<%BANNER%>

Simulating the Regrowth Dynamics of Tifton 85 Bermudagrass as Affected by Nitrogen Fertilization

Permanent Link: http://ufdc.ufl.edu/UFE0022777/00001

Material Information

Title: Simulating the Regrowth Dynamics of Tifton 85 Bermudagrass as Affected by Nitrogen Fertilization
Physical Description: 1 online resource (115 p.)
Language: english
Creator: Alderman, Phillip
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: analysis, bermudagrass, cropgro, cynodon, dactylon, growth, modeling, nitrogen
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Warm-season perennial grass pastures are an important part of beef and dairy production in Florida as a feed source and for effluent management. Thus, understanding the effects of different management strategies on the yield, forage quality and nutrient uptake capabilities of these pastures can have significant effects on both Florida agriculture and environmental quality. However, current best management practices do not take into account changing climatic conditions. Dynamic crop growth simulation models have the potential to provide producers with useful information for modifying their management practices in light of current climate conditions. The CROPGRO-Forage model is one such model, whose generic structure lends itself to adaptation for new species. Currently, species parameters exist only for bahiagrass (Paspalum notatum Fl?gge). Because of the prominence of hybrid bermudagrass Cynodon dactylon (L.) Pers. in improved pastures in Florida, species parameters for modeling bermudagrass growth are needed. Therefore, the objectives of this study were to quantify regrowth dynamics for bermudagrass as affected by nitrogen (N) fertilization and to simulate those dynamics with the CROPGRO-Forage model. To achieve the first objective, established Tifton 85 bermudagrass plots were supplied with four nitrogen (N) rates (0, 45, 90 and 135 kg N/ha/cutting) in 2006 and 2007. The soil was a Pomona sand (sandy, siliceous, hyperthermic Ultic Haplaquod). Dry matter (DM) yield, crude protein concentrations (CP), and in vitro digestible organic matter concentrations (IVDOM) were measured for four 28-day harvests each year from mid-July to mid-October. Regrowth dynamics and canopy photosynthesis were measured for two of the four 28-day regrowth cycles each year. For simulating bermudagrass regrowth, minor adjustments were made to the CROPGRO-Forage model code and data from the field experiment were used for species parameter estimation. Increasing N fertilization increased harvested DM, CP, and IVDOM linearly for all harvests. Quadratic effects on DM yield and CP were present in most harvests indicating a lack of response to N rates beyond 90 kg N/ha/cutting. Nitrogen fertilization effects were reduced in the latest cycle for both years. Total stem and leaf dry matter, tissue N concentration, leaf area index, and canopy photosynthesis increased significantly with N fertilization rate (P < 0.05). Stem and rhizome total nonstructural carbohydrate (TNC) concentrations decreased significantly with increasing N fertilization (P < 0.05) due to increased tissue N concentration and more carbohydrate utilization for shoot growth. However, significant seasonal trends in TNC concentrations were not observed. Simulation of regrowth dynamics was good for dry matter growth in plant components with d-statistic values for most regrowth cycles above 0.80. Canopy photosynthesis simulations were moderately good with d-statistic values for most cycles between 0.30 to 0.56. Simulated N concentration and TNC concentration dynamics were not sufficiently responsive to N fertilization or day of regrowth cycle compared to the noticeable variation in observed N and TNC concentrations. Future model development should incorporate leaf and stem tissue cohorts, which would allow for more mechanistic simulation of N and TNC dynamics as well as providing a foundation for future modeling of forage nutritive value.
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 Phillip Alderman.
Thesis: Thesis (M.S.)--University of Florida, 2008.
Local: Adviser: Boote, Kenneth J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022777:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022777/00001

Material Information

Title: Simulating the Regrowth Dynamics of Tifton 85 Bermudagrass as Affected by Nitrogen Fertilization
Physical Description: 1 online resource (115 p.)
Language: english
Creator: Alderman, Phillip
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: analysis, bermudagrass, cropgro, cynodon, dactylon, growth, modeling, nitrogen
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Warm-season perennial grass pastures are an important part of beef and dairy production in Florida as a feed source and for effluent management. Thus, understanding the effects of different management strategies on the yield, forage quality and nutrient uptake capabilities of these pastures can have significant effects on both Florida agriculture and environmental quality. However, current best management practices do not take into account changing climatic conditions. Dynamic crop growth simulation models have the potential to provide producers with useful information for modifying their management practices in light of current climate conditions. The CROPGRO-Forage model is one such model, whose generic structure lends itself to adaptation for new species. Currently, species parameters exist only for bahiagrass (Paspalum notatum Fl?gge). Because of the prominence of hybrid bermudagrass Cynodon dactylon (L.) Pers. in improved pastures in Florida, species parameters for modeling bermudagrass growth are needed. Therefore, the objectives of this study were to quantify regrowth dynamics for bermudagrass as affected by nitrogen (N) fertilization and to simulate those dynamics with the CROPGRO-Forage model. To achieve the first objective, established Tifton 85 bermudagrass plots were supplied with four nitrogen (N) rates (0, 45, 90 and 135 kg N/ha/cutting) in 2006 and 2007. The soil was a Pomona sand (sandy, siliceous, hyperthermic Ultic Haplaquod). Dry matter (DM) yield, crude protein concentrations (CP), and in vitro digestible organic matter concentrations (IVDOM) were measured for four 28-day harvests each year from mid-July to mid-October. Regrowth dynamics and canopy photosynthesis were measured for two of the four 28-day regrowth cycles each year. For simulating bermudagrass regrowth, minor adjustments were made to the CROPGRO-Forage model code and data from the field experiment were used for species parameter estimation. Increasing N fertilization increased harvested DM, CP, and IVDOM linearly for all harvests. Quadratic effects on DM yield and CP were present in most harvests indicating a lack of response to N rates beyond 90 kg N/ha/cutting. Nitrogen fertilization effects were reduced in the latest cycle for both years. Total stem and leaf dry matter, tissue N concentration, leaf area index, and canopy photosynthesis increased significantly with N fertilization rate (P < 0.05). Stem and rhizome total nonstructural carbohydrate (TNC) concentrations decreased significantly with increasing N fertilization (P < 0.05) due to increased tissue N concentration and more carbohydrate utilization for shoot growth. However, significant seasonal trends in TNC concentrations were not observed. Simulation of regrowth dynamics was good for dry matter growth in plant components with d-statistic values for most regrowth cycles above 0.80. Canopy photosynthesis simulations were moderately good with d-statistic values for most cycles between 0.30 to 0.56. Simulated N concentration and TNC concentration dynamics were not sufficiently responsive to N fertilization or day of regrowth cycle compared to the noticeable variation in observed N and TNC concentrations. Future model development should incorporate leaf and stem tissue cohorts, which would allow for more mechanistic simulation of N and TNC dynamics as well as providing a foundation for future modeling of forage nutritive value.
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 Phillip Alderman.
Thesis: Thesis (M.S.)--University of Florida, 2008.
Local: Adviser: Boote, Kenneth J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022777:00001


This item has the following downloads:


Full Text

PAGE 1

1 SIMULATING THE REGROWTH DYNAM ICS OF TIFTON 85 BERMUDAGRASS AS AFFECTED BY NITROGEN FERTILIZATION By PHILLIP D. ALDERMAN 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 2008

PAGE 2

2 2008 Phillip D. Alderman

PAGE 3

3 ACKNOWLEDGMENTS I would like to first thank Dr. Kenneth Boote (m y committee chair) for introducing me to the process of agronomic rese arch and crop modeling. Our numerous conversations on the appropriate way to explain and model crop grow th phenomena have been truly insightful. I would also like to thank Dr. Lynn Sollenberger for his help in understa nding the thesis-writing process as well as in maintaining a balanced perspective of both pr ofessional and personal priorities. Thanks also to my other committee members, Dr s. Keith Ingram and Adegbola Adesogan, for their helpful suggestions on how to improve the field project and their encouragement to contin ue asking good questions. I also want to express my appreciation to Jason Hupp, with whom I spent many hot summer days in the field. His help with sa mpling and conversations regarding his eclectic hobbies made the field work pass much more easily. Susan Sorrel, Miguel Castillo, Dwight Thomas, and Sid Jones also helped in the lab and in the field. Many hours of hard work were put toward the success of this project. I very much appreciate their help at various points along the way. A special thank you goes out to Dr. Sam Coleman who gave generously of his time and resources in allowing me to scan samples with the NIR equipment at his lab and teaching me NIR theory and methodology. Th e experience gave me a greate r appreciation for the technology in addition to saving me from weeks of lab analyses. I would like to thank Heidi Hebebrand, Jonath an Oliver, and Lauren Alderman for their help processing samples on one or two occasions when I was feeling overwhelmed by it all. Their support along with the pray ers and moral support from other friends and family genuinely made it possible for me to finish this project. I would like to acknowledg e my grandparents Phil and Shirley Weidler who warmly opened their ho me to me on numerous occasions for a quiet place to study and write without distractions. I would also like to express my gratitude to my

PAGE 4

4 parents, Vince and Cynthia Alderman, for their en couragement and example of what it means to love God and share his love with other peopl e. I find their example of perseverance and dedication to their responsibilities to be a c ontinuing source of strength and guidance in living a full and meaningful life. I am forever indebted to my loving and capab le wife, Dani Alderman, who has truly seen me through this endeavor. From helping me processing samples at midnight to accompanying me in the field to words of encouragement spoken at just the right time, I could not have finished this project without her. Her words of wisdom have helped me keep perspective and remember why it is I started it all in the fi rst place. Her prayers, words a nd actions on my behalf have made it all possible. For these and many other reasons, I am continually grateful to have her with me through it all. Finally, I give thanks to God, without whom non e of this would make a difference. I am grateful to Him for bringing hope and meaning to a world that woul d otherwise be utter chaos. His love and grace have sustained me through times of joy and despair. I offer this thesis to Him as an expression of my desire to be a good steward of the talent s and abilities with which He has blessed me. I ask for His help in using what He has given me in loving service to Him and to those around me.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................9 ABSTRACT...................................................................................................................................12 CHAP TER 1 INTRODUCTION..................................................................................................................14 2 LITERATURE REVIEW.......................................................................................................16 Bermudagrass................................................................................................................... ......16 Yield and Nutritive Value...............................................................................................17 Physiology.......................................................................................................................19 Photosynthesis..........................................................................................................19 Regrowth dynamics..................................................................................................21 Dormancy.................................................................................................................24 Modeling Crop Growth and Development.............................................................................26 CROPGRO......................................................................................................................26 CROPGRO-Forage..........................................................................................................27 Parameter Estimation and Model Evaluation.................................................................. 29 3 TIFTON 85 BERMUDAGRASS REGROWTH DYNAMIC S AS AFFECTED BY NITROGEN FERTILIZATION............................................................................................. 32 Introduction................................................................................................................... ..........32 Materials and Methods...........................................................................................................33 Yield and Nutritive Value...............................................................................................34 Regrowth Dynamics........................................................................................................35 Canopy Photosynthesis....................................................................................................37 Results and Discussion......................................................................................................... ..39 Yield and Nutritive Value...............................................................................................39 Regrowth Dynamics........................................................................................................41 Canopy Photosynthesis....................................................................................................47 Conclusions.............................................................................................................................49 4 SIMULATING THE REGROWTH DYNAM IC S OF TIFTON 85 BERMUDAGRASS WITH THE CROPGRO-FORAGE MODEL......................................................................... 72 Introduction................................................................................................................... ..........72 Materials and Methods...........................................................................................................73

PAGE 6

6 Datasets....................................................................................................................... .....73 Set-up of Simulation Experiment.................................................................................... 73 Model Code Adjustments................................................................................................ 74 Parameter Selection......................................................................................................... 75 Dry matter................................................................................................................ 75 Leaf area...................................................................................................................76 Photosynthesis..........................................................................................................76 Carbohydrate dynamics............................................................................................ 77 Nitrogen dynamics...................................................................................................78 Parameter Optimization................................................................................................... 78 Assessment of Model Predictions................................................................................... 79 Results and Discussion......................................................................................................... ..80 Dry Matter.......................................................................................................................80 Leaf Area.........................................................................................................................82 Photosynthesis.................................................................................................................83 Carbohydrate Concentration Dynamics.......................................................................... 83 Nitrogen Concentration Dynamics.................................................................................. 84 Conclusions.............................................................................................................................85 5 SUMMARY AND CONCLUSIONS...................................................................................105 LIST OF REFERENCES.............................................................................................................108 BIOGRAPHICAL SKETCH.......................................................................................................115

PAGE 7

7 LIST OF TABLES Table page 3-1 Probability levels (P) for the effects of year, m onth of harvest, and N fertilization rate (N rate) on harvested dry matter (D M), in vitro digestible organic matter (IVDOM), and crude protein (CP)..................................................................................... 66 3-2 Interaction effects of N rate, harves t month, and year on harvested DM during 2006 and 2007 .............................................................................................................................66 3-3 Interaction effects of N rate, harves t month, and year on crude protein (C P) concentrations of harveste d material during 2006 and 2007............................................. 67 3-4 Interaction effects of N ra te, harvest month, and year on in vitro digestible organic m atter (IVDOM) of harvested material during 2006 and 2007......................................... 67 3-5 Probability levels (P) for the effects of year (Yr), regrowth cycle (Cy), day of regrowth (D ay), and N fertilization rate (N rt) on shoot dry matter (Shoot), leaf dry matter (Leaf), percent leaf (P leaf)), leaf nitrogen concen tration (Leaf N), stem dry matter (Stem), stem nitrogen concentra tion (Stem N), stem total nonstructural carbohydrate concentration (Stem TNC), rhizome dry matter (Rhiz), rhizome nitrogen concentration (R hiz N), and rhizome total nonstructural carbohydrate concentration (Rhiz TNC)..................................................................................................68 3-6 Canopy height of Tifton 85 bermudagrass grown at Gainesville, FL at harvest for four N fertilization rates for three harvest dates in 2007................................................... 69 3-7 Probability levels (P) for the effects of year, m onth of sampling, and N fertilization rate (N rate) on root dry matter (Root DM ), root total nonstructural carbohydrate concentration (Root TNC), and root nitrogen concentration (Root N).............................. 69 3-8 Probability levels (P) for the effects of year (Yr), regrowth cycle (Cy), day of regrowth (D ay), and N fertilization rate (Nrt) on leaf area in dex (LAI), and canopylevel photosynthesis (Cphot).............................................................................................. 70 3-9 Percent light interception of Tifton 85 berm udagrass grown at Gainesville, FL under four N fertilization rates for Cycle 2 of 2006.................................................................... 70 3-10 Effects of N rate on leaf-level photosynthesis by m easurement date................................ 71 4-1 Soil profile characteristics of the Pom ona sand (sandy, siliceous, hyperth ermic Ultic Haplaquod) used for model simulations............................................................................ 97 4-2 Definitions of selected bermudagrass sp ecies p arameters used with the CROPGROForage model................................................................................................................... ..98

PAGE 8

8 4-3 Initial and optimized Tifton 85 bermudagra ss sp ecies parameter values used with the CROPGRO-Forage model............................................................................................... 100 4-4 D-statistic and RMSE values for shoot mass predictions based on optimized berm udagrass parameters for four N rates....................................................................... 102 4-5 D-statistic and RMSE values for leaf m ass predictions based on optimized bermudagrass parameters for four N rates....................................................................... 102 4-6 D-statistic and RMSE values for st em m ass predictions based on optimized bermudagrass parameters for four N rates....................................................................... 102 4-7 D-statistic and RMSE values for canopy photosynthesis predictions based on optim ized bermudagrass parameters for four N rates......................................................103 4-8 D-statistic and RMSE values for stem total nonstructural car bohydrate concentration predictions based on optim ized bermudagr ass parameters for four N rates.................... 103 4-9 D-statistic and RMSE values for rhizom e total nons tructural carbohydrate concentration predictions based on optimized bermudagrass parameters for four N rates.......................................................................................................................... ........103 4-10 D-statistic and RMSE values for leaf N concentration predictions based on optim ized bermudagrass parameters for four N rates....................................................................... 104 4-11 D-statistic and RMSE values for st em N concentration predictions based on optim ized bermudagrass parameters for four N rates......................................................104 4-12 D-statistic and RMSE values for rhiz om e N concentration predictions based on optimized bermudagrass parameters for four N rates......................................................104

PAGE 9

9 LIST OF FIGURES Figure page 3-1 Monthly rainfall at the University of Fl orida Beef Research Unit, Forage Evaluation Field Laboratory for 2006 and 2007 ..................................................................................51 3-2 Monthly average maximum temperature at the U niversity of Florida Beef Research Unit, Forage Evaluation Fiel d Laboratory for 2006 and 2007........................................... 51 3-3 Shoot mass of Tifton 85 bermudagrass grow n at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.........................................................................................52 3-4 Leaf mass of Tifton 85 bermudagrass grow n at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.........................................................................................53 3-5 Percent leaf of Tifton 85 bermudagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.............................................................................54 3-6 Leaf nitrogen (N) concentrations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1..................................................55 3-7 Stem mass of Tifton 85 bermudagrass grow n at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.........................................................................................56 3-8 Stem total nonstructu ral carbohydrate (TNC) concen trations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.............................................................................................................................57 3-9 Stem nitrogen (N) concentrations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1..................................................58 3-10 Rhizome mass of Tifton 85 bermudagrass gr own at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.............................................................................59 3-11 Rhizome total nonstructural carbohydr ate (TNC) concentrations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.............................................................................................................................60

PAGE 10

10 3-12 Rhizome nitrogen (N) concentrations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1............................61 3-13 Root mass, root nitrogen (N) con centration, and root total nonstructural carbohydrates of Tifton 85 berm udagrass gr own at Gainesville, FL for 2006 and 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.................................62 3-14 Leaf area index of Tifton 85 bermudagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.............................................................................63 3-15 Mid-day canopy photosynthesis of Tifton 85 bermudagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.......................................................64 3-16 Canopy photosynthesis versus leaf area index with asym ptotic exponential trend line for 2006 and 2007..............................................................................................................65 4-1 Predicted and observed shoot mass of Ti fton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1..................................................87 4-2 Predicted and observed l eaf m ass of Tifton 85 bermudagr ass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1..................................................88 4-3 Predicted and observed stem mass of Ti fton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1..................................................89 4-4 Predicted and visually estimated leaf area index of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1............................90 4-5 Predicted and observed mid-day canopy photosynthesis of Tifton 85 bermudagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1...............91 4-6 Predicted and observed stem total nons tructural carbohydrate c oncentrations (TNC) of Tifton 85 berm udagrass grown at Gaines ville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.......................................................................................................92 4-7 Predicted and observed rhizome total nonstructural carbohydrate concentrations (TNC) of Ti fton 85 bermudagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.........................................................................................93

PAGE 11

11 4-8 Predicted and observed leaf N concentrations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1............................94 4-9 Predicted and observed stem N concentr ations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1............................95 4-10 Predicted and observed rhizome N concen trations of Tifton 85 berm udagrass grown at Gainesville, FL for Cycle 2 of 2006, Cycle 4 of 2006, Cycle 2 of 2007, and Cycle 4 of 2007 for 0 kg N ha-1, 45 kg N ha-1, 90 kg N ha-1, and 135 kg N ha-1.........................96

PAGE 12

12 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 SIMULATING THE REGROWTH DYNAM ICS OF TIFTON 85 BERMUDAGRASS AS AFFECTED BY NITROGEN FERTILIZATION By Phillip D. Alderman December 2008 Chair: Kenneth J. Boote Major: Interdisciplinary Ecology Warm-season perennial grass pastures are an im portant part of beef and dairy production in Florida as a feed source and for effluent ma nagement. Thus, understanding the effects of different management strategies on the yield, forage quality and nutrient uptake capabilities of these pastures can have significan t effects on both Florid a agriculture and environmental quality. However, current best management practices do not take into account changing climatic conditions. Dynamic crop growth simulation mode ls have the potential to provide producers with useful information for modifying their mana gement practices in light of current climate conditions. The CROPGRO-Forage model is one such model, whose generic structure lends itself to adaptation for new species. Currently, species parameters exist only for bahiagrass (Paspalum notatum Flgge). Because of the prominence of hybrid bermudagrass [Cynodon dactylon (L.) Pers.] in impr oved pastures in Florida, sp ecies parameters for modeling bermudagrass growth are needed. Therefore, the objectives of this study were to quantify regrowth dynamics for bermudagrass as affected by nitrogen (N) fertiliz ation and to simulate those dynamics with the CROPGRO-Forage model. To achieve the first objective, established Tifton 85 bermudagrass plots were supplied with four nitrogen (N) rates (0, 45, 90 and 135 kg N/ha/cutting) in 2006 and 2007. The soil was a

PAGE 13

13 Pomona sand (sandy, siliceous, hyperthermic Ultic Haplaquod). Dry matter (DM) yield, crude protein concentrations (CP), and in vitro digestible organic ma tter concentrations (IVDOM) were measured for four 28-day harvests each year from mid-July to mid-October. Regrowth dynamics and canopy photosynthesis were measured for two of the four 28-day regrowth cycles each year. For simulating bermudagrass regrowth, minor ad justments were made to the CROPGRO-Forage model code and data from the field experiment were used for species parameter estimation. Increasing N fertilization increased harves ted DM, CP, and IVDOM linearly for all harvests. Quadratic effects on DM yield and CP were present in most harvests indicating a lack of response to N rates beyond 90 kg N/ha/cutting. Nitrogen fertilization effects were reduced in the latest cycle for both years. Total stem and leaf dry matter, tissue N concentration, leaf area index, and canopy photosynthesis increased signifi cantly with N fertilization rate (P<0.05). Stem and rhizome total nonstruc tural carbohydrate (TNC) concentrations decreased significantly with increasing N fertilization (P<0.05) due to increased ti ssue N concentration and more carbohydrate utilization for shoot growth. However, signifi cant seasonal trends in TNC concentrations were not observed. Simulation of regrowth dynamics was good for dry matter growth in plant components with d-statistic values for most regrowth cycles above 0.80. Canopy photosynthesis simulations were moderately good with d-stat istic values for most cycles between 0.30 to 0.56. Simulated N concentration and TNC concen tration dynamics were not sufficiently responsive to N fertilization or day of regrowth cycle compared to the no ticeable variation in observed N and TNC concentrations. Future model development should incorporate leaf and stem tissue cohorts, which would allow for more mechanistic simulation of N and TNC dynamics as well as providing a foundation for future mode ling of forage nutritive value.

PAGE 14

14 CHAPTER 1 INTRODUCTION War m-season perennial grasses are an importa nt part of beef and dairy production in Florida. Grazing and harvesting hay provides an important animal feed source. Due to their extensive root systems and excellent nutrient uptake ability, these gras ses are also a vital component of the effluent management strategi es for a number of dairies. Consequently, understanding the effects of different management strategies on their yield, forage quality, and nutrient uptake capabilities can have signifi cant effects on both Florida agriculture and environmental quality. Current recommended best management practices for forage grasses are fairly static with fixed optimal harvest interv als and nutrient applica tion rates (Staples, 1995; Chambliss and Dunavin, 2003). However, pasture growth and nutrient up take are dynamic and are greatly affected by fluctuat ing environmental conditions (W oodard et al., 2002; Mislevy and Martin, 2006). Ideally, BMPs would change wi th these annual and seasonal fluctuations. Crop growth simulation models have the pot ential to provide help ful dynamic decisionsupport for producers. Using su ch models, producers could ta ilor their harvest timing and nutrient management practices to current environmental conditions allowing them to achieve desired yields and forage quality while minimi zing nutrient loss and environmental degradation. Further, combining these models with reliable seasonal climate foreca sts would allow producers to make management decisions early to take a dvantage of favorable c onditions or to mitigate risks associated with unfavorable future environmental conditions. CROPGRO is a process-based dynamic crop growth simulation model whose basic model structure is generic. It si mulates general biological proc esses such as photosynthesis and nitrogen (N) uptake using parameters unique to each species in order to predict crop growth under a variety of conditions. Species parameters governing various aspects of crop growth such

PAGE 15

15 as the timing of growth stages, sensitivity to photoperiod, and partitioning of photosynthate to different plant components are stored in parameter files. Because these parameters are separate from the model itself, CROPGRO can be readil y adapted for new species by estimating these parameter values. Species parameter files currently exist for soybean ( Glycine max L.), peanut ( Arachis hypogea L.), dry bean ( Phaseolus vulgaris L.), faba bean (Vicia faba L.), and tomato ( Lycopersicon esculentum Mill.) (Scholberg et al., 1997; Boote et al., 1998a, 1998b, 2002). CROPGRO-Forage is a modified version of CROPGRO that was developed to better model the growth of tropical perennial grasses. Rymph (2004) incorporated changes in the model to accommodate the C4 photosynthetic pathway, perennati ng organs such as stolons and rhizomes, and photoperiod-induced dorm ancy during winter. Bahiagrass ( Paspalum notatum Flgge) was the first species used for the deve lopment of CROPGRO-Forage. Because of the widespread use of hybrid bermudagrass cultivars [ Cynodon dactylon (L.) Pers.] in improved pastures in Florida, species parameter values for modeling bermudagrass are needed. However, the kind of datasets documenting regrowth and photosynthesis dynamics for forage bermudagrass required for model development are ve ry limited. As a result, field research is needed to document these dynamics for estimating bermudagrass species parameters for the CROPGRO-Forage model. The objectives of this study were to better quantify regrow th dynamics of bermudagrass under varying N fertilization rates and to estimate species parameter values for simulating these dynamics with the CROPGRO-Forage model. Experimental data and a review of relevant scientific literature were used for bermudagrass species parameter estimation and optimization. Once parameter values were estimated, model pred ictions were compared with observed data to assess CROPGRO-Forage simulations of bermudagrass regrowth dynamics.

PAGE 16

16 CHAPTER 2 LITERATURE REVIEW Bermudagrass Berm udagrass [ Cynodon dactylon (L.) Pers.] is a stolonifero us and rhizomatous perennial C4 grass that likely originated in southeast Africa (Harlan et al., 1970; Bogdan, 1977). Wild varieties can grow up to 60-cm tall with leaves ranging from 3to 12-cm long and 2to 4-mm wide (Bogdan, 1977). Tolerance to salinity, flooding, drought, a nd diverse soil conditions has allowed bermudagrass to spread across the globe (Skerman and Riveros, 1990). Harlan and de Wet (1969) refer to it as a ubiquitous, co smopolitan weed because of this worldwide prevalence. The earliest documented case of bermudagra ss introduction to the USA was in the mideighteenth century when Gover nor Henry Ellis brought the speci es to Savannah, GA (Burton and Hanna, 1995). Although lauded by some for its gr eat importance as a pasture grass, most farmers in the southern USA focused their effo rts on eradicating it from their fields. This widely-held perception of bermudagrass as a pest species did not change significantly until the release of Coastal berm udagrass in 1943 (Burton, 1954, c ited by Burton and Hanna, 1995). With higher yields and better quality fo rage than common bermudagrass, Coastal bermudagrass was characterized by larger and longer leaves, stems, and rhizomes, resistance to several foliage diseases, and grea ter tolerance to frost and drought (Burton et al., 1957). Taller growth, combined with greater forage production in the late summer and early fall, made it a good hay crop. Its low viable seed count and less pernicious rhizomes also made it appealing since it was not a risk of becoming a weed (B urton and Hanna, 1995). Since the release of Coastal bermudagrass, a number of other cultivars ha ve been developed for improved yields, higher forage quality, greater resistance to pests and disease, and better tolerance to drought and

PAGE 17

17 frost. Among them are several interspecific hybrids involving bermudagrass and stargrass ( C. nlemfuensis Vanderyst). C. nlemfuensis is similar to bermudagrass in size and growth characteristics but is strictly stoloniferous. Cultivars involving C. nlemfuensis include Tifton 68 bermudagrass, which is actually a cross between two varieties of C. nlemfuensis collected in Kenya (PI 255450 and PI 293606), and Tift on 85 bermudagrass (Bogdon, 1977; Burton and Monson, 1984; Burton, 2001). Tifton 85 berm udagrass is the ster ile pentaploid F1 progeny (2n=5x=45) of a cross betw een Tifton 68 and PI 290884 ( C. dactylon ) from South Africa and is characterized as being taller and darker green, having larger stems, broader leaves, and more rapidly-spreading stolons than other bermudagrass hybrids (Bur ton et al., 1993; Burton, 2001). Some of these characteristics derive from its stargrass parentage. Yield and Nutritive Value In general, Tifton 85 yields higher dry m atter (DM) than previous bermudagrass hybrids (Burton et al., 1993). Reported annu al yields from established Tifton 85 in the southeastern USA range from 14.7 to 31.6 Mg DM ha-1 yr-1 (Burton et al., 1993; Brink et al., 2004; Sistani et al., 2004; Mislevy and Martin, 2006), although yields of up to 39.3 Mg DM ha-1 yr-1 have been reported in fertilized, irrigated field plots in Brazi l (da Fonseca et al., 2007). In a 3-yr clippedplot study, Tifton 85 outyielded Tifton 68 by an average of 3.4 Mg DM ha-1 yr-1 and Coastal by 3.1 Mg DM ha-1 yr-1(Hill et al., 1993). Johnson et al. (2001) showed that DM yield increased on average from 660 to 1810 kg ha-1 cutting-1 in response to an increase of N fertilization from 0 to 157 kg ha-1 cutting-1. This N response is consistent w ith other studies in volving bermudagrass (Prine and Burton, 1956; Beaty et al., 1975; Over man et al.,1993; Brink et al., 2004; Silveira et al., 2007). A number of studies have indi cated that Tifton 85 has consis tently high nutritive value and forage quality. Hill et al. (1993) show ed that Tifton 85 had a relatively high in vitro dry matter

PAGE 18

18 disappearance (IVDMD) of 603 g kg-1, which was higher than Coastal (543 g kg-1) but less than Tifton 68 (636 g kg-1). Other studies have indicated I VDMD values for Tifton 85 from 659 to 487 g kg-1, depending on the age of the harvested mate rial and season of harvest (Hill et al., 1996; Mandebvu et al., 1998a, 1998b, 1999; Johnson et al., 2001; Mislevy and Martin, 2006). Johnson et al. (2001) reported a quadratic response of in vitro organic matter digestibility (IVOMD) to N fertilization for Tifton 85. As N fertilization was increased from 0 to 157 kg ha-1 cutting-1, IVOMD increased on average from 580 to 600 g kg-1. This effect contradicted results from other studies, which showed no significant effect of N fertilization on IVDMD (McCormick, 1974; Harvey et al., 1996; Vendramini et al., 2008). However, the authors gave no explanation about the mechanism fo r this increase in digestibility. Crude protein (CP) concentrations are also relatively high for Tifton 85. Mislevy and Martin (2006) reported CP concentrations from 95 to 222 g kg-1 depending on month of harvest. Several other studies also indicate CP concentrations within that range (Hill et al., 1993; Mandebvu et al., 1998a, 1998b; Sinclair et al., 2003; Brink et al ., 2004). Johnson et al. (2001) showed that CP concentrations in creased on average from 98 to 181 g kg-1 when N fertilization was increased from 0 to 157 kg ha-1 cutting-1. This increase in CP concentration with increasing N rate is consistent with ot her research on bermudagrass (Prine and Burton, 1956; Beaty et al., 1975; Brink et al., 2004; Silveira et al., 2007). High yields combined with high nutritive value give Tifton 85 high forage quality when compared with other improved forages. Hill et al. (1993) reported an av erage daily gain (ADG) of 0.67 kg for steers grazing Tifton 85. Though not significantly different from the ADG of 0.65 kg day-1 for Tifton 78, because of longer DM production, Tifton 85 pastures could be grazed longer resulting in a body weight (BW) gain of 1,160 kg ha-1 versus the 790 kg ha-1 gain

PAGE 19

19 for Tifton 78. Average daily gain for heifer s grazing Tifton 85 pastures was reported as 0.52 kg day-1 while gain ha-1 was reported as 650 kg ha-1 (Pedreira, 1995). Physiology Photosynthesis From a physiological perspective, bermudagra ss is similar to othe r tropical perennial grasses. As a C4 grass, it thrives in high temperatur e and high light environments. This phenomenon is largely due to the CO2-concentrating shuttle of C4 species in which rapid fixation to a C-4 acid by phosphenol pyruvate (PEP) car boxylase in mesophyll cells and subsequent decarboxylation of the C-4 acid in bundle sheath cells saturates the rubisco enzyme by increasing the CO2 concentration in the bundle sheath cells where the rubisc o enzyme is sequestered (Hatch, 1976). This system protects the rubi sco enzyme from the mo re rapid decline in solubility of CO2 than O2 and the increase in O2 competition over CO2 under higher temperatures (Ku and Edwards, 1977). Because of these efficien cies, in general, leaf-level photosynthesis for C4 species is not light-saturated even at full sunlight (Jones, 1985; Moore et al., 2004). However, beyond the inherent effici encies associated with the C4 pathway at the leaf-level, tropical grasses also make optimal use of high lig ht through their canopy stru cture. In general, tropical grasses maintain high leaf area indices at nearly erect leaf angles resulting in light being evenly distributed among leaves within a canopy. This leaf orientation allows each leaf to operate at more optimal light levels allo wing higher canopy photosynthetic rates under full sunlight conditions once the canopy has achieved close to full light in terception (Jones, 1985). For bermudagrass, Burton and Hanna (1993) su ggest that optimal growth occurs when mean daily temperature exceeds 24C and that littl e to no growth occurs below 6 to 9C. The optimal instantaneous temperature for photosynthesis is reported at about 37C, well within the range reported for other C4 pasture species (Ludlow and Wilson, 1971; Bogdan, 1977).

PAGE 20

20 Alexander and McCloud (1962) attempted to qu antify leaf-level and canopy-level carbon exchange rate for Coastal bermudagrass. They reported single leaf net photosynthesis as saturating at about one-third of full s unlight at a rate of about 8 mol CO2 m-2 s-1, which is low for a C4 species, but gas exchange methodology was in its infancy at that time. They also measured canopy-level carbon exchange ra te (CER) ranging from 8 to 15 mol CO2 m-2 s-1 depending on leaf area index (LA I) and age of canopy leaf material However, the exact age of leaf material and the CO2 and temperature conditions for thes e measurements were not clearly stated, which raises questions about what other factors may ha ve contributed to such low photosynthetic rates. In contra st, Hart and Lee (1971) reported single-leaf net photosynthesis for Coastal bermudagrass at approximately 34 to 38 mol CO2 m-2 s-1 for most young leaves at 30C in full sunlight under ambient CO2 conditions, but as high as 50 mol CO2 m-2 s-1 for some leaves. These values are more consis tent with values reported for other C4 grasses, such as bahiagrass ( Paspalum notatum Flgge) (Fritschi, 1996). Howeve r, Hart and Lee (1971) also showed that leaf-level photosynt hetic rates decreased ra pidly with leaf age. Morgan and Brown (1983) reported canopy CER for Coastal bermudagrass cut at weekly and monthly intervals as a function of light intensity and LA I. For the weekly cutting inte rval, they measured apparent canopy CER under full sunlight at about 19 mol CO2 m-2 s-1 for LAI values of 1 to 2, and 41 mol CO2 m-2 s-1 for LAI values of 2 to 3. For the monthly cutting interval, canopy CER under full sun was approximately 32 mol CO2 m-2 s-1 for LAI of 1 to 2, 47 mol CO2 m-2 s-1 for LAI of 2 to 3, 44 mol CO2 m-2 s-1 for LAI of 3 to 4, and 54 mol CO2 m-2 s-1 for LAI greater than 4. The corresponding dark respiration from canopy -root-soil-microorganisms did not change significantly with increasing LAI and ranged from 8 to 12 mol CO2 m-2 s-1.

PAGE 21

21 Photosynthetic response to N fertilization ha s not been reported for bermudagrass. However, increases in leaf-level photosynthetic rate in response to N fertilization have been reported for several C4 species. Several studies have docum ented up to a fourfold increase in photosynthesis with increasing leaf N concentrations for several Panicum species (Wilson, 1975; Bolton and Brown, 1980; Brown and Wilson, 1983; Wilson and Brown, 1983). Similarly, Taub and Lerdau (2000) showed increases in leaf phot osynthesis with increasin g leaf N concentration for six different C4 grasses. Ranjith and Meinzer ( 1997) observed similar trends for two sugarcane cultivars (Sacharum spp. hybrid), but dem onstrated that the degree of response in leaflevel photosynthesis varied significantly, even between genetically similar cultivars. Canopylevel photosynthesis response to N fertilization has not been docum ented directly, though Gastal and Lemaire (2002) cite numerous studies in dicating that N ferti lization affects canopy photosynthesis by increasing leaf N concentrations within the canopy and increasing total leaf area growth. Regrowth dynamics Another aspect of tropical perennial grass physiology is the presence of perennating organs, such as stolons and rhizom es, which a llow new shoots to regrow after injury to or removal of most leaf and stem material. Richards (1993) cites numerous studies indicating that the presence of active meristems is the prim ary factor affecting ra pid refoliation after a defoliation event. The presence of these growing points in stolons and rhizomes at or below the soil surface (i.e., below grazing or cutting) largely determines plant persistence under regular defoliation, from cutting or grazing (Chapman and Lemaire, 1993). Rhizomes and stolons also serve as storage organs that maintain pools of carbohydrates and N upon which the plant can draw during initial stages of regrowth (Richards, 1993). This function should not be overemphasized as many st udies of regrowth have shown that dependence

PAGE 22

22 on stored carbohydrates for recovery from defoliation is usually limited to the first few days of regrowth (Harris, 1978; Humphreys, 1991). Ne vertheless, storage pools are particularly important in cases where the majority of photosynt hetic tissue is removed or has senesced, such as after a severe defoliation ev ent or during regrowth after dorm ancy (Richards, 1993). Sampaio et al. (1976) demonstrated this important role clearly for bahi agrass by removing all leaf tissue daily for 13 wk. Bahiagrass stolons survived on stored carbohydrate rese rves until finally dying after 13 weeks. Because of the importance of these pools under conditions of repeated severe defoliation, total nonstructural ca rbohydrate (TNC) and N concentrations in stolon and rhizome tissue are used as indicators of species persiste nce (Chaparro et al., 1996; Pedreira et al., 2000). Pedreira et al. (2000) evaluated pasture composition, light interception and TNC reserves of grazed Florakirk bermudagrass at different residual stubble heights and after different rest periods. They reported rhizome TNC concentrations from 44 up to 107 g kg-1 OM. Additionally, stubble TNC concentrat ions ranged from 13 to 38 g kg-1 OM. Beaty et al. (1976) reported somewhat lower concentrations for co mbined root and rhizome tissue samples for Coastal bermudagrass. Combined root and rhiz ome tissue N concentrations ranged from 4.8 to 10.2 g kg-1 as applied N increased from 0 to 672 kg ha-1 year-1. Another important factor affecting plant pe rsistence is the degree to which phenotype adapts to regular defoliation (Chapman and Le maire, 1993). Morgan and Brown (1983b) studied the regrowth dynamics of Coastal bermudagrass sw ards subjected to weekly and monthly harvest frequencies. Plots were harves ted to a residual stubble height of 6 cm, and weekly LAI, aboveground biomass, percent leaf, and below-ground biomass were measured. Monthly harvesting resulted in near complete defoliation. Though LAI for the swards increased to 6.6, 6.7, and 10.4 during the three monthly regrowth cycles, post-harvest LAI ranged from 0.9 in June to 0.03 late

PAGE 23

23 in September. Similarly, leaf percentage of post-harvest stubble with monthly harvesting remained between 1 and 2% throughout the growing season. For weekly harvesting, postharvest LAI increased through the season from 0. 9 in June to 2.1 in September, while leaf percentage of post-harvest stubble in general remained within the range of 15 to 20%. Aboveground biomass increased from a residual stubble of about 300 g m-2 to about 1000 g m-2 during regrowth between monthly harv ests. Weekly harvests maintained above ground biomass between 300 and 500 g m-2. Below-ground biomass ranged from 50 to 100 g m-2 for both weekly and monthly harvests throughout the season. Fagundes et al. (1999) investigat ed the effects of maintaining different steady-state sward heights (5, 10, 15, and 20 cm) on light intercep tion, leaf area index, and light extinction coefficient (k) in Tifton 85 bermudagrass. The k value of a canopy indicates the amount of leaf area required for full light inter ception. Lower values indicate more erect leaves in the canopy requiring higher leaf areas for full light intercep tion. Seasonal averages of light interception were 24.4, 68.1, 91.5, and 96.2% for the 5-, 10-, 15-, and 20-cm swards, respectively. Average LAI for these were 0.87, 1.5, 1.91, and 2.80 for the 5, 10-, 15-, and 20-cm swards, respectively. The LAI values for the 15and 20-cm swards are somewhat low for light in terception values of 91.5% and 96.2% especially for a grass canopy, but these low estimates may be due to their methods for measuring leaf area. Because of the small leaf size, it is possible that leaf laminae rolled or curled after being removed, thus causing the measured leaf area to be smaller than the actual leaf area. Average k values for th ese swards were estimated at 0.29, 0.85, 1.29, and 1.38 for the 5-, 10-, 15-, and 20-cm swards, respectiv ely. The latter two k values are unusually high for a grass canopy and might be due to not accounting for stem area light interception and the low estimates of LAI mentioned above. Because swards were grazed instead of clipped in this

PAGE 24

24 study, any comparison between these results and t hose of clipped-plot studies should be made cautiously. However, Morgan and Brown (1983a ) estimated a more reasonable k value for Coastal bermudagrass of 0.62 based on measurements taken over a season at solar noon. They also reported a range of k values from 0.62 to nearly 1.3, estimated based on diurnal variation in solar elevation angle, given the latitude and sola r hour of measurements at the experiment site. Prine and Burton (1956) investigated the effect s of increasing N rate and harvest interval on the yield, CP, and morphologi cal characteristics of Coasta l bermudagrass. Increasing N fertilization from 0 to 1011 kg N ha-1 yr-1 increased the N concentr ation, stem length, plant height, length of longest leaf, l eaf-blades per stem, number of visible internodes per stem, and length of internodes. However, leaf percentage d ecreased on average from 76.3 to 68.9% with increasing N. Beaty et al. (1975) studied the dynamics between partitioning of DM and N to above-ground versus below-ground tissue for bermuda grass. As N fertilization increased from 0 to 672 kg ha-1 yr-1, they reported an increase in the ra tio of below-ground to above-ground DM from 2.9 to 4.1 for common bermudagrass and from 1.7 to 2.1 for Coastal bermudagrass. Above-ground and below-ground N concentrations also increased significantly with increases in N fertilization. For above-ground herbage, N concentrations increased from 7 to 15 g kg-1 for common bermudagrass and from 8 to 14 g kg-1 for Coastal. Below-ground N concentrations increased from 5 to 13 g kg-1 for common bermudagrass and from 5 to 10 g kg-1 for Coastal. Dormancy Dor mancy, usually associated with a decrease or cessation of shoot growth and increased partitioning of carbohydrates to rh izomes and stolons, was strongl y correlated with daylength and solar radiation for Coastal bermudagrass (Burton et al., 1988) Cooler temperatures and lower rainfall also seemed to increase dormanc y, though neither contributed significantly on its own. Studies like this have led some to conclu de that these grasses are photoperiod sensitive and

PAGE 25

25 that dormancy is a daylength effect similar to that observed in determinate crops as they transition from vegetative to reproductive gr owth in response to shortening photoperiod. Though the exact mechanism of this e ffect is still unknown, Sinclair et al. (2001; 2003) demonstrated convincingly that the driving force for dormancy was photoperiod in several tropical forage grasses. They exposed plots of Pensacola bahiagra ss, Tifton 85, Florakirk, and Florona bermudagrasses to extended photoperiod and compar ed the plant partitioni ng and growth trends with plots growing under natural daylength. Ther e were significantly higher DM yields for extended photoperiod as compared to natural ph otoperiod during the winter season despite being exposed to otherwise similar conditions. In general, below-ground mass and N concentration were not significantly different for bermudagrass. However, they observed that bermudagrass plots under the extended photoperiod treatment di d not appear to accumulate as high TNC concentrations in the fall as those with natural daylength leading to sign ificant differences in TNC concentrations between trea tments in the spring. Their conclusion was that photoperiod was the primary factor affecting dor mancy. Sinclair et al. (2004) la ter showed that light intensity was also a contributing factor, but more as an effect on the degree of dor mancy, with higher light intensities reducing dormancy effects. Several studies have indicated varying effects on late-season rhizome TNC and N concentration in response to N fe rtilization in dwarf bermudagrasses used for turf. Schmidt and Blaser (1969) reported a decrease in late -season rhizome carbohydrat e concentrations in response to large N applications for Tifgreen de spite an increase in net photosynthesis. They explain that large N applicati ons stimulated shoot growth, thereby reducing the amount of carbohydrate available for partitioning to rhizomes. They also reported an increase in total plant N concentration in response to large N applica tions. Goatley et al. (1994) also reported a

PAGE 26

26 decrease in late-season rhizome TNC concentrations of Tifgreen with increased N rates for one year. However, there were no significant N e ffects for the subsequent 2-yr of the study. Similarly, Trenholm et al. (1998) reported increases in rhizome and stolon TNC concentrations for FloraDwarf and Tifdwarf in response to different photoperiod treatments, but no significant N effect was found. Modeling Crop Growth and Development CROPGRO The CROPGRO model is a proc ess-based m odel of crop growth and development. It simulates general biological pr ocesses such as photosynthesis and N uptake using parameters unique to each species in order to predict crop growth under a variety of conditions. CROPGRO originated with the developmen t of three separate models: S OYGRO, designed to simulate the growth and development of soybean (Glycine max L.), PNUTGRO, designed to simulate the growth and development of peanut ( Arachis hypogaea L.), and BEANGRO, designed to simulate the growth and development of dry bean ( Phaseolus vulgaris L.). Because these three models simulated basic physiological processes in much the same way, the main code from each was consolidated to form CROPGRO and the parameters describing traits specific to species and cultivar were moved to parameter files (Boot e et al., 1998a, 1998b). Separation of parameter files from the main code makes CROPGRO a ve rsatile model adaptable to new varieties and species upon estimating values for these paramete rs. Because of this flexibility, CROPGRO has been used to simulate additional annual crops, such as faba bean ( Vicia faba L.) and tomato ( Lycopersicon esculentum Mill.) (Scholberg et al ., 1997; Boote et al., 2002). The CROPGRO model code has since been conv erted to a more modular format, with one physiological process per module. These modules have been combined with other modules for crop, weather, and soil processes to form the Cropping Systems Model (CSM) (Jones et al.,

PAGE 27

27 2003). The CSM was incorporated into th e Decision Support System for Agrotechnology Transfer (DSSAT) shell in Version 4. A more complete overview of DSSAT, CSM, and the new modular structure of CROPGRO is found in the documentation for DSSAT v4 (Hoogenboom et al., 2003) and in Jones et al. (2003). CROPGRO-Forage Recently, R ymph (2004) rewrote part of the CRO PGRO model code to better simulate the growth of tropical perennial grasses such as ba hiagrass. The new code includes a module that simulates rhizome and stolon growth, a module th at simulates dormancy, and a change in the module governing plant response to freeze events that allows for pr ogressive freeze damage. In addition, species and cultivar parameters aff ecting photosynthesis, partitioning of DM, C and N remobilization, growth, senescence, and plant ph enology were adjusted to reflect bahiagrass growth and development based on literature va lues and parameter optimization (Rymph et al., 2004). Under the previous version of CROPGRO, phot osynthesis could be simulated either as daily canopy photosynthesis or as hourly leaf-level photosynthesis (Boote et al., 1998a). The daily canopy approach was simply an asympto tic light response of canopy photosynthesis to daily solar radiation. The hourly leaf-level approach is derived from principles described in Boote and Loomis (1991). In essence, canopy photosynthesis is predicted as a function of sunlit versus shaded leaves (based on LAI, leaf a ngle, etc.) as well as parameters for quantum efficiency (QE), light-saturated leaf photosynt hetic rate (under conditions of high leaf N concentration, 30C, and a given specific leaf we ight), and effect of l eaf N concentration. A more detailed description is given in Boote and Pick ering (1994). Previous values for these parameters were derived based on C3 species. As a result, in adapting CROPGRO for tropical grasses, Rymph et al. (2004) had to adju st these parameters to better reflect C4 photosynthesis.

PAGE 28

28 This adaptation included an increase in QE a nd a new photosynthesis response curve to leaf N concentration because photosynthesis of C4 species occurs at lower le af N concentrations than for C3 species. In addition, light-sat urated leaf photosynthetic rate was input based on literature values for bahiagrass. Partitioning of DM and remobilization of C a nd N were also processes that had to be adjusted. Partitioning of DM in the original CROPGRO is a function of vegetative stage (Vstage), indicated by leaf number. Initially, when plants have fe w leaves, those leaves receive greater partitioning, however partitioning to stem increases with V-stage. The addition of a storage organ required additional parameters governing DM partitioning to the storage organ (Rymph, 2004). Related to DM partitioning are the dynamics of C and N remobilization. In CROPGRO, remobilization is primarily from l eaf and stem tissue to support reproductive growth. In CROPGRO-Forage, remobilization ca n occur from old tissue to new leaves. In addition, remobilization can occur from stolons, but this is normally low except when accelerated by low LAI in order to simulate new shoot growth after a severe defoliation event. Both partitioning and remobiliza tion are affected by the new dormancy module. Daylength triggers dormancy based on the findings of Sinclair et al. (2001, 2003) that daylength was the driving factor. Decreasing daylength increases partitioning to and decreases remobilization from the storage organ in the ba hiagrass model (Rymph, 2004). Processes of growth, senescence, and phenol ogical development were kept essentially intact in CROPGRO-Forage. However, minor ch anges were made in the model to accommodate repeated resetting of phenological stage after a harvest. New valu es for the parameters affecting each of these processes were estimated based on data and scientific literature for bahiagrass.

PAGE 29

29 Parameter Estimation and Model Evaluation Param eter estimation is an important part of model development. There are a number of techniques from which to choose. Makowski et al. (2006) described th e relative merits and shortcomings of a number of different possible parameter estimation approaches. Ultimately, the best approach for a given situation often depe nds on the modelers methodological predisposition and the objective of the model under development. For relatively simple models, such as in linear regression, some form of least squares estimation may be adequate. However, when dealing with complex nonlinear models, optimization of parameter values requires more sophisticated techniques. A modeler working with complex models can rarely measure the exact values necessary for adequate model function. As a result, there is an in herent uncertainty about parameter values that must be accounted for. The generalized likelihood uncertainty estimation (GLUE) method is a Bayesian approach that has been shown to be particularly robust in just such situations (Shulz et al., 1999; Makowski et al., 2002; Wang et al ., 2005). A Bayesian approach to parameter estimation treats parameters as random variables each with its own distribution. The objective of the GLUE approach is to estimat e the mean for this distribution based on prior knowledge about the parameters in question, a se t of observed data, and a likelihood function. Prior knowledge comes from previous studies and scientific lite rature and provides information about the possible distribution for each parameter being estimated, called the prior distribution. Based on these distributions, sets of randomly sel ected parameter values can be generated using a random sampling method to form different scenarios to be run in the model. Once these scenarios have been simulated, the likelihood functi on is used to calculate the probability that the values in a given scenario represent the means fo r each parameter. The posterior distribution for each parameter is estimated based on the probability associated with each scenario. The mean of the posterior distribution provides an estimate of the parameter value and the posterior variance

PAGE 30

30 indicates the uncertainty surrounding that value. If done properly, the poste rior distribution will show a significantly reduced uncertainty from that of the prior distribut ion (Bevin and Binley, 1992). In the process of model development, there is also a need for methods of evaluating the product of the process. It is imperative in the evaluation of a model to st ate clearly the intended objectives of the model. The objectives of the model must determine th e accuracy required of the model. With this in mind, Wallach (2006) ex plained the principles behind model evaluation. In order to have an accurate idea of how well a model can predict across all possible scenarios, model evaluation must be done using datasets that were not used to develop the model. The dilemma to modelers is that the more data used for model development, the better the model will be. However, model evaluation also improves as the number and variety of datasets used in evaluation increases. The simplest means of deali ng with this in a situation where datasets are limited is to use some datasets for model de velopment and parameter estimation and leaving others for model evaluation. More complicated evaluation methods, such as cross-validation and bootstrap estimation attempt to make the best of both worlds by estimating statistics such as mean square error of prediction (MSEP) based on changing the portion of the data to be used for model calibration and the portion to be used for model evalua tion (Wallach, 2006). Techniques like these can be particularly useful when data are limited and the models being used are fairly simple. However, they require repeated refitting of model parameters generally making these techniques too cumbersome for use with larger more complex models. Statistics used for model ev aluation range from simply summing the difference between observed and predicted values, to calculating more complicated statistics, such as the concordance correlation coefficient (Lin et al., 2002). Two statistic s that have been used widely

PAGE 31

31 for evaluating model performance are the root mean squared error (RMSE) and the Willmott agreement index (Willmott, 1981; Willmott et al., 1987). Both methods are measures of the difference between simulated and observed valu es. The RMSE is calculated by squaring the difference between the predicted and observed va lues at each point of comparison and then calculating the mean of the squared differe nces, as given by the following equation: N i iiYY N1 2) ( 1 (Eq. 2-1) where N is the number of data points for comparison, iY is the observed value, and iY is the value predicted by the model. A better model prediction will produce a smaller RMSE. The calculation of the Willmott agreement index, or d-index, is presented in Willmott (1981). Though slightly more complicated than RMSE, it incorporates the deviation from the mean of both predicted and observed values. It is calculated using th e following equation: N i iiii N i iiYYYY YY1 2 1 2) ( ) ( (Eq. 2-2) where N is the number of data points, iY is the observed value, iY is the value predicted by the model, and Y is the mean of the observed data. The d-index ranges from 0 to 1 with values near 1 indicating good model predictions. Neither meas ure of accuracy is ad equate on its own, but when used with an understanding of the dynamics of the system being modeled they give a good indication of how well the model is working.

PAGE 32

32 CHAPTER 3 TIFTON 85 BERMUDAGRASS REGROWTH DYNAMICS AS AF FECTED BY NITROGEN FERTILIZATION Introduction Nitrogen (N ) fertilization decisi ons are some of the most impor tant decisions agricultural producers can make. From a stric tly production perspective, excessi ve N fertilization results in a waste of resources. However, because of nitr ate N susceptibility to leaching, excessive N fertilization can cause environmen tal problems. Therefore, unders tanding the effects of different N fertilization rates can make the difference betw een an efficient, environmentally responsible operation and an inefficient operation that leads to environmental degradation. Numerous studies have been conducted demonstr ating significant effect s of N fertilization rate on harvested dry matter (D M) and nutritive value of berm udagrass (Johnson et al., 2001; Brink et al., 2004; da Fonseca et al., 2007; Silveira et al., 2007). Nitrogen fertilization rate has also been shown to affect plant height, percent leaf, leaf lengt h, live leaf number per stem, and visible internode number for forage bermudagr ass (Prine and Burton, 1956). Similar responses have also been observed for hybrid dwarf bermud agrass (Stanford et al., 2005). Increases in leaflevel photosynthesis in response to N fertilization have been well demonstrated for other C4 grass species (Wilson, 1975; Wilson and Brown, 1983; Brown and Wilson, 1983; Bolton and Brown, 1980; Ranjith and Meinzer, 1997). However, neither leaf-level nor ca nopy-level photosynthetic response to N has been reported for berm udagrass. Additionally, though rhizome total nonstructural carbohydrate (T NC) concentration has been shown to decrease in response to increasing N fertilization for dwarf bermudagrass in some studies (Blaser, 1969; Goatley et al., 1994), other studies have not observed similar responses (Trenholm et al., 1998). These studies explored some of the effects of N rate on various aspects of bermudagrass growth, but the precise relationships were not elucidated.

PAGE 33

33 The main objective of this study was to more clearly quantify and e xplain the effects of increasing N fertilization on Tifton 85 bermudagra ss yield, nutritive value, regrowth dynamics and canopy photosynthesis. Based on the previous studies cited above, increasing N fertilization was hypothesized to increase harv ested DM, nutritive value, shoot, stem, and leaf mass, shoot, stem, and leaf growth rate, le af area expansion, stem, leaf, and rhizome N concentration, and canopy photosynthetic rate. Stem and rhizome TNC c oncentrations were expected to decrease with higher N rates. Materials and Methods The experim ent was conducted at the University of Florida Beef Research Unit near Gainesville, FL (29N, 82W) in 2006 and 2007 on a Pomona Sand (sandy, siliceous, hyperthermic Ultic Haplaquod). A 39x 39-m area of established Tifton 85 bermudagrass pasture was separated into four blocks (randomi zed complete block design) of four 6x 6-m plots with 1-m alleys. Each year, an initial applic ation of 500 kg ha-1 of a 10-4-17 (N-P-K) fertilizer was applied to all 16 plots in May. The whole area was then staged by cutting to a 10 cm stubble height in mid-June. At the beginning of each regrowth cycle, treatments of 0, 45, 90, and 135 kg N ha-1 were applied as ammonium nitrate. Following the June staging harvest, harvest samples were taken every 28 d when plot s were cut to a 10-cm stubble height, until the final harvest in mid-October. During the 2nd (July-August) and 4th (September-October.) regrowth cycles, weekly plan t samples and photosynthesis measurements were taken. The experiment was rainfed except for a 38-mm irrigati on application in late April 2007 just prior to fertilizer application and a nother 13-mm irrigation application just following a Cimmeron application in early May 2007 to control bahiagrass ( Paspalum notatum Flgge) that had invaded several plots. Total daily rainfall was measur ed on site and daily solar radiation, and daily

PAGE 34

34 minimum, maximum, and average air temperatur es were taken from the Florida Automated Weather Network (FAWN) database. Yield and Nutritive Value Harvest sam ples were taken from each plot ev ery 28 d by cutting a 1x 4-m strip to 10 cm and collecting the cut forage. The fresh mass of the entire harvest sample was taken on site using a milk scale. An approximately 600 g fresh subsample was taken and weighed using a smaller balance and the remaining harvest sample discarded. After all sub-samples were taken, they were placed in a forced-air drying oven at 60C fo r 48 h. Once dried, each sub-sample was again weighed to determine the percent DM of the harvest sample. This fraction was used to calculate the total DM of harvested he rbage based on the fresh mass of the whole sample. Each subsample was then ground through a 1-mm screen in a Wiley mill and the ground samples were analyzed for in vitro digestible organic matter (IVDOM) us ing a modification of the two-stage technique as described in Moor e and Mott (1974). Crude protein (CP) concentration was also measured using a modification of the aluminum block digestion procedure of Gallaher et al. (1975). A sample weight of 0.25 g was combined with 1.5 g of a 9:1 K2SO4:CuSO4 catalyst and digestion was conducted for at leas t 4 h at 375C using 6 ml of H2SO4 and 2 ml H2O2. Nitrogen in the digestate was determined by semiautomated colorimetry (Hambleton, 1977). Crude protein concentration was calculated as N multiplied by 6.25. Statistical analyses of yield, IVDOM, and CP concentration were performed with a linear mixed effects model using the lme function in R (R Development Core Team, 2006). The model included year, harvest month, N fert ilization rate, and their interac tions as fixed effects. Block and its interactions were random Where interactions were signi ficant, treatment means were presented separately. Orthogonal po lynomial contrasts (linear, qua dratic, cubic) were used to test the type of response to N fertilization.

PAGE 35

35 Regrowth Dynamics Core sam ples of 20x 20x 20-cm were take n by hand from each plot weekly during the 2nd and 4th regrowth cycles. Two shovels and a metal corer were used to cut the sod and remove the soil core. Tiller number was counted and abov e-ground material was cut from each core at the soil level and placed in a sealed plastic bag for transport to the laboratory. Soil was removed from below-ground plant material by rinsing the core with water over a fine screen. The rinsed below-ground material was placed in a separate sealed plastic bag and kept cool for transport to the laboratory. Because of the amount of time a nd labor required for rinsing and collecting root material, this sampling activity was limited to the beginning (Day 0) and end (Day 28) of the cycle. In effect, this amounted to four meas urements at 28 day intervals throughout the season. When root samples were not taken, cores were rinsed adequately to cut and discard the roots and save rhizome and stolon material, which was further rinsed and placed in a bag. Once transported to the laboratory, remaining root material was separate d (and discarded) from rhizomes and stolons. All plant samples were kept refrigerated in the lab until they could be further processed. Plant samples were processed by separating leaf stem, stolon, rhizome, and root tissue. Leaf tissue was defined as emerged leaf blade and was separated at the ligule. Stem tissue included the leaf sheath and any immature, unemerg ed leaves. Stolon and rhizome tissues were considered together as rhizome tissue. Wh en root samples were taken, root tissue was categorized separately from rhizome tissue. Leaf and stem areas (one side) were measured using a LI-COR model 3100 leaf area meter (LI-COR Inc. Lincoln, NE). All fractions were then dried in a forced-air drying oven at 60C for 48 hours. Once dried, the mass of each tissue fraction was measured and the samples were ground through a 1-mm screen in a Wiley mill.

PAGE 36

36 Ground samples (1088 total) were analyzed for N and TNC concentration by near-infrared reflectance (NIR) spectroscopy calibrated with data based on analyzing 10% of the samples for N and TNC using laboratory anal ysis. Spectral data were collected on all samples, with an average of 32 scans per sample, using a NIR Systems 6500 spectrophotometer (Foss Int., Laurel, MD) equipped with a static sample cup. The N concentration of samples used for calibration was determined by digesting samples using a modification of the aluminum block digestion procedure of Gallaher et al. (1975) as described above. Fo r TNC analysis, samples were incubated with invertase and amyloglucosidase to hydrolyze sucrose and starch, respectively. The resultant total hexoses were analyzed by a reducing sugar assay (Chr istiansen et al. 1988). Principle component analysis wa s conducted on the spectral da ta and on a subset selected for calibration using the > Select = procedure of the software InfraSoft In ternational (ISI, State College, PA) based on spectral dissimilarity of samples (Schenk and Westerhaus, 1991a). Reference laboratory data for N and TNC concentra tions were compared with the spectral data for the calibration samples and equations were deve loped with the ISI software using partial least squares regression (Schenk and Westerhaus, 1991b). The N mean, standard error of validation, and r2 for the equation used were: 27.6 g kg-1, 12.8 g kg-1, and 0.92, respectively. The TNC mean, standard error of validation, and r2 for the equation used were: 36.6 g kg-1, 26.1 g kg-1, and 0.98, respectively. These equations were then used to predict N and TNC concentrations for all samples, including those used for the calibration. Statistical analyses of dry mass and the NIR predicted values of N and TNC for each tissue were performed with a linear mixed effects model using the lme function in R (R Development Core Team, 2006). Fixed effects for the model include d year, cycle, day of cycle, N fertilization rate, and their interactions. Bl ock and its interactions were included as random effects.

PAGE 37

37 Interactions including year and cycle were significant so data ar e presented separately by year and cycle. Shoot, stem, and leaf growth rate with in cycles was estimated by linear regressions of shoot, stem, and leaf mass and the resulting slopes were compared for differences between N rates. Because root mass measurements were spaced evenly throughout the season each year, root mass and fraction root were analyzed with a simplified model including year, sample date, N fertilization rate and their interactions as fi xed effects with block and its interactions as random effects. Canopy Photosynthesis Net canopy photosynthesis was m easured appr oximately weekly on two replications during the 2nd and 4th regrowth cycles. Because of the need for a clear sky for accurate readings, weather concerns occasionally re quired measuring replications on separate days or deviating from a precise weekly schedule. For canopy measurements, an open 0.25-L leaf chamber, attached to an LI-COR LI-6200 portable photos ystem (LI-COR Inc., Lincoln, NE), was placed within a larger 0.49-m3 aluminum-frame, clear plastic chamber. A chamber base with 0.56-m2 land area was inserted into the ground in each m easurement site using flat shovels and soil was pressed against the sides of the bottom of the ba se to ensure an adequate seal. Canopy carbon exchange rate (mol CO2 m-2 land s-1) measurements were made at each of five light levels ranging from full sunlight to complete dark. Clot hs of varying thickness we re used to cover the chamber to change light levels of the canopy. Measurements were begun about 10-15 seconds after changing light levels to allow the canopy to adjust to the new light environment. Approximate light levels were: PAR >1500 mol m-2 s-1, 900-1100 mol m-2 s-1, 600-800 mol m-2 s-1, 200-400 mol m-2 s-1, and 0 mol m-2 s-1. Three 16-second measurements were taken at each light level for each plot. The canopy chamber was opened as needed in between measurements to maintain close to ambient levels of humidity and CO2 concentration in the

PAGE 38

38 chamber. Measured carbon exchange rates und er full dark conditions were considered to represent canopy, root, and soil respiration. Gross canopy photosynthesis was calculated by adding the absolute dark respiration to the observed apparent photosynthesis. Leaf-level CER at full sunlight was also m easured for the same replications on which canopy CER was measured. Unfavorable weathe r conditions and other scheduled research activities did not permit taking leaf-level measurements every time canopy measurements were taken. However, when conditions and time were permitting, leaf-level measurements at full sunlight were taken on one fully-emerged leaf from each plot for the same two replicates on which canopy measurements were taken. Canopy light response data were fit to an asymptotic exponential model (Boote et al., 1985): ) 1(*)/*( maxmaxPPARAQE grossePP (Eq.3-1) using the nls function in R (R Development Core Team, 2006), where Pgross is the gross photosynthetic rate (mol CO2 m-2 s-1), Pmax is maximum photosynthetic rate in saturating light (mol CO2 m-2 s-1), AQE is the apparent quantum efficiency, which is the initial slope of the CO2 assimilation:incident PAR response (mol CO2 mol-1 absorbed photons), and PAR is photosynthetically active radiation (mol m-2 s-1). Values for Pmax and AQE were estimated using nonlinear least squares regression and gross canopy photos ynthesis at PAR = 1500 mol m-2 s-1was calculated. The primary purpose of Eq. 3-1 was to calculate gross photosynthetic rates at PAR = 1500 mol m-2 s-1 so that they could be compared across treatments and regrowth periods despite variation in full sunlight levels. Statistical analyses on canopy photosynthetic ra tes were performed with a linear mixed effects model using the lme function of R (R Development Core Team, 2006). The model

PAGE 39

39 included year, cycle, day of cycle, and N fertili zation rate for fixed effects with block and its interactions as random effects. Because leaf level photosynthesis measurements were limited by inclement weather and other research activities, leaf-level photosynthetic rate was analyzed within measurement date using a simplified mode l with N fertilization rate as the only fixed effect and block as a random effect. Results and Discussion Yield and Nutritive Value In general, N fertilization significantly in crea sed yield. In addition, month of harvest was also highly significant (Table 3-1). There were linear and quadratic effects of N application on yield at each harvest in both years except for July 2007 when only the linear effect was significant (Table 3-2). The quadratic effect occurred because there was generally no yield response as N rate increased from 90 to 135 kg ha-1 cutting-1. In several harvests, the cubic effect was significant, generally because the rate of in crease in yield began to slow between 45 and 90 kg N ha-1 cutting-1 and then slowed further between 90 and 135 kg N ha-1 cutting-1. These results are consistent with earlier findings for Tifton 85 bermudagrass (Johnson et al., 2001; Brink et al., 2004; Silveira et al., 2007). Though year di d not have signifi cant effects, year N rate and year month of harvest interactions were significant. In 2006, the highe st yields for all treatments were obtained in July, while in 2007 yields were greater or tended to be greater in August than in July. This was likely the result of several hi gh rainfall events during the period of regrowth leading up to the August 2007 harvest (Fig. 3-1). This may have caused increased leaching of plant-available N from the root zone causing so il N conditions to be similar for plots with applied N. Nutritive value also improved with in creasing N fertilization. Crude protein concentrations showed a significant linear respon se to increasing N rate for all harvest dates

PAGE 40

40 (Table 3-3). Greater increases in CP concentrat ions occurred from increasing the N rate from 0 to 45 kg N ha-1 cutting-1, than from 90 and 135 kg N ha-1 cutting-1. This is consistent with other research findings (Johnson et al., 2001; Brink et al., 2004; Silveira et al. 2007). Because CP is determined directly from tissue N concentration, it follows that increasing N fertilization should increase CP concentration up to some saturation point. Beyond the satu ration point there should be no change even if N rates are increased furthe r. Seasonal effects were also observed with a progressive increase in CP c oncentrations during 2006. For 2007, a more distinct shift in CP concentration was observed between the August and September harvests. Nitrogen fertilization rate effects on IVDOM we re also significant. Increasing N rate increased IVDOM linearly for each harvest in both years (Table 3-4). These results are consistent with results published by Johnson et al. (2001), but contradict other studies finding no effect of N on digestibility (Harvey et al., 1996; Vendramini et al., 2008). Contrary to expectations, seasonal effects on IVDOM were only significant for 2006. In 2006, IVDOM concentrations were higher for later harvests, wh ich is consistent with numerous other studies on Tifton 85 (Hill et al., 1996; Mandebvu et al., 199 8a, 1998b, 1999; Johnson et al., 2001). Mislevy and Martin (2006) explained similar seasonal tren ds for both CP and IVDOM value as primarily due to more rainfall and higher temperatures in middle of the growing season as compared to conditions early and late in the season. Thes e conditions increased the rate of growth and development causing a more rapid increase in plant maturity and decline in nutritive value. This may explain why seasonal effects on IVDOM con centrations were observed in 2006, but not in 2007. Greater rain and higher temperatures la te in 2007 countered normal seasonal weather patterns thereby affecting seasonal trends for IVDOM (Figures 3-1 and 3-2).

PAGE 41

41 Regrowth Dynamics As with yield, N fertilization rate and season of growth had significant effects on shoot mass (Table 3-5). In addition, highly significan t effects were observed for year and day of regrowth cycle. In general, shoot mass wa s increased by N fertilization (Fig. 3-3). The significant N rate by day in teraction (P<0.01) indicates shoot gr owth rate was also significantly increased by greater N fertilization rates ( P <0.05). Linear regressions of shoot mass within each cycle indicated that shoot growth increased from about 50 to 130 kg ha-1 day-1 for Cycle 2, and from about 10 to 60 kg ha-1 day-1 for Cycle 4 with increasing N rate from 0 to 135 kg N ha-1 cutting-1. The significant differences in these growth rates between cycles are shown by the significant Cycle by Day by N rate interaction (P<0.01) in Tabl e 3-5. These differences are likely due to shorter photoperiod, lo wer solar radiation, and lower temperatures in Cycle 4 as compared to Cycle 2 (Burton et al., 1988; Sinclair et al., 2001). Leaf growth appeared to be favored over stem growth initially during cycle 2 of both years with increases in leaf mass within 1 wk after cutt ing (Fig. 3-4). Initial high partitioning to leaf would be expected after a severe defoliation ev ent as the plant refoliates the canopy to support adequate photosynthesis for regr owth (Chapman and Lemaire, 1993) Leaf growth more or less followed a linear trend throughout most cycles, w ith the rate of growth increasing as N rates increased (Fig. 3-4). Somewhat surprisingly, pe rcent leaf was not significantly affected by N rate (Table 3-5), even though da y of regrowth, cycle, year, and their interactions were highly significant. The variation in th e amount of residual leaf mass between trea tments and the rapid increase in leaf percentage for higher N rates caused day by N rate interactions to be significant (P<0.01). Increasing N rate decreased residual leaf mass for both years, though this may not be apparent from Figure 3-4 due to the difference in magnitude between resi dual (Day 0) and final (Day 28) leaf mass. However, percent leaf of post-harvest stubble showed a significant

PAGE 42

42 difference with respect to N rate for both cycles of both years (Fig. 3-5). This is likely a function of a closed canopy and taller shoot growth in the 90 and 135 kg N ha-1 treatments (Table 3-6). As full canopy closure occurred, leaves lower in the canopy senescenced due to self-shading. This coupled with taller shoot growth resulted in most of the leaf tissu e being distributed higher in the canopy. By the end of the regrowth cycle, the majority of the leaf tissue for high N rates was above the cutting height of 10 cm, whereas for lower N rates shoot growth was shorter and more leaf tissue remained below the cutting heig ht. Despite this initial disadvantage, higher growth rates for the high N treatments allowed a rapid recovery of the canopy and within the first week of regrowth leaf percenta ge was similar for all treatments for both cycles of both years (Fig. 3-5). This is supported by the significant day by N rate eff ect (P<0.01) demonstrating that the slope of percent leaf over time was affected by N fertilization rate (Table 3-5). Nitrogen fertilization increased leaf N concen tration fairly uniformly across the regrowth cycle (Fig. 3-6, Table 3-5). The time trend appears to be quadrati c, with leaf N first increasing then decreasing through each cycle, though this is much less pronounced for the 0 kg N ha-1 cutting-1 rate. One would expect leaf N concentra tions to be similar on Day 0 and Day 28 since Day 28 is the same as Day 0 of the next cycle. Leaves remaining after harvesting would be older and growing under low-light conditions, both of which would tend to lower the leaf N concentration. New leaf tissue tends to have a relatively high N concentration, whereas more mature leaf tissue tends to have a lower N con centration. As the canopy grows, new leaf growth represents a smaller percentage of the total leaf mass and the overall leaf N concentration decreases. The smaller time trend for the 0 kg ha-1 rate may relate to differences in the amount of residual leaf tissue describe d above. Because the 0 kg ha-1 rate resulted in a larger amount of residual leaf tissue, the proportion of older leaf tissue carried over into the next cycle is much

PAGE 43

43 greater. As a result an overall increase in leaf N concentration from ne w leaf growth would be reduced by the large percen tage of older leaves. Stem growth appeared to lag behind leaf growth for 1 to 2 wk for most cycles. Stem mass did not increase for the first week for either cycl e in either year, and even appeared to decrease initially for Cycles 2 and 4 of 2006 and Cycle 2 of 2007 (Fig. 3-7). However, the lag in stem mass increase may not indicate a lag in stem growth. Following harvest, part of the residual stem mass was tissue that had been cut by the mo wer, but had not yet senesced. During the initial phase of regrowth it is possible that stem tissues that lacked an active meristem senesced to remobilize resources to new actively growing shoots. The lag in stem mass increase likely represents the period of time over which senescence of old stem tissue was being off-set by new stem tissue growth. This is also consistent with visual observations of senesced stem internode tissue as the regrowth cycle progressed. Nitrogen fertilization signifi cantly decreased (P<0.01) stem TNC concentrations (Table 35, Fig. 3-8). Through each regrowth cycle, stem TNC concentrati on decreased initially and then recovered to the approximate concentration observed for Day 0. These dynamics could support the conclusion that carbohydrat e remobilization does not end until sometime between 7 and 14 days into the cycle (Fig. 3-8). These findings, however, contradict other research indicating that remobilization only occurs for the first 3 days of regrowth (Harris, 1978; Humphreys, 1991). These dynamics might also be explained by carbohydr ate utilization for regrowth. In this case, growth and maintenance respiration would deplet e these TNC reserves until adequate leaf area index can be restored to adequately meet th e carbohydrate demand with da ily photosynthate. At that point, any surplus photosynthate would be stored in stem tissue thereby increasing TNC concentrations. Stem TNC concentrations mi ght also be affected by changes in stem N

PAGE 44

44 concentration. With stem tissue N (crude protei n) increasing, even if the mass of TNC did not change, the TNC concentration would decrease. Thus, this might explain a portion of the observed dynamics. Stem N concentration varied noticeably thr ough the regrowth cycle (Fig. 3-9). Not surprisingly, N fertilization significantly increase d stem N concentration (Table 3-5). For all treatments, there was an initial increase in st em N concentration followed by a gradual decrease as the cycle progressed. Most of these dynamics were likely caused by luxury uptake following N fertilization on Day 0. As DM regrowth occurred with time, this excess N was incorporated into new tissue thereby decreasing the N concentration. Howeve r, luxury uptake of applied N cannot explain the whole of these dynamics because stem N concentrations for the 0 kg N ha-1 cutting-1 treatment also showed an increase and subs equent decrease. Init ial N uptake was more rapid for Cycle 2 than for Cycle 4 in both year s as evidenced by a more distinct spike in N concentration. Significant cycl e and cycle by day effects (P<0.01) confirm this difference in dynamics between Cycles 2 and 4 (Table 3-5). Another contributing factor to the changes in stem N concentration during regrowth was the proportional increase in leaf sheath tissue during the first part of the cycle. Initially, new shoot growth consists mainly of leaf and leaf sheath tissue with very little true stem tissue. Consid ering that leaf sheath tissu e tends to have higher N concentrations than true stem tissue and that leaf sheath tissue was considered stem for the purposes of this study, the increase in leaf sheath tissue during the first part of the regrowth cycle likely contributed to the initial in crease in stem N concentration. Rhizome mass did not show a significant response to N fertilization rate. However, year and cycle effects were both signifi cant and day of cycle approached significance (Table 3-5). In 2006, rhizome mass appeared to incr ease over both early and late regr owth cycles (Fig. 3-10). In

PAGE 45

45 2007, however, it appeared to remain fairly c onstant. An increase in rhizome mass during 2006 would not be unusual since the pasture was establis hed the previous year. However, because of the high variability in the rhizome mass data it is hard to draw any firm conclusions. Surprisingly, rhizome TNC concentration showed no significant seasonal effects that might indicate increased carbo hydrate partitioning to rhizome tissue late in the season (Table 3-5 and Fig. 3-11). Nevertheless, rhizome TNC concentra tions showed significant response to N rate and day of cycle (Table 3-5). As expected, higher N rates generally resulted in lower rhizome TNC concentrations. This phenomenon has been attrib uted to higher TNC utilization for increased shoot growth under higher N fertilization (Goatley et al., 1994). Higher TNC concentrations for low N rates are a result of TNC accumulation because of inadequate available N for shoot growth. Dynamics within all cy cles showed a depletion of rhiz ome TNC during the first half of the cycle and a restoration of TNC levels during the latter half of the cycle. Taken at face value, these trends resembled the trend in stem TN C concentration and appeared to represent remobilization of TNC from rhizomes to new shoots into the second week of regrowth. However, other studies using carbon isotopes have indicated that th is type of remobilization is limited to the first 3 or 4 d after defoliation (H arris, 1978; Humphreys, 1991). Assuming that the observed TNC dynamics do not represent remobilization of carbohydrates or N, a possible explanation might be that while remobilization fr om rhizomes to shoots is minimal, partitioning of daily photosynthate to rhizome tissue may be significantly reduced during canopy expansion as new shoot growth provides a stronger sink fo r photosynthate than existi ng rhizome tissue. If this is the case, normal respiration of rhiz ome tissue may be significant enough to reduce TNC concentrations during the time when little to no daily photosynthate is being partitioned to

PAGE 46

46 rhizomes. This would also explain why the restor ation of TNC levels occurs after the majority of canopy expansion has occurred. Rhizome N concentration did not change very much over the cycle (Fig. 3-12). When it did change, N concentrations increased slightly at the beginning of the cy cle and then decreased through the latter part of the cy cle, the opposite trends observed for rhizome TNC concentrations (Fig. 3-11). The changes in TNC concentrations may have caused the apparent changes in N concentration of the rhizome tissu e. With TNC concentrations decreasing, rhizome tissue with the same N content would have had a higher N con centration despite little to no change in the actual amount of N in the tissue. Neverthele ss, increased N fertilization rate significantly increased rhizome N concentr ation (Fig. 3-12, Table 3-5). Root mass showed highly significan t effects from year and mont h (Table 3-7). N rate was almost significant (P =0.05). In 2006, root mass of three treatments showed an increase from July to September with a subsequent d ecrease to October (Fi g. 3-13). The 90 kg ha-1 treatment did not appear to change much through the season. However, in 2007 the highest root mass for all treatments was in July with the main decreas e of root mass occurring between the July and August sampling dates. This may be due to dry conditions early in 2007 that stimulated more root growth. With adequate rainfall later in the season, unnecessary root tissue likely senesced. The root N and TNC concentrations did not ch ange much through each season (Fig. 3-13). No consistent N rate effects were found for either of these variables. However, root TNC decreased for all treatments between the August and Septem ber sampling dates for both years. The reason for this is not entirely clear. Leaf area index was affected by year, cycle, da y of cycle, and N rate (Table 3-8). Measured LAI values were exceedingly low for grass species. This is especially true given that

PAGE 47

47 high N plots achieved greater than 95% light cap ture by the fourth week of Cycle 2 in 2006 (Table 3-9). The low values are likely due to le af rolling prior to scan ning leaf area. Despite attempts to keep samples cool and in sealed plastic bags to prevent th em from drying out, a large amount of the leaves were at least partially ro lled prior to scanning. Thus, leaf area indices reported here are suspected to be 50% lower than the actual numbers. Ne vertheless, the N effect was significant over each cycle (T able 3-8, Fig. 3-14). Residual LAI after harvest was notably higher for the two low N rates than for the two hi gh N rates in the field. The higher percent leaf of residual stubble mentioned above supports that observation. De spite starting regrowth with very little residual leaf area, th e two high N treatments were able to rapidly refoliate after cutting, whereas lower N treatments had a slower canopy r ecovery. In general, each treatment followed a logistic trend and appeared to saturate around its own ma ximum LAI, with high N rates achieving the highest LAI. Presumably for the high N rates, this maximum LAI represents the leaf area needed for full canopy light capture. Se nescence due to self-shading and reduced leaf growth likely contributed to maintaining LAI at these levels. For the 0 kg N ha-1 rate, maximum LAI levels were probably related to slower growth and higher rate s of leaf senescence due to remobilization of N from older leaves to younge r leaves higher in the canopy. Despite not having full canopy interception for low N rates, the N status of the soil was probably too low for the N demand for new leaf growth to be met by root N uptake. Plants were required to remobilize N from older tissue in order to sustai n growth. Visual observa tions about the degree of leaf senescence as well as the low leaf N concentrations for the 0 kg N ha-1 treatment discussed above also support this explanation (Fig. 3-6). Canopy Photosynthesis Canopy photosynthesis had significa nt year, cycle, day of regrowth and N rate effects (Table 3-8). Photosynthetic rates ap peared to follow a logistic cu rve within each cycle with the

PAGE 48

48 plateau varying with N rate and cycle (Fig. 3-15 ). Increasing N fertilization rate up to 90 kg N ha-1 cutting-1 resulted in a higher photosynthetic rate, a nd Cycle 2 in both years had a higher rate than Cycle 4. Leaf mass correlated fairly we ll with photosynthesis over the first two weeks of growth. Linear models of photosynthetic rate as a function of leaf mass showed r2 values between 0.81 and 0.89 depending on the cycle and year for the first tw o weeks of growth. However, increases in leaf mass through the remainder of the cycle did not show any effect on canopy photosynthesis, which appear ed to plateau by the second week of regrowth for all N rates. While N fertilization rate showed si gnificant effects on canopy photosynthesis, leaf N concentration did not. For the most part photosynthesis followed the same logistic trend as LAI. Because photosynthesis is a function of light in terception it follows that under full sunlight conditions, increasing LAI would correlate with increasing photosynthesis up to an LAI level that approaches full light capture (Morgan and Brown, 1983a). An asymptotic exponential regression of canopy photosynthesis as a response of LAI showed an r2 of 0.75 indicating that LAI was likely to be at least an important factor in determ ining canopy photosynthesis (Fig. 316). Leaf-level photosynthesis only had significant N rate effects on two dates in 2006 and one date in 2007 (Table 3-10). The mean rates for le af-level photosynthesis are in the range expected and averaged 35, 36, 38, and 41 mol m-2 s-1 for the 0, 45, 90, and 135 kg N ha-1 cutting-1 rates, respectively. On the dates where N rate effects we re significant, the trends were not consistent. This lack of consistency cont rasts with the clear N rate effects observed for canopy-level photosynthesis. A possible reason why more clea r dynamics were not observed may have been because the number of leaves measured on each day was too small. However, given the other research activities going on, this was mostly unavo idable. In any case, le af-level photosynthesis

PAGE 49

49 measurements alone could not have fully refl ected the dynamics of canopy-level photosynthesis because they did not capture the co ntribution of total leaf area index. Conclusions The m ain objective of this study was to more clearly quantify and e xplain the effects of increasing N fertilization on Tifton 85 bermudagra ss yield, nutritive value, regrowth dynamics and canopy photosynthesis. Increas ing N fertilization increased harvested DM, CP, and IVDOM linearly for all harvests. Quadratic effects on DM yield and CP we re present in most harvests indicating a lack of response to N rates above 90 kg N ha-1 cutting-1. The exact mechanisms for N effects on IVDOM remain unclear. The signif icant differences in the degree of N effects between years indicates that there are other conf ounding factors that still need to be accounted for. As expected, leaf, stem, and overall shoot grow th were favored by increasing the N rate. Unexpectedly, partitioning between leaf and stem did not appear to be affected significantly beyond the effects of N rate on the amount of resi dual leaf tissue after ha rvest. Though percent leaf was hypothesized to increase with higher N a pplications, no significant differences in leaf percentage were observed by the end of each regrow th cycle. Percent leaf in the residual stubble mass and the initial rate at which percent leaf increased were affected by N fertilization rate. Leaf area expansion was enhanced by higher N fertilization allowing gr eater and more rapid increases in leaf area after defo liation. Leaf, stem, and rhizome N concentrations were increased by N fertilization. Stem and rhizome TNC c oncentrations decreased with increasing N fertilization as was expected, due to more carbohydrate utiliza tion for shoot growth and to greater N concentration. Cycle e ffects on stem and rhizome TNC were also evident with minima between 7 and 14 days after harvest. However, contrary to expectations no significant seasonal trends in rhizome TNC concentrations were observed. Canopy photosynthetic rate was increased

PAGE 50

50 by N fertilization. This effect appeared to be pr imarily due to greater leaf area expansion under higher N treatments rather than by incr eases in leaf-level photosynthesis. Of practical significance, fertilizi ng at rates higher than 90 kg N ha-1 cutting-1 are unlikely to increase yield or nutritive value enough to warr ant the added expense. Additionally, fertilizer applications after August may not be worthwhile as seasonal effects would largely minimize any potential benefit. Further, the use of resources such as seasonal climate forecasts may also help producers determine whether trends in the current season are likely to enhance or counteract the effects of applying a higher N rate.

PAGE 51

51 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 50 100 150 200 250 2006 2007Monthly Rainfall (mm) Figure 3-1. Monthly rainfall at the University of Florida Beef Re search Unit, Forage Evaluation Field Laboratory for 2006 and 2007. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 15 20 25 30 35 40 2006 2007Average Maximum Temperature (C) Figure 3-2. Monthly average maximu m temperature at the University of Florida Beef Research Unit, Forage Evaluation Fiel d Laboratory for 2006 and 2007.

PAGE 52

52 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1) 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1) 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1) 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1)(A) (B) (C) (D) Figure 3-3. Shoot mass of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic, C = cubic. L L,Q L,Q Q Q L,Q L,Q L C L L L Q Q

PAGE 53

53 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1) 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1) 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1) 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1)(A) (B) (C) (D) Figure 3-4. Leaf mass of Tifton 85 bermudagrass gr own at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic. L,Q L L L,Q L L L L,Q L L L,Q L,Q L L Q Q

PAGE 54

54 07142128 0 20 40 60 80 100 Day of RegrowthPercent Leaf 07142128 0 20 40 60 80 100 Day of RegrowthPercent Leaf 07142128 0 20 40 60 80 100 Day of RegrowthPercent Leaf 07142128 0 20 40 60 80 100 Day of RegrowthPercent Leaf Figure 3-5. Percent leaf of Tift on 85 bermudagrass grown at Gaines ville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic. Q Q L,Q L,Q L L L,Q L L L L L

PAGE 55

55 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1) 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1) 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1) 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1)(A) (B) (C) (D) Figure 3-6. Leaf nitrogen (N) conc entrations of Tifton 85 bermudagr ass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = lin ear, Q = quadratic, C = cubic. L L,Q,C L,Q L L L L,Q L,Q L,Q L,Q,C L L,Q L,Q L,Q L L,Q L,Q L L,Q L,Q

PAGE 56

56 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1) 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1) 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1) 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1)(A) (B) (C) (D) Figure 3-7. Stem mass of Tifton 85 bermudagrass gr own at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic. Q L,Q L,Q L,Q Q Q L,Q L Q Q

PAGE 57

57 07142128 0 25 50 75 100 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1) 07142128 0 25 50 75 100 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1) 07142128 0 25 50 75 100 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1) 07142128 0 25 50 75 100 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1)(A) (B) (C) (D) Figure 3-8. Stem total nonstructural carbohydr ate (TNC) concentrations of Tifton 85 bermudagrass grown at Gainesville, FL fo r (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic, C = cubic. L L L,Q L L L,C L,Q L L L L L L L L,Q

PAGE 58

58 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1) 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1) 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1) 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1)(A) (B) (C) (D) Figure 3-9. Stem nitrogen (N) concentrations of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial c ontrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic, C = cubic. L L L,Q L,Q L L L L L L L,Q L L L L L,Q,C L,Q L L L,Q,C

PAGE 59

59 07142128 0 2500 5000 7500 10000 Day of RegrowthShoot Mass (g m-2)R h i z o m e M a s s ( k g h a1) 07142128 0 2500 5000 7500 10000 Day of RegrowthShoot Mass (g m-2)R h i z o m e M a s s ( k g h a1) 07142128 0 2500 5000 7500 10000 Day of RegrowthShoot Mass (g m-2)R h i z o m e M a s s ( k g h a1) 07142128 0 2500 5000 7500 10000 Day of RegrowthShoot Mass (g m-2)R h i z o m e M a s s ( k g h a1)(A) (B) (C) (D) Figure 3-10. Rhizome mass of Tifton 85 bermudagra ss grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic, C = cubic. Q L,Q C C Q

PAGE 60

60 07142128 0 50 100 150 200 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1) 07142128 0 50 100 150 200 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1) 07142128 0 50 100 150 200 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1) 07142128 0 50 100 150 200 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1)(A) (B) (C) (D) Figure 3-11. Rhizome total nonstructural car bohydrate (TNC) concentrations of Tifton 85 bermudagrass grown at Gainesville, FL fo r (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic, C = cubic. L L L L L L L L L L L,Q L L L,Q L L L L L,Q L,Q

PAGE 61

61 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1) 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1) 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1) 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1)(A) (B) (C) (D) Figure 3-12. Rhizome nitrogen (N) concentrations of Tift on 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrow th: L = linear, Q = quadratic, C = cubic. L L L L L L L L L L L L L L L,Q L L,Q L,Q L,C L

PAGE 62

62 Jul Aug Sep Oct 0 1500 3000 4500 6000 Shoot Mass (g m-2)R o o t M a s s ( k g h a1) Jul Aug Sep Oct 0 1500 3000 4500 6000 Shoot Mass (g m-2)R o o t M a s s ( k g h a1)(A) (B) Jul Aug Sep Oct 0 5 10 15 20 Shoot Mass (g m-2)R o o t N ( g k g1) Jul Aug Sep Oct 0 5 10 15 20 Shoot Mass (g m-2)R o o t N ( g k g1)(C) (D) Jul Aug Sep Oct 0 25 50 75 100 Shoot Mass (g m-2)R o o t T N C ( g k g1) Jul Aug Sep Oct 0 25 50 75 100 Shoot Mass (g m-2)R o o t T N C ( g k g1)(E) (F) Figure 3-13. Root mass for (A) 2006 and (B) 2007, root nitrogen (N) concentration for (C) 2006 and (D) 2007, and root total nonstructura l carbohydrates for (E) 2006 and (F) 2007 for Tifton 85 bermudagrass grown at Gainesville, FL for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic, C = cubic. L Q Q L C L

PAGE 63

63 07142128 0 0.5 1 1.5 2 2.5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x 07142128 0 0.5 1 1.5 2 2.5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x 07142128 0 0.5 1 1.5 2 2.5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x 07142128 0 0.5 1 1.5 2 2.5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x Figure 3-14. Leaf area index of Tifton 85 bermudag rass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic. L L,Q L,Q Q L L L L L,C L,Q L L,Q L Q

PAGE 64

64 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1)(A) (B) (C) (D) Figure 3-15. Canopy photosynthesis of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ), 45 kg N ha-1 ( ), 90 kg N ha-1 (), and 135 kg N ha-1 (). Letters represent significant (P<0.05) polynomial contrasts of N fertilization rates within day of regrowth: L = linear, Q = quadratic. L L L L L L L L,Q L,Q L L,Q

PAGE 65

65 0123 0 20 40 60 80 C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1)L e a f A r e a I n d e x Figure 3-16. Canopy photosynthesis ve rsus leaf area index for Tifton 85 bermudagrass grown at Gainesville, FL during 2006 and 2007 with an asymptototic expone ntial trend line.

PAGE 66

66 Table 3-1. Probability levels (P) fo r the effects of year, month of harvest, and N fertilization rate (N rate) on harvested dry ma tter (DM), in vitro digestib le organic matter (IVDOM), and crude protein (CP). Source of variationHarvested DMIVDOMCP Year 0.500.390.57 Month <0.01<0.01<0.01 N rate <0.01<0.01<0.01 Year Month <0.01<0.01<0.01 Year N rate 0.020.83<0.05 Month N rate <0.010.14<0.01 Year Month Nitrogen0.600.05<0.01 Table 3-2. Interaction effects of N rate, harvest month, and year on harvested DM during 2006 and 2007 04590135 kg ha-12006 Jul 1130a2940a3640a4430aL**,Q**,C170 Aug 1040a2530ab3280 b 3140 b L**,Q**230 Sep 980a1770 b 2800c3010 b L**,Q**140 Oct 220 b 600c990d970cL**,Q**,C60 337078401071011550L**,Q**390 2007 Jul 1130a2430a2920a3720aL**380 Aug 1390 b 3510 b 3450a3710aL**,Q**,C*280 Sep 770c1620c2080 b 2040 b L**,Q**110 Oct 620c1360c1200c1060cL**,Q**,C*100 39108920965010530L**,Q**,C*640 kg ha-1Season Season Harvest date Nitrogen (kg ha-1 cutting-1) Polynomial contrast SE Comparison of month means within an N rate a nd year. Means in same column and year with same letter are not significantly different (P>0.05). Orthogonal polynomial contrast of Nitrogen effect within harvest date. L=linear, Q=quadratic, C=cubic; letter with no symbol (P 0.1); *(P 0.05); **(P 0.01).

PAGE 67

67 Table 3-3. Interaction effect s of N rate, harvest month, a nd year on crude protein (CP) concentrations of harveste d material during 2006 and 2007 04590135 g kg-12006 Jul 98a105a128a140aL**7 Aug 88a110a137a164 b L**5 Sep 113 b 132 b 175 b 188cL**,Q*6 Oct 100ab162c216c221dL**,Q**,C*5 2007 Jul 97a107a133a146aL**,C4 Aug 94a105a139a149aL**,C8 Sep 116 b 144 b 188 b 200 b L**,Q*,C**4 Oct 121 b 144 b 184 b 189 b L**,Q*,C**4 g kg-1Harvest date Nitrogen (kg ha-1 cutting-1) Polynomial contrast SE Comparison of month means within an N rate a nd year. Means in same column and year with same letter are not significantly different (P>0.05). Orthogonal polynomial contrast of nitrogen effect within harvest date. L=linear, Q=quadratic, C=cubic; letter wi th no symbol (P 0.1); *(P 0.05);**(P 0.01). Table 3-4. Interaction effects of N rate, harvest month, and year on in vitro digestible organic matter (IVDOM) of harvested material during 2006 and 2007 04590135 g kg-12006 Jul 532a567a571ac579acL**9 Aug 467 b 504 b 523 b 540 b L**11 Sep 486 b 520 b 549ab556abL**11 Oct 492 b 574a594c604cL**,Q**7 2007 Jul 500a533a555a576aL**11 Aug 512a534a551a545aL*11 Sep 512a555a557a583aL*11 Oct 499a549a540a549aL16 g kg-1Harvest date Nitrogen (kg ha-1 cutting-1) Polynomial contrast SE Comparison of month means within an N rate a nd year. Means in same column and year with same letter are not significantly different (P>0.05). Orthogonal polynomial contrast of nitrogen effect within harvest date. L=linear, Q=quadratic, C=cubic; letter wi th no symbol (P 0.1); *(P 0.05);**(P 0.01).

PAGE 68

68Table 3-5. Probability levels (P) for the effects of year (Yr), regrowth cycle (Cy), day of regrowth (Day), and N fertilization rate (Nrt) on shoot dry matter (Shoot), leaf dry matter (Leaf), percent leaf (Pleaf )), leaf nitrogen concentr ation (Leaf N), stem dry matter (Stem), stem nitrogen concentration (Stem N), stem total nonstructural carbohydrat e concentration (Stem TNC), rhizome dry matter (Rhiz), rhizome nitrogen concentra tion (Rhiz N), and rhizome total nonstructural carbohydrate concentration (Rhiz TNC). Source of variationShootLeafLeaf NStemStem NStem TNC Yr<0.01<0.01<0.010.79<0.01<0.010.37<0.01<0.010.98 Cy<0.01<0.010.4<0.010.20<0.010.10<0.05<0.010.29 Day<0.01<0.01<0.010.29<0.010.250.400.08<0.01<0.05 <0.01<0.010.16<0.01<0.01<0.01<0.010.67<0.01<0.01 Yr Cy0.070.20<0.010.12<0.01<0.050.250.37<0.01<0.01 Yr Day0.70<0.01<0.010.68<0.01<0.05<0.01<0.01<0.01<0.01 0.330.420.91<0.010.060.290.110.1<0.01<0.05 Cy Day<0.01<0.010.430.49<0.01<0.01<0.050.740.940.94 <0.01<0.010.990.960.170.40<0.010.19<0.010.82 <0.01<0.01<0.010.78<0.010.15<0.05<0.050.100.84 Yr Cy Day0.110.250.250.35<0.010.330.220.530.69<0.01 0.43<0.050.360.670.380.610.760.98<0.050.35 0.780.720.80.880.060.440.180.970.100.89 <0.01<0.050.710.56<0.010.910.320.160.360.90 0.700.090.660.66<0.010.460.400.300.960.59 Pleaf RhizRhiz NRhiz TNC Nrt Yr Nrt Cy Nrt Day Nrt Yr Cy Nrt Yr Day Nrt Cy Day Nrt Yr Cy Day Nrt

PAGE 69

69 Table 3-6. Canopy height of Tifton 85 bermudagrass grown at Gainesville, FL at harvest for four N fertilization rates for three harvest dates in 2007. Harvest date SE 04590135 cm cm Aug 34445255L**,Q*1.2 25364345L**,Q**1.0 Oct 24363840L**,Q**,C**1.0 Nitrogen (kg ha-1 cutting-1) Polynomial Contrasts Sep Orthogonal polynomial contrast of N effect w ithin harvest date. L=linear, Q=quadratic, C=cubic; *(P 0.05); **(P 0.01). Table 3-7. Probability levels (P) for the effects of year, month of sampling, and N fertilization rate (N rate) on root dry matter (Root DM ), root total nonstructural carbohydrate concentration (Root TNC), and root nitrogen concentration (Root N). Source of variationRoot DMRoot TNCRoot N Year <0.010.170.50 Month <0.01<0.050.06 N rate 0.050.070.07 Year Month <0.010.710.99 Year N rate 0.21<0.010.08 Month N rate 0.390.510.75 Year Month Nitrogen0.990.710.38

PAGE 70

70 Table 3-8. Probability levels (P) for the effect s of year (Yr), regrow th cycle (Cy), day of regrowth (Day), and N fertilization rate (Nrt) on leaf area in dex (LAI), and canopylevel photosynthesis (Cphot). Source of variationLAICphot Yr<0.01<0.01 Cy<0.01<0.01 Day<0.01<0.01 Nrt<0.01<0.01 Yr Cy0.200.28 Yr Day<0.010.105 Yr Nrt0.420.16 Cy Day<0.01<0.05 Cy Nrt<0.010.10 Day Nrt<0.01<0.01 Yr Cy Day0.250.72 Yr Cy Nrt<0.050.15 Yr Day Nrt0.720.74 Cy Day Nrt<0.050.47 Yr Cy Day Nrt0.09370.83 Table 3-9. Percent light inter ception of Tifton 85 bermudagrass grown at Gainesville, FL under four N fertilization rates for Cycle 2 of 2006. SE 04590135 percentpercent July 26 45454038 L*5 Aug 2 52737472 L**,Q**3 Aug 10 78919495 L**,Q*4 Aug 15 80929597 L**,Q*3 Measurement date Nitrogen (kg ha-1 cutting-1) Polynomial Contrasts Orthogonal polynomial contrast of N effect with in measurement date. L=linear, Q=quadratic; *(P 0.05); **(P 0.01).

PAGE 71

71 Table 3-10. Effects of N rate on leaf-lev el photosynthesis by measurement date. 04590135 mol m-2 s-12006 Jul 2633373844L**1 Aug 237453446NS5 Aug 932323533NS5 Aug 1531273629NS6 Sep 2738464347NS4 Oct 436413748NS7 Oct 1133364345L*4 2007 Aug 140295334C*6 Aug 735323945NS7 Aug 1327262633NS9 Oct 842433745NS6 mol m-2 s-1Measurement date Nitrogen (kg ha-1 cutting-1) Polynomial contrast SE Orthogonal polynomial contrast of N effect with in measurement date. NS=not significant, L=linear, C=cubic; *(P 0.05); **(P 0.01).

PAGE 72

72 CHAPTER 4 SIMULATING THE REGROWTH DYNAM IC S OF TIFTON 85 BERMUDAGRASS WITH THE CROPGRO-FORAGE MODEL Introduction The CROPGRO model is a proc ess-based m odel of crop growth and development. It simulates general biological pr ocesses such as photosynthesis and nitrogen (N) uptake using parameters unique to each species and cultivar in order to predict crop growth and development under a variety of conditions. Species parameters governing various aspects of crop growth and development such as the timing of growth stages sensitivity to photoperiod, and partitioning of photosynthate to different plant co mponents are stored in parameter files. These parameters are separate from the model itself making CROPGRO a versatile model adaptable to new species upon estimating values for these parameters. For annual crops, species parameter files currently exist for soybean ( Glycine max L.), peanut ( Arachis hypogea L.), dry bean ( Phaseolus vulgaris L.), faba bean (Vicia faba L.), and tomato ( Lycopersicon esculentum Mill.) (Scholberg et al., 1997; Boote et al. 1998a, 1998b, 2002). CROPGRO-Forage is a modified version of CROPGRO that was developed to better model the growth of tropical pere nnial grasses. This version in corporates changes in the model to accommodate the C4 photosynthetic pathway, a storage organ to simulate rhizome growth, mobilization of carbohydrate and N from rhizomes for re-growth, and a function for winter dormancy (Rymph, 2004). The species used for developing the model was bahiagrass ( Paspalum notatum Flgge). However, no set of species parameters for bermudagrass [ Cynodon dactylon (L.) Pers.] has been estimated. The objec tives of this study were to estimate species parameters for bermudagrass and to assess mode l predictions of dry ma tter (DM) partitioning, leaf area index, photosynthesis carbohydrate, and N dynamics.

PAGE 73

73 Materials and Methods Datasets Param eter estimation and model evaluation we re based on data from field research conducted in 2006 and 2007 at the University of Flor ida, Beef Research Unit, Forage Evaluation Field Laboratory near Gainesville, FL. A detailed description of the experiment is given in Chapter 3. In brief, four N rates (0, 45, 90, and 135 kg N ha-1 cutting-1) were applied to established Tifton 85 bermudagrass pasture at start of each of four 28-d regrowth cycles for both years. Yield data were collected for each of thes e regrowth cycles. For tw o of the four regrowth cycles, weekly canopy photosynthesis, leaf, stem and rhizome mass, leaf, stem, and rhizome N concentration, and stem and rhizome TNC concentr ation were measured. Daily solar radiation, maximum temperature, and minimu m temperature data for the experiment were taken from the Alachua weather station of the Florida Au tomated Weather Network (FAWN) database (http://fawn.ifas.ufl.edu). Rainfall was measured and recorded on site. Soil profile data were taken from the 1985 Alachua County soil survey (Soil Survey Staff, 1985). Set-up of Simulation Experiment A sim ulation experiment was created based on fi eld research described in Chapter 3 using the DSSAT crop management data editing pr ogram XBUILD (Uryasev et al., 2003a). Information such as simulation starting and endi ng dates, initial crop weight and age, and crop management information were included in the crop management data file. In addition, the soil profile data shown in Table 4-1 for a Pom ona sand (sandy, siliceous, hyperthermic Ultic Haplaquod) were taken from the 1985 Alachua County soil survey and incor porated into a soil data file with the DSSAT soil data editing progr am SBUILD (Soil Survey Staff, 1985; Uryasev et al., 2003b). Information such as soil texture, bulk density, and organic matter concentrations for each profile layer were included in the file. Values for soil water-holding traits were

PAGE 74

74 calculated based on texture and bulk density. Solar radiation, maximum temperature, minimum temperature, and rainfall data were put into weather data files using the DSSAT weather data editing software called W eatherman (Wilkens, 2003). Model Code Adjustments Two m inor model code adjustments were n ecessary prior to estimating bermudagrass parameters. The first involved the algorithm by which CROPGRO-Forage simulates harvesting. Previously, harvesting was simulated by reporti ng a residual DM from which the required fraction of DM removal from leaf and stem tis sue could be calculated. Once calculated, this fraction was applied equally to leaf and stem tis sue. However, this did not match the dynamics measured for post-harvest residual leaf and stem material for bermudagrass. In the field, the fraction of leaf removed was much higher than the fraction of stem removed, with nearly all leaf material being removed by harvesting for highe r N treatments. To better simulate observed harvests, the model code was change d to calculate DM removal in kg ha-1, remove 95% of leaf mass (based on field observations), and remove the difference between the two from stem mass. The other code change involved the soil or ganic matter (SOM) module. Soil organic matter dynamics are handled in CROPGRO-Forage by a module based on the CENTURY model. A more detailed descripti on of this module is given in G ijsman et al. (2002). In essence, it simulates SOM decomposition based on environm ental conditions and soil data represented by three SOM fractions. The first fraction (SOM 1) represents active SOM and is readily decomposed. The second fraction (SOM2) repres ents SOM that decomposes more slowly. The passive SOM fraction is represented by SOM3 and is essentially stable. Before changing the code, simulated N mineralization produced more than 100 kg N ha-1 during cool months from January to April of both 2006 and 2007. The resu lt was very little simulated N stress for the lowest N rate until late in the season. Several adjustments of SOM2 a nd SOM3 fractions were

PAGE 75

75 attempted but resulted in excessive N stress du ring summer months. To address the issue, parameters governing temperature effects on decomposition rate were adjusted to reflect a base temperature of 15C, which was consistent w ith results reported in Jackson (2003). The resulting simulated N mineralization dynamics were deemed to be adequate for the purposes of this study. Parameter Selection Param eters selected for estimation were based on expected differences in growth dynamics between bahiagrass and bermudagrass as described in the literature. Definitions of all selected parameters are given in Table 4-2. Where literature values were lacking for bermudagrass, bahiagrass parameter values were used for initial values. Initial parameter values taken from the literature are given in Table 4-3. Parameters selected for estimation were grouped into those affecting dry matter partitioning, leaf area, photosynthesis, car bohydrate, and nitrogen dynamics. Dry matter Dry m atter partitioning is primarily govern ed by the parameters XLEAF, YLEAF, YSTEM, and YSTOR. Values for XLEAF(1-8) represent progressive vegetative stages (Vstage) based on node number. Vegetative stage is itself a function of cumulative thermal time and is reset at each harvest. Once V-stage increases beyond a given XLEAF value, partitioning between leaf, stem, and rhizome is set to the corresponding YLEAF, YSTEM, and YSTOR values (code actually interpolates between successi ve XLEAF values). Partitioning to root at a given V-stage is calculated as the differen ce between 1.0 and the sum of YLEAF, YSTEM, and YSTOR. Simulated water or N stresses increase partitioning to roots. In addition, the dormancy function added to the model by Rymph (2004) increases partitioni ng to rhizome while correspondingly decreasing partitioning to leaf, stem, and root as photoperiod decreases. Initial

PAGE 76

76 values for YLEAF, YSTEM, and YSTOR were set to values estimated for bahiagrass by Rymph et al. (2004). Leaf area Selected param eters affecting leaf ar ea included FINREF, SLAREF, SLAMAX, and SLAMIN. FINREF and SLAREF ar e the specific leaf areas (SLA ) of a species at emergence and during peak vegetative growth, respectivel y, under optimal temperat ure, water and light conditions. SLAMAX and SLAMIN are the maximu m and minimum values for SLA for a given species under limiting low light and saturating hi gh light, respectively, and these two parameters actually are the primary driver s of SLAREF. Simulated SLA is calculated based on the temperature, water, and light conditions within the range set by these parameters. Values for all four of these parameters were calculated base d on ranges of SLA for bermudagrass as reported by Morgan and Brown (1983b). Photosynthesis Estim ated parameters governing photosynthesis included FNPGN, FNPGL, SLWREF, XLMAXT, LNREF, and LFMAX. LFMAX is ma ximum leaf-level photosynthetic rate under saturating sunlight, at 30C, 350 ppm CO2, 21% oxygen, at a defined specific leaf weight (SLWREF). The SLWREF is the specific leaf we ight (SLW) at which LFMAX is set and can be used to adjust photosynthesis based on simulated SLW. Parameters FNPGN and LNREF are relate leaf N concentration to photosynthe tic rate, while FNPGL and XLMAXT modify photosynthetic rate for minimum night temperat ure and instantaneous hourly temperature conditions, respectively. These parameters were selected for estimation based on literature indicating that there is high va riability in leaf-level photosyn thesis and response to leaf N concentration from species to species (B olton and Brown, 1980; Brown and Wilson, 1983; Wilson and Brown, 1983; Ranjith and Meinze r, 1997; Taub and Lerdau, 2000). Also,

PAGE 77

77 parameters for photosynthetic response to temperature were selected for estimation based on Burton and Hanna's (1995) description of bermudag rass as being more cold-tolerant than some other C4 grasses. Initial values were set based on values op timized for bahiagrass (Rymph, 2004). Carbohydrate dynamics Carbohydrate dynam ics primarily include partitioning of daily photosynthate to different organs, as well as carbohydrate remobilizati on from stem and rhizome tissue to support regrowth. Daily photosynthate is allocated to various tissues based on the maintenance and growth respiration requirements for each tissue. Amount of growth respiration for a given tissue is calculated based on DM partitioning and carbohydrate costs associat ed with producing the tissue. PCARLF, PCARST, and PCARSR are the initial total carbohydr ate (structural and nonstructural) concentrations for leaf, stem and rhizome tissue, respectively. These are calculated as the remaining fraction of tissue afte r protein, lipid, lignin, or ganic acid, and mineral fractions are accounted for, and all of these are us ed to estimate growth respiration. In addition, ALPHS and ALPHSR are the maximum total nonst ructural carbohydrate fractions for stem and rhizome tissue, respectively, but these are placed in the tissues as part of the defined PCARST and PCARSR, and are part of the respective DM partitioning to the tissues using YLEAF, YSTEM, and YSTOR. If there is not sufficient N to grow DM at the defined critical N concentrations for each tissue, then surplus ca rbohydrate is partitioned to stem and rhizome tissue in a ratio defined by another species parameter. If maintena nce and growth carbohydrate demand cannot be met by daily photosynthate, ca rbohydrate from stem and rhizome tissue is remobilized to meet this demand, which is particularly important when residual leaf area index is low or zero after harvest. Rates governing carbohydrate remobilizat ion are given by the

PAGE 78

78 parameters CMOBMX for stem tissue, and CMOBSRX and CMOBSRN for rhizome tissue. These parameters were also initially set to values optimized for bahiagrass (Rymph, 2004). Nitrogen dynamics Nitrogen dynam ics in the model are a function of N demand, N uptake, and N remobilization. The N demand is set by DM partitioning and tissue composition. Parameters PROLFI, PROSTI, and PROSRI are defined as the maximum (luxury) tissue protein fractions for leaf, stem, and rhizome tissue, respectively. Parameters PROLFG, PROSTG, and PROSRG are the minimum tissue protein fractions required for growth of new leaf, stem, and rhizome tissue, respectively. If the N demand set by these parameters cannot be met by N uptake, N is remobilized from other tissue, but carbohydrate is accumulated in rhizome and stem. Parameters PROLFF, PROSTF, and PROSRF represent the protein fractions below which no more N remobilization can occur for leaf, stem, and rhizome tissue, respectively. Rates of N remobilization are controlled by NMOBMX for stem tissue and NMOBSRX and NMOBSRN for rhizome tissue. Simulated N deficit occurs when N demand cannot be met by N uptake and remobilization. Parameters affecting N dynamics were selected for estimation because of differences in tissue N concentration between bahiagrass and bermudagrass (Johnson et al., 2001; Sinclair et al., 2003). Initial values for these parameters were initially set to bahiagrass values (Rymph, 2004). Parameter Optimization The genera lized likelihood uncertainty es timation (GLUE) approach was used for parameter optimization. The objective of the GLUE approach is to estimate expected values of parameters based on prior knowledge, a set of ob served data, and a like lihood function. Prior knowledge comes from previous studies and scientific literature and provides information about the possible distribution of each parameter being estimated, called the prior distribution. Based

PAGE 79

79 on these distributions, sets of randomly selected values for each parameter can be generated using random sampling methods to create scenario s of parameter values to be run with the model. A likelihood function is then used to calc ulate the probability that the values in a given scenario represent the expected values for each parameter, called the likelihood value. Likelihood values are normalized by dividing each likelihood valu e by the sum of the likelihood values for all scenarios. This normalized like lihood is called the like lihood weight for that scenario. The value of a given parameter within a scenario is multiplied by the likelihood weight and then summed across all scenarios resulting in the expected value for that parameter. More detailed discussions of this method are given in Bevin and Binley (1992) a nd Shulz et al. (1999). The likelihood function used in this study was the same as that used in Wang et al. (2005): NieYLMSE MSE iii,1, )()min( (Eq. 4-1) where )(iiYLis the likelihood va lue of scenario i MSEi is the mean square error of scenario i min(MSE) is the lowest mean square error of all simulated scenarios, and N is the total number of scenarios. Bermudagrass parameter values were optimized using observed data for the 135 kg N ha-1 cutting-1 rate. Optimized values were estimat ed for each parameter group based on model simulations of 500 scenarios and data for vari ables affected by that group of parameters. Because optimized values were estimated separa tely for each group, this process was repeated for each group using the parameter estimates from the first optimization for the other groups. Results of the second optimizati on run are given in Table 4-3. Assessment of Model Predictions Using optim ized species parameter values ob tained using GLUE, growth and development were simulated for all four N rates. Results of these simulations were then compared with

PAGE 80

80 observed data. Root mean square error (RMSE) and the Willmott agreement index (d-statistic) were used to assess how well the model simulated observed data (Willmott 1981; Willmott et al., 1987). The equation for RMSE is: N i iiYY N1 2) ( 1 (Eq. 4-2) where N is the total number of data points for comparison, iY is a given observed value, and iY is the corresponding value predicted by the mode l. A better model prediction will produce a smaller RMSE. The Willmott agreement index (d-statistic) is given by the following equation: N i i i N i iiYYYY YY1 2 1 2) ( ) ( (Eq. 4-3) where N is the number of observed data points, iY is a given observed value, iY is the corresponding value predic ted by the model, and Y is the mean of the observed data. D-statistic values range from 0 to 1 with values near 1 indicating good model predictions. Results and Discussion Dry Matter Overall, dry matter predictions were good afte r the optimization procedures. For shoot mass, d-statistic values were all above 0.85 except for the 0 kg N ha-1 treatment in Cycle 4 of 2006 (Table 4-4). Root mean square error values were also fairly low. Simulated growth showed a treatment separation similar to the obser ved data with the growth of 45, 90, and 135 kg N ha-1 treatments following similar trends and the 0 kg N ha-1 treatment growth being much lower (Fig. 4-1). However, predicted shoot ma ss did not show any lag in growth as was observed in the field. This is probably becau se none of the residual stubble after simulated

PAGE 81

81 harvests was treated as damaged in the simulation. As a result, increased senescence of residual shoot (mostly stem) tissue following harv est was not mimicked by simulations. Under-prediction of shoot mass at the end of Cycle 2 of 2006 was due to simulated water stress. The fact that a similar decrease was not ob served in the field data may indicate that either soil profile data regarding soil water holding cap acity were inaccurate or the soil moisture threshold at which water stress is simulated ma y require adjustment for bermudagrass. Another possibility is that the model is over-predicting evapotranspiration (ET) an d as a result is more rapidly depleting simulated soil moisture th an is actually occurring in the field. The model calculates potential ET as a function of leaf area index with a solar extinction coefficient of 0.55, but without corrections for reduced stomatal conductance. Consequently, on cold winter days when simulate d daily photosynthesis is zero, when stomata should be closed and ET should be reduced, ET calculations are based on the LAI only, without considering stomatal conductance. This is not as great a concern for simulating growth and development of annual crops, which do not maintain a canopy year-round. However, for a perennial crop such as bermudagrass, which does maintain a canopy for most of the year, simulated ET under a small residual LAI could reduce the soil moisture enou gh to cause unwarranted simulated water stress, especially in cool winter and sp ring seasons with reduced rainfall. Leaf mass predictions were also very good. Most d-statistics were higher for leaf mass than for shoot mass except for the 0 kg N ha-1 cutting-1 rate, in part, becaus e there was less lag in leaf growth than for shoot growth (Table 4-5, Fi g. 4-2). This indicates that functions in the model that preferentially alloca te resources to leaf growth under conditions of low leaf area index worked well. The fact that there were differences in dynamics from low to high N rates shows that simulated N stress restricted leaf grow th appropriately. The ov er-prediction of leaf

PAGE 82

82 mass for high N rates and the under prediction for low N rates in cycle 2 of 2007 may indicate a need for adjusting some of the parameters gover ning N dynamics. Another possibility is that soil organic matter and mineral N data may be slightly inaccurate. If this were the case, simulated soil N mineralization may have been lower than actual N mineralization and N stress for the 0 kg N ha-1 treatment would not have been as se vere in the field as was simulated. Predicted stem mass dynamics were much less accurate than leaf mass dynamics. The slight lag in simulated stem grow th was not as distinct as observed in the field (Fig. 4-3). Further, field data showed an initial decrease in stem mass that was not properly captured by simulations. As a result, d-st atistic values were low even though RMSE values were not especially high (Table 4-6). This may be due to the reasons discussed above, specifically increased senescence of damaged residual tissue afte r harvest. Simulation of additional datasets is needed to determine whether these dynamics ar e realistic. If so, changes in model code may be needed to account for this. The issue coul d potentially be resolved by increasing simulated senescence of stem tissue, especially during th e 1to 2-week phase of remobilization of carbohydrate and N from those tissues for regrowth. Leaf Area Measured leaf area values were unrealistically low in the fiel d experiment as described in Chapter 3. Therefore, d-statis tic and RMSE values were not re ported for LAI. However, Figure 4-4 does show predicted LAI compared with vi sual estimates of LAI. Despite inherent inaccuracies due to the subjectivity of the estimat es, they were considered to be more accurate than measured LAI. Overall, predicted LAI fo llowed the treatment differences described above for leaf mass. In both years, simulated LAI wa s close to estimates for all treatments during Cycle 2, but lower than LAI estimates in Cycl e 4. This under-prediction of LAI in the fall occurred despite good leaf mass predictions. Beca use LAI in the model is a function of leaf

PAGE 83

83 mass and SLA, this indicates a potential problem with predicted SLA in fall that could be addressed with a code change to reduce SLA duri ng shorter daylengths of fall. A lack of good leaf area data prevented optimization of paramete rs affecting seasonal variation in SLA. Thus, further parameter optimization w ith datasets including good measurem ents of LAI are needed to improve model predictions. Photosynthesis Photosynthesis predictions were good except for rather precip itous drops caused by cloudy weather (Fig. 4-5). These drops are why d-statistic values ar e low despite having low RMSE values (Table 4-7). On cloudy days, canopy phot osynthesis was measured between clouds under full sun whereas simulated photosynthetic rate was based on the daily solar radiation and was correspondingly low. Effects of cloudy weather notwithstanding, differences in simulated canopy photosynthesis were evid ent between the 0 kg N ha-1 and the other three treatments. Differences in simulated leaf-level photosynthesis were insufficient to account for differences in canopy-level photosynthes is, particularly be tween the 45 kg N ha-1 cutting-1 rate and the 90 and 135 kg N ha-1 cutting-1 rates. This is consistent with measured leaf rates. In addition, differences in canopy photosynthetic rate are co rrelated with differences in LAI. Together these demonstrate that the driving force behind higher simulated canopy photosynthesis for higher N rates results from effects on total leaf area index, which is consistent w ith conclusions drawn from the field study in Chapter 3. Carbohydrate Concentration Dynamics Carbohydrate concentration dynamics were not pr edicted well as shown clearly for stem and rhizome tissue by low d-statistic values and high RMSE values (Tables 4-8 and 4-9). Model predictions were consistently higher than obs erved TNC concentrati ons for both stem and rhizome tissues (Fig. 4-6 and 4-7). This is prim arily due to the way carbo hydrate is stored when

PAGE 84

84 plant growth is restricted by N stress. In the model, when the carbohydrate supply from daily photosynthesis cannot be used for crop growth b ecause of limited available N, the unused carbohydrate is stored in stem and rhizome tissu e. As a result, prolonged N stress causes extremely high stem and rhizome TNC accumulation. In some ways this partially matched the dynamics of the observed data. Lower N rates tended to have higher stem and rhizome TNC concentrations; however, the degree to which this occurs in the model is much higher than in observed data. The extent of this effect within the model may need to be reevaluated in order to better predict observed carbohydrat e dynamics. One possible approach would be to discard unused daily photosynthate that exceeds the ma ximum carbohydrate storage capacity for existing tissues. From a physiological perspective this approach could represen t a type of feedback reduction in daily photosynthesis caused by excessive carbohydrate accumulation in photosynthetic tissue. Nitrogen Concentration Dynamics Model predictions of N concentration dynamics were not particularly good either. Though better than carbohydrate dynamics, mo st d-statistic values were still fairly low. Leaf N concentration dynamics were predicted some what well, especially for the 0 kg N ha-1 rate (Table 4-10). Predicted leaf N concentrations appeared to peak at around 35 g kg-1 whereas observed values reached as high as 45 g kg-1 (Fig. 4-8). Predicted stem and rhizome N dynamics were noticeably inaccurate as N concentrations remain ed constant across regrowth cycles (Fig. 4-9 and 4-10). No cycle for any treatment had d-stat istic values over 0.50 for stem N concentration and 0.57 for rhizome N concentration predictions (Tables 4-11 and 412). Some N dynamics might have been masked by the excessive stem and rhizome TNC concentrations discussed above; however, this should have been minimized at the highest N rate. Certainly further work in this area is needed. Accurate predictions of leaf and stem N con centration are critical

PAGE 85

85 particularly for predicting forage nutritive va lue well. One reason for inaccurate simulated dynamics might be the simulation of leaf, stem a nd rhizome as only three pools of tissue, rather than distinguishing between old an d new plant tissue within an organ type. Model additions to incorporate plant tissue cohorts mi ght allow better prediction of these dynamics. Remobilization of carbohydrate as well as N would likely be improved as movement of these resources from old to new tissue could be more mechanistically simu lated. In that case remobilization would not only occur between different types of tissue (i.e. rh izome to stem or leaf). Remobilization could also be simulated from old stem tissue to new stem tissue, with the abscission of the older mobilizing tissue. All of this would allow more realistic simulation of these dynamics. Conclusions Overall model predictions of bermudagra ss growth dynamics were fairly good. Concurrence between predicted and observed dry matter partitioning, leaf area, and photosynthesis indicate that the model simulate d these aspects of growth realistically. Nevertheless, there is still room for model improve ment, particularly in relation to the simulation of initial post-harvest growth patterns. Differ ences in predicted and observed carbohydrate and N dynamics were much greater, demonstrating that more work is needed in these areas. The effects of simulated water and N stress on these dyna mics especially require more consideration. The adjustments made to the model code were relatively minor but were needed for simulating harvests for bermudagrass and N mi neralization adequately. Though allowing for more accurate bermudagrass parameter estimation, neither change was meant to be permanent. More comprehensive solutions to these issues should be developed and incorporated into future model versions. Future model development should include furthe r optimization of parameters for leaf area using other datasets, adjustments to estimated ET calculations, model sensitivity to water stress,

PAGE 86

86 and development of tissue cohorts. Other useful model improvements would be algorithms for simulating above-ground plant ti ssue distribution and canopy archit ecture. This would allow simulated harvests to be set ba sed on residual stubble heights in stead of residual dry matter. Because forage management is generally based on residual stubble height rather than dry matter, this might make simulation results of greater us e to producers. Additionally, predicting canopy architecture dynamics could be pot entially useful for simulating di fferences in plant response to grazing as opposed to cutting. Model components for simulating forage nutritive value dynamics would also be useful. If leaf and stem tissue cohorts were incorporat ed into the model as suggested above, it would allow relatively simple prediction of digestibility dynamics based on tissue age, with more mechanistic predicti ons of nutritive value to be added later.

PAGE 87

87 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1) 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1) 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1) 07142128 0 1500 3000 4500 6000 Day of RegrowthShoot Mass (g m-2)S h o o t M a s s ( k g h a1)(A) (B) (C) (D) Figure 4-1. Simulated and observed shoot mass of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 88

88 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1) 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1) 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1) 07142128 0 500 1000 1500 2000 2500 Day of RegrowthShoot Mass (g m-2)L e a f M a s s ( k g h a1)(A) (B) (C) (D) Figure 4-2. Simulated and obser ved leaf mass of Tifton 85 berm udagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 89

89 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1) 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1) 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1) 07142128 0 750 1500 2250 3000 Day of RegrowthShoot Mass (g m-2)S t e m M a s s ( k g h a1)(A) (B) (C) (D) Figure 4-3. Simulated and observed stem mass of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 90

90 07142128 0 1 2 3 4 5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x 07142128 0 1 2 3 4 5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x 07142128 0 1 2 3 4 5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x 07142128 0 1 2 3 4 5 Day of RegrowthShoot Mass (g m-2)L e a f A r e a I n d e x(A) (B) (C) (D) Figure 4-4. Simulated and visually estimated le af area index of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 91

91 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) 07142128 0 20 40 60 80 Day of RegrowthShoot Mass (g m-2)C a n o p y P h o t o s y n t h e s i s ( m o l C O2 m2 s1) Figure 4-5. Simulated and observed mid-day canopy photosynthesis of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 92

92 07142128 0 75 150 225 300 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1) 07142128 0 75 150 225 300 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1) 07142128 0 75 150 225 300 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1) 07142128 0 75 150 225 300 Day of RegrowthShoot Mass (g m-2)S t e m T N C ( g k g1)(A) (B) (C) (D) Figure 4-6. Simulated and observed stem total nonstructural carbohydrate (TNC) concentrations of Tifton 85 bermudagrass grown at Gaines ville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 93

93 07142128 0 100 200 300 400 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1) 07142128 0 100 200 300 400 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1) 07142128 0 100 200 300 400 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1) 07142128 0 100 200 300 400 Day of RegrowthShoot Mass (g m-2)R h i z o m e T N C ( g k g1)(A) (B) (C) (D) Figure 4-7. Simulated and observed rhiz ome total nonstructura l carbohydrate (TNC) concentrations of Tifton 85 bermudagrass grow n at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 94

94 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1) 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1) 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1) 07142128 0 10 20 30 40 50 Day of RegrowthShoot Mass (g m-2)L e a f N ( g k g1)(A) (B) (C) (D) Figure 4-8. Simulated and observe d leaf nitrogen (N) concentrat ions of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 95

95 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1) 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1) 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1) 07142128 0 10 20 30 40 Day of RegrowthShoot Mass (g m-2)S t e m N ( g k g1)(A) (B) (C) (D) Figure 4-9. Simulated and observed stem nitrogen (N) concentrations of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ).

PAGE 96

96 Figure 4-10. Simulated and observed rhizome nitrogen (N) concentr ations of Tifton 85 bermudagrass grown at Gainesville, FL for (A) Cycle 2 of 2006, (B) Cycle 4 of 2006, (C) Cycle 2 of 2007, and (D) Cycle 4 of 2007 for 0 kg N ha-1 ( ) 45 kg N ha-1 ( ), 90 kg N ha-1 ( ), and 135 kg N ha-1 ( ). 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1) 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1) 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1) 07142128 0 5 10 15 20 Day of RegrowthShoot Mass (g m-2)R h i z o m e N ( g k g1)(A) (B) (C) (D)

PAGE 97

97Table 4-1. Soil profile characteristics of the Pomona sand (sandy, siliceous, hyperthermic Ultic Haplaquod) used for model simulations. SLB SLLL SDUL SSAT SBDMSLOCSH2O SNH4 SNO3 SOM1 SOM2 SOM3 13 0.065 0.141 0.475 1.30 1.28 0.14 0.0 0.0 0.01 0.65 0.34 23 0.037 0.088 0.420 1.47 0.38 0.09 0.0 0.0 0.01 0.65 0.34 41 0.097 0.179 0.389 1.56 0.17 0.18 0.0 0.0 0.01 0.65 0.34 51 0.097 0.239 0.438 1.39 1.85 0.24 0.0 0.0 0.01 0.65 0.34 61 0.101 0.245 0.441 1.38 1.90 0.25 0.0 0.0 0.01 0.65 0.34 81 0.049 0.101 0.405 1.51 0.45 0.10 0.0 0.0 0.01 0.65 0.34 109 0.039 0.080 0.357 1.65 0.12 0.08 0.0 0.0 0.01 0.65 0.34 119 0.094 0.143 0.297 1.82 0.02 0.14 0.0 0.0 0.01 0.65 0.34 175 0.142 0.198 0.350 1.67 0.19 0.20 0.0 0.0 0.01 0.65 0.34 213 0.082 0.125 0.362 1.64 0.06 0.13 0.0 0.0 0.01 0.65 0.34 SLB, depth to base of soil layer (cm); SLLL, soil lower limit (cm3 cm-3); SDUL, soil drained upper limit (cm3 cm-3); SSAT, soil saturated upper limit (cm3 cm-3); SBDM, soil bulk density, moist (g cm3), SLOC, soil organic carbon (%); SH2O, Initial soil water content, (cm3 cm-3); SNH4, Initial ammonium, (g elemental N Mg-1 soil); SNO3, Initial soil nitrate, (g elemental N Mg-1 soil); SOM1, Initial microbial soil organic matter fractional composition (un itless); SOM2, Initial intermed iate soil organic matter fractio nal composition (unitless); SOM3, Initial passive soil organic matter fractional composition (unitless); SOM1 + SOM2 + SOM3 = 1.0.

PAGE 98

98Table 4-2. Definitions of selected bermudagrass species parameters us ed with the CROPGRO-Forage model. Parameter Definition Dry Matter Partitioning XLEAF(1-8) Vegetative stage (leaf number) at which corresponding YLEAF, YSTEM, and YSTOR values occur for dry matter partitioning YLEAF(1-8) Fraction of dry matter partitioned to leaf tissue at a given vegetative stage YSTEM(1-8) Fraction of dry matter partitioned to stem tissue at a given vegetative stage YSTOR(1-8) Fraction of dry matter partitioned to rhizome tissue at a given vegetative stage Leaf Area SLAREF The specific leaf area (cm2 g-1) of the standard reference cultivar at peak early vegetative phase, under optimal temperature, no water stress, and high light FINREF The specific leaf area (cm2 g-1) of leaves at emergence scaled based on cultivar parameters SLAMAX The maximum specific leaf area (cm2 g-1) of a species when grown under limiting low (near zero) light at optimum temperature conditions SLAMIN The minimum specific leaf area (cm2 g-1) of a species when grown under saturating high light at optimum temperature conditions Photosynthesis FNPGN(1-4) Parameters defining a quadratic function describing leaf photosynthetic res ponse to leaf N concentration with photosynthetic rate at 0 for leaf N concentrations below FNPGN(1) and at maximum photosynthetic rate for leaf N concentrations at and above FNPGN(2) FNPGL(1-4) Parameters defining a qu adratic function describing the effect of minimum night temperature on the next days leaf-level light-s aturated photosynthetic rate with a photosynthetic rate of 0 for previous minimum night temperatures at or be low FNPGL(1) and a nonlimited photosynthetic rate for previous minimum night temper atures at or above FNPGL(2) XLMAXT(1-6) Look-up function defining the relative photosynthetic electron transport rate versus temperature YLMAXT(1-6) Relative photosynthetic electron transport rate s corresponding to values for XLMAXT(1-6) SLWREF The specific leaf weight (g cm-2) at which standard light-s aturated LFMAX is defined LNREF The leaf N concentration at which standard light-saturated LFMAX is defined LFMAX Light-saturated leaf photosynthetic rate (mg CO2 m-2 s-1) at 30 C, 350 ppm CO2, and 21% O2

PAGE 99

99Table 4-2. (continued) Parameter Definition Carbohydrate Dynamics PCARLF, PCARST, PCARSR Total carbohydrate (structural and non-structural) concentrations for leaf, stem, and rhizome tissue ALPHS, ALPHSR Initial nonstructur al carbohydrate concentr ations for new stem, and new rhizome tissue CMOBMX Maximum carbohydrate mobilization rate from vegetative tissues (not including rhizome tissue), fraction of available carbohydrate pool per day CMOBSRX Maximum carbohydrate mob ilization rate from rhizome tissue, fraction of available carbohydrate pool per day CMOBSRN Normal carbohydrate mobiliza tion rate from rhizome tissue, fraction of available carbohydrate pool per day Nitrogen Dynamics PROLF I, G, and F Maximum (I), normal growth (G), and final (F) protein con centrations of leaf tissue PROST I, G, and F Maximum (I), normal growth (G), and final (F) protein con centrations of stem tissue PROSR I, G, and F Maximum (I), normal growth (G), a nd final (F) protein concentrations of rhizome tissue NMOBMX Maximum N mobilization rate from vegetativ e tissues (not including rh izome tissue), fraction of available N pool per day NMOBSRX Maximum N mobilization rate from rhizome tissue, fract ion of available N pool per day NMOBSRN Normal N mobilization rate from rhiz ome tissue, fraction of available N pool per day

PAGE 100

100Table 4-3. Initial and optimized Tifton 85 bermudagrass specie s parameter values used with the CROPGRO-Forage model.. Parameter Initial Value Optimized Value Units Dry matter partitioning XLEAF(1-8) 0.0, 2.0, 3.0, 5.0, 7.0, 10.0, 30.0, 40.00.0, 2.0, 3.0, 5.0, 7.0, 10.0, 30.0, 40.0V-stage YLEAF(1-8) 0.60, 0.40, 0.30, 0.25, 0.20, 0.20, 0.20, 0.200.47, 0.44, 0.43, 0.39, 0.35, 0.22, 0.14, 0.14 YSTEM(1-8) 0.10, 0.10, 0.10, 0.10, 0.05, 0.05, 0.05, 0.050.16, 0.15, 0.20, 0.20, 0.24, 0.24, 0.24, 0.24 YSTOR(1-8) 0.15, 0.20, 0.30, 0.40, 0.45, 0.50, 0.50, 0.500.18, 0.21, 0.21, 0.27, 0.27, 0.27, 0.27, 0.27 Leaf Area SLAREF 200 FINREF 200 SLAMAX 530 SLAMIN 160 Photosynthesis FNPGN(1-4) 0.75, 3.00, 10.0, 10.0 0.77, 3.35, 10.0, 10.0 %N FNPGL(1-4) 7.0, 22.0, 45.0, 57.0 6.0, 20.0, 45.0, 57.0 C XLMAXT(1-6) -5.0, 10.0, 26.0, 45.0, 57.0, 60.0 -5.0, 8.0, 38.7, 50.0, 55.0, 60.0 C YLMAXT(1-6) 0.0, 0.0, 1.0, 0.8, 0.0, 0.0 0.0, 0.0, 1.0, 0.8, 0.0, 0.0 SLWREF 0.0050 0.0053 LNREF 3.00 3.35 %N LFMAX 1.760 2.053 fraction d-1fraction d-1fraction d-1cm2 g-1cm2 g-1cm2 g-1cm2 g-1 g cm-2mg CO2 m-2 s-1

PAGE 101

101Table 4-3. (continued) Parameter Initial Value Optimized Value Units Carbohydrate dynamics PCARLF,PCARST,PCARSR0.642, 0.697, 0.711 0.631, 0.699, 0.679 fraction ALPHS 0.08 0.08 fraction ALPHSR 0.30 0.17 fraction CMOBMX 0.025 0.028 fraction d-1CMOBSRX 0.050 0.050 fraction d-1CMOBSRN 0.020 0.020 fraction d-1Nitrogen dynamics PROLF I,G, and F 0.22, 0.11, 0.05 0.248, 0.099, 0.050 fraction PROST I,G, and F 0.11, 0.07, 0.033 0.111, 0.070, 0.032 fraction PROSR I,G, and F 0.092, 0.064, 0.056 0.124, 0.065, 0.050 fraction NMOBMX 0.02 0.02 fraction d-1NMOBSRX 0.06 0.06 fraction d-1NMOBSRN 0.01 0.01 fraction d-1

PAGE 102

102 Table 4-4. D-statistic and RMSE values fo r shoot mass predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statistic RMSE 2006 20.880.880.940.94331745624660 40.540.950.870.91329195398311 2007 20.920.950.930.98243460669366 40.850.920.900.89190314349361 Nitrogen (kg ha-1 cutting-1) kg ha-1 Table 4-5. D-statistic and RMSE values fo r leaf mass predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statistic RMSE 2006 20.790.890.980.98352414238223 40.780.950.930.97170144215130 2007 20.870.980.960.95199165295304 40.910.960.960.91110146139202 Nitrogen (kg ha-1 cutting-1) kg ha-1 Table 4-6. D-statistic and RMSE values fo r stem mass predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statistic RMSE 2006 20.610.790.850.80264442427499 40.330.510.560.63445181315236 2007 20.610.880.850.94358324461305 40.140.740.710.85291239254162 Nitrogen (kg ha-1 cutting-1) kg ha-1

PAGE 103

103 Table 4-7. D-statistic and RMSE values for canopy photosynthesis predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statisticRMSE 2006 20.520.560.470.5511233129 40.190.350.330.3112212324 2007 20.570.540.550.5413252930 40.330.430.440.4915232624 Nitrogen (kg ha-1 cutting-1) mol CO2 m-2 s-1 Table 4-8. D-statistic and RMSE values for stem total nonstructural carbohydrate concentration predictions based on optimized bermudagr ass parameters for four N rates. Cycle 0459013504590135 D-statistic RMSE 2006 20.190.160.260.201231847556 40.210.110.140.24511445842 2007 20.250.290.220.2551463636 40.250.340.260.2360313032 Nitrogen (kg ha-1 cutting-1) g kg-1 Table 4-9. D-statistic and RMSE values for rhizome total nonstructural carbohydrate concentration predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statistic RMSE 2006 20.150.180.210.16117189113115 40.310.190.270.25381369283 2007 20.790.380.260.2317607776 40.520.270.260.2130688181 Nitrogen (kg ha-1 cutting-1) g kg-1

PAGE 104

104 Table 4-10. D-statistic and RMSE values for leaf N concentration predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statisticRMSE 2006 20.630.660.700.693.67.55.46.8 40.940.330.520.551.86.78.47.8 2007 20.830.670.750.514.07.24.55.1 40.860.650.710.632.74.24.16.4 Nitrogen (kg ha-1 cutting-1) g kg-1 Table 4-11. D-statistic and RMSE values fo r stem N concentrati on predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statisticRMSE 2006 20.460.500.480.454.68.57.810.2 40.350.430.400.377.410.113.014.0 2007 20.460.490.470.457.011.913.214.7 40.340.410.370.388.610.513.615.2 Nitrogen (kg ha-1 cutting-1) g kg-1 Table 4-12. D-statistic and RMSE values for rhizome N concentration predictions based on optimized bermudagrass parameters for four N rates. Cycle 0459013504590135 D-statisticRMSE 2006 20.390.570.410.291.71.61.52.7 40.430.160.220.222.71.42.44.7 2007 20.360.500.280.093.82.62.84.6 40.280.570.170.072.01.05.07.1 Nitrogen (kg ha-1 cutting-1) g kg-1

PAGE 105

105 CHAPTER 5 SUMMARY AND CONCLUSIONS The objectiv es of this study were to better quantify regrowth dyna mics of bermudagrass under varying N fertilization rates and to estimate species parameter values for simulating these dynamics with the CROPGRO-Forage model. To that end, an experiment was conducted by subjecting established Tifton 85 bermudagrass pasture to four N fertilization rates. Yield, nutritive value, regrowth dynamics and photosynth esis were measured for 2006 and 2007. In addition, a simulated experiment was created based on the field experiment for estimating bermudagrass species parameters and assessing model performance. Regrowth dynamics were simulated for 2 yr and the results were compared to observed data from the field experiment. Field research showed that increasing N fe rtilization increased harvested DM, CP, and IVDOM linearly for all harvests. The exact mechanisms for N effects on IVDOM remain unclear. That the degree of N effects was strongly influenced by clim atic conditions warrants further research into the interaction between th ese factors and N fertilization. Leaf, stem, and overall shoot growth were enhanced by increasi ng the N rate. However, partitioning between leaf and stem were not affected beyond the effects of N rate on the amount of residual leaf tissue after harvest, which was greater for the low N treatment. Though percent leaf was hypothesized to increase with higher N applications, no significant differences in leaf percentage were observed in this study other than those of percen t leaf in the residual stubble mass. Leaf area expansion was enhanced by higher N fertilization allowing greater and more rapid increases in leaf area after defoliation. Leaf stem, and rhizome N concentrati ons also increased. Stem and rhizome TNC concentrations decreased under increas ing N fertilization, as was expected, due to more carbohydrate utilization for shoot growth. Ho wever, contrary to ex pectations no significant seasonal trends in rhizome TNC concentrations were observed. Canopy photosynthetic rate was

PAGE 106

106 increased by N fertilization primarily due to greater leaf area expansion under higher N treatments rather than by increa ses in leaf-level photosynthesis. Practically speaking, fertilizing at rates beyond than 90 kg N ha-1 cutting-1 would not likely increase yield or nutritive value enough to warrant the added expense. Additionally, fertilizer applications after August may not be as wort hwhile because seasonal effects would largely minimize any potential benefit. Further, the use of resources such as seasonal climate forecasts may also help producers determine whether trends in the current season are likely to enhance or counteract the effects of applying a higher N rate. Prior to estimating bermudagrass species parame ters, two adjustments had to be made to the model code. The algorithm for simulating harvests had to be changed to more accurately simulate bermudagrass harvests. In particular, changes were needed to decrease the amount of residual leaf material left after simulated harvests. In addition, model code governing temperature sensitivity for SOM decomposition ha d to be modified slightly to ensure N mineralization dynamics were being adequately captured. Parameters were changed to raise the base temperature for decomposition, effectivel y lowering N mineralization rates during winter and early spring. Both of these changes were m eant to be temporary fixe s for the purposes of estimating bermudagrass species parameters more accurately. More comprehensive solutions to these issues should be incorporat ed into future model versions. Overall, CROPGRO-Forage predicted bermudagrass regrowth dynamics fairly well. Predicted dry matter partitioning, leaf area, and photosynthesis more or less followed observed data indicating that the model is simulating these aspects of growth more or less realistically. Nevertheless, there is still room for model improve ment, particularly in relation to the simulation of initial post-harvest regrowth patterns. In addition, fairly poor model prediction of

PAGE 107

107 carbohydrate and N dynamics demonstrate that more wo rk is needed in these areas. Effects of simulated water and N stress on these dynamics especially require more improvement. Future model development should include furthe r optimization of parameters for leaf area index development using other datasets, adjustments to estimated ET calculations, model sensitivity to water stress, and development of tissue cohorts. Other useful model improvements include algorithms for simulating above-ground pl ant tissue distribution a nd canopy architecture. This would allow simulated harvests to be ba sed on residual stubble hei ghts instead of residual dry matter. Because forage management is generally based on residual stubble height rather than residual dry matter, this would make translation of simulated results to management options more clear for producers. Additionally, pred icting canopy architecture dynamics could be potentially useful for simulating differences in plant response to grazing as opposed to cutting. Model components for simulating forage nutritive va lue dynamics would also be useful. If leaf and stem tissue cohorts were inco rporated into the model, this would allow relatively simple prediction of digestibility dynamics based in part on tissue age. Eventually, more mechanistic predictions of nutritive value could be developed by simulating primary and secondary cell wall growth, and tissue lignification. All of thes e are important next steps for improving our understanding and prediction of tr opical forage grass growth.

PAGE 108

108 LIST OF REFERENCES Alexander, C. W., and D.E. McCloud. 1962. CO2 uptake (net photosynthesis) as influenced by light intensity of isolated bermudagrass leaves contrasted to that of swards under various clipping regimes. Crop Sci. 2:132-135. Beaty, E.R., K.H. Tan, R.A. McCreery, and J.B. Jones. 1975. Root-herbage production and nutrient uptake and retention by bermudagrass and bahiagrass. J. Range Manage. 28:385389. Beven, K., and A. Binley. 1992. The future of distributed models: model calibration and uncertainty prediction. Hydr ological Processes 6:279-298. Bogdan, A.V. 1977. Tropical Pasture and Fodder Plants (Grasses and Legumes). New York: Longman Inc. pp.92-98. Boote, K.J., J.W. Jones, and G. Hoogenboom 1998a. Simulation of crop growth: CROPGRO model. p. 651-692. In R.M. Peart and R.B. Curry (eds.) Agricultural Systems Modeling and Simulation. Marcel Dekker, Inc., New York. Boote, K.J., J.W. Jones, G. Hoogenboom, a nd N.B. Pickering. 1998b. The CROPGRO model for grain legumes. p. 99-128. In G.Y. Tsuji, G. Hoogenboom, and P.K. Thornton (ed.) Understanding Options for Agricultural Pr oduction. Kluwer Academic Publishers, Dordrecht, The Netherlands. Boote, K.J. and R.S. Loomis. 1991. The pr ediction of canopy assimilation. p. 109-140. In K.J. Boote and R.S. Loomis (eds.) Modeling Cr op Photosynthesisfrom Biochemistry to Canopy, CSSA Special Publication no. 19. ASA-CSSA, Madison, Wisconsin. Boote, K.J., M.I. Mnguez, and F. Sau. 2002. Adapting the CROPGRO legume model to simulate growth of faba bean. Agron. J. 94:743-756. Boote, K.J., and N.B. Pickering. 1994. Mode ling photosynthesis of ro w crop canopies. Hort. Science 29:1423-1434. Bolton, J.K., and R.H. Brown. 1980. Photosynthesi s of grass species in differing carbon dioxide fixation pathways V. Response of Panicum maximum, Panicum miloides, and tall fescue (Festuca arundinacea) to nitrogen nutrition. Plant Physiol. 66:97-100. Brink, G.E., K.R. Sistani, and D.E. Rowe 2004. Nutrient uptake of hybrid and common bermudagrass fertilized with broi ler litter. Agron. J. 96:1509-1515. Brown, R.H. and J.R. Wilson. 1983. Nitrogen response of Panicum species differing in CO2 fixation pathways. II. CO2 exchange characteristic s. Crop Sci. 23:1154-1159. Burton, G.W. 1954. Coastal bermudagrass. Ga. Agric. Exp. Sta. Bull. NS2.

PAGE 109

109 Burton, G.W. 2001. Tifton 85 bermudagrassearly history of its creation, selection, and evaluation. Crop Sci. 41:5-6. Burton, G.W., R.N. Gates, and G.M. Hill. 1993. Registration of Tifton 85 bermudagrass. Crop Sci. 33:644-645. Burton, G.W. and W.W. Hanna. 1995. Bermudagrass. p. 421-429. In R.F. Barnes, D.A. Miller, and C.J. Nelson (eds.) Forages. Vol 1: An Introduction to grassland agriculture. 5th ed. Iowa State University Press, Ames, Iowa. Burton, G.W., J.E. Hook, J.L. Butler, and R.E. He llwig. 1988. Effect of temperature, daylength, and solar radiation on produc tion of Coastal bermudagra ss. Agron. J. 80:557-560. Burton, G.W. and W.G. Monson. 1984. Registra tion of Tifton 68 bermudagrass. Crop Sci. 24:1211. Burton, G.W., G.M. Prine, and J.E. Jackson. 19 57. Studies of drought tole rance and water use of several southern grasses. Agron. J. 49:498-503. Chambliss, C.G. and L.S. Dunavin. 2003. Ti fton 85 bermudagrass. SS-AGR-57. Institute of Food and Agricultural Sciences, University of Florida, Agronomy Department, Florida Cooperative Extension Service. Chaparro, C.J., L.E. Sollenberger, and K.H. Quesenberry. 1996. Light interception, reserve status, and persistence of clipped Mott elephantgrass swards. Crop Sci. 36:649-655. Chapman, D.F. and G. Lemaire. 1993. Morphogenetic and structural de terminants of plant regrowth after defoliation. p.95-104. In Proc. Int. Grassl. Congr., 17th, Palmerston North, New Zealand. 8-21 Feb. 1993. Pa lmerston North, New Zealand. Christiansen, S., O.C. Ruelke, W.R. Ocumpa ugh, K.H. Quesenberry, and J.E. Moore. 1988. Seasonal yield and quality of Bigalta, Redal ta, and Floralta limpograss. Trop. Agric. 65:49-55. da Fonseca, A.F., A.J. Melfi, F.A. Monteiro, C.R. Montes, V.V. de Almeida, and U. Herpin. 2007. Treated sewage effluent as a so urce of water and nitrogen for Tifton 85 bermudagrass. Agric. Water Manage. 87:328-336. Fagundes, J.L., S.C. da Silva, C.G.S. Pedreira A.F. Sbrissia, C.A.B. de Carvalho, R.A. Carnevalli, L.F de M. Pinto, L.K. Molan, and I.N.S. Rolim. 1999. Light interception and herbage accumulation in Tifton-85 swards grazed by sheep under c ontinuous stocking. In A. de Moraes, C. Nabinger, P.C. de F. Carv alho, S.J. Alves, and S.B.C. Lustosa (eds.) Proc. Int. Symposium Grassland Ecophysiolo gy and Grazing Ecology. Curtiba, Paran, Brazil. 24-26 Aug. 1999. Fritschi, F.B. 1996. Establishment growth of perennial peanut and bahiagrass in response to carbon dioxide and temperature. M.S. thesis. Univ. of Florida, Gainesville.

PAGE 110

110 Gallaher, R. N., C.O. Weldon, and J.G. Futral. 1975. An aluminum block digester for plant and soil analysis. Soil Sci. Soc. Amer. Proc. 39:803-806. Gastal, F., and G. Lemaire. 2002. N uptake and distribution in crops : an agronomical and ecophysiological perspective. J. Exp. Bot. 53:789-799. Gijsman, A.J., G. Hoogenboom, W.J. Parton, and P.C. Kerridge. 2002. Modifying DSSAT crop models for low-input agricultural systems using a soil organic matter-residue module from CENTURY. Agron. J. 94:462-474. Goatley, J.M., V. Maddox, D.J. Lang, and K.K. Crouse. 1994. Tifgreen bermudagrass response to late-season application of nitroge n and potassium. Agron. J. 86:7-10. Hambleton, L. G. 1977. Semiautomated method for simultaneous determination of phosphorus, calcium and crude protein in anim al feeds. J.A.O.A.C. 60:845-852. Harlan, J.R., and J.M.J de Wet. 1969. Sources of variation in Cynodon d actylon (L.) Pers. Crop Sci. 9:774-778. Harlan, J.R., J.M.J. de Wet, and K.M. Rawal. 1970. Geographic distribution of the species of Cynodon L.C. Rich. (Gramineae). East Afr. Agric. For. J. 36:220-226. Harris, W. 1978. Defoliation as a determinant of the growth, persiste nce and composition of pasture. p. 67-85. In J.R. Wilson (ed.) Plant relati ons in pastures. CSIRO, East Melbourne, Australia. Hart, R.H. and D.R. Lee. 1971. Age vs. net CO2 exchange of leaves of coastal bermudagrass. Crop Sci. 11:598-599. Harvey, R.W., J.P. Mueller, J.A. Barker, M.H. Poore, and J.P. Zublena. 1996. Forage characterstics, steer performance, and water qua lity from bermudagrass pastures fertilized with two levels of nitrogen from swine lagoon effluent. J. Anim. Sci. 74:457-464. Hill, G.M., R.N. Gates, and G.W. Burton. 1993. Forage quality and grazing steer performance from Tifton 85 and Tifton 78 bermudagrass pastures. J. Anim. Sci. 71:3219-3225. Hill, G.M., R.N. Gates, J.W. West, and P.R. Utley. 1996. Intake and digestibility of bermudagrass hays harvested at two matur ity stages. J. Anim. Sci. 74(Suppl. 1):18 (Abstr.). Hoogenboom, G., J.W. Jones, C.H. Porter, K.J. Boote, W.D. Batchelor, L.A. Hunt, A.J. Gijsman, P.W. Wilkens, U. Singh, and W.T. Bowen. 2003. DSSAT v4 cropping system simulation model. In G. Hoogenboom, J.W. Jones, C.H. Porter, P.W. Wilkens, K.J. Boote, W.D. Batchelor, L.A. Hunt, and G.Y. Tsuji (eds .) Decision Support System for Agrotechnology Transfer Version 4.0. Volume 1. Un iversity of Hawaii, Honolulu, HI. Humphreys, L.R. 1991. Tropical pasture utilisation. Cambridge University Press, Cambridge, U.K.

PAGE 111

111 Jackson, B.A. 2003. Nitrogen mineralization of incorporated bahiagrass sod and its effect on nutrient uptake and yield of maize. MS Thesis. Univ. of Florida. Gainesville. Johnson, C.R., B.A. Reiling, P. Mislevy, and M.B. Hall. 2001. Effects of nitrogen fertilization and harvest date on yield, digestibility, fiber, and protein fractions of tropical grasses. J. Anim. Sci. 79:2439-2448. Jones, C.A. 1985. C4 Grasses and Cereals: Growth, deve lopment, and stress response. John Wiley & Sons, New York. Jones, J.W., G. Hoogenboom, C. H. Porter, K.J. Boote, W.D. Batchelor, L.A. Hunt, P.W. Wilkens, U. Singh, A.J. Gijsman, and J.T. Ritchie. 2003. The DSSAT cropping systems model. Europ. J. Agron. 18:235-265. Ku, S.B., and G.E. Edwards. 1977. Oxygen inhibition of photosynthesis. I. Temperature dependence and relation to O2/CO2 solubility ratio. Plant Physiol. 59:986-990. Lin, L., A.S. Hedayat, B. Sinha, M. Yang. 2002. Statistical methods in assessing agreement: models, issues and tools. J. Amer. Stat. Assoc. 97:257-270. Ludlow, M.M., and G.L. Wilson. 1971. Photosynthesi s of tropical pasture plants. I. Illuminance, carbon dioxide concentration, leaf temperature, and leaf-air vapor pressure difference. Aust. J. Biol. Sci. 24:449-470. Makowski, D., J. Hillier, D. Wallach, B. Andrieu, and M.-H. Jeuffroy. 2006. Parameter estimation for crop models. p.101-140. In D. Wallach, D. Makowski, and J.W. Jones (eds.) Working with Dynamic Crop Models: Ev aluation, Analysis, Pa rameterization, and Applications. Elsevier B.V., Amsterdam. Makowski, D., Wallach, D., and Tremblay, M. 2002. Using a Bayesian approach to parameter estimation: comparison of the GLUE and MCMC methods. Agronomie 22:191-203. Mandebvu, P. J.W. West, G.M. Hill, R.N. Gates, R.D. Hatfield, B.G. Mullinix, A.H. Parks, and A.B. Caudle. 1999. Comparison of Tifton 85 and Coastal bermudagrasses for yield, nutrient traits, intake, and digestion by grow ing beef steers. J. Anim. Sci. 77:1572-1586. Mandebvu, P., J.W. West, R.N. Gates, and G.M. Hill. 1998a. Effect of hay maturity, forage source, or neutral detergent fiber conten t on digestion of diet s containing Tifton 85 bermudagrass and corn silage. Anim. Feed Sci. Technol. 73:281-290. Mandebvu, P., J.W. West, R.N. Gates, and G.M. Hill. 1998b. In vitro digestion kinetics of neutral detergent fiber extracted from Tift on 85 and Coastal bermudagrasses. Anim. Feed Sci. Technol. 73:263-269. McCormick, M.E. 1974. Effects of forage variety, fertilizer, and cutting on nutritive value and yield of various forages. M.S. thesis. Louisiana State University, Baton Rouge.

PAGE 112

112 Mislevy, P. and F.G. Martin. 2006. Biom ass yield and forage nutritive value of Cynodon grasses harvested monthly. Soil Crop Sc i. Soc. Florida Proc. 65:9-14. Moore, J. E., and G. O. Mott. 1974. Recovery of residual organic matter from in vitro digestion of forages. J. Dairy Sci. 57:1258-1259. Moore, K.J., K.J. Boote, and M.A. Sanderson. 2004. Physiology and developmental morphology. In L.E. Moser, B.L. Burson, and L.E. Sollenberger (eds.) Warm-Season (C4) Grasses. Agron. Monogr. 45. ASA, CSSA, SSSA, Madison, WI. Morgan, J.A., and R.H. Brown. 1983a. Photosynth esis and growth of bermudagrass swards. I. Carbon dioxide exchange characteristics of swards mowed at weekly and monthly intervals. Crop Sci. 23:347-352. Morgan, J.A., and R.H. Brown. 1983b. Photosynthe sis and growth of bermudagrass swards. II. Growth patterns as estimated by harvest a nd gas exchange techniques. Crop Sci. 23:352357. Overman, A.R., M.A. Sanderson, and R.M. Jone s. 1993. Logistic response of bermudagrass and bunchgrass cultivars to applied ni trogen. Agron. J. 85:541-545. Pedreira, C.G.S.. 1995. Plant and animal responses on grazed pastures of Florakirk and Tifton 85 bermudagrasses. PhD Dissertation. Univ. of Florida, Gainesville. Pedreira, C.G.S., L.E. Sollenberger, and P. Mislevy. 2000. Botanical composition, light interception, and carbohydrate reserve status of grazed Florakirk bermudagrass. Agron. J. 92:194-199. Prine, G.M., and G.W. Burton. 1956. The effect of nitrogen rate and clipping frequency upon the yield, protein content and cer tain morphological characteris tics of Coastal bermudagrass (Cynodon dactylon (L) Pers.). Agron. J. 48:296-301. R Development Core Team. 2006. R: A language a nd environment for statistical computing. R Foundation for Statistical Computing, Vienna Austria. URL http://www.R-project.org. Ranjith, S.A. and F.C. Meinzer. 1997. Physiol ogical correlates of variation in nitrogen-use efficiency in two contrasting suga rcane cultivars. Crop Sci. 37:818-825. Richards, J.H. 1993. Physiology of plants recovering from defoliation. p. 85-93. In Proc. Int. Grassl. Congr., 17th, Palmerston North, New Zealand. 8-21 Feb. 1993. Palmerston North, New Zealand. Rymph, S.J. 2004. Modeling growth and compositi on of perennial tropical forage grasses. PhD Dissertation. Univ. of Florida, Gainesville. Rymph, S.J., K.J. Boote, A. Irmak, P. Misle vy, and G.W. Evers. 2004. Adapting the CROPGRO Model to Predict Growth and Compos ition of Tropical Grasses: Developing Physiological parameters. Soil Crop Sc i. Soc. Florida Proc. 63:37-51.

PAGE 113

113 Sampaio, E.V.S.B., E.R. Beaty, and D.A. As hley. 1976. Bahiagrass regrowth and physiological aging. J. Range Manage. 29:316-319. Schmidt, R.E., and R.E. Blaser. 1969. Effect of temperature, light, and nitrogen on growth and metabolism of Tifgreen bermudagrass (Cynodon spp.). Crop Sci. 9:5-9. Scholberg, J.M.S., K.J. Boote, J.W. Jones, and B.L McNeal. 1997. Adap tation of the CROPGRO model to simulate the growth of field-grown tomato. p. 133-151. In M.J. Kropff, P.S. Teng, P.K. Aggarwal, J. Bouma, B.A.M. Boum an, J.W. Jones, and H.H. van Laar (eds.) Proceedings of the second international symposium on systems approaches for agricultural development. Kl uwer Academic Publ., London. Schenk, J.S., and M.O. Westerhaus. 1991a. Popul ation definition, sample selection, and calibration procedures for near infrared re flectance spectroscopy. Crop Sci. 31:469-474. Schenk, J.S., and M.O. Westerhaus. 1991b. Populati on structuring of near infrared spectra and modified partial least squares regression. Crop Sci. 31:1548-1555. Shulz, K., K. Beven, B. Huwe. 1999. Equifina lity and the problem of robust calibration in nitrogen budget simulations. Soil Sci. Soc. Am. J. 63:1931-1941. Silveira, M.L., V.A. Haby, and A.T. Leonard. 20 07. Response of Coastal bermudagrass yield and nutrient uptake efficiency to nitr ogen sources. Agron. J. 99:707-714. Sinclair, T.R., J.D. Ray, L.M. Premazzi, a nd P. Mislevy. 2004. Photosynthetic photon flux density influences grass responses to extende d photoperiod. Environ. Exp. Bot. 51:69-74. Sinclair, T.R., P. Mislevy, and J.D. Ray. 2001. Short photoperiod inhib its winter growth of subtropical grasses. Planta 213:488-491. Sinclair, T.R., J.D. Ray, P. Mislevy, and L. M. Premazzi. 2003. Growth of subtropical forage grasses under extended photoperiod during short-daylength months. Crop Sci. 43:618623. Sistani, K.R., G.E. Brink, A. Adeli, H. Te wolde, and D.E. Rowe. 2004. Year-round soil nutrient dynamics from broiler litter application to three bermudagrass cultivars. Agron. J. 96:525-530. Skerman, P.J. and F. Riveros. 1990. Tropical Gr asses. Food and Agriculture Organization of the United Nations: Rome. pp.310-315. Soil Survey Staff. 1985. Soil survey of Alachua County, Florida. United States Department of Agriculture, Soil Conservation Service. U.S. Government Printing Office. Staples, C.R. 1995. Bermudagrass: Growing, stor ing, and feeding for dairy animals. Circular 1140. IFAS, Univ. of Florida, Florida Cooperative Extension Service.

PAGE 114

114 Taub, D.R. and M.T. Lerdau. 2000. Relationship between leaf nitrogen and photosynthetic rate for three NAD-ME and three NADP-ME C4 grasses. Am. J. Bot. 87:412-417. Trenholm, L.E., A.E. Dudeck, J.B. Sartain, a nd J.L. Cisar. 1998. Bermudgrass growth, total nonstructural carbohydrate concen tration, and quality as influenced by nitrogen and potassium. Crop Sci. 38:168-174. Uryasev, O., J.W. Jones, G. Hoogenboom, a nd C. Porter. 2003. DSSAT v4 crop management data editing program (Xbuild). In G. Hoogenboom, J.W. Jones, C.H. Porter, P.W. Wilkens, K.J. Boote, W.D. Batchelor, L.A. Hunt, and G.Y. Tsuji (eds.) Decision Support System for Agrotechnology Transfer Version 4.0. Volume 2. University of Hawaii, Honolulu, HI. Vendramini, J.M.B., L.E. Sollenberger, A.T. Ades ogan, J.C.B. Dubeaux, Jr., S.M. Interrante, R.L. Stewart, Jr., and J.D. Arthington. 2008. Protein fractions of Tifton 85 and ryeryegrass due to sward management practices. Agron. J. 100:463-469. Wallach, D. 2006. Evaluating crop models. p.11-50. In D. Wallach, D. Makowski, and J.W. Jones (eds.) Working with Dynamic Crop Models: Evaluation, Analysis, Parameterization, and Applicati ons. Elsevier B.V., Amsterdam. Wang, X., He, X., Williams, J.R., Isaurralde, R.C., and Atwood, J.D. 2005. Sensitivity and uncertainty analyses of crop yields and so il organic carbon simulated with EPIC. Trans ASAE 48:1041-1054. Wilkens, P.W. 2003. DSSAT v4 weather data editing program (Weatherman). In G. Hoogenboom, J.W. Jones, C.H. Porter, P.W. Wilk ens, K.J. Boote, W.D. Batchelor, L.A. Hunt, and G.Y. Tsuji (eds.) Decision S upport System for Agrotechnology Transfer Version 4.0. Volume 2. Univer sity of Hawaii, Honolulu, HI. Willmott, C.J. 1981. On the validation of models. Physical Geography 2:184-194. Willmott, C.J., S.G. Ackleson, R.E. Davis, J.J. Feddema, K.M. Klink, D.R. Legates, J. ODonnell, C.M. Rowe. 1987. Statistics for th e evaluation and comparison of models. J. Geophys. Res. 90:8995-9005. Wilson, J.R. 1975. Comparative response to nitrogen deficiency of a tropi cal and temperate grass in the interrelation between photosynthesis, growth and accumulation of non-structural carbohydrate. Neth. J. Agric. Sci. 23:104-112. Wilson, J.R. and R.H. Brown. 1983. Nitrogen response of Panicum species differing in CO2 fixation pathways. I. Growth analysis a nd carbohydrate accumulation. Crop Sci. 23:11481153. Woodard, K.R., E.C. French, L.A. Sweat, D.A. Graetz, L.E. Sollenberger, B. Macoon, K.M. Portier, B.L. Wade, S.J. Rymph, G.M. Prin e, and H.H. VanHorn. 2002. Nitrogen removal and nitrate leaching for forage systems recei ving dairy effluent. J. Environ. Qual. 31:1980-1992.

PAGE 115

115 BIOGRAPHICAL SKETCH As the child of an Air Force officer, Phillip D. Alderm an grew up internationally. By the time he had graduated high school from the American School in Japan, he had lived or spent time in eleven countries outside the USA. The cross-cultural and cross-linguistic experiences of Phillips childhood gave him a broa d perspective from which to view the world, and when he arrived at the University of Flor ida as an undergraduate, he pursued a Bachelor of Arts degree in linguistics. His experience study ing linguistic theory and diffe rent language systems expanded his analytical capacity and honed his abilities to perceive underl ying patterns behind observed phenomena. Nevertheless, while Phillip enjoyed his linguistic studi es, he sought a more practical means by which to serve others. A conversation with Dr. Jerry Bennett, a fellow church member and chair of the Agronomy De partment, led him to explore possibilities in agronomy. After much consideratio n, Phillip added a minor in plan t science to his undergraduate studies. As he began studying agronomy, he also star ted working as an OPS research assistant doing research with Dr. Kenneth Boote in the Agronomy Department. Through these experiences, Phillip became interested in crop growth simulation modeling and its potential as a tool for agronomic research and agricultural devel opment. It was these interests that led him to pursue a Master of Science degree related to crop growth modeling in preparation for future work in international agricultural development.