labile inorganic P pool include s the P in the soil solution. Be cause roots do not develop at planting, the initial soil volume in direct contact with roots is assu med to be zero at the beginning of the simulation and the total am ount of inorganic phosphorus availa ble at that time is assigned to the no-root zone. If transplant s are used, an initial soil root volume is estimated to initialize the simulation. As the roots develop, a proportional mass of P is added to the root zone pools and subtracted from the no-root pools in proportion to the soil volume adjacent to the new root growth. Phosphorus transformations be tween the inorganic pools Per day phosphorus transformations between the three pools occu r according to the following first-order relationships: pool active the to pool labile the from flow P PiLabile KLA (3-1) pool labile the to pool active the from flow P PiActive KAL (3-2) pool stable the to pool active the from flow P PiActive KAS (3-3) pool active the to pool stable the from flow P PiStable KSA (3-4) Where the P flows between the differe nt pools are in units of mg [P] kg-1 [Soil] day-1. The coefficients KLA, KAL, KAS, and KSA are the respective transf ormation rate constants, in unit of day-1. The values of KLA, KAL and KAS depend on the phosphorus availability index (Table 3-1) (Jones et al., 2005a; Jones et al., 1984a; Sharpley et al., 1984, 1989). KLA, KAL and KAS are calculated as follow: 5 01 03 0 x PAvailInde x PAvailInde KLA (3-5) x PAvailInde K KLA AL 3 (3-6) 05 7 77 1 x PAvailInde ASe K (3-7) 0001 0SAK Where PAvailIndex = P availability index defined in Table 3-1.

PAGE 69

69KAS = rate constant for transformation from active P to stable P. KSA = rate constant for transformation from stable P to active P. Mineralization and immobilization of phosphorus from the organic matter are handled in the soil organic module. A net mineralized P amount is calculated and is added proportionally to the root and no-root zones when its value is positiv e and subtracted from the labile P pool if its value is negative. Phosphorus uptake calculated in the plant model is subtracted directly from the PiLabile pool in the root zone. P fertilizer applied is direc tly added to the labile and activ e pools. The amount of applied P that enters those pools depends on the soil cat egory and the application method. Fertilizer applied in bands or hills is used more efficien tly by the plant. When th ese application methods are used, all of the P is applied directly into the root soil volume. Wh en broadcast or other application methods are used, the fertilizer is pr oportioned to the root and no-root zones to the depth of incorporation. A P fertilizer availability function is com puted using soil composition (Table 3-2) based on studies by Jones (1984) and Shar pley et al. (1984, 1989). The P fe rtilizer availability index is expressed as a fraction of fertilizer which enters the labile pool. The remaining P fertilizer is added to the active pool. Phosphorus availability for uptake by plants The available phosphorus for plant uptake is the soluble phosphorus that is in the root zone. A fraction of the root labile Pi is assume d to be soluble, and that fraction defines the portion of the root labile Pi in the soil solution that is sensed and is available for extraction by plant roots on a daily basis. leRoots SoilPiLabi FracPSol l SoilPiAvai (3-8) Where SoilPiAvail is the plant available inorganic phosphorus;

PAGE 70

70 FracPSol is the fraction of root labile Pi that is soluble (i.e. enters the soil solution). SoilPiLabileRoots is the inorganic lab ile P that is in the root zone. Soil Organic Module The soil organic module describes transformati ons of organic materi als that eventually contribute P to or extract P from the inorganic P pools through mi neralization or immobilization. Organic phosphorus pools The soil organic P module has of four litte r pools and two soil organic matter (SOM) pools (Gijsman et al., 2002a): Organic residues added to the surface of the so il become either surface litter or soil litter. The residue materials themselves are divide d into easily decomposable or metabolic materials (i.e. sugars and proteins) and recalcit rant or structural ma terials (i.e. lignin and other fibers) (Figure 3-1). As a consequence, four litter pools can be defined: a surface structural litter pool, a surface metabolic litte r pool, a soil structur al litter pool, and a soil metabolic litter pool. Microbial activity creat es two active pools, one on the surface and another in the soil (SOM1 pools). A stable pool exists in the soil only (SOM23) and is the combination of the slow SOM (SOM2) and passive SOM (SOM3) pool s for carbon (Gijsman et al., 2002a). The soil SOM1 and SOM23 are the main pool s that control inorganic phosphorus dynamics in the soil. The surface litter pools ge nerate flows of carbon a nd nutrients into the surface SOM1 and the soil litter po ols through tillage (Figure 3-1), and the surface and soil litter pools eventually become pa rt of the soil SOM1 pool. Phosphorus movements between the differe nt organic P pools follow carbon flows according to a carbon to phosphorus ratio at which phosphorus is allowed to enter a specific pool (Table 3-3). The different fl ows and their directions are summarized in the soil organic phosphorus processes section of Figure 3-1.

PAGE 71

71Phosphorus flows between the organic pools Phosphorus flow from any organic pool A to any organic pool B (PflowAB) is proportional to the carbon flow between the same pools. The terms pool A and pool B as used in this section refer to any combinati on of the four litter pools and two soil organic phosphorus pools between which P flow can occur (F igure 3-1). A typical fl ow can be described by the following equation: P flow (from pool A to pool B in a specific layer) in kg [P] ha-1 = A AC CFlowAB P (3-9) Where PA is the amount of phosphorus in pool A of that layer (kg [P] ha-1). The amount of phosphorus in each of the five pools (3 inorganic and 2 organic) is defined at initialization. CA is the amount of carbon in pool A of that layer (kg [C] ha-1). The partitioning of the measured total organic carbon defines the amount of carbon that be longs to the three different soil organic matter pools. The fractions of carbon in SOM1 (active), SOM2 (slow) and SOM3 (passive) are defined by the m odel user according the cultivation history of the soil. Typical partitioning of the total SOM respectively into SOM1, SOM2 and SOM3 for a previously cultivated, irrigated and highly fertilizer loamy soil is 2%, 39% and 59% (Parton et al., 1988, 1994), but this can be varied but the user. CFlowAB is the carbon flow from pool A to pool B (kg [C] ha-1) for that layer. The flow of carbon out of a pool is calc ulated as follow (Gijsman et al., 2002a): C flow (out of pool A) in kg [C] ha-1 d-1 = OTHER DEFAC CUL DEC CA A A (3-10) Where CA is the carbon content of pool A (kg [C] ha-1); DECA is the maximum decomposition rate of pool A under optimal c onditions and without increased decomposition due to soil disturbance (day-1). The maximum decomposition rates of various pools are listed in Table 3-3. CULA is the effect of cultivation on the decomposition rate of pool A. CULTA functions as a multiplier on the maximum decomposition rate (0 to 1); DEFAC is the decomposition factor that repres ents the effect of temperature and low soil water conditions on the decomposition rate parameter. DEFAC functions as a multiplier on the maximum decomposition rate (0 to 1); OTHER represents the effect of other fact ors on the maximum decomposition rate. These factors include the lignin content of the structur al material and the cl ay content of the soil which are used to reduce the decomposition rate of the structural litter and the soil SOM1. The lignin concentration of the structural litter is used to pa rtition its total carbon flow to SOM1 and SOM23. The non-lignin portion enters SOM1 and the lignin portion flows into SOM23 with carbon and phosphorus.

PAGE 72

72Phosphorus mineralization and immobilization The material flowing out of pool A is a llowed to enter pool B only under a certain C:P ratio that is computed assuming a potential immo bilization rate of phosphorus This constraint is depicted by the following equation: CPB = IMMOB PFlowAB CFlowAB (3-11) Where CPB is the C:P ratio of the material allowed to enter th e receiving pool B; IMMOB is the immobilization of P (kg [P] ha-1) from the inorganic labile pool; CFlowAB and PFlowAB are respec tively the C flow and the P flow from pool A to pool B in (kg [C or P] ha-1). When the material flowing from pool A to pool B has a C:P ratio that is larger than the C:P ratio of the material that is allowed to enter pool B (CPB), an immobilization of phosphorus from the inorganic labile pool occurs to compensate fo r the deficit of P in the material flowing. The amount of phosphorus immobilized is derived from equation (3-11): IMMOB = PFlowAB CPB CFlowAB (3-12) Mineralization (MINERAB) occurs only when the actual flow (PFlowAB) exceeds the expected flow CPB CFlowAB. MINERAB = CPB CFlowAB PFlowAB (3-13) Each carbon flow is accompanied by respirat ion losses in the form of carbon dioxide (CO2), which is a flow of carbon that doe s not enter the rece iving pool. Phosphorus mineralization is also concomitant with this lo ss of carbon to CO2. The amount of phosphorus that is mineralized during the respiration process (PFlowCO2) is calculated as: P flow CO2 (from A to B) in kg [P] ha-1 = A AC FlowA CO P 2 (3-14)

PAGE 73

73 Where CO2FlowA is the CO2 flow out of pool A (kg [C] ha-1); PA and CA are respectively the amount of phosphorus and carbon in pool A (kg[C or P] ha-1). The total P mineralization (MINERTOT) resulting from the carbon flow and the respiration losses to CO2 is therefore: MINERTOT = 2PFlowCO MINERAB (3-15) The net phosphorus mineralized The flow of carbon or phosphorus from pool A to pool B generates an immobilization of phosphorus in the material flowing and a minera lization of phosphorus th at does not enter pool B. Immobilization holds phosphorus and depletes the soil inorgani c P but mineralization releases phosphorus that can be made available for plan t uptake. Total P mineralized (SUMPMIN) and total P immobilized (SUMPIMM) are com puted by summing up the mineralization and immobilization P from all the different flow s. A net P mineralized corresponding to the difference (SUMPMIN SUMPIMM) is calculated and added to the inorganic labile pool for plant uptake (Figure 3-1). However, if the total P immobilization and other P takeoff are greater than the amount of P available in the soil, the SOM and litter decomposition are reduced by a reduction factor, so that the amount of P need ed for immobilization equals the amount of P available in the soil. Description of the Plant Phosphorus Model The plant phosphorus component models P ta ken up from the soil and stored in four different plant parts: roots, s hoots (leaves plus stems), shells and seeds. Phosphorus supplied by the soil to meet the plants demand and phosphorus exported by the crop at harvest are external to the plant P component.

PAGE 74

74Phosphorus in the Plant The P accumulated in the whole plant is the su m of the P taken up into the different plant parts (Jones et al., 2005b ; Daroub et al., 2003). Seed Shell Shoot Root PlantP P P P P (3-16) The P in the different parts of the plant is computed as P concentr ations (g [P] g [shoot]-1 for instance). The mass of P (in kg P ha-1) is further calculated after its combination with growth data provided by the appropriate crop growth model. Minimum and optimum concentrations of P fo r maize defined at three growth stages are derived from literature and stored in a speci es file (Table 3-4). The optimum shoot P concentration was calculated using growth st age dependent equations developed by Jones (1983). The minimum shoot P concen tration was taken as 60% of the optimum values (Daroub et al., 2003). Initial P concentration values for the di fferent plant parts are se t to the optimum when the plant emerges from the soil. Because the model uses a daily time step to compute phosphorus in the plant and the optimum and minimum concentr ations of P are availabl e at discrete growth stages only, linear interpolation between the growth stages is used to determine the optimum and minimum P concentrations ever y day (Figure 3-2). These inte rpolations depend on the actual plant growth as influenced by cultivar characteri stics, soil and weather conditions. The N:P ratio is handled in a similar way to constrain the upta ke of P when nitrogen is limiting (Figure 3-3). Actual phosphorus accumulated in the plant on any day is increased by uptake. Amount of P mobilized (from roots, shells and shoots only) and lost due to senescence, pest and disease is furthermore subtracted from the P in the plant part considered.

PAGE 75

75Uptake The amount of phosphorus available for uptake by the whole plant is the minimum of demand and soil supply. PTotal_Uptake = Supply Soil Demand TotalP P MIN_ _, (3-17) The maximum and minimum N:P ratio comput ed daily by linear interpolation from Figure 3-3 is used to limit P uptake if on any da y the actual N:P ratio is below the minimum. PTotal_Uptake_Nlimited = Factor duction Uptake Uptake TotalP P_ Re (3-18) The P uptake reduction factor utilized when th e actual N:P ratio falls below the minimum value is calculated as follows: PUptake_Reduction_Factor = 0 1 : :Minimum ActualP N P N MIN (3-19) Where N:PActual is the actual N:P ratio and N:PMinimum is the minimum N:P ratio. Soil Supply Soil P supply is the amount of root zone labi le P computed in the soil inorganic module. Only a fraction of the soil supply is considered available (soluble) to meet the plant demand on any day. That fraction is a parameter changeable by the user. The value currently used is 0.2 meaning that 20% of the labile inorganic P in the soil zone adjacent to roots on any one day can be taken up by the plant during that day. This va lue was obtained based on a best-fit compromise between simulated and measured biomass from the Wa experiment described in Chapter 2. Plant Demand and P Mobilization Pools Plant concentration of phosphorus at any one time during the growing cycle dropping below the optimum concentration (specified in the species file) is considered a deficit and induces stress (Figure 3-4). Dema nd is calculated for each plant part based on the amount of P

PAGE 76

76 required to bring the P concentration in each of the plant parts up to the optimum, plus P required for new growth. PDemand = growth New Actual OptimumP P P_ (3-20) Where PDemand is the amount of P in kg ha-1 required to bring the act ual concentration of P to the optimum; POptimum is the computed optimum P in kg ha-1 using linear interpolation; PActual is the amount of P in kg ha-1 present in the plant part concerned; PNew_growth is the amount of P in kg ha-1 needed for new growth. High biomass accumulation can cause PActual to be greater than POptimum resulting in a negative PDemand for any plant part at any moment during the growth. The excess P accumulated is therefore stored in a mobiliz ation pool for each plant part. Th ere is no P mobilization pool for seeds. Demand for each plant part is first met by P st ored in the mobilization pools. P moves from root and shoot mobilization pools to satisfy P dema nd in shells and seeds. The P leftover stays in the respective mobilization pools. Total P demand is recalculated and is possibly met by the soil supply. If the soil supply is insufficient to meet this demand, uptake and subsequently P concentration in plants parts are reduced and will increase P stress. Partitioning and Translocation During the reproductive phase, tota l phosphorus taken up by the plant (PTotal_Uptake) from the soil is first used to meet seed demand. If the seed demand is greater than the amount of phosphorus available to meet this demand, phosp horus translocation from roots, shoots and shells to the seeds occurs. The available P for translocation is calculated as follows: PTranslocation = Min Shells Actual Shells Min Shoots Actual Shoots Min Roots Actual RootsP P P P P P_ _ (3-21) The maximum amount of P that can be mined fr om the shells and the vegetative tissue in one day (PTranslocation_Max) is a fraction of the available P for translocation.

PAGE 77

77 PTranslocation_Max = ion TranslocatP FracPMobil MAX 0 0 (3-22) Where FracPMobil is the fraction of the translocat ed P that can be used by the seeds in one day. FracPMobil is defined as a parameter in the species file. The P translocated is used to meet the seed demand if it is still positive after total P uptake from the soil has been used up. The remaining P after seed demand is fully met is used by shells. Vegetative tissues (roots and s hoots) are supplied with P after reproductive organs (seeds and shells) demand is met. Stress Factors Two stress factors are co mputed based on a P stress ratio wh en the actual shoot phosphorus concentration falls below the optimum. The stress ratio is computed as follows: PStress_Ratio = ion Concentrat Min Shoots ion Concentrat Optimum Shoots ion Concentrat Min Shoots ion Concentrat Actual ShootsP P P P MIN_ _ _, 0 1 (3-23) PStress_Ratio = 0 means maximum stress and PStress_Ratio = 1 means no stress. P stress effects on photosynthesis and P parti tioning are modeled differently. Thresholds values are defined in the species file and are used to compute stress factors for photosynthesis and P partitioning: PStress_Factor_Photosynthesis = 0 1 ,_SRATPHOTO P MINRatio Stress (3-24) PStress_Factor_Partitioning = 0 1 ,_SRATPART P MINRatio Stress (3-25) Where SRATPHOTO is the minimum value of th e ratio of P in vegetative tissue to the optimum P, below which reduced photosynthesis will occur. SRATPART is the minimum value of the ratio of P in vegetative tissue to the optimum P, below which vegetative partitioning will be affected. The two P stress factors, which are given different weights, are used to reduce photosynthesis and P partitioning on any day during the growth of the plant when the actual shoot P concentration falls below the computed optimum shoot P concentration. Values of

PAGE 78

78 SRATPHOTO and SRATPART as read from the species file are resp ectively 0.80 and 1.00 meaning that P deficits in shoot tissue will first affect root-shoot partitioning before it affects photosynthesis (Figure 3-4). Model Inputs and Outputs The soil-plant phosphorus model does not r un as a standalone application but is intrinsically linked to DSSAT cr op growth models. As a conseque nce the soil-plant P model also uses the basic inputs required to run the crop growth models. Additional inputs and parameters required to run the model are summarized in Tabl es 3-5 and 3-6. The model essentially modifies the crop growth models outputs to allow them to be phosphorus-limited in P-limiting environments. Sensitivity Analysis Sensitivity analysis is an important assessm ent tool that assists with evaluating the uncertainty and variability associated w ith model structure and inputs during model development, calibration and validation. Introduction A simulation model of crop growth and developmen t is the result of several cycles of fine tuning of model theory and structure, parame ter estimations and adjustment of number of required input variables. The ultimate objective of the continuing model refinement is to obtain a model that is as close to the ideal model as possible, predicting measurable outputs with maximum accuracy. Scientists admit, however, that even the most carefully-built simulation model is not expected to give si mulations that exactly equal observations. Uncertainty associated with model equations, measured model input variables and estimated model parameters will always remain an integrated part of the mode l and will contribute a great deal to simulation biases. The uncertainty is not an accident; it may be the substance of the scientific method itself

PAGE 79

79 (Saltelli, 2002). More specifically, the role of sensitivity analyses is to help apportion the uncertainty in the model output to the different sources of uncerta inty and variability in inputs (Saltelli, 2005). Uncertainty and variability just ifying the usefulness of sensitiv ity analysis can stem from various sources: Choice of an appropriate complexity. Modeling agricultural and biological systems requires an appropriate choice of components th at are meaningful for the system and will eventually form the structure of the model. The same real world system can be approached differently by various scientific communities although they may set the same objectives and have similar technical backgrounds. The main classi cal theories may be the same but the way scientists see the system can be influential in the way they model it. For example, modelers of phosphorus-limited production have used differe nt P pools (Probert, 2004; Daroub et al., 2003). Because the components modeled and th e structure used set the mathematical representation, the choice of the model comp lexity can be subjective and introduce some uncertainty with respect to the processes involved. Parameter estimation. Uncertainty can also come from parameters estimated based on weak evidence or not-so-well es tablished experimental resu lts, especially during model development. Measured model inputs. Another important source of vari ability is model input variables that may have been measured from field experi ments or obtained from various data sources. Field measurements (even replicated) can includ e important precision errors, sometimes due to variability in natural processes. Errors of this kind can propagate to model outputs more than proportionally.

PAGE 80

80Expert systems. Parameters or input variable that cannot be easily measured are sometimes indirectly estimated using expert opin ions or empirical relationships such as pedotransfer functions (Gijsman et al., 2002b). When confidence limits are not provided for these methods, reliability on the indirect estimates may be questionable. Modeling language and untrained model users. When the modeling language is not English-like, model users, sometimes performing calibration with no capability to read the programming language, may have to use it as a bl ack box. It is not evident that members of interdisciplinary teams in which a model was de veloped are aware of th e way inputs are mapped to outputs in the total model (O berkampf et al., 2004). With co mplex models requiring hundreds of parameters and input variab les to operate, it may become unclear how the model behaves independently of any evaluati on with real world datasets. Sensitivity analysis has become an ingredient of modeling (Saltelli et al., 2000) and been used in many studies at various stages of model development (Makowski et al., 2005; Ratto et al., 2001; Rahn et al., 2001). General objectives of a sensitivity analysis are (Monod et al., 2006): i) to verify that the model behaves as expect ed when inputs are changed; ii) to quantify the magnitude of the influence of parameters on output s; iii) to identify the model parameters that require maximum accuracy in their estimation; iv) to identify input variables to the model that need to be measured accurately for the simulati ons to be correct; and v) to isolate possible parameter interactions effects on outputs. This section describes and presents results of a sensitivity analysis performed on some major P-related parameters of the soil-plant pho sphorus model in DSSAT. The overall objective of this sensitivity analysis was to assess the ef fect on maize biomass, gr ain yield and P uptake of

PAGE 81

81 major inputs and parameters that are directly related to plant response to P and P stress. Specific questions for this sensitivity analysis were Question 1: Does the model respond to P fertilizer as expected? Question 2: How does biomass and grain yield react to changes in initial PiLabile where plant uptake occurs, initial orga nic P that add P to this uptak e pool, and the fraction of this uptake pool that is soluble? Question 3: What is the magnitude of the effect of variability in initial PiLabile and initial organic P, two P pools that are estimated in the model with un certainty, on biomass and grain yield in the model? Question 4: How does the model react to variabili ty (as found in the literature), in parameters that are used to compute P stress, optimum and minimum s hoot P concentration? Question 5: Would variability in optimum and mi nimum seed P concentration have any effect on biomass accumulation by th e plant or crop grain yield? Materials and Methods The sensitivity analysis requires the specifica tion of a computer expe riment, a sensitivity analysis method and inputs factors. These three components of the sensitivity are described next. Computer experiment A sensitivity analysis can be regarded as a highly controlled experiment carried out using specific treatments applied to a specific crop growing in a specific environment under specific management conditions. The mention of highly c ontrolled carries an important meaning for a sensitivity analysis because only the effects of the treatments are investigated and all other growing factors are fixed at c onstant values. While this kind of experiment can bear a high resemblance to a field station trial, some peculiar characteristics must be pointed out:

PAGE 82

82 The aim of a sensitivity analys is is to study the behavior of a model whereas the aim of a field station trial is to examine the behavior of nature. The sensitivity analysis of a crop model can be thought of as an experiment wh ere nature is replaced by the simulated crop model (Monod et al., 2006); The experiment described through a sensit ivity analysis can be hypothetical. Some conditions relevant to the experiment cannot be met in a station tr ial due to limits to control nature or ethical considerations. In the sensitivity analysis experiment, there is no limit to the achievement of conditions required to isolate treatments effects. A station trial examines the behavior of nature through independent factors usually external to the field but that are known or su spected to influence dependent variables. In a sensitivity analysis, the input factors are ne cessarily sampled or chosen from the input variables to the model or the model parameters. In a station trial, rep lications are key components to statistic al analysis of the results of the experiment due to field variabil ity that result in measurement error. When an analysis of variance is performed, the meas urement error is used as a pr oxy to assess the amount of variation accounted for by each factor under stu dy. In sensitivity analysis experiments, there is no need for repeating the same treatments as long as the model is deterministic. As a result, measurement error (actually simula tion error) cannot be computed and formal hypothesis testing has no scien tific meaning and cannot even be performed (Monod et al., 2006). To differentiate the sensitivity analysis experi ment from real world station trials, we will call it a computer experiment. The treatments w ill be called scenarios. Each scenario is the combination of levels of factors that will be named input factors (Monod et al., 2006). Settings for the computer experiment The computer experiment was performe d using agro-ecological and modified management information from the phosphorus experiment conducted in Kpeve, Ghana (6o 40.80 N, 0o 19.20 E, altitude 67 m above sea leve l), in 2006 and described in Chapter 2. The experiment was conducted during the main rainy season (April to August) on a loamy soil (Tables 3-7 and 2-15). Daily rainfall, solar radiation, a nd maximum and minimum temperatures were monitored using an automatic weather station located within the research station. A medium-duration cu ltivar, Obatampa (Table 2-1) was sown on May 27, 2006. The plant population at emergence was 6.25 plants m-2. To remove any nitrogen stress, a total of 500

PAGE 83

83 kg N ha-1 was applied in the form of urea and split as follow: 100 at planting; 150 at 14 days after planting; 150 at 27 da ys after planting; and 100 at 41 days after planting. Water stress was controlled by automatic irrigation when the avai lable water in the first 50 cm dropped below 70% of the drained upper limit. Input factors, scenarios and model outputs Three input variables to th e soil modules and three plant module parameters were selected and constituted two categ ories of input factors for the se nsitivity analysis (Table 3-8). Each input factor had 3 levels. The factor levels were specified in a way that a low, a medium and a high setting of the factor were investigated through the sensitivity analysis. Wherever applicable the medium level reflected the default (or the nominal) values in itially specified in the model and defined the control s cenarios (Tables 3-8 and 3-9). The input factor values were selected ba sed on the knowledge of their uncertainty around a nominal value or the medium settings. To an swer specific questions addressed in this sensitivity analysis, six input f actors were selected. The input f actors were selected from model parameters (Table 3-5) and inputs (Table 3-6). Th e six inputs factors for the sensitivity analysis are discussed next. Initial Inorganic labile P and organic P. Although accurate knowledge of labile inorganic phosphorus and total organic phosphorus pr esent in the soil at planting is crucial for good quality simulations of growth and development, it was antic ipated that most model users will not have access to the data measured as needed. If an alternate method is not provided for indirect estimation, potential model users could possibly resort to indi cative values found in literature or eventually conclude that the model is not of any pract ical use because required input data are not readily available. Access to heavy data requirem ent by the model user is a known constraint of model use in research and developm ent in large parts of th e world (Struif-Bontkes

PAGE 84

84 and Wopereis, 2003; Matthews and Stephens, 200 2; Walker, 2000). Since organic carbon, pH and available phosphorus are routinely measured in most traditional agronomic experiments, developing relationships th at can make use of those data and provide reasonable estimates of inorganic labile P and organic P was thought to be helpful. Some soil properties are related to each other and may be estimated from selected measurements; however, indirect estimation of inorganic and organic phosphorus can pose seri ous uncertainty problems due to the complex chemistry of P in soils. In this specific situa tion, the uncertainty comes from two main sources: 1) measurement errors of the soil parameters; 2) regression errors associat ed with developing the equations for estimating indirectly the soil phosp horus. This was a strong motivation for studying the effect on variable initial inorganic labile P and total organic P on some key model outputs. Studies by Sharpley et al. ( 1984 and 1989) who developed linear relationships to predict inorganic labile P and organic P for different cate gories of soils provided a basis for the ranges of values used in the sensitivity analysis. Most so ils considered in Sharpley (1984) have measured PiLabile in the range 0-15 ppm and total organi c P in the range 50-200 ppm. The PiLabile in the first 20 cm was 6.5 ppm at Wa and 16 ppm at Kp eve. The estimated organic P (from Table C-5, Appendix C, Slightly weathered soils) in th e top 20 cm was 37 ppm at Wa and 133 ppm at Kpeve. These two soils clearly have different P-supply capabilities in the ranges explored by Sharpley, and they present indicative starting poin ts for the specific levels of the input factors initial PiLabile and organic P. Since some soils in Sharpleys survey would be more PiLabiledepleted than the Wa soil, the low level of th e input factor PiLabile was set at 2.0 ppm. The medium level of the input factor Initial PiLabile was set at 8 ppm, around the middle of the range 0-15 ppm found in Sharpley (1984). The high level of the same input factor was set at the upper boundary of that range.

PAGE 85

85 The organic P level in the Wa soil (37 ppm) seems representative of a very low level so the low level of the input factor In itial Organic P was set at 40 ppm. The medium level of the input factor Initial Organic P was se t at 100 ppm, around the middle of the range 50-200 ppm found in Sharpley (1984). The upper limit of the range 50 -200 ppm (that is 200 ppm) was considered as the high level of the input factor Initial Organic P. The low, medium and high levels of the two input factors Initial PiLabile and Initial organic P were therefore respectively 2, 8 a nd 15 ppm for Initial PiLabile and 40, 100 and 200 ppm for Initial organic P. The original soil ra tio between organic C and P (Table 3-7) was maintained for the soil used in the analysis m eaning that the soil carbon input value was changed along with the organic P. P fertilizer. Fertilizer application is one of the most important tactical management strategies used to balance nutrient requirements by crops. Studying and understanding a crop growth models behavior to varying fertilizer le vels is key to checking on the models ability to simulate variable fertilizer i nput feasibilities. Small applications of phosphorus (20-40 kg [P] ha1) to degraded soils have been recommended to rest ore progressively the soil P status (Shapiro et al., 2003) and reported to increase inorganic soil labile P (Nziguheba et al., 2002) and improve grain yield in maize (Fofana et al., 2005). The levels of phosphorus fertilizer were set to 0, 30 and 60 kg P ha-1. The P fertilizer was managed in the same way as in the Kpeve experiment to ensure efficient use by the plant, split-applied in bands, 50% at planting and 50% 14 days after planting. Fraction of root labi le P that is soluble. The maximum fraction of available phosphorus which can be taken up in a day has a value of 0. 20 in the phosphorus model and is not allowed to vary regardless of growth stage, cultivar variat ion or differences in phos phorus uptake efficiency.

PAGE 86

86 However, the uptake rate of phosphorus by cereal plants varies with pl ant age and intrinsic cultivar differences (Johnston, 2000). If there is a large supply of phosphorus in the soil, restricting the soluble P to 20% throughout the season could result in P shortage at the maximum uptake period. Phosphorus uptake can also be greatly increased in mychorizae-colonized environments, which will critically modify any pa rameter setting for P availability under normal conditions. These reasons may have motivated uptake of nutrients including phosphorus by biological systems to be modeled as a substrat e saturation process with end-product inhibition using the Michaelis-Menten f unction (Wilson and Botkin, 1990; Lehman et al., 1975a). To quantify the effects on some mode l outputs of possible variations of the maximum fraction of available phosphorus which can be taken up in a day around the approximated average of 0.20, three uptake fraction levels 0.10, 0.20 and 0.80 we re tested in the present sensitivity analysis. Shoot P and seed P. Optimum and minimum values of shoot and seed phosphorus concentration used in the plant module were deri ved from literature (Daroub et al., 2003; Jones, 1983). Although those concentration limits are essentially associat ed with crop physiology, they can be subject to cultivar variab ility. In addition, P modelers ha ve used different optimum shoot P concentrations. For example, Dar oub et al. (2003) used an initia l shoot P concentration of 0.7% whereas Probert (2004) used 0.5%, which is lo wer than the shoot P le vel of maize found by Jones (1983) in his survey. Other studies reported even lower in itial shoot P concentration under non-limiting P conditions (e.g. 0.45, Ziadi et al., 2007) This sensitivity analysis was designed to cover the range of optimum s hoot P variability found in literature. The optimum shoot P concentration was increased or decreased by 50% around the nominal to obtain the high and low levels of the input factor shoot P concentr ation (Table 3-9). The minimum shoot P concentration was taken as 60% of the optimum (Daroub et al., 2003). The optimum seed

PAGE 87

87 concentration was also increased or decreased by 50% around the nominal value of 0.35% to obtain the high and low levels of the input fact or seed P concentration. The ratio of 2:1 between optimum and minimum seed P concentration was kept for all the input factor levels. The six input factors and their levels are summarized in Tables 3-8 and 3-9. Model outputs. The effects of the total input space constituted of the six input factors were assessed on three model outputs: aboveground biomass, grain yield and total plant uptake. Method and design of th e sensitivity analysis The analysis was conducted using a global ap proach where the input factors and their interactions were explored simu ltaneously (Saltelli, 2004). The six factors, each having 3 levels were combined in a complete factorial design yielding 36 = 729 observations or simulation runs. Sensitivity index An analysis of variance was carried out on the model outputs using the SAS software (SAS, 2002). Since formal hypothesis testing cannot be performed due to the lack of a valid error term, the most useful assessment procedure was to compare the sum of squares contributions from each factor and interactions to the total sum of squares. This is denoted by the name sensitivity index and is calculated as follow: SI (Main effect of F) = SS Total FactorF SS_ (3-26) SI (Interactions involving F) = SS Total nsF Interactio SS_ (3-27) SI (Total effect of F) = SI (Main effect of F) + SI (Interactions involving F) (3-28) Where SI represents sensitivity index; F represents factor F; SS represents ANOVA Sum of Squares; InteractionsF represents interacti ons involving factor F.

PAGE 88

88 The sensitivity index was used to measure th e effect of factors a nd their interactions on outputs. It has values between 0 and 1 with 0 indi cating that the model is not sensitive at all to the factor for the particular output, and 1 indicative of maximum sensitivity. Results and Discussion Phosphorus fertilizer and initial PiLabile ha d the most influentia l effects on the output variables. In the absence of P fertilizer, the effect of the fraction of labile P in solution and the shoot P became also important. Soil inputs effects Figures 3-5A to C show the response of cr op total aboveground biom ass, grain yield and crop total phosphorus uptake to the three levels of the soil input factors (t he levels low, medium and high along the abscissa have different meani ng depending on the input factor considered and are explained in Table 3-9). Among the soil input factors tested initial PiLabile and P ferti lizer had the most influential effects on the three output variab les. Aboveground biomass, grain yield and total plant P uptake responded clearly to variable levels of initial PiLa bile and P fertilizer application with the same pattern. Biomass, grain yield and uptake increa sed and their correspondi ng standard deviations decreased with increasing levels of initial PiLa bile and P fertilizer (Fi gures 3-5A, 3-5B, and 35C). The sensitivity indices of the re sponse of initial PiLabile and P fertilizer to the three output variables were relatively high co mpared to the other input factor s (Figure 3-7). These two input factors alone explained 54%, 47% and 41% of the total sum of squares respectively for aboveground biomass, grain yield and tota l plant P uptake (Tables 3-10 to 3-12). The total sensitivity indi ces of P fertilizer and initial Pi Labile were about two times their main effects showing that these two input factors were also infl uential in terms of interaction

PAGE 89

89 with the other input factors. The total sensitivity index of P fertilizer was 0.73 for biomass (Table 3-10), 0.71 for grain yield (Table 3-11) and 0. 52 for plant P uptake (Table 3-12). The total sensitivity index of initial PiLabile was 0.27 for biomass (Table 3-10), 0.31 for grain yield (Table 3-11) and 0.24 for P uptake (Table 3-12). Organic phosphorus did not contribute much to the variation in any of the output variables over the range tested. This range that was deri ved from Sharpleys studies (1984) reflected approximately the variability in organic P in mo st P-depleted soils (Brady and Weil, 2002). The sensitivity indices for this factor were smaller than 0.01 (Tables 3-10 to 3-12). The response to the input factor s initial PiLabile and P fertil izer varied with the output variable and the level of the input factor. For ex ample, when the six input factors are considered together, the response was more pronounced for the output variable to tal P uptake than for biomass and grain yield. Concerni ng the individual levels low, medium and high, of the input factors, the response of the out put variables was especially marked at the low levels. Variability in the input factors initial PiLa bile and P fertilizer did not result in a proportional variability in the outpu t variables. For example, a decr ease in initial PiLabile of 75% relative to the nominal value resulted in 16% decrease in biomass and grain yield relative to the nominal values, and 24% decrease in total P uptake on average (Table 3-14). An increase in initial PiLabile of 88% resulted in only 8% in crease in biomass and grain yield and 11% increase in total P uptake on average. Biomass increa sed by 37%, grain yield by 33% and total P uptake by 42% on average when P fertilizer was increased from 0 to 30 kg [P] ha-1 (Table 3-14). The increase in the output variables barely exceeded 5% when P fertilizer increased from 30 to 60 kg [P] ha-1.

PAGE 90

90 The reason initial PiLabile, P fertilizer and th eir interactions have so much effect on the output variables is that in the model, plants ta ke up phosphorus directly from the labile pool. For slightly weathered soils with no base saturation measured, as the one used in this sensitivity analysis, 85% of the P fertilizer applied enters di rectly the labile pool making this pool the most important for plant uptake and produ ctivity. The higher sensitivity of total P uptake is associated with the fact that the uptake is the primary process directly c onnected to phosphorus availability. The behavior of the model in terms of res ponse of crop productivity and uptake to initial PiLabile and fertilizer is supporte d by numerous fertilizer studies (Colomb et al., 2000; Fofana et al., 2005; Nziguheba et al., 2002; Pe llerin et al., 2000) and does not per se raise new issues. Initial organic P had virtually no effect on the output variables suggesting that over a growing season, the organic matter contributio n to phosphorus uptake may be more dependent on the rate constants controlli ng the mineralization and immobili zation of organic P than the amount of organic P available at the beginning of the growing season. In f act in this sensitivity analysis, the total P mineralized from organic ma tter in the soil profile varied from 45 to 55 kg ha-1 at the end of the season, but the fraction that is actually c ontributing to plant uptake was small. This is because 1) the roots do not e xplore the whole soil profile and therefore cannot access all the P mineralized; 2) th e portion of organic P mineralized that is soluble for plant uptake is only that amount that enters the volume of the soil where roots are present at the time of the mineralization. This means that minerali zed P not seen by the plant at any one time in the season because the volume of roots is small is transformed over time into insoluble forms that cannot be used by the plant. The synchroniza tion between the av ailability of the mineralized organic P and the accessibility of the plant to it was an influential factor of the small sensitivity observed of the input fact or Initial organic P.

PAGE 91

91Plant parameters effects The variability in plant parameters studied had less influence on the output variables than the soil parameters (Figures 3-6A to C). The grap hs were scaled uniformly to Figures 3-5A to C to highlight the differences in response among the two groups of input factors (soil and plant). Aboveground biomass, grain yield and plant uptake varied over a smaller range, which resulted in main sensitivity indices between 0.02 and 0. 15 (Tables 3-10 to 3-12). Shoot P had a higher influence on biomass (main SI of 0.04, Table 310), Fraction of labile P had a higher influence on grain yield (main SI of 0.03) and seed P had a higher influence on total P uptake (main SI of 0.15). Increase in fraction of labile P from 0.2 to 0.8 resulted in about 7% increase on average in the output variables. Decrease in this fraction fr om 0.2 to 0.1 caused an 8% decrease on average in the output variab les (Table 3-14). Variation of the shoot P concentration from the medium to the low level (Table 3-9) caused the biomass to decrease by 13%, the grain by 3% on average. However, this decrease in shoot P concentration increased the total P uptake by 39%. When the shoot P levels were increased from medium to high, the biomass and the grain yield decreased by 10% but the total P uptake increased by 18% (Table 3-14). Variation in seed P from medium to low d ecreased biomass and grain yield by about 4% on average. Increase in seed P from medium to high increased biomass and grain yield by only 1% on average. The total P uptake was more re sponsive to seed P variation. The uptake was reduced by 39% between the medi um and the low seed P level and increased by 18% on average between the medium and high seed P levels (Table 3-14). It is possible that higher eff ects could be detected at wider ranges of the three plant parameters. However, realistic motivations must support such analyses becau se the variability in

PAGE 92

92 optimum phosphorus concentration in maize shoo ts and seed for instance is not of any huge magnitude (Jones, 1983). Interactions As might be expected from their high contri butions to the total sum of squares Initial PiLabile and P fertilizer produced relatively high interactions with other factors. The interactions between Initial PiLabile and P fer tilizer and P fertilizer and shoot P were the strongest (Tables 310 to 3-12). The interaction plot (Figure 3-8) shows how the strong response to P fertilizer at low initial PiLabile disappeared very quickly as initial PiLabile increases. This has important implications when simulating a fertilizer trial: accurate measur ements of initial PiLabile must be obtained in order to achieve good simulation of crop production. On the contrary, the response to Initial PiLabile did not seem to be affected by the frac tion of labile P in solution. Increasing Initial PiLabile resulted in an increase of biomass at all fraction of labile P in solution levels. However, the rate of biomass increase did not depend on the le vel of fraction of labile P in solution (Figure 3-9). Special case of zero P fertilizer When the sensitivity indices were recalculated considering the 0P fertilizer level only, the effect on the relative order of importance of the input factors was small (Table 3-13). Initial PiLabile was the most influential factor (mai n SI of 0.47, Table 3-13). Shoot P had a much higher effect on the biomass (main SI of 0.25) th an the case when P fertilizer was included. The Fraction of labile P in solution had also a much higher effect on the biomass (main SI of 0.15) than the case when P fertilizer was included. The effects of Initial organic P and Seed P remained relatively small (main and total SIs less 0.01, Table 3-13).

PAGE 93

93 The sensitivity indices of Initial PiLabile, S hoot P and Fraction of labile P in solution increased because the dominant factor P fertiliz er was removed from the input space. The main sensitivity index of Initial PiLabi le and Fraction of labile P in so lution increased by a factor of 4 and the main sensitivity index of Shoot P increas ed by a factor of 6 (Tables 3-10 and 3-13). The persistence of main sensitivity indices less than 0.01 for Initial organic P and Seed P suggested that the effects of these two f actors on biomass were small per se and were not masked by the dominant effect of P fertilizer. However, interactions between factors were w eak (highest interaction SI for biomass was 0.04, Table 3-13, compared to highest interactio n of 0.15 with P fertilizer, Table 3-10) showing that most of the interactions between factors were due to the presence of P fertilizer. In fact, the presence of P fertilizer as a fact or had two effects: i) making all in teractions involvi ng P fertilizer relatively stronger than all othe r interactions between factors (Table 3-13); ii) decreasing interaction SIs between other factors by accoun ting for more variability. For example, the interaction SI between In itial PiLabile and Shoot P was 0.01 w ith P fertilizer (Table 3-10) and 0.04 without P fertili zer (Table 3-13). Conclusion While errors in observations tend to be the direct and most evident target when discrepancies between simulations and measuremen ts are recorded, inaccurately-measured input variables and model parameters estimated based on weak evidences or regression analyses can contribute a great deal to simulation errors. The sensitivity analysis of some key model parameters and input variables help to learn the behavior of the model and to diagnose in advance sources of possible simulation errors. A sensitivity analysis on three of the models parameters and three of the models input variab les revealed that i) initial PiLabile and P fertilizer had the greatest impact on grain yield, total plant biomass and total plant uptake of P; ii)

PAGE 94

94 initial organic P had little effect on plant producti on in the range tested and over a single growing season; iii) the fraction of labile P in solution, the optimum shoot and seed P concentrations had smaller effect on grain yield, total plant bioma ss and uptake in the range of sensitivity used in this analysis; iv) the relative order of importance of th e input factors was not affected by P fertilizer application but generally, interaction be tween the factors tested are strengthened in the presence of P fertilizer. Accurate estimation of initial PiLabile presen t in the soil is crucial to simulating crop productivity in phosphorus deficien t cropping systems especially wh en phosphorus fertilizer is simulated. Failure to estimate accurately cultivars optimum shoot and seed P concentration by 50% around the nominal value used in the D SSAT soil-plant phosphorus model can result in biomass, grain yield and total P uptake varia tion of up to 39%. Inaccurate estimation of the fraction of labile P in solution can also become a cause of poor simulation results. Summary and Conclusion The soil-plant phosphorus model described in this chapter integr ates soil and plant phosphorus processes that are linked using modu lar programming techniques to crop growth models in the DSSAT CSM. Th e model simulates phosphorus in plants and soils based on integrated processes between i) inorganic phosphor us present in the soil in three pools, labile, active and stable; ii) or ganic phosphorus present in two pools, active and stable; iii) plant phosphorus present in roots, shoots, shells and seeds. Phosphorus-limited production including plant biomass, grain yield and pl ant uptake can also be calculated thanks to the linkage to crop growth models in DSSAT. A sensitivity analysis of the model limited to si x input factors showed that initial PiLabile and P fertilizer were the most important forces driving simu lations of plan t production. The fraction of root labile P, the shoot and seed P appeared to ha ve less impact on biomass, grain

PAGE 95

95 yield and P uptake of maize. Accura te predictions require therefore that at least initial PiLabile be measured or estimated correctly. If PiLabile cannot be measured directly and has to be indirectly estimated based on available P like P-Br ay1 or Olsen, careful attention needs to be paid to the relationships deri ved and their agronomic validity. Diagnosing causes of poor model predictions should not only focus on checking meas urements compared to simulations but also verifying the validity of input data, initial PiLabi le in this instance.

PAGE 96

96 Table 3-1. Soil category-dependent ca lculation of P availability index Soil category P availability index Calcareous 60 0 3 0058 0 CaCO Slightly weathered 70 0 11 0 0034 0 0043 0 pH PiLabile aturation TotalBaseS Highly weathered 68 0 100 log 30 0 CLAY Other soils PiLabile00023 0 40 0 Source: Singh, U. 1985. A crop grow th model for predicting corn (Zea mays L.) performance in the tropics. PhD thesis, Univ ersity of Hawaii, Honolulu. Table 3-2. Soil category-dependent calcula tion of P Fertilizer Availability Index Soil category P Fertilizer Availability Index Calcareous 72 0 3 0042 0 CaCO Slightly weathered 50 0 11 0 0034 0 0043 0 pH PiLabile aturation TotalBaseS Highly weathered 70 0 100 log 19 0 CLAY Other soils PiLabile0002 0 60 0 Source: Singh, U. 1985. A crop grow th model for predicting corn (Zea mays L.) performance in the tropics. PhD thesis, Univ ersity of Hawaii, Honolulu. Table 3-3. Summary of decomposition rates for the soil organic pools an d C:P ratios at which phosphorus is allowed to enter the specific pools Pool generating the flow Pool location Maximum rate at which flow occurs (d-1) Pool C:P Metabolic litter Surface 0.040550 Soil 0.050680 Structural litter Surface 0.010680 Soil 0.013420 SOM1 Surface 0.016440 50 Soil 0.020000 50 SOM2 Soil 0.000548 SOM3 Soil 0.000012 SOM23 Soil 100 Source: Parton, W.J., Ojima, D.S., Cole, C.V ., Schimel, D.S., 1994. A general model for soil organic matter dynamics: Sensitivity to litte r chemistry, texture and management. In: Bryant, R.B., Arnold, R.W. (Eds), Quantita tive modeling of soil forming processes. SSSA Spec. Publ. 39. SSSA, Madison, WI, pp 147-167. Parton, W.J., Stewart, J.W.B., Cole, C.V., 1988. D ynamics of C, N, P and S in grassland soils: A model. Biogeochemistry 5:109. Gijsman, A.J., Hoogenboom, G., Parton, W.J ., Kerridge, P.C., 2002. Modifying DSSAT crop models for low-input agricultural systems using a soil organic matter-residue module from CENTURY. Agronomy Journal 94, 462-474.

PAGE 97

97 Table 3-4. Optimum and minimum phosphorus cont ent (%) in different pl ant parts and maximum and minimum plant N:P ratio at three growth stages, as used in the model for maize Plant part Emergence Effective grain filling/ End of leaf growth* Physiological maturity Root Optimum 0.041 0.041 0.041 Minimum 0.020 0.020 0.020 Shoot Optimum 0.700 0.250 0.200 Minimum 0.400 0.150 0.100 Shell Optimum 0.500 0.500 0.050 Minimum 0.250 0.250 0.025 Seed Optimum 0.350 0.350 0.350 Minimum 0.175 0.175 0.175 Plant N:P ratio Maximum 25.000 15.000 9.300 Minimum 4.200 2.700 2.100 Source: Jones, C.A. 1983. A survey of the variability in tissue nitrogen and phosphorus concentrations in maize and grain sor ghum. Field Crops Research 6, 133-147. Daroub, S.H., Gerakis, A., Ritchie, J.T., Frie sen, D.K., Ryan, J., 2003. Development of a soilplant phosphorus simulation model for cal careous and weathered tropical soils. Agricultural Systems 76, 1157-1181. *The end of leaf growth applies to shoots only.

PAGE 98

98 Table 3-5. Summary of parameters in the soil-plant phosphorus model Parameter Unit P transformations between pool s and P availability Rate constant for transformation from labile P to active P d-1 Rate constant for transformation from active P to labile P d-1 Rate constant for tran sformation from active P to stable P d-1 Rate constant for transformation from stable P to active P d-1 P availability index unitless Fraction of root labile inorganic P that is soluble unitless Shoot P concentrations Optimum shoot P concentration at emergence g g-1 Optimum shoot P concentration at tasseling g g-1 Optimum shoot P concen tration at physiological maturity g g-1 Minimum shoot P concentration at emergence g g-1 Minimum shoot P concentration at tasseling g g-1 Minimum shoot P concentration at physiological maturity g g-1 Root P concentrations Optimum root P concentration at emergence g g-1 Optimum root P concen tration at effective grain filling g g-1 Optimum root P concentration at physiological maturity g g-1 Minimum root P concentration at emergence g g-1 Minimum root P concentration at effective grain filling g g-1 Minimum root P concen tration at physiological maturity g g-1 Shell P concentrations Optimum shell P concentration at emergence g g-1 Optimum shell P concentration at effective grain filling g g-1 Optimum shell P concen tration at physiological maturity g g-1 Minimum shell P concentration at emergence g g-1 Minimum shell P concen tration at effective grain filling g g-1 Minimum shell P concentration at physiological maturity g g-1 Seed P concentrations Optimum seed P concentration at emergence g g-1 Optimum seed P concen tration at effective grain filling g g-1 Optimum seed P concentration at physiological maturity g g-1 Minimum seed P concentration at emergence g g-1 Minimum seed P concen tration at effective grain filling g g-1

PAGE 99

99 Table 3-5. continued Minimum seed P concentration at physiological maturity g g-1 N to P ratios Maximum vegetative N:P ratio at emergence unitless Maximum vegetative N:P ratio at effective grain filling unitless Maximum vegetative N:P ratio at physiological maturity unitless Minimum vegetative N:P ratio at emergence unitless Minimum vegetative N:P ratio at effective grain filling unitless Minimum vegetative N:P ratio at physiological maturity unitless P mobilization and stress Maximum fraction of P which can be mobilized from shoot per day unitless Minimum value of the ratio of P in vegetative tissue to the optimum P below which reduced photosynthesis will occur unitless Minimum value of the ratio of P in vegetative tissue to the optimum P below which vegetative part itioning will be affected unitless Table 3-6. Summary of additional inputs required to run the soil-plant phosphorus model in DSSAT Input Unit Initial labile inorganic P ppm Initial active inorganic P ppm Initial stable inorganic P ppm Initial active organic P ppm Initial stable organic P ppm P in residue (if applied) % P fertilizer (if applied) kg ha-1 Soil CEC cmolc kg-1 Soil texture % Soil CaCO3 content %

PAGE 100

100 Table 3-7. Selected physical and chemical properties of the Kpeve soil used in the sensitivity analysis, as estimated from pedo-transfer functions in DSSAT SLB SLLL SDUL SSAT SRGF SBDM SLCF C:P 10 0.180 0.260 0.460 1.000 0.83 40.0 138 20 0.070 0.140 0.280 1.000 1.08 40.0 136 30 0.040 0.080 0.160 0.607 1.47 35.0 130 40 0.060 0.120 0.240 0.497 0.74 74.5 138 50 0.040 0.060 0.120 0.407 0.47 88.1 123 60 0.050 0.090 0.180 0.333 0.56 82.3 218 70 0.080 0.150 0.300 0.273 0.97 61.9 200 80 0.060 0.110 0.220 0.223 0.77 72.9 127 90 0.090 0.160 0.320 0.183 1.04 57.9 124 SLB, depth, base of soil laye r (cm); SLLL, soil lower limit (cm3 cm-3); SDUL, soil upper limit, drained (cm3 cm-3); SSAT, soil upper limit, saturated (cm3 cm-3); SRGF, soil root growth factor (unitless); SBDM, soil bulk density, moist (g cm3), corrected for gravel content; SLCF, soil coarse fraction or gravel cont ent (%); C:P, ratio of orga nic carbon to organic phosphorus (unitless). Table 3-8. Summary of inputs factors and outputs for the sensitivity analysis of the P model Nominal value Variability limits Input and output variable or parameter (medium value) LowerUpper Unit Inputs Soil Initial inorganic labile P 8 2 15 ppm Initial organic P 100 40 200 ppm P Fertilizer 30 0 60 kg P ha-1 Plant Maximum P uptake fraction 0.2 0.1 0.8 unitless Shoot P concentration Medium low high g / g Seed P concentration Medium low high g / g Outputs Total plant aboveground biomass kg ha-1 Grain yield kg ha-1 Total plant uptake of P kg ha-1

PAGE 101

101 Table 3-9. Specification of the di fferent levels of the input factors Shoot P and Seed P for the sensitivity analysis of the P model Shoot P concentration (g/g) Emergence End of leaf growth / Effective grain filling* Physiological maturity Low Optimum 0.0035 0.0013 0.0010 Minimum 0.0021 0.0008 0.0006 Medium Optimum 0.0070 0.0025 0.0020 Minimum 0.0040 0.0015 0.0010 High Optimum 0.0105 0.0038 0.0030 Minimum 0.0063 0.0023 0.0018 Seed P concentration (g/g) Low Optimum 0.0018 0.0018 0.0018 Minimum 0.0009 0.0009 0.0009 Medium Optimum 0.0035 0.0035 0.0035 Minimum 0.0018 0.0018 0.0018 High Optimum 0.0053 0.0053 0.0053 Minimum 0.0026 0.0026 0.0026 *End of leaf growth for shoot P a nd effective grain filling for seed P. Table 3-10. Main, interactions, and total sensitivity indices (unitle ss) of biomass for factors used in the sensitivity analysis Main SI Interactions SI Total SI PiLabile Organic P P fertilizer Frac LabileP Shoot P Seed P PiLabile 0.11 0.00 0.15 0.00 0.01 0.00 0.27 Organic P 0.00 0.00 0.00 0.00 0.00 0.00 0.00 P fertilizer 0.43 0.15 0.00 0.05 0.11 0.00 0.73 FracLabileP 0.04 0.00 0.00 0.05 0.00 0.00 0.09 Shoot P 0.04 0.01 0.00 0.11 0.00 0.00 0.15 Seed P 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FracLabileP: Fraction of Labile P that is soluble

PAGE 102

102 Table 3-11. Main, interactions, a nd total sensitivity indices (unitl ess) of grain yield for factors used in the sensitivity analysis Main SI Interactions SI Total SI PiLabile Organic P P fertilizer Frac LabileP Shoot P Seed P PiLabile 0.11 0.00 0.19 0.00 0.01 0.00 0.31 Organic P 0.00 0.00 0.00 0.00 0.00 0.00 0.00 P fertilizer 0.36 0.19 0.00 0.04 0.12 0.00 0.71 FracLabileP 0.03 0.00 0.00 0.04 0.01 0.00 0.08 Shoot P 0.02 0.01 0.00 0.12 0.01 0.00 0.15 Seed P 0.00 0.00 0.00 0.00 0.00 0.00 0.01 FracLabileP: Fraction of Labile P that is soluble Table 3-12. Main, interactions, a nd total sensitivity indices (unitl ess) of plant uptake of P for factors used in the sensitivity analysis Main SI Interactions SI Total SI PiLabile Organic P P fertilizer Frac LabileP Shoot P Seed P PiLabile 0.11 0.00 0.09 0.00 0.01 0.02 0.24 Organic P 0.00 0.00 0.00 0.00 0.00 0.00 0.00 P fertilizer 0.30 0.09 0.00 0.02 0.10 0.01 0.52 FracLabileP 0.03 0.00 0.00 0.02 0.00 0.00 0.05 Shoot P 0.09 0.01 0.00 0.10 0.00 0.00 0.20 Seed P 0.15 0.02 0.00 0.01 0.00 0.00 0.18 FracLabileP: Fraction of Labile P that is soluble Table 3-13. Main, interactions, a nd total sensitivity indices (unitl ess) of biomass for a special case of zero P fertilizer. The P fertilizer was also removed as a factor. Main SI Interactions SI Total SI PiLabile Organic P Frac LabileP Shoot P Seed P PiLabile 0.47 0.00 0.01 0.04 0.00 0.52 Organic P 0.00 0.00 0.00 0.00 0.00 0.00 FracLabileP 0.15 0.01 0.00 0.01 0.00 0.17 Shoot P 0.25 0.04 0.00 0.01 0.00 0.30 Seed P 0.00 0.00 0.00 0.00 0.00 0.00 FracLabileP: Fraction of Labile P that is soluble

PAGE 103

103 Table 3-14. Mean aboveground biomass, grain yi eld and total P uptake corresponding to each level of the input factors us ed in the sensitivity analys is. Each mean contains 243 = 729/3 observations. Input Factors Output Variables (kg ha-1) Biomass Grain yield Total P uptake PiLabile (ppm) 2 8282 3007 18 8 9885 3567 23 15 10704 3839 26 Organic P (ppm) 40 9634 3474 22 100 9608 3465 22 200 9629 3474 22 P fertilizer (kg ha-1) 0 6872 2604 15 30 10867 3885 25 60 11132 3925 27 Fraction of solution P (unitless) 0.1 8886 3228 20 0.2 9643 3493 22 0.8 10342 3693 24 Shoot P (See Table 3-12) Low 9084 3494 26 Medium 10401 3621 19 High 9387 3299 22 Seed P (See Table 3-12) Low 9488 3403 27 Medium 9738 3530 17 High 9646 3481 23

PAGE 104

104 Figure 3-1. Processes in the integrated soil-plant phosphorus model in DSSAT PSenesced Shoots PSenesced Roots PSurface SOM1 PSoil SOM1 PSoil SOM23 Metabolic PSoil Litter Structural Structural PSurface Litter Metabolic Soil Organic Phosphorus Processes Net P Mineralized PRoots PShells PSeeds PShoots Plant Phosphorus Processes Remobilization Partitionin g U p take NoRoot Labile Pi Fertilizer P NoRoot Active Pi NoRoot Stable Pi Root Labile Pi Root Active Pi Root Stable Pi Soil Inorganic Phosphorus Processes

PAGE 105

105 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 EmergenceEnd of leaf growthPhysiological Maturity Harvest growth StageP concentration (g P [g shoot]-1) Optimum Minimum Figure 3-2. Optimum and minimum P concentrati on in maize shoots used in the plant P model 0.000 5.000 10.000 15.000 20.000 25.000 30.000 EmergenceEnd of leaf growthPhysiological Maturity Harvest growth Stageplant N:P ratio Maximum Minimum Figure 3-3. Maximum and minimu m N:P ratios used in the plant module to limit uptake of P

PAGE 106

106 0.00 0.20 0.40 0.60 0.80 1.00 1.20 0102030405060708090100 days after plantingshoot P concentration (%) or P stress Optimum shoot P Minimum shoot P Actual shoot P P stress partitioning P stress photosynthesis Figure 3-4. Relationship between maize shoot P concentrations and P stresses affecting vegetative partitioning and photosynthesis 0 5 10 15 20 25 30 35 40 LowMediumHigh input factor leveltotal phosphorus upt ake (kg ha-1) PiLabile Organic P P Fertilizer Figure 3-5. Simulated plant total aboveground bi omass, grain yield and plant uptake of P at different levels of initial PiLabile, initial organic P and P fertiliz er. Error bars shown represent one standard deviat ion. A) Aboveground biomass. B) Grain yield. C) Plant uptake of P. 0 2000 4000 6000 8000 10000 12000 14000 16000 LowMediumHigh input factor leveltotal aboveground biomass (kg ha-1) PiLabile Organic P P fertilizer 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 LowMediumHigh input factor levelgrain yield (kg ha-1) PiLabile Organic P P Fertilizer A C B

PAGE 107

107 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 LowMediumHigh input factor levelgrain yield (kg ha-1) Uptake fraction Shoot P Seed P 0 5 10 15 20 25 30 35 40 LowMediumHigh input factor leveltotal phosphorus uptake (kg ha-1) Uptake fraction Shoot P Seed P Figure 3-6. Simulated total plant aboveground bi omass, grain yield and plant uptake of P at different levels of maximum uptake fraction, optimum shoot and seed P concentrations. Error bars shown represen t one standard deviation. A) Aboveground biomass. B) Grain yield. C) Plant uptake of P. 0 2000 4000 6000 8000 10000 12000 14000 16000 LowMediumHigh input factor leveltotal aboveground biomass (kg ha-1) Uptake fraction Shoot P Seed P A B C

PAGE 108

108 0.000.100.200.300.400.500.600.700.800.901.00 PiLabile Organic P P fertilizer Fraction of solution P Shoot P Seed P PiLabile*Organic P PiLabile*P fertilizer PiLabile*Fraction of solution P PiLabile*Shoot P PiLabile*Seed P Organic P*P fertilizer Organic P*Fraction of solution P Organic P*Shoot P Organic P*Seed P P fertilizer*Fraction of solution P P fertilizer*Shoot P P fertilizer*Seed P Fraction of solution P*Shoot P Fraction of solution P*Seed P Shoot P*Seed P Sensitivity index for total aboveground biomass Figure 3-7. Sensitivity indices for the si x input factors and their interactions

PAGE 109

109 0 2000 4000 6000 8000 10000 12000 03060 P fertilizer levels (kg ha-1)total aboveground biomass (kg ha-1) PiLabile=2 PiLabile=8 PiLabile=15 Figure 3-8. Simulated response of total plant aboveground biomass to phosphorus fertilizer at different levels of PiLabile 0 2000 4000 6000 8000 10000 12000 2815 Inorganic labile P levels (ppm)total aboveground biomass (kg ha-1) Fraction of labile P=0.10 Fraction of labile P=0.20 Fraction of labile P=0.80 Figure 3-9. Simulated response of total plant abov eground biomass to PiLabile at different levels of fraction of labile P

PAGE 110

110 CHAPT ER 4 FIELD TESTING OF THE DS SAT PHOSPHORUS MODEL Introduction The complex nature of relationships between components present in agricultural systems suggests the use of simulation techniques to study those systems rather than directly experimenting continuously on the systems them selves. Simulation models can assist with assessing alternatives and making decisions that would consume the entire career of an agronomist (Struif Bontkes and Wope reis, 2003; Matthew s et al., 2000). Because models developed in a certain e nvironment can be adapted and applied in different agroecological c onditions, the suitability of a model to simulate processes of interest is a major criterion in order to achieve meaningful inferences. Models have become so complex and been described with so many variables and pa rameters that their degrees of freedom have increased drastically. With an a ppropriate choice of input variable and parameter values they can be made to produce realistic outpu ts that can agree erroneously with real world measurements. Therefore, in addition to the suitability criteri on, model testing or evaluation using the right combination of inputs and parameters is an importa nt step that diagnoses the ability of the model to capture appropriately the essence of crop-envir onment interactions and th eir variability at the meso and micro scales. Annino and Russell (1981) underlined the risks associated with the application of simulation models that have not passed the test of a sound scientific assessment. They cited the use of an untested or invalid mode l as one of the seven most frequent causes of failure in many simulation modeling studies. To enable crop models in DSSAT to simula te phosphorus dynamics in cropping systems, a soil-plant phosphorus model was modified from initial studies by Da roub et al. (2003) and implemented in the software. Modifications appl ied to the Daroub et al version of the model

PAGE 111

111 include: 1) linkage of the mode l to the CENTURY module to allow simulations of organic P transformations in soils; 2) implementation of a generic, modular crop P module that is usable by all crops in DSSAT, and 3) addi tion of algorithms for initializa tion of the different phosphorus pools using measured soil phosphorus data. Daroub et al. (2003) reported th at the initial soilplant phosphorus model simulated with good a ccuracy P uptake for maize grown under acidic conditions when linked to the DSSAT CSM. Th e redesigned soil-plant phosphorus model still operates as an experimental version and has not been tested yet. This chapter is centered on evaluating the soil-plant phosphorus model describe d in chapter 3. The da tasets used for the evaluation of the model are two phosphorus expe riments conducted in Ghana and described in Chapter 2. Results from these experiment s are also discussed in Chapter 2. The main question addressed in this chapte r was: can the soil-plant phosphorus model simulate the responses of maize biomass and gr ain yield to different levels of phosphorus as observed in the field in Ghana? The objectives of the present chap ter are: 1) to describe select ed methods for evaluating the performance of the soil-plant phos phorus model; 2) to present an assessment of the ability of the soil-plant phosphorus model to simulate soil and crop conditions in two locations in Ghana (Kpeve and Wa) using those tools. To meet these objectives, some model parame ters (genetic coefficients describing the cultivar used and the fraction of labile P in so lution) were first calibrated using essentially the dataset from Wa and partially the dataset from Kpeve. The evaluation reported in this chapter focused on the following key outputs: grai n yield, aboveground biomass and shoot P concentrations.

PAGE 112

112 Materials and Methods The Kpeve and Wa datasets described in Chapter 2 were used to evaluate the performance of the model. The Soil-Plant Phosphorus Model The model simulates phosphorus transforma tions between 1) three inorganic pools: labile, active and stable; 2) two or ganic pools: active and stable, a nd 3) four plant parts: roots, shoots, shells, and seeds. The model was impl emented in DSSAT for CERES-Maize to enable the maize model to predict nitrogen and phos phorus-limited maize production as affected by cultivar, soil, weather, an d management information. The soil inorganic P module of the model si mulates phosphorus transformations between a labile, active and stable pool. The soil organi c P module simulates phosphorus transformations between a surface litter, a mi crobial pool, and a stable pool. The model accounts for the mineralization of organic P to inorganic pools and the immobilization of P to organic pools. Available phosphorus for uptake by plants is de scribed as being provided by the labile pool within 2 mm of plants roots. Phosphorus taken up by the plant is partitioned to seeds, shells and vegetative tissues. During the reproductive phase, phosphorus accumu lated in the vegetative tissues can be remobilized and translocated to seeds. Plan t growth is limited by phosphorus between two thresholds that are species-specific optimum a nd minimum concentrations of P defined at three stages in the growth of the plant. Phos phorus stress factors are computed to reduce photosynthesis, dry matter accumulation and partitioning. A sensitivity analysis of the model to some key phosphorus-related parameters established that the model responds well to phosphorus fertili zer applications on soils with low initial available phosphorus (Chapter 3). The analysis al so isolated the initial soil inorganic labile

PAGE 113

113 phosphorus and the fraction of P that is availa ble for uptake per day as two important soil parameters that have significant influence on major model outputs. Datasets for Testing the Model The datasets used to evaluate the model came from two phosphorus experiments carried out in Ghana in 2004 and 2006. A description of th e experiments is provided in Chapter 2. The treatments at Kpeve, 0P, 10P, 30P, 80P, received respectively 0, 10, 30, and 80 kg [P] ha-1. At Wa, the treatments were combinations of levels of 2 factors: nitrogen fertilizer, 3 levels, 0, 60, and 120 kg [N] ha-1; phosphorus fertilizer, 3 levels, 0, 60, and 90 kg [P] ha-1. The experiment implemented in Kpeve in 2006 did not respond to phosphorus although availa ble P measured as Bray-1 was low. It was found that the soil in Kpeve had relatively high organic matter content (1.8%) and other phosphorus forms that could have been made available to the plant during the growing season. This observation may help expl ain the lack of respons e to phosphorus observed at Kpeve. The second experiment conducted in Wa responded well to phosphorus and nitrogen fertilizer applications. Parameters and Inputs for the Model Tests The parameters for the model tests included genetic coefficients and phosphorus-related parameters. Inputs included soil and weather co nditions. The parameters and inputs used are described next. Weather conditions The experiment in Kpeve, Southern Ghana (6o 40.80 N, 0o 19.20 E, altitude 67 m above sea level, Figures B-1 and B-2) was conducted in 2006 during the primary rainy season (March to July). The site has a bimodal rainfall pa ttern with an average annual rainfall of 1300 mm falling in two rainy seasons, March to July and September to October (Figure 2-1). The average annual temperature is 28 degrees C.

PAGE 114

114 The experiment in Wa, Northern Ghana (10o3 N, 2o30 W, altitude 320 m above sea level, Figures B-1 and B-2) was carried out during the only rainy season in 2004. The rainfall pattern in Wa is unimodal. The average annual rainfall is 1100 mm falling main ly between April and September (Figure 2-3). The mean annual temperature in Wa is 27 oC. Soil conditions The soil in Kpeve has a sandy loam texture and is classified as Haplic Lixisol which has a dark grayish brown topsoil and grayish brown to brown subsoil (Adiku, 2006). Soil analysis (Table 2-15) showed that the soil has good organic carbon content and available phosphorus (Bray1) that is at the limit between sufficiency and deficiency (11.69 ppm). The relatively high Mehlich1 P (90.44 ppm) value obtained from a differe nt soil testing laboratory suggested that this soil may not be severely P-deficient. The soil in Wa has a loamy sand texture with very low levels of organic carbon, organic nitrogen, available P (Bray1) and exchangeable K (Table 2-18). Genetic coefficients The genetic coefficients for the cultivar used, Obatanpa were calibrated based on the growth and development data ob tained essentially from Wa fo r the high N and P treatments. Because the experiment in Kpeve was affected by a drought spell starting at silking, the dataset from this location was not quantitatively involved in the calibration of the genetic coefficients for Obatanpa. However, qualitative comparisons were us ed to ensure that the model predicted well the anthesis date (that was not affect ed by the drought) at Kpeve as well. Since the cultivar Obatanpa was described as a medium-maturing variety with a maturity period of 105-110 days (Anonymous, 1996), the ca libration starting values of the genetic coefficients (Table 4-1) were from a mediumduration cultivar taken from the DSSAT database of cultivars. The coefficients were manually ad justed until an agreement between simulated and

PAGE 115

115 measured days to anthesis and to physiological maturity, biomass and grain yield was obtained. The development coefficients (P1 and P5) were adju sted first using measured days to silking and physiological maturity from the Wa experiment. The growth coefficients G2 and G3 were calibrated next, using measured end-of-season bi omass and grain yield from Wa. The coefficient PHINT for thermal time between the appearances of two successive leaf tips was not changed because leaf number data was not available. Since water stress affected the phenology and probably the biomass and grain yiel d in the Kpeve experiment, the data from this experiment was not used in any genetic coefficient calibration. Th e calibrated coefficients for Wa was used with no further altering for model evaluation purposes at Kpeve. Phosphorus parameters Optimum and minimum P concentrations in ro ots, shells, seeds as well as maximum and minimum N:P ratios were taken from the litera ture (Jones, 1983; Probe rt and Okalebo, 1992; Daroub et al., 2003; Probert, 2004). Optimum shoot P concentration at different stages of growth was estimated using the follo wing equations (Jones, 1983): At emergence and end of leaf growth : Optimum shoot P concentration (%) = X108 0 684 0 At physiological maturity: Optimum shoot P concentration (%) = X 0056 0 238 0 Where: X is the growth stage. Emergence was defined as growth stage 0 (X = 0), end of leaf growth as growth stage 4, and physiological maturity as growth stage 10 (J ones, 1983). Minimum shoot P concentration was taken as 60% of the estimated optimum (Daroub et al., 2003). For the soil parameters, the rate constants for inorganic P transformation from labile to active pools (KLA), active to labile pools (KAL), and active to stable pools (KAS) were estimated from the value of the P availability index respec tively using equations 35, 3-6 and 3-7. The P

PAGE 116

116 availability index was approximated as 0.40 (Table 3-1, Other soils). The fraction of soluble P was adjusted until the lowest error between simu lated and measured grain yield was obtained using the Wa dataset. Initial conditions Initial PiLabile was calculated from measured P Bray1 and exchangeable K (Table C-1) as (1.09*PBray1) + (10.59*ExchangeableK) + 2.71 for both sites, Kpeve and Wa. Initial PiActive and PiStable were calculated using equations 3-4 and 3-7. Initial total organic P was calcu lated from the values of or ganic carbon and pH (Tables 215 and 2-18 as OrganicC pHe e 10 0 12 10 5 11 9002 (Singh, 1985). This total organic P was partitioned as 6% active and 94% stable (Par ton et al., 1988, 1994; Gijsman et al., 2002). Measured soil parameters that included soil organic carbon a nd nitrogen (Tables 2-15 and 2-18) were used as input to th e crop model. Other soil parameters not measured but necessary to run all DSSAT models were estimated using pe dotransfer functions in DSSAT. Soils water lower limit (SLLL), drained upper (SDUL) and upper limit saturated (SSAT) for Kpeve were taken from Adiku (2006). The bulk density used at Kpeve was corrected fo r gravel content using equation 2-1. At Wa, the SLLL, SDUL, SSAT an d bulk density values were estimated using pedotransfer functions in DSSAT. Initial soil water at planting was set at the SDUL level for both sites. Initial nitrate-N and ammonium-N were not measured and assumed to be 0.01 ppm. Model Evaluation Simulation of exact real world values by mode ls would not generally be expected because of the many simplifications with which the mode l approximates reality. The primary concern of model evaluation is comparing simulations and measurements and explaining possible

PAGE 117

117 deviations. In this study, attenti on will be given to the analysis of these deviations to assess the model performance and gradually introduce modi fications to get more understanding of the causes of simulation error. The evaluation presented here was a first step in testing the ability of the model to mimic the wide differences in responses to P. Model evaluation tools Simple scatter plots were used wherever appr opriate to stimulate in tuitive and preliminary evidence of model performance. The simulations and measurements were compared globally using a standardized mean deviation, the root mean square error (RMSE): 5 0 2 N t Measuremen Simulation RMSE (4-1) Where N is the number of pairs of measurements and simulations. The RMSE estimates the dispersion between si mulated and measured data (Du Toit et al., 1997). The RMSE can be expressed rela tive to the mean of measur ements to visualize how the deviation compares to the average observation: 100 M RMSE RRMSE (4-2) Where RRMSE is the relative RMSE (in percent) and M is the mean of measurements. Deviations between simulations and measur ements can be furthermore explored by partitioning the overall RMSE into components that relate to specific types of discrepancies (Kobayashi and Salam, 2000; Gauch et al., 2003). If the simulations and the measurements agreed perfectly, the (simulation, measurement) pairs of points would be aligned along the 1:1 line in a scatter plot and the RMSE would be equal to zero. This perfect agreement situation would mean the following: 1) the mean of simulations, S, equals the mean of measurements, M;

PAGE 118

118 2) if a regression analysis was performed, the sl ope of the equation would be equal to 1; and 3) the coefficient of determination R2 resulting from a simple linear regression analysis would be equal to 1. An RMSE different from zero can th erefore be envisioned as the result of three potential problems: Problem 1: The model failed to simulate the mean of measurements, introducing a simulation bias: there is shift in the fitted re gression line from the original perfect agreement line. This situation can be quantified by the Squared Bias: 2M S SB (Kobayashi and Salam, 2000). SB reveals a possible trend of the mode l to overestimate or underestimate the measurements. Problem 2: The deviation is the result of the model failing to simulate correctly the magnitude of fluctuation among the measurements : there is rotation of the fitted regression line around the perfect agreement line with the axis of rotation passi ng through the origin. This condition can be measured by the Squared Difference between the Standard Deviations, SDSD, 2 m sSD SD SDSD (4-3) Where SDs is the standard deviation of simulations and SDm is the standard deviation of measurements. Problem 3: The deviation is attributable to the failur e of the model to simulate the pattern of the fluctuation across the measur ements: the pairs of points woul d appear in a random pattern in a scatter plot. This situation can be quantified by the Lack of positive Correlation weighted by the standard deviations, r SD SD LCSm s 1 2 (4-4) Where r is the Pearson coefficient of correlation.

PAGE 119

119 The LCS can also be interpreted as the resi dual error sum of squa res after removing SB and SDSD. Kobayashi and Salam (2000) found that the three components SB, SDSD and LCS add up to the Mean Squared Error, MSE LCS SDSD SB MSE (4-5) Since 2) (RMSE MSE equation (4-5) can be rewritten to relate the RMSE to the components of the model error: LCS SDSD SB RMSE 2) ( (4-6) The advantage of using the partitioned MSE resi des in the possibility to investigate what components of the overall model deviation were most important. Results and Discussion The soil-plant P model was able to capture the response of maize to P fertilizer as observed at both sites. Results of genetic coefficients calibration and P-related parameters estimation are also described next. Weather The crop in Kpeve experienced a drier than aver age July (2006) (Figure 2-1) that affected maize phenology and growth. The major season, whic h normally ends in la te July, ended earlier (in June) at a critical stage during crop growth. The total rain fall in July was only 40.40 mm, which was below the calculated 2003-2005 averag e for that month (Figure 2-1). Although the total rainfall received in 2006 from planting to harvest was higher than the amount received during the same period in 2003-2005 (713 mm in 2006, 464 mm in 2005, 612 mm in 2004, and 526 mm in 2003), the rainfall received in the mont h of July 2006 was low: the July total rainfall was 60 mm in 2003, 105 mm in 2004 and 26 mm in 2005 compared to 40.40 mm during the year of the experiment.

PAGE 120

120 As a response to this unexpected drought, the field was sprinkler-watered for three days from July 26th to July 28th (60 to 62 days after planting). Be cause irrigation equipment was not set up on the field at the commencement of the trial, the sprinkler-watering was improvised, which delayed the water application for about 10 days after drought symptoms were first observed, and provided only about 15 mm of wate r. The adverse effects of rainfall variability and unreliability at Kpeve in recent years were also point ed out by Adiku (2004 and 2006). Genetic Coefficients Thermal time related to days to anthesis (P1) was set (Table 4-1) so that the model could simulate correctly the measured anthesis dates at Kpeve and Wa. Thermal time related to days to physiological maturity (P5) was set (Table 4-1) so that the model could simulate correctly the measured physiological maturity dates at Wa. Th e potential kernel number per plant (G2) was increased from 700 to 900 and the potential kernel growth rate (G3) decreased from 8.50 to 6.50 mg/day to obtain the best fit to the measured grain yield at Wa. Phosphorus Parameters Optimum and minimum P concentrations in roots, shells, seeds and maximum and minimum N:P ratios taken from the literature (Jones, 1983; Probert and Okalebo, 1992; Daroub et al., 2003; Probert, 2004) are pr esented in Table 4-2. Estimate d optimum and minimum shoot P concentrations using the equations developed by Jones (1983) are presented in Table 4-2. To reflect the fact that phosphorus stress should affect vegetative partitioning before photosynthesis, the minimum value of the ratio of P in vegetative tissue to the optimum P below which reduced photosynthesis occurs was set to 1. 0, and the minimum value of the ratio of P in vegetative tissue to the optimum P below which vegetative partitio ning will be affected was set to 0.8. It was assumed that the maximum fraction of P which can be mobilized from shoot per day cannot exceed 0.10.

PAGE 121

121 The estimated soil P parameters were identical for both Kpeve and Wa (Table 4-3). This was because all of the soil P parameters (except th e fraction of labile P in solution) depend on the value of P availability index. Th e dependency of P availability i ndex on Initial PiLabile as shown in the equation 0.40 + 0.00023Initial PiLabile is weak and the calculati on essentially yields 0.40 for Initial PiLabile values 1 ppm (Initial PiLabile was 16.52 ppm at Kpeve and 6.49 ppm at Wa). The adjustment to the fraction of labile P in solution was challenging. Specific studies were not conducted on this fraction in the way it is used in the soil -plant phosphorus model discussed here. Studies that suggested a value of 0.015-0.020 related the fr action directly to the total PiLabile pool (Daroub et al., 2003). Sin ce in the present phosphorus model, the fraction applies to the part of PiLabile in the root zone only, it was certain that the calibrated value of this fraction would be higher than 0.020. Through calib ration using the Wa da taset a value of 0.20 was obtained. This value was also used to evaluate the model at Kpeve. A sensitivity analysis on this fraction of labile P in solution showed that this parameter did not have as much influence on model outputs as the size of the initial PiLabile pool itself and the optimum shoot P concentration (Tables 3-10 and 3-13). Initial Conditions Initial sizes of the different phosphorus pools for both sites are given in Table 4-4. The initial PiLabile in the soil at Kpeve was nearly three times that of Wa (Table 4-4). Organic P was relatively high at Kpeve (Table 44). Other soil parameters are summarized for Kpeve in Table 4-5 and for Wa in Table 4-6. Model Evaluation at Kpeve The soil-plant P model was able to capture the lack of response to P as observed in the experiment at Kpeve.

PAGE 122

122 In-season growth Simulation of accumulated biomass over time was in good agreement with measurements (Figure 4-3). The RMSE of 87 kg ha-1 at 17 dap increased with biomass over time but remained at about 470 kg ha-1 between 31 dap and final harvest (108 dap). The RMSE increase during the season was due to increasing biomass values. In relative terms, the simulation actually improved over time. The RRMSE was only 5% at final harvest (Table 4-7). The early season error was mostly due to an ove rprediction by the model that resulted in a high squared bias (Figures 4-4 and 4-3). At anthes is (52 dap) and final harvest (108 dap), the SB was less and the error due to the pattern of va riation among the measurements (LCS) became the important component of MSE (Figur e 4-4), but the overall errors were actually small (Table 4-7). The negative correlation coefficients observed at 17 dap and anthesis ( 52 dap) were caused by two opposite trends in the variati on of the biomass: measured bi omass decreased while simulated biomass increased at those periods with incr easing P applications. The observed decreasing biomass with P additions was not significant. Final grain yield Predicted grain yields were in good agreem ent with measurements (Figure 4-1). The RMSE was 255 kg ha-1 representing 8% of the mean of measurements. The model captured well the lack of response to P at Kpeve as shown by the non significant differences among the measured grain yields (Figure 4-2) even though the growth data from Kpeve was not used in calibrating the genetic coefficients of the cultiv ar Obatanpa. The simulation error was mainly due to the pattern of variat ion of grain yield among the four trea tments (Figure 4-2). This was also reflected in the low correlation coefficient ob served between measurements and simulations (0.36). The statistical non significance of the gr ain yield means that th e slight differences observed in grain yield among the four treatm ents were not determined by the phosphorus

PAGE 123

123 applied but rather to other causes such as m easurement error or fiel d variability. Since a deterministic model does not account for such fluctuations, the low RMSE suggested good performance for final yield. Wa At Wa, the model predicted th e response of maize to both nitrogen and phosphorus with higher error than Kpeve. In-season growth The aboveground biomass was underpredicted by the model at most planting dates in all treatments (Figure 4-9). The RMSE varied with sampling date between 216 and 2574 kg ha-1, which corresponded to 19-57% RRMSE values (Tab le 4-7). However, th e different components of the MSE showed that the general tendency of the model to underpredict the biomass did not actually affect its ability to effectively capture most of the responses of biomass to nitrogen or phosphorus fertilizer. The LCS or the SDSD, whic h correspond to the failure of the model to simulate correctly the pattern or the magnitude of fluctuation among the measurements, generally represented the smallest portion excep t for days after planting 46 (Figure 4-8). The correlation coefficient between simulated and measured biomass at each sampling date can lead to misleading interpretations when used alone. For example, the correlation coefficient had the same value of 0.97 at days 46 and 61 af ter planting. However, the RRMSE doubled from 46 to 61 dap (from 24 to 45%) (Table 4-7). The increase in the RRMSE is an indicator of a progression towards a poorer perfor mance of the model, but at the same time the persistence of a high correlation coefficient suggests that the st rong linear association between simulations and measurements has not been lost. The problem w ith the use of correlation and linear regression alone for model evaluation is that when simulati ons and measurements are treated as dependent and explanatory variables, many assumptions of the analysis are violated (Mitchell, 1997).

PAGE 124

124 In-season shoot P concentration The performance of the simulation of shoot P concentration at Wa depended on N and P fertilization of the crop: 1) When neither nitrogen or phosphorus were applied (tr eatment 0N 0P, Figure 4-10), simulation of shoot P concentra tion followed a similar pattern as in phosphorusfertilized treatments (Figure 410); 2) In the no phosphorus trea tments that received nitrogen (treatments 60N 0P, 120N 0P, Fi gure 4-10), simulated shoot P c oncentration was less variable and remained closer to the minimum shoot P con centration than the measur ements (Figure 4-10); In treatments that received both nitrogen and phosphorus fertilizer, simulated and observed shoot P concentration were sim ilar during the vegetative phase (Figure 4-10). After this phase, simulated shoot P concentration remained stable at a higher level than measured (Figure 4-10). These response patterns are summarized in Figure 4-11. At least three problems of incompatibility between simulations and measur ements are highlighted in this figure: 1) Simulated shoot P concentration in the 0P treatment was lower th an in the 60 and 90P treatments, which logically reflects low availabl e soil P in the 0P treatment, low P uptake and low P status in the plant due to high P stress on biomass growth. Th is is the expected relationship between shoot P concentrations in plants grown on P-limiting and non P-limiting soils as found in several studies surveyed in Jones, 1983. In the experiment reported here, the shoot P concentrations did not show much variati on between P-limiting and non P-limiting conditions (Figure 4-11). Shoot P concentration varied in the experiment between 0.58% and 0.05% on average regardless of the treatment consider ed. 2) The measured decrease in shoot P concentration from 0.57 at 28 dap to 0.05% at 125 dap was also in contrast to the simulated 0.50 to 0.20% during the same period in the well-su pplied phosphorus treatments (Figure 4-11). Inseason variability in shoot P concentration found in other studies under non P-limiting conditions is 0.50% 0.25% (Plenet et al., 2000a) and 0. 45% 0.15% (Ziadi et al., 2007). The shoot P

PAGE 125

125 concentration measured at maturity (125 dap) was lower than any of these values in the experiment reported here (0.05%). Final grain yield The grain yield in Wa was simulated w ith a low error. The RMSE of 266 kg ha-1 represented 13.6% of the mean measurement. The mean difference between simulations and observations (bias) was only 3 kg ha-1. The simulations agreed well with the data (Figure 4-5). The model was able to capture the three yiel d ranges that are dependent on the amount of nitrogen applied with a correlation coefficient of 0.99 (Figure 4-5). The differential responses to nitrogen and phosphorus fertilizer were equally well simulated (Figure 4-6). The decomposition of the error revealed that much of the overall MSE could essentially be partitioned among the SDSD and the LCS (Figure 4-7). The SB was very small because the mean measurement was well predicted (mean of measurements = 1961 kg ha-1; mean of simulations = 1958 kg ha-1). The prediction of the measured variance of yield was also good (s tandard deviation of measurements = 1336 kg ha-1; standard deviation of simulations = 1472 kg ha-1). The variances of the grain yield themselves were generally high because of the wide range in fertilizer inputs (0, 60 and 120 kg ha-1 for nitrogen and 0, 60 and 90 kg ha-1 for phosphorus). These high grain yield variances reflected the low nitrogen and phosphor us status of the soil prior to the start of the experiment, which made the soil responsive to the application of either nutrient. The relatively high LCS (compared to the other mode l error components) does not mean, in this particular situation, that the m odel failed to simulate correctly the pattern of variation among the measurements because 1) the overall simulation error was low and 2) the LCS is a weighted product of the standard deviations which ar e inherently high in this experiment.

PAGE 126

126 Conclusion The assessment presented in this paper s howed that the soil-p lant phosphorus model simulated maize grain yield and biomass with a good degree of accuracy both under phosphoruslimiting (Wa) and non phosphorus-limiting (Kpeve) conditions in Ghana. Grain yield was simulated with an RRMSE of 8% at Kpeve and 14% at Wa. Final biomass was simulated with an RRMSE of 5% at Kpeve and 30% at Wa. The highe r errors at Wa were mostly due to more bias in biomass simulati ons, but the model actually simulated well the response to P fertilizer. Simulation of shoot P concentration at Wa wa s generally good and in agreement with inseason shoot P variability found in the literature. However, the shoot P concentrations measured in the Wa experiment at 81 dap and harvest matu rity (125 dap) were exce ptionally low (0.05%). The soil-plant P model captured the observed re sponse to P fertilizer at Wa, and lack of response to P fertilizer at Kpeve. These results are promising because this is a first evaluation of the model across two contrasting conditi ons of P availability to plants. However, the calibration of the fraction of labile P in solution was challenging mostly because it is difficult to measure directly and sufficient information was not available in the literature. Different values of this parameter might work better on other soil types. This parameter was found to be influential on plan t P uptake, biomass and grain yield when P fertilizer was not applied. Methods for indirect estimation the initial si zes of the inorganic a nd organic P pools that play an important role in the response of the model to P have uncertainties associated with them. Model performance can be expected to improve as refinements are intr oduced in these methods.

PAGE 127

127 Table 4-1. Growth and development genetic coeffi cients for the Obatanpa cultivar used at both sites, Kpeve and Wa, for testing the phosphorus model Definition DSSAT ID Starting Value Obatanpa Degree days (base 8oC) from emergence to end of juvenile phase P1 200 300 Photoperiod sensitivity P2 0.00 0.00 Degree days (base 8oC) from silking to physiological maturity P5 800 830 Potential kernel number (/plant) G2 700 900 Potential kernel growth rate (mg/day) G3 8.50 6.50 Phyllochron PHINT 38.90 38.90

PAGE 128

128 Table 4-2. Plant parameters used for test ing the phosphorus model at Kpeve and Wa Parameter Unit Value Shoot P concentrations Optimum shoot P concentration at emergence % 0.70 Optimum shoot P concentration at tasseling % 0.25 Optimum shoot P concen tration at physiological maturity % 0.20 Minimum shoot P concentration at emergence % 0.40 Minimum shoot P concentration at tasseling % 0.15 Minimum shoot P concen tration at physiological maturity % 0.10 Root P concentrations Optimum root P concentration at emergence % 0.041 Optimum root P concentration at effective grain filling % 0.041 Optimum root P concentration at physiological maturity % 0.041 Minimum root P concentration at emergence % 0.020 Minimum root P concentration at effective grain filling % 0.020 Minimum root P concen tration at physiological maturity % 0.020 Shell P concentrations Optimum shell P concentration at emergence % 0.50 Optimum shell P concentration at effective grain filling % 0.50 Optimum shell P concen tration at physiological maturity % 0.050 Minimum shell P concentration at emergence % 0.25 Minimum shell P concen tration at effective grain filling % 0.25 Minimum shell P concentration at physiological maturity % 0.025 Seed P concentrations Optimum seed P concentration at emergence % 0.35 Optimum seed P concen tration at effective grain filling % 0.35 Optimum seed P concentration at physiological maturity % 0.35 Minimum seed P concentration at emergence % 0.175 Minimum seed P concentration at effective grain filling % 0.175 Minimum seed P concentration at physiological maturity % 0.175 N to P ratios Maximum vegetative N:P ratio at emergence unitless28.0 Maximum vegetative N:P ratio at effective grain filling unitless15.0 Maximum vegetative N:P ra tio at physiological maturity unitless9.3 Minimum vegetative N:P ratio at emergence unitless4.2 Minimum vegetative N:P ratio at effective grain filling unitless2.7

PAGE 129

129 Table 4-2 continued Minimum vegetative N:P ra tio at physiological maturity unitless2.1 P mobilization and stress Maximum fraction of P which can be mobilized from shoot per day unitless0.10 Minimum value of the ratio of P in vegetative tissue to the optimum P below which reduced photosynthesis will occur unitless0.80 Minimum value of the ratio of P in vegetative tissue to the optimum P below which vegetative partitioning will be affected unitless1.00 Source: Jones, C.A. 1983. A survey of the variability in tissue nitrogen and phosphorus concentrations in maize and grain sor ghum. Field Crops Research 6, 133-147. Jones, C.A., Cole, C.V., Sharpley, A.N. W illiams, J.R., 1984a. A simplified soil and plant phosphorus model: I. Documentation. Soil Scie nce Society of America Journal 48, 800805. Daroub, S.H., Gerakis, A., Ritchie, J.T., Frie sen, D.K., Ryan, J., 2003. Development of a soilplant phosphorus simulation model for calcareous and weathered tropical soils. Agricultural Systems 76, 1157-1181. Probert, M.E., 2004. A capability in APSIM to m odel phosphorus responses in crops. In: Delve, R.J. Probert, M.E. (Eds), Modelling Nutrie nt Management in Tropical Cropping Systems. ACIAR Proceedings No. 114. pp. 92-100. Table 4-3. Soil parameters used for testing th e phosphorus model at Kpeve and Wa. The values correspond to the top layer of the soils (0-10 cm for Kpeve and 0-20 cm for Wa) Parameter Unit Kpeve Wa Rate constant for transformation from labile P to active P d-1 0.03674 0.03674 Rate constant for transformation from active P to labile P d-1 0.00490 0.00490 Rate constant for transformation from active P to stable P d-1 0.00043 0.00043 Rate constant for transformation from stable P to active P d-1 0.00010 0.00010 P availability index unitless 0.40 0.40 Fraction of root labile inorganic P that is soluble unitless 0.20 0.20

PAGE 130

130 Table 4-4. Values of additional inputs required to run the soil-plant phosphorus model for the Kpeve and Wa experiments. Values correspond to the top layer of the soils (0-10 cm for Kpeve and 0-20 cm for Wa) Input Unit Kpeve Wa Initial labile inorganic P ppm 16.52 6.49 Initial active inorganic P ppm 123.91 48.70 Initial stable inorganic P ppm 495.66 194.82 Initial active organic P ppm 7.99 2.22 Initial stable organic P ppm 125.15 34.78 P in residue (if applied) % 1.1 Not measured P fertilizer (if applied) kg ha-1 0, 10, 30, and 80 0, 60, and 90 Soil CEC cmol kg-1 17.8 10.0 Soil Clay % 18.3 7.5 Source: estimated from soil compos ition data from the experiments. Table 4-5. Estimated initial condition soil parameters for Kpeve SLB SLLL SDUL SSAT SRGF SBDM C:P SLTX 10 0.180 0.260 0.460 1.000 0.83 138 Sandy Loam 20 0.070 0.140 0.280 1.000 1.08 136 Loam 30 0.040 0.080 0.160 0.607 1.47 130 Sandy Loam 40 0.060 0.120 0.240 0.497 0.74 138 Clay 50 0.040 0.060 0.120 0.407 0.47 123 Sandy Clay 60 0.050 0.090 0.180 0.333 0.56 218 Sandy Clay 70 0.080 0.150 0.300 0.273 0.97 200 Sandy Clay 80 0.060 0.110 0.220 0.223 0.77 127 Sandy Clay Loam 90 0.090 0.160 0.320 0.183 1.04 124 Sandy Clay SLB, depth, base of soil layer (cm); SLLL, so il lower limit (cm3 cm-3); SDUL, soil upper limit, drained (cm3 cm-3); SSAT, soil upper limit, sa turated (cm3 cm-3); SR GF, soil root growth factor (unitless); SBDM, soil bulk density, moist (g cm3), corrected for gravel content; C:P, ratio of organic carbon to organic phosphorus (unitless); SLTX, so il texture (unitless). Table 4-6. Estimated initial condition soil parameters for Wa SLB SLLL SDUL SSAT SRGF SBDM SSKS SLTX 20 0.085 0.155 0.383 1.000 1.54 2.59 Loamy Sand 40 0.122 0.190 0.362 0.549 1.57 2.59 Sandy Loam 60 0.124 0.170 0.204 0.368 1.52 0.12 Sandy Clay 90 0.059 0.079 0.088 0.223 1.38 0.06 Clay SLB, depth, base of soil laye r (cm); SLLL, soil lower limit (cm3 cm-3); SDUL, soil upper limit, drained (cm3 cm-3); SSAT, soil upper limit, saturated (cm3 cm-3); SRGF, soil root growth factor (unitless); SBDM, soil bulk density, moist (g cm3); SSKS, saturation hydraulic conductivity (cm h-1); SLTX, soil texture (unitless).

PAGE 131

131 Table 4-7. Summary of aboveground biomass error statistics for the Kpev e and Wa experiments Kpeve Days after planting 17 31 52 108 (harvest) RMSE (kg ha-1) 87 475 470 470 RRMSE (%) 83 51 9 5 Correlation Coefficient -0.63 0.88 -0.88 0.53 Wa Days after planting 28 46 61 81 125 (harvest) RMSE (kg ha-1) 216 481 2574 1048 1479 RRMSE (%) 57 24 45 19 30 Correlation Coefficient 0.74 0.97 0.97 0.99 0.99

PAGE 132

132 0 1000 2000 3000 4000 0103080P level (kg ha-1)grain yield at harvest (kg ha-1) measured simulated Figure 4-1. Comparison of simulated and measured grain for different ph osphorus levels in the Kpeve experiment 7.94 11.40 80.66 0 20 40 60 80 100 Bias SquaredSDSDLCSMSE components (%) Figure 4-2. Decomposition of th e grain yield MSE for the Kpev e experiment, using the method developed by Kobayashi and Salam (2000)

PAGE 133

133 0 3000 6000 9000 12000 173152108 days after plantingtotal biomass (kg ha-1) measured simulated 0 3000 6000 9000 12000 173152108 days after plantingtotal biomass (kg ha-1) measured simulated 0 3000 6000 9000 12000 173152108 days after plantingtotal biomass (kg ha-1) measured simulated 0 3000 6000 9000 12000 173152108 days after plantingtotal biomass (kg ha-1) measured simulated Figure 4-3. Comparison of simulated and measur ed biomass on four samples taken during the season for the four treatments tested in Kpeve. A) Treatment 0P. B) Treatment 10P. C) Treatment 30P. D) Treatment 80P. 0 20 40 60 80 100 173152108 days after plantingMSE and components (% ) Squared Bias SDSD LCS Figure 4-4. Decomposition of the in-season biom ass MSE for the Kpeve experiment, using the method developed by Kobayashi and Salam (2000) A B C D

PAGE 134

134 Figure 4-5. Comparison of measured and simulate d maturity grain yield obtained in the Wa experiment using the 1:1 line 0 1000 2000 3000 4000 5000 0P60P90P0P60P90P0P60P90P 0N60N120N treatments (kg ha-1)maturity grain yield (kg ha-1) measured simulated Figure 4-6. Measured and simulate d responses of maturity grain yi eld to different combinations of nitrogen and phosphorus levels in the Wa experiment 0 1000 2000 3000 4000 5000 010002000300040005000 measured grain yield (kg ha-1)simulated grain yield (kg ha-1)

PAGE 135

135 0.01 26.32 73.67 0 20 40 60 80 100 Bias SquaredSDSDLCSMSE components (%) Figure 4-7. Decomposition of th e grain yield MSE for the Wa experiment, using the method developed by Kobayashi and Salam (2000) 0 20 40 60 80 100 28466181125 days after plantingMSE and components (% ) Squared Bias SDSD LCS Figure 4-8. Components of the biomass MSE for the Wa experiment at five sampling times

PAGE 136

136 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated 0 4000 8000 12000 28466181125 days after plantingtotal biomass (kg ha-1) measured simulated Figure 4-9. Measured and simula ted responses of cumulative biom ass to different combinations of nitrogen and phosphorus levels in the Wa experiment. A) Treatment 0N 0P. B) Treatment 0N 60P. C) Treatment 0N 90P. D) Treatment 60N 0P. E) Treatment 60N 60P. F) Treatment 60N 90P. G) Treatme nt 120N 0P. H) Treatment 120N 60P. I) Treatment 120N 90P. A B C D E F G H I

PAGE 137

137 0.00 0.20 0.40 0.60 0.8028466181125days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated 0.00 0.20 0.40 0.60 0.80 28466181125 days after plantingshoot P concentration (%) measured simulated Figure 4-10. Measured and simulated responses of shoot P concentration to different combinations of nitrogen and phosphorus leve ls in the Wa experiment. A) Treatment 0N 0P. B) Treatment 0N 60P. C) Treatme nt 0N 90P. D) Treatment 60N 0P. E) Treatment 60N 60P. F) Treatment 60N 90P G) Treatment 120N 0P. H) Treatment 120N 60P. I) Treatment 120N 90P. 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 28466181125 days after plantingmeasured s hoot P concentration (%) 0 P 60 P 90 P 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 28466181125 days after plantingsimulated shoot P concentration (%) 0 P 60 P 90 P Figure 4-11. Variation of the shoot P concentrat ion during plant growth as affected by three phosphorus levels in the Wa experime nt. A) Measured. B) Simulated. A B C D E F G H I A B

PAGE 138

138 CHAPTER 5 SUMMARY AND CONCLUSIONS The soil-plant phosphorus model in the DSSAT CSM integrates information on phosphorus in soils and plants to simulate phosphorus transformati ons in soils and their effects on plant production. Information on soil phosphorus includes the quantity of inorganic, readilyavailable phosphorus (labile P), slowly availabl e phosphorus (active P), very slowly available phosphorus (stable P) and orga nic phosphorus. Transformation constants control the way phosphorus is moved among these pools. The model differentiates between soils with different P sorption capacities to partition fertilizer applied to the inor ganic P pools. A fraction of the phosphorus in the readily-available pool becomes soluble and may be taken up on any day that a plant is growing on the soil. Information on the plant includes optimum and minimum phosphorus concentration in differe nt plant parts (roots, shoots, shells and seeds). The plants demand for phosphorus is estimated as the P defi cit relative to a seasona lly-varying optimum P concentration. This demand is satisfied by P upt ake from the readily available inorganic P pool. If this uptake is not sufficient to meet the demand of seeds present, phosphorus can be removed from other vegetative organs. Phosphorus not removed with harvest constitutes a capital investment in the soil in the organic form. A sensitivity analysis of the model, limited to six key factors, showed that P fertilizer application and the initial value of the readily available P were the most important P-related inputs affecting the predictability of plant biom ass, yield and P uptake. The fraction of readilyavailable P that is soluble, the shoot and seed P were also influential but to a smaller extent. However, these parameters have more influe nce on the model outputs in the absence of P fertilizer. Accurate predictions require therefore that at least initial readily available P be measured or estimated correctly. In this regard, different names of readil y available P have been

PAGE 139

139 used in the literature and can be the source of model input error. In the DSSAT soil-plant phosphorus model, the readily available P that provides so luble P for plant uptake is approximately the inorganic labile P extracted wi th resin. If the resin P measurement is not available but any of the following extractants were used to measure available P, Bray1, Colwell, Mehlich1, Morgan, Olsen, Truog, and water, the model will use empirical relationships from an expert system to indirectly estimate the readily available inorganic P. The correct specification of the quantity of fertilizer app lied can become another source of error. Although in many agronomic experiments, P fertilizer a pplication is expressed as phosphate (P2O5), the amount of phosphorus applied is expressed as pure P in th e model. There is a 2.29 factor for converting between the two. The contrasting results obtained from the two experiments used to evaluate the phosphorus model provided an ideal situ ation for testing the robustn ess of the model under opposite conditions. The available phosphorus (Bray1) was re latively low at Kpeve (southern Ghana) but other important phosphorus sources such as chemi cal contributions of or ganic matter (organic matter content in the soil top 20 cm at Kpeve was 1.8%) not accounted for by the Bray1 extraction could have been res ponsible for high indigenous phos phorus supply in the soil. No significant difference in measured plant phenol ogy, aboveground biomass, green leaf area and grain yield was found between fertilized and unfertilized treatments at this site. The soil at Wa (Northern Ghana) was relatively low both in available P (2.5 ppm Bray-P in the top 20 cm) and organic carbon (0.49% in the top 20 cm). Mai ze responded well to phosphorus fertilizer application on this soil. Leaf area index and aboveground biomass were low in no nitrogen and no phosphorus treatments throughout the season. The highest reduction in leaf area index and biomass occurred at the same time, which supports the reported finding that poor

PAGE 140

140 biomass accumulation in P deficient conditions is associated with reduced photosynthetically absorbed radiation by the plant, due to reduced leaf area. The re duction in grain yield could have been a result of indirect and direct e ffects of N and P stress on photosynthesis. Testing of the phosphorus model under bot h P-limitation (Wa) and no P limitation (Kpeve) conditions showed that plant biomass and grain yield were quite predictable. Grain yield was simulated with an RRMSE of 8% at Kpev e and 14% at Wa. Final biomass was simulated with an RRMSE of 5% at Kpeve and 30% at Wa. Although the simulation skill was lower at Wa, the model reasonably captured the response of biom ass and grain yield to P fertilizer at both sites. The soil-plant phosphorus model described, anal yzed and tested with field data performed acceptably well over specific and known soil phos phorus conditions. The potential exists for using the model as an application tool or in decision-support because model simulation of crop response to P fertilizer is promis ing. However, the current level of confidence in the model must be enhanced through further tes ting and validation studi es. Some P model parameters are highly uncertain and must be estimated from other, more easily measurable variables. For example, the initial inorganic labile P that has a major influence on crop respons e need greater precision in its estimation. This confidence raising process includ es: 1) verification or re-verification of the model; 2) more accurate estimation of the inorga nic labile P from measured available P when new data become available for calibration of the expert system; 3) special study on the estimation of the fraction of inorganic labile P that is soluble for a specific soil and how this fraction changes with soil properties like the P-sorbing capacity.

PAGE 141

141 APPENDIX A MEASURED GROWTH DATA AT KPEVE Table A-1. Monthly total rainfall in 2006 (one standard deviation of rainfall), mean daily solar radiation, and mean daily temperature collected during the Kpeve experiment in 2006 Month Rain (mm) Solar Radiation (MJ m-2 day-1) Maximum Temperature (oC) Minimum Temperature (oC) March 107.6 (7.4) 14.8 35.2 23.4 April 84.0 (8.8) 14.0 35.0 24.3 May 257.4 (12.3) 14.4 32.7 22.8 June 202.4 (18.3) 14.5 31.5 22.7 July 40.4 (3.6) 11.6 30.2 22.9 August 21.0 (1.8) 10.5 30.0 22.7 Table A-2. Days to tasseling (one standard deviation of four repli cations), days to anthesis (one standard deviation of four replications), and days to silking (one standard deviation of four replications) for the experiment in Kpeve, Ghana P level (kg ha-1) Tasseling (day) Anthesis (day) Silking (day) 0 48 (1.0) 51 (0.6) 59 (3.9) 10 49 (0.8) 51 (1.0) 57 (3.3) 30 48 (0.6) 50 (0.5) 57 (5.1) 80 48 (2.1) 51 (0.6) 63 (1.9) Table A-3. Measured mean aboveground biomass (one standard deviation of four replications) for four phosphorus treatments, sampled four times during the growing season in the Kpeve experiment P level (kg ha-1) 17 dap 31 dap 52 dap 108 dap 0 110 (22) 913 (145) 5681 (739) 9802 (1572) 10 102 (18) 897 (91) 5436 (805) 10002 (949) 30 105 (13) 914 (106) 5047 (883) 9816 (706) 80 101 (9) 1013 (81) 5083 (680) 8926 (914) dap = days after planting. Data are reported in kg ha-1.

PAGE 142

142 Table A-4. Mean green leaf area (one standard deviation of four replications) for four phosphorus treatments, measured seven times during the growing season in the Kpeve experiment P level (kg ha-1) 17 dap 24 dap 31 dap 38 dap 45 dap 52 dap 68 dap 0 327 (47) 1010 (44) 2409 (176) 4495 (309) 5269 (481) 5554 (187) 4258 (470) 10 332 (77) 1078 (318) 2634 (656) 4842 (881) 5699 (1034) 6235 (1067) 4543 (1309) 30 373 (47) 1175 (189) 2807 (235) 5102 (454) 5612 (411) 5946 (454) 4330 (464) 80 320 (25) 1004 (78) 2576 (235) 4605 (315) 5345 (202) 5473 (278) 4090 (217) dap = days after planting. Data are reported in cm2 plant-1. Table A-5. Mean maize height ( one standard deviation of four replications) for four phosphorus treatments, measured seven times during the growing season in the Kpeve experiment P level (kg ha-1) 17 dap 24 dap 31 dap 38 dap 45 dap 52 dap 68 dap 0 26 (5) 42 (10) 76 (10) 124 (22) 195 (32) 236 (36) 240 (42) 10 29 (5) 45 (8) 76 (14) 129 (22) 208 (22) 250 (27) 259 (33) 30 26 (5) 47 (6) 84 (8) 138 (15) 215 (19) 250 (29) 256 (31) 80 25 (7) 44 (9) 81 (12) 126 (21) 188 (37) 234 (35) 240 (32) dap = days after planting. Da ta are reported in cm plant-1. Table A-6. Mean soil moisture ( one standard deviation of four replications) in four phosphorus treatments plots, measured using TDR ei ght times during the growing season in the Kpeve experiment P level (kg ha-1) 0 dap 16 dap 24 dap 30 dap 37 dap 45 dap 53 dap 69 dap 0 12.1 (3.1) 8.8 (2.2) 17.1 (4.8) 13.7 (3.1) 11.5 (2.6) 9.4 (2.6) 6.4 (2.1) 13.7 (7.3) 10 11.8 (2.1) 8.3 (1.4) 15.3 (2.9) 12.1 (1.4) 10.2 (1.5) 8.0 (1.9) 6.0 (1.9) 9.4 (1.6) 30 12.7 (2.5) 8.4 (1.1) 17.3 (4.0) 13.3 (3.1) 10.8 (1.6) 8.3 (1.6) 5.6 (1.2) 10.6 (3.3) 80 13.2 (3.2) 8.6 (1.7) 17.9 (5.4) 13.7 (3.6) 11.1 (1.7) 9.0 (1.7) 5.7 (1.4) 12.0 (4.5) dap = days after planting. Data are reported in %.

PAGE 143

143 Table A-7. Measured mean grain yield (one standard deviation of four replications), unit grain weight (one standard deviati on of four replications), and grain number (one standard deviation of four replications) for four phosphorus levels in the Kpeve experiment P level (kg ha-1) Grain yield (kg ha-1) Unit grain weight (g grain-1) Grain number (# m-2) 0 3286 (683) 0.23 (0.04) 1655 (193) 10 2859 (384) 0.25 (0.03) 1344 (405) 30 3025 (358) 0.24 (0.02) 1492 (471) 80 2918 (411) 0.23 (0.04) 1467 (334)

PAGE 144

144 APPENDIX B MAPS OF THE EXPERIMENT SITES LOCATIONS Figure B-1. Map of the African continent s howing Ghana, the country where the field experiments were carried out Ghana

PAGE 145

145 Figure B-2. Map of Ghana showing the location of the two study sites, Kpeve in the South and Wa in the North

PAGE 146

146 APPENDIX C INITIALIZATION OF SOIL INORGANIC AND ORGANIC PHOSPHORUS POOLS IN THE SOIL-PLANT PHOSPHORUS MODEL Initial values of the three so il inorganic P pools (labile, ac tive and stable) and two soil organic P pools (active and stable ) described in Chapter 3 are n eeded to simulate phosphorus in soils and plants. These values would ideally co me from P fractionation studies on the soil of interest following the procedure developed by He dley et al. (1982), Ti essen et al. (1984), and Tiessen and Moir (1993). The Hedley/Tiessen fractionation procedur e is a thorough phosphorus extraction method that treats a so il sample with increas ingly aggressive chemicals. The soil is first shaken with water plus resin (to extract th e most labile part of P), then treated with NaHCO3, NaOH, (and sometimes NaOH with soni cation), diluted HCl and hot concentrated HCl. Each chemical extracts a more resistant form of phosphorus that escaped the previous extractant. The residual phosphorus still remaining in the soil is measured after digestion of the soil sample with perchloric or sulfuric acid. Th is procedure sequentially extracts both inorganic and organic forms of P. The data obtained from th e P fractionation are used to determine directly the sizes of the inorganic and organic pools. However, few researchers make use of the P fractionation method probably because it is expensive and also because simple and inexpens ive extraction methods, such as resin and Bray1 methods, would generally answer ph osphorus availability questions that arise in most agronomic experiments. If an alternate method is not provided for i ndirect estimation of the pools sizes, potential model users could possibly resort to indicative values found in the literature or eventually conclude that the model is not of any practical use because required input data are not readily available. Since organic carbon, pH and availa ble phosphorus are routinely measured in most

PAGE 147

147 traditional agronomic experiments, developing relati onships that can make use of those data and provide reasonable estimates of inorganic labile P and organic P was thought to be helpful. The aim of this appendix was to present direct and indirect me thods of estimation of initial inorganic labile, active and stable P, and initia l organic active and stable P for use by the soilplant phosphorus model in DSSAT. The relations hips discussed in this appendix are based on studies by Singh (1985), Sharpley (1984, 1989). Initialization of Inorganic Phosphorus Pools From P Fractionation Data The quantities of inorganic labile active, and stable P (Table C-1) initially present in the soil can then be derived, in mg kg-1, from the fractionation data for each soil layer as (Jones et al., 2005a): Initial PiLabile = 3 sin Re PiNaHCO Pi (C-1) Initial PiActive = PiNaOH 5 0 (C-2) Initial PiStable = sidual P PiHClHot PiHCl c PiNaOHSoni PiNaOHRe 5 0 5 0 (C-3) Where PiResin is inorganic P extracted with water and resin. PiNaHCO3 is inorganic P extracted w ith bicarbonate of sodium. PiNaOH is inorganic P extracte d with sodium hydroxide. PiNaOHSonic is inorganic P extracted with sodium hydroxide plus sonication. PiHCl is inorganic P extracted with diluted HCl. PiHClHot is inorganic P extracted with hot concentrated HCl. PResidual is residual P measured after digestion of the remaining sample with perchloric or sulfuric acid. From Measured Available P Using the Anion Exchange Resin Method P extraction using this method can be approximate d as a direct measurement of PiLabile in the soil. The anion exchange resin technique extracts phosphorus from the soil in the same manner as plants and has been reported as a reliable method for measuring plant available phosphorus (Myers, 2005; Abdu, 2006). PiActive and Pi Stable are assumed to be in equilibrium initially and are calculated based on the valu e of PiLabile as follo w (Jones et al., 1984a):

PAGE 148

148 PiActive = AL LAK K PiLabile (C-4) Where KLA = rate constant for transformation from labile P to active P KAL = rate constant for transformation from active P to labile P. 5 01 03 0 x PAvailInde x PAvailInde KLA (C-5) x PAvailInde K KLA AL 3 (C-6) Where PAvailIndex is used as a measure of the activity level of P in the soil. The calculation of the PAvailIndex depe nds on soil category (Sharpley et al., 1984, Table C-2) and is provided in Table 3-1. According to Jones et al. (1984a), PiStable is four times as large as PiActive: PiStable = PiActive 4 (C-7). From Other Methods If fractionation data are not av ailable and resin measurements were not made, the initial PiLabile can be estimated from other P ex traction methods based on regression equations between resin P and extractable P (such as Bray1 a nd Olsen P) that were used to build an expert system. The equations that appear in the expert syst em are based on studies conducted by Sharpley et al. (1984, 1989) and Singh (1985). The expert system in its cu rrent version has not been tested independently for its ability to estimate accurately labile P from different available P extraction methods. It is used in the soil modules of the P model as an experimental version to estimate the initial inorganic and organic phosphorus pool sizes based on th e method used for measuring available P and the soil category concerned. The criteria used for assigning soil categories are presented in Table C-2.

PAGE 149

149 The inorganic labile P is computed first and the active Pi is derived from the labile Pi in such a way that the two pools remain in equilibri um (Equation C-4). The stable Pi is calculated using the size of PiActive (Equation C-7). The following P extraction methods are used in the expert system proposed by Singh (1985) to compute initial PiLabile in the soil: water, Bray 1, Ol sen, Mehlich 1, Truog, Morgans solution and Colwell (Table C-3). In slightly we athered soils, measured exchangeable potassium is used in combination with Bray 1, Olsen, Mehlich 1 and Truog for a more accurate estimation of PiLabile (Table C-3). The soil-plant phosphorus model cannot be initia lized if none of these measured P data is available. Initialization of Soil Organic Phosphorus Pools The division of the organic residues added to the surface of the soil into metabolic and structural components (Figure 3-1) is governed by the lignin to N ratio (lignin:N) of the residues. The metabolic fraction is estimated as equal to 0.85 0.013*(lignin:N ratio) (Gijsman et al., 2002a). The procedures for estimating the initial sizes of the active and stable soil SOM (SOM1 and SOM23) from P fractionation data or from the expert system (using organic C and pH) are described next. Initialization from P Fractionation Data The initial values of the SOM1 and SOM23 pools can be obtained from Hedley/Tiessen soil P fractionation data (Table C-4) (Gijsman and Porter, unpublished): Initial PoActive (SOM1) = PoNaOH PoNaHCO 3 (C-8) Initial PoStable (SOM23) = ) Re 5 0 ( sidual P PoHClHot PoHCl c PoNaOHSoni (C-9) Where PoNaHCO3 is organic P extracted with bicarbonate of sodium. PoNaOH is organic P extracted with sodium hydroxide.

PAGE 150

150 PoNaOHsonic is organic P extracted with s odium hydroxide plus sonication. PoHCl is organic P extracted with diluted HCl. PoHClhot is organic P extracted with hot concentrated HCl. PResidual is P recovered after digestion w ith perchloric or sulfuric acid. Initialization from Measured Organic P If only a pooled total organic P value is known, the partitioning betw een active and stable P depends on the land and crop use history of th e soil (previous crop in DSSAT). The initial active organic P is set to 3% and the initial stable P to 97% of the total organic P if the previous crop is bahia grass or grass weeds. For all other previous crops, the initial active and the initial stable organic P represent respectively 6% and 94% of the measured total organic P. Initialization from Organic C and soil pH Indirect estimation of the active and stab le organic P (if total organic P was not measured) through the expert system uses measur ed soil organic C and pH (Table C-5). These soil properties are known to be correlated with soil organic P; equations relating them to total organic P in the soil have been developed base d on studies by Sharpley et al. (1984, 1989) and Singh (1985). The distribution betw een active and stable P is exactly the same as if the total organic P was directly measured.

PAGE 151

151 Table C-1. Relationship between inorganic P pools and P extracted using the Hedley procedure P fractionation methods Inorganic P pools Resin NaHCO3 NaOH NaOH and SonicationHCl Hot HCl Residual Labile + + Active 2 / 1 Stable 2 / 1 + + + 2 / 1 Source: Gijsman and Porter (2005) Table C-2. Specification of soil categories Soil Category Criteria Andisol Soil description or taxonomy includes the terms "ANDOSOL" or "ANDISOL" or VOLCAN" or "ANDEPT" Calcareous CaCO3 content > 15% Slightly Weathered Ratio 16 100 / CLAY CEC Highly Weathered Ratio 16 100 / CLAY CEC Other Soils --Soil description or taxonomy does not include the terms "ANDOSOL" or "ANDISOL" or "VOLCAN" or "ANDEPT"; --CaCO3 and CEC are not measured. Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in the tropics. PhD thesis, University of Hawaii, Honolulu. Table C-3. Equations for calculatin g initial inorganic P labile fr om different extraction methods for different soil categories Soil Category P or K data available PiLabile (mg kg-1) Calcareous Olsen 18 0 17 1 POlsen Bray 1 88 1 1 81 1 PBray Mehlich 1 (double acid 1:5) 20 10 1 10 0 PMehlich Water 09 0 92 5 PWater Slightly Olsen 53 6 76 0 POlsen Weathered Olsen and Exchangeable K 62 2 09 10 62 0 ExchK POlsen Bray 1 77 6 1 37 1 PBray Bray 1 and Exchangeable K 71 2 59 10 1 09 1 ExchK PBray Mehlich 1 (double acid 1:5) 82 5 1 71 2 PMehlich Mehlich 1 & Exchangeable K 42 2 58 9 11 16 2 ExchK PMehlich Truog 35 3 34 0 PTruog Truog and Exchangeable K 48 1 85 5 30 0 ExchK PTruog Morgan's Solution 87 11 30 187 PMorgan

PAGE 152

152 Table C-3 continued Highly Olsen 19 2 50 2 POlsen Weathered Bray 1 30 0 1 88 2 PBray Mehlich 1 (double acid 1:5) 21 0 1 97 5 PMehlich Truog 49 1 07 1 PTruog Mehlich 1 (double acid 1:10) 72 5 1 64 0 PMehlich Colwell 21 4 43 0 PColwell Andisol Olsen 56 2 41 1 POlsen Bray 1 11 2 1 88 2 PBray Mehlich 1 (double acid 1:5) 67 6 1 52 4 PMehlich Truog 73 0 27 0 PTruog Unknown Olsen 39 11 74 0 POlsen Or Bray 1 24 10 1 35 1 PBray Not specified Mehlich 1 (double acid 1:5) 39 9 1 65 2 PMehlich Truog 15 6 28 0 PTruog Singh, U. 1985. A crop growth model for predicting corn (Zea mays L.) performance in the tropics. PhD thesis, University of Hawaii, Honolulu. Table C-4. Relationship between organic P pools and P extracted using the Hedley procedure P fractionation method Inorganic P pools NaHCO3 NaOHNaOH and Sonication HClHot HCl Residual Active + + Stable + + + 2 / 1 Source: Gijsman and Porter (2005) Table C-5. Equations for calcula ting initial total organic P from soil organic carbon (OrgC) and pH for different soil categories Soil Category Organic P (mg kg-1) Calcareous OrgC pHe e 55 0 6 3 8 11 2002 Highly Weathered OrgC pHe e 35 0 6 3 85 11 2002 Slightly Weathered OrgC pHe e 10 0 12 10 5 11 9002 Other Soils OrgC pHe e 135 0 8 7 5 11 5202 Singh, U. 1985. A crop growth model for predicting corn ( Zea mays L.) performance in the tropics. PhD thesis, University of Hawaii, Honolulu.

PAGE 153

153 LIST OF REFERENCES Abdu, N., 2006. Soil-phosphorus extraction me thodologies: a review. African Journal of Agricultural Research 1 (5), 159-161. Adiku, S.G.K., 2004. Exploring options for im proving crop productivity within some farming zones of Ghana using coupled climatic-crop models, Final Report fo r the period November 2002 to August 2004. Department of Soil Science, University of Ghana, Legon, Ghana. Adiku, S.G.K., 2006. Measuring and Assessi ng Soil Carbon Sequestration by Agricultural Systems in Developing Countries, Final Re port for the period March to July 2006. Department of Soil Science, University of Ghana, Legon, Ghana. Annino, J.S., Russell, E.C., 1981. The most freque nt causes of simulation analysis failure and how to avoid them. Interfaces 11 (3), 59-63. Anonymous, 1996. Obatanpa, a new maize variety w ith high quality protein for farmers. Crops Research Institute Factsheets, Animal Science Department, Ministry of Food and Agriculture, Accra, Ghana. Bolanos, J., Edmeades, G.O., 1993. Eight cycles of selection for drought tolerance in lowland tropical maize. II. Response in reproductiv e behaviour. Field Crops Research 31, 253-268. Brady, N.C., Weil, R.R., 2002. The Nature and Properties of Soils. Pearson Education, NJ. Colomb, B., Kiniry, J.R. Debaeke, P., 2000. Effe ct of soil phosphorus on leaf development and senescence dynamics of fi eld-grown maize. Agrono my Journal 92, 428-435. Daroub, S.H., Gerakis, A., Ritchie, J.T., Frie sen, D.K., Ryan, J., 2003. Development of a soilplant phosphorus simulation model for cal careous and weathered tropical soils. Agricultural Systems 76, 1157-1181. Del Campillo, M.C., Van Der Zee, S.E.A.T. M., Torrent, J., 1999. Modelling long-term phosphorus leaching and changes in phosphorus fertility in ex cessively fertilized acid sandy soils. European Journal of Soil Science 50, 391-399. Du Toit, A.S, Booysen, J., Human, J.J., 1997. Us e of linear regression an d correlation matrix in the evaluation of CERES3 (Mai ze). South African Journal of Plant and Soil 14, 177-182. Dzotsi, K.A., Agboh-Noameshie, A., Struif Bont kes, T.E., Singh, U., Dejean, P., 2003. Using DSSAT to derive optimum combinations of cultivar and sowing date for maize in Southern Togo. In: Struif-Bontkes, T.E., Wopereis, M.C.S. (Eds), Decision-Support Tools for Smallholder Agriculture in Sub-Saharan Afri ca: A Practical Guide. IFDC and CTA, pp. 100-113. Eghball, B., Shanahan, J.F., Varvel, G.E., G illey, J.E., 2003. Reduction of high soil test phosphorus by corn and soybean varie ties. Agronomy Journal 95, 1233-1239.

PAGE 154

154 FAO, 2003. Fertilizers and their use. Food and Agriculture Orga nization of the United Nations, Rome, Italy. FAO, 2005. Fertilizer Use by Cr op in Ghana. Land and Plant Nutrition Management Service, Land and Water Development Division, Food a nd Agriculture Organization of the United Nations, Rome, Italy. Fofana, B., Tamelokpo, A., Wopereis, M.C.S., Breman, H., Dzotsi, K.A., Carsky, R.J., 2005. Nitrogen use efficiency by maize as affected by a mucuna short fallow and P application in the coastal savanna of West Af rica. Nutrient Cycling in Ag roecosystems, 71 (3), 227-237. Gauch, G.H., Jr., Hwang, J.T.G., Fick, G.W., 2003. Model evaluation by comparison of modelbased predictions and measured va lues. Agronomy Journal 95, 1442-1446. Gijsman, A.J., Hoogenboom, G., Parton, W.J ., Kerridge, P.C., 2002. Modifying DSSAT crop models for low-input agricultural systems us ing a soil organic matter-residue module from CENTURY. Agronomy Journal 94, 462-474. Gijsman, A.J, Jagtap, S.S., Jones, J.W., 2002b. Wading through a swamp of complete confusion: how to choose a method for estimating soil wa ter retention parameters for crop models. European Journal of Agronomy 18(1-2), 75-105. Gijsman, A.J., Porter, C.H., 2005. The Soil Or ganic Phosphorus Module of the DSSAT Cropping Systems Model. University of Florida, Gainesville. Havlin, J.L., Beaton, J.D., Tisdale, S.L., Nelson, W.L., 1999. Soil fertility and fertilizers: an introduction to nutrient management. Pr entice Hall, Upper Saddle River, NJ. Hedley M.J., Stewart, J.W.B., Chauhan, B.S ., 1982. Changes in inorganic and organic soil phosphorus fractions induced by cultivation practices and by laboratory incubations. Soil Sci. Soc. Am. J. 46, 970-976. Hunt, L.A., Boote, K.J., 1998. Data for model opera tion, calibration, and evaluation. In: Tsuji, G.Y. et al. (Eds), Understanding Options fo r Agricultural Production. Kluwer Academic Publisher, Boston, pp. 9-39. Jagtap, S.S., Abamu, F.J., 2003. Matching im proved maize production technologies to the resource base of farmers in a moist savanna. Agricultural Systems 76, 1067-1084. Jagtap, S.S., Abamu, F.J., Kling, J.G., 1999. Long-term assessment of nitrogen and variety technologies on attainable maize yields in Nigeria using CERES-Ma ize. Agricultural Systems 60, 77-86. Janssen, B.H., 1993. Integrated nutrient managemen t: the use of organic and mineral fertilizers. In: Van Reuler, H. Prins, W.H. (Eds.). The Role of Plant Nutrients for Sustainable Food Crop Production in Sub-Saharan Africa. Dutch A ssociation of Fertilizer Producers (VKP), Leidschendam, The Netherlands, pp. 89-105.

PAGE 155

155 Janssen, B.H., Lathwell, D.J., Wolf, J., 1987. Modeling long-term crop response to fertilizer phosphorus. II. Comparison with field re sults. Agronomy Journal 79, 452-458. Johnston, A.E., 2000. Soil and Plant Phosphate. In ternational Fertilizer Industry Association. Paris, France. Jones, C.A. 1983. A survey of the variability in tissue nitrogen and phosphor us concentrations in maize and grain sorghum. Fiel d Crops Research 6, 133-147. Jones, C.A., Cole, C.V., Sharpley, A.N. W illiams, J.R., 1984a. A simplified soil and plant phosphorus model: I. Documentation. Soil Scie nce Society of America Journal 48, 800805. Jones, J.W., Hoogenboom, G., Porter, C.H., Boote, K.J., Batchelor, W.D., Hunt, L.A., Wilkens, P.W., Singh, U., Gijsman, A.J., Ritchie, J.T., 2003. The DSSAT cropping system model. European Journal of Agronomy 18, 235-265. Jones, J.W., Luyten, J.C., 1998. Simulation of biol ogical processes. In: Peart, R.M., Curry, R.B. (Eds), Agricultural Systems Modeling and Si mulation. Marcel Dekker, New York, pp. 1962. Jones, J.W., Singh, U., Porter, C.H., 2005a. Soil Inorganic Phosphorus Module. Research Report. University of Florida, Gainesville. Jones, J.W., Singh, U., Porter, C.H., Bostick, W.M., 2005b. Generic Plant Phosphorus Module. Research Report. University of Florida, Gainesville. Jones, J.W., Tsuji, G.Y., Hoogenboom, G., Hunt, L. A., Thornton, P.K., Wilkens, P.W., Imamura, D.T., Bowen, W.T., Singh, U., 1998. Decisi on support system for agrotechnology transfer: DSSAT v3. In: Tsuji, G.Y. et al. (Eds ), Understanding Options for Agricultural Production. Kluwer Academic P ublisher, Boston, pp. 157-177. Kaboneka, S., Nivyiza, J.C., Sibomana, L., 2004. Effects of nitrogen and phosphorus fertilizer addition on wheat straw carbon decomposition in a Burundi acidic soil. In: Bationo, A. (Ed.), Managing Nutrient Cycles to Sustain Soil Fertility in Sub-Saharan Africa. Academy Science Publisher, Nairobi, Kenya, pp. 151-161. Karpinets, T.V., Greenwood, D.J., Ammons, J. T., 2004. Predictive mechanistic model of soil phosphorus dynamics with readily available i nputs. Soil Science Soc. Am. J. 68, 644-653. Kelling, K.A., Bundy, L.G., Combs, S.M., Peters, J.B., 1998. Soil test recommendations for field vegetable, and fruit crops. Univ. Wis. Coop. Extn. Ser. Bull. A2809. Khan, M.A., Hussain, N., Abid, M., Masood, M.U., 2005. Effect of phosphorous levels on growth and yield of maize ( Zea mays L.) cultivars under saline conditions. International Journal of Agriculture and Biology 7 (3), 511-514.

PAGE 156

156 Kobayashi, K. Us Salam, M., 2000. Comparing simulated and measured values using mean squared deviation and its compone nts. Agronomy Journal 92, 345-352. Koopmans, G.F., Chardon, W.J., Eh lert, P.A.I., Dolfing, J., Suurs, R.A.A., Oenema, O., van Riemsdijk, W.H., 2004. Phosphorus availability for plant uptake in a phosphorus-enriched noncalcareous sandy soil. Journal of Environment Quality 33, 965-975. Lehman, J.T., Botkin, D.B., Likens, G.E., 1975a. The assumptions and rationales of a computer model of phytoplankton population dynamics Limnology and Oceanography 20, 343-364. Lewis, D.R., McGechan, M.B., 2002. A review of field scale phosphorus dynamics models. Biosystems Engineering 82 (4), 359-380. Littell, R.C., Henry, P.R., Ammerman, C.B., 1998. Statistical analysis of repeated measures data using SAS procedures. Journal of Animal Science 76, 1216-1231. Makowski, D., Naud C., Monod, H., Jeuffroy, MH., Barbottin, A., 2005. Global sensitivity analysis for calculating the contribution of genetic parameters to the variance of crop model prediction. In: Hanson, K.M., Hemez, F.M. (Eds), Sensitivity Analysis of Model Output. Proceedings of the 4th International Conference on Sensitivity Analysis of Model Output held in Santa Fe, NM, 8-11 March 2004. pp. 307-316. Matthews, R., Stephens, W., Hess, T., Masson, T., Graves, 2000. Application of Crop/Soil Simulation Models in Developing Countries Final Report. Institute of Water and Environment, Cranfield Universi ty, Silsoe, Bedfordshire, U.K. Miao, Y., Mulla, D.J., Batchelor, W.D., Paz, J. O., Robert, P.C., Wiebers, M., 2006. Evaluating management zone optimal nitrogen rates with a crop growth model. Agronomy Journal 98 (3), 545-553. Mitchell, P.L. 1997. Misuse of regression for empirical validation of models. Agricultural Systems 54, 313-326. Monod, H., Naud, C., Makowski, D., 2006. Uncertain ty and sensitivity analysis for crop models. In: Wallach, D., Makowski, D., Jones, J.W. (Eds), Working with Dynamic Crop Models: Evaluation, Analysis, Parameterization and Applications. Elsevier, Amsterdam, The Netherlands. pp. 55-96. Myers, R.G., Sharpley, A.N., Thien, S.J., Pier zynski, G.M., 2005. Ion-sink phosphorus extraction methods applied on 24 soils from the continenta l USA. Soil Sci. Soc. Am. J. 69, 511-521. Naab, J.B., 2005. Measuring and Assessing Soil Carbon Sequestration by Ag ricultural Systems in Developing Countries, 2004 Annual Report. Sa vanna Agricultural Research Institute, Wa, Ghana.

PAGE 157

157 Newman, S., Lynch, T., Plummer, A.A., 2000. Aspects of success and failure of decision support systems in agriculture. In: Abbass, H.A., To wsey, M. (Eds), Application of Artificial Intelligence, Optimization and Bayesian Methods in Agriculture. Queensland University of Technology Press, Brisbane, Australia, pp. 17-30. Nziguheba, G., Merckx, R., Palm, C.A., 2002. Soil phosphorus dynamics and maize response to different rates of phosphorus fertilizer applie d to an Acrisol in western Kenya. Plant and Soil 243, 1-10. Oberkampf, W.L., Helton, J.C., Joslyn, C.A., Wojtkiewicz, S.F., Ferson, S., 2004. Challenge problems: uncertainty in system response given uncertain parameters. Reliability Engineering and System Safety 85, 11-19. Odum H.T., Odum, E.C., 2000. Modeling for All S cales: An Introduction to System Simulation. Academic Press, San Diego. Pellerin, S., Mollier, A., Plenet, D., 2000. Phosphorus deficiency affects the rate of emergence and number of maize adventitious nodal roots. Agronomy Journal 92, 690-697. Osiname, O.A., Meppe, F., Evere tt, L., 2000. Response of maize ( Zea mays ) to phosphorus application on basaltic soils in Northw estern Cameroon. Nutrient Cycling in Agroecosystems 56, 209-217. Parton, W.J., Ojima, D.S., Cole, C.V., Schimel, D.S., 1994. A general model for soil organic matter dynamics: Sensitivity to litter chemistr y, texture and manageme nt. In: Bryant, R.B., Arnold, R.W. (Eds), Quantitative modeling of soil forming processes. SSSA Spec. Publ. 39. SSSA, Madison, WI, pp 147-167. Parton, W.J., Stewart, J.W.B., Co le, C.V., 1988. Dynamics of C, N, P and S in grassland soils: A model. Biogeochemistry 5:109. Pellerin, S., Mollier, A., Plenet, D., 2000. Phosphorus deficiency affects the rate of emergence and number of maize adventitious nodal roots. Agronomy Journal 92, 690-697. Pheav, S., Bell, R.W., White, P.F., Kirk, G.J.D ., 2003. Fate of applied phosphorus in a highly weathered sandy soil under lowland rice croppi ng, and its residual effect. Field Crops Research 81, 1-16. Plenet, D., Etchebest, S., Mollier A., Pellerin, S., 2000a. Growth analysis of maize field crops under phosphorus deficiency. I. Leaf growth. Plant Soil 223, 117-130. Plenet, D., Mollier, A ., Pellerin, S., 2000b. Growth analys is of maize field crops under phosphorus deficiency. II. Ra diation-use efficiency, biomass accumulation and yield components. Plant Soil 224, 259-272. Probert, M.E., 2004. A capability in APSIM to m odel phosphorus responses in crops. In: Delve, R.J. Probert, M.E. (Eds), Modelling Nutrie nt Management in Tropical Cropping Systems. ACIAR Proceedings No. 114. pp. 92-100.

PAGE 158

158 Probert, M.E., Okalebo, J.R., 1992. Effect of phosphorus on the growth and development of maize. In: Probert, M.E. (Ed), Semi-Arid Ea stern Kenya. ACIAR, Australia, pp. 55-62. Rahn, C., Mead, A., Draycott, A., Lillywhite, R ., Salo, T., 2001. A sensitivity analysis of the prediction of the nitrogen fertilizer require ment of cauliflower crops using the HRI WELLN computer model. Journal of Agricultural Science 137, 55-69. Ratto, M., Tarantola, S., Saltelli, A., 2001. Sensit ivity analysis in model calibration: GSA-GLUE approach. Computer Physics Communications 136, 212-224. Reddy, K.R., Wang, Y., DeBusk, W.F., Fisher M.M., Newman, S., 1998. Forms of soil phosphorus in selected hydrologic units of the Florida Everglad es. Soil Sci. Soc. Am. J. 62, 1134-1147. Saltelli, A., 2002. Sensitivity analysis for importa nce assessment. Risk Analysis 22 (3), 579-588. Saltelli, A., 2004. Global sensitivity analysis: an introduction. In: Hanson, K.M., Hemez, F.M. (Eds), Sensitivity Analysis of M odel Output. Proceedings of the 4th International Conference on Sensitivity Analysis of Model Output held in Santa Fe, NM, 8-11 March 2004. pp 27-43. Saltelli, A., Tarantola, S., Campolongo, F., 2000. Sensitivity analysis as an ingredient of modeling. Statistical Sc ience 15 (4), 377-395. Sartain, J.B. 1980. Mobility and extractability of phosphorus applied to the surface of tifway bermudagrass turf. Soil and Crop Science Society of Florida Proc. 39, 47-50. SAS, 2002. SAS version 9.1 for Window s. SAS Institute Inc., Cary, NC. Schachtman, D.P., Reid, R.J., Ayling, S.M., 1998. Phosphorus uptake by plants : from soil to cell. Plant Physiology 116, 447-453. Schmidt, J.P., Buol, S.W., Kamprath, E.J., 1996. Soil phosphorus dynamics during 17 years of continuous cultivation: a method to estimate longterm P availability. Geoderma 78, 59-70. Shapiro, C.A., Fergusson, R.B., Hergert, G.W., Dobermann, A.R., Wortmann, C.S., 2003. Fertilizer Suggestions for Corn. NebGuide G 74-174-A. Cooperate Extension, Institute of Agriculture and Natural Resources, Univer sity of Nebraska-Lincoln, Lincoln. Sharpley, A.N., Jones, C.A., Gray, C. Cole, C.V., 1984. A simplified so il and plant phosphorus model II. Prediction of labile, organic, and sorbed phosphorus. Soil Sci. Soc. Am. J. 48, 805-809. Sharpley, A.N., Singh, U., Uehara, G., Kimble J., 1989. Modeling soil and plant phosphorus dynamics in calcareous and highly weathered soils. Soil Sci. Soc. Am. J. 53, 153-158. Singh, U. 1985. A crop growth model for predicting corn ( Zea mays L.) performance in the tropics. PhD thesis, University of Hawaii, Honolulu.

PAGE 159

159 Singh, U., Thornton, P.K., Saka, A.R., Dent, J. B., 1993. Maize modeling in Malawi: a tool for soil fertility research and development. In: De Vries, F.P. et al. (Eds). Systems Approach for Agriculture Development. Proceedings of the International Symposium on Systems Approach for Agriculture Development held in Bangkok, Thailand, 2-6 December 1991. pp. 253-273. Singh, U., Wilkens, P.W., Chude, V., Oikeh, S., 1999. Predicting the effect of nitrogen deficiency on crop growth durati on and yield. In : Robert, P.C. et al. (Eds.), Proceedings of the Fourth International Confer ence on Precision Agriculture he ld in St. Paul, Minnesota, 19-22 July 1998. American Society of Agronomy, pp.1379-1393. Struif Bontkes, T.E., Wopereis, M.C.S., (Eds) 2003. Decision Support Tools for Small-Holder Agriculture in SubSaharan Africa, A Pratical Guide. IFDC and CTA. Sturgul, S.J., Bundy, L.G., 2004. Understanding soil phosphorus: an overview of phosphorus, water quality and agricultural management pr actices. Nutrient and Pest Management Program, University of Wisconsin-Madison, Madison. Tiessen, H., Moir, J.O., 1993. Ch aracterisation of available P by se quential extraction. In: Carter, M.R. (Ed.), Soil sampling and methods of analysis. CRC Press, Boca Raton. pp. 75-86. Tiessen H., Stewart, J.W.B., Co le, C.V., 1984. Pathways of phosphorus transformations in soils of differing pedogenesis. Soil Sci. Soc. Am. J. 48, 853-858. Tsuji, G.Y., Hoogenboom, G., Thornton, P. K., (Eds) 1998. Understanding Options for Agricultural Production. Kluwer Academic Publisher, Boston. Uehara, G. 1998. Synthesis In: Tsuji, G.Y. et al. (Eds), U nderstanding Options for Agricultural Production. Kluwer Academic P ublisher, Boston, pp. 389-392. Vincent, K.R., Chadwick, O.A., 1994. Synthesi zing bulk density for soils with abundant rock fragments. Soil Sc. Soc. Am. J. 58, 455-464. Wafula, B.M. 1995. Applications of crop simulati on in agricultural extens ion and research in Kenya. Agricultural Systems 49, 399-412. Walker, D.H., 2000. Decision support, learning and rural resource management. Agricultural Systems 73, 113-127. Wallach, D., Makowski, D., Jones, J.W., (E ds) 2006. Working with Dynamic Crop Models: Evaluation, Analysis, Parameterization, a nd Applications. Elsevier, Amsterdam, The Netherlands. Wilson, M.V., Botkin, D.B., 1990. Models of si mple microcosms: emergent properties and the effect of complexity and stability. American Naturalist 135 (3), 414-434. Wolf, J., De Wit, C.T., Janssen, B.H., Lathwe ll, D.J., 1987. Modeling long-term crop response to fertilizer phosphorus. I. The model. Agronomy Journal 79, 445 451.

PAGE 160

160 Ziadi, N., Belanger, G., Cambouris, A.N., Trem blay, N., Nolin, M.C., Claessens, A.. 2007. Relationship between P and N concentrati ons in corn. Agronomy Journal 99, 833-841.

PAGE 161

161 BIOGRAPHICAL SKETCH Kofikuma Adzewoda Dzotsi was born in Lo me, Togo (West Africa). He obtained his GED in 1996 and in fall 1996 entered the School of Agronomy at the Univer sity of Lome (UL), Togo. During his fifth and last year at the School of Agronomy (2001), Kofi kuma participated in a training workshop on systems analysis and mode ling organized by the African division of the International Center for Soil Fertility and Agri cultural Development (IFDC). This program was his first exposure to systems analysis and simu lation modeling applied to soils and crops and he decided to do his thesis research in this field. In March 2001, Ko fikuma joined IFDC to carry out his thesis research on Long-term assessment of variety and sowing time effects on grain yield of maize in southern Togo that was defended in November 2002 and Kofikuma graduated as an agronomy engineer from the De partment of Plant Productions School of Agronomy, UL. Between September and December 2002, Kofikuma worked as a research assistant at IFDCs Systems Approach Unit in Lome. In January 200 3, he took up the position of agronomist in IFDCs Natural Resource Management Program in Lome and worked on developing integrated soil fertility management (ISFM) options for ba sil. During his tenure in the program, he was responsible for evaluating a nd fine-tuning some ISFM-orien ted decision support tools like DSSAT, QUEFTS, SIMFIS. He al so supervised two agronomy e ngineer theses in 2004 and 2005. After spending almost 3 years in the program he decided to pursue a graduate program at the University of Florida (UF). In fall 2005, Kofikuma joined the McNair Bostick Simulation Laboratory in the Agricultural and Biological En gineering department at UF as a masters student. Kofikuma married Pascaline Akitani-Bob in April 2005. They have one son, Eyram R. Dzotsi.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101204_AAAABB INGEST_TIME 2010-12-04T12:38:42Z PACKAGE UFE0021774_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 21167 DFID F20101204_AAAWWB ORIGIN DEPOSITOR PATH dzotsi_k_Page_087.QC.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
4b67bff661eb06684c24e20dd1ed8bf7
SHA-1
46f82b6edae46ae11bdb99839549a8102859dde2
6035 F20101204_AAAWVM dzotsi_k_Page_073thm.jpg
930487780d0f359fb01eeafed1c0c058
299e982dad24169a23fb8ad05380de28476b33d4
8228 F20101204_AAAWUY dzotsi_k_Page_050.QC.jpg
9416d7bffd12f1a1fbe22227318d1903
5dc2ceed17ae1b7b71cc71a2e74f6bbfcb0622fe
112258 F20101204_AAAVSK dzotsi_k_Page_110.jp2
27f9e9a2cce29bd0f38ae0d3905c749b
e0408e4a0362e65b6f4f6164e80489662e4388e4
23841 F20101204_AAAVRW dzotsi_k_Page_030.QC.jpg
549901a30e598448f133af4d3d196564
7c198259c069fe3aaf0f765735189766b67690bd
5904 F20101204_AAAWWC dzotsi_k_Page_087thm.jpg
2027725dde0afaabcb37cab5ff81761c
a0f311f94c01c31123efacc34d285287c158f706
6172 F20101204_AAAWVN dzotsi_k_Page_074thm.jpg
3e11f81d18e560243d5f03ef5b8d9abb
0376219f79eff0dea4cdf9bd7246c3afb68fa92c
2563 F20101204_AAAWUZ dzotsi_k_Page_050thm.jpg
b7250dd22e5a4e432ccfb941f6d78bc9
e3153ed7eb3c2318549e65fc708231795c94bd76
2058 F20101204_AAAVTA dzotsi_k_Page_070.txt
7772a338df8ea038bc5c8cdd16fbfd02
db42a5693bda3c7380d07c2b73fc91cc6e437024
57624 F20101204_AAAVSL dzotsi_k_Page_090.pro
53ab85bf164ea66e4cb4dc8092851b80
7bebfc30b8d3a5b4a217031511f8b6342ddd9ff6
72357 F20101204_AAAVRX dzotsi_k_Page_151.jpg
a2ea2785b94b7cb67c9863b981f1d6b1
68435d2ebefa88dfca102162666ad2caa8a1dab6
23768 F20101204_AAAWWD dzotsi_k_Page_088.QC.jpg
1edc35e72b0f45703807731834f23094
7c78f8f7b0b493bb6074894e4b503c8ea6ca7ae6
19731 F20101204_AAAWVO dzotsi_k_Page_075.QC.jpg
a6b5500ff31bb00f591491298de43ac5
2461952c40777def2a4fd010082053faa27af726
54868 F20101204_AAAVTB dzotsi_k_Page_059.jpg
f46cb26e868fe8b723a03d951012e0f8
6f8f6c4c1ba8c55d0fc214bc06d4ec25613f0312
16618 F20101204_AAAVSM dzotsi_k_Page_150.QC.jpg
a76e2a27e80070fce94c47a68b8ba1fb
76a2354a7618551c00c1c82e471713fabd5b5788
25271604 F20101204_AAAVRY dzotsi_k_Page_012.tif
5aabb374dab96e5ba4c5c0d4b1b7ec4d
a027b6fdd7b301802522cd182b0d845d22a94b6f
23998 F20101204_AAAWWE dzotsi_k_Page_091.QC.jpg
216d060d36f5ccd1410f1906139a9c5e
62567d8d97913ac48f954e58200e0811fb19cb07
20653 F20101204_AAAWVP dzotsi_k_Page_077.QC.jpg
c53a9c0bd27b307e8b7b22d2354d518e
5e70a84f1a3b033c1c4a85e2314b94a62ad0e9a5
17431 F20101204_AAAVTC dzotsi_k_Page_104.pro
9911f06f7280321d925d1d67e9fe6842
73ce26505ede7dee3824fd1bbb05e61ec9fa671e
5098 F20101204_AAAVSN dzotsi_k_Page_017thm.jpg
17d599eb24e3cb7c552925cac7d7cc2c
238f46749a99e819d236af95402c96fc45ea241c
28184 F20101204_AAAVRZ dzotsi_k_Page_154.QC.jpg
c1558f5e2b926b83077b825489d83c0e
d316782712263f247aa195b969cb9d1deee07be1
25582 F20101204_AAAWWF dzotsi_k_Page_093.QC.jpg
035ed7860bbe7db3539a372ab17e225b
51641b5f042bff75da238815b7b373d420d5572b
22056 F20101204_AAAWVQ dzotsi_k_Page_079.QC.jpg
5b97efb5e77cac38a125e19cb83c6512
79cef05115431ef121fc038f32af8a82c4823284
5742 F20101204_AAAVTD dzotsi_k_Page_015.pro
8a4cda124500af6faaad1b57df8cab6b
775784a982211da1f7142da6c062a88dbdeff41a
1053954 F20101204_AAAVSO dzotsi_k_Page_119.tif
9ec2a1ee0e420af5c1075129166d14c2
6cd96a6e644c948f1b1a7811e0f763a77b08e083
22251 F20101204_AAAWWG dzotsi_k_Page_098.QC.jpg
6006e29d422d938a1f7feeef05bf07ef
881be4b64d50bd92672e848ef4942127b312e90c
6474 F20101204_AAAWVR dzotsi_k_Page_079thm.jpg
2e189ebcdb08dc343991f855b00f65d3
6781a74af85d41833dc76f88a079b5da11add40b
1206 F20101204_AAAVTE dzotsi_k_Page_107.txt
3bb617208e3309a6265050cfbdc4c239
6a6977902b687fd0526af1e8e809c1cc61a709f1
6984 F20101204_AAAVSP dzotsi_k_Page_031thm.jpg
a832d3f19b230f122096fd2995d2b434
35bc2ed7402c5053f9bc499dc1a125f26ec40ad0
18955 F20101204_AAAWWH dzotsi_k_Page_100.QC.jpg
67a9a96026911f174fb0762a015a2bcf
88ca6220047b064c44b79c5f5b4fbb2d43b11d34
6654 F20101204_AAAWVS dzotsi_k_Page_080thm.jpg
5a23e171b3616cb21033776287cc8321
8c9a8193f7c7952a1381068d3f85b0ba4e173bae
1805 F20101204_AAAVTF dzotsi_k_Page_100.txt
59a41d2a2d1184f45b411cdf1e76acd2
af9c3a1663d6d68041ebfd378670e9e3fbe8d8b7
1706 F20101204_AAAVSQ dzotsi_k_Page_036.txt
76c9a16a7ad73b65d0ead06188bf4711
94a97e9e773315cd7c02088f8a168de991b35621
11774 F20101204_AAAWWI dzotsi_k_Page_103.QC.jpg
4ac70616d4480e6e74691a0d33c52275
af4e89d61ecfa424b447a4a9a6855c121f3c3e1c
22933 F20101204_AAAWVT dzotsi_k_Page_081.QC.jpg
f63915320262ed5562e27e04c4b9edf3
8db2fedd868e7d645cf37b6b5a6cb57535cc3c81
6996 F20101204_AAAVTG dzotsi_k_Page_032thm.jpg
9c47a9a40cf4ff50fe1745ce12fa5f42
38a39abdb7cebfee8c4a9356fbe5c510902de6b5
F20101204_AAAVSR dzotsi_k_Page_101.tif
b8d0a20211a22e41732c76a6d28d1aad
27524e0e43e0e4b06b8f6f751b974d85014aa5b5
7353 F20101204_AAAWVU dzotsi_k_Page_082thm.jpg
a9604353fd841e1a4436c08c51de9708
39e72a847d9254dd21cb87ad45d0470faf55f103
2241 F20101204_AAAVSS dzotsi_k_Page_147.txt
7a3b331c159ed06d05f0f02310d73823
d0ab240687003887a7bc192df033cc9f3630449e
16255 F20101204_AAAWWJ dzotsi_k_Page_106.QC.jpg
b1957f93a8aa7726e5a5557db89df92e
b963960ea7b7c46cc27e2ba7f746df0d998d942e
25466 F20101204_AAAWVV dzotsi_k_Page_083.QC.jpg
80ee86df6779ce51fac0196f1eea1e92
5ed5d27bcea646b3d6f1afffd2188c7bd88bc581
20968 F20101204_AAAVTH dzotsi_k_Page_056.QC.jpg
ec792249939d0a41b5302eba860762c3
2175e61b447fde50846910c04f69ba01c7caf699
F20101204_AAAVST dzotsi_k_Page_016.tif
826d4eb8e023316f17c6d7ab8bd69216
464103f699b4d464ec34a757d21f406806338fd6
4896 F20101204_AAAWWK dzotsi_k_Page_106thm.jpg
2c707b8427f4a2d8b5d4bfcfe068a149
8098d974f3b308e06f8c5514b367321eb74ec5da
26039 F20101204_AAAWVW dzotsi_k_Page_084.QC.jpg
c12224fd808955fa2f2a892d86bca6ba
af417078ab58fec634e9523bcf564881885e279a
70634 F20101204_AAAVTI dzotsi_k_Page_016.jpg
6c984eb25a7c93a63ebf7f8ca094f730
26fe102232af4440580b4b9e2ca289ae55859226
6820 F20101204_AAAVSU dzotsi_k_Page_094thm.jpg
2eabf08d0b67f52a501b6b0b173f2c35
a76d83a4fe40ec553bc1be8f16a2ae8073750ef0
13754 F20101204_AAAWXA dzotsi_k_Page_136.QC.jpg
ba10f66f85f0215b49f597e4427319a3
ba6c68d476d204cc521c6bc6c435bd83f3c2401f
10823 F20101204_AAAWWL dzotsi_k_Page_108.QC.jpg
d2d4d42f11d3815498db94c42dc0895e
c5ee7e4e148919d040928e90244c2a7b5cd842f6
24720 F20101204_AAAWVX dzotsi_k_Page_085.QC.jpg
28f0b0873fe6791fe3257e59af58d4b9
65d4f556db8dbdcf3a6a4ee452a17e4c0c672799
479770 F20101204_AAAVTJ dzotsi_k_Page_062.jp2
f37190bd18295fe60ce93e4bf54202c4
0c2672a7e973b26fa23ccf0a69c5e2f3b1f75374
4774 F20101204_AAAVSV dzotsi_k_Page_160.pro
ab7e76bc410aa9aebc34cb73201b6e99
f54461b956a3161a62028af9db11869ef04a0cb3
5827 F20101204_AAAWXB dzotsi_k_Page_137thm.jpg
9b1cbc2c472db2e3b8ff5b83e5a21623
be4ff623503d13e95de86abdc3d0010ef7972c23
12397 F20101204_AAAWWM dzotsi_k_Page_109.QC.jpg
01cb7463f9be7fc3c4b29e8ca986e868
cb473cb6686209a62ad77ec45b9545f950423f66
6840 F20101204_AAAWVY dzotsi_k_Page_085thm.jpg
740f26cf64dd43fa4b6061677555f250
ddcb820254e2fde6d85d7a64b3e0357e3407f7d1
81169 F20101204_AAAVTK dzotsi_k_Page_017.jp2
a4724824feffa8ed3c6d0422e44e3a77
8e57399f4e1dadd0259da8c064be46e1a5e5b4e6
F20101204_AAAVSW dzotsi_k_Page_020.tif
c9b203da798dd1f1fda6a33dac088204
059c538a9330ac8a7269bb50ef9a5586021456a4
25629 F20101204_AAAWXC dzotsi_k_Page_138.QC.jpg
ac4f596d3afd3c7a6501af2fe72bc553
6aed813264507e5c087cc536d368803e95d1a1bf
6461 F20101204_AAAWWN dzotsi_k_Page_111thm.jpg
b7ede7b7d3c2e1c716a6ffe3b4cdc90d
63746021a63e41b408af53b8261285045107a0bc
26425 F20101204_AAAWVZ dzotsi_k_Page_086.QC.jpg
e6e7a4b8ed223d2801a31c5d208a32f3
3ef82214ea6f42120b66bbaad531e60cbddb6b75
6216 F20101204_AAAVSX dzotsi_k_Page_126thm.jpg
b25d690cb9ad86085b088031371dd4b3
e186a1ac2c482ef2bc7342213fccf5ac55a4d830
F20101204_AAAVUA dzotsi_k_Page_089.tif
e479b8359228b582d2e49b47debc1869
014116de9344e1e66e37fe38fa8bf7d7ebf577ab
24919 F20101204_AAAVTL dzotsi_k_Page_120.QC.jpg
657e29d10138a8271a2787da3d3491ba
148793b6735e324a53640a458b2174c8daa41ce5
25235 F20101204_AAAWXD dzotsi_k_Page_139.QC.jpg
a98825237a9b951017767761db1d6cbd
fccf6639cb2174d5ae04f6999fdca865cf95163e
6245 F20101204_AAAWWO dzotsi_k_Page_117thm.jpg
51e2e9f239b4074f89b1c0f0dab83a46
db031d1038df818d11586a80b7c0ae2b71d7ab28
80204 F20101204_AAAVSY dzotsi_k_Page_084.jpg
87c8d603a8f9a2c0f80833768cf42eed
e69769be3ddbe9f482597fbf23976dabe408fc86
22525 F20101204_AAAVUB dzotsi_k_Page_026.pro
9c36b273cae60558cb9fc93dcb7b43c0
230e69486d48e69689d29225669395e128584976
26869 F20101204_AAAVTM dzotsi_k_Page_106.pro
47d30e1121291ee7f401ec7129a4fd6c
5baaa4b77785f310fb006789f5a7427bbf0c5401
6893 F20101204_AAAWXE dzotsi_k_Page_139thm.jpg
04dfd35c7fb7fcc21a732ea9389ed569
8220f581fd27cfe71d3a30fbe8e574a3425d40bc
6563 F20101204_AAAWWP dzotsi_k_Page_121thm.jpg
42ee62109e1cb2230ea986ead8f8ae99
9777fed952362577d1106c34297880db1cc97f0d
13228 F20101204_AAAVSZ dzotsi_k_Page_134.QC.jpg
fa5cf321e5e3fff22a7e744a06518af8
5626a496452a5966226121cf967a270bc02b13de
51442 F20101204_AAAVUC dzotsi_k_Page_003.pro
2fb3a81f2e1d6a405365496904436c6c
88069832e7672dcb0bcfd73248c5e577d6b3e956
29685 F20101204_AAAVTN dzotsi_k_Page_014.QC.jpg
a2fdef739b504f90875c978241a64faf
5bbb762366f7b3ae4e51ceaf4ac0445444f31354
23515 F20101204_AAAWXF dzotsi_k_Page_140.QC.jpg
38435fb6918b455adb16fccb1c492177
e61d9c02ee389a4525c15222dc8548fd591643bd
7117 F20101204_AAAWWQ dzotsi_k_Page_124thm.jpg
026c1d8fdfc51252ebc4213d4144b5a3
b9671932e642fbe30b8d388b926c5c44ae54050a
92061 F20101204_AAAVUD dzotsi_k_Page_005.jpg
72dae97ad8cdb22a45455018a10b6623
3e880600cdc7148dd0bf85fa529c3eb235e390ad
51554 F20101204_AAAVTO dzotsi_k_Page_056.pro
61274ecdef32ab5eed886301521d7a3c
0a12e6c07305844480ef34298de0652ec559920a
2519 F20101204_AAAWXG dzotsi_k_Page_143thm.jpg
b9cbf28b895c3a2bf2d7dc2d83a6d6d3
3ea4931da9bd51936171457c1bf47bdc07be7a11
23792 F20101204_AAAWWR dzotsi_k_Page_125.QC.jpg
9dedbd1018e2ad69207699371315fe2f
9d6378219eea2ef79b12ae511e8bb634ee7cbc0d
2925 F20101204_AAAVUE dzotsi_k_Page_019.txt
3df60b2f87b8016151967152fe0d1dd7
b41ea430bf40754f623e9fc7124efde1a3355ecc
6707 F20101204_AAAVTP dzotsi_k_Page_123thm.jpg
4f8e761324bb145ef2d6d07cec2ca04c
fdd4a4ca00a0356ce96fc3007d3b7f5b1a3d5cc5
6639 F20101204_AAAWXH dzotsi_k_Page_145.QC.jpg
2db02a420e117ca21e0035d79c6557b8
8fedadd4e848f8eb6ba1760ddb9ceb2248aacf9f
22091 F20101204_AAAWWS dzotsi_k_Page_126.QC.jpg
40ff17c989e6308602ebaf8920254129
98cf798ec8f2c8da03cdac544a90a4d4fa758c53
80257 F20101204_AAAVUF dzotsi_k_Page_010.pro
4d57692cb44949bb403d71315bd8138d
5b9c4d972ffead5f7b63079f2292a915d8d1671c
66893 F20101204_AAAVTQ dzotsi_k_Page_081.jpg
d5a170efd394b846b95e3a014d2e1180
d2d2e7905f59104fc2b3c821500d7601cfe91e82
24296 F20101204_AAAWXI dzotsi_k_Page_147.QC.jpg
434af7bfc13139cd3742e48415359c87
f016837b937ddb02212287939d7cf85511cc017f
8843 F20101204_AAAWWT dzotsi_k_Page_127.QC.jpg
c8005f20108127043170517bd9df0120
6defe7277065ee773c6442feed7f17ef3fe41e8a
2143 F20101204_AAAWAA dzotsi_k_Page_031.txt
b454a8cf5106aa58e5ba4691cdf9a1cc
7acd23aa2ca16cf8fb2c851d9cdf5d63d4ec2398
1346 F20101204_AAAVUG dzotsi_k_Page_133.txt
15d9c3480a623b9e14b47b0105e6741f
b0e666ed3897c4b2b55126e6aa0fa073be3a77f1
2158 F20101204_AAAVTR dzotsi_k_Page_022.txt
e6aec427c84f28ba7a0a39a3c0fdd26e
cba6c2dd20e73c928aa0fcbd3f80536c5acf161b
5606 F20101204_AAAWXJ dzotsi_k_Page_148thm.jpg
2f68bccbd87da6695cd11262e5807c03
475aa2726d63290484aa9fd3aa5d31b108656071
2734 F20101204_AAAWWU dzotsi_k_Page_127thm.jpg
b9202b1a4919cb709d42bb74c774d5cb
688e3d6f2b4a5d7b0503b725db6dd2f28c59f50c
1051962 F20101204_AAAWAB dzotsi_k_Page_144.jp2
3613879682d8cc5cb2341f000d7f3b7d
b167a653612c4d52f43c5d3cf3da1c5129ac1853
20942 F20101204_AAAVUH dzotsi_k_Page_073.QC.jpg
c59bc42f5ff6adb89d108a9322377158
27cba57ff31e5a3a46e3b9c3c4adc0166f4b357f
3908 F20101204_AAAVTS dzotsi_k_Page_105thm.jpg
5cc0e16b905af6c95adfff03beb043fd
c66cfefc7c4d2ea76778adb5673ac847868b7d5c
7739 F20101204_AAAWWV dzotsi_k_Page_131.QC.jpg
90accf06f37ed301a693f1f689b01c87
9596f450494de3b03581597bf9102fc8406ab4ba
99291 F20101204_AAAWAC dzotsi_k_Page_092.jp2
8cec00e65f1edbec1343677f17d7f308
66102b8f48350dafb397ea645c0cf54ff9f5eb26
1982 F20101204_AAAVTT dzotsi_k_Page_149.txt
dea199d9729ad6905eb8037654e919aa
b423bcebfcefb9a5ff74c69c2e0250aeec82911f
21060 F20101204_AAAWXK dzotsi_k_Page_149.QC.jpg
b38ec2d03b56c21a62c85321f359ca28
d0de7ab944ad3c41f29fc6ed71fc7ac8510470ff
10566 F20101204_AAAWWW dzotsi_k_Page_132.QC.jpg
54292f284f124f5b48f819ad58cb093a
c363672e9e61964f8439d3e0596226f08f9b1440
F20101204_AAAWAD dzotsi_k_Page_120.tif
93507c69eb325a4a02d288d1d9a663fd
4e02e6a87c6dee4f67421edb64402adf10a846e8
117303 F20101204_AAAVUI dzotsi_k_Page_021.jp2
53986fea22dbdf86a5988951108c51cb
bda1ec84f77e7e078904b15bb42ea11f974432d3
27044 F20101204_AAAVTU dzotsi_k_Page_153.QC.jpg
58c1e64595c32187f0220cea268f585f
83f16bcc944feaef505be5f1ee4e65f106e61d93
6519 F20101204_AAAWXL dzotsi_k_Page_149thm.jpg
4c635e3587c8c962ee804b62f8682835
face468abb36882993425ca0e4a7647ee9af7e0a
3446 F20101204_AAAWWX dzotsi_k_Page_132thm.jpg
0f2df99fb48b822e217cc986604d2f1e
d70024ff54bd62611cf3056432b1a69a82e3bfae
100139 F20101204_AAAWAE dzotsi_k_Page_154.jpg
0a91098b809e2b8efcf0ff6c482ca8b9
b88884f5c5ef2cd4754679c40b6518a554c30443
41620 F20101204_AAAVUJ dzotsi_k_Page_148.pro
38b54862e9475cda488a05e37e0477b8
f4c381b0bffbfb91d7db0f997c6ed98feaafbae1
59462 F20101204_AAAVTV dzotsi_k_Page_075.jpg
04ec3ed09f9988ec7ebfd051abb45374
3b8b8c99dba991a6ed5d79a5302cb68278eceacb
6129 F20101204_AAAWXM dzotsi_k_Page_151thm.jpg
47470e408fdff9da05cc3c48fdd1911c
48bf005046e5864c7e269caff461039608cdcae1
11440 F20101204_AAAWWY dzotsi_k_Page_135.QC.jpg
97e498c5216c174bff5f3a977ee90352
688f3e1a5c74103658011c4767913fb4f30375d7
2172 F20101204_AAAWAF dzotsi_k_Page_077.txt
69be1a321973c9349a38bdb59b36b490
4f2688c9b11ded96b8fb4567aede1d1c97ba2a39
375784 F20101204_AAAVUK dzotsi_k_Page_107.jp2
0113f2aa6e48ebe37bf006b80c691b4e
ab76ac2ae42d50f7705b49000caa154be2e1bb44
F20101204_AAAVTW dzotsi_k_Page_084.tif
87813c447b4e34a809f4db7f374e46d7
27de0ab8565d4c9f8fd0088bdcc00ffbd4f40ac6
7406 F20101204_AAAWXN dzotsi_k_Page_153thm.jpg
3caf91331e8709e39a36beadbd8d23a5
3b9b2582595d83260fa5583003cb3ae795b9e06c
3882 F20101204_AAAWWZ dzotsi_k_Page_135thm.jpg
2010044347115833ddd4568cd1d1b655
125df5d81c8378107b95d19b4cb63adb8f89bc46
F20101204_AAAWAG dzotsi_k_Page_106.tif
ff290e42401ef05336a3f23500705bed
a2794242116acd3f558924586d61f01d4c3471b1
49955 F20101204_AAAVVA dzotsi_k_Page_077.pro
42c2686d158283addd3757ab450c44ca
653b4ab397cf669f522f330a0b8462aa4d52a3b2
F20101204_AAAVUL dzotsi_k_Page_150.tif
aefbdedd79c9d2bd323418cd578ba402
8ddcba4848dd18ed1b7b7fab94add85e5cf0c0dd
73594 F20101204_AAAVTX dzotsi_k_Page_044.jpg
28a6bf41011e80a024a3ef3865e14ef3
c7a21988d3448d6d4f079e24bbe39daed5802440
7225 F20101204_AAAWXO dzotsi_k_Page_155thm.jpg
b7b24f72c82cfc3d0ab954cfcb713a01
b13e0c6c3cf991143cd75fdb1b4420aa7f72e127
50532 F20101204_AAAWAH dzotsi_k_Page_076.pro
e4fb4cc033256f5389614fd01886a00c
2ae15886e88177733641591b829f01aac8dc2d3d
7126 F20101204_AAAVVB dzotsi_k_Page_013thm.jpg
a283be851342050fff5acfc242f9effc
981966994a0acfa5410fc4541afb081c8af82be2
725009 F20101204_AAAVUM dzotsi_k_Page_059.jp2
4d3b48faf50a2dd00ac0180ba0b710ec
d2942a8dc3d6f0adf844f4fd1512cfb7ede9509a
119698 F20101204_AAAVTY dzotsi_k_Page_006.pro
be9acc1e62895011cb13803952b5007e
d9d8195e7b6deb92577d8a3b63619071c1033f85
27390 F20101204_AAAWXP dzotsi_k_Page_157.QC.jpg
0faa3cb2ba599ebe6d43ce7ebcb65c74
9bb8da6a18fc39d048eb279de3ad8aa55aec2c58
26521 F20101204_AAAWAI dzotsi_k_Page_124.QC.jpg
7adb285baf8dd00b040458e95c42ad44
c21665be2de08fab22edc84e827b2346c331e76a
12912 F20101204_AAAVVC dzotsi_k_Page_097.QC.jpg
be6b11bcbd45c9b88fd425babac5b3f4
cfda21989feb44d30eadb4603058e865f482c8d4
23373 F20101204_AAAVUN dzotsi_k_Page_034.QC.jpg
bf10f159c213ecaac01af3c602b47a60
91a980cba473ced19a67850590d9de9b94ce0226
6903 F20101204_AAAVTZ dzotsi_k_Page_091thm.jpg
c8da01436191c541e2119ade4c5a45c1
d921db87ce5b5d65fb81407e1674addf54e1ffb2
7426 F20101204_AAAWXQ dzotsi_k_Page_157thm.jpg
a5bdacd0ab2e154953c38912553f4b80
be5b9627b351cfc10f462eb1c071d7f66bc073eb
1396 F20101204_AAAVVD dzotsi_k_Page_150.txt
5b4882f2c08d98631b5fd319c6b3e843
9661f3e11a56976132fb2bfdb45a2212e30b1176
67202 F20101204_AAAVUO dzotsi_k_Page_130.pro
31e671ad006157f45f85684370dfd03d
fbeb637d0a324ba608e9c332d4e2d670f132d8e8
2482 F20101204_AAAWAJ dzotsi_k_Page_145thm.jpg
6108a756d2d43622c5c999442e934f67
8db3fc6767e933d27191b92b7271a137f5c2be84
28795 F20101204_AAAWXR dzotsi_k_Page_158.QC.jpg
55abb1195cb030f0e0fd5025256e6b3c
53bd7d49f9172b7b856ff895834cecc413062879
33849 F20101204_AAAVVE dzotsi_k_Page_095.jp2
971910df3f9168d8280de17f67e03d2e
40016477360d3076477516e9f59c19d8633b53d9
24558 F20101204_AAAVUP dzotsi_k_Page_039.QC.jpg
964d6c4519712e68da7557f7ec571547
076d7164a03f078d0542ead291404bf21fcfaa98
2783 F20101204_AAAWAK dzotsi_k_Page_154.txt
1a20d11c931bcce5040bb8a27e5d2bba
33db76db46a367cf216235b185c38537401b8ef3
7407 F20101204_AAAWXS dzotsi_k_Page_158thm.jpg
832f6d75256cefe06be87adb9c715849
41b99404ee83c01da103e37b7a8313567929701b
51120 F20101204_AAAVVF dzotsi_k_Page_080.pro
cb4fb23a5effa061d6143faedaa9a8ad
5b01ad6e5f5d5a4c519c52f67e315edbe1a03073
95396 F20101204_AAAVUQ dzotsi_k_Page_157.jpg
4d9b136aad826eda1f54b6e7522c497e
6212d8318ba15e193463e7e3addd5e31928c0d7a
15609 F20101204_AAAWAL dzotsi_k_Page_060.pro
d292e3d709eff0f5885a9bdb6232194b
fc316ced968829a351e549a551237bc47bb89dda
24038 F20101204_AAAWXT dzotsi_k_Page_161.QC.jpg
3559361d6ad95430e501fd5129a3d469
f083107f00f932388c3dc10fdae79b6b4d0ab158
52301 F20101204_AAAVVG dzotsi_k_Page_137.pro
bbff2d315153e31a2d2512eba256f12e
3111b9cb7a2dec6c329bd55f5de090ec000f44ef
35690 F20101204_AAAVUR dzotsi_k_Page_135.jpg
53f785dce34fd3e0f4720acbf3fad55d
61423661b1fd3be40ad376ddae0469dc59af543a
6281 F20101204_AAAWBA dzotsi_k_Page_042thm.jpg
886c453b141a31db380adf71f66f88f3
74e2ca546007190864cb37faa5cccaa4f7b5e8b6
54805 F20101204_AAAWAM dzotsi_k_Page_017.jpg
7e72b79464050c87461dff16222d8209
be086b6e378131409552d368f77660fb99adb4dc
185591 F20101204_AAAWXU UFE0021774_00001.mets FULL
804e8283987a6133e1e1a4fff458619a
f35f58dd0227383ab3bc2993be032fb528525c4f
7131 F20101204_AAAVVH dzotsi_k_Page_156thm.jpg
8a8b9242ac7001866cc2c7f0f08a05ff
e5a42cea01c1dc5f2b6d0b4bd03999ab452d13cc
4941 F20101204_AAAVUS dzotsi_k_Page_027.pro
c5b939b5401566c409d5d434fff23deb
898d252b0f2feeefe0137fc0f8974016255636b0
116241 F20101204_AAAWBB dzotsi_k_Page_122.jp2
2224222e75e861ec42a54d08e5bc58bd
f7278f2965224eca5e9d05ee116ffae7ad9337ed
26769 F20101204_AAAWAN dzotsi_k_Page_155.QC.jpg
be1a0e27ddb38ca0138b105e8973d1a9
a78b29bb3bc4bbe3d91d803e98133bf23a02dc19
25394 F20101204_AAAVVI dzotsi_k_Page_094.QC.jpg
62cdeac20295bcc7f794c9c627a18db0
600ee756c0c8fea1403e08153b36cf409cedf958
56726 F20101204_AAAVUT dzotsi_k_Page_148.jpg
7d96ed6aa892b8516d202c7e17339a9f
ce99c8360c1b9d4c51f046c55aa4c196c7cdc313
92564 F20101204_AAAWBC dzotsi_k_Page_153.jpg
b83e19ea301de8bd4941cb05f304980f
f9f1b06180ff7eec2a4338e97bcb05cfcd1d060a
99086 F20101204_AAAWAO dzotsi_k_Page_013.jpg
7d3d1136d4592368a5ddb7a5668c97e5
a509afae6233bc0842bf09a91ff500b953a47cb1
96843 F20101204_AAAVUU dzotsi_k_Page_117.jp2
dbc66b0bb00224bbb9a887010dcffe60
273a6856d93ffee4f0864b63421b7e51c75dee08
1161 F20101204_AAAWBD dzotsi_k_Page_058.txt
15c45461cbbc5ed71035e9b2fca96f0f
f890d13a8ab7265318ab408310d992147a907fc3
F20101204_AAAWAP dzotsi_k_Page_138.tif
37ac0b583953476e9d551213cb7c205b
890382126b2ca333470a916277fe11295b2c5f27
1147 F20101204_AAAVVJ dzotsi_k_Page_104.txt
63bf08b2df0d0aa9783bf62721e541f4
9f053d04d4b0df300b4c3801a1998fde9185d576
F20101204_AAAVUV dzotsi_k_Page_081.tif
5fa7c59d4c9f395c83dc4d24d59cb20c
008192602ed5a4e6bcdfe0adfe46869fed51a8d9
54368 F20101204_AAAWBE dzotsi_k_Page_094.pro
cf6914313d4cd5656a70d542c8a17615
8fc57463a5cd27389614d2f081abe8310ccdcb6e
48184 F20101204_AAAWAQ dzotsi_k_Page_111.pro
f81844b90c34cd32852ce02a11546510
7e77876dae1360135b0b29a0348a233f216b3b08
4337 F20101204_AAAVVK dzotsi_k_Page_099thm.jpg
3d74a8e375de8f6cdc39a4481fbdb74e
d2311dc8851034f3fdc9aaf3630f06f465720c50
601 F20101204_AAAVUW dzotsi_k_Page_132.txt
2663e5f01e643d8fc4e9960a406dca91
885ec50f328db4b024d5b8e7cc8cd1623e1029b7
76610 F20101204_AAAWBF dzotsi_k_Page_066.jpg
39eb2d03f3d550d61185b5583d74bf40
f5c3e1942acebdaeea0423a361ac16ca101cd5dd
6458 F20101204_AAAWAR dzotsi_k_Page_052thm.jpg
1166071ba26d8cdb89e9c0c092b00140
578ee1eecba7c1aa377040459972eeee32177913
104869 F20101204_AAAVVL dzotsi_k_Page_098.jp2
b25de82472a17c1f24a709a19732eedf
2215174a28ab8202b683e4a2620bab777a0170b7
95935 F20101204_AAAVUX dzotsi_k_Page_071.jpg
ebd02c7d0173d978e7cd8c9a294dae21
8ceab74f00a83c60a47fcb3f53ff2ca041f4a69d
1816 F20101204_AAAWBG dzotsi_k_Page_042.txt
4f44f295dfc6c5d986b7e281eb3b9c1b
84f5345650c6c64f7bb10c85713f449ac2cec1ab
632290 F20101204_AAAVWA dzotsi_k_Page_136.jp2
ca0c27cd30bf52d1b16cc43c54291542
ffd61d405dcf9874c41499203fdbda44075d4651
2184 F20101204_AAAWAS dzotsi_k_Page_020.txt
458720e09a98eee4a2f7d640d96af52e
57767a681a4f57ca2fab64768b8bbfd6a65f568e
1366 F20101204_AAAVVM dzotsi_k_Page_002thm.jpg
1d4155c4aac06a1fa5ed6c73a920cac5
968be84fd386a90fa473dc370a60ac88620d55fe
F20101204_AAAVUY dzotsi_k_Page_001.tif
76415449bbb455beb726f16e65dc41ec
fd7411cfc37fedd30a14be8c11fc867c91f5d1f5
2227 F20101204_AAAWBH dzotsi_k_Page_129.txt
83cd373e73aabd30c2e1f6ba628926bf
697303affaca5eb1104bd26e6e03e170e1f86036
5895 F20101204_AAAVWB dzotsi_k_Page_072thm.jpg
c8cceb1179e5a4f8724968c7e06e8519
89c426fcf71431226d746fda45b4cb9768e05149
46905 F20101204_AAAVVN dzotsi_k_Page_116.pro
121a9d63008f3a853581f834b8b4ae92
062b1606a05707e0c0cd6663085c6eff2261ef92
20907 F20101204_AAAVUZ dzotsi_k_Page_058.pro
c891fa6ae0d3b63f37f22a6c1606714a
494bc7e719e2ccb9b7073cc3ab8d43c28b4083b9
1001 F20101204_AAAWBI dzotsi_k_Page_109.txt
7eca499cc54846fe664aaaa1b1989af8
c0f2a065b0ccf2a38ea4281782671c5e0404b4a6
101969 F20101204_AAAVWC dzotsi_k_Page_011.jpg
29dd23a4548778d8f90e788142009d0f
db2377f984d4bb2b9e7f84eb40aa39c2762c52ce
6584 F20101204_AAAWAT dzotsi_k_Page_064thm.jpg
08fcbed379fea620ed3f79e8598fbf3b
0a16ddc7d66bad6c3ee3dee172e0e140e021b0fe
27786 F20101204_AAAVVO dzotsi_k_Page_027.jp2
44a3467e1ab744c1d3c0efbec7e51247
d51caa2a28f70b5e7b24c1559b7fd7dfe735d462
26144 F20101204_AAAWBJ dzotsi_k_Page_057.jpg
4965ed899adf764a4d70eee49714e3e9
96b0bed6fe6655bf97ebc2c5d567485f58f3137f
141044 F20101204_AAAVWD dzotsi_k_Page_158.jp2
dd20a753f20a24c0163085a71064c586
9648d6a71156198d0db5f0dc2248bb5530a43ea9
F20101204_AAAWAU dzotsi_k_Page_144.tif
d70843c9342733ea50cc76912b0f2ba5
c3724798fa370a139ebe58fb62e49457eadbaaa3
46591 F20101204_AAAVVP dzotsi_k_Page_016.pro
0b559378b920755f25d5a12259155d6f
80b82911c06fb20c8a0e99105640e85c9442462e
F20101204_AAAWBK dzotsi_k_Page_046.tif
8bf16bc468ba25afd2ad30c4a74073c7
6ccc2fc1447f95ca83b85241d3196a71ecd22c8c
6392 F20101204_AAAVWE dzotsi_k_Page_076thm.jpg
87fa264f0eb56db86ca6c1bde0659fd4
1d1792c77c87274e5401d91efa8c68a7b56a9384
1051972 F20101204_AAAWAV dzotsi_k_Page_007.jp2
102121fb4cf79cc8cfa135639b40e7f7
8f286df538acabcaca439c30d30224a7ee250c39
79429 F20101204_AAAVVQ dzotsi_k_Page_051.jpg
9840558d1de9ac1db8bf034a02cd6aa1
34c63b301fd2481dbb8dcd82cdac198de845caae
49029 F20101204_AAAWBL dzotsi_k_Page_133.jpg
517c043dfa9900aa0eeaab3736cc5037
a563042c4bc61b713e4e82eca145950046074501
7259 F20101204_AAAVWF dzotsi_k_Page_010thm.jpg
9d580650cced8ba6972204571c250fd4
9f271855035779b9a671b7e55fb4c5c4b9c9e272
F20101204_AAAWAW dzotsi_k_Page_143.tif
63c21e557d2a8929c5140110907de887
5d311e5f808ebbd85c3e5b296b5e84f434570cbf
66927 F20101204_AAAVVR dzotsi_k_Page_077.jpg
dc6b1bea7697d475d2a63502e59966e3
439c518a6f84c5bbd9f38747f30ed626af70242b
21541 F20101204_AAAWCA dzotsi_k_Page_102.QC.jpg
77cc9c72b42e72fd7e7c019cdcbfa3c7
db4dc9fc6de35adc81091272e7763be051fab773
7402 F20101204_AAAWBM dzotsi_k_Page_001.QC.jpg
a90d5481b94d0d82a9f81d4bd1769b75
0440e525296b05fb3615d719aa2e7a440c95f6ad
5798 F20101204_AAAVWG dzotsi_k_Page_118thm.jpg
a26f694038d2fd439abd5ef98b373520
6541f9ba8eb815e9b0bbc118e52ad2e53c2488b1
2128 F20101204_AAAWAX dzotsi_k_Page_110.txt
234fb064eb72cc0d7fbb1d4a73994235
6a0680209bff42bb5a9e8387359aee5c28dc8da6
23029 F20101204_AAAVVS dzotsi_k_Page_033.QC.jpg
2a3bb6881acbc985f53c5d6c33f29652
0c166c29e6a613dcc7dc5d7ecc5c69b7bfec6b49
60556 F20101204_AAAWCB dzotsi_k_Page_004.jp2
6527047cb26b10003b56c52e35f22eb5
490211984fa72b5877b6ca59564f5b0f86df7c30
79604 F20101204_AAAWBN dzotsi_k_Page_065.jpg
acf2e68d5d3c52bbaafbb346ced839f9
dc12aceae428adae12b14f47f5610cb689a52d5d
4817 F20101204_AAAVWH dzotsi_k_Page_160.QC.jpg
49bfd9f4f1fd172b48bfe4d25fec47f8
ca0edaef4cc61475ebb6619362689777d7a69de5
90712 F20101204_AAAWAY dzotsi_k_Page_072.jp2
6ec38cd10a7e24bff586d919df891086
5890e7e2f6199bd6c5836cc198e5ebbbe23ffa5b
72045 F20101204_AAAVVT dzotsi_k_Page_035.jpg
51529d7394bfbc255db3164ed5dbfd1e
63adb402ff7855aeca7afa3248c7f07046183e84
115259 F20101204_AAAWCC dzotsi_k_Page_093.jp2
9ab5a22b4a75cfb6147a7df1fbd46f93
be94d936ccf924ca6f8e2e6c7b430d9d35e0a832
26748 F20101204_AAAWBO dzotsi_k_Page_006.QC.jpg
97d84f0776a563da28c8a9969edd673e
a6632371c1c8b62d5ef51ae0cfe2bc677bcdfa08
2019 F20101204_AAAVWI dzotsi_k_Page_008thm.jpg
ac279044ed5b84924b9fc94af63a4058
f2fc9d64239604b571d9afebeaa0650b9e37b296
1949 F20101204_AAAWAZ dzotsi_k_Page_117.txt
b3f900d9921fa9f68e2afabfbcccbf7c
8977c4a4ebfddafb80e8ff7e5ae241d9f8619442
46255 F20101204_AAAVVU dzotsi_k_Page_074.pro
6dc1ab7983253e1ea847998b1620aa58
40029df2087374b4d9bf2820a4a69af5dedb62ac
23894 F20101204_AAAWCD dzotsi_k_Page_131.jpg
4ae877dec2d47767ac4c78e482348c85
07fe4047369a3923ebcb50f853b37fce09932d92
F20101204_AAAWBP dzotsi_k_Page_103.tif
934f883c8687f3d0d2158a99b8f1faed
bb369dc793ba753597ae62042186ca0bb06a8657
6291 F20101204_AAAVWJ dzotsi_k_Page_142thm.jpg
95b0664ef8055b409102f15964184802
5ba099971d09e3367fd842c5d019521ecb9474d4
70926 F20101204_AAAVVV dzotsi_k_Page_113.jpg
9274be835d6258dcb494f90ccb0d1380
ded9d683261567bd2bf7c1e969aa97721ec6f749
93551 F20101204_AAAWCE dzotsi_k_Page_102.jp2
875c71cb4f5a79a6e21251d1c1952c31
3cd02e8816f8807f9ae6907b4ad6f91eff72ec81
2026 F20101204_AAAWBQ dzotsi_k_Page_080.txt
73ff61725f33003127d3e3a09d3d3e73
2db599ec781528a35b888a88ea7c0ecdceb57387
17547 F20101204_AAAVVW dzotsi_k_Page_101.QC.jpg
5ca514d5fb79e80d033daf51902051a5
f0f5d2df594b672479404aea8226ff1dcb03b1d6
97645 F20101204_AAAWCF dzotsi_k_Page_074.jp2
3188c20b751389ca38ef94b0c3d9f054
2ec1ce96ec2ce358bfa5308c789ee5aca8caee80
109431 F20101204_AAAWBR dzotsi_k_Page_052.jp2
6d03d5883a09e30ad5c1a471da8f9f70
ec590b512209dcdb217c3fa144ba76a032641e6e
F20101204_AAAVWK dzotsi_k_Page_124.tif
c2f18c716961484a2a902b3bd54bea29
20b8846969a26133a58fa94583add9344676e156
51063 F20101204_AAAVVX dzotsi_k_Page_048.pro
43d0a6746fc9c3144c36b9fb40b2410f
c499b2a076c2d8d0788342d0e1f658966d4a0057
29049 F20101204_AAAWCG dzotsi_k_Page_046.QC.jpg
6a5f4083d3728480d08a43dd06624b5d
5a5679592a7de4bb8ea9e9b0024fe980560ebc04
76889 F20101204_AAAVXA dzotsi_k_Page_093.jpg
75924ceabbf30c044591491b92b347a7
f1ffb2830242fb8e63d29501f14a4f2604906c41
14388 F20101204_AAAWBS dzotsi_k_Page_134.pro
8ccf9831ea1ea96f2b28df339f792842
07caccc0dac2351fd1ba04a41baa64d4d04e5fb4
6729 F20101204_AAAVWL dzotsi_k_Page_048thm.jpg
6ce5430828b22c867a46416df6436942
00bbf7969b769e8da2af4de2664735edfa705334
22527 F20101204_AAAVVY dzotsi_k_Page_043.QC.jpg
eeb6b4d30a17722d8831d4be737650ff
f4121a526b6dc29f66255d351129b90dbe55d1cc
70159 F20101204_AAAWCH dzotsi_k_Page_140.jpg
2d27271210dc9a98b069793adf7040a5
b600772e69a83e01ec90403d364155fcc696e912
2800 F20101204_AAAVXB dzotsi_k_Page_159.txt
4db0e06590b7335fc9d8e2589ad6a3c9
0cedd06686931979f7273fe622858479ef28bbea
6788 F20101204_AAAWBT dzotsi_k_Page_066thm.jpg
a03fd5db4108fedf8fadcebf4fd0db51
b13d779c6421f23a76ad42f9d34bf12f2ea90660
23303 F20101204_AAAVWM dzotsi_k_Page_113.QC.jpg
0b25b834a20e5d2296920783dd94f82b
0819fa0a51ce902a17b366f6bfefb4291193393d
23321 F20101204_AAAVVZ dzotsi_k_Page_053.QC.jpg
40d5761f3440a802357c14fa604c679a
2f77f178e28ba1d56b857b90238e1225782d6617
23826 F20101204_AAAWCI dzotsi_k_Page_008.jpg
73a6922ab53375e4bf419b7da9d20b13
0c8f793203f3a2ff9c15f8f163d3f73371774915
23074 F20101204_AAAVXC dzotsi_k_Page_005.QC.jpg
14b5caf77c3ff78fce8c0cf1a3d203e6
fd05db3bae61c5cb306e7e27bde30f07b45268b1
79183 F20101204_AAAVWN dzotsi_k_Page_014.pro
4e7c13b6a9925ee9e22ed3a52bf2345e
84c807a16620682d6c833a3827fb8ad71324dc6b
116570 F20101204_AAAWCJ dzotsi_k_Page_049.jp2
3ad412a0b65a1b8dcc7f67f5d19fb6b7
639b51ec4babb5bb62fbd42eef08e29d18596757
56638 F20101204_AAAVXD dzotsi_k_Page_038.pro
883d084045d56a82955c037060ab18fc
7ac80daccdfabed2ce0cd4441c88e3c4334abe44
24656 F20101204_AAAWBU dzotsi_k_Page_110.QC.jpg
62171a7510b48f4acc53ca7d2aa30256
df55ee9955f18930a8f5b9595a8cce58d44f5ffd
4618 F20101204_AAAVWO dzotsi_k_Page_063thm.jpg
cb33362eabadfb7ed3ba1490d305c7c8
3ecb4ec164d1f4faf8c8f8a1e465b87bb85eaa72
48621 F20101204_AAAWCK dzotsi_k_Page_140.pro
2d0f10b217cefcb66bee5a8e36a2b98e
59b6ecce3246108898f0e1d838eea071b72e9763
1051979 F20101204_AAAVXE dzotsi_k_Page_014.jp2
eddfb963388f1c81200cf96a46c7abb3
99d67e2d9c5ba7890f8f57722d5c7114505167ae
760 F20101204_AAAWBV dzotsi_k_Page_008.txt
c80aacbc2dd7c89b2effb905aacfac17
19f0137ebe030dd6b0e82b24cc165440a1e82f5a
22643 F20101204_AAAVWP dzotsi_k_Page_042.QC.jpg
cd7910d1efe9e194dd77d394aa2c4a20
7c14447fb97ecaf16202acb2bf2b345d83b05556
118541 F20101204_AAAWCL dzotsi_k_Page_065.jp2
3d15a66affa153c968f1a9f3c1377e5a
552b1c7cf26687870e8691e65191f29897c4dedf
67064 F20101204_AAAVXF dzotsi_k_Page_155.pro
86be4c9d492fd987a616fb73de355c4a
c916bb96b6a30d483249ea6c615000ca990829a0
836 F20101204_AAAWBW dzotsi_k_Page_057.txt
ed109dd6f44e56517b72d3a6046c14cf
7441295eb22e7a6afd4cd67a0dd2b8577e0266c2
481 F20101204_AAAVWQ dzotsi_k_Page_001.txt
1e839b55b59edd074a867ba09f4faffb
8ab8fec2bab4d632eaffdead82935a94d8ff0abd
5169 F20101204_AAAWCM dzotsi_k_Page_141thm.jpg
366181d4d703c0c960fa4948c2eb2b69
3c99b4d95d52539d1d45802b9e04241d854bb6a0
F20101204_AAAVXG dzotsi_k_Page_035.tif
49ce4ceac7e8c1a25ea0fc1c1a8b6d3e
30beb220b4cac1eec85fb0c149c5a5a6b0263858
68960 F20101204_AAAWBX dzotsi_k_Page_076.jpg
b5f62480988f6836aeb4a55e393f515b
5a72688c241dd85f9e05f0563eb0cf71194045b3
4667 F20101204_AAAVWR dzotsi_k_Page_060thm.jpg
17ac8f1bb58645fae033cb43cecbd994
98795e39dd8472a3ba459f4918d98dd4bdefe681
21498 F20101204_AAAWDA dzotsi_k_Page_117.QC.jpg
81d4cdde980dfba22bacd8a1a0fd3b48
667b989b4f985e2839e359125e31c65d3b123ea8
25545 F20101204_AAAWCN dzotsi_k_Page_032.QC.jpg
f76c76e1bc7a32ad82de126e0cafb93d
c11b443aba0ac30660d816ab1709ac251cc6af19
6890 F20101204_AAAVXH dzotsi_k_Page_093thm.jpg
bb5f9ad46c873b8548ca8a2aa113a49e
ac8306b8ce311d5efd74fef5085fd2c2920a41e1
2077 F20101204_AAAWBY dzotsi_k_Page_120.txt
57c2d85baab929747f742b461364cf45
40f8fbab88db19954e7351d0545908da556b72a4
3006 F20101204_AAAVWS dzotsi_k_Page_009.txt
8733309ef1b87eeb7229eec90d3e90e7
eacf0de2212a757633ff62b2272ce6ca3373ac8e
F20101204_AAAWDB dzotsi_k_Page_121.tif
4dde7466798491425db8c7c9697190f0
bf2a7945326d1f9afe400f63488652f4726be1c6
64076 F20101204_AAAWCO dzotsi_k_Page_117.jpg
655b4f8e31ce211f7746c8aefb731f16
8665154423a45ef2ba69ff412a8baa7b89e65708
6651 F20101204_AAAVXI dzotsi_k_Page_110thm.jpg
e2306e46af8a3d9f63fa9bc61e6b40b9
d1e7fa5b7973c918ab3c05d3743f99ad2767bc50
2771 F20101204_AAAWBZ dzotsi_k_Page_158.txt
ea43ceb3965c2787bb67d05955dee5ff
91a6d785915f73d6a725768b8549a320d1540e91
97834 F20101204_AAAVWT dzotsi_k_Page_009.jpg
eae477e2e7efb8797edb4cc36c6c2357
eb183df8bdd1e494fe9f305f9542bfbcdf5a2ffa
F20101204_AAAWDC dzotsi_k_Page_050.tif
585acb8d96c1a7ac0db8916dd8e8ea55
50ba2d5da444f38646cf104bd6e1f7306331df65
2820 F20101204_AAAWCP dzotsi_k_Page_157.txt
856a335d7166321082488d2c30a46d8c
cfd1c1b98ec65781dbd82f47599705245a0a8a48
25833 F20101204_AAAVXJ dzotsi_k_Page_021.QC.jpg
b58f9d39c0c8c8580949e5a67eff0ca2
d0e53aac47a2ee1e634de8ea3d546f3f92e0bf3a
25182 F20101204_AAAVWU dzotsi_k_Page_041.QC.jpg
14f4d52f165f28545b8a06045f975bb1
6a493ad0eea893f1a2145ee976e9e261058e9e9f
7065 F20101204_AAAWDD dzotsi_k_Page_084thm.jpg
bbc741e9269d831af538c6b1313cbead
479639be298fe1502e5c0b48c0834602ded84dd7
106596 F20101204_AAAWCQ dzotsi_k_Page_112.jp2
5af3c71d8669791791d6e8058fc6b5d1
a63a203af00ccc9addba996076122249c6d69fc8
22110 F20101204_AAAVXK dzotsi_k_Page_067.QC.jpg
2ab151b508e97a334b93cd61928a2323
b611d25c4db857e5012d8bc1743be444fa74e953
6077 F20101204_AAAVWV dzotsi_k_Page_081thm.jpg
da81ea469f25d15bb4cca88a860d3e86
83a6d96b658a3d776cf15d23e1a304988b5d5e1d
68384 F20101204_AAAWDE dzotsi_k_Page_153.pro
fc5839134e9ffe061b60e50c06cecf32
3d1ce1c8a72042cf84d5f0b2f7c8109c0d7a3ac9
69783 F20101204_AAAWCR dzotsi_k_Page_082.pro
cc8b3af927d5cd7b890fb2bc48c0151c
a5a2a37ff1b130524f1fac085abb40d9c0bedfca
47539 F20101204_AAAVWW dzotsi_k_Page_060.jpg
571c3773d405a1b2226f41fbc94f901e
c6ae76cedc0fcee26e12b7fd5744e4160a8e2a25
17366 F20101204_AAAWDF dzotsi_k_Page_063.pro
da7745707bff15945712580c74f66314
1373d701e052a0e9cbc2b3022306cfd0aebed637
24109 F20101204_AAAVYA dzotsi_k_Page_116.QC.jpg
48df935a86c9addf6a638ce3be6778ce
9420473168075376ac153427b3105c865f78999e
592 F20101204_AAAWCS dzotsi_k_Page_127.txt
5077c91a0a6f5251c7467ae6035c5b24
ca7e69e62a86ddca5f4b55edc52989cb01ce2c43
50485 F20101204_AAAVXL dzotsi_k_Page_088.pro
c59da1b3e2dbee9b20782b1521de5377
0fc5940f1cafa633f5ee07bbd6dee4b6edbfb21c
101685 F20101204_AAAVWX dzotsi_k_Page_043.jp2
3edee0e6fc87a471b8f491aa1c3b6abc
7f3421b28be167e1576bddb12f028b6252d3c83f
1996 F20101204_AAAWDG dzotsi_k_Page_136.txt
92323cbcbd935dc2b20f76adadf1083e
d2a5e0d909c0fe59f0e8966bf50a3cc9dd6360df
15713 F20101204_AAAVYB dzotsi_k_Page_062.pro
3595b44866ab27f712bc6fe18c314fe9
1eb77954a839eea2cf9b84a46ecd864da35d3418
77434 F20101204_AAAWCT dzotsi_k_Page_139.jpg
8ff2250a66597835fc385d5822764b6a
8254dc902e4bd71da72d30a7db4a0d1ff786f4f0
F20101204_AAAVXM dzotsi_k_Page_027.tif
b3be371d86ee5180d6a3a3e8693170f3
7b4d199df9281cf17a1a109196e4ceda94f9725b
114567 F20101204_AAAVWY dzotsi_k_Page_138.jp2
96e037e5597ffbab3e407dc506c8f6fa
5b0cc4241ff219e6d01bead437e0e3898ab73400
78904 F20101204_AAAWDH dzotsi_k_Page_047.jpg
0ec467fe73e057ea6de2b440d5139c6a
74bd728d920cfa17e34d26d6e15c4e35f0f5541e
119057 F20101204_AAAVYC dzotsi_k_Page_022.jp2
6e292c6578792f8f1ce789a24f73e902
6d1c6ee52d738fcf98a4185e6e5311fe7b04eedd
1051924 F20101204_AAAWCU dzotsi_k_Page_159.jp2
6e6eea16cd1521f08a8bc2b8221e747e
3d54fdf308610e65a54ac49250aa09a5d6796fe6
23890 F20101204_AAAVXN dzotsi_k_Page_121.QC.jpg
27d5dd5e8a24dafb61a68d91bcd66eda
fe7fc995cfe3a41da048be03882c2a73254fbab9
2121 F20101204_AAAVWZ dzotsi_k_Page_023.txt
114a90b8a07b5641e5dfc61451d62b5a
e5ca5c2e927d10f4c7b23fda9ea6ebe469dfe2b4
142701 F20101204_AAAWDI dzotsi_k_Page_071.jp2
9100a015f4adbdcff24a60f99c0bf4a2
983fcc9f09a8f07e9330c651eb2701cd0a1f72cd
4627 F20101204_AAAVYD dzotsi_k_Page_150thm.jpg
2c7273bebfb08fb74be7450b66096b87
bc4480f847fc1b1ba12a96a0b486a7b386343073
71569 F20101204_AAAVXO dzotsi_k_Page_098.jpg
4d7ba5b0a1967a8510b19fcc348a2541
378deac29f57177880dd53abc1f6b06740b2f92f
31203 F20101204_AAAWDJ dzotsi_k_Page_050.jp2
6454309673bf067c8d6e3ab591cc7990
b421122d58740f4f642852cb4e52da0f3d4c9db3
74494 F20101204_AAAVYE dzotsi_k_Page_120.jpg
5bbf234b792c6fab782f37c75d9143c8
b9f5592a009ecd4a0f40b95181bb3531222ae030
52611 F20101204_AAAWCV dzotsi_k_Page_120.pro
741b13469346763fab9a0c69650e54fc
85ab08daefee92f595b632bbe7f8b22babc8e870
7902 F20101204_AAAVXP dzotsi_k_Page_159thm.jpg
938f8a93039742fded864a505b22cd53
7eb4107e3d658c589fda2b5e05726e24c29b3604
100750 F20101204_AAAWDK dzotsi_k_Page_149.jp2
1e3e55277307bf7cb7aab10e423a57ac
b46d55d7620574677077b4600ef11a23e98b1e93
51494 F20101204_AAAVYF dzotsi_k_Page_110.pro
c971d65385feb7c1c57a36267d453c12
cc71637df011fabc085712d071e08dfc5662e2dc
1918 F20101204_AAAWCW dzotsi_k_Page_113.txt
bd3afc30302427d1ea40897f06a2ec2c
cd1ecba3ae1637d9ae4ddaed6e091d49adc26f7c
2010 F20101204_AAAVXQ dzotsi_k_Page_069.txt
f7261acf4e554f1aa85c22eb100de2ea
521bff38bc422c523fea6483b5399fc788a543f1
27881 F20101204_AAAWDL dzotsi_k_Page_009.QC.jpg
edf4bf0d134278f4a614a893e67c0d5e
92918f9856cf2c2f1f1aec64f58c325cfa9a3ee8
23444 F20101204_AAAVYG dzotsi_k_Page_128.QC.jpg
bb90364e7959e0d26fc1632622066804
bcb11c4980e6dc8eab549ea0e44586dbc514af36
F20101204_AAAWCX dzotsi_k_Page_076.tif
446295d57ca078fcd0653f51f1be06e1
2c782fe9c87d1fd5bf9ec67402a9acda5b6dd582
F20101204_AAAVXR dzotsi_k_Page_083.tif
cbb8fb7595eead1362de627b08d14bf9
d9abf3c83f103c6e8a5ce38942534e8d4bfd1c85
43353 F20101204_AAAWEA dzotsi_k_Page_004.jpg
d55280f932124c3c00777202a0efb151
1ecbb11fa81171a648a11c13c4c8c37fe8bc5621
6270 F20101204_AAAWDM dzotsi_k_Page_043thm.jpg
b94e40f7f1da8623f39c94cadd584fe9
7d8ce5bf69a8a710c3e32b3ed5e532ac1c19c543
F20101204_AAAVYH dzotsi_k_Page_128.tif
f40becb24ab2603cf7468727b4c8743a
452b48c63be3cefec306e0cf459df0c000c77162
2216 F20101204_AAAWCY dzotsi_k_Page_038.txt
61478da316eff896cc5d1e937a99fed8
1ad231648babe44ea703d4667fb34e5f32184a1c
F20101204_AAAVXS dzotsi_k_Page_024.tif
6f9cc30e747eefcae03b2aa02884a84c
7235334130e41a12bb5eee918be4f7dd244f7d3c
F20101204_AAAWEB dzotsi_k_Page_033.txt
6526d724a3e126147b4b19a9098484cf
9de3c9e3d3484f81836d29102ee224b13b29e535
22700 F20101204_AAAWDN dzotsi_k_Page_092.QC.jpg
d0455f5ebb685a10837596fc32660ba2
9679dff26ae41798b5aa025214bdfcdac6ac3eee
46066 F20101204_AAAVYI dzotsi_k_Page_054.pro
2b23c282d437d3799ef12907ba0bf12e
972c557dcbeb5f248e9845215707690e9084af39
F20101204_AAAWCZ dzotsi_k_Page_022.tif
88993eb05a7b35b8362617d5acf8850d
1dee4a0a773f913db6fb84de0ecfa41b8e43d43d
24834 F20101204_AAAVXT dzotsi_k_Page_122.QC.jpg
ed4c2863121d9ce7561e1e241cb96d11
a8c7b7bea2b57a6cbfd142cf3bb946deb80a638c
1926 F20101204_AAAWEC dzotsi_k_Page_140.txt
2e0fe46692f1403e3c97783fe29e5d33
fad6095dc4aa235eae712034d41e1166920536d8
101366 F20101204_AAAWDO dzotsi_k_Page_079.jp2
aa5ddb1b852c90edcd05a24393825e5e
fb91f936935da47c662359c10d3df027cdcd103a
2630 F20101204_AAAVXU dzotsi_k_Page_096.txt
10176f766fb6deb5104bac06399f6b44
733b1c0fdbd71d1291ce9d86b2ef490195c5223c
2521 F20101204_AAAWED dzotsi_k_Page_052.txt
20c05970b22cf17b01cf22ee01545943
8ae4810f3d8eefcad392b7e1b34d644a82163bcc
6915 F20101204_AAAWDP dzotsi_k_Page_122thm.jpg
6c325af9be5a63d86b1805b6a93dc898
5439e41b0f7359eb86a793e6f7328cda50e74182
F20101204_AAAVYJ dzotsi_k_Page_111.tif
94409cdad7563f3080f54f8324512ad9
55ae741d113f55423b4dc2c738401f2bb5d2fc36
45547 F20101204_AAAVXV dzotsi_k_Page_042.pro
a97f546ae23f86e591e9e93c592f6417
f7b3757a8946ee34371969ad6765fd000240de2a
2189 F20101204_AAAWEE dzotsi_k_Page_084.txt
0a9c90c8bd4cdfaa33e96a232cce4cdd
6e1b0dee04e8abaa391789c8ba8b4d1c1745852f
F20101204_AAAWDQ dzotsi_k_Page_030.tif
8c8da538ca23a4ad764c02d9f2cf40a3
6b639185c7a77947bca2b5094f594280fc8ff1bd
13284 F20101204_AAAVYK dzotsi_k_Page_050.pro
5af004264868487dcbe8f970ffe517fb
cd7db44d575fa345efde069fe81e917f13006e88
3343 F20101204_AAAVXW dzotsi_k_Page_107thm.jpg
9475be49f0dffddfc67c561f8d09a12d
046bf11450ef4b88b4077eb41ddf758bb85d0cea
93847 F20101204_AAAWEF dzotsi_k_Page_019.jpg
f85a438854dd55ba7117096a63a2b79b
1d2761f15125567d9ae10ea99e16f1a1779facfc
25056 F20101204_AAAWDR dzotsi_k_Page_123.QC.jpg
becb4794cf3362bd5d880b196953ab84
f8368744dded91f5c4e444bba6446f86901c0b7a
F20101204_AAAVYL dzotsi_k_Page_123.tif
d5b42d44f58f842ec43d022fd31a0f5c
5c6a08a584c77cbd3b8af9ccccbf8ef93f068e34
65599 F20101204_AAAVXX dzotsi_k_Page_156.pro
6808396c4044d498d965c5e4ca733d42
e58f86cadd466317e2092462104c93d1db674490
1917 F20101204_AAAWEG dzotsi_k_Page_148.txt
061ba30428325cb83eb267c7aa6cd276
c8efa554598824a38bc6091f06bb2362baa72ec6
1860 F20101204_AAAVZA dzotsi_k_Page_043.txt
06911efa11271e2a08e4e1b23069522d
fe9f0438ab4269b630db13ec0243011cdbabdb04
41504 F20101204_AAAWDS dzotsi_k_Page_026.jp2
6a3247a3ddd5c64840c52db59623752c
ec7f9d829f12727a34324fa31379e3f2b4577b52
F20101204_AAAVXY dzotsi_k_Page_114.tif
3dd59b1c9417bdd31b9faf0a76fa831e
cb63e19759e87cef71b494104da6d8442527ed9c
F20101204_AAAWEH dzotsi_k_Page_109.tif
4a48c48a0b5b62e5d5a571973b2a08ae
79ec8d6efceedb72429159703da439ab2724fb5e
24475 F20101204_AAAVZB dzotsi_k_Page_040.QC.jpg
9453b40709b099154b81abec651b1d68
85ed72d5676ecb2b9fcf21fcefaf7e9f97d52aea
783 F20101204_AAAWDT dzotsi_k_Page_134.txt
eb34b4b9f99f6356b4a88f948380b0af
b0916c9592c07ddda29baf871222e96a41adafb2
1980 F20101204_AAAVYM dzotsi_k_Page_115.txt
0e587ed9eaf66b1568d63a56ff31f78c
293207227024dec05ab7f11a9d68d6e75d936cd8
73791 F20101204_AAAVXZ dzotsi_k_Page_080.jpg
16d113494404a69ae4583ceaa13c3902
d8ec6cdf5b580865abfd50a679ada1578686cce5
5237 F20101204_AAAWEI dzotsi_k_Page_101thm.jpg
e2f0bd42970c887f9b2fe05223e538a0
aa6aef7b5d823095647f240353c73dfcba650410
6355 F20101204_AAAVZC dzotsi_k_Page_147thm.jpg
ea3cdc6b8e7b5817f468152a828aa51b
bf7e7d004077ee8796a2793b69f76f836d4b4402
26164 F20101204_AAAWDU dzotsi_k_Page_001.jp2
8899b414be5799ea7589a175412c5ccb
2d854d0ac430071f4334b6f7f089dbaf1d9e096e
2913 F20101204_AAAVYN dzotsi_k_Page_013.txt
64277e34171e8b64c42a581b3e09d901
37d08089e1f918ab42e30c9870b059e35de556eb
633493 F20101204_AAAWEJ dzotsi_k_Page_133.jp2
47ddc0713349971ca67b664c65cc35fc
c85b6b56db0c1aa99c43537e6d1ce4b1daa76962
20080 F20101204_AAAVZD dzotsi_k_Page_129.QC.jpg
c469adcd85cda7bb5160fdd09468c3b2
052f82157de867eef9b040b2b372775dc510aa6d
1552 F20101204_AAAWDV dzotsi_k_Page_099.txt
3bc10791e0b14da26dd67caa262f8edd
1620730cd07cd135d6ff5b8afb8c909f8e2f437d
20371 F20101204_AAAVYO dzotsi_k_Page_107.pro
61fcec914038eb7394940b9444d1ecde
4240d6245f8d550838ad10f7ebf861822aa331e6
42382 F20101204_AAAWEK dzotsi_k_Page_075.pro
5e3149d2babbc05dcacc001371dbac13
4ea51ef424fda2f02b7dcc4e62ce51ca04988bf4
7096 F20101204_AAAVZE dzotsi_k_Page_130thm.jpg
9d795cb0e4168b51f843244adae7037c
3124346e873b7bb5f234734195cbd75cdef1f085
F20101204_AAAVYP dzotsi_k_Page_058.tif
52a2eb8b81e449ea0e505a450ed031aa
4528df2fdc1bf3b86e8a4a0d1517f765c5d0279c
F20101204_AAAWEL dzotsi_k_Page_052.tif
a82f215ab86510eab2bf4195b8b610b9
a9d46a1914d48c25c5fd9ee8f9feb68757d5a9c2
2639 F20101204_AAAVZF dzotsi_k_Page_102.txt
49f83fa6f3fea9501a780ffb19a245b2
cefedbe9b6a1a1ae153904a695a3393c57eafdb8
1984 F20101204_AAAWDW dzotsi_k_Page_112.txt
389218ee63104dc5a99147d6446c9582
6c43979184ed029485bd0dbf247c194aaa756896
27061 F20101204_AAAVYQ dzotsi_k_Page_027.jpg
691138c2ad155b509ae1b5e95ca0ad03
ea8cd66b5d9793fa0c7004aa89d02e5ff231436d
121024 F20101204_AAAWFA dzotsi_k_Page_086.jp2
8b55e44fecf06d23daa0315682a105fb
4ac6dd1e1b25103b78e1108a4df0d6a46dcb12e6
110022 F20101204_AAAWEM dzotsi_k_Page_007.pro
14f09c9f48af4f35510d286a1d30e9f0
5dd912372e83abb1de4dcec5513e459541ebfabd
51630 F20101204_AAAVZG dzotsi_k_Page_129.pro
af11cee9a10f81d50c1ce14711697908
cf438b34fb8f74e7659b8123fed7c4c6a1af0e31
52814 F20101204_AAAWDX dzotsi_k_Page_066.pro
5d65dec808029ef920f40f0c6ca74e6e
fbfcd60cb33f807f7d0f8fcf1d8ca369ea8aee15
106216 F20101204_AAAVYR dzotsi_k_Page_088.jp2
7e15408f43aa613833082fb63f2fabf2
9b85ee52fe1519cdd16cab6b24bfff6888720434
105639 F20101204_AAAWFB dzotsi_k_Page_033.jp2
673c3122bbbbe8d485ed4bdd3dd2d254
deb504a7d39c651834491fb4c431b9c86eb2eeea
113234 F20101204_AAAWEN dzotsi_k_Page_114.jp2
30f87528f7307c9eb823b59dd234de37
063afabbdc5e07556f2fc08dd5d0bea4ed7a99ea
57758 F20101204_AAAVZH dzotsi_k_Page_128.pro
9101be4f702e1691e917633fb0704465
61cea39d9c7cc449218fd93d8c3eb9dcfc20feb0
22071 F20101204_AAAWDY dzotsi_k_Page_070.QC.jpg
c4f5c1d94a7b29c8ef3f54bf41965002
3a2c409bfc5c1cf485a04df087bbd0f461db3eb3
29142 F20101204_AAAVYS dzotsi_k_Page_131.jp2
115c72a7623aeea610a27cf9b4cbe8e4
c6b35a475ecf5a8e7172a10ec87bd285feb62ac9
2674 F20101204_AAAWFC dzotsi_k_Page_095thm.jpg
db45a1028c6726862701ade4749c2764
85ff2dd21883bdf6fa428d22b44b6656973f3fab
13461 F20101204_AAAWEO dzotsi_k_Page_160.jp2
e3e4693431766a4bd2a01239cdeec404
4cc2f3869aea1a06aec3878c867bcf4ba0dbe67a
19634 F20101204_AAAVZI dzotsi_k_Page_152.QC.jpg
6d716eda16d018f9799d18cd07793ee1
83f738f128981f2d2bfaebd347011aaaf4d0ca2c
76564 F20101204_AAAWDZ dzotsi_k_Page_040.jpg
275f5a65bfed82b6b2779c527b9071e1
12a81018e8c122733c34677695a83e142761a87e
6103 F20101204_AAAVYT dzotsi_k_Page_002.jp2
6a38a9cb954c33506c4eaf6637925fc2
b800dbc0216c2748ef6d7c79f25130a808092c94
87029 F20101204_AAAWFD dzotsi_k_Page_054.jp2
d1f58beada934fd2188f0a00885cfbf7
9c08c0bbbb36d4af6c580efadd9e2776b0384581
6169 F20101204_AAAWEP dzotsi_k_Page_068thm.jpg
709e2f1bf85fea96f619a6434f85d087
1a0273a45ac1e647fc421f05f4388c0c22d08a37
7028 F20101204_AAAVZJ dzotsi_k_Page_090thm.jpg
b10051ee9b9e468b4119f805d6b7a943
1adc799d1c9c31d0f51262f62ec9068757d35426
6811 F20101204_AAAVYU dzotsi_k_Page_138thm.jpg
a6e66316fa8594abb3e29e71f5574c1e
0ce3a9fc3827d976d0d951113e20d78ed2130cfb
14594 F20101204_AAAWFE dzotsi_k_Page_061.QC.jpg
edc4405702709abf4c1518aa44add64a
ba12f745333118a532db92c7ce59535ce198087a
46892 F20101204_AAAWEQ dzotsi_k_Page_126.pro
38d3445ac20d18894b1b460f28379550
155fc7abe0033c20f9299a549e401eb97d451f7a
40970 F20101204_AAAVZK dzotsi_k_Page_109.jpg
04af375c6bfbc3b366ba4b0807818902
b50dc69e366cf32dca2cd759424214e7daaadeb9
6549 F20101204_AAAVYV dzotsi_k_Page_006thm.jpg
e803aa14977ca20ed67a3662cc1da438
c2bd7682848e5cf07f431c5ec826abbcba5af7a8
25828 F20101204_AAAWFF dzotsi_k_Page_090.QC.jpg
efee3ad2e94f0f2b0b22a71ecb8e4aca
2b74fe2c65587e2bb8925db663f549a281feabbe
57555 F20101204_AAAWER dzotsi_k_Page_124.pro
c50aa81c33b5b903a9077081f001a3a3
711949d391d114d9c66c60f01eb53ebf91d7350f
F20101204_AAAVZL dzotsi_k_Page_154.tif
349754a29b978e7cf711280a6e61861d
2688e9ad08c2f787a6ce637bc0a5ee68988a73f1
11764 F20101204_AAAVYW dzotsi_k_Page_131.pro
98c182118ff1ee2886c50c636f3f1dfb
aa65feb269a47f8c7b4826ed7807cbdb9a4d3aba
F20101204_AAAWES dzotsi_k_Page_049.tif
f996287b2a440bd0c12217e95a7031f5
d1bfba0ca681f802b5c8e09387d8c12558181c40
F20101204_AAAVZM dzotsi_k_Page_033.tif
6e054f5bf1916abfcb8889db5559ea1d
96a2a97b880877dc8bce96365c87905400f62c2e
4076 F20101204_AAAVYX dzotsi_k_Page_005.txt
3baa82ce3eda4bfd3f3910df0afc5a27
a55a8105d9c532b3bea7158a894ec75422c4012d
102743 F20101204_AAAWFG dzotsi_k_Page_046.jpg
0e3a7050ef56c0cb5bddae36abc849a8
47a518c2717ded09c4a5e56362492fa429b32613
5327 F20101204_AAAWET dzotsi_k_Page_129thm.jpg
8fc23c639f5570509257cbb3ff407778
2c32c7a134b8cf51d1362f63d423952c1bf4e279
108064 F20101204_AAAVYY dzotsi_k_Page_125.jp2
5c8ed63f1c72bc7cf35b4eb964e72b1b
a4b328f3d873baf9eec8cbd77c60e1c6c4448d68
7130 F20101204_AAAWFH dzotsi_k_Page_024thm.jpg
88dd65dc7638b579d9ea3a08a315c32a
ff81e107f62fe77643a1b17f920542dc2595c108
3202 F20101204_AAAWEU dzotsi_k_Page_011.txt
042ba8155b5dc7fab0dd4823b243184c
cd770ba63ecf5bc6d1b15597abeb8537b38c9458
8347 F20101204_AAAVZN dzotsi_k_Page_001.pro
b824443bda89533aafa65555587a64d7
d2136ef4788f0ebf5c19f9b95d8e9620949a67f1
F20101204_AAAVYZ dzotsi_k_Page_100.tif
78ef3b961cb7887a15a4c38c5d069864
21e77fd6c13d0db9b92625a76b47d34fd82503e0
21744 F20101204_AAAWFI dzotsi_k_Page_016.QC.jpg
64c4d2ae15c477aba76ff2b260c3211d
de15ee3c87b987e478384d379c6153027e2b4d4e
698004 F20101204_AAAWEV dzotsi_k_Page_063.jp2
ac4fca1af9656bdcf1fc018a3b8c5263
208549abe4b6eaf5d7923f3f343d6f3f124c4cb9
100301 F20101204_AAAVZO dzotsi_k_Page_042.jp2
190504af419761e4be98d41938111b8e
e94002cbc25ce9a7e1e47a4bd8ed51c85c435114
3838 F20101204_AAAWFJ dzotsi_k_Page_103thm.jpg
3e29682d4c514535a6f588db0adb7cea
c20b32392558844e6315563f873b0035dceda586
F20101204_AAAWEW dzotsi_k_Page_066.tif
c52481c5c90c1c5341c0bd4219915078
23295fa37880c7d678236f271d8cdb0a7350dc76
15882 F20101204_AAAVZP dzotsi_k_Page_061.pro
b7150b38cb3b5efe261f7d8afe8a222f
70cb010deddcd39a1ff60eeefe935d4a116a9aef
1404 F20101204_AAAWFK dzotsi_k_Page_106.txt
8f1028b668f40d71c56b544f2c242a24
33ab72e8fe5b115908466e6419425f187cbf08df
20149 F20101204_AAAVZQ dzotsi_k_Page_036.QC.jpg
e8351ebc328a25fe26342ddf3bb665a3
55db4056f65758196f441d5b4b4738cf4e9b6f25
71780 F20101204_AAAWFL dzotsi_k_Page_070.jpg
8ccaf0f8c9e240461330de1bbf32a40a
94cbdfb29f5343c8d5c32ef3aa1f90376e95f202
108900 F20101204_AAAWEX dzotsi_k_Page_028.jp2
2ff92313b8e9e31cf5cfd973990481b4
f2e1fd34457d9688d5997216f49ab77297fb9621
59660 F20101204_AAAVZR dzotsi_k_Page_142.pro
ee3c6bf4b30bf9ecca70746f83b3883d
ee54fb8ef346f823398462c7e38e94f3af859009
103332 F20101204_AAAWGA dzotsi_k_Page_129.jp2
b0121a975fa18e595861defa3458d11c
09ac271f544d0abdec0e5baba29ff02ff284bbc2
70232 F20101204_AAAWFM dzotsi_k_Page_111.jpg
4791d26c2ee3c68e6f01d693a16b8bb4
bf7e8a7ee618415c3d5cbcd83da477fe7e602ca3
5993 F20101204_AAAWEY dzotsi_k_Page_077thm.jpg
1c736437052e26d077bc739242395d43
cac6418a10cc35d2cfdb173429477b19e40ada85
F20101204_AAAVZS dzotsi_k_Page_155.tif
f0c7cc7d59f5d699c33a76a653a095ca
4db840c333b5e9edcdf0bd3ee0cd92a96e63399f
55972 F20101204_AAAWGB dzotsi_k_Page_084.pro
ee7cf527572162819c283aa7b4c8c976
352c2eab781929894a83cccc8d97ad0751f3495f
1051986 F20101204_AAAWFN dzotsi_k_Page_013.jp2
37aa38bef817e78b47d5193ec6d58164
03b776e04ad19c283b42174bcafe41c84e0877e8
F20101204_AAAWEZ dzotsi_k_Page_153.tif
fd489cfab0df257b8dc07ba40630fa00
0335dbd27026583c9c1d4b13d8387ae71843e82c
551158 F20101204_AAAVZT dzotsi_k_Page_109.jp2
16bbd715c2ba06036a7a27067cef2181
e1ee11ac3c307315b002cb77df229c226e10ce02
14990 F20101204_AAAWGC dzotsi_k_Page_025.QC.jpg
e210f4521c1abc6bec3bd56455f0499f
361c458b9e85af8a690f6e5cca30f73ba814ff0d
F20101204_AAAWFO dzotsi_k_Page_158.tif
82f7d2ef135572b097a221c82b6b25f1
0587ae29fe962ea40928c4d74ab23fce6ce260b3
35920 F20101204_AAAVZU dzotsi_k_Page_105.jpg
11c0d6015758e5d9b1fe8d6031d19d7c
664a8d55fc4da36bb5fd78b57446e25dece77741
F20101204_AAAWGD dzotsi_k_Page_149.tif
836db46da03ac56472e0b074565a39a2
4f97eb9bebacb5a444fb6a8cc9a6fb543253b045
62461 F20101204_AAAWFP dzotsi_k_Page_119.jpg
e5dbca9b9fe96e0312862d832b0c37ab
d6737942f257a03810c0d979f0fa23017fa1c190
52668 F20101204_AAAVZV dzotsi_k_Page_150.jpg
57349c94b83805935e269acc953f13d4
027326cd896b2eb66fc0e5f98c9b9940f15f6e49
36771 F20101204_AAAWGE dzotsi_k_Page_103.jpg
8578b1f5768f87e803133b9fd53a22ee
7c6ff75bb749ed20e817c9c3312071bcf2e6dd00
6640 F20101204_AAAWFQ dzotsi_k_Page_146thm.jpg
84ba16029174c14e78a2b13b9d4ea024
ec770e550b54e81472205b59c68d63635a83942a
48020 F20101204_AAAVZW dzotsi_k_Page_151.pro
3e8e50c4f18a2e733a60373dc09ff594
d4e71d4bfc82ac61ca266b9ee0059c12dc6fc619
73772 F20101204_AAAWGF dzotsi_k_Page_048.jpg
18404a016deba9f3130f925a85c1918f
f8d243b78360941110d2e1c83f7489472aa0ed91
70125 F20101204_AAAWFR dzotsi_k_Page_069.jpg
79fef2b4a0014e728ee8454494ea3ef6
c8230a376f8f42d81c89332b9eb9c06771ec33e2
86455 F20101204_AAAVZX dzotsi_k_Page_141.jp2
4083f27c69ab4a8fb65183f707d9b609
1ee25ad96a3aca821f1f3460e63ef7a15a2ce99b
75265 F20101204_AAAWGG dzotsi_k_Page_128.jpg
baccacf70d33cb8a625c9ba6bdd56fdd
b62197b9adbf0e2c761556f5316b12fde90ef928
4197 F20101204_AAAWFS dzotsi_k_Page_109thm.jpg
7bf4a53d536f97edc4f1d8b86aaa3e67
1d369da8366224843a5d37da45cc6cedec1cadd7
6973 F20101204_AAAVZY dzotsi_k_Page_045thm.jpg
7667c51c1724645ed1ffdfb8f9a493e2
ae3d2be0dbbc054b9f4fd2f607d059ec41a2b49c
115659 F20101204_AAAWGH dzotsi_k_Page_040.jp2
b817f438f3f13b6f2f6a79005d65e3f3
eeb9edb1e5548bec9bc0b7c3b210728c59b5087b
18628 F20101204_AAAWFT dzotsi_k_Page_008.pro
7be2d4f1f5fcdba5c48a9b50bd6bb73a
f2c6b74eb1d2c15ce95729e56334f7320186097f
49718 F20101204_AAAVZZ dzotsi_k_Page_144.jpg
2c1193287560ea17d3add1f2a778a48f
b81a4147dd1be53879bd532b3059ecb5164cceec
75239 F20101204_AAAWGI dzotsi_k_Page_129.jpg
84d339997f2ead9730e97f2167bc4b58
0f8d973fb3037e7544f33fac8effba36fbef91ae
F20101204_AAAWFU dzotsi_k_Page_040.tif
f795ea287530772f41411504a53d8be5
67d0a2c65ac9eb9dcbaee182a1655625c7f38cd1
45256 F20101204_AAAWGJ dzotsi_k_Page_141.pro
8855e08a05580b050f7fd9cb826a5459
eb726e93b5fd6864b940d242db28b633ea17c8f2
F20101204_AAAWFV dzotsi_k_Page_019.tif
283d3a56048a27739158a091b598d07b
0c552f592e28f69d2ff3241e89c8484694fafffb
25149 F20101204_AAAWGK dzotsi_k_Page_096.QC.jpg
640f46662d912b4b2aa6bb06f0b3b992
44b709763b0e1c874d8393af1d001381fe7096b6
90659 F20101204_AAAWFW dzotsi_k_Page_100.jp2
545b93c2a47fe503b502165c08672f83
048b4c5011ae46d7b2f39f9e4d639f95661bc9b9
6436 F20101204_AAAWGL dzotsi_k_Page_044thm.jpg
71042fcd11821e29b043c06e75c609f4
34bd0e514a738674f1db88018f79636b4294de8d
54079 F20101204_AAAWFX dzotsi_k_Page_055.jpg
b71f26f22e89139f0fa507dec23b8740
fb5ab7f280e5891bed921ccbe0a4ac49f87b2015
11144 F20101204_AAAWHA dzotsi_k_Page_105.QC.jpg
5455cbb3d8bd12a12b84ce53ffe652f7
e7e9233f609f938238c35ea141a8781109f56249
5066 F20101204_AAAWGM dzotsi_k_Page_133thm.jpg
f4daba80c12a301bd7f9bdc203b8cc53
4e74573193eda2ecd1c5e907b568e0e909725518
70105 F20101204_AAAWHB dzotsi_k_Page_121.jpg
43d6c067f8f9322df5687017edcc459f
9831b4e20dc0c1bc4d13c6880f408b4aff0e3e2c
142002 F20101204_AAAWGN dzotsi_k_Page_082.jp2
3e18500174da82d5c865a3cb7461457d
69318b151f0fa0155ff225eace1ab1e8dc57d325
74081 F20101204_AAAWFY dzotsi_k_Page_091.jpg
81a44bcf8592292d7f065c3017b6f295
d401954e1f34696243e2d23f74045618ffd1d49e
F20101204_AAAWHC dzotsi_k_Page_161.tif
5035c14839e55935928b57b5fff1a0ad
296cb5f482f1a3b1c7b1fa253583be83fc201178
6674 F20101204_AAAWGO dzotsi_k_Page_089thm.jpg
1d3b38a47e1adccbac494785f913fac0
d0c84e85d6d167357063e402185e74709100c377
51151 F20101204_AAAWFZ dzotsi_k_Page_039.pro
5a8db539f5cf96f8cd9ce35f59e03483
c99abe692794a23eaf109e01a8922a36ca92547a
2125 F20101204_AAAWHD dzotsi_k_Page_028.txt
f82caa35ca7b1bbca7b121e2db227f28
bef174a03a4ab4e23a0fbbd440db3d6922c8093e
54971 F20101204_AAAWGP dzotsi_k_Page_045.pro
356115532bb90b82d16daab847dc7176
c1ebee01b2c988f54f5f9fea9ba1e4a10365612c
6969 F20101204_AAAWHE dzotsi_k_Page_071thm.jpg
151cd2ad797945a34c30fb3c9e5d05ef
a3612d9dd7e59182c0ab94094a08c9e9c8bbfe2e
1955 F20101204_AAAWGQ dzotsi_k_Page_078.txt
14504f7d5bb3cce56aa393a69d791581
e1938bcb6f9303f0537aad308279915343ecf45c
F20101204_AAAWHF dzotsi_k_Page_010.tif
566163d53ab1a7012e661d7a3638429b
6c1082ac111cfddc0d4ab649558715a7c7a96dfb
117866 F20101204_AAAWGR dzotsi_k_Page_090.jp2
e6651e54e6fda53f179701df1e345453
756b779a214518fd306ddc2bb33f581c29a9edaa
3646 F20101204_AAAWHG dzotsi_k_Page_097thm.jpg
d8bf5908f961eeba67a0daf178b79540
6b9c2da559a84856f7604da1078b3695be212e6c
1877 F20101204_AAAWGS dzotsi_k_Page_074.txt
2266f6b6804a98b67721f15206432b20
ba3fb27eff423b8d00480cebbae8d6052284a6aa
27680 F20101204_AAAWHH dzotsi_k_Page_130.QC.jpg
67e2419dc74c5b083f3383a161b84013
efc6faf4a3e9aacdae8b698a9ce3de8ded09ce83
6404 F20101204_AAAWGT dzotsi_k_Page_092thm.jpg
146cf542761b00f918e6eb5099caacae
6284302ff10fab52d45ca522a6b6df13d4a5c378
116257 F20101204_AAAWHI dzotsi_k_Page_047.jp2
b0d71bafaa71fd0d2eff1f0607c98206
2ecacb2b639ab58a1b0fa5f5e78b09f4f2141942
6051 F20101204_AAAWGU dzotsi_k_Page_053thm.jpg
5798ec52303153bc03e4f18345b9e18e
cc016461b1606bf7d531292471ef8ca2494650d9
6796 F20101204_AAAWHJ dzotsi_k_Page_112thm.jpg
20d6d1ced705dacdcb236d61710ffc77
b1cf8d580f8ba2e8bbc696d361a52b91dafea088
23331 F20101204_AAAWGV dzotsi_k_Page_111.QC.jpg
ef2d275a759b6113700638af248030b3
9e21e7fad7d8b5a81f566750e561515c6262446f
51956 F20101204_AAAWHK dzotsi_k_Page_114.pro
bfcbe2adbd05a7cd03b14d818b5e8ce4
d2fbf12600050731960a2ba974c270456bf57ced
71062 F20101204_AAAWGW dzotsi_k_Page_071.pro
c2955d9dfbd83835514eae074efa3b96
836a6bc9f89849ef0a67339d1f330a7aa3199938
74156 F20101204_AAAWHL dzotsi_k_Page_064.jpg
3c94331311b99bc320405d876b8400ed
094ad17170ded1e4c79037a77bf40c5c74a43047
5681 F20101204_AAAWGX dzotsi_k_Page_152thm.jpg
3d8b215bc0bc7338db0682dacf8f874e
504790d3ebf59acb602539a0943947d9d98988ac
2039 F20101204_AAAWHM dzotsi_k_Page_072.txt
cc8c54d40da68ab78dfecb3c82ff17f7
d71ec99e837830f468b5717a0e41bb6793795219
75812 F20101204_AAAWGY dzotsi_k_Page_146.jpg
ae44dff5802ff21ac52021afc0afc2f5
f62afc4d26550413bb10ef4a661d34ce0a34cf59
6634 F20101204_AAAWIA dzotsi_k_Page_161thm.jpg
8068290a74f9183058f83e33e34d9b68
115dd05af4b57ab453c65123e2a3fe2644909379
96432 F20101204_AAAWHN dzotsi_k_Page_087.jp2
d1dd220850d692357bb690c1622f609d
76a2ba6da52f4cd972d908c3f2f70b2f403edd92
1915 F20101204_AAAWIB dzotsi_k_Page_126.txt
2fbd2a11dc1876c933b7e799bbf0bdfa
374d0d371f1382bafdcbf54b719931422a61f285
F20101204_AAAWHO dzotsi_k_Page_051.tif
7add35998d59472fdea9d0229d9ccf03
e74fdf1f27b063ba3281459118b2a7bd6f107a80
F20101204_AAAWGZ dzotsi_k_Page_107.tif
5b59a595e403f1d6efca9264677af26e
b510e2c31b01c6c1e513d110f406c808a6f2e5c7
2002 F20101204_AAAWIC dzotsi_k_Page_044.txt
c9ad65648860abf70031db4641108c95
34d152ea5605202884b2500c88c57e980ed1099e
4971 F20101204_AAAWHP dzotsi_k_Page_006.txt
6c3acaf7f86d344fb9b5f4340ec8a9f6
0086cb5f40507d82f047b9937cbda609cfbb8c61
F20101204_AAAWID dzotsi_k_Page_142.tif
812e39d448340c064f31164b27fe77b6
527c6b6b21cc7c9771373af0c2fddc3bd5722302
141951 F20101204_AAAWHQ dzotsi_k_Page_019.jp2
a814d456e85581bdb836743a4c5bd467
7d53887fab8d91771d244bd8f076395a68e54341
102985 F20101204_AAAWIE dzotsi_k_Page_069.jp2
14882c10de0e4fa7a1dc3a0faa9641ec
c25a6e881ae559417a1b32cdcb8627dc2f2e2bf2
10932 F20101204_AAAWHR dzotsi_k_Page_107.QC.jpg
57d0606fd3b17fbc5b71d5039f2c9db6
17501be63537efc294bf3377621534d6b7b9821b
66852 F20101204_AAAWIF dzotsi_k_Page_067.jpg
677eb1cb0ee9166282950a47350718f8
d6bfac6ad9d5704606d4a3007043f434fd9952b3
40614 F20101204_AAAWHS dzotsi_k_Page_134.jpg
8dc379f8414573bbb90a7a3c549558a1
467dbca20bda26d7a3bab6a2a1eb88e786445006
54187 F20101204_AAAWIG dzotsi_k_Page_023.pro
e1925b3812c7b80a489b9f90301cb37f
7f75df85b2a54593d7ee930139623c9371e5c376
69189 F20101204_AAAWHT dzotsi_k_Page_158.pro
0dd64deb8eb464e3fdf4ee2925f4e8ef
98103f06b55d3c3a9e61a0fb83e61aae93eb273b
17290 F20101204_AAAWIH dzotsi_k_Page_109.pro
329831e16c4b463751e39cd3d55cfd9a
1405a6887e6006501ddc763f4c91c628bfad7357
118003 F20101204_AAAWHU dzotsi_k_Page_045.jp2
581092cbc0081b793d8d2aba33525c03
77866b284608734f742331dd725986ac4cc51e5e
26001 F20101204_AAAWII dzotsi_k_Page_024.QC.jpg
1833ddb1209bcaa30613afe8cb06f970
da950fb9d5290ed7f740f0379644dfce9fadd47e
57262 F20101204_AAAWHV dzotsi_k_Page_101.jpg
7b3f2cc2cf8eb5423a1de4cf83625386
939984cc5653ce257c3d59aba2a730224fc14b61
59374 F20101204_AAAWIJ dzotsi_k_Page_054.jpg
95e407286cfa4a3b6d7f28efa142da18
24053e0e95b290f308159cbdd7a01d7d5059ffc1
94722 F20101204_AAAWHW dzotsi_k_Page_056.jp2
2353df13127a5597287b9cf4bb84cd83
09ab716cc44e5a6fd3beb373d823155c7f958b3b
882 F20101204_AAAWIK dzotsi_k_Page_061.txt
f25ed85641cf4e3482158a1a99c11138
adad28c157f91a3d1419a4fec8fdf6666fe13af0
5635 F20101204_AAAWHX dzotsi_k_Page_075thm.jpg
ffddefdca267a887d48132250ef4672a
9fe05eac2250374cd1946e767db3150889051b60
6406 F20101204_AAAWIL dzotsi_k_Page_113thm.jpg
945014c3a64a68862573646ce2e808a3
b7a41c648f4a1b2a84f768a220e55f610aef4d1f
3110 F20101204_AAAWHY dzotsi_k_Page_145.pro
c7087d80478ab2e6497aee7b5fefd6d6
03e9868e013aac6ff4987c88bd25cc7a8e45f116
43488 F20101204_AAAWJA dzotsi_k_Page_072.pro
63ed3ba439dfc32b185ef1e49c60538c
aa9d1c261fef783c8530f92786e0a337af428625
340403 F20101204_AAAWIM dzotsi_k_Page_108.jp2
c93200b629d85e4f461b9b6191d2fdb7
d79e01ebebb848c9d135ad6a70f8c8b1a8386b6d
2116 F20101204_AAAWHZ dzotsi_k_Page_041.txt
0fe97e5f0b666f478a95f92d34c39086
795c164cbc041e0f7dc3b2b1d2bd79872c5a45e8
26000 F20101204_AAAWJB dzotsi_k_Page_065.QC.jpg
55b1818ae90c48ccb151dff91fb41509
d14ec738bf80dd468e39ebc137f591bfcfe924a4
F20101204_AAAWIN dzotsi_k_Page_025.tif
260c010c0444c10ad1ac5eda8e149525
ef5fce76c803aa5d9ad202ee855f10baa1c39b6a
41303 F20101204_AAAWJC dzotsi_k_Page_012.pro
41bf485fa155e6ec806725679937fc78
39a2eac99eb398470617bc8b346dee5b7ad3ff25
1612 F20101204_AAAWIO dzotsi_k_Page_144.txt
1b02b19212e86ff5c7a6946e69c51571
6cff0b572d98f1b0639a9d3096d8ae6a3e42d55d
F20101204_AAAWJD dzotsi_k_Page_043.tif
1b77d0af41e4bf8cc2ab2e9612bf25ed
cd05c41c12c26a1f17ca79fd2c087e7fc0ccb214
72977 F20101204_AAAWIP dzotsi_k_Page_089.jpg
1c236c18f9b9711a82527ed4bd22093a
fe69da4e9033c35c9cf9db426545c29b9d33de72
38122 F20101204_AAAWJE dzotsi_k_Page_058.jpg
70650a94d14dc6592da5a4a6ff64045a
e4fd30d101481b2889772c00b43df53ae1d676d0
20171 F20101204_AAAWIQ dzotsi_k_Page_072.QC.jpg
53eec217508b32f738b5cddbed59691d
391f8f201af14fdf7ac9cbe0587b524a8b301158
27546 F20101204_AAAWJF dzotsi_k_Page_082.QC.jpg
424f7a1ad01a08d2edbd100b658cdf00
6adf72264664560730b58487a2727d4fd382d2d9
76126 F20101204_AAAWIR dzotsi_k_Page_122.jpg
ea67962078291fa014d1518146b75861
4ee290438f33096253b135d0b325e908d41ac4d5
7154 F20101204_AAAWJG dzotsi_k_Page_011thm.jpg
c277ef7d2d477dc98f63cfd4be2cfd22
f0a18535552ff21a43706bea9a4726177a2040b9
23635 F20101204_AAAWIS dzotsi_k_Page_080.QC.jpg
6911407a5ef0a11f0815187e324a9f17
226d8150ee542ba7e1406aec157bd5c1cea1e984
70006 F20101204_AAAWJH dzotsi_k_Page_157.pro
e86db925ed9368f23fb1ddde78d81f98
18110aed919e257f3ebc4b8214507f5e16115a6d
32909 F20101204_AAAWIT dzotsi_k_Page_143.jp2
9d1a620471f3273ea7d7c8f83faeaf86
e6b926f737bd7155636531bf9e05c02c816c0fa6
F20101204_AAAWJI dzotsi_k_Page_094.tif
337883f37952c878ea83e891001b9d5a
bff1b1e7859fe8742b6bb69a58092c418520921b
447775 F20101204_AAAWIU dzotsi_k_Page_105.jp2
dd9f8cccdda4131f5f153384f0b1e259
9496d12204dbfd069501bb80cfd3f5eee757c84c
F20101204_AAAVGH dzotsi_k_Page_096.tif
36726fbd6ba20fb02eaa3c2b8f57867a
68ea4183d68b45b9715145088c9556cf3909021e
102088 F20101204_AAAWJJ dzotsi_k_Page_126.jp2
55c194a8f52306a8191e6fb50e96d0fc
ba0d8b149324da8f1876979717cff443506d1e0c
18430 F20101204_AAAWIV dzotsi_k_Page_141.QC.jpg
5296e5add4d00d962ccdaaa437904eaf
d37fabbb2261f7eccd30a7fff62ffe5ba81eb29b
77939 F20101204_AAAVGI dzotsi_k_Page_020.jpg
a5c6a481e2453e70d09f0baba84b89d5
98417a171980d4d9e1cedd83e1cbe2d144193643
22576 F20101204_AAAWJK dzotsi_k_Page_068.QC.jpg
6ac62909f4541d55dc706b2bf9ad9c45
dde777e07120a2b5bcb96c4af07400f8cd3c8c51
78820 F20101204_AAAWIW dzotsi_k_Page_038.jpg
6d9434a8802fcdcf71bf21771f03a15e
33a2a96989f75780fb4ec32a63f6518624bd7dad
2067 F20101204_AAAVGJ dzotsi_k_Page_152.txt
7d8b3919d16746cf4c5143b697f330cf
5fc0b3b2c4b0b4a79b923417b896b45ab1b29327
43482 F20101204_AAAWJL dzotsi_k_Page_073.pro
3727d245306aac63843cb871de2f37fe
b41d4eb740650b8d16517504e44660c90be7edb1
59438 F20101204_AAAWIX dzotsi_k_Page_097.jp2
9580613ff0de25143488bd9bb8529a21
1818a0fc9a90d07562835cda23c5bfe1d361bf3f
106022 F20101204_AAAWKA dzotsi_k_Page_014.jpg
6f6bb12db51660de62487038174d3739
9d21fe0740b201521b16f2dc2ee2ac34ce49ee41
12402 F20101204_AAAVGK dzotsi_k_Page_132.pro
0ee5f8b10b10bcead17111baf61846fa
1bab27e395557f3682e09e9dd4cbfd249fda502e
F20101204_AAAWJM dzotsi_k_Page_125thm.jpg
1422b35f3ad79289fb524f2b52687ab5
7e2998b91e4244388c8c66ddb5e234944151d163
F20101204_AAAWIY dzotsi_k_Page_053.tif
d68bf83216df8fdc8a6f5a115eabc3cb
7e4772c3905b46ad877c0ac94ba0a43cc5b2334a
14548 F20101204_AAAWKB dzotsi_k_Page_015.jpg
5b035c7bf5765327c3bb41458a16e9e8
1bbae13b7f6108152a572f8b8a4dcd459c3d3d6a
71323 F20101204_AAAVGL dzotsi_k_Page_125.jpg
b7a61236ea9f34e52624ac41c16e39ae
82415cf1fbfaeeb94a9c134a098880c3f33a828f
19260 F20101204_AAAWJN dzotsi_k_Page_148.QC.jpg
d089fe48528779481a24d07c81a1e6ab
a90b5a42355f350ff008691a1e303c441dcf3bfc
107108 F20101204_AAAWIZ dzotsi_k_Page_068.jp2
aa31ec8fd6f34e6d51af2b1c069300b3
b47bafdff4c68ade56533c5e849886b6b7325659
76837 F20101204_AAAWKC dzotsi_k_Page_023.jpg
37ebe89b5cfeb26df84bb31586d0bf2b
e919570628a63fccbee272bf68972d20fb4d5de5
F20101204_AAAVGM dzotsi_k_Page_061.tif
4c17a37e9073dd850e1d2857b9d8d11d
ec66550e4354748f19c8a53291ca98830bbbeb4a
54798 F20101204_AAAWJO dzotsi_k_Page_098.pro
ea2f6ac7670da587868c0138a5c404eb
0b9c06389ef0f9da4ef0123995697805bfe3c0e6
1796 F20101204_AAAVHA dzotsi_k_Page_073.txt
34f003973ac48e79ec25bbbe755f2e9b
7befca413817caf9cefb675e4fb76364368dcae2
3146 F20101204_AAAVGN dzotsi_k_Page_046.txt
155a581e51df0c3c45cd21c6b81a40d0
5ef41cfae0f6f019822b8c349fa7b7bd011dc3de
109017 F20101204_AAAWJP dzotsi_k_Page_044.jp2
8b680fd8ae5625ee8a103a7faa8644fe
69836cf7d222a9b37d3c64009623dd2a34c600c6
46430 F20101204_AAAVHB dzotsi_k_Page_079.pro
f2b21631cd250aaa6f2416ecf829a83f
68dc2fbc7b9a2f7e823dd9904bb39e0ae7dd6633
46334 F20101204_AAAWKD dzotsi_k_Page_025.jpg
96495c1b1282a4a1211dc26fc53984d3
31347b9936658a9deee1d9a4a8413b3f589a0f23
118160 F20101204_AAAVGO dzotsi_k_Page_020.jp2
b23a84699035f18b122ddeacf121efd3
d2b0998ad10a9b8d5862f261f954f1251561d68f
55601 F20101204_AAAWJQ dzotsi_k_Page_065.pro
54f7a0b7a0b3aec9df42f5b4d87f6d63
85c2d8c9eb8984f1fa322e8bd2a8ebd5aff2542a
65775 F20101204_AAAVHC dzotsi_k_Page_056.jpg
3d15fcb1ff389c8278e0b3eac9379db5
40a38ac9d17e5092302d62c986bbd6ea6d6c290a
74065 F20101204_AAAWKE dzotsi_k_Page_030.jpg
af94e5e111db82c9073d3c83e4496fb2
9a42a7ac77035053778c7fa5cce6ffc424bd7c31
5998 F20101204_AAAVGP dzotsi_k_Page_102thm.jpg
0916017b1e501bd83e5bfba7dd85b967
4eb4a95c2fa30ca415ab8f2c131956ffc2473c65
77241 F20101204_AAAWJR dzotsi_k_Page_046.pro
49b790143d7c7229d5953f773427ee17
8f9b4601dd4336e6ef1308581c463c12bc756ede
3482 F20101204_AAAVHD dzotsi_k_Page_108thm.jpg
ebe4ef612f7bd792abb0aeeb21ce234c
92f29e7821d24b2bbc1b8d5b2abe0b1508c08054
77850 F20101204_AAAWKF dzotsi_k_Page_031.jpg
6147abbdf7bcee1a22562388f15193a6
f048eab830172c33d6044822eab1a7a9f51fe4ef
102580 F20101204_AAAVGQ dzotsi_k_Page_078.jp2
bada1d04491de7e668edbb57e048acd0
409e58db88a202d1513b1722fccb37d38bafbe40
F20101204_AAAWJS dzotsi_k_Page_047thm.jpg
e6c09507b0059c5a7d84060c1078a95d
7b1c7cb84346428c79cd1d6f18462d2dcbed1da3
74056 F20101204_AAAVHE dzotsi_k_Page_028.jpg
ffdcd28e0c735e3f8fa1db035a019526
c33f2c919f835b48939c80c30aae4dd076355a57
76354 F20101204_AAAWKG dzotsi_k_Page_032.jpg
773610012ff656509f03cf831db33785
2d0a1efa1c41d4208a0a4e02bcf61881f0c643b5
2017 F20101204_AAAVGR dzotsi_k_Page_034.txt
a41a51632ee8b03e224b029fccfe5ea4
cb1f470734fcd9e0f946ff171eb0641a1cab93cd
240368 F20101204_AAAWJT UFE0021774_00001.xml
efc7fe6a03f9b29ea76a2e93fe6b9dd6
63750ddab6d868132bd1c59867af39aa1e386cb1
52764 F20101204_AAAVHF dzotsi_k_Page_085.pro
8733406e91d396be91b9469bf916c8e8
9039c59b70c98a476126c4b1bc68ece7967dd4e5
72302 F20101204_AAAWKH dzotsi_k_Page_034.jpg
26d8b7b358c5e51dc00cf31c55a5ac34
5151026565aa357213b1dec6ed49edafed11b259
62417 F20101204_AAAVGS dzotsi_k_Page_118.jpg
88da056c88d462c6d90631a64387abba
445a3342f526e3a41be73f9a624042d920f3fca3
F20101204_AAAVHG dzotsi_k_Page_030.txt
e4eef1b81201ab3bc4d1986376e602d3
74a17181cda05f28be630946e2dd90e279a22ffc
62861 F20101204_AAAWKI dzotsi_k_Page_036.jpg
4a1de5cdef942d4fb98ec0837c6c4fee
3f4c3ca76fcb5f9642812de702bba6362b291f40
29940 F20101204_AAAVGT dzotsi_k_Page_159.QC.jpg
1d2a5d4f511f0a81e4927b52be493c73
cde0c64c067670ada20965fa7037d92543cce7e0
F20101204_AAAVHH dzotsi_k_Page_137.tif
dc86e93157a52086d6648ec83ed0213c
8c176433d8ba22ca1739145724a5d8acf82d31e0
66418 F20101204_AAAWKJ dzotsi_k_Page_042.jpg
6183a4df646410ae00db79cc1d535de8
4975ec7e4a741571329f3080d12d97ecc6ae028a
4446 F20101204_AAAVGU dzotsi_k_Page_134thm.jpg
37528ca70dc13313c1aab340799a5c20
9b006d8eca099d7facfb356777fcb317943d0d5b
10030 F20101204_AAAWJW dzotsi_k_Page_002.jpg
495d432d3607a18d25bdd6ba921a16fe
1eb8dd5b7467c0594782255bef38f4b19b4170aa
27557 F20101204_AAAVHI dzotsi_k_Page_013.QC.jpg
a8e2a5afed3b02e7f588ddde3bc85f62
37daec200440c8914337ee6effccab3e3fdce387
67950 F20101204_AAAWKK dzotsi_k_Page_043.jpg
f622ab92c6a75a87c2f808ccfddd27b0
28aca76bde55e67f5f184b8016fc40f473d2a0b1
6159 F20101204_AAAVGV dzotsi_k_Page_056thm.jpg
62619ef2dfa8339ab647134d75bb36b2
ffbbcb914f8b18e81c84fe925b011fd0ab17e76e
74618 F20101204_AAAWJX dzotsi_k_Page_003.jpg
5deddb78d828aa5d4af0b6e723157eb1
8d5b263473dd4457a1f63d8b0ebc6f4ad4db433c
119296 F20101204_AAAVHJ dzotsi_k_Page_024.jp2
ee651ed0218fd813cadc14dd9d0ef32a
7cf5ed05306692d2b4a21e88bbb5b121d7a3ec74
77539 F20101204_AAAWKL dzotsi_k_Page_045.jpg
115c0fa7ffcf6cef1f6bb3ab3926d9d7
1b075cdbccafb9a764fa9ad50cae82862f1deb28
39333 F20101204_AAAWLA dzotsi_k_Page_104.jpg
02e0c4c6dc773637d96b874d9297de8a
586f302b39d5cda24cd909fc75f7f0b9affd4c5c
14967 F20101204_AAAVGW dzotsi_k_Page_160.jpg
32e7f00e958f37039b2d27e0bc64c8f8
f3d96a6c53be9890b2a74c7d1957b6d513a0ea14
110089 F20101204_AAAWJY dzotsi_k_Page_006.jpg
1eaa3f52956b739ba2e60d9aa62c1190
e29103e70ad34955aeacbd1b71a69dffe0429be6
70663 F20101204_AAAVHK dzotsi_k_Page_033.jpg
26e0187f1897bd236499d84380a640a6
d02840af5c42c42aaac04dfd5d2692479837b24c
77298 F20101204_AAAWKM dzotsi_k_Page_049.jpg
0fce7e0cb0d1afa70e35bbfe279cd8a5
3d491699d72f2b94a1b9c806c47221575c5442b8
53531 F20101204_AAAWLB dzotsi_k_Page_106.jpg
412cd329beee95eea9a592ce116c8cb5
c21342b5b359efcb6a762cbf84cb22d7ed4e7a08
42268 F20101204_AAAWKN dzotsi_k_Page_061.jpg
e5e7d80693f94f23729e7ecf79ab2cbd
36813f3046ffcadbd3a9196ab8abfb7c3a57f7e3
135355 F20101204_AAAVGX dzotsi_k_Page_155.jp2
f7e97d5b000975db69a5455af600b515
29916c6cdf72cc8bd444c5e8843ad564aaa67c5b
60156 F20101204_AAAWJZ dzotsi_k_Page_012.jpg
5c8f0b4817caad4fb8795cec6632d20f
760c7a24fce93f4776293fe5273d2929da17e215
F20101204_AAAVHL dzotsi_k_Page_009.tif
730e8338d565f2174d955350c1252d93
eb0bef132e6eb0bf4e9304b0df6dccf44b164bde
35086 F20101204_AAAWLC dzotsi_k_Page_107.jpg
7e164f6cf63eef39d01bde16172dfb3f
d1f0ee7faefc1f2c647b6069e7e61f2bae5d3ec5
40863 F20101204_AAAWKO dzotsi_k_Page_062.jpg
241bdbeaf474b74235e3a13af7a968f2
878e3234d32306b76a75668b7eb9ab9a2f4bfa3f
54248 F20101204_AAAVGY dzotsi_k_Page_138.pro
a226e2f0a1ee1d605cef5efdc9714b9a
4292d734734e4bbb0c7ecbcb3602a46006e0d1de
50086 F20101204_AAAVIA dzotsi_k_Page_125.pro
930dfbb4921aac5fde1fea821dba8461
d22b969dfd891f10bd8861dee9df6f1ae2c09a6c
1095 F20101204_AAAVHM dzotsi_k_Page_002.pro
e76de9c7f315796903a437dce9fd5fab
543c47e8c069b4b72aee2297de133251d4e7a928
74483 F20101204_AAAWLD dzotsi_k_Page_110.jpg
3cc75b858520e89b9e0330f188d8a804
007ae222011f6906fd2508030afeef93fd8baf55
47021 F20101204_AAAWKP dzotsi_k_Page_063.jpg
152c23300c5b67604bf700198efb5935
e81a3d52a5b994d00fb143430d8f542424c704ab
1192 F20101204_AAAVGZ dzotsi_k_Page_025.txt
27eabaa25af01040b227e000465aec52
8b4cfa4f78cf896e07e26a326f80a75fb78ae226
21742 F20101204_AAAVIB dzotsi_k_Page_074.QC.jpg
30fd26c00a90354aa50d231aeae412a8
bbca3c4b35345bb72f5d16e65bd18c8a4f0b8d80
F20101204_AAAVHN dzotsi_k_Page_003.tif
d565b5a1ae541e0a0f684b1587085d79
258ab9a4e50a59436b0340270ee78b018a87fb2b
71336 F20101204_AAAWLE dzotsi_k_Page_112.jpg
0fd13a0ed58d67e56aa5937be38d3539
4a7ad2a2d02afbba2bfbe1b914cb60c0e9415408
62606 F20101204_AAAWKQ dzotsi_k_Page_072.jpg
145c88e8d83db97373e08bed19b1e5bc
3603103afe7f782c84069af850b7f4592e62b17e
6671 F20101204_AAAVIC dzotsi_k_Page_096thm.jpg
c746cf71fe192fad301988448cf5f53a
693805dd602f2b87b5157b4a0bf2712956ecb690
F20101204_AAAVHO dzotsi_k_Page_047.txt
5546dcc264c352f4922303192cc8d368
eb178d183340a32aa27a82a5a728a384eddd1203
74727 F20101204_AAAWLF dzotsi_k_Page_114.jpg
fc26c7633731153cce4951ba33e858d3
870dc02d640cf5603735e9dcc149164b96bb291b
64949 F20101204_AAAWKR dzotsi_k_Page_073.jpg
6371e2b0926e4c9ed1d404af8a40149e
72916ebb4a33034434437792ca1ab3a6840d771f
34118 F20101204_AAAVID dzotsi_k_Page_099.pro
5510d9b1b4a2353265980b136d65d2e9
dae53ada68fc8dca780d0e2d47934b26ff86e315
28155 F20101204_AAAVHP dzotsi_k_Page_010.QC.jpg
143543d179f4533c19cf9863ca8aa650
bcdce2033d78a4fda30496788e964fc09488e62f
73102 F20101204_AAAWLG dzotsi_k_Page_115.jpg
f132b5b7c0dc1945182b00ecb6efb906
5852661743efe142469ae7f57eab8bc715dfc482
66038 F20101204_AAAWKS dzotsi_k_Page_074.jpg
a179f5034c824771e6493e497f3f5538
0b1b2006133d4d3a36cc24a96a41f22a499e4a1a
3065 F20101204_AAAVIE dzotsi_k_Page_026thm.jpg
2dd26ac2548f46ecc0b4fda0501daaa0
1847e6782b62234502f3805369cc0375daaf4eda
110084 F20101204_AAAVHQ dzotsi_k_Page_039.jp2
94c6a33f47acefb985e2297f52c8ea7f
7b642ee9d9f84e80fd31779cdfe8cc363f95ae7d
69171 F20101204_AAAWLH dzotsi_k_Page_116.jpg
24d1baab537e5a96778d4d6594731cc7
e5a7d53c913e0ba4b5f7780b7ddb943ff49cbbbc
76165 F20101204_AAAWKT dzotsi_k_Page_083.jpg
9869ace7ce3137e3da5e71cf172f584e
ec890e0c4f47d93f54002f173d5681eed7e033c6
2162 F20101204_AAAVIF dzotsi_k_Page_045.txt
a94518ea0035378840d2d518eeccdbad
ff51c7458e3e1daf3e59f147c9f772610ea47207
103324 F20101204_AAAVHR dzotsi_k_Page_116.jp2
a6172c57024c2eae2157b168688957ed
15344c175a1420b22f8815128acb6c6c3ab6d4c2
28120 F20101204_AAAWLI dzotsi_k_Page_127.jpg
5f16ff79ac32c3c5105092d768b84055
df5f95408d3ad54919b6327274fa9131efd963c3
74984 F20101204_AAAWKU dzotsi_k_Page_085.jpg
971044afd0663a53935c43360429187d
0c47f3d6e632edfab034cc38d47b0e4ab66666f8
24750 F20101204_AAAVIG dzotsi_k_Page_114.QC.jpg
7ba7d3eefaeded9fc8960e0d4577a8a3
7d9e03b3ddcda7c9dd8fcc0ab70401f6e0f3e314
5056 F20101204_AAAVHS dzotsi_k_Page_054thm.jpg
c9bd38933343e7259526cd3cff08687d
b16a05a716a54e844af6c1ae42ab1bce47ca0095
95840 F20101204_AAAWLJ dzotsi_k_Page_130.jpg
0255b0fd63c6d4bb9959051173c6622f
ec6440d6eb50ebc129ae396a7f1783ef3cdc1998
70099 F20101204_AAAWKV dzotsi_k_Page_088.jpg
598a6a71b040a7cfb041026d901b3476
56503a58a6990879fdf4aab29aa9c75803896e8f
47642 F20101204_AAAVIH dzotsi_k_Page_092.pro
db2d55a646badbc789ac6082ecd6f951
ab6bf2e28b03ede1da39e8b9ccb7ee16c9332d27
6482 F20101204_AAAVHT dzotsi_k_Page_088thm.jpg
98d5346168f5464677f1310b0df970db
60546fd62dc3ba82f8aade6e536588dbef94692b
45194 F20101204_AAAWLK dzotsi_k_Page_136.jpg
938882d641838989071df82312fcb075
bcaabf40cc394e26e2488442e26915891c757865
79448 F20101204_AAAWKW dzotsi_k_Page_090.jpg
775f4a19294c133eec80cd413433666d
8d53c860b1188a58a21492cdbf12acf69c8c8d2a
1636 F20101204_AAAVII dzotsi_k_Page_160thm.jpg
ae44db9ebd6378201e78271c7a51e004
ee29a80eca663678572262a39933e8b8f74c57a0
3330 F20101204_AAAVHU dzotsi_k_Page_002.QC.jpg
32dacef5ef2b60f7fb1caeb50df4a4ae
0cd78159537121291736e409232a9ecacc000114
62191 F20101204_AAAWLL dzotsi_k_Page_137.jpg
f8e661b39504d915b5be138eca7356b4
c2d82fd7bd83be34aedde52a9674bb3d693eff4d
83241 F20101204_AAAWKX dzotsi_k_Page_096.jpg
9991e4f25406de742e5a7a4459ef7ff8
83d3e2a0d0e7bcc9f674b3ba346dc597aa4c24a9
75900 F20101204_AAAVIJ dzotsi_k_Page_041.jpg
1a3f78b0afe87be2562c418a31e584fe
e5c17264b488af89df8c3814cf85582b9028dc95
89639 F20101204_AAAVHV dzotsi_k_Page_155.jpg
59edf63dcd1125bf95b844331b60d891
1a220c24941d53037aec47831c19c2c49dcdd092
103088 F20101204_AAAWMA dzotsi_k_Page_016.jp2
d5b9456c3f73d81ea57ca3da688fe245
2d2e1a3118780cb4a7a0bab2e53504731e8fdf0d
61774 F20101204_AAAWLM dzotsi_k_Page_141.jpg
bbc99218678bc1931f72583ce76b4b87
29fc412226f283d2cdb1bcba534b05669b1fde3e
49721 F20101204_AAAWKY dzotsi_k_Page_099.jpg
dd9f0498b4b9745e6199d856f2b45a1a
4610f45398f156dbbf9f2badd3b3767bd916765e
482451 F20101204_AAAVIK dzotsi_k_Page_008.jp2
8bb979f53a3a716285934ce94d8162b2
30eceabe6a9e3977a57bcbfa32c08c0536022cd7
F20101204_AAAVHW dzotsi_k_Page_048.tif
9d97dcd83649f795e5fd40cc08bd8d61
81107ae689536347f6225a1105707f297b0a0dc4
115783 F20101204_AAAWMB dzotsi_k_Page_023.jp2
5a8247ab398b8bdcf6c459fabe19b464
7813a9f01d4b7abf986458a03ee3299cf94ff6a3
76799 F20101204_AAAWLN dzotsi_k_Page_142.jpg
72f9835c7a47a4770573c4b3504b6a92
308c40a13ce270953516e8cd95f1a24114966fd7
65189 F20101204_AAAWKZ dzotsi_k_Page_100.jpg
b98ce0c04162a7a70599eb80fe27719d
ae6e9a6ddc9e7cafd38e2fb19d5c193692e469bb
18317 F20101204_AAAVIL dzotsi_k_Page_054.QC.jpg
9312ddeb3cd2e9dc4076ed1288609a7d
eb5c0c7eca266d6b7b4adf7fb5a32e42b0c02488
15735 F20101204_AAAVHX dzotsi_k_Page_133.QC.jpg
8c40bbd77ad2bf4170093c178b9953a9
76c1cae14d54606f71483b41471df0abb380da9f
119186 F20101204_AAAWMC dzotsi_k_Page_029.jp2
c6fff570bed5f03c83c0a069552fde92
2f1cb755f266a022030246f666c04579c9c659e2
26872 F20101204_AAAWLO dzotsi_k_Page_143.jpg
73d6fb46ee337376da885f2ff79fc9cc
ccd0e82f1e915705683fd188468e3983dbac0706
58944 F20101204_AAAVJA dzotsi_k_Page_096.pro
75c5ca0628ff8d4b0286c8e4c26837b7
c61e8ef22a8bb3653a240de1880a1304d60e8e79
51731 F20101204_AAAVIM dzotsi_k_Page_028.pro
3c0ec100191e143a7ff35a867c81473c
a5b27ea29060e5f21b5da67b03b0b5ae33b0be0a
49051 F20101204_AAAVHY dzotsi_k_Page_112.pro
60e488469f574270c6d03e0d15d06080
352868683bc9275685df7f620b3fbd81e8355388
108152 F20101204_AAAWMD dzotsi_k_Page_030.jp2
e4a81286ee00020da14f6009268fee7c
eb848e1237482ec4c9d09f258c670f5d50e9da7b
74371 F20101204_AAAWLP dzotsi_k_Page_147.jpg
4cbc746cdf12c92ffe33b3108011a856
8c5033e46d8c3d5f870eb5d00d7598ce83f807ef
66540 F20101204_AAAVJB dzotsi_k_Page_025.jp2
2738d7b492d79c8108b68828bc30c283
897ebeea615473f9ff551b45af4012431f0dc2cd
25884 F20101204_AAAVIN dzotsi_k_Page_020.QC.jpg
52c7553d1b488db2d89decc66e0ffce6
33ca4e3f90cc1e3b0c65306ec281bf68f7e13eb5
2576 F20101204_AAAVHZ dzotsi_k_Page_131thm.jpg
8bae91f23ebd3fdf7513f607492de890
b98377a60faa0d6a2170a5654b4dc0337c09c5c6
117040 F20101204_AAAWME dzotsi_k_Page_031.jp2
554063c84e67777f905cef170b55388a
14a8ca1f6bd71f8051e27bfb62797e2d16dc6107
67449 F20101204_AAAWLQ dzotsi_k_Page_149.jpg
654f5e8e3fc875e578324aaddffec025
065e4ed6f4b08d0b1e8eb3333041cba5710f79c2
3574 F20101204_AAAVJC dzotsi_k_Page_104thm.jpg
63b5865fffcf440ff79eeefab32dbc96
95f28a11feaa59fc2e36f02a19efc39b45a93911
95595 F20101204_AAAVIO dzotsi_k_Page_036.jp2
602f8f105b9a1628ef8782184d2606d1
c84ba918e0beb54b4483daffd4150f0ce28eeba8
115085 F20101204_AAAWMF dzotsi_k_Page_032.jp2
afc8f78b399186da7475ce1b47c8dae7
4fa71779655b8a0770b4238e3ca9351d51fc6013
63353 F20101204_AAAWLR dzotsi_k_Page_152.jpg
69e8f0946d4aa40d9301987ebb80a378
a11817664007d09ebfb2d7f1028d1e8f67cd6115
F20101204_AAAVJD dzotsi_k_Page_047.tif
5ef208f9dc94e533cd2a7b0701ae78aa
a4da8894c27fde7ca70f755adf8b2640f25ec824
51193 F20101204_AAAVIP dzotsi_k_Page_102.pro
fbf4f26fa20fc7ffba49509363cb4ffa
ac1ed4772249c0049f73e3ead115e365486a16d3
120669 F20101204_AAAWMG dzotsi_k_Page_038.jp2
b5c589c8a98909f541e1ddf491670dfe
b6c10696945293522cccda4c17f8e059a80c8a4e
90271 F20101204_AAAWLS dzotsi_k_Page_156.jpg
9e91bc40a9ceb35f299dd9714c396de6
18244e4d551d0c9e94e69bd375c116b4ef3ffd28
75432 F20101204_AAAVJE dzotsi_k_Page_150.jp2
5cd18ebf6b4b4d568cd82d308f96b70b
20c081c612f10d13d42fb3d53d14634c892495a5
F20101204_AAAVIQ dzotsi_k_Page_098.tif
5b1c2dfc672deb4b38b41e7876373764
5042bb6f1123b54680e36ed3c204b13160276894
33759 F20101204_AAAWMH dzotsi_k_Page_057.jp2
e9a7c80c7e9f4d9880c7efc3db3a621c
5aeafe9b6f1c1738c9d1f8af9765eecce3dd7f86
93540 F20101204_AAAWLT dzotsi_k_Page_158.jpg
94b84e4158cce9df51e8ae1451507f97
5918ad73e1bc251a81a09907f72dbe947bd89c5c
F20101204_AAAVJF dzotsi_k_Page_013.tif
dddda55ad5dc61f75744f1feaa6f89ab
8b2b49979191bee49887ac302d7c5e6290ed90d4
F20101204_AAAVIR dzotsi_k_Page_139.tif
ed359670b64481f41986e81b62ab52da
1bf9872834aec4212415ff69695db311edac8670
472625 F20101204_AAAWMI dzotsi_k_Page_058.jp2
f4ecd32850814babc5be3be916450c1a
a1559aa6ffa9b5e24470cfefbe8f6f2dc5982b75
107114 F20101204_AAAWLU dzotsi_k_Page_159.jpg
5b4ebcc819550c2c79db7dc2825013fa
99e13c6195dbb48e2a70702fd787a008d2ee36e7
F20101204_AAAVJG dzotsi_k_Page_039.tif
15c7a606de733164b74fbb7a1f03676c
2d3fcb73fae5f022a44dc6f27270a760d33791a4
77717 F20101204_AAAVIS dzotsi_k_Page_138.jpg
013c11eb1fb49bf5650f68559e1e0477
f3fad5a719b234da45943a4b0ca8e118e77db810
548293 F20101204_AAAWMJ dzotsi_k_Page_060.jp2
b824408117c506302a6414f038be4cff
2bd2d2aabe78e7086c979e284ff3d50a247fa330
74004 F20101204_AAAWLV dzotsi_k_Page_161.jpg
359e1247ed884ee18346734d328297e9
82879f1d9ba9fdb41681e700acab7d3efb9f6fdb
50374 F20101204_AAAVJH dzotsi_k_Page_115.pro
7db9ecd2b8ab79f15ed8467a0ca436d5
0b376f4543b1f685bcb0528f0bbbed955ad29371
68371 F20101204_AAAVIT dzotsi_k_Page_079.jpg
1414ea7f8ab95e2ce5e4e13312e53c34
dd026d42814fa5b6004d12ec5806a851835fc514
98879 F20101204_AAAWMK dzotsi_k_Page_067.jp2
1c76d0e673281d18e9eded55737c9d35
baf64f4117855365a9c8cd5834ee2263e3072ce7
110252 F20101204_AAAWLW dzotsi_k_Page_003.jp2
ef744c32e4a645b61be4ddb9415271fa
73e75b5f5010756585aff86699cb087c40b88667
6358 F20101204_AAAVJI dzotsi_k_Page_128thm.jpg
725fa25711d42ebdf25e3ce4a5dcb57a
54c0a5d40df1d1529c33382006613c9cff813063
53776 F20101204_AAAVIU dzotsi_k_Page_041.pro
00701d506ef2351d77486df270265646
742870adf11b85ea83bb5e14f87351f017f55a73
105444 F20101204_AAAWML dzotsi_k_Page_070.jp2
7544f5cf87f7808ce1fb61ba25aaf1da
543bd7896384476fb032a8c4b5df32f556062855
1051955 F20101204_AAAWLX dzotsi_k_Page_006.jp2
14ae1181d3546d030d3bfedada73e3ad
df06b8c8f3acbd944f66e78cd582b05200769b38
F20101204_AAAVJJ dzotsi_k_Page_042.tif
07ce2e0093cce1a82d884ea5b5469e39
df08305d1862fffe4c1235cc146a8db563b63c83
51113 F20101204_AAAVIV dzotsi_k_Page_034.pro
d2c780dad0dee82226d431529629959d
05c4d8d04ebc53ad41dcf58e3703619971ec6637
1051969 F20101204_AAAWNA dzotsi_k_Page_130.jp2
91a2668fc38e11ff36a45a94049247ff
cab43e89b8c58fbc0c832dcc0091db1796193439
102174 F20101204_AAAWMM dzotsi_k_Page_077.jp2
43697335d09163eb7ca18730b008f326
0e14cf804f56edea3918956ad7c43df26e4f9865
1051985 F20101204_AAAWLY dzotsi_k_Page_011.jp2
71b973db1b9e5245a46ec784c16c6e31
80f1ef4ee11a19aca662071ed8c45278c9da6ac3
25085 F20101204_AAAVJK dzotsi_k_Page_066.QC.jpg
8344082916455e29e74160ea13cbdc70
16fb86d4dd8696abc41147b3b5a846c0b2c7d500
78654 F20101204_AAAVIW dzotsi_k_Page_024.jpg
18433853460e99c1672072ecc79b3d3c
673f87622b55897e5fd11031857c8dd4bc25e7fa
507679 F20101204_AAAWNB dzotsi_k_Page_134.jp2
1256d21f23945ed1f3804058107d9fb1
dcc9561a0b6c164e5a5bfb944c336017330d24ad
112046 F20101204_AAAWMN dzotsi_k_Page_085.jp2
1e01cf997d908737d4951f7d5899c8e3
871670ed6a8dc83dde99c38cb0e1fd13e6a43c63
191558 F20101204_AAAWLZ dzotsi_k_Page_015.jp2
caf7296319507a9e22a8c6aa374e272b
eaaeb71f46196f8884b6bb4d6f783271d8f497d6
116560 F20101204_AAAVJL dzotsi_k_Page_066.jp2
4a33e4b18f26df285c80e44470abdfbb
949e91f668428a4a5d177f7efb4a6de486f7ea2d
53541 F20101204_AAAVIX dzotsi_k_Page_147.pro
22a08c64a987c99732aeb57e8d7367dc
b285cd61c7af334b12c47702350051a78962e2e7
366727 F20101204_AAAWNC dzotsi_k_Page_135.jp2
1b3838d32e95dcff19dae24c9c6aacac
b5fd096518260e700320a5c5319ecee93fd39c74
108022 F20101204_AAAWMO dzotsi_k_Page_089.jp2
877e02623b13c0160f8781713eae87e7
ddda91614bd5f7fe708272b468475a69f75fffb9
78635 F20101204_AAAVJM dzotsi_k_Page_029.jpg
e7b3ac649077f41db5de69d95051fcd1
19ca86192890b6fd49bda0714eed11131e0a31cb
1878 F20101204_AAAVIY dzotsi_k_Page_081.txt
6d373c1fcc8ee73bc762684bc0aa9b58
0a2a425982cc45e473b12631e535bb45e61335b6
F20101204_AAAVKA dzotsi_k_Page_055.tif
16878e24582b49bc780a0444c3a26142
39915e5540aaddeffb891e56d650bd1758fbdaa7
105442 F20101204_AAAWND dzotsi_k_Page_140.jp2
14d68bd420b7ad64cc64c61b9af44fe3
399734a6815a37266bfd4e0b58fe18b5e12fd393
120222 F20101204_AAAWMP dzotsi_k_Page_096.jp2
89a8a4abbee8f7ad685105b88667c62b
49187606d5a53419603f960a6be05fd50273601a
24066 F20101204_AAAVJN dzotsi_k_Page_115.QC.jpg
ca83c4d31f273ad3eeb1fab07908071a
4197ce941018af3382214f2e089ac369c7bbf9ee
18747 F20101204_AAAVIZ dzotsi_k_Page_137.QC.jpg
7394f9a2ea495fd02f7c83b4ae9e8321
65e3921520c4e2a75fb6123c9b1fa65774090df7
54025 F20101204_AAAVKB dzotsi_k_Page_049.pro
aa91cbbed51089a45feaad4a0a84f2ce
92af19a3d5924cfaa2d6f43be96ab316c1be037d
282451 F20101204_AAAWNE dzotsi_k_Page_145.jp2
86c15ebf093022352f668e38aba415a9
f1cc4b2b5f3c644bfbf68c79ff937441710d1ee4
69439 F20101204_AAAWMQ dzotsi_k_Page_099.jp2
22bf330b70486c4ca2d8bffa77516fba
7e4d0840d95a66ece7b2ffa4f764ff664c39aa8e
22914 F20101204_AAAVJO dzotsi_k_Page_078.QC.jpg
54741d488c93e01922cffc4675a93f51
e2eaa5c7b584d665f46b6d041193293cecab103b
21836 F20101204_AAAVKC dzotsi_k_Page_059.pro
918f85e7d46b0bb06b99fd15b5bcacb1
69635e71a6c54201fb404541130d9e0b08d0ef59
112882 F20101204_AAAWNF dzotsi_k_Page_146.jp2
47ab7beb417ab01c65485c94354602ec
7d1f8320357685d50160a6fb62eadcb87fcf83ee
79405 F20101204_AAAWMR dzotsi_k_Page_101.jp2
6a706f7af872220f5fe9c712d766629f
87dcd0f7043d65a46adcf4eb8b04ac440a10cd4e
1051965 F20101204_AAAVJP dzotsi_k_Page_010.jp2
f174cee8fdeadf66191bcd3ca2c2cc3c
3771b7b501e95a337890294d4dbee44682eb469a
1051951 F20101204_AAAVKD dzotsi_k_Page_012.jp2
275bf064f872b3300b7324938febb540
2635dbc764c4c822f89c7a394979a1503931d026
112104 F20101204_AAAWNG dzotsi_k_Page_147.jp2
c42d7e9366cb10419ab4db0ef384b551
95fbb1b45b748c5ada0b6a9bffd1fdb69de59095
615377 F20101204_AAAWMS dzotsi_k_Page_104.jp2
0da56ce27e9c217388a7aa57e0bf7667
e918f90e91853d5b3fae28706b3eb9b1b66e3a00
79669 F20101204_AAAVJQ dzotsi_k_Page_011.pro
8f457e015dc45017cbb17d2006485760
72af04341e963b86398562a12b5c7a2687c8c331
34287 F20101204_AAAVKE dzotsi_k_Page_026.jpg
3c5b24d1f65d8897dc73a4be78c8ebc8
94bdc81fed2ef7350acfaad91332ac8c55285faf
86091 F20101204_AAAWNH dzotsi_k_Page_148.jp2
5378b7ebf8748bd5c8a700e00e8b6d58
5c4ddd86b5f254c176c05bc84d904734e25c698f
776720 F20101204_AAAWMT dzotsi_k_Page_106.jp2
32fa84405830422b159b19f809abc7ef
366e7854d86e3bc1ec4b23539319660a89e4e6cb
11823 F20101204_AAAVKF dzotsi_k_Page_058.QC.jpg
a88c0c3bd07797367e6ad89a19e4ec33
f8a0c64d8432968a2b4242bf415776ad7ac62feb
F20101204_AAAVJR dzotsi_k_Page_145.tif
a9ad12be7069dffa415383986089c12a
9840c3ad80a0dfc2a10aee37736146aa1bb45b9e
101824 F20101204_AAAWNI dzotsi_k_Page_151.jp2
6147a42e47ee2b14c8ff83e780d51ce7
d1897ae5ab6776c269aac07f49de5bfd80de8543
91147 F20101204_AAAWMU dzotsi_k_Page_118.jp2
47ea18b2f5a128dbdef50fcc780380db
f8e9f77d75770319d5d7ee52dfc0b937fce98ad9
25480 F20101204_AAAVKG dzotsi_k_Page_133.pro
31455a0b7324322b98e73d7cd8aa3092
3798380863d3e3504cf8dc5c3993a88195019245
77336 F20101204_AAAVJS dzotsi_k_Page_037.jpg
90db16d8855e5ce666355b1a2ae08bb7
adcc74fefab7876425c137e9a6bd6a85d6f4083f
89178 F20101204_AAAWNJ dzotsi_k_Page_152.jp2
6beba2808f4af8f3956f1b4c85c289ed
f7d7606afa073c95f2dfda9101a247c0a92dfcbd
92387 F20101204_AAAWMV dzotsi_k_Page_119.jp2
e07e1c0dc1d34b6413dc08a45879e96d
d1022de864b331df418201ef93ff8e38400f1842
26633 F20101204_AAAVKH dzotsi_k_Page_156.QC.jpg
d0fa996cd4f071409e2911bb6419929e
93bf69070c789d512965dc1027d6d57c20c9a824
104678 F20101204_AAAVJT dzotsi_k_Page_111.jp2
b234f59dbdbd9872821a541217e81890
3e79371a613ec4c6c345afbe75ddb37496f7c5f5
136878 F20101204_AAAWNK dzotsi_k_Page_156.jp2
4792299d4234efad0c21b952c7ee23bb
3058ba7652fa4602ef58746046cd664ece921e75
113698 F20101204_AAAWMW dzotsi_k_Page_120.jp2
d4e56ab5ecac8955f4dc626f5e1aa127
6631c02c05974cb254f36f0c1ff9f15376bc03c8
73163 F20101204_AAAVKI dzotsi_k_Page_039.jpg
92f8c32b228b82bc51beb4c81ef453ce
17a055e9e4f05d0ef16e394872601390b736e52a
5971 F20101204_AAAVJU dzotsi_k_Page_119thm.jpg
4ca76ebe54c168f3b8d34a14ecbdb2ae
56732f1af3b346ee8d69b707524a75323ff0e27b
F20101204_AAAWOA dzotsi_k_Page_036.tif
014af8e5ee9962986d22363373686719
3ea3318207fc2cb19d604d8531c6d05c5622b997
143388 F20101204_AAAWNL dzotsi_k_Page_157.jp2
718703670bdce3f2593a7891e8a1ad45
d9f48d57679ba5fb5e0526cf8972d20a133e7814
103542 F20101204_AAAWMX dzotsi_k_Page_121.jp2
39736a3ab60f88fb86b96066b505c2ef
87c44d504b5fc22c75d7764f118bb27d87fe6e21
4501 F20101204_AAAVKJ dzotsi_k_Page_144thm.jpg
d0f6fd8594d1f2ece1a7885f51335506
46f7903fb4c153cced6fe123dc69bb88629c3cc0
2005 F20101204_AAAVJV dzotsi_k_Page_018.txt
ab413e85a312c6c0761d41eb8c38e16c
50c8ee6dc88cf07dd4bbda8b1a309f654f3bc1bb
F20101204_AAAWNM dzotsi_k_Page_002.tif
79cfd72a56dfd315e596be847b83470b
c1a39008634dbe0c9d5831ae94d319e01c37a164
112001 F20101204_AAAWMY dzotsi_k_Page_123.jp2
030058bf532433ae577acb4d9722e195
c50405a1257fddcc1946c4cd96368248f9470f3a
22008 F20101204_AAAVKK dzotsi_k_Page_151.QC.jpg
c16dc28dc1cdb8b380f5b4b09a662fe3
280b64b265901f3fce7a33f2321bfa124c38fb34
F20101204_AAAVJW dzotsi_k_Page_146.tif
258b34a5782195cb71562ac45e568ce2
508ce7e773ced0c2e3edd562ff7238306389d49f
F20101204_AAAWOB dzotsi_k_Page_037.tif
4ec825d3d993c6e96de1e8fe88b22f12
3362848f259bccc4cd9caf59b8840b00f463904c
F20101204_AAAWNN dzotsi_k_Page_004.tif
40b15c183904cb53d6ef097fa8f419e1
c3ff9f33b3d613d7f16198f0bf96e6a24ba7d6e1
33884 F20101204_AAAWMZ dzotsi_k_Page_127.jp2
072052b1fcc7e2341e376e78d676dfe4
822c24e99de9b106fcba3e0e07a733589f27cbee
108105 F20101204_AAAVKL dzotsi_k_Page_035.jp2
1b61399747405c358db5e41f4bea78c9
98abed7505e531733410d3247a174781aac2e619
102994 F20101204_AAAVJX dzotsi_k_Page_076.jp2
838805dd199e781b62784022e46df277
08c154f4b902d57378f39169c1c42bc070d307ee
F20101204_AAAWOC dzotsi_k_Page_038.tif
63db67a514a2e4b337556cd4b978227d
72e8268c8bd97d1df81a9598dc6ccc7595549d8a
F20101204_AAAWNO dzotsi_k_Page_005.tif
1d6095b96900f9af730bc3a680999d21
6058258dc688364e81cb7d25646e3143c39092fa
114806 F20101204_AAAVLA dzotsi_k_Page_094.jp2
02e254b382fe8dfcfe8e070d5cc5f021
c4aefe1d6410f469254915112bf844b3e00add24
F20101204_AAAVKM dzotsi_k_Page_082.tif
b936c5c8c836dc614d3ab5791d5a51cd
1807145b9eea761a07feb1d4b2d2e860a55dd3cf
7964 F20101204_AAAVJY dzotsi_k_Page_143.QC.jpg
9b908d6e578d9d5082633a96789cc016
594c40b48808a1bc0efd7ad3d41aeb491190f987
F20101204_AAAWOD dzotsi_k_Page_044.tif
b91196af3ebbd2f3939096e786e9fc02
84db136f122de91aba49a82f7196a66e12bd31a5
F20101204_AAAWNP dzotsi_k_Page_006.tif
f32d278c259ce5d4ef468e22209486fb
afffb314ab41f171c5bc2066deddbc56d8c58a16
25375 F20101204_AAAVLB dzotsi_k_Page_045.QC.jpg
4b8400195e104b98617701022c6101bb
2609d882022b713c95f21853520de8dc432f055f
2750 F20101204_AAAVKN dzotsi_k_Page_128.txt
1c5282f09eda0589d08c293a24ecd627
53d8b0d42a586e935827228e9c004fa4a3038b2d
102 F20101204_AAAVJZ dzotsi_k_Page_002.txt
d5f506218f760c528b89916e80d7a43a
a0afd9a3771cce86c9e9b3bcb6957507b69b3721
F20101204_AAAWOE dzotsi_k_Page_054.tif
4be2ac9ddd5ba4e0f50b38296748b731
c9bafa031e6b1f871bfe6a8a9d7150e2d5bce362
F20101204_AAAWNQ dzotsi_k_Page_011.tif
43c1142f9df0b855a38032b8178e10a7
6acafd4ca22f13d631319c3c2a94b21c9b50189c
44815 F20101204_AAAVLC dzotsi_k_Page_067.pro
98f838c9f9987c42b61fa9cc84f0b1d7
fe1a2bbe4fb993c1698ae3d380783fe3472462de
111269 F20101204_AAAVKO dzotsi_k_Page_091.jp2
b42c8c3c9a79d3034fcfb84fc3144260
b1ac2658423cdd8ccfaac3974e3d61304dea2dc8
F20101204_AAAWOF dzotsi_k_Page_060.tif
78c7e1c2d2c2c88a48475e5de7bf8911
553cb0eeb62bdaa0241c7db226ffc1badc0ab882
F20101204_AAAWNR dzotsi_k_Page_014.tif
8a57d39951c7db9e92a7d2b573c9962f
39f7773c35ca6451ca836af4eed3353b535b218e
27385 F20101204_AAAVLD dzotsi_k_Page_071.QC.jpg
b2450002ed9a015af0e4aebfc6233b39
b3bde311944e4900f30b0b9c26755f5398ddaa6c
100877 F20101204_AAAVKP dzotsi_k_Page_081.jp2
e00ac5b6ebdd162996dc53a7f816c593
4cb8c4f8501280dd7085430f0b34f600959e83bc
F20101204_AAAWOG dzotsi_k_Page_064.tif
863e80707e9f4e133dde4a9ccf49091b
7d42f7992070f487a0fcbeae5dcc7d9bc26ad8c4
F20101204_AAAWNS dzotsi_k_Page_015.tif
ce2e89de693864142d79a95773d8e7a3
235251a8b9f0615e17f123d406c3d63179df28df
7002 F20101204_AAAVLE dzotsi_k_Page_029thm.jpg
7e1e33e5d14a44c31ba21c71d6ab149f
76587c42f3e11a323144a567644b880d17a3f06b
F20101204_AAAVKQ dzotsi_k_Page_023.tif
8cb4d31fe257172e7567402351e3523f
b0d411e97e37f625c7f2b7f39057fc2320f0f4aa
F20101204_AAAWOH dzotsi_k_Page_065.tif
adc0a23b062a953406fb434ce19b8bcb
e25f5eaa55af901a4945e1135e3f0ce6b4ee155c
F20101204_AAAWNT dzotsi_k_Page_017.tif
34c3a9a1e4b2eb9cd40d1f00af82dd3d
693d5d84ee30fcd046c6f51f114cef105fc4bad1
6257 F20101204_AAAVLF dzotsi_k_Page_033thm.jpg
5d3fc8ee3836bf4be33813f38d04b650
9a12aa62d36788500451f7e609628b8eafa344f1
951658 F20101204_AAAVKR dzotsi_k_Page_137.jp2
00e68b0f8668a79accaeaef378f4db18
b0e99017b44fbcc5801d546864735c6174c0be79
F20101204_AAAWOI dzotsi_k_Page_067.tif
1b97007daf2baad7dafacc3d57a59b0b
f7605a200d6aa34b0c4c65a5e843d7bcfb2eba3a
F20101204_AAAWNU dzotsi_k_Page_018.tif
e9284a97bc212491321f8fb50b3b6e31
8aebc7746c8d4735bbc6dfb1fe6b5fb2318beed5
F20101204_AAAVLG dzotsi_k_Page_056.tif
3d1f3161d4a044b610050306ad18d7e1
95fc6b348c3d3ced9915eae090a55afed5b72b97
78210 F20101204_AAAVKS dzotsi_k_Page_022.jpg
a561dc8217c345d433004ac1356dd8f5
1a842a89ee9e66efded094bd79445d11b00a9403
F20101204_AAAWOJ dzotsi_k_Page_068.tif
4e5aa982c15bb5ff1af4a8a03ea3ee84
61d9e5dbcb0df39a65f6474cb99253a75f5b2909
F20101204_AAAWNV dzotsi_k_Page_021.tif
db66681c658d9add434d5de00933b1e2
6e8b378f0b33f59b1c42e1fb6f9486bc85c2bcd6
F20101204_AAAVLH dzotsi_k_Page_116.tif
3bd9bf7241f2c171b5b8b43d91096bd4
2d758e3b76c9c6c8469b3d06330b784b6f26dc66
7077 F20101204_AAAVKT dzotsi_k_Page_009thm.jpg
05ad9ea3e19d58b3eab49058dd913ca9
e706c33f5739a4397abcd906fbd2cc921926e35d
F20101204_AAAWOK dzotsi_k_Page_070.tif
07e641733b9ce8a9bb90b536bea26aec
cfb3cb04f88c109032f292798481febcdeae9963
F20101204_AAAWNW dzotsi_k_Page_026.tif
596c418ef94873b66ad9d9ca4d2c41f4
e6290a9acba78a8196723ed34a43739d3155a066
F20101204_AAAVLI dzotsi_k_Page_088.tif
99e9376eb55cb839ea44b7754c346217
d33fc0bd0bc0723987aee4f1c1502014f038cc26
93160 F20101204_AAAVKU dzotsi_k_Page_082.jpg
9e532a65cf6d9a2d414536340ad20755
3a3bc5354341a85a83fdb167c6859f067ebba913
8425398 F20101204_AAAWPA dzotsi_k_Page_104.tif
a7174347e6b32b06255047c050a608bd
09e72e01788486024a27a5932c4757d4f2002474
F20101204_AAAWOL dzotsi_k_Page_071.tif
3c28603d41025b6aae411ff5a9427b95
0f2dd360e63c9cfc53242119ef8b2b57509de475
F20101204_AAAWNX dzotsi_k_Page_031.tif
d982fe6aa7a275d784ba2c376dc56cc9
328688a756fd872ea575b81daf644185b922e78f
F20101204_AAAVLJ dzotsi_k_Page_118.tif
6c21be6f6e1ab8677ae330460f77d83b
f2168718fd15567d057b19b28f90a7563784837b
F20101204_AAAVKV dzotsi_k_Page_062.tif
6ca992c4295275bb2fbefa75380222dc
d7eae52c8d7b6ef02d4bb17283d28d127014aab3
F20101204_AAAWPB dzotsi_k_Page_105.tif
66869cb7b1b25cb6e3c9a90dd1271d53
175114056b9ba0ae88206984f30db3d846b0dd83
F20101204_AAAWOM dzotsi_k_Page_072.tif
71500fc0313f665178d05500c8491387
0f6819aca057307327dc696b7517d94cee518255
F20101204_AAAWNY dzotsi_k_Page_032.tif
2a7a2ce004614b8f7f40778c8fb9abf0
f89fb764c647469c5e5ca9de4eb4ce6a11199125
141971 F20101204_AAAVLK dzotsi_k_Page_154.jp2
972f136215d1d3a723e9c5713f0177c9
c29d2f144928a924838a8b8dce56417d142e775d
2202 F20101204_AAAVKW dzotsi_k_Page_055.txt
77e3b28b100ddc548fa1ba0272d74250
c7331739de3a466c56c423b9d71d63e8b117a139
F20101204_AAAWON dzotsi_k_Page_073.tif
dce1e7727595111146f8218a7611e19b
eca8419fa1dff4a4a1c8996c0199965d9d7ea42c
F20101204_AAAWNZ dzotsi_k_Page_034.tif
38fb8c794fc4fb39ede8f9c5e54ba4b4
528754e8b02d534ddbf57328c8889ea13dbb5b9b
110419 F20101204_AAAVLL dzotsi_k_Page_080.jp2
1fff34142c3c47b489b6f566b0d329ab
1874870d482209147596cb5759f575f500d52ae0
1051983 F20101204_AAAVKX dzotsi_k_Page_009.jp2
4e2ab7e8a536acc6da1874ac3d9a69eb
8f8af85dd0bcc15058b7c793d92d37689faccb79
F20101204_AAAWPC dzotsi_k_Page_110.tif
f8382e38b53a8c219c0a7b819e6bba67
00e0a41850cd2507c67351c3f79185809a08bb4c
F20101204_AAAWOO dzotsi_k_Page_074.tif
374c7412eabfc48971890f4fe92bd6bd
5553446a66466973fa8500737cdca32be386c574
F20101204_AAAVLM dzotsi_k_Page_147.tif
43aada11bce0694d27c527cb4b8c1923
8fd9a112ed105089730bc4a5d1c708da53fdb2cc
F20101204_AAAVKY dzotsi_k_Page_157.tif
979879e28dcad22e2afd874a903c0a74
e8bb05d0178f3ce2007fefb134de59191d3a92c2
F20101204_AAAVMA dzotsi_k_Page_005.jp2
05638516c23ffa76da77dd20d9d8f481
b5fe07f930fee3ea1dea1f920f8f69355a84f661
F20101204_AAAWPD dzotsi_k_Page_113.tif
c3219ffd612add0afee94f9aa092a5d7
a85bcff6f54c12d5528c88a5330a66e0db393628
F20101204_AAAWOP dzotsi_k_Page_075.tif
8b8ab168bc8b690b1abfb227e9491353
e3d7e5c94f4c75efad0f4d5587b064dd86a1c92e
2068 F20101204_AAAVLN dzotsi_k_Page_039.txt
41005b74a0b5bb2c689cd21af59979fb
e908903fd8812957a8da61961d3375bb04fd7bba
73539 F20101204_AAAVKZ dzotsi_k_Page_123.jpg
6c9ad81d24fe8505a50d99fbefe731dc
eb72d1a6906718e63268f924d7101e159ec7a8b6
67772 F20101204_AAAVMB dzotsi_k_Page_092.jpg
e9694cbbd6c2fdb7d639393222ffc2a2
38f34b892eacf332c068e2664c1097efcb0a873d
F20101204_AAAWPE dzotsi_k_Page_115.tif
c75b74338a2494175fe96f2a083239bb
ebba6fcebc8e91baeda1d4ab8f47b80399fc6c43
F20101204_AAAWOQ dzotsi_k_Page_077.tif
d8b936c8ae5203998f5373f2f1085d8c
503558f55aa8d1fe1d1e8e235db86ccde62e00d7
113549 F20101204_AAAVLO dzotsi_k_Page_128.jp2
798c39671dc860dd166e71626bbdee55
81044d5bbf9063a8359311b817586eca2143ca2e
51039 F20101204_AAAVMC dzotsi_k_Page_089.pro
d0745b28bdd3d9487d57b2735fdd8bd0
d60234d16adf6c61134d1bab9832fa8211a5e367
F20101204_AAAWPF dzotsi_k_Page_117.tif
6356f2dcbf0906a3a04fb554bf7a5af0
94156ea8659d4a25aa1a66cc3f80e8d5f1e67781
F20101204_AAAWOR dzotsi_k_Page_079.tif
9dfb187c5068adfdec26ce4aca1ef7fd
d9dd83dfb2f69cb072ff917580fb3224226e2f6b
30191 F20101204_AAAVLP dzotsi_k_Page_132.jpg
fae5ece80564dc56c592c595cf71f813
eab25709c0f683b726327aacb16eb1d4a400bc95
3099 F20101204_AAAVMD dzotsi_k_Page_027thm.jpg
d1160b81136cc5a05ac6dad520c8012e
03bcae4ee23e960007ebb59d264074522fa1d577
F20101204_AAAWPG dzotsi_k_Page_125.tif
b1d8b006897cffc7deed0f23d2b80787
7b71271dffdf7164c917a58d8b21d25ce53761aa
F20101204_AAAWOS dzotsi_k_Page_085.tif
56b24dde72af334dbcce12a4b1241f52
bca10eadbd6a5d139dba8bab7986847264271c0c
3052 F20101204_AAAVLQ dzotsi_k_Page_142.txt
6d37255f1eed6079ab964cc60aaa4f52
104114b739e37f18d18e83596e443a01dfedca15
55261 F20101204_AAAVME dzotsi_k_Page_068.pro
22772a7b95db241cb49e5311c974be1e
e961acb7d240f58b76b862c3f1f5cfcdfb7a3ccd
F20101204_AAAWPH dzotsi_k_Page_126.tif
542acc230c9675651e80649034108d94
802256a8a0bf8ac659bce13d32e1debd6ce6f0c3
F20101204_AAAWOT dzotsi_k_Page_087.tif
78bf78867019d6db3fd7500ece5e1848
37363711ed45a783f1580bc0345d2d51ddce7710
F20101204_AAAVLR dzotsi_k_Page_078.tif
74db060ac9d3d592d793402fd9efd62f
53db7d56b05d172df5349d48d1c99b200ea674b7
1886 F20101204_AAAVMF dzotsi_k_Page_092.txt
90f836ce7b22b2344debe42aaf64c599
c5e2b68ca4ee827a18d88ace8393052078df67d2
F20101204_AAAWPI dzotsi_k_Page_127.tif
338c383747796c359245a310dfe77612
50e578cdf5c260d5474112c507e2869f42898ac7
F20101204_AAAWOU dzotsi_k_Page_090.tif
dda2cab2ac0f30a669317cb63461464a
7058b02be8412b20ca41331a1bfd6414abdd1766
2111 F20101204_AAAVLS dzotsi_k_Page_146.txt
ef8b77bfa509a58e2091a5f5db46f395
f5bd70602bcc57876189c62f3b4f758c397995bf
23275 F20101204_AAAVMG dzotsi_k_Page_142.QC.jpg
496a198bddcfd4dff7d5d4ea81dca7f1
176a081f587447de48d8c01c994d7b628de5b6d2
F20101204_AAAWPJ dzotsi_k_Page_129.tif
eeeec343358e78a8bf91294a06b2af44
6897342ffe83cecea03591f27f55bb88f8afd306
F20101204_AAAWOV dzotsi_k_Page_091.tif
40a366efaa3f1dca7e408d41c8e89b0b
9b39339ec654333ab032ea39e2f1225c0cc28ed5
116480 F20101204_AAAVLT dzotsi_k_Page_139.jp2
63037e75e5c936c5a10d9f8500f3e035
da92acd417f4d4ec41ed6b84801dc3bc5f99b97d
F20101204_AAAVMH dzotsi_k_Page_148.tif
6c126e44b6c72a6e5c56fad1d4444216
6c272fcc36fab1e4264a9f4a2f9eabd70051f358
F20101204_AAAWPK dzotsi_k_Page_131.tif
2a134f30986cc096dff2523ba10149f6
cb3f09d5b76fb743d488c61116a5f153acc50bea
F20101204_AAAWOW dzotsi_k_Page_092.tif
d5267f3646e54749cac3bd0c6b2fc442
9270221f00a5efd97f420c9be080cd0931633acc
46082 F20101204_AAAVLU dzotsi_k_Page_149.pro
3a44cdfaa416c962dca75bc796dc76da
fc807577b3e334e11235451c32aa701f46183715
F20101204_AAAVMI dzotsi_k_Page_003.txt
8de3dc3b6cbc844ba73bd24b32bb311c
213d14e1611d32c49228c5fd224798f54c2150d6
56322 F20101204_AAAWQA dzotsi_k_Page_029.pro
8328ce1288f4c61f0b0de185b8182ede
a903193b72359bcb45d1becd29b69aeaa42d976d
F20101204_AAAWPL dzotsi_k_Page_132.tif
370a2e8f3e34a42dff61d6aac0b841c1
43a138993ba4c3667a2617bdb3bd9d56280a2084
F20101204_AAAWOX dzotsi_k_Page_093.tif
0aac85b7a4b4ed57b0d1a96bf12f3199
3851d7ef608f32568cf724fa887f627d735f4a14
7512 F20101204_AAAVLV dzotsi_k_Page_154thm.jpg
bbf130f6ed21f4368d5355b2f1001ba7
22700a783898365fc056fcbd4ba14e07d97c2037
4503 F20101204_AAAVMJ dzotsi_k_Page_061thm.jpg
4d05772080b5588cbc396f59bdcd0810
8d28f7341c8057b53a0f966a2fe2cca19c32b01f
50058 F20101204_AAAWQB dzotsi_k_Page_030.pro
129b64fcb9fd0aa600e7a42f963aec0b
41881b30ab5794f8f23b3ad0fbfa1a43cc609ccc
F20101204_AAAWPM dzotsi_k_Page_134.tif
cee13593665fac6ced6c867fa178f6e9
d454078ac27ca99ad48c456b1319ee5650ba923e
F20101204_AAAWOY dzotsi_k_Page_097.tif
5e6ec2b5cb05b3fae79fedfe931b8242
1a431e22f799cc4341aa6d2149040f85eafcdeb8
6315 F20101204_AAAVLW dzotsi_k_Page_016thm.jpg
a34415836ae97014f06cd97ff2916d33
5504c0c31359024069e313f3dda73e3256bae852
45710 F20101204_AAAVMK dzotsi_k_Page_097.jpg
79de40c96653da6baa4baeffc5565088
dcaf97c2ea2082e680bde523b6335e95427eb69a
54202 F20101204_AAAWQC dzotsi_k_Page_031.pro
2cf2231c0ddecdc909fb1cada312cbb0
68b6ef9560e8891cd93553f7e1d431b4f99d4a84
F20101204_AAAWPN dzotsi_k_Page_136.tif
c06728906c0cf8ba8cf0b44c733dff2c
48193ad3d57e37b563caf88406c24fd9d6601d41
F20101204_AAAWOZ dzotsi_k_Page_102.tif
e78bca4f22996d75af8f521da1ddbd68
2d275dea38b80c6bdd1689f3c182ef40e8cdef5c
235 F20101204_AAAVLX dzotsi_k_Page_015.txt
226e45bd63fc9e171716e8b99c79fd7c
4e1ca0b2b0a4e4e9e535b167e8224b67e82a921e
12174 F20101204_AAAVNA dzotsi_k_Page_104.QC.jpg
4847c89389de4f180167f4c18f2fa312
7aee220f7556d83c41038b3942457bbf2283cebd
22513 F20101204_AAAVML dzotsi_k_Page_076.QC.jpg
a2a9e723c9fcc801534a2a9b424b6639
e241cfc5625ffca2a6a00ecf4abcbcf2b1526c2f
F20101204_AAAWPO dzotsi_k_Page_141.tif
2269191f511e52183e657f7f598c7dc4
5baa3c5a784ba66a6f37c7370128d0c6658629fa
1779 F20101204_AAAVLY dzotsi_k_Page_103.txt
08eb3df53ae814bebb7414fc8e4cd638
164e231d36a97d781e15c46046b7d46aa75cc8ec
6912 F20101204_AAAVMM dzotsi_k_Page_114thm.jpg
247967499b08b157c01bac05ecd0a5e9
59e92f1db24a552a88387cd0105df0a943bda90d
54195 F20101204_AAAWQD dzotsi_k_Page_032.pro
73ec51d4cabc8b7ae02373cc0ada555f
6de26d8c3aa22b2bcd27232d30dee94425953f09
F20101204_AAAWPP dzotsi_k_Page_151.tif
4dabec4f2d5ec7256f67622327892617
8af446f898e1a4f32e4de17d0c5af108194cf005
111344 F20101204_AAAVLZ dzotsi_k_Page_115.jp2
30a9fa89ab45c4b9565480170deafe3e
480d0a52a2e47bc661b6b6a5b81af1196f8168a3
F20101204_AAAVNB dzotsi_k_Page_140.tif
8bb3ba7701e955f51d694b86e2db7aea
edee74eb655b27c499fb2921b841765c3dbe1a87
4825 F20101204_AAAVMN dzotsi_k_Page_059thm.jpg
0e4377067beee635fc57cd1fac7e2747
eb9f46fa13ad111c365b1eb6e8df030d1f198f49
49475 F20101204_AAAWQE dzotsi_k_Page_033.pro
26c310a14499ab9993e62062cb0e2f9a
492d55ef409f74448cae8b102f4bc07027b3334a
F20101204_AAAWPQ dzotsi_k_Page_152.tif
5d3da6d9ebdfd665a954cdbb7b691aff
af6edde8e0037c296f002b37462f9a26af925ce0
50555 F20101204_AAAVNC dzotsi_k_Page_044.pro
002f1224e2e02ca992a2450f0c75b579
d64b74cd6138db571471f46252984e7be1a394bb
55648 F20101204_AAAVMO dzotsi_k_Page_040.pro
ddc36e2f090db2eabd68f56a05b57f54
24619567dec6727dcdd3c94e10cb3ecae1169af6
49291 F20101204_AAAWQF dzotsi_k_Page_035.pro
7842ad627dc8a9ccd773da8654d12af0
05f42ee55fcb86ecd367a5f47631045e1d424a4d
F20101204_AAAWPR dzotsi_k_Page_160.tif
df893d68e357de2ab739264b6a2190c2
293d8a347f42911bf0fb233b50fa0bd2d052ede7
6975 F20101204_AAAVND dzotsi_k_Page_019thm.jpg
30a5bce2381c8e80c566ada58ef839c8
1862114388bfd34884e641127cb2cd3e5b0cff4f
80152 F20101204_AAAVMP dzotsi_k_Page_086.jpg
b20281c19d3c322eb0c2c31e8c81de30
58330ba7e4b6e10958e639d31ee0b9c2d4259c5a
53871 F20101204_AAAWQG dzotsi_k_Page_037.pro
e4e6ca612712a42cb6274732d6385ca4
9248ee7d051c700e457bef3ff425db39e9bdcc7c
27233 F20101204_AAAWPS dzotsi_k_Page_004.pro
109e5be25b58c739d35f6cc3c316cbb3
66c71602c8fe12e1ed8bd29f99e3237d6c95a177
102546 F20101204_AAAVNE dzotsi_k_Page_018.jp2
0f09d8ea7b6ab0703d9cf87d026c28ed
372adf926f172d7639d8c5fdce18e25792116f58
76256 F20101204_AAAVMQ dzotsi_k_Page_094.jpg
16e3a392b3fc2d4085071cdf7b30906c
d171e4231ad25adc44f2a8044d727258b5df730b
46886 F20101204_AAAWQH dzotsi_k_Page_043.pro
cc5f7692a8d077426d5e5dce6cd542f9
79ae56df8af65e3afd027c17dd5484480c4f7c58
73864 F20101204_AAAWPT dzotsi_k_Page_009.pro
5a6f2673533900e6d52f55cb6a11d081
3cd294342a9aaa4e29ea621848c3793a4636faf4
F20101204_AAAVNF dzotsi_k_Page_099.tif
65e8b316c950b06b640e47755627f6d4
90e8776cbf9e44f52a8dddb4b554c078502925f3
30912 F20101204_AAAVMR dzotsi_k_Page_103.pro
aa4ae3525fa0326f9c175f0dc15d1185
9426f8cd90e7102fa60f7c373ec6784a96a82491
54391 F20101204_AAAWQI dzotsi_k_Page_047.pro
9910bcb869f14f3aea53014924d1b5e8
7b71533810fbf9b4a48dfb6c1f8944538a573cd5
71337 F20101204_AAAWPU dzotsi_k_Page_013.pro
d8ae75a80f3e7a310cffe88364c260af
1f2d38e0dcf1f832fd4e95825bc967e41ecf0f6f
23458 F20101204_AAAVNG dzotsi_k_Page_112.QC.jpg
6bc02148dbff7264db19d2de0e4117e7
d2300d8db2f7b3669ff5e076ce0efc307e7b58f9
6838 F20101204_AAAVMS dzotsi_k_Page_023thm.jpg
21b9d5f1577b38af89197e175fe7fcc9
f19333d7885d1bafc6a57ec392d118a408a84826
53532 F20101204_AAAWQJ dzotsi_k_Page_052.pro
b375dba1ca0e649d50e3bd54a3e811d8
accde4f87965bb1e5fe545b1d50d2e12fd7a75ed
47112 F20101204_AAAWPV dzotsi_k_Page_018.pro
acd11d95d75efb3dbfe429299e876f23
4f9effa4ddaa5c483df11dc3f2c9a7c11f944ccc
7556 F20101204_AAAVNH dzotsi_k_Page_046thm.jpg
a97affc64b7653ac85d139866c58a65c
f32356befe6233d2a5c519affe81e9c2c11e26dc
93911 F20101204_AAAVMT dzotsi_k_Page_073.jp2
a8ca0463bcdceaaa1aae5597d6638af7
662616b85c9e70d02bbce12b0b8c6860e9b87fe0
56207 F20101204_AAAWQK dzotsi_k_Page_053.pro
535d2beb865e2d54d86b5165af6476cb
d3939e89b4cb7435c4cb5e3d1b93733c5c16faae
71858 F20101204_AAAWPW dzotsi_k_Page_019.pro
45ee89c7c6229fa909c3e508bf02be53
03a829a364cee893d0b37fd363d77bcfd809ccd9
88275 F20101204_AAAVNI dzotsi_k_Page_075.jp2
ad11a73658a8156ae4f48caf44b6cf77
eaa3b5d55a88ef6875c4c27d7d68b5be979937ea
15987 F20101204_AAAVMU dzotsi_k_Page_143.pro
5d1a218443e93fcc218b24ef9b3b5075
2b75a4601b135b482760a62ca19411bebf5dfa05
16017 F20101204_AAAWRA dzotsi_k_Page_105.pro
2d2e1daf6eb974d45e8ba693ad4f290f
03f9b16e5cf07230944704ea45498bf87ace855b
42204 F20101204_AAAWQL dzotsi_k_Page_055.pro
26534aa84676e7f8b207bc4f538cbceb
d36d635ef3fa44b130d24ddc5d7a1bc5c78b17cf
55450 F20101204_AAAWPX dzotsi_k_Page_020.pro
f98beb9c003834b9f07c8c89c371e0c4
b24a035e6b86219fee0fd0fe97e77b94f6741106
7113 F20101204_AAAVNJ dzotsi_k_Page_065thm.jpg
8a1fdd9b60fc0b0636b8a04a6ec4045f
9c4d0ff4fa2c25e552d9096a756b2f2bbdae308b
F20101204_AAAVMV dzotsi_k_Page_015thm.jpg
41a261b22f5cfc869860d91c4fb97bd6
c699431ebc902e3df5fe9fe15c5b69bdb3dcee67
16969 F20101204_AAAWRB dzotsi_k_Page_108.pro
1d67ca1b8f2d38d89d3f0a9292f31270
135236471c7d95cbc6c355c690559b76317732f6
16383 F20101204_AAAWQM dzotsi_k_Page_057.pro
fd1fabc77ea3b0bdf4e2e2c0b1f5fd8a
39ba1e4dedd382ff626163036cd138c08c3a900c
54733 F20101204_AAAWPY dzotsi_k_Page_024.pro
82d902aaa394e42e443bcb9787da5200
5ced3c89219d2dd88b22fc8b90694a5920ec46e6
1439 F20101204_AAAVNK dzotsi_k_Page_017.txt
a560110a441f14598db3f30d1721b4fd
cce51ec0b08980435f23595215fbd4942e4a0906
6556 F20101204_AAAVMW dzotsi_k_Page_078thm.jpg
7e5c7e6284b8086fab6aa0d7d357e5ad
d346a9067428e0cee6a97e39816d269a0f815fb7
48426 F20101204_AAAWRC dzotsi_k_Page_113.pro
9cb127ab91ccb5dc1690f8287d927d7a
efd184a60ff6056ca3454ab89a2d2aa1992c0e3a
50780 F20101204_AAAWQN dzotsi_k_Page_064.pro
2cfe746822df5963f8d2e0ef87e7e204
cde6799f83cf14f7426e6cab92d517219611cca0
29094 F20101204_AAAWPZ dzotsi_k_Page_025.pro
87110215bce8b3f409020c9d10598031
695e4ea2999c35260f4a9579c8e966cf84c9da85
108373 F20101204_AAAVOA dzotsi_k_Page_064.jp2
f05a7967e58d516aa7785835ef831edf
ef2ce33bb74d2c689c471ab8120ea25e986e09b3
6473 F20101204_AAAVNL dzotsi_k_Page_008.QC.jpg
3d7c4ef68da5cd04f13b9fb171495dca
b755204a49daedb09dfa05c7240bb3381b984efa
114954 F20101204_AAAVMX dzotsi_k_Page_083.jp2
a55b59f6ca73068dcb371ec85d11ce0e
53ae0517f35765de209cb118ea1c0f76f464937f
45027 F20101204_AAAWRD dzotsi_k_Page_117.pro
02182cdb353444738f793feb26f2de26
2feb8849d9b8d5865f68cb8a3d8ce56e6d820449
49313 F20101204_AAAWQO dzotsi_k_Page_069.pro
306a7a7e5688411406050d6aa9c0b1f5
5cf08c6d2a5cf61cb8090140f20affc17b65df4c
4274 F20101204_AAAVOB dzotsi_k_Page_062thm.jpg
a90d239fd43532ea51207e0072d1483a
d6d01cdce01aa67da45965705d3f513be38cb953
5628 F20101204_AAAVNM dzotsi_k_Page_100thm.jpg
8582e55b5373f9c48f2be36da8e85878
0eaaeaa1eab2fddd6ef282b7be26c12417d58c7c
19964 F20101204_AAAVMY dzotsi_k_Page_145.jpg
7179cf91566cd425acd7ae0d3d87fc23
5d0433c29c85d42f9621a4a45763066bf8b4e2ac
50347 F20101204_AAAWQP dzotsi_k_Page_070.pro
dcea15bbb806b885b236d92deee633a5
dcaafed93db5c29c34e258708b954b851308ca2f
24640 F20101204_AAAVNN dzotsi_k_Page_146.QC.jpg
a659cc7f9156b3ba7fcfda073b500f73
9a123a3813103c88b0df77a08a660832f47dbf7e
109854 F20101204_AAAVMZ dzotsi_k_Page_053.jp2
6b6fda3f7a435c3cea1a219a54197a39
3034212db9329bd8a9fb362a8d69f4f9f954025d
43246 F20101204_AAAWRE dzotsi_k_Page_118.pro
4251e8e0715f3581a8c85c8eac4a268f
17ea17270c6d675a61f2ca75974dfb5784b24356
47876 F20101204_AAAWQQ dzotsi_k_Page_078.pro
e241e7e85541356a74084b2e4cf1f4c8
49995f0b79c4276a902321ae2d55140d78117835
F20101204_AAAVOC dzotsi_k_Page_130.tif
64d10c6e88f138125624c280d8e2b1d3
f489c951a6a44c67506b6732b108422499c9a932
31433 F20101204_AAAVNO dzotsi_k_Page_108.jpg
ce1ad26fc92dfc752d926c9239e2532e
add2ef68138d1cb44d392ee7bc390077eb4e0829
42450 F20101204_AAAWRF dzotsi_k_Page_119.pro
704ccf3202f81e74b85c63fc4d62ec93
cdff1df371d9fe707edfe6e55d86997436c259a2
54515 F20101204_AAAWQR dzotsi_k_Page_083.pro
28c22f442c030d26f904ecf6846f347a
81609910053a23dc9b4c9ce41e89f057c01b6918
2179 F20101204_AAAVOD dzotsi_k_Page_151.txt
c3ebad17f5665123aea8b2105ba7200d
6ab5de29ffb3578a5879c5f377c37a1e12ab9f59
69826 F20101204_AAAVNP dzotsi_k_Page_159.pro
4313f5801ab15d3aca410bf30eefa86d
e91f9cb822f68b0614d0b8bcdf50eb995e906426
48878 F20101204_AAAWRG dzotsi_k_Page_121.pro
5a77a370de601d5b5f8f7e9068958e32
1e6cd080e5436092ca8fd2552b84bef893b4b508
56628 F20101204_AAAWQS dzotsi_k_Page_086.pro
28805aaf62815c1300ede523bcdf055e
38522ebbf23d4074d0e60d06adb88c3400333dcd
25175 F20101204_AAAVOE dzotsi_k_Page_037.QC.jpg
c0bbb3a989483eb6a324e22d3184e61d
4982822755efb5e967bed958e4a7ff44027a9b10
44315 F20101204_AAAVNQ dzotsi_k_Page_152.pro
d066bb025d4da8f6cdcfe3d9eaa6d206
303a6a2892026359c1e49d4801b40b82c9f3d54d
13972 F20101204_AAAWRH dzotsi_k_Page_127.pro
81b858d9a2b43ae0654e326a19dc607e
4a6d5243a7f94827c07ba691057cae222e5dc7e1
45245 F20101204_AAAWQT dzotsi_k_Page_087.pro
77c077a72c38bbeb52defa8f2527e1ff
d05c410803c0b9b8bfb69d361937be2faa9142ec
F20101204_AAAVOF dzotsi_k_Page_063.tif
b0a43287734e4452ba182387dc68b0ce
72259610bd1e9ae25499943e44c6a777037beef2
6443 F20101204_AAAVNR dzotsi_k_Page_034thm.jpg
9465bad28ba6ec17723bf03c5d85472a
58051477f17992a962d4fe4a7f09a5fa5003c971
38118 F20101204_AAAWRI dzotsi_k_Page_136.pro
8eb871ff9f95f10b8e1d933391f6d261
498a36d6439529035759a79d34bd1b2867089c74
51450 F20101204_AAAWQU dzotsi_k_Page_091.pro
4eeee8198ca4c30bb64cdeb354b76738
7f7351c968c8dd260d21908f7fb0a961012750cf
2036 F20101204_AAAVOG dzotsi_k_Page_016.txt
e0782f3c65ff921005bf0c089853b3ed
bc686ddee2e26ae704ef28e90d7025f28bb5f347
112014 F20101204_AAAVNS dzotsi_k_Page_142.jp2
5d4f458878e685bd30ba70a9ae609564
e3df3d23e7ce0729184d4b08554e744309f54487
54430 F20101204_AAAWRJ dzotsi_k_Page_139.pro
ef09cfe83d74d2e58baf7ca09a7fd811
8ac93e2ba01cff7e9e296429f07f1cca6016572b
55144 F20101204_AAAWQV dzotsi_k_Page_093.pro
83a660c06303a73f1f81cb19af1e7770
36167e1685a837697fe065e48d276d964c6babde
53648 F20101204_AAAVOH dzotsi_k_Page_021.pro
1a444c5d76c61df6f694d9856fda9d76
801fc2d4cdeac4c80bc8a95b7b1ce81bdd91be25
21300 F20101204_AAAVNT dzotsi_k_Page_118.QC.jpg
4eedfae8b24afba00f6b891ebf109472
32f0be81777de84fc50df1a45297a30555462349
28742 F20101204_AAAWRK dzotsi_k_Page_144.pro
feaa67405a9344a8007b3019844b41de
70b253b6a681df08a47becf0dad2d5263d2cce55
14506 F20101204_AAAWQW dzotsi_k_Page_095.pro
bd783f64e34f34f76b8c1102f7e9febd
fb9c34f2bacfd3a1540ab835581f83599bc066d6
F20101204_AAAVOI dzotsi_k_Page_069thm.jpg
053a33a3f70895dbd7126fc811f418aa
6d74d4fcd660a496384e480c438f15e4b59b6a89
872 F20101204_AAAVNU dzotsi_k_Page_062.txt
f8faaf01f391efb36609b2e36cd6a156
661e120aa2f4957f1b13d36caa79f0088fd4e40d
2715 F20101204_AAAWSA dzotsi_k_Page_051.txt
988b5200b4f3d6b2ab19abadc4a508ec
e937475db0ad833f79e05511731cb0658cf3b5a9
51941 F20101204_AAAWRL dzotsi_k_Page_146.pro
3ed02b90466eba33694eba12d2a51342
ad70220e1ac1e37f2e7f86391a4146d11b26af4f
30304 F20101204_AAAWQX dzotsi_k_Page_097.pro
8ef1161d454fc8286a52b1c1da9f7be4
df748d11fe64ca7c8c36089fe2fba161ad96040d
121201 F20101204_AAAVOJ dzotsi_k_Page_124.jp2
5f635c04e52752ff341f736c4423366b
e6225cfd4c5026afc9d485624b862a5b5d07b843
3079 F20101204_AAAVNV dzotsi_k_Page_130.txt
91e3eaf711ec49c646ead2ab32c14625
b645560e3ef46ac812feb9c028fa8d9edde1c7dc
2696 F20101204_AAAWSB dzotsi_k_Page_053.txt
077d2b82ca26f5e1efe02d9ae9cb0911
1d9f9e46fa5897a6a345ea3aa15441b4782489e5
69269 F20101204_AAAWRM dzotsi_k_Page_154.pro
97ae582728e9b2b523832262af4a4411
0d5ca00638b7fb45d3841fb960bc55c42b031ced
42625 F20101204_AAAWQY dzotsi_k_Page_100.pro
c8f35e32a51868bde653bf5bd2f6a2e8
6a71e9712d4982b8f3526e1d12b14561ff66a21e
2738 F20101204_AAAVOK dzotsi_k_Page_153.txt
fb9437ebd219a2309072217e582aa5ee
fec71dc26807583f2a51b157f696c16c3c49882d
F20101204_AAAVNW dzotsi_k_Page_028.tif
b765ca21a9378e1beede3c8bb17291b0
ba28a37d2ea4d8b38ace9b50b698acb945218c67
2288 F20101204_AAAWSC dzotsi_k_Page_054.txt
0f6d52eac98bf5a55ea254c53ef433bb
4e6dba636fa4cbf01b14fe011eb4dd974c5763e3
48967 F20101204_AAAWRN dzotsi_k_Page_161.pro
8c05bb70734aa041032e5f23c093a621
d9bb08a822d05fc54a0cc3928ea8cd45cb658248
37722 F20101204_AAAWQZ dzotsi_k_Page_101.pro
11c7f8b78c695edaad766fd2bb53dc52
44d98f74e0b8c564e21d4a8640f919f99b1a6513
F20101204_AAAVOL dzotsi_k_Page_141.txt
eb7b577292c7f29ffc9b7b65513bf375
0987485929fa2f232a04a42fcbcb8ee22b45817e
111483 F20101204_AAAVNX dzotsi_k_Page_034.jp2
bd32ecd713e0d9204e5cb3771ad3ec32
eec8d4a4fb51f5d3a651a18ae19f33e8f2241ca9
F20101204_AAAVPA dzotsi_k_Page_007.tif
775db87d3005e74d747696ba5bf92d93
c132d12738b551e12a116b047d792c29f3f205e0
2557 F20101204_AAAWSD dzotsi_k_Page_056.txt
8102a0c4adbf011b473fc4536a511865
467efaf81caeb10c9277059cfb5f6472bf0f2a9c
4491 F20101204_AAAWRO dzotsi_k_Page_007.txt
2d10ab16b8139fc6161446e32f33301c
27cd2e52c849ae07511470554aee9c306e74a220
77032 F20101204_AAAVOM dzotsi_k_Page_053.jpg
b813cd04f7b3e584eabcc873737a2042
8b9d0b7a0f48a3f1a3174081d5c495b94bbce7c9
F20101204_AAAVNY dzotsi_k_Page_095.tif
e820125f9afabf1be4fb0af31ed6df8a
ced033ed64acc21314c3db33af53898cdc1ca86a
55143 F20101204_AAAVPB dzotsi_k_Page_022.pro
e2ab39812f27a953e8f392494eaca946
2c0658410613609310e19c02bac67bd03ec51f24
1040 F20101204_AAAWSE dzotsi_k_Page_059.txt
ddab17c2911f4d217f09309f06ae38f2
6021eec2ebc96cca4caa1ec6dcd252437a3a5b15
3223 F20101204_AAAWRP dzotsi_k_Page_010.txt
31eb065aa6231eb8864aeb98ac4fb890
0878de4e7d4cb45ffabdfe7da71b487ef5d9865d
36003 F20101204_AAAVON dzotsi_k_Page_017.pro
ae492a3f1201506018f396fcd6aaa960
9b8e7303a84596537fe5ff4af0c6743f96a04c93
7086 F20101204_AAAVNZ dzotsi_k_Page_083thm.jpg
4541b5cfebb1d6de679b1aee73e9b10f
6fa09bd10bacbc6871ea383fbd58c99d47f5825e
97595 F20101204_AAAVPC dzotsi_k_Page_005.pro
2571cf7b76cbd709daa9f00056b52282
42c75c45a37697eb3e7a9435a95dfb79bfc06aba
1679 F20101204_AAAWRQ dzotsi_k_Page_012.txt
871ca90ea2d030448139783223e86f6a
9c172afa5d5197d7bdbb4e2a8e6377a479d9e680
80974 F20101204_AAAVOO dzotsi_k_Page_124.jpg
58102601f3f3aed0e3040e09671c32a4
e8797b16aadf238e3f5e622b4443589e7d310b59
841 F20101204_AAAWSF dzotsi_k_Page_063.txt
98402e51cb1001bd500108595bece869
00a67886635cef89f75a0b0c3ab7cb97b0a43559
2126 F20101204_AAAWRR dzotsi_k_Page_021.txt
15c21d5aa85a7981b4ece7f141f2baa1
7fad785e9cdbf23785ff4974ffcf12517f51aecb
23553 F20101204_AAAVPD dzotsi_k_Page_001.jpg
af8515b9a176ccdedb73212c9d572df7
c957467e9a2b686602e239ceb94be594ec980bdc
528852 F20101204_AAAVOP dzotsi_k_Page_061.jp2
15d18bff9078e3c436ec0bf2ab0b67c1
a887316ccd788daefd91115c9385bd09fd78cce7
2105 F20101204_AAAWSG dzotsi_k_Page_064.txt
c2c67331594f5d1fb21a80a6f4d373d7
e4f6671f2b59efed3e7ebdddf921c312f5d2ab5e
2164 F20101204_AAAWRS dzotsi_k_Page_024.txt
4aa7bd2b8a809d4a3a82b2d72e8c99b3
cd4222a602176624cc019386cf340d88d870a8b5
67391 F20101204_AAAVPE dzotsi_k_Page_126.jpg
ad8c42d12678509988aafb3dad1e3211
e760df1edadf27928aaff62282f52809fb74d818
F20101204_AAAVOQ dzotsi_k_Page_159.tif
0c653c842625a3a8a9e9b23e5ba55364
fe6bfbe46dc499dadcc4c8b23fcc065ef6bdc354
2212 F20101204_AAAWSH dzotsi_k_Page_065.txt
4d08c43e1f6490a7fff4dedcbaf21b54
5e1fa7c5487888a4d505ff88537686f467e60e17
1171 F20101204_AAAWRT dzotsi_k_Page_026.txt
ae1882ccf26b06ca5a8a0f4315d45642
389d110e349c0c6730abc5fdce8b8b0220979345
24623 F20101204_AAAVPF dzotsi_k_Page_050.jpg
01cc5b37be14524e837b33f4101a7f72
8a0bdc9fa8380e8430d05f017088f0c5225b9802
263 F20101204_AAAVOR dzotsi_k_Page_027.txt
dfb29be8f4fd9b8aa06770efe3e63dd1
4087f44fac0420396a3b219d80b9e8f94e04094f
1812 F20101204_AAAWSI dzotsi_k_Page_067.txt
2fb0b31c1b4cc6cdb89b27a7b4ff6533
09c5b1abde78bdf8d72f2dd8ddd81a46acf12190
2261 F20101204_AAAWRU dzotsi_k_Page_029.txt
dde424496ed1a3b9d50c1ec13f2fe7b5
43be056c6f32cb386e61d48522169a3a4c1fc45a
1802 F20101204_AAAVPG dzotsi_k_Page_101.txt
3321b7f4775b80a8bb0eeb9999350287
b1d719797c2b414e72ecc11086d4da989fbae46b
56233 F20101204_AAAVOS dzotsi_k_Page_051.pro
e523cf131da09790cb6014d6280eef3f
ae9ce2237ce74ffe49a2eebeecc10af42abc0b5c
3022 F20101204_AAAWSJ dzotsi_k_Page_071.txt
4c8668695357da28f03b05330b0c5f83
559745bbd77ed3e0710c301057eb785fcb93dd5a
2127 F20101204_AAAWRV dzotsi_k_Page_032.txt
ded26172b6f585abd606be7f90d354ff
d289ce759ec1b560b11c66de537b1db9f677be08
51123 F20101204_AAAVPH dzotsi_k_Page_103.jp2
af3cf2308be4e05643c1f6d383284c71
e9cc93e4689812029e479b50c9b5f6f36eb88203
25915 F20101204_AAAVOT dzotsi_k_Page_031.QC.jpg
18c6be79c9c1b92f7cada280d78f1550
46c862ae13cd934f9d932eac8604f263aced688c
F20101204_AAAWSK dzotsi_k_Page_075.txt
f3f5ee1cca4cae65b7c0a8c1c3299cfb
6fe0bc5362b204159d3456062011f76046550d5a
1961 F20101204_AAAWRW dzotsi_k_Page_035.txt
dd46e4b844e9a9154ca6cdb80cf8ff49
eaf1d1957b02f03bb2452f5d16f626ec6a2c67cd
21322 F20101204_AAAVPI dzotsi_k_Page_119.QC.jpg
51a90000207fac3e9ad32b906602aed2
90825d9c894330f3ce4f56203c50b53b3943cacd
3185 F20101204_AAAVOU dzotsi_k_Page_014.txt
46fc35b9f7e9aeb54f172fa5c794cc31
f5985a09c05b2d20705f06b4fb2f5c64bcfc50f2
2097 F20101204_AAAWTA dzotsi_k_Page_122.txt
274ff4a16226808af96ef979d8f8a3a4
709fa43219a8bd8b2ed8e016c6a9491cfa83f18a
2167 F20101204_AAAWSL dzotsi_k_Page_076.txt
215329d0c60be8e353b6b1c811074410
c60dfc22ee83ea98c4875044e2feefac1c15fb32
F20101204_AAAWRX dzotsi_k_Page_037.txt
dc9a93aa2e58c8c80ef5122251cd18a8
c54ff216373a60d32f22fc0a3d08826996d88cdd
109911 F20101204_AAAVPJ dzotsi_k_Page_161.jp2
c849c16be1277b4be4d0498c9227e7fe
ba62eb2bf23ed50945539221c0e41b39ee881dac
72663 F20101204_AAAVOV dzotsi_k_Page_068.jpg
29da5622204e183d2923f6de54e16a43
b87f6530e0854d01d7b865e276c525affedf95f2
2011 F20101204_AAAWTB dzotsi_k_Page_123.txt
3180c5c6f6ce98ee633c3f14c74ed696
726df2e006f65d2165b067e1f001527764b66831
F20101204_AAAWSM dzotsi_k_Page_083.txt
b018a88365f96d25a03c68e47cfa3346
34ec9001a1d87e0f5e8cf3256404dfd42bac5efb
2013 F20101204_AAAWRY dzotsi_k_Page_048.txt
3ed9bc2cf66634f59513d04f48615e47
0fa9032ee92e9becc20912d2558af5c81a9c32b9
F20101204_AAAVPK dzotsi_k_Page_057.tif
c4c213991e3e2d7c608d723c24554abb
410c9188165197d8e0b6f7fd5da04c9052942d26
F20101204_AAAVOW dzotsi_k_Page_086.tif
414ee869c9830007deab19665d60329a
aec877cc922b36f8b2b4ff3f1b8f0e6ff343bf00
1977 F20101204_AAAWTC dzotsi_k_Page_125.txt
138edac527802e6918c38693bbd2d0b4
911f267319d9214ad718abec79516b9170e646fa
2090 F20101204_AAAWSN dzotsi_k_Page_085.txt
1742856d870ef3b1a1e4c88e552cf6ba
f5c0cebb3d62786751878e3a0807fca96e0dd43e
2166 F20101204_AAAWRZ dzotsi_k_Page_049.txt
42be51e99dea8cc952f06142e432d50a
a889e4a351ff4f54656e8a0acd69c5e9d00286c2
6661 F20101204_AAAVQA dzotsi_k_Page_003thm.jpg
2209bcf7f403d9e4aec82b6796b28ee7
aa680777a9c066badba5f7be7e6e7c2bdfba2e31
6150 F20101204_AAAVPL dzotsi_k_Page_067thm.jpg
1d611e45e38641fc2db7e14b5c680787
74c3d8f2a86221720bd7d912f8115ffc42d93d2e
6558 F20101204_AAAVOX dzotsi_k_Page_140thm.jpg
7de5e7a781a5a77e18fdef3ffac98d0d
1d4396d354689f07be8906e9a5eea601c5af904c
506 F20101204_AAAWTD dzotsi_k_Page_131.txt
0e06e6fe5866996d95809d94d63b3342
6e66c7b54b1a8b23caa78cee7dae577bdd3cd809
2009 F20101204_AAAWSO dzotsi_k_Page_088.txt
6569cdd57a4a2a5e797ce50487d243d3
997e463d9e1c829f66aa5b9b6289d94cac7119ad
105616 F20101204_AAAVQB dzotsi_k_Page_113.jp2
532474063cd89f8bf640f86386ad56bc
b78431733f02b6cf44a06574f3516627f2fccb73
2024 F20101204_AAAVPM dzotsi_k_Page_089.txt
d52c1ceb1a1f11072431e0fd623950fa
f4f3f05d2cad68394594bbe8451a58482de3f4eb
26479 F20101204_AAAVOY dzotsi_k_Page_038.QC.jpg
e8b1e911283bab570832c18404ca391b
fa68c93c3bfa6cc7a072e99ebf89c4da8d2f4215
487 F20101204_AAAWTE dzotsi_k_Page_135.txt
3921a3c44e148fc7a40a52ceb2698b8b
cb3c73bf041cfa9e1c0ecc3b80cc4782cb2aa484
F20101204_AAAWSP dzotsi_k_Page_093.txt
fbfa0c21894b2be2f370b6df01a2e12b
c684a3bc89abe6537a46ac22d99bec2d6dd751bc
F20101204_AAAVQC dzotsi_k_Page_112.tif
e7357596907abd3ee158f8cd72c668e0
d51da29f8adfa0b0c9e9ea070f006e33eaaf3e1b
79820 F20101204_AAAVPN dzotsi_k_Page_052.jpg
2cdd37199eeb77643e3619d102244424
529da04f245b75e626a134ea16c6d631daa102f6
F20101204_AAAVOZ dzotsi_k_Page_135.tif
f71e23a6c3d5156170a5114a77d87737
6041707eb1461b95b03ea8f85cbaa403f0fd362d
2375 F20101204_AAAWTF dzotsi_k_Page_137.txt
0649ab8d47a4797e20218780223e0717
79b94c2011bc85b710caa6b6075be31eb4d043bc
2174 F20101204_AAAWSQ dzotsi_k_Page_094.txt
c1ec436e941b8e85d26d4225591ff42a
b46a32516b926d93900a13346798278d7bfe4756
6717 F20101204_AAAVQD dzotsi_k_Page_120thm.jpg
99b8dee7aaf5fd43418c1fae13e7185a
4dc8809a3ed588709557a213aeef1632328c2d50
42437 F20101204_AAAVPO dzotsi_k_Page_036.pro
e3a96f7de42f9435f0c080e1ac026854
5eda02936d76b50593b04152dc7a27918b205a37
576 F20101204_AAAWSR dzotsi_k_Page_095.txt
4557446c6c40e9b03547361c4ccb834b
1cc9e165874af6de01fef4e343b99489bac8ef93
6049 F20101204_AAAVPP dzotsi_k_Page_098thm.jpg
1ccc29ff74b65a728cce930b8d8912b7
a5b3b72e530c1314df91c713f2d2cf8f754f96ea
2198 F20101204_AAAWTG dzotsi_k_Page_138.txt
f83c98b2942c0f78794e451d5e220588
74d204aea28b2ceedda7ee82ae45349b89a011a9
1576 F20101204_AAAWSS dzotsi_k_Page_097.txt
65bf0fb6fc72afb6c6f2cbb3f9cebf7e
3eca5410c7ca49a934ea1c8921a654d4cffac3b5
69054 F20101204_AAAVQE dzotsi_k_Page_078.jpg
23dc8861deeb8e184a652cccaaabf156
88a9b256bd811113448be6e587425499c0d006b9
153515 F20101204_AAAVPQ dzotsi_k_Page_046.jp2
f19d3256c90487788d506758f581ae01
e83dcd6ef5a571b96dde98bffc717165aee1f215
2137 F20101204_AAAWTH dzotsi_k_Page_139.txt
ba528ef170864bac45cf41022f7fc377
70c45a6e3f7369ebe75008e4ea4f3ed32ddb1b08
2553 F20101204_AAAWST dzotsi_k_Page_098.txt
5c8b1a12ccf3023cf176a90f550315e8
5066c139123895444e63009e57860da92c64a9a7
1111 F20101204_AAAVQF dzotsi_k_Page_108.txt
e604e69320216c1001ca70e4c60076f7
f6134fa37d08e8d80f083ec6f687461cc1ad3ee2
28133 F20101204_AAAVPR dzotsi_k_Page_019.QC.jpg
94593c448b93f608eaf4d4dcede13ac9
21cb4eb2213362eaa26f999f584a8d8f8b31e86c
771 F20101204_AAAWTI dzotsi_k_Page_143.txt
88f2e6ca3081e38bb85b10a1838378aa
df68654beac8b43a0073515320734b34476b3a07
1184 F20101204_AAAWSU dzotsi_k_Page_105.txt
26c0033f1afeef0d352fed81ff482ad3
a01ce64e8c9be9ab81ec5173bd7d983aa70d522a
7323 F20101204_AAAVQG dzotsi_k_Page_014thm.jpg
b1e8f18382408c60cc900df264fbceed
4afca94ea8e195a9e17e71a586a3180f328d9b24
26511 F20101204_AAAVPS dzotsi_k_Page_095.jpg
47ebfcc1d3056e43e1c56a896a0d8cc8
db2ef7ad98202c55b38f8d01a38a6b80f6323dec
204 F20101204_AAAWTJ dzotsi_k_Page_145.txt
c4fdac80941eab851a7e2bb078c9c239
9d4a2660f6eab2205d23f94d46c93f1a614dab07
1914 F20101204_AAAWSV dzotsi_k_Page_111.txt
37c302920f1972fec698ab8ac794a450
228186ac9052c4e28241af461954387809f7f8d2
F20101204_AAAVQH dzotsi_k_Page_029.tif
8d36f4a6e0a7567feb1995303165a6ad
4d56bb07de9098fa51c33912d8b8b9190f5cab20
115491 F20101204_AAAVPT dzotsi_k_Page_037.jp2
b5fe346ae51a127d33a55104cfa76049
8bc94978c214067decca90588700e12893896a8b
2699 F20101204_AAAWTK dzotsi_k_Page_155.txt
39aa67b5df8a1351cd17e1333ccb659e
5ed41c19080d016ddac8fb4f18e375032a614d90
2061 F20101204_AAAWSW dzotsi_k_Page_114.txt
a89204e78f23e2700e5eeecac15504bb
55363185ef4c513f87b662b3aca3a822d05681e0
F20101204_AAAVQI dzotsi_k_Page_108.tif
1f006370d74af493350b303ed6251e8a
4d6f0fea0db29bd6b9f9e9e2d65b09f213a8a610
6335 F20101204_AAAVPU dzotsi_k_Page_018thm.jpg
41307337cf53777daf478fed25859f68
e87c78546851e0b92b0978b0ca9e29ee1a0726bf
23626 F20101204_AAAWUA dzotsi_k_Page_018.QC.jpg
721a16949f5ffcbc7ef0845ba4da7ebd
50b5f75cf604964f5cd6b35aacda3f307e1be6da
2651 F20101204_AAAWTL dzotsi_k_Page_156.txt
bf5f26fbb5437964c5ba897c7ddcdf32
0b948b8d6c92335c0b63c6bda8241964814300a5
1890 F20101204_AAAWSX dzotsi_k_Page_116.txt
8631d853f0169f9ffd19539cff21ced5
0a8b5cc4546c0fe50cf8418e454f2923dcba0a11
2041 F20101204_AAAVQJ dzotsi_k_Page_091.txt
6a31c4da530cc15a36c126c4f49e950c
82bdbd26e94a0b23978e7d9698c4373f0bea78ed
34328 F20101204_AAAVPV dzotsi_k_Page_150.pro
4d9aca04701dde3c88e60cb9c388ad1a
a6542eb14acfbf6a324e452514dc23065afcc3e5
7004 F20101204_AAAWUB dzotsi_k_Page_020thm.jpg
ec5182bbe2ff44901ae2c113bbb89125
56596305971ec7c267ba26fbf5045d5da71cc918
195 F20101204_AAAWTM dzotsi_k_Page_160.txt
6b86bc15472e0ebcdf7255dbf89006b4
efe7c160db66c63398266f0f96c23d48130616c0
1843 F20101204_AAAWSY dzotsi_k_Page_118.txt
6cedb020d5de369ea20cd1d5e6e9064b
1eef8b18fd100767b13e3d5fa3200bf90fbd002b
2831 F20101204_AAAVQK dzotsi_k_Page_082.txt
7b149f9216f685d024a191ed02661875
3c97c032037dc85a47920f335ca31f51e8db5ae9
F20101204_AAAVPW dzotsi_k_Page_041.tif
9e1f8351f72da02a7fd42dcf73c1f1ba
ce0e22a3b995a6263782c2d1e1a3aced2ecbd696
7092 F20101204_AAAWUC dzotsi_k_Page_021thm.jpg
d92c2ef70f77e8fe48e7cb0f5b814cff
ae22615a131683464c4e2c31f8720e8d5aab79ba
1964 F20101204_AAAWTN dzotsi_k_Page_161.txt
ac51f4a7f7e324f9fbc708d30ae89dba
2d4fd8b73da8fcd6e89fa85708896ed30dc1a467
1951 F20101204_AAAWSZ dzotsi_k_Page_121.txt
71a12f650bdace6b16c8597b652f507f
c8d282365032266d8d9a04f066e078a2523d26e7
70312 F20101204_AAAVQL dzotsi_k_Page_018.jpg
235c23ad731c347f1b876f822c86096b
25cc3caed39435d960962a94929ba8055eda9574
24300 F20101204_AAAVPX dzotsi_k_Page_064.QC.jpg
e29c721ada1e6cf75091501f48abfce0
62dce664224194142c0079a465a642fa06230b6f
14896 F20101204_AAAVRA dzotsi_k_Page_099.QC.jpg
6d323209c64ece2028a07a8b2dfc6e4c
01165356206535987563dd49a4cdcba6066ff20b
25579 F20101204_AAAWUD dzotsi_k_Page_022.QC.jpg
a16cb8ef59a45410634f878ed0b6989c
0ecf448e560f18f38c5574c8a3bb70221865e5b4
2311 F20101204_AAAWTO dzotsi_k_Page_001thm.jpg
e37d1e01a6247b9c1fe482d0c5ff1342
ad51b52268551c8afe25c63a1fd685fea28fd0fb
F20101204_AAAVQM dzotsi_k_Page_133.tif
0e3e6299ad987deb36e6cd006a890bba
09e34d5f203dc54586820ab9929246f2981bd2b6
6055 F20101204_AAAVPY dzotsi_k_Page_036thm.jpg
2a551a3a035dd2452a49925010c0275b
5d517d3437352b23e0eebb88affdbfab79e4ec32
2262 F20101204_AAAVRB dzotsi_k_Page_090.txt
d5dbabc5e7d01f47e78ac03b6c8e6c97
546eac7d5cd6a51ab6135a261f181872dd7f5925
6850 F20101204_AAAWUE dzotsi_k_Page_022thm.jpg
8def8bc22afdbecc86c13f23e43c5b18
8a2239d4fec9e1526b852243109c2180729350c7
731904 F20101204_AAAWTP dzotsi_k.pdf
ac506f9971398c307ba838b77c7bf12c
89d0dd15e4a6cdbe9db7c2522a6348a33cc60e54
101236 F20101204_AAAVQN dzotsi_k_Page_010.jpg
71a765276b5466497846e49fd8198d16
97c679db637bce82a436e5956d0d14d0c5b7fbde
14826 F20101204_AAAVPZ dzotsi_k_Page_144.QC.jpg
c3edbad0d9d5e9720dbc805b2b6b8909
bdde1af7e5d1ac6f164744bc0bd284426a220d55
F20101204_AAAVRC dzotsi_k_Page_156.tif
7639fe41c9d26b966854028248c0b85d
58062540861647f332b69426399c3e724fa1ec64
25348 F20101204_AAAWUF dzotsi_k_Page_023.QC.jpg
d343481d99329901fcb3777acd802322
2ffefe080c943d7dbba7a816680cb5c6d750dd84
24672 F20101204_AAAWTQ dzotsi_k_Page_003.QC.jpg
9b016dfe540641336653cd7609b1f701
0cfefc6dc04712c2f558eca32d9df4bb021f493d
76278 F20101204_AAAVQO dzotsi_k_Page_055.jp2
71147b5c6c7518b44dd9434aa0a204a7
96c7548c9061523c3f1d84efa9951896f4afd731
1793 F20101204_AAAVRD dzotsi_k_Page_119.txt
39f2d1b17235097c86796e6e212a648b
d34f5c497f604f25202123adcf7169434f7cc0b6
4413 F20101204_AAAWUG dzotsi_k_Page_025thm.jpg
72e15818110d4f864b00ee02be5aba8a
325499d4d213dbd9fb1b315caac7568356f091a3
14489 F20101204_AAAWTR dzotsi_k_Page_004.QC.jpg
cd16135c0b3d793ab81ed0b7bb837a21
32d52d144f2df62c8912c4645f224e90f5edd68d
2015 F20101204_AAAVQP dzotsi_k_Page_087.txt
05819123cde6959a8a68483b2d8dbd62
56e716adabb4be32e3d52886ef436207d3d5f6c0
140285 F20101204_AAAVRE dzotsi_k_Page_153.jp2
47e91d08c5e5153d3e5a91c4b74dab9a
374059c7598e2f3fb6fe99f42c6266146b3b962b
4043 F20101204_AAAWTS dzotsi_k_Page_004thm.jpg
49c9e75e196f151ac967875be126e88f
a6e381aa99bfee64017e79f7a145ad5223f2e433
825 F20101204_AAAVQQ dzotsi_k_Page_060.txt
bddbb9519cc963dade2971744723037e
c3607bf334e754af3fa4d9334f5a9c16827169e1
8694 F20101204_AAAWUH dzotsi_k_Page_027.QC.jpg
4ccad9cc287025b3aa3ab2207061720f
68ef740a4ac09b68518d2b54d9670da350cd6baa
6113 F20101204_AAAWTT dzotsi_k_Page_005thm.jpg
628155f38449cfbc147d01698e79ebbd
7b3af34191f5133a5b3f717a2a8fcebedf30c203
2217 F20101204_AAAVQR dzotsi_k_Page_086.txt
caf3d2524684fcde0aac08b45f6e1337
96176029f934720f67827622a01bdff0960b3584
F20101204_AAAVRF dzotsi_k_Page_080.tif
59f3e15085b33fd768adb27fa9757d90
db56518f1edf38fbacf95facd3725ae400fb22c7
24473 F20101204_AAAWUI dzotsi_k_Page_028.QC.jpg
eae13bd42f3be30ea9245bd7cfb9c298
d96b422a2c8db8e2ab173ae2913a2c044fa6822e
23786 F20101204_AAAWTU dzotsi_k_Page_007.QC.jpg
a6dd8c2d104a17b1b593069d61135752
5f4a52edf6c8d10385a80bfea3734644bc1989b1
112174 F20101204_AAAVQS dzotsi_k_Page_048.jp2
1c3f6125f455d5e64098d88ae97cd19b
133ef91b7575242e3d228da916b9f26b979bd490
63777 F20101204_AAAVRG dzotsi_k_Page_087.jpg
4fad3dba176cd4578697382a72124baa
c0a6552de5635d45f316ea738a646227623234da
6688 F20101204_AAAWUJ dzotsi_k_Page_028thm.jpg
2a4833623f8a6a277cc37b49ab8222b8
d61752de52efcb8efb1c5dc8c9bb78f3e1bedd3a
6023 F20101204_AAAWTV dzotsi_k_Page_007thm.jpg
dcdf90be075497aaa1b5ac49ff11e534
cc62c8659719a323b70fa3abf0ccf60c6d51aeff
535 F20101204_AAAVQT dzotsi_k_Page_050.txt
903e4c31248dc19157f14e4be42f9739
d1ccd443a816c6f44e5c24df4651a546eb9d8ded
6600 F20101204_AAAVRH dzotsi_k_Page_116thm.jpg
4d90b791f5ff8f340b7b3a135ad238ae
18dbec0503ce85d477f34f19f865154a9a85182f
26147 F20101204_AAAWUK dzotsi_k_Page_029.QC.jpg
eeda4261352250e4536da2825fc9f468
7ef4285f0d5706ae207494e1bf3a53b63fad1b96
28349 F20101204_AAAWTW dzotsi_k_Page_011.QC.jpg
945c433a90df4d61f6c28d1cb10326dc
ac82c1816e9a3da122017bc89e06e51ce37e09e2
2441 F20101204_AAAVQU dzotsi_k_Page_068.txt
85f4a08efa77cb87b70993a536e744fb
c41a3bc5d3f5f268ff793749552ed10472cd6a24
2079 F20101204_AAAVRI dzotsi_k_Page_066.txt
e6b49a6ff8ca268c19f7ec86786a92e6
0b2b623eaa69dfc20b03bfb182ed76162d681387
23918 F20101204_AAAWVA dzotsi_k_Page_051.QC.jpg
167f39c9b04f744fa1c5aafce7844443
e0b52e75b6c6999a111bf6a4ccfad49ca181ec34
6507 F20101204_AAAWUL dzotsi_k_Page_030thm.jpg
b99ba9b28cebbeea572fa5cdcd48b8a3
61dcc72e54cd47dc9799516c3fb4e23e533f60af
15963 F20101204_AAAWTX dzotsi_k_Page_012.QC.jpg
e7b519cd07d78a0a22c9bced0ecc8555
b4568c6eea65cdc09c4f30ac5437450c71769364
51068 F20101204_AAAVQV dzotsi_k_Page_123.pro
3f24a40d5da68ff097e6f33a1ed516aa
b08d2ce8b0b485a81f27e60afc554c86bb8c7a9f
9589 F20101204_AAAVRJ dzotsi_k_Page_135.pro
2ef12bf2e8973e894a6554078fb7e896
f4eae7ddcb77fad227ca841d407dbdc5adbe8b0f
6341 F20101204_AAAWVB dzotsi_k_Page_051thm.jpg
ff32ba1ada907c61d0e5915024484bd5
7d7f5cb8fcb83bf2bce430fb77d8768521c2dad8
23375 F20101204_AAAWUM dzotsi_k_Page_035.QC.jpg
8ba10ad45a4121fc25bb4d333c957aa5
9345cf967bb104f5937ada5ed53da56b562ae0a1
4361 F20101204_AAAWTY dzotsi_k_Page_012thm.jpg
4886494ee916d379153a62e9048197d4
7db977cf05bd6c548cdcd601e4f12d0ac8f847be
2255 F20101204_AAAVQW dzotsi_k_Page_040.txt
19459877845669f9a5b2ca1dc9c84411
28c1cb470a6a18d8d61a88f031c612f8ef0c0f50
F20101204_AAAVRK dzotsi_k_Page_059.tif
38258a876357816e54f7fd7a1020462a
d76c6558a0114bb16d19719421af94aff0d02462
24471 F20101204_AAAWVC dzotsi_k_Page_052.QC.jpg
d7f8fe0fc4c8482c597156336e895e5e
32f000b01f183c856d0bf5f6c7dd8bfb51132374
6795 F20101204_AAAWUN dzotsi_k_Page_035thm.jpg
2ab8cde3831c4b0eb9bb16424b96ccea
1c5f42f776bf0cc0e40316f2b6d148316082e7ee
18386 F20101204_AAAWTZ dzotsi_k_Page_017.QC.jpg
2d873ef6b775f98baff74400df02283a
f05a5d92545822544bb1379a95c7b9606bc94bed
4278 F20101204_AAAVQX dzotsi_k_Page_136thm.jpg
51fbe03b64ca32b6c31fbfd4ce39ad17
6805e4a8e510f054926dd66924076a45b469811f
3576 F20101204_AAAVSA dzotsi_k_Page_058thm.jpg
4f39148d9fea8c0c440f59b0926be66c
a6abf611d777a763c0b32f1f826372cb0536f96d
F20101204_AAAVRL dzotsi_k_Page_069.tif
fbd8a1b83543774acaa4035dbe43a664
c90ce6d955ffdab77f30c607b43998625c3fb22d
17577 F20101204_AAAWVD dzotsi_k_Page_055.QC.jpg
b70e44a1183b293c5452f0475ccf2511
d5b5336c018769d9dac5bb644063961ec50c22f5
7090 F20101204_AAAWUO dzotsi_k_Page_037thm.jpg
e1d8f2c7f1f65fd1065eae1f525a62ad
90bd22c50e69c832316aeb40e5ed41517eb565e6
F20101204_AAAVQY dzotsi_k_Page_008.tif
bc2ea52644a246c1da63022db4be5c02
25897500963331472197ab238c8a7e4c2732fed0
110860 F20101204_AAAVSB dzotsi_k_Page_051.jp2
640d1c0c07726854cfbe5e916140c39d
a063f71bcf29540ed8b2fb941b3be60d7a1a8420
53294 F20101204_AAAVRM dzotsi_k_Page_122.pro
cf3cd3037c5518f98bc4801ff6645994
fbaca84eeb8297e6ed95f501a1f87b0c123f7aef
8669 F20101204_AAAWVE dzotsi_k_Page_057.QC.jpg
7b8ea63b32629af73e0750fab13a0d71
9b700030f6d21c42f1e870d749dbf37ab66cfd04
7246 F20101204_AAAWUP dzotsi_k_Page_038thm.jpg
b310aa867fad12a707c4922321912d2c
87fd94ac7bae40c7c4835b4ac3d56659ae63a1cc
24079 F20101204_AAAVQZ dzotsi_k_Page_089.QC.jpg
0cc91ddf94a906c3185b867d9fa5d447
c8462dc4c9a8332279cd34eadb3b3e26a011cea2
5308 F20101204_AAAVSC dzotsi_k_Page_055thm.jpg
0332951fa8736c793322c31e85e308da
f67e9de0518c9920974cafd5ba0016ac37ac05f7
10372 F20101204_AAAVRN dzotsi_k_Page_026.QC.jpg
79181079433975954a2c7d3c8a1077bc
910d68429d168a9d69e000d5931e3649ecc85d2c
2809 F20101204_AAAWVF dzotsi_k_Page_057thm.jpg
3ce5e1a64a376f3ee3b8cb9445c9974c
31f126d184ceea522ef0407380f3a6d491156078
6678 F20101204_AAAWUQ dzotsi_k_Page_039thm.jpg
c0c5948da03470160caad9386c8ffb94
e9a90a241fc22509079c9b7644c0fe278708db0e
4485 F20101204_AAAVSD dzotsi_k_Page_015.QC.jpg
111bd09dae44641b0fbfb0e0f5f63c7c
bd3c9c2752cdddbda6f5b60271c16b2177912df1
78351 F20101204_AAAVRO dzotsi_k_Page_021.jpg
d38f3e758c8f90f490759bead2e7e5d3
c365dea820f4adc89076f1b005674d524af1b40c
16381 F20101204_AAAWVG dzotsi_k_Page_059.QC.jpg
24a24496f287792bd7daf13e2d49b012
66d00b331f4d7bc641f7266a8bd2cdb8c580999e
6851 F20101204_AAAWUR dzotsi_k_Page_040thm.jpg
5a2e6e82d3d997b36388c45467aae432
53bb227d053dc59acdb7f9af0dc13212b771eec4
357436 F20101204_AAAVSE dzotsi_k_Page_132.jp2
76d6b62aa77d6c25fe028be18e676a42
858666e4c441a59188179351e07b48f3043ed8da
1847 F20101204_AAAVRP dzotsi_k_Page_079.txt
daf81316211674780c9167a32e8c5ac8
bbe27e8a7820192ccbb2a27c7f23735db3bd657f
14930 F20101204_AAAWVH dzotsi_k_Page_060.QC.jpg
44e96e7ede62e34744b2d31ae76bf062
5aa504c06d62765aae5c6e422d4b1b21e5f26f13
6878 F20101204_AAAWUS dzotsi_k_Page_041thm.jpg
cc51f070ccfc0b2af31193b81fa93236
d66bd41d8ad95cb74a9a00cf3c6cdedb4c009358
F20101204_AAAVSF dzotsi_k_Page_122.tif
c3b9e8d5c3ba0e4af0768179ef82511a
bd6e58adcb45ec5959a27142ac133684f306e2d6
8867 F20101204_AAAVRQ dzotsi_k_Page_095.QC.jpg
d02a4ec42869c16e6296fb13231c126d
582f7830a43507fa5e2cbe99d7338bdf8ee364a0
23896 F20101204_AAAWUT dzotsi_k_Page_044.QC.jpg
5d11b5462b488d968314628658315a80
107055248397f5e98c43212108cd041b7343226b
2253 F20101204_AAAVRR dzotsi_k_Page_124.txt
bf69c0f44756f2aa1946bfd2002d02db
58cbe9605ef0412f5ee828375f4b2c99fff77c04
12775 F20101204_AAAWVI dzotsi_k_Page_062.QC.jpg
ca7c23bde9d8e1cdbcba79e7aa8c1174
1419901f6736a92cb0c855daa7cbad2bcfd8649b
25853 F20101204_AAAWUU dzotsi_k_Page_047.QC.jpg
482eb54ad9b291a232234d24db577704
fd49ccd2923cc75dc4c13aeda05d6643284f98c0
114989 F20101204_AAAVSG dzotsi_k_Page_041.jp2
61bb6f1b5d1deea0a261c04b945289e5
b49c6f3d4b3f756bb40c043127d01d15566a8535
F20101204_AAAVRS dzotsi_k_Page_045.tif
031054fed3ca0b05ef734d0c3028981f
1bc8850a8e462f96021694f66e0b167e657a3f26
14240 F20101204_AAAWVJ dzotsi_k_Page_063.QC.jpg
38039724e3b015fa3418bfd2912eba9d
3b559e6f8e31b761c1135db92fc50ffb3ff05e80
24179 F20101204_AAAWUV dzotsi_k_Page_048.QC.jpg
3121fa64ba51a650c446ec7e91cd9817
6a2b35e07286302dbe9b6d03feb96547683bfe0e
69835 F20101204_AAAVSH dzotsi_k_Page_102.jpg
4bea91b15a18945a5397c3394aac4cb4
172c0e757024e72e44b85f71bb5918981cf6de87
119689 F20101204_AAAVRT dzotsi_k_Page_084.jp2
a10fa6fa45893e37e41604ca6c2afd74
f218f24698dc66cce11f76e643775063fa084444
21615 F20101204_AAAWVK dzotsi_k_Page_069.QC.jpg
8d801ec86b64ff89d3a8af85ca0c8866
a86568abb48d9d2b2def22402458502d4d13eacd
25572 F20101204_AAAWUW dzotsi_k_Page_049.QC.jpg
b0b48de6cb340a7861b4e9ea955d9d76
dc89720d6505e3e5e1fbd9d5d79adb2b58f94d96
1102 F20101204_AAAVSI dzotsi_k_Page_004.txt
836d366b153817c1848f5a3e9af00d77
d871d0ab0654f885010718be4feca4d470a770b2
96516 F20101204_AAAVRU dzotsi_k_Page_007.jpg
f181e2eac6bd8792548d2c15f9ea53bb
c9b176718e282b1a41d84a1c755ff148cad0fc41
7098 F20101204_AAAWWA dzotsi_k_Page_086thm.jpg
2a5c1e73ffe60a8a45225617664d100f
3a383c1ec4759e7a551347793ec3233704a1f3c9
6266 F20101204_AAAWVL dzotsi_k_Page_070thm.jpg
bf8c7b5ac3d88a16a66201fabca04b4f
9012c18b6d91da8a05aef900868538c266fdfc38
6927 F20101204_AAAWUX dzotsi_k_Page_049thm.jpg
7671e094a0b529f045c5c4f886903f30
f5048454fdbdf7fd388e155d3107760dffed4fa0
6764 F20101204_AAAVSJ dzotsi_k_Page_115thm.jpg
9d33a8e33bb2556cc576d7a1d9c0364c
818155949adc03bcfcaae8ab5d7c0524a462ce00
46897 F20101204_AAAVRV dzotsi_k_Page_081.pro
d4108c810e3997032f509c26cf2c8693
07357284abb1bc7d438eeefd7fd3b9533493dc66