<%BANNER%>

Improving a Tomato Growth Model to Predict Fresh Weight and Size of Individual Fruits

Permanent Link: http://ufdc.ufl.edu/UFE0024298/00001

Material Information

Title: Improving a Tomato Growth Model to Predict Fresh Weight and Size of Individual Fruits
Physical Description: 1 online resource (221 p.)
Language: english
Creator: Rybak, Maria
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: fresh, fruit, model, tomato, weight
Agronomy -- Dissertations, Academic -- UF
Genre: Agronomy thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: A new version of the CROPGRO-Tomato model was developed to include a subroutine for predicting dry matter concentration, fresh weight and size of fruits over time. This subroutine builds on already existing crop model predictions of dry mass per fruit and fruit thermal age. The model was calibrated with field data obtained on tomato cultivar ?Florida 47? in Gainesville Florida during spring 2007 and evaluated with an independent data set obtained during spring 2006. The dynamics of dry weight accumulation, fresh weight, dry matter concentration, and fruit size were simulated by the model and compared with independent data of fruits tagged at anthesis in the field study at three successive weekly dates. These results demonstrated that the model was able to explain and predict the time-series growth and development of individual fruits on a cohort basis, to include delayed growth, slower growth, and smaller size of progressively later-set fruits. Overall, the standard deviation of model error was less than 20 percent of the mean for all the variables evaluated (dry weight, fresh weight, fruit dry matter concentration, and fruit size). In addition, the Willmot index was always above 0.9. The effects of water and N stress on the growth of individual fruits were studied by withholding drip irrigation for three periods and withholding N supply beginning at 14 days after transplanting. The main effects of water stress on the growth of individual fruits were reduced fresh mass and size and an increase in dry matter concentration, while N stress on the other hand, caused a decrease in dry matter concentration of fruits. The existing water and N stress factors in the CROPGRO model were linked to the new equations that affect dry weight growth and dry matter concentration of individual fruits, and were successfully used to predict the stress-induced variations in dry matter concentration, and reductions in dry and fresh weight growth and size of single fruits. The newly-added subroutine for predicting dry matter concentration, fruit fresh weight, and size, along with coupling to water and N stresses make the improved model a valuable risk assessment tool for predicting fresh market production and quality of tomato.
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 Maria Rybak.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Boote, Kenneth J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024298:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024298/00001

Material Information

Title: Improving a Tomato Growth Model to Predict Fresh Weight and Size of Individual Fruits
Physical Description: 1 online resource (221 p.)
Language: english
Creator: Rybak, Maria
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: fresh, fruit, model, tomato, weight
Agronomy -- Dissertations, Academic -- UF
Genre: Agronomy thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: A new version of the CROPGRO-Tomato model was developed to include a subroutine for predicting dry matter concentration, fresh weight and size of fruits over time. This subroutine builds on already existing crop model predictions of dry mass per fruit and fruit thermal age. The model was calibrated with field data obtained on tomato cultivar ?Florida 47? in Gainesville Florida during spring 2007 and evaluated with an independent data set obtained during spring 2006. The dynamics of dry weight accumulation, fresh weight, dry matter concentration, and fruit size were simulated by the model and compared with independent data of fruits tagged at anthesis in the field study at three successive weekly dates. These results demonstrated that the model was able to explain and predict the time-series growth and development of individual fruits on a cohort basis, to include delayed growth, slower growth, and smaller size of progressively later-set fruits. Overall, the standard deviation of model error was less than 20 percent of the mean for all the variables evaluated (dry weight, fresh weight, fruit dry matter concentration, and fruit size). In addition, the Willmot index was always above 0.9. The effects of water and N stress on the growth of individual fruits were studied by withholding drip irrigation for three periods and withholding N supply beginning at 14 days after transplanting. The main effects of water stress on the growth of individual fruits were reduced fresh mass and size and an increase in dry matter concentration, while N stress on the other hand, caused a decrease in dry matter concentration of fruits. The existing water and N stress factors in the CROPGRO model were linked to the new equations that affect dry weight growth and dry matter concentration of individual fruits, and were successfully used to predict the stress-induced variations in dry matter concentration, and reductions in dry and fresh weight growth and size of single fruits. The newly-added subroutine for predicting dry matter concentration, fruit fresh weight, and size, along with coupling to water and N stresses make the improved model a valuable risk assessment tool for predicting fresh market production and quality of tomato.
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 Maria Rybak.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Boote, Kenneth J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024298:00001


This item has the following downloads:


Full Text

PAGE 1

1 IMPROVING A TOMATO GROWTH MODEL TO PREDICT FRESH WEIGHT AND SIZE OF INDIVIDUAL FRUITS By MARIA RAQUEL RYBAK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQU IREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Maria Raquel Rybak

PAGE 3

3 To my Family

PAGE 4

4 ACKNOWLEDGMENTS I wish to express my thanks and compliments to my supervisory commit tee, Dr. Kenneth Boote, Dr. James Jones Dr. Jerry Bennett, Dr. Eduardo Vallejos and Dr. Barry Tillman for being part of my committee and reading and criticizing the manuscript. I also owe appreciation to Dr. Bennett, who was my first contact at the Univer sity of Florida and diligently connected me with my advisor. My most considerable debt is to Dr. Kenneth J. Boote who was generous with his time and expertise in this research. His knowledge, patience and constant support were equally important to the completion of this dissertation In the academies one finds many intelligent professors but my honor and privilege has been to find two wise professors. My admiration and gratitude go to Dr. Ken Boote and Dr. Jim Jones, now and always. Special thanks go to Dr. Johan Scholberg and Cheryl Porter for their indispensable help. They generously donated hours of time and shared valuable data with me. I also want say thank you to Dr. Lincoln Zotarelli, Dr. Michel Dukes, Mr. Wayne Williams, Mr. Richard Fethiere and Mr. Jason Hupp for their technical assistance. Sincere gratitude is extended to the faculty, staff members and other people in the Agronomy Department for the help they extended anytime I needed it. I owe gratitude to the National Institute of Agricultural Tec hnology in Argentina for authorizing a nd assisting me in this program. E specially thanks go to Ing. Nestor Oliveri who has supported me throughout my career. Thanks to my family; they fill my heart with unimaginable love. Especially thanks to my beloved mo ther Veronica and sister Myrian. They are also my best friends and a source of constant encouragement. Myrian is always guiding my steps down a path that I can follow with confidence beca use she has been there before. My brothers, sisters and nephews are true gifts in

PAGE 5

5 my life. I also want to honor the precious memory of my father. His love, integrity and wisdom always kept me going in the right direction. I would like say thanks to Nidia, Carmen and Luca who have been as my own family in Gainesville. My m emories go to my friend McNair who is gone but his example and kindness will be in my heart forever. Also thanks go to Sara Jane, Welch and Julie who gave me so much warmness. Thanks to my friends who gave me motivation from distant lands, and those I me t here who shared their friendship with me. Finally and always, thanks to my unfailing and amazing God for his everlasting love and grace which sustain me.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES ............................................................................................................................ 10 ABSTRACT ........................................................................................................................................ 18 CHAPTER 1 INTRODUCTION ....................................................................................................................... 20 2 LITERATURE REVIEW ........................................................................................................... 22 Introduction ................................................................................................................................. 22 Influence of the Environment on Growth and Development of Tomato Plants ...................... 2 2 Light ...................................................................................................................................... 22 Temperature ......................................................................................................................... 32 Water .................................................................................................................................... 43 Nitrogen ................................................................................................................................ 47 Partitioning of Assimilates Within a Tomato P lant .................................................................. 49 Genetic Traits Related to Tomato Yield .................................................................................... 56 Tomato Models ............................................................................................................................ 59 3 UPDATING PARAMETERS IN CROPGRO AND TESTING THE MODEL FOR PREDICTING GROWTH AND DRY MATTER ACCUMULATION ................................ 65 Introduction ................................................................................................................................. 65 Materials and Methods ................................................................................................................ 65 Results and Discussion ............................................................................................................... 69 Conclusions ................................................................................................................................. 74 4 EQUATIONS FOR GROWTH OF TOMATO FRUITS UNDER NON LIMITING CONDITIONS ............................................................................................................................ 89 Introduction ................................................................................................................................. 89 Materials and Methods ................................................................................................................ 92 Results and Discussion ............................................................................................................... 93 Conclusion ................................................................................................................................... 95 5 IMPROVING A TOMATO GROWTH MODEL TO PREDICT FRESH WEIGHT AND SIZE OF INDIVIDUAL FRUITs .................................................................................. 101 Introduction ............................................................................................................................... 101

PAGE 7

7 Materials and Methods .............................................................................................................. 102 Results and Discussion ............................................................................................................. 108 Calibration Results (Year 2007) ....................................................................................... 108 Validation Results (Year 20 06) ........................................................................................ 113 Ability of the Model to Predict Total Yield and Size Distribution. ............................... 115 Summary and Conclusion ......................................................................................................... 116 6 EFFECTS OF WATER AND NITROGEN STRESS ON FRUIT GROWTH AND TOMATO YIELD ..................................................................................................................... 136 Introduction ............................................................................................................................... 136 Materials and Methods .............................................................................................................. 138 Results ........................................................................................................................................ 141 Water Stress ....................................................................................................................... 141 Ni trogen Stress ................................................................................................................... 143 Growth of Individual Fruits under Water Stress .............................................................. 144 Growth of Individual Fruits under Nitrogen Stress ......................................................... 145 Discussion .................................................................................................................................. 146 Summary and Conclusion ......................................................................................................... 149 7 CROPGRO TOMATO SIMULATION OF INDIVIDUAL FRUIT GROWTH UNDER WATER AND NITROGEN LIMITATIONS. ........................................................................ 167 Introduction ............................................................................................................................... 167 Materials and Methods .............................................................................................................. 170 Results and Discussion ............................................................................................................. 173 Water Stress ....................................................................................................................... 173 Nitrogen Stress ................................................................................................................... 177 Conclusion ................................................................................................................................. 180 8 CONCLUSIONS ....................................................................................................................... 197 LIST OF REFERENCES ................................................................................................................. 200 BIOGRAPHICAL SKETCH ........................................................................................................... 221

PAGE 8

8 LIST OF TABLES Table page 3 1. Parameters represent temperature dependence of tom ato phenology, fruit and seed growth, default V4.0 values and updated values in specie file. .......................................... 76 3 2. Additional modified parameters in ecotype file, default V4.0 values and calibrated values ..................................................................................................................................... 77 3 3. Tomato plant age at transplanting in CROPGRO default V4.0 values and calibrated values in experimental file (X). ............................................................................................. 77 3 4. Cultivar coefficients in CROPGRO Tomato model, default and calibrated values after updating the cardinal temperatures in SPE file. Standard cultivar. ..................................... 78 3 5. Cultivar coefficients in CR OPGRO Tomato model, default and calibrated values after updating the cardinal temperatures in SPE file. Cultivar: Solarset. .................................... 79 3 6. Root mean square error and Willmott d index for simulations using default genetic parameter values. .................................................................................................................... 79 3 7. Root mean square error and Willmott d index for simulations using updated genetic parameter values. .................................................................................................................... 80 4 1. Estimated coefficients solved by fitting the 2007 data to a Gompertz function to predict DW, FW and fruit diameter. .................................................................................................. 97 4 2. Means comparisons among cohorts for dry weight, fresh weight, dry matter concentration and fruit diameter of individual tomato fruits, measured during the spring of 2006 and 2007 at Gainesville, FL. ........................................................................ 97 5 1. Observed vs. CROPGR O simulated final dry weight, fresh weight, dry matter concentration and fruit size of individual tomato fruits, tagged at three different dates. Observed data came from calibration experiment conducted at Gainesville, Fl during spring in 2007. .......................................................................................................... 119 5 2. Estimated coefficients solved by fitting data of the experiment 2007 to equation 1 to predict fruit DMC over time for the three cohorts and to equation 3 to predict fruit diameter, (Infostat software). .............................................................................................. 119 5 3. Observed vs. CROPGRO simulated dry weight, fresh weight, dry matter concentration and fruit size of individual tomato fruits, tagged at three different dates. Observed d ata came from validation experiment conducted at Gainesville, Fl during spring in 2006. ...................................................................................................................................... 120 6 1. Standard size classes for fresh tomatoes in the U.S ............................................................... 151

PAGE 9

9 6 2. Means comparison among three treatments for tomato leaf gas exchange and N leaf, stem and fruit concentration in Gainesville Fl, during spring of 2007. ............................ 151 6 3. Means comparison among three treatments for tomato total above biomass, total fruit dry and fresh fruit yield, harvest index, fruit number and maximum LAI at harvest date on Jun 25 (NS), Jun 28 (WT) and Jul 05 (CT), in Gainesville, Fl, during spring o f 2007. ................................................................................................................................. 151 6 4. Means comparison among three treatments of growth variables of individual tomato fruit at harvest in Gainesville Fl, during spring of 2007. ................................................... 152 6 5. Contribution of each commercial fruit size class to total yield fruit at harvest in Gainesville Fl, during spring of 2007. ................................................................................ 153 7 1. Physical, chemica l and water holding capacity characteristics of soil series Tavares, (uncoated sand soil). Source Carlisle et al. (1988). .......................................................... 182

PAGE 10

10 LIST OF FIGURES Figure page 3 1. Simulated and observed Leaf Area Index for tomato standard cultivar in Bradenton, Fl during spring of 1991 and 1992 using updated parameters (Table 34). ............................ 81 3 2. Simulated and o bserved Leaf Area Index for tomato standard cultivar in Bradenton, Fl during spring of 1994 and 1995 using updated parameters (Table 34). ............................ 81 3 3. Simulated and observed Leaf Area Index for tomato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using updated parameters (Table 3 4). ....................... 82 3 4. Simulated and observed Leaf Area Index for tomato in Quincy, Fl durin g 1995 standard cultivar (spring) and S olarset cultivar (fall) using updated parameters (Table 3 4). ............................................................................................................................. 82 3 5. Simulated and observed total above biomass for tomato standard cultivar in Bradent on Fl, during spring of 1991 and 1992 using updated parameters (Table 3 4). ...................... 83 3 6. Simulated and observed total above biomass for tomato standard cultivar in Bradenton, Fl during spring of 1 994 and 1995 using updated parameters (Table 3 4). ....................... 83 3 7. Simulated and observed total above biomass for tomato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using upda ted parameters (Table 3 4). ............................................................................................................................................. 84 3 8. Simulated and observed total above biomass for tomato in Quincy, Fl during 1995 standard cultivar (spring) and S olarset cultivar (fall) using updated para meters (Table 3 4 and 35). ................................................................................................................ 84 3 9. Simulated and observed total fruit dry weight for tomato standard cultivar in Bradenton, Fl during spring of 1991 and 1992 using updated parameters (Table 3 4). .... 85 3 10. Simulated and observed total fruit dry weight for tomato standard cultivar in Bradenton, Fl during spring of 1994 and 1995 using updated parameters (Table 3 4). .... 85 3 11. Simulated and observed total fruit dry weight for tomato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using updated parameters (Table 3 4). ............................................................................................................................................. 86 3 12. Simulated and observed total fruit weight for tomato in Quincy, Fl during 1995 standard cultivar (spring) and S olarset cultivar (fall) using updated parameters (Table 3 4 and 35). ................................................................................................................ 86 3 13. Simulated and observed fruit number for tomato standard cultivar in Bradenton, Fl during spring of 1991 and 1992 using updated parameters (Table 34). ............................ 87

PAGE 11

11 3 14. Simulated and observed fruit number for tomato standard cultivar in Bradenton, Fl during spring of 1994 using updated parameters (Table 3 4). ............................................ 87 3 15. Simulated and observed frui t number for tomato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using updated parameters (Table 34). ............................ 88 4 1. Individual fruit growth curves (g dry matter per fruit) over time in 2007, fitted to a three parameter Gompertz function ( a mean of four fruits. .............................................................................................................. 98 4 2. Individual fruit growth curves (g fresh mass pe r fruit) over time in 2007, fitted to a three parameter Gompertz function ( a mean of four fruits. .............................................................................................................. 98 4 3. Individual fruit growth curves (fruit diameter in cm) over time in 2007, fitted to a three parameter Gompertz function ( a mean of four fruits. .............................................................................................................. 99 4 4. Individual fruit growth curve s (dry matter concentration) over time in 2007, fitted to a four parameter Gompertz function with displacement ( .................................................................. 99 4 5. Individual fruit growth rate (g dry matter per day) over time, obtained by fitting the 2007 data to a three-parameter Gompertz function and calculating the first derivative of the function. ..................................................................................................................... 100 5 1. Predicted vs. observed dry weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 121 5 2. Predicted vs. observed dry weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 121 5 3. Predicted vs. observed dry weight of individual fruits of cohort # 3 in Gainesville Fl during spring 2007 (each point is a mean of four fruits). .................................................. 122 5 4. Fruit dry matter concentration over time fitted to a four parameter CROPGRO function in 2007 ( ..... 122 5 5. Predicted vs. observed dry matter concentration of individual fruits, cohort # 1 (each point is a mean of four fruits) in Gainesville Fl, during spring 2007. .............................. 123 5 6. Predicted vs. observed dry matter concentration of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 (each poi nt is a mean of four fruits). ......................... 123 5 7. Predicted vs. observed dry matter concentration of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 (each point is a mean of four f ruits). ......................... 124

PAGE 12

12 5 8. Predicted vs. observed fresh weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 124 5 9. Predicted vs. observed fresh weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 125 5 10. Predicted vs. observed fresh weight of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 125 5 11. Predicted vs. obse rved fruit diameter of individual fruits cohort # 1 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 126 5 12. Predicted vs. observed fruit diameter of individual fruits cohort # 2 in Gainesville Fl, during spring 2007 (each point is a mean of four fruits). .................................................. 126 5 13. Predicted vs. observed fruit diameter of individual fruits cohort # 3 in Gainesville F l, during spring 2007 (each point is a mean of four fruits). .................................................. 127 5 14. Predicted vs. observed dry weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2006 (each point i s a mean of four fruits). .................................................. 127 5 15. Predicted vs. observed dry weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). .................................................. 128 5 16. Predicted vs. observed dry weight of individual fruits of cohort # 3 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). .................................................. 128 5 17. Predicted vs. observed dry matter concentration of individual fruits of cohort # 1 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). ......................... 129 5 18. Predicted vs. observed dry matter concentration of individual fruits of cohort # 2 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits. .......................... 129 5 19. Predicted vs observed dry matter concentration of individual fruits of cohort # 3 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). ......................... 130 5 20. Predicted vs. observed fresh weight of individual fruits of cohort #1 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). .................................................. 130 5 21. Predicted vs. observed fresh weight of individual fruits of cohort #2 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). .................................................. 131 5 22. Predicted vs. observed fresh weight of individual fruits of cohort #3 in Gainesville Fl, during sprin g 2006 (each point is a mean of four fruits). .................................................. 131 5 23. Predicted vs. observed fruit diameter of individual fruits cohort # 1 in Gainesville Fl, during spring 2006 (each point is a mean of fo ur fruits). .................................................. 132

PAGE 13

13 5 24. Predicted vs. observed fruit diameter of individual fruits cohort # 2 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). .................................................. 132 5 25. Predicted vs. observed fruit diameter of individual fruits cohort # 3 in Gainesville Fl, during spring 2006 (each point is a mean of four fruits). .................................................. 133 5 26. Predicted vs. observed total fruit dry weight in Gainesville Fl, during spring 2006 and 2007. ...................................................................................................................................... 133 5 27. Predicted vs. observed total fruit fresh weight in Gainesville Fl, d uring spring 2006 and 2007. ............................................................................................................................... 134 5 28. Prediction over time of fruit fresh weight as contributed by each marketable size class of tomatoes in Gainesville Fl, during spring 2007. ............................................................ 134 5 29. Predicted vs. observed total tomato fruit fresh weight contributed by each marketable size class at harvest in Gainesville Fl, during spring of 2007. .......................................... 135 6 1. Leaf water potential of mature tomato leaves under three treatments in Gainesville Fl, during spring of 2007. .......................................................................................................... 154 6 2. Daily soil water content based on 20 minute values from TDR probes buried at 20 cm depth, in experiments conducted in Gainesville Fl, during spring of 2007. ..................... 154 6 3. Daily soil water content based on 20 minute values from TDR probes buried at 40 cm depth, in experiments conducted in Gainesville Fl, during spring of 2007. ..................... 155 6 4. Photosynthesis rate of mature tomato leaves under three treatments in Gainesville Fl, during spring of 2007; symbols represent means of 16 leaves per treatment. ................. 155 6 5. Stomatal resistance of mature tomato leaves under three treatments in Gainesville Fl, during spring of 2007 (symbols represent means of 16 leaves per treatment. ................. 156 6 6. Effects of water and nitrogen on Leaf Area Index (LAI) over time of tomato plants in Gainesville Fl, during spring of 2007. ................................................................................ 156 6 7. Effects of water and nitrogen stress on Specific Leaf Area (SLA) over time of tomato plants in Gainesville Fl, during spring of 2007. ................................................................. 157 6 8. Effects of water and nitrogen stress on total above biomass over time of tomato plants in Gainesville, Fl, during spring of 2007. ........................................................................... 157 6 9. Effects of water and nitrogen st ress on total fruit dry weight of tomato plants over time in Gainesville, Fl, during spring of 2007. ........................................................................... 158 6 10. Effects of water and nitrogen stress on total fresh yield of tomato plants over time in Gainesville, Fl, during spring of 2007. ............................................................................... 158

PAGE 14

14 6 11. Leaf blades N concentration over time of tomato plants under N and water stress grown in Gainesville, Fl, during spring of 2007. ............................................................... 159 6 12. Stem N concentration over time of tomato plants under N and water stress grown in Gainesville, Fl, during spring of 2007. ............................................................................... 159 6 13. Fruit N concentration over time of tomato plants under N and water stress grown in Gainesville, Fl, during spring of 2007. ............................................................................... 160 6 14. Effects of water and nitrogen stress on dr y weight accumulation over time in individual tomato fruits for cohort number 1 in Gainesville, Fl, during spring of 2007. ...................................................................................................................................... 160 6 15. Effects of water and nitrogen stress on fresh weight accu mulation over time in individual tomato fruits for cohort number 1 in Gainesville, Fl, during spring of 2007. ...................................................................................................................................... 161 6 16. Effects of water and nitrogen stress on dry weight accumulation over t ime in individual tomato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007. ...................................................................................................................................... 161 6 17. Effects of water and nitrogen stress on fresh weight accumulation over time in indivi dual tomato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007. ...................................................................................................................................... 162 6 18. Effects of water stress on dry weight accumulation over time in individual tomato fruits for cohor t number 3 in Gainesville, Fl, during spring of 2007. ............................... 162 6 19. Effects of water stress on fresh weight accumulation over time in individual tomato fruits for cohort number 3 in Gainesville, Fl, during spring of 2007. ............................... 163 6 20. Effects of water and nitrogen stress on dry matter concentration over time in individual tomato fruits fro cohort number 1 in Gainesville, Fl, during spring of 2007. ...................................................................................................................................... 163 6 21. Effects of water and nitrogen stress on dry matter concentration over time in individual tomato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007. ...................................................................................................................................... 164 6 22. Effects of water stress on dry matter concentration over time in individual tomato fruits for cohort number 3 in Gainesville, Fl, during spring of 2007. ............................... 164 6 23. Effects of water and nitrogen stress on the diameter over time of individual tomato fruits for cohort number 1 in Gainesville, Fl, during spring of 2007. ............................... 165 6 24. Effects of water and nitrogen stress on the diameter over time of individual tomato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007. ............................... 165

PAGE 15

15 6 25. Eff ects of water stress on the diameter over time of individual tomato fruits for cohort number 3 in Gainesville, Fl, during spring of 2007. .............................................. 166 7 1. CROPGRO simulated (lines) and measured (poi nts) leaf area index in Gainesville, FL during spring of 2007 comparing well irrigated with water stressed tomato plants. ....... 183 7 2. CROPGRO simulated (lines) and measured (points) total above biomass in Gainesville, FL during spring of 2007 comparing well irrigated with water stressed tomato plants. ....................................................................................................................... 183 7 3. CROPGRO simulated (lines) and measured (points) total fruit dry weight in Gainesville, FL during spring of 2007 comparing well irrigated with water stressed tomato plants. ....................................................................................................................... 184 7 4. CROPGRO simulated (lines) and measured (points) total fruit fresh weight in Gain esville, FL during spring of 2007 comparing well irrigated with water stressed tomato plants. ....................................................................................................................... 184 7 5. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohor t # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 185 7 6. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 185 7 7. CROPGRO simulated (line) vs. observed (points) dry weight of ind ividual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 186 7 8. CROPGRO simulated (line) vs. observed (points) dry ma tter concentration of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ....................................................................... 186 7 9. CROPGRO simulated (line) vs observed (points) dry matter concentration of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ....................................................................... 187 7 10. CR OPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ....................................................................... 187 7 11. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 188

PAGE 16

16 7 12. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed plants. ...................................................................................................................... 188 7 13. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 189 7 14. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 189 7 15. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plan ts. ......................................................................................................... 190 7 16. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. ......................................................................................................... 190 7 17. CROPGRO simulated (lines) and measured (points) leaf area index in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants. ...... 191 7 18. CROPGRO simulated (lines) and measured (points) above ground biomass in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants. ....................................................................................................................... 191 7 19. CROPGRO simulated (lines) and measured (points) total fruit dry weight in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants. ....................................................................................................................... 192 7 20. CROPGRO simulated (lines) and measured (points) total fruit fresh weight in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants. ....................................................................................................................... 192 7 21. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants. ......................................................................................................... 193 7 22. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants. ......................................................................................................... 193 7 23. CROPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants. ............................................................................. 194

PAGE 17

17 7 24. CROPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants. ............................................................................. 194 7 25. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing we ll fertilized with N stressed tomato plants. ......................................................................................................... 195 7 26. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants. ......................................................................................................... 195 7 27. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 1 in Gainesville Fl, during spri ng 2007 comparing well fertilized with N stressed plants ....................................................................................................................... 196 7 28. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 2 in Gainesville Fl, during spri ng 2007 comparing well fertilized with N stressed plants ....................................................................................................................... 196

PAGE 18

18 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degre e of Doctor of Philosophy IMPROVING A TOMATO GROWTH MODEL TO PREDICT FRESH WEIGHT AND SIZE OF INDIVIDUAL FRUITS By Maria Raquel Rybak May 2009 Chair: Kenneth J. Boote Major: Agronomy A new version of the CROPGRO Tomato model was developed to include a subroutine for predicting dry matter concentration, fresh weight and size of fruits over time. This subroutine builds on already existing crop model predictions of dry mass p er fruit and fruit thermal age. The model was calibrated with field data obtained on tomato cultivar Florida 47 in Gainesville Florida during spring 2007 and evaluated with an independent data set obtained during spring 2006. The dynamics of dry weight accumulation, fresh weight, dry matter concentration, and fruit size were simulate d by the model and compared with independent data of fruits tagged at anthesis in the field study at three successive weekly dates. These results demonstrated that the model was able to explain and predict the time -series growth and development of individu al fruit s on a cohort basis, to include delayed growth, slower growth, and smaller size of progressively later -set fruits. Overall, the standard deviation of model error was less than 20 percent of the mean for all the variables evaluated (dry weight, fres h weight, fruit dry matter concentration, and fruit size). In addition, the Willmot index was always above 0.9. The effects of water and N stress on the growth of individual fruits were studied by withholding drip irrigation for three periods and withho lding N supply beginning at 14 days after transplanting. The main effects of water stress on the growth of individual fruits were reduced

PAGE 19

19 fresh mass and size and an increase in dry matter concentration, while N stress on the other hand, caused a decrease i n dry matter concentration of fruits. The existing water and N stress factors in the CROPGRO model were linked to the new equations that affect dry weight growth and dry matter concentration of individual fruits, and were successfully used to predict the s tress -induced variations in dry matter concentration, and reduction s in dry and fresh weight growth and size of single fruits The newlyadded subroutine for predicting dry matter concentration, fruit fresh weight, and size, along with coupling to water an d N stresses make the improved model a valuable risk assessment tool for predicting fresh market production and quality of tomato.

PAGE 20

20 CHAPTER 1 INTRODUCTION Tomato production in Florida contributes greatly to the economy of the state. The total production i n the US for the 200607 season totaled 13,524,530 million tons (11,484,800 tons processing tomatoes and 1,679,730 tons fresh market) from 175,150 hectares planted (126,910 hectares processing and 48,240 hectares fresh market). Florida accounted for 39 per cent of the total U.S. tomato fresh production, followed by California with 33.2 percent. Eleven other states such as Georgia, Virginia and Ohio, accounted for the remaining 27.8 percent (Foolad, 2007). The value of the 200607 U.S. tomato crop was $2.05 billion, and of this $1. 277 bill i on corresponded to fresh market, as published by the USDAs Statistical Service (Foolad, 2007). The yield of vegetables for fresh market is usually defined by plant/fruit fresh weight, which is directly related to dry weight production (Marcelis et al ., 1998). However, the relationship between growth in fresh matter and dry matter is poorly understood (Marcelis and Gijzen 1998). A general process of prediction of fresh weight comparable to the photosynthesis -based process u sed in mechanistic models for dry matter growth, is needed (Heuvelink et al ., 2004) Several mo dels using tomatoes have focused o n the carbon b alance that leads to prediction of fruit growth on the basis of dry mass. Most of these are used for the purpose of growth and development studies and for the control of greenhouse environ ments. Some examples are TOMGRO (Jones et al ., 1991), TOMSIM (Heuvelink and Bertin, 1994), TOMPOUSSE (Gary et al ., 1997) and CROPGRO, in which tomato was one of the species included (Scholberg et al., 1997). When fresh vegetables are the crops to be modeled, the photosynthesis -based models need to be modified, taking into consideration the high water content of these crops. For fresh

PAGE 21

21 vegetables, the flow of water and carbon into indi vidual harvestable organs dictates growth. However, most models simulate water relationships for the whole canopy based on the hydric balance of the soil -plant atmosphere continuum Moreover, fresh vegetables such as indeterminate tomato have cyclic patter ns of yield influenced by the environment and this variability has not been well modeled despite the potential practical application. On the other hand, the photosynthesis driven models like CROPGRO have many advantages. Among these advantages is the fact that CRO PGRO has been validated sufficiently for use in Florida. This is important considering that Florida is the leading producer s t ate of fresh tomatoes in the USA, and CROPGRO has robust mechanistic support. It is desirable to know how, starting with the carbon balance, models like CROPGRO can be improved in order to develop practical applications for growers concerned with the prediction of fresh weight, graded size, and weekly yield variabil ity of fresh market crops like tomato. The general purpose o f this work is to improve the existing CROPGRO -tomato model through the fulfillment of the following objectives: T o update parameters in CROPGRO based on recent published literature. To investigate the functional relationships between fresh weight and dry weight growth of tomato fruits under different growth conditions. To develop a sub-model to simulate the fresh weight and the size of individual fruits on a cohort basis, and to validate the model with independent data.

PAGE 22

22 CHAPTER 2 LITERATURE REVIEW Int roduction Mechanistic crop models are valuable tools of research and can be used to integrate most of our knowledge of crop behavior (Acock and Reynolds, 1989; Heuvelink, 1995). In addition, validated models are well suit ed for predict ing harvest tim e, a p iece of information which is essential for optimal crop management and commercial planning strategies. Although CROPGRO Tomato is one of the most complete models developed for tomato since it was developed more than 10 years ago, no further improvements or validations have been done. In this chapter a bibliographical review of research related to how environmental factors such as radiation, temperature, water and nutrients influence the growth and development of tomato is presented. The focus is on those functional relationships that could be useful for improving tomato modeling. In addition, genetic traits related to tomato yield are briefly reviewed. Moreover, a short section is devoted to describe the state of art of tomato models and related topics. Gene ral aspects about the specie s are not covered but some comprehensive text books are available that describe deeply the taxonomic, morphological and physiological traits of the specie s as well as production systems. Examples are: The tomato crop: a scientif ic basi s for improvement (Ath erton and Rudich, 1986), Cultivo del Tomate (Nuez, 1995) and more recently Crop Prod uction Science in Horticulture: Tomatoes (Heuvelink 2005). Influence of the Environment on Growth and Development of Tomato Plants Light Light interception: The light is the most important factor affecting the productivity of tomato and a complete review was written by Papadopoulos and Parajasingham (1997). Heuvelink (1995) found that the average tomato crop growth rate is highly related (R2 = 0 .87; p

PAGE 23

23 < 0.001) to the photosynthet ically active radiation (PAR). From a practical point of view, the incident light may be less important in open field, where normally there are less light limitations as opposed to greenhouses; this is because the tomato specie is affected by chilling injuries and therefore shorter days typical o f winter are normally avoided. Consequently, when the tomato is grown in open field, issues regarding the light interception may be more important than the incident radiation per se. The light measurement for photosynthesis is expressed as photosynthetic photon flux density (PPFD) in the range wavebands corre sponding to 400700 nm (Gardner 1985). For calculating the intercepted radiation it is accepted that light is exponentially attenuated within a canopy according to a function which derives from optical laws. This is expressed in general as I/I0 = e kLAI where I and I0 are the total PPFD within and above the canopy, LAI is the leaf area index above I and k is t he extinction c oefficient (Gijzen, 1992), Acock et al. (1978) found that the k value for tomato is equal to 0.63 for the upper leaf layers and the stem and it decreases to 0.52 dee p in the canopy. Scholberg et al. (2000) reported that k values dropped from 1 early in th e season to about 0.2 when the fully canopy was developed. The idea of making k variable throughout the season in tomato models has not been incorporated, and most models use a typical k value equal to 0.75 (CROPGRO) or 0.72 (TOM SIM) or some value in betw een. Therefore, performing sensitivity studies about k could be useful in order to improve the s imulation of intercepted light. The concept of LAI is normally applied for estimating the PPFD interception but the tomato canopy is usually structured in row a rrangements. This has important consequences on light interception. In the first place, the edge of the canopy row receives additional PPFD from the sides ( Gijzen, 1992), second the canopy is confined to discrete rows so different leaves at a

PAGE 24

24 given depth c an receive different amounts of radiation (Longuenesse et al., 1993). Last, the practice of tying the plants may limit its expansion and not all the assigned space is truly occupied by plants. Wilson et al (1992) measured the light interception of a tomat o canopy growing in greenhouse. They found that the interception of incident PPFD on the top of the canopy was 76 % while about 20 % was lost through gaps between rows. Simultaneously an average of 13 % reflectance and 23 % transmittance of the incident PP FD by the tomato canopy was determined. Several experiments about light penetration into a tomato canopy as a function of height for different plant spacing were carried o ut by Papadopoulos and Omrod ( 1988a) According to their results the proportion of av ailable PPFD intercepted increased with closer spacing but the vertical distribution of PPFD was better (more uniform) with wider plant spacing. Moreover, the effects of plant spacing were less important in spring than in autumn and they attributed these results to different growth stages of the crop which impact the extinction coefficient k. The authors argued that closer spacing was only important in the early part of the season but when enough LAI is developed the crop will intercept most of the availa ble PPFD regardless of the initial spacing. Harper et al (1979) found that tomato plants planted in autumn intercepted 30 40 % of incident radiation; by contrast, about 70 % of the radiation was intercepted by the plants in spring. Also, Pa pa dopoulos and Omrod (1988a ) found a close r relationship between LAI and PPFD interception than directly to plant density; therefore, interpretation of the Harper et al. (1979) data is difficult because they did not separate the effect of the season on the growth stage of the plant from the plant spacing effect. Scholberg et al. (2000) reported that even with fully developed canopy, field grown tomato only intercepted 50 to 60 % of the incident light due to the wide space between rows which resulted in an incomplete ground cover. Efforts have been made to improve the simulation of PPFD interception when the

PAGE 25

25 crops are grown in rows. The concept of effective LAI for example was used by Acock (1991) where the effective LAI is the ratio between crop leaf area and shaded area a nd w h ere the shadow area is simulated as a function of the shadow length measured at right angles to the row, the height of the crop row and the apparent solar elevation at right angles to the row ( Gijzen 1992). The advantage of this approach is that it is simple but it does not account for the PPFD scattering within the canopy so the light distribution is not well simulated. A more sophisticated improvement was proposed by Boote and Pickering (1994) to simulate PPFD interception by row crops. They incorporated the concepts of sun leaves (intercepting direct solar radiation) and shade leaves (intercepting scattered radiation originated from non -sun angles). The shadow projected by the canopy according their model is a function of the canopy dimensions (hei ght and wide), latitude and azimuth angle, day of the year and time of the day. Using this shadow projection LAI is allowed to intercept PPFD and is assumed that the soil shaded by the plant canopy is the ground area occupied by the plant for photosynthesi s. Likewise, the Gijzen and Goudriaan (1989) model was developed for simulating light distribution and photosynthesis in row crops Photosynthesis: The reported relationship between solar radiation and net photosynthesis is that photosynthesis initially in creases linearly with irradiance level until the relationship become s asymptotic (Papadopoulos, 1985; Longuenesse et al, 1993). The saturation level for photosynthesis inside greenhouse is 170 W m2 (Papadopoulos and Omrod, 1988b ) or ranges between 100 to 150 W m2 (Cockshull, 1988) respectively. Outdoors the values of saturating irradiance for photosynthesis reach 210 W m2, likely higher because of a higher CO2 concentration. Furthermore, Gijzen (1995) determined that for closed canopies photosynthesis did not saturate up to 2000 mol m2 s1. Acock et al. (1978) reported that in tomato the upper

PAGE 26

26 layer, which represented 23 % of the leaf area, assimilated the 66 % of the net CO2 fixed by the canopy. In addition, a reduction from 8.6 to 5.2 in the leaf a rea index did not affect significantly the rate of net photosynthesis. However, when LAI was reduced to 2 (removing the middle leaves), the canopy photosynthesis was reduced by one third and the da rk respiration by one quarter. These authors also measured the photosynthesis rate of individual leaves at upper, middle and lower layers of the canopies and concluded that the differences should be attributed to different amounts of light intercepted and not to leaf age differences. Papadopoulos and Omrod (1988b) reported that tomato plant spacing had a negligible effect on net photosynthesis of upper leaves but net photosynthesis of lower leaves was reduced for plants at a narrow spacing. In addition, photosynthesis is not a linear function of irradiance but rather a non linear function with dependence on the nitrogen concentration of the leaves which have a saturating response to increase d nitrogen concentration except at high irradiance level. This reveals the role of N on the Calvin cycle enzymes and the t hylakoids which represent the majority of the leaf N content (Dewar, 199 6) Growth: The crop growth rate of tomato is directly related to the net assimilation rate and the leaf area index. The crop growth rate (CGR) is defined as the product of the net ass imilation rate multiplied by the leaf area index (LAI) (Hunt, 1982). The PPFD interception is closely related to an early development of LAI. A fast covering of ground by leaves in tomato rows can be accomplished by optimizing the plant population (Harper et al., 1979; Papadopoulos and Omrod, 1990). According to Scholberg et al (2000), tomatoes growing in open field reached a maximum LAI equal to 3.8 about 11 weeks after transplanting. The near optimal light interception occurred about 4 weeks after trans planting for plants spaced 45 cm in the row while at 60 cm the near optimal interception occurred 6 weeks after transplant ing G rowth can also be

PAGE 27

27 expressed as relative crop growth rate (RGR) which is the rate of gain in biomass in relation to the total bio mass. The RGR is the product of the net assimilation rate (NAR) multiplied by the leaf area ratio (LAR), which in turn is defined as the leaf area for a given total biomass available for photosynthesis or leaf leafiness (Hunt, 1982). Bruggink and Heuvelink (1987) determined that for tomato the relationship between PPFD and RGR is well represented by a rectangular hyperbola and therefore at low daily light integrals the RGR will decrease. They simulated the growth of tomato plants and found that increas ing N AR increased RGR. Hurd and Thornl e y (1974) observed that high light integrals induced a decline in specific leaf area (SLA). They determined a SLA equal to 0.069 m2 g1 at 0.58 MJ m2 d ay1 and equal to 0.015 m2 g1 at 9.5 MJ m2 day1 in plants of 21 to 3 0 days age, and attributed these results to a differential biomass partitioning within leaves at high light integrals. In other words if the leaf thickness increases due a low sink source relationship the result will be a lower SLA. Similarly, Gary et al (1993) found a negative correlation between tomato SLA and the mean light integral of the previous 3 days. In commercial practices De Koning (1993) measured a LAI equal to 3 in May but then he observe d a linear decrease until it reached 1 in October. The S LA also decreased from 250 cm2 g1 in full grown leaves during spring to 100 cm2 g1 in summer. An average value of SLA for tomato equal to 196 cm2 g1 was reported by Heuvelink (1995). The seasonal variation in SLA is difficult to explain because several factors influence it at the same time. Ontogenetic changes, environmental influences (light, temperature, CO2) and sin k source effects are involved in these variation s (Gijzen ., 1992). The SLA can be expressed also as the inverse which is called specific leaf weight (SLW or 1/SLA) in g m2. The SLW according to Papadopoulos and Omrod (1988b ) decreases when the plant spacing is reduced. They found that the tomato SLW was 59 and 29 g m2 in plants spaced at 45 and 23 cm respectively for plants

PAGE 28

28 inside green house. In plants outdoors and spaced at 30 cm the SLW was 32 g m2. According to these results, increasing the integral radiation in tomato can increase RGR via its effects on SLW of new leaves, which in turn increases NAR Yield: The slope of the relationship between crop biomass and intercepted light is usually referred as radiation use efficiency (RUE) which on average has a value equal to 2.5 g MJ1 (PAR) for tomato grown in greenhouse (De Koning, 1993). The RUE is equal to 1.05 g dry weight MJ1 m2 (t otal radiation) for tomato growing in open field (Scholberg et al ., 2000). According to Cockshull et al. (1992), the tomato fruit yield is in direct proportion to the accumulated solar radiation and 2.01 kg of fresh fruit weight are produced per 100 MJ of radiation incident on the crop. De Koning (1989a ) found similar values reporting radiation use efficiency equal to 2.07 kg of tomato fruit per 100 MJ of incident radiation. The theoretical RUE is 1.0 g dry mass MJ1 of global radiation, which equals to 3. 1 g dry mass MJ1 of PAR inside a greenhouse (Challa and Bakker, 1998). In addition, values of 2.37 and 2.5 g dry mass MJ1 of PAR in greenhouse were reported by Heuvelink (1995) and Sandri et al. (2003). The last authors reported that when tomato plants w ere grown at two irradiance levels (12.4 MJ m2 day1 and 5.0 MJ m2 day1) the number of fruits per square meter did not differ significantly between treatments. However the dry matter from the last harvest of unshaded and shaded plants, di ffered signifi cantly being 975 g m2 and 762 g m2 for total dry mass, 550 g m2 and 420 g m2 g for fruit dry mass, and 425 g m2 and 343 g m2 for vegetative organ dry mass respectively. The dry matter partitioning is not largely impacted by PPFD but it affects th e total available assimilates for distribution. Cockshull et al (1992) grew tomato in low shaded (6.4% light reduction) and highly shaded environments (23.4% light reduction) compared to the control. They found that the reduction in fruit fresh weight wa s 7.5 % and 19.9%, respectively.

PAGE 29

29 The total above ground biomass was reduced by 6.2% and 16.5% respectively. They also reported that over 14 weeks of harvest the dry weight accumulation was in direct proportion to the incident PPFD and regardless of the tr eatment 2.01 kg of fresh weight of fruit was harvested per 100 MJ received. The yield of tomato fruits depends on the number and the weight of the fruits. Cockshull et al. (1992) reported that fruit number per truss was positively correlated with the recei ved radiation and a stronger effec t was observed when less than 1. 5 MJ m2 day1 was received at the time of anthesis of the first truss. Indeed, it has been repeatedly reported that the major influence of irradiance on tomato fruit yield is through its ef fects on the number of fruits that reach a marketable size. This effect is basically due to a high proportion of fruit abortion when the sink/source ratio is increased if low light integrals are accumulated at the time of flower initiation. Papadopulos and Ormrod (1990) reported that the plant spacing affected the percentage of tomato fruit set more markedly in spring than on autumn. For instance, t hey found that plant spacing of 60, 45 and 23 cm in autumn resulted in fruits sets of 62, 64, and 46%, respect ively; whereas the same plant spacings in the spring led to fruit sets of 58%, 52%, and 13% Heuvelink (1995) reported that the plant density had no direct effect on biomass allocated to the fruits but the effects were indirect because it changed the frui t number (sink capacity) and that finally resulted in differences in bi omass allocated to the fruits. Pe a rce et al. (1993) found that the rate of fruit growth was maintained during at least the first 20 hours after a tomato plant was placed on darkness. Th is suggests that current photosynthesis does not limi t the rate of fruit expansion. L ight and carbon dioxide affect the yield pattern in tomato (Adams et al ., 2001a; Adams and Valdez 2002). These authors found that when light levels were increased for a s hort time, there was no immediate effect on the yield. Yields subsequently increased, particularly four to

PAGE 30

30 six wee ks after the treatment started. This increase was mainly due to an increase in mean fruit size. Similarly, decreasing CO2 concentration at wee ks 24 to 26 caused depressed yields from weeks 28 to 31. Light levels and CO2 concentrations affect photosynthesis and consequently the amount of assimilates available. Changes in the amount of available assimilate greatly affect fruits that are growing ra pidly at that time. Newly set or nearly mature fruits will not be affected as much because their growth rates are low. Fruit set can be affected by the availability of assimilates and the fruit load on the plant Individual fruit growth: The cumulative gro wth pattern of tomato fruit is sigmoidal and follows three phases (Ho and Hewitt, 1986). According to Wang et al. (1993) there is an initial slow growing period from approximately 0 to 10 days after anthesis (DAA), then a tomato fruit accumulates most of i ts dry matter from 10 to 40 DAA. Subsequently, about 10 days after the first break of color the importation of carbohydrates ceases totally (Ho and Hewitt, 1986). Finally there is a maturation period characterized by biochemical changes that leads to the fruit maturation. The fruit size is primarily determined by the cultivar. The fruit size also depends on its position on the truss, the competition for assimilates with other sinks and the water availability. The competition for assimilates depends on the number of fruits set on each truss, the number of plants per square meter and the duration of fruit growth. Bailey and Hunter (1988) and Cockshull et al. (1992) reported that under shade the proportion of tomato fruits with small size increased. They att ributed this decrease in fruit size to a lower assimilate supply in plants growing under shade which was proportionally reduced more than the reduction in fruit number under the same shade condition. Cockshull and Ho (1995) found that tomato plants grown a t higher density produced more fruits per square meter. They attributed these results to a higher LAI for radiation interception in plants grown at high density compared with low density.

PAGE 31

31 The overall weight of marketable fruit produced per plant however, f ell from 19.7 kg at low density to 15.1 kg at high density. Papadopoulos and Omrod (1990) found a linear decrease in fruit weight (g fruit1) when increasing the plant density. Rates of fruit growth are found in the literature. Bertin (1993) for instance, measured a maximum rate of 1.04 g dry matter day1 for a beefsteak cultivar, a very high value compared with other published data. Heuvelink and Marcelis in 1989 measured a rate of 0.2 g dry matter day1 whereas values of 0.27 g dry matter day1 and 0.37 g dry matter day1 were published by Jones et al. (1991) and Ho et al (1983) respectively. In agreement with the last authors Aikman (1996) showed that a tomato fruit that takes 54 days from anthesis to maturity reached the maximum growth rate at day 23 and that rate was equal to 0.34 g dry matter day1. Bertin et al. (1998a ) studied the effects of cultivar, fruit position and seed content on the variability of the fruit weight in tomato under low and high competition for assimilates (one fruit and se ven fruits per truss). They concluded that the variability in the potential fruit weight was mainly related to the cultivar and the fruit position on the truss. On average (of the first 15 trusses) the potential weight of the first fruit was 234 g, 283 g and 391 g for long life, round, and beefsteak cultivars, respectively. Likewise, the final weight of the fifth fruit was 219 g, 226 g and 315 g respectively, that is 6 %, 20 %, and 19 % lower than for the first fruit. Increasing the number of sinks caused the weight of the first and fifth fruit to decrease at maturity regardless of the cultivar. Current tomato yield -predicting models do not take into account competi tion among sinks under low assimilate supply; this oversight could lead to prediction errors (Bertin et al. 1998). The reduction in fruit weight was not correlated with the truss position along the stem, but varied from 0 to 75 % among the first 15 trusses. Under competitive conditions, distal fruits were more affected by competition than proxima l fruits in all cultivars. According to the authors, the

PAGE 32

32 greater ability of proximal fruits to compete for assimilates under limited supply results from a faster accumulation of dry matter during the 25 days after anthesis due to the higher capacity to rec eive the phloem mobile assimilates in comparison with distal fruits. Bohner and Bangerth (1988) observed that the difference in cell number at anthesis could be responsible for fruit size at maturity. Bangerth and Ho (1984) showed that distal fruits before anthesis contain 18 % fewer ovules than proximal fruits and this may be a reason for the higher sink strength of proximal fruits. Regarding the relation between number of seeds and fruit weight they found that even when there was a correlation, this was n ot stable among cultivars and treatments Temperature Although tomatoes are grown in many parts of world in open field and greenhouse, it is rather a subtropical specie because is subject to chilling injur y when it is exposed to low temperatures above free zing for prolonged time. Tomato has been reported as responding positively to thermoperiodicity (Went, 1945). In addition, several authors have demonstrated the capacity of tomato plants to int egrate temperature (Calvert, 195 7; Hurd and Graves, 1984; De Ko ning, 1990). This means, for example, that a temperature regimen of 20/24C (night/day) can be considered similar to a constant temperature of 22C. De Koning (1990) demonstrated that tomato plants can integrate temperature over several days and under lar ge temperature amplitude s This information however needs to be carefully interpreted because although different temperatures may have the same integral and have little effect on predictions of accumulated yield, this may not be true in analyzing the fl uctuations of yield on weekly bases in which case the daily variations in temperature should be considered (Adams and Valdes, 2002). Moreover, temperature compensation is not possible for those processes with a strong interaction between temperature and some other environmental factor. The temperature has a large influence on growth and yield of the tomato crop and it is the major driving force of all the

PAGE 33

33 development phases. According to Charles and Harris (1972) and Aung (1978), the optimum temperature f or the tomato crop depends on the developmental state, cultivar and the org an being considered. Jones (1999) suggested some optimum ranges for tomato plant growth: air temperature 18 29C, rooting temperature 1824C, and canopy temperature 2023C. Challa et al (1995) gave the following ranges: germination 25/25C (day/night) and after transplanting 22/17C (day/night). Plant development: The developmental processes such as flowering, number of leaves and fruit set have a strong impact on tomato yield (Heuvelink, 2005). To predict the plant development as a function of temperature, Monteith (1977) introduced the concept of thermal time which accumulation is given by the summation of tem peratures above a threshold called base temperature (Ferreira et al ., 1997). The base temperature used for tomato is inconsistent among authors and values of 4C 6C, 7C, 8C, and 10C have been reported (Warnock and Isaacs, 1969; Folker, 1979; Calado and Portas 1987; Scholberg et al., 1997). W arnock (1973) for example, in Davis California (USA) used a base temperature of 6C to calculate the degree days accumul ated for tomato from seeding to emergence, anthesis, fruits of 2.54 cm in diameter, breaker fruit and ripe fruit. The inconsistencies in the base temperature values may have several explanations but one possible reason is that the base temperature is no t identical for all the life phases of the plant, therefore responses to temperature during the seedling phase, for example, may be different from responses during fruit setting or fruit ripening. However, one common base temperature is often considered fo r all stages of the crop. Another cause is reflected in the study published by Calado and Portas (1987). They worked at three locations during three consecutive years in Portugal and on different sowing dates. Their results, using the degree day approach by the intercept method, showed that the calculated base temperature for the same

PAGE 34

34 tomato cultivar from anthesis to harvest was equal to 6C 8C and 10C for the three locations respectively, where the base temperature increased from areas with a warm early spring to cooler ones. The hypothesis in crop modeling is that cardinal temperatures for a given specie and cultivar should hold cons tant across weather and sites. The failure to do so, is evidence of incorrect cardinal temperatures and temperature response equations. Another source that makes difficult the interpretation of published data is the methodology used to derive the cardinal temperatures. The most common method is the intercept method but the methodologies used are not always clearly established in the papers. Additionally, the method used to calculate the accumulated degree days for each phase may be different among authors. For instance, most models use the daily average temperature minus the base. CROPGRO, however, uses hourly temperature, which is obtained from the maximum and minimum temperature data plus a decay function at night that predicts the hourly temperature accur ately. The importance of these different approaches is that under the first approach even when linearity is assumed, this in reality may be not true and during a period of time the temperatures could be higher than temperatures at which a linear response i s assumed and therefore the results may be biased. Another aspect to be considered is the type of function that the model uses to account for the temperature dependence of processes. To account for the dependence on temperature on several processes, the CR OPGRO tomato model uses a four point function represented by four cardinal temperatures: base temperature (Tb, the temperature below which the rate of the process is zero), optimum one (Topt1 or temperature at which the rate reaches the maximum), optimum t wo (Topt2, highest temperature at which the rate of the process is still at it maximum) and maximum (Tmax or temperature above which the rate is zero). Therefore, CROPGRO allows a punishment or deceleration of the process when temperature is beyond Topt2 a ccounting in this way for heat

PAGE 35

35 stress that delays the processes. Other models by contrast, only consider no increase in rate beyond the opti mum temperature but there is no punishment by high temperature. The shape among this four point function in CROPGRO depends on the process. It is linear for development rate but may have different shapes for other processes (Boote and Scholberg, 2006). The default ( V 4.0) version of CROPGRO uses the same values of cardinal temperatures for the three phenologic al phases (vegetative, early reproductive and late reproductive p hases) and those values are 10C, 28C and 55C for base, optimum (both Topt1 and Topt2) and Tmax, respectively. Just as for basal temperature the maximal temperatures for tomato specie also vary according to authors and values of 48C (De Koning, 1994) and 50 C (Scholberg et al., 1997) have been published. In addition, the maximal temperature is a val ue rarely found in literature. Early vegetative and reproductive phases: According to Van der Ploeg and Heuvelink (2005), the optimum temperature for early vegetative growth of tomato plants is 25C Tomato vegetative development can be thought as the leaf rate appearance and it is also related to the truss appearance rate since the specie is a sympod io with approximately one inflorescence every three leaves (Heuv elink, 2005). De Koning (1994) showed a linear increase of leaf appearance rate with increase of average air temperature, increasing from 0.2 leaf d1 at 12 C to a maximum of 0.5 leaf d1 at 2 8C and decreasing thereafter until reaching zero at 48 C This author found that the relationship between temperature and leaf unfolding is linear in the range of 17 C to 23 C and 2.5, 3, and 3.5 leaves were unfolded each week at 17 C 20C and 23C respectively. According to Adams et al (2001b ), vegetative development in tomato has an optimum one (Topt1) equal to 22 C and an optimum two (Topt2) equal to 26 C. The base temperature for this phase accordin g to their calculation is 7 C. They did not mention maximum (Tmax) values but as proposed by De Koning this value seems to be equal to 48 C. A linear relation between

PAGE 36

36 flowering and air temperature has been observed by Abreu and Meneses (2000). Truss appearance rate increased linearly with a rate of 0.11 to 0.17 truss day1 when average temperature was raised from 17 to 23 C (De Koning, 1994). According to De Koning (1993), the number of truss per week is enhanced by temperature with 0.05 truss week1 C. Similarly Heuvelink (2005) found that in the range of 18C to 23C the truss appearance rate enhance ment is 0.001 truss day1 C. The cardinal temperatures for the early reproductive phase or progression to anthesis according to Adams et al. (2001b ) showed values similar to vegetative development having a base temperature equal to 7.2 C. In addition, they gave no data for maximum temperatures Because this data is not available from other sources the same Tmax as used for vegetative development could be assumed. Therefore 7.2 C, 22 C, 28 C and 48 C for Tb, Topt1, Topt2 and Tmax respectively seems to be reliable to use as cardinal temperatures for early reproductive development. Stem growth: The response of internode length growth to temperature, according to Langton and Cockshull (1997) followe d a very marked linear trend (p < 0.001). They exposed tomato plants growing in cabinets to a factorial of 24 combinations of day (DT) and night (NT) temperatures ranging from 12 to 32C Their experimental data for tomato growing over a 10 day period at 1 6, 20, 24, and 28C average NT and DT air temperatures, showed that the corresponding lengths reached by the internodes were 12, 18, 27 and 30 mm, respectively. Their experiments showed increased extension growth with increase in DT, and indicated that the optimum DT for extension growth is close to 28 C Similar, but less marked linear growth (P < 0.05) was shown for NT, with the optimum NT for extension growth being around 24C The DT X NT temperature i nteraction was not significant. The stem growth is n ot directly used as an indicator of development in most models, however because the response to temperature is clearly

PAGE 37

37 linear simulating more precisely the stem growth could be a path for differentiating the growth of determinate versus indeterminate culti vars Effect of temperature on NLPI: Dieleman and Heuvelink (1992) reviewed the factors affecting the number of leaves preceding the first inflorescence (NLPI) in the tomato plants. This factor is very important in tomato development because it determines the earliness of the crop and the transition from vegetative to generative growth. They stated that a lower temperature either in day or night during the sensitive period of tomato seedling growth causes a smaller NLPI independent of the cultivar or the t omato plant type. However the temperature does not affect NLPI if the light intensity is high. At low light intensity and high temperature, the NLPI increases. The effect of temperature on NLPI depends on the mean diurnal temperature while the root tempera ture has no effect within an ample range (1235C). In addition, the authors pointed that tomato seeds do not respond to vernalization and therefore the temperature effect on NLPI is on the plant not on seeds. In brief, light and temperature effects on NLP I interact on tomato in such a way that higher light intensity decreases NLPI and higher temperature and low light increase it. This interaction is explained by the nutrient diversion theory proposed by Sachs and Hackett (1969) which has not been refuted until now. According to this theory a certain amount of photoassimilates need to be accumulated in the apex before tomato flower initiation takes place. Thus at low light conditions during the sensitive phase not enough assimilates are accumulated to switch from vegetative to reproductive growth. This is enhanced in high temperature because vegetative organs grow faster consuming the few assimilates available, and a high respiration rate at high temperature also consumes more assimilates. In high light inten sity conditions, the effect of temperature is less marked because enough photoassimilates are produced. An elegant integration of that theory was recently published by Uzun (2006) who

PAGE 38

38 performed a multiple regression analysis of the effects of temperature a nd light integral on the NLPI. His results confirmed the Sachs and Hacketts theory, the NLPI declined linearly with temperature and particularly at the lowest daily mean light integral. Because this role in determining the switch from vegetative to repro ductive growth and also the earliness of the crop, the concept of NLPI could be very useful if it were incorporated into tomato m odels which until now simulate this transition only through the thermal time and allocation rules. Since NLPI is a function of light and temperature, both variables could be used to simulate NLPI. A caution note should be made; however, according to the Uzun studies this response to light temperature interaction is not linear and therefore normalizing the equation for descr ibin g it may be difficult. Fruit set and pollination: The temperature range for fruit setting in tomato is narrow and the night temperature is critical. The optimal range reported for fruit setting in tomato is 15 to 20C (Went, 1945), and 18 to 20C (De Konin g, 1994). Fruit set is also low when the average maximal day temperature is above 32C and the average minimal night temperature is above 21C (Moore et al., 1952; Benedictos and Yavari, 2000). These last authors did not find cultivar differences in tomato with respect to flower abortion at high temperatures (37/21C day/night). Tomato fertility is affected by high temperature and this has a direct effect on yield through the number of fruits that are set on the plant. The simulation of this effect by tomat o models has been poor in general. The CROPGRO tomato species files includes a function that modifies partitioning limit (XFRT) to reproductive if temperature is too hot (reduced above 28C and falling to zero at 33C) and this function is a substitute f or elevated temperature effects on fertilit y (Boote and Scholberg, 2006). For fruit pollination, the value for ceiling or failure temperature is critical because the fertility of tomato flowers is compromised at high temperature. The same values as propose d by Adams et al. (2001b ) for progression to anthesis appear to be valid for fruit addition and

PAGE 39

39 pollination e xcept for ceiling temperature. According to Moore et al. (1952) and Benedictos and Yavari (2000) above 32 C is the failure day temperature for fruit setting and pollination while above 21 C is the night failure temperature. According to Heuvelink (2005) the range of flower fertilization in tomatoes is reduced at temperatures outside of the range 5 C to 37 C. The growth rate of pollen is increased between 10 C to 35 C but is reduced outside this range. According Atherton and Harris (1986) the failure temperature for fruit pollination is 40 C and the most critical stage appears to be meiosis, which occurs about 9 days before anthesis. The optima l temperature for pollination according to those a uthors is between 17 to 24 C. For base temperature, a value of 5 C was proposed and as a cons equence, from Atherton and Harris the cardinal temperature for pollination seems to be: 5C, 17 C, 24 C and 40 C for Topt1, Topt2, and Tmax respectively. Since the Adamss et al. experiments were carried out in controlled temperature environments it seems reasonable to adopt their values for modeling purposes and because of the strong effect of high temperatur es on fruit setting to use a rather conservative value for ceiling temperature. As a consequence for fruit setting and pollination reliable cardinal temperature values appears to be 7.2 C, 22 C, 26 C and 32 C for base, Topt1, Topt2, and Tmax respecti vely. Late reproductive phase: Assuming that the fruit growth period can be described by relating linearly its reciprocal to temperature, a critical temperature summation needs to be reached in order for fruits achieve maturity (Heuvelink, 2005). For this late reproductive phase (fruit development rate and progression to maturity), Adams et al. (2001b ) found values of 5.7 C for base and 22 C for Topt1. De Koning (1994) proposed 4 C for base temperature, around 21 C for optimum and presented no data for Topt2 and ceiling. Aikman (1996) proposed that the time from anthesis to maturity for tomato is 806 degree days using a base temperature equal to

PAGE 40

40 4.75 C. Using the 4C of De Koning (1994) this time is 940 degree days while Scholberg et al. (1997), calcul ated 722 degree days using a b ase temperature equal to 10 C. Fruit growth rate: Temperature is the climatic factor that most affects the fruit rate growth in tomato (Walker and Ho, 1977; Pearce et al., 1993). Hurd and Graves (1984) suggested a Q10 value e qual to 1.7 for tomato fruit growth and equal to 2 for fruit m aturation (Hurd and Graves, 1984). For fruit growth rate (dry matter and water accumulation) Rylsky (1979) found an optimum equal to 26 C while Adams et al. (2001a ) found a regimen of 25/25 C (day/night) to be optimum for fruit growth. Ho et al. (1983) found that the maximal rate of dry matter accumulation in tomato fruit at 19.3C occurs around day 23 after anthesis (335 degree days over a base temperature equal to 4.75C). Unfortunately, no data were reported for ceiling temperatures. De Koning (1994) proposed 10 C for base tempe rature but no data for ceiling. The ceiling temperature on the fruit growth rate seems more complicated to obtain since the water relationships in the system may influence the effect of high temperature on the fruit growth rate. Therefore, quality data for this parameter requires that both temperature and water should be controlled precisely. According to De Koning (2000), tomato fruit growth rate is relatively indep endent of temperature; however temperature appears to be the principal factor determining the duration of the tomato fruit growth period. His results showed this period to be 73 days at 17C and 42 days when te mperature increased to 26C. Si milar results w ere found by Rylsky (1979). In addition, Verkerk (1955) found that the time inte rval from anthesis to harvest was equal to 90 days at 13C, 53 days at 19C, and 40 days at 26 C. In his experiment, De Koning divided the growth period in five states, and found different responses to temperature depending the fruit age. High temperature shortened the growth period in two phases, first at the young developmental state; the middle phase was insensitive to temperature, and again close to maturity

PAGE 41

41 when temperatur e had a great impact on days to harvest. Adams et al. (2001) and Adams and Valdes (2002) found that when tomato plants were grown at 14, 18, 22 and 26C, fruits ripened after 95, 65, 46 and 42 days, respectively, and the maximum fruit growth rate was achie ved at 25/25C. According to their results, fruits were more sensitive to elevated temperature in their later stages of maturation. In addition, their results showed a decline in days to harvest of 8.7 to 11.2 days when temperatures were increased by 7C f or a 3 week period above a value of 18C. In summary, taking both processes together (fruit development and maturation) cardinal temperatures values according to these authors seems to be 5.7C, 26C, 28C and 48C for base, Topt1, Topt2, and Tmax re spectively. Maximum temperature is not reported and therefore until other data is available, the same value s as used for vegetative growth will be assumed. Yield: The fresh weight of individual tomato fruits is strongly related to the number of locules pre sent. Sawheney (1983), and Sawhney and Polowick (1984), studied the influence of the temperature on the size and locule number of tomato fruits. The study was conducted in temperature-controlled chambers with three thermal regimens: low (LRT) 18/15 C (day /night), intermediate (IRT) 23/18 C (day /night) and high (HRT) 28/23 C (day /night). The fresh weight, fruit diameter and locule number were significantly different under the three regimens. On ave rage the fresh weight was 133 g, 96 g, and 73 g for low intermediate and high temperature regimens respectively. The fruit diameter was equal to 6.99 cm, 6.05 cm and 5.42 cm for the LRT, IRT and HRT treatments. The locule number was equal to 7.4, 8.5 and 6.99 (average). Houghtaling (1935) reported a correlat ion coefficient equal to 0.87 between the ovary size and the fruit size in tomato. Similarly, Young and McArthur (1947) reported that the gene lc which controls the number of locules in tomato fruits is correlated w ith the fruit size. Gibberellin applicati on on flowers produce d a similar effect as low temperature and produced a larger ovary

PAGE 42

42 and a greater locule number in tomato fruits which resulted in bigger fruits. Overall, most of reports support that cool temperature increases fruit size (Went 1957; Ch arles and Harris 1972; Sawheney, 1983). Rylsky ( 1979) reported that at even low er temperature the size of the fruits is reduced because of parthenocarpy. Sawhney and Polowick (1984) observed that less pollen was produced at very low and high temperature a nd more pollen at intermedi ate values. However, because all flowers were pollinated by hand the effects of a poo r pollination were not studied. Temperature often has strong influences on the demand for assimilates (sink strength) by the growing organs and therefore on the biomass allocation. In cucumber fruits, for example, the sink strength of individual fruits increases at increasing temperature, resulting in an increase of dry matter allocation to the fruits (Marcelis 1992). In tomato, De Koning (1989) found similar results. However, he reported that further increase in temperature also reduced the number of fruits in such a way that the biomass allocation to all the fruits was not altered. Heuvelink (1995) found in a spring experiment that 56 to 60 % o f cumulative dry matter was distributed to fruits, 28 33% to leaves and 12 13 % to stems. On the other hand, late in autumn, when temperature decreased, the same author found only 3538 % of dry matter was distributed to fruits, 44 45 % to leaves and 18 2 1 % to stem. However, due to lower radiation late in autumn that produced a poor fruit set it is difficult separate the effects of light and temperature on allocation. Bertin et a l. (2003) established that when the temperature is increased the tomato frui t growth rate increases; however she pointed out that the size and final mass of the fruit could be unchanged because the increased growth rate is compensated by a shorter growth period. The effect of temperature on tomato yield according to Adams and Vald es (2002) is correlated to both the earliness of the production and the yield fluctuations on weekly bases. Thus high temperatures during ripening phase results in a flush of ripe fruits. However, rapid fruit ripening

PAGE 43

43 results in a decrease of the number of younger fruits approaching maturity on the plant, which then reduces yield in subsequent weeks. As a consequence, those authors proposed using temperature as driving variable to simulate the harvest dynamics of tomato Water Fruit growth: Under non stress conditions, tomato fruit growth follows a sigmoid pattern that results from cell division during about the first two weeks after anthesis, and thereafter increase in volume (cell expansion) until the fruit reaches its final size (Coombe, 1976; Gillaspy et al ., 1993, Monselise, 1978). The last process depends on the inflow of water which is linked to the carbon movement into the fruit. As a result, the growth of tomato fruit depends largely on the rate of water accumulation (Li et al ., 1989). According to H o et al. (1987) a large amount of water with assimilates is transported in the phloem sap to the fruit. Consequently, for modeling tomato fruit growth, the water relationships need to be understood. Tomato models in general simulate the fruit growth as a function of dry matter accumulation rate and the growth in fresh mass is generally ignored or deduced empirically from the dry matter accumulation. Lee et al (1989) and Johnson et al. (1992) showed that the water flow into tomato fruits depends on the wat er potential difference between the plant and the fruit and consequently it controls the expansive growth of tomato fruit via phloem. Because transpiration from tomato fruits is negligible with respect to the whole plant (Johnson et al., 1992), the water m ovement in the plant can be analyzed considering the water flow to leaves as only transpiration and water flow to fruits as only growth (Ya Ling et al ., 2004). Ninety percent of water and dry matter entering the tomato fruit is via phloem, and water impo rt via xylem ceases after the maximal growth of the fruit has been reached (Ethret and Ho, 1986; Ho et al., 1987). Johnson et al (1992) found a strong correlation between the fruit -stem water potential gradient and changes in fruit diameter. They showed t hat for tomato plants grown in high radiation, a threshold value between 1.1 to

PAGE 44

44 1.3 Mpa of stem water potential stopped the fruit growth presumably because the pressure gradient was too low to drive water flow into the fruit. These results are in agreem ent with data presented by Guichard et al (1999) who found that high vapor pressure deficit (VPD) conditions reduced the fresh yield of tomato fruits mainly by its effect on water rather than dry matter influx, which was not affected. The fruit water pote ntial ( 0.4 Mpa) was not sensitive to VPD whereas the stem -fruit water potential gradient was reduced under high VPD conditions. On the contrary, low VPD increased the difference of water potential between fruit and stem maintaining the stem water potentia l around 0.28 Mpa and allowing fruit growth. Recent investigations published by Ya Ling et al. (2004) confirmed that the apoplastic stem -fruit water potential gradient seems to be a good indicator of both fruit growth rate and fruit weight. Considering th at under adequate root water supply the fruit osmotic potential varies little, it is valid to propose that stem water potential will establish the gradient between fruit and stem. Therefore, only stem water potential could be considered in fruit growth pre diction. Kawabata et al. (2005) studied the role of transpiration from tomato fruits in phloem transport and fruit growth. Their results showed that transpiration does not serve as a limiting step of carbohydrate transport to tomato fruits. Furthermore, al though salinity effects will not be considered here, this issue has been well documented (Ethret and Ho, 1986; and Cuartero and Fernandez Munoz 1999; Mitchell et al ., 1991). All these authors reported that salt stress influence s both dry matter accumulatio n and water import into the fruits by effects on osmotic potential, maintaining dry matter accumulation but decreasing fruit fresh weight because less water is transported in xylem to the fruit and cons equently the size is reduced. Photosynthesis: The decr ease in net photosynthesis in plants induced by water stress has been often reported. Mild stress may decrease photosynthesis by decreasing CO2 diffusion due to

PAGE 45

45 stomata closure and under severe stress biochemical reactions may be affected as well (Gimenez et al., 1992; Lawlor, 1995). Castrillo et al (2001) studied the effects of mild and severe water stress on tomato Rubisco activity, stomata conductance and leaf protein content. They found that a reduction in soil water poten w) produced a decrease in w. During water stress w decreased gradually reaching a minimum 2.50 MPa on the 14th day and it had the lowest values un til the end of the stress. The stomat a conductance showed a gradual decrease durin g the stress starting to decrease at 0.7 MPa. The Rubisco activity in water stressed plants was stable at beginning of the stress, then began to decrease at a leaf w of -1.70 MPa and reached the low er value at w of 2 .50 MPa. The tomato leaf protein content showed w of 1.50 MPa it decreased and was low est at w of 2.50 MPa. Rahman et al (1999) found that the photosynthesis rate, transpiration rate, leaf water potent ial and water use efficiency were reduced in tomato plants under water stress as compared with well irrigated plants while the stomata resistance and the leaf temperature was increased and those responses were cultivar depe ndent. Especially important was the difference among drought sensitive and tolerant genotypes related to the capacity of recovering under re watering. Thus, the tolerant cultivars recovered shortly after re -watering while the sensitive ones took a significantly longer time to reestablish photosynthesis and trans pira tion rates. Yield: The effect of water stress o n tomato fresh yield is dynamic and complex to analyze. F actors such as timing of osmotic adjustment, interactions with nutrient status, especially nitrogen and potassium, time of occurrence, magnitude and dur ation of the stress, effects of the stress on reproductive sink activity and growth component being considered affect the response T he reports of osmotic adjustment in tomato are not consistent. In some studies partial osmotic adjustment was reported (Ala rcon et al ., 1994; Torrecillas et al ., 1995). I n other studies however,

PAGE 46

46 no osmotic adjustment was observed (Perez Alfonsea et al ., 1993). One weakness of those reports as pointed by Alian et al (2000) is that the period of stress imposed was likely too short to see clear osmotic adjustment response The studies on water stress are focused on fruit size, which is largely a measure of the fresh weig ht (Berman and Dejong, 1996). Research on tomato suggested that water stress limits fleshy fruit water accumula tion but does not affect carbon partitioning to the fruits (Ehret and Ho, 1986). Scholberg (1996) evaluated the effects of water supply on tomato fruit yield grown in field conditions. His experimental data showed that fresh fruit weight was more strongly affected by water stress than fruit dry weight and the decrease of yield under stress was attributed to smaller and fewer fruits as compared with the control. These results agree with Ehret and Ho (1986), Mitchell et al (1991) and Van Ieperen et al. (2005) who showed that water stress during growth of tomato fruit reduced fruit size by 30%. In addition, Garcia et al. (2004) found that one important response of tomato plants to early season water stress was a significant reduction in the number of floral bu ds and there was a delay in the floral buds appearance. Wolf and Rudich (1988) studied the effects of water stress on the tomato dry weight accumulation and fruit growth. The rate of dry matter accumulation was found to be higher in early than in later set ting fruits and was unaffected by the water regime. Water stress however, shortened the duration of fruit growth and accelerated the ripening. Under water stress more than 40 % of the final dry weight yield was contributed by the fruits set during the firs t week of flowering. On the other hand, many investigations have reported that irrigating at a rate of 120 or 140% of evapotranspirative demand ha s increased significantly the fresh weight yield in toma to crops (Ortega Farias, 2003). Bussiers (1993, 1994) studied the potential dry matter and water import rates in the tomato fruit in relation to fruit size. He found that the fruit radius is an important fruit size parameter

PAGE 47

47 controlled by changes in water and dry matter imported into the fruit. This is becaus e both flows are the product of two factors: 1) fruit surface area, which is radii dependent and 2) dry matter and water import rate per unit of fruit surface area. His results suggested that is more appropriate to consider the strength of the sink as bein g proportional to the product of fruit surface area multiplied by sink activity. Based on that he developed models of water and dry matter import rates in tomato fruits and predicted partitioning to fruit as a function of fruit surface area and proposed t hat this relation could be used in fruit size simulations. Those models, although well sustained from a biophysical point of view, are difficult to incorporate into mechanistic models developed for tomato which are photosynthesis based and the water inflow into individual fruits is ignored Nitrogen The ionic forms of nitrogen (N) used by tomato plants are ammonium (NH4 +) and nitrate (NO3 -). Tomato plants grow better with NO3 sources. When NH4 + is the major N source, toxicity can occur resulting in yield r eduction (Barker and Mills, 1980; Jones, 1999). This is especially true in reproductive stages when, under NH4 + excess, the incidence of blossom end rot (BER) increases dra matically (Pill and Lambeth, 197 7; Pill et al., 1978; Jones, 1999). In early vegetat ive growth, however, NH4 + may be beneficial because of faster utilization and therefore a faster crop growth early in the cycle. In addition, Hartman et al. (1986) reported that a high NH4 + nutrition increases the phosphorus (P) concentration of the plant but reduces p otassium (K), calcium (Ca) and m agnesium (Mg). In the fruit, the P increased and K decreased while Ca and Mg were not affected (Jones, 1999). According to Adams (1986) N is essential for plant growth, influencing fruit number and yield. Howeve r, excessive N will increase the percentage of unevenly ripened (blotchy) fruits (Grie r son and Kader, 1986; Jones, 1999). During the crop cycle tomato can uptake 150 to 300 kg ha1 of N but the response to N is more pronounced in the

PAGE 48

48 range of 0 to 100 kg ha1 (Larouche et al., 1989). During rapid growth N uptake rates may exceed 4.3 kg ha1d1 (Dumas, 1990; Rinaldi et al., 2007). According to Warn er et al. (2004) the N requirements of tomato are higher during the vegetative than reproductive growth and their results agree with Scholberg (1996) and Scholberg et al. (2000) who studied the effect of N stress on growth and yield of tomato growing in field conditions. The results showed that severe N stress reduced significantly LAI, biomass and fruit yield by 60 to 70 % as compared with well fertilized plants. In addition the concentration of N in vegetative organs such as leaves and stems was reduced under N stress, while the fruits on other hand, maintained a relatively stable N concentration. Under N stre ss the RUE (dry matter produced per unit light intercepted) was reduced by about 30 % and the authors related this reduction to a lower leaf N concentration which in turn affected the photosynthesis rate. The leaf photosynthesis rate also was reduced by 30 % compared to control plants although the leaf N concentration was dramatically reduced (40 to 15 mg g1). One of the reasons was that SLW increased under N stress, thus causing specific leaf nitrogen ( SLN ) to be considerably maintained despite large decr ease in N concentration. Lawlor (2002) published a complete review about the relationship between photosynthesis and SLN. It is largely accepted that the leaf photosynthesis response to irradiance depends on the SLN (Evans 1989). The maximum rate of photosynthesis (Amax) increases asymptotically to the leaf N content and this relationship is true either considering SLN per leaf area unit or per leaf weight unit (Grindlay, 1997). The maintenance of a relatively high SLN in plants under N stress concurs with the Scholberg et al. (2000) observations who found only 30 % of reduction in leaf photosynthesis rate of plants exposed to N stress as compared with the control treatment although the N concentr ation was dramatically reduced. The impact of N is frequentl y evaluated in leaf photosynthesis terms and the effect of N on building the source of assimilates but

PAGE 49

49 information about the fruit sink response to N limitations or the fruit growth rate is scarce. Huett (1986) and Huett and Dettman (1988, 1991) reported that increased yield of semi -determinate tomatoes in response to increasing N was associated with higher ratios of fruit dry matt er to total plant dry matter. That means that in tomato crop where the fruits are the dominant sinks apparently the N level in fluences the dry matt er partitioning to the fruits. Hebbar et al. (2004) reported an average of 33 % of higher fruit yield in fertigated tomatoes especially related to N fertilization and drip irrigation method, they attributed this yield increment to a hi gher LAI, number of fruits, dry matter production and fertilizer use efficiency Partitioning of Assimilates Within a Tomato Plant Partitioning of assimilates among plant organs is often described by the concepts of source and sink. Source is a plant part that exports more carbon than it imports, while sink is a plant part that imports more than it exports (Ho et al., 1989). The ability of a sink to import assimilates is referred as sink strength. The sink strength can be quantified by the potential growth rate of a sink, i.e. the growth rate under conditions of nonlimiting assimilate supply (Heuvelink and Marcelis, 1989). According to Farrar (1988), Ho (1988) and Marcelis (1992), the distribution of assimilates among sinks is primarily regulated by the sin ks themselves. Tomato fruit s are very strong sink s for carbohydrates (Ho, 1984; Wang et al., 1993). In tomato crop the fruits are the most important sinks (De Koning, 1993), therefore it is expected that the dry matter distribution between vegetative and g enerative sinks is mainly regulated by the fruits. Under normal conditions, photosynthetic supply does not limit fruit growth in tomato because the source exceeds the sink demand (Khan and Sagar, 1969; Tanaka and Fuyita, 1974; Heuvelink and Buiskool, 1995) De Koning (1994) reported a source sink ratio for tomato of 0.5 for indeterminate greenhouse grown cultivars which means a weak source limitation compared to other fruit crops. Hurd et al. (1979) and De Koning and De Ruiter (1991) observed that when the

PAGE 50

50 number of fruits per tomato plant was increased, the fruit growth increased at the expense of vegetative growth. However, a balance in the source sink ratio needs to be reached for indeterminate greenhouse cultivars because if the source to sink ratio is too low (less leaf area and light interception) flower/fruit abortion may decrease the fruit yield (Bertin, 1995). Heuvelink (1997) published the results of experiments that were carried out in order to establish the relationship between the fruit load and the dry matter partitioning in tomato plants. He found that the dry matter distributed to the fruits can be described by a saturation type function of the number of fruits retained by the truss. According to this relationship the generative sink strength was proportional to the number of fruits (in a range of two to seven). The fraction partitioned to fruit was almost halved when fruit number was decreased from 7 to 2. The average sink strength of the vegetative sink (three leaves and the internodes betwee n trusses) was 2.96 times the average sink strength of one fruit. Consistent with these results the average dry matter partitioning to the fruit over time increased with the average fruit number on the plant during the time interval. The weight of individual fruits was increased when the number of fruit decreased but this diminution was less than proportional. At higher than seven fruits per truss the relationship was not more proportional. Indeed, De Koning (1994) observed a low potential fruit weight whe n the fruit number per truss was over seven. Heuvelink and Buiskool (1995) demonstrated that reducing the number of fruits per plant strongly di minished the biomass assignment to the fruits, whereas the total dry matter production was hardly affected. In a ddition, temperatures in a range of 9 to 23 C, and irradiance in a range of 8 to 15 MJ m2, hardly influenced partitioning. The knowledge of the relations mentioned above is useful in simulation models. Heuvelink and Marcelis (1989), for instance, used a constant value of 3 g per day per plant for total vegetative sink strength. The simulated average sink strength of a fruit was 0.129 g

PAGE 51

51 per day. Those values are consistent with the constant ratio value between vegetative unit and one fruit which, as was m entioned, is equal to 2.96. Thus in a tomato plant with 8 growing vegetative units (De Koning 1994), results in 2.96 *0.129*8 sink strength as published in the experiments of Heuvelink and Marcelis. Regarding partitioni ng of dry matter to fruits average values between 54 60 % and maximum of 64 % and 67 % were reported by Heuvelink (1997) and Aikman (1989) respectively. Heuvelink (1995) found that in spring 56 to 60 % of cumulative dry matter was distributed to fruits, 28 33% to the leaves and 1213 % to the stems. However, in autumn only 3538 % of dry matter was distributed to the fruits, 4445 % to the leaves and 1821 % to the stem. De Koning (1993) found a fruit partitioning equal to 72 % in commercial crops and ove r the whole season while Cockshull et al. (1992) estimated that the 69 % of the produced biomass was allocated to the fruits. The results of Aikman (1996) are close to those values since he reported that 67 % of the dry matter is allocated to fruits in tom ato. He calculated that 1 mol of carbon assimilated as CO2 would give about of 0.4 mol harvested in carbohydrate, about 12 g harvested biomass or 200 g harvested fresh weight. Lower values were found by Ehret and Ho (1986) and Scholberg et al (2000) who reported that 58 % of total above dry matter produced by tomato crop was allocated to fruits. The fruit dry matter concentration may be influenced by the fruit number, the temperature and the duration of the growth season. The average fruit dry matter con ce ntration was 5.6 % according to Ehret and Ho (1986). In agree ment with this value s of 5.46 % were reported by Winsor and Adams (1987), 5.5 6 % by De Koning (1993) 6 % by Aikman ( 1996) and 5 7.5 % by Davies and Hobson (1981) It could be argued that the sink strength is strongly dependent on the developmental stage of the fruit as was demonstrated by De Koning in 1994 but the same author argued that when the

PAGE 52

52 development is normal there is a correlation between the fruit number and the fruit age as we ll. Therefore the relationship found between the fruit load and the dry matter partitioning to the fruits is still valid. However, it does not imply that always the number of fruits on a plant will determine the dry matter allocation to the fruits O ther aspects such as severe stress and competition between fruits within a truss are frequently mentioned as factors that may alter the biomass allocation to fruit s (Aikman, 1996; Bertin, 2005). According to Huevelink and Buiskool (1995) the influence of the sink demand on dry matter production per unit of intercepted radiation and probably on leaf photosynthetic rate in commercial tomato production can be ignored. Therefore and different from other crops in which feedback inhibition is reported, reducing frui t number (sink strength) in tomato plants apparently does not reduce the photosynthetic rate. They found that while the production of dry matter was not influenced by sink -source ratio (except in the extreme case of one fruit per truss), the distribution o f dry matter between fruits and vegetative parts was greatly affected. Thus, of the same amount of dry matter produced, the partitioning was highly variable and a fraction of 0.3 to 0.6 was allocated to the fruits depending on the fruit number. Lack of inf luence of sink -source ratio on dry matter production in tomato plants was also supported by Ho (1992) who concluded that only in extreme cases (pruning all the fruits) a low sink demand will negatively influence the photosynthetic rate. According to Huevel ink and Buiskool (1995) the source sink ratio did not affect the plant development and the number of visible leaves at the end of the experiment was the same for pruned and unpruned plants. However, the specific leaf area and the internode length decreased (heavier stems and thicker leaves) and the plant leaf area increased when the sink source ratio was reduced. The truss appearance rate and fruit growth period are hardly influenced by sink -source ratio (De Koning, 1994). While the plant leaf area is not a ffected by

PAGE 53

53 competition for assimilates except under extreme conditions, the leaf mass per area (SLW) or the inverse the specific leaf area (SLA) show large fluctuations when the source sink activities are altered. Bertin and Gary (1998) showed that under a rtificial manipulation of the sink -source ratio through shading, CO2 enrichment or fruit removal, the dry weight of the leaves and as a consequence the SLW or SLA underwent large and rapid fluctuations. For example, a reduction in 60 % of photosynthetic a ctive radiation (PAR) led to a 24 % decrease in SLW after 10 days. Carbon dioxide enrichment and fruit removal produced an increase in SLW of 45 and 15 % respectively on plants with two fruiting trusses, but hardly affected SLW of producing plants. They di d not observed cultivar differences in these responses Based on their results those authors proposed that the structural SLW varies between a maximum and a minimum according to the ratio of assimilate supply and the demand during leaf development. They al so proposed that when a maximum SLW occurs, a storage pool of assimilates might have accumulated in leaves during periods of high supply and low demand. The concepts of SLW or SLA are very important in crop models because they are frequently used to predic t leaf area expansion and consequently light interception based on the dry weight of the leaves. Reduction in sink number also induces variation of SLW in tomato leaves ( Bertin and Gary 1992). The variations can be related to changes in the level of starch and hexoses in tomato leaves ( Shaw et al., 1986). According to Bertin and Gary (1998) the fluctuations in sugars and starch accounted for 29 % of the daily variation in SLW when the sink source ratio is manipulated; therefore they hypothesized that fluct uations in other components in leaves, besides sugars and starch, may be involved in the changes of SLW. In some models, the SLA is a constant or it varies in responses to climate or the season. In other models, the leaf area growth is mainly a function of the plant leaf age and temperature ignoring mass Still there are models in which leaf expansion is

PAGE 54

54 function mainly of temperature but water or nitrogen status is allowed to modify the potential growth. The source sinks interactions, however, as affecting the leaf weight normally has not been considered in crop models to describe fluctuations of SLA. This is a weakness and may affect the accuracy of the leaf growth predictions. Gary et al. (1995) proposed to simulate SLA as a function of leaf age an d tempe rature but allowing structural SLA to vary between a minimum (full satisfaction of growth demand) and a maximum (minimum leaf thickness) (Gary et al. 1995; Heuvelink, 2005). This approach might be a path to improve the simulation of SLA fluctuations and th erefore of the leaf expansion as well. Temperature often has strong influences on the demand for assimilates (sink strength) by the growing organs and therefore on the biomass allocation. In cucumber fruits for example the sink strength of individual fruits increased at increasing temperature, resulting in an increased dry matter allocation t o the fruits (Marcelis, 1992). Similar results were found by De Koning (1989) in tomato. However, they reported that the increase in temperature also reduced the numb er of fruits in such a way that the biomass allocation considering all the fruits together could be not altered. The re are several reports of the role of the irradiance on dry matter partitioning Papadopoulos and Ormord (1990) for instance, found that th e dry matter distribution to the tomato fruits increased with increased plant spacing and they argued that this effect can be attributed to an increase in the light interception per plant. Cockshull et al. (1992) found that a reduction in solar radiation b y 23 % had no effect on the dry matter distribution to tomato fruits although the number of fruits was reduced. Yoshioka and Takahasi (1981) on other hand, fixed the fruit number and observed decreased assimilate allocation to the fruits when the ir radiance increased. The distribution of assimilates in tomato plants is not affected by osmotic stress within

PAGE 55

55 certain limits. The tomato specie is rel atively tolerant to salinity. Ehret and Ho (1986), reported that salinity values up to 6 mS cm1 did not affect t he dry matter partitioning in tomato plants. Grange and Andrews (1993) established that there is close relationships between the rates of assimilate importation during the rapid growth phase and the final weight per fruit. The rate of assimilate importation is affected by temperature and under normal circumstances the increase of temperature increases the rate and in consequence the fruit weight. However, this is true if the assimilation and water supplies are not limited. Under water stress an increase in temperature likely will increases the canopy and fruit transpiration which will affect the fruit weight mainly because a reduction in its volume The dry weight however, may be less affected because even when water stress reduces the phloem sap volume, t he concentration of dry matter in fruit increases therefore the rate of importation of assimilates remains relatively stable (Ho et al ., 1987). Osmotic stress may have similar effect on the dry weight assimilation rates. The potential strength of differ ent sinks is related to the developmental stage of the plant. In addition, the potential strength of the fruits is related to the phyllotaxy of the plant and to hormonal regulation as well. In concern for hormonal regulation, Ho cited the reports of Kinet (1987) and Kinet et al (1978, 1986) who established that the apparent role of the auxins on the fruit strength is related to the cell division activity in the ovary which is hormonally regulated. This theory is also the basis for the close and positive re lationships frequently found between the weight of tomato fruits and the seed number since it has been argued that the capacity of seeds to produce auxins appears to stimulate the cell division in the ovary. The duration of fruit growth from anthesis to ma turity is not highly variable among cultivars being in the range of 40 to 65 days. Because there are huge variations in fruit size among cultivars, one can infer that the source of this variation is different rates of dry matter

PAGE 56

56 accumulation among them. In fact, mutants of tomato plants with altered number of cells in the ovary of the fruit during pre anthesis have demonstrated that the cell number is a good measure of the sink size and determine s the strength of the fruit to attract assimilates thus affect ing the final size of the fruits. In addition, even though the auxin production has been reported as the reason for good correlation between the number of seeds and the fruit weight, other authors suggested that this correlation is in fact related to the n umber of seeds which correlates with the number of ovules per se and this in turn determine the cell number in the ovary. However, parthenocarpic fruits that still reach a commercial size make difficult the last interpretatio n. The photosynthesis of tomato fruits is negligible after reaching a diameter of 15 mm (Bertin et al, 2001). Therefore, the carbon balance of tomato fruits depends on the importation of assimilates from leaves. Sucrose is the main sugar t ransported (Walker and Ho, 1977); however, sucro se does not accumulate in fruits but it is converted in hexoses, which account for 75 % of the soluble solids in a ripe tomato fruit. According to Ho et al. (1983), during the first week after anthesis sugars and starch account for 10 % of the dry weight o f the fruit. In a fruit of 3 weeks, 20 % of the dry matter is starch. After that the sugar proportion increases until a steady value which range s between 50 and 75 %. Even when a tomato fruit does not accumulate significant amount of starch except in the ea rly development, Dinar and Stevens (1981) showed a positive relationship between the amount of soluble solids in a ripe fruit and the rate of starch accumulation during early gre en stages. This relationship has been explained as the starch assimilation in early cell enlargement provides an extra capacity to store imported sugars Genetic Traits Related to Tomato Yield Fruit and yield characteristics of tomato plants are controlled by gene tic and environmental factors (Sch olberg, 1996). There are more vari eties of tomato sold worldwide that any other vegetable (Foolad, 2007). However, selection for yield per se is not successful in general because

PAGE 57

57 yield is a complex trait influenced by genetic and non genetic factors, and therefore, the heritability for yie ld is very low in most crops including tomato (Foolad, 2007). Consequently, the breeders normally bred for components of yield that contribute to it. In tomato for example targets for breeding have been: disease resistance (Bournival et al ., 1990; Scott, 2003), tolerance to abiotic stresses (Vallejos, 1991; Foolad, 2005), earliness and fruit solid content (Fridman et al., 2000; Georgelis et al, 2004) among others. According to Foolad (2007) the history for improved tomato fruit yield through different att ributes has been very successful. For example, between 1920s and 1990s, fruit yield of processing tomato cultivars in the U.S increased from 10.1 to 72.4 tons ha1 (Warren, 1998). On average, half of this increment is attributed to breeding and the other half to improvement in management practices (Duvick, 1986). In the fresh market cultivars, the University of Florida s breeding program has achieved increases in yield by breeding for heat tolerance for production under hot and humid conditions (Scott, 1997 ). Genetic studies have shown that most tomato traits of economic importance that distinguish cultivated tomatoes from their related wild species are due to quantitative trait loci (QTLs). Molecular mapping for instance has revealed the presence of several dozen QTLs associated with variations in tomato fruit size. For many tomato varieties, the differences in fruit size are established by the time of flowering and the size of the ovary becomes a good predictor of the fina l size of the fruit (Yeager, 1927; Grandillo et al., 1999). Szymkowiak and Sussex (1992) showed that in tomato the meristem size and the number of carpels is determined by the number of cells in an internal layer of the shoot meristem and this layer determine s the eventual sink strength of the developing fruits. A QTL of tomato called fw2.2 has been cloned and characterized and seems to make the largest contribution to size variation between the cultivated tomato genotypes and their small wild species relatives. The QTL fw2.2 appears to det ermine the

PAGE 58

58 difference in fruit weight and diameter (Liu et al ., 2003). According to Frary et al. (2000) allelic differences at fw2.2 increase fruit weight by 30 percent. Nesbitt et al. (2001) studied whether the change in tomato fruit size caused by fw2.2 is associated to other phenotypic changes that indirectly affect fruit size. They concluded that the primary effect of fw2.2 is on controlling frui t size although there are other secondary aspects associated, such as fruit number and photosynthate distribu tion. The number of QTLs reported as affecting tomato fruit fresh weight varies from 3 (Bernachi et al. 1998) to more than 18 (Esehd and Zamir, 1995). Grandillo et al. (1999) estimated that a common set of 28 QTLs for fruit fresh weight can be identified for tomato that shows segregation in at least two independent studies. Zeng et al. (1990) suggested a similar number of QTLs affecting fruit fresh weight in tomato. In similar way diverse amount of QTLs affecting the shape of tomato fruits have been ident ified and the number varies from 2 to more than 16 QTLs. Moreover, a common set of 18 QTLs for fruit shape can be identified for tomato that shows segregation in at least two independent studies. The two major fruit shape QTLs, fs2.1 and fs8.1 have been c haracterized and round shape is partially dominant over the more elongate shape (Grandillo et al., 1996). On the other hand, several qualitative genes have been mapped in different chromosomes of tomato that exert effect on fruit size or the number of locu les. For example the gene ovate reduces the fruit weight by half whereas the gene faciated increases the number of locules and seeds increasing the fruit weight by 60 % (MacArthur and Butler, 1938). The potential use of real genetic parameters in models instead of the empirical ones has been frequently proposed as a possible tool to improve the physiological bases of m echanistic models. This task is technically very difficult in cases like tomato in which the main genes that define yield and quality trait s are quantitative and therefore the inheritance patterns are normally

PAGE 59

59 polygenic. Consequently, it is almost impossible to assume independency of characters for parameterization purposes. Still another approach for this goal could be to study the inheritan ce of other traits not directly involve d in yield, but related to it. For example, the traits that define the earliness and the life cycle of the specie. Molecula r techniques that allow cloning and characterizing QTLs are potentially valuable tools that c a n help in the future to obtain better results in this area but the lin k with modeling is still weak. In 2006, Thornley quoted Does the massive amount of data now being produced by genomics, proteomics and metabolomics change fundamentally the modeling pr oject ? His short answer was no and nothing has changed because qualitatively the biology is the same. He cited Popper (1959) to say models still operate within the loop of hypothesis, analysis, synthesis, prediction, test against real data and back to hy pothesis What could change with what Thornley called the omics data avalanche is that there is now an increased opportunity for relating predictions to properties defined at molecular level with all the possibilities that such knowledge may provide. Ac cording to Thornley try ing to construct models going from molecular level to the whole system in one bound is not viable but what is possible and really matters is to establish connections between the top and the bottom of the system, step by step, to all the intermediate levels of organ tissues, cell and sub cellular and when those connections are made, then the understanding and therefore the ability to simulate and intervene the systems will be maximized Tomato Models Horticultural models were develope d significantly in the eighties as consequence of the need for quantitative information tools to improve decision making to control greenhouse environments (Gary et al., 1998). Garys paper was a complete review describing the state of art of horticultural models. Other reviews about specific topics of horticultural models were written by Le Bot et al. (1998) who reviewed the modeling of mineral nutrition in horticultural crops,

PAGE 60

60 Jones and Tardieu (1998) who focused their review on water relationships while Marcelis et al (1998) reviewed the modeling of biomass production and yield of horticultural crops. After 10 years of those reviews the topics pointed as still being in their infancy continue in that sate and models fail to consider, for example, the interaction between pest s and diseases coupled directly to horticultural models, the use of real genetic parameters as commented above, the simulation of plant architecture and morphogenesis, the simulation of fresh matter production rather than only dry mat ter production and the modeling of quality traits among others. Models for tomato have focused on the carbon balance that leads to predictions of fruit growth in dry mass. Some examples are TOMGRO (Jones et al., 1991), CROPGRO tomato (Scholberg et al., 1997 ), HORTISIM (Gijzen et al 1998), TOMSIM (Heuvelink and Bertin, 1994), TOMPOUSSE ( Gary et al ., 1997) SI MULTOM ( Sauviller et al ., 2002). In crop oriented process models the biomass production is based on the carbon balance which depends on the gross phot osynthesis of the canopy in turn closely related to light interception and therefore to the leaf area, leaf photosynthesis and the losses of carbon through respiration process. Leaf photosynthesis is modeled using different equations depending on the mode l, but independent of the function shape used, most of them follow a curve of saturation type where the main parameters are Amax (maximum photosynthetic rate) and the initial slope of the light use efficiency (Gary et al., 1998). Scholberg et al (2000) r eported an Amax value for tomato equal to 1.36 mg CO2 m2 s1 a higher value than values used by other authors such as Heuvelink (1995) who uses a value equal to 29.3 mol CO2 m2 s1 which is equivalent to 1.2 mg CO2 m2 s1. Some models calculate the gross photosynthesis integrating it from leaf level to canopy level. The most simple take the projected leaf area and just multiply it by the unit leaf

PAGE 61

61 photosynthetic rate. All consider also the extinction coefficient of the light. In addition, some mode ls calculate photosynthesis at leaf level using a Farquar s kinetics approach or at canopy level using a rectangular hyperbola function or an exponential curve. The light distribution for tomato canopy which grows in rows needs to take account of the di stribution of direct and diffuses light, the light scattering and leaf angle. An example is the hedge row mod el (Boote and Pickering, 1994). Jones and Tardieu (1998) reviewed the modeling of water relationships in horticultural crops. According to them th e task of simulating water relationships by the models was related to simulat ing the water requirements by the crops for transpiration purpose and soil evaporation. Therefore, models ha ve been based o n the water balance that consider s water inputs (precipi tation and irrigation) and water outputs (canopy transpiration, soil evaporation) and diverse losses such as runoff, deep percolation etc. The evapotranspiration is mostly modeled using energy balance approaches such as the PenmanMonteith or Priestly -Ta ylor equations. The second aspect that has been included in some models is the influence of the plant water status on functions such as stomata aperture, organ growth in extension and assimilates fluxes. However, the tomato fruit is about 95 % water (Davie s and Hobson, 1981) thus the water flux into the fruit is one of the key aspects in the fruit yield formation of this specie. The fundamentals to simulate water and assimilate movement through both phloem and xylem into the fruit were published by Guichar d et al. (1999). A virtual model was published also by Genard and Lescurret (2005) for peach specie. In those models the xylem flux depends basically on the stem water potential while the phloem flux depends mainly on the dry matter concentration of the p hloem sap. Recent ly published is a model for simulat ing carbon and water fluxes into individual fruits of tomato developed by Liu et al. (2007). The water efflux from tomato fruits

PAGE 62

62 was modeled by Leonardi et al. (2000) as a function of radiation and the va por pressure deficit (VPD) Bussieres (1994) on other hand, proposed that the tomato fruit transpiration is a small and constant proportion of the fruit mass. The last approach may be realistic considering that the transpiration of tomato fruit is actually limited due the resistance exerted by the epidermis. Even when a few advances have been published, the task of modeling the water and assimilate fluxes into individual fruits and the n coupling it to a more integrated whole plant model is still in its infa ncy. However, this may be an unavoidable step in order to improve the fruit growth and quality simulation. On the other hand, the latest version of DSSAT (4.5) not yet officially released has incorporated a routine for deriving the fresh weight of indivi dual fruits from the carbon balance and then converting the fresh weight to fruit size. This routine appears to work reasonably well according to initial testing (Boote and Scholberg, 2006 ) but further validation with field data needs to be done. One issue in this derivation is how the model deals with water stress in view of the small er impact of water stress on the carbon balance as compared to its impact on fresh production T herefore further functional modifications might be ne eded As for carbon, mech anistic sub models have accounted for the mineral balance in the plant. However, only the nitrogen is generally considered explicitly and the process of uptake, canopy accumulation and redistribution of N according to partitioning rules and orga n ageing ar e often simulated. With r espect to the soil nitrogen some models ha ve incorporated modules for simulating organic matter and dynamic s of both organic and mineral N in the soil. One example is CROPGRO tomato model which allows the users t o couple the simul ation to specific models for soil organic matter budget such as modified CENTURY ( Gijsman et al., 2002). In summary, the tomato fruit growth has been modeled based on the canopy dry matter production and subsequent accumulation into the fruit. The biomass production by the canopy is

PAGE 63

63 photosynthesis driven. The integrated net photosynthesis normally on a daily basis is converted to biomass and it in turn is converted in to accumulated dry matter according to established rules for partitioning and fruit growth In addition, the biomass production is regulated by the efficiency of conversion of the intercepted radiation in dry matter The water inflow into the fruit has been ignored at least until recent efforts particularly led by French researchers. Bussieres (1993 ) for example published a pioneer ing work simulating the rate of water importation into tomato fruits as dependent on the difference in water potential between the stem and the fruit. Previously Lee (1990) and Gen ar d and Huget (1996) simulated the wate r uptake and transpiration of tomato fruit per area unit as a constant or a variable function, respectively. Regarding crop development, most tomato models are based on the fact that development depends mainly on the temperature and in the range of a toma to plant growing conditions this dependence is linea r. Thus, most tomato models simulate the timing of organ appearance, development, and phase transitions as a function of the accumulated thermal time. Some models use simple equations that originally were formulated to describe the responses of enzymatic process to temperature and follow the classical bell shape curves. Other models are more elaborate; CROPGRO for instance ha s four point functions that follow a rectangular or sinusoidal shape and where ea ch point repr esents a cardinal temperature: b ase, optimal one, optimal two and maximum (failure) temperatures. In addition in some models the fruit number is an input. More realistic are those models such as CROPGRO that predict the phase transition from tomato flowering to fruit setting and then establish a fruit population based on a s equence of cohorts added after the anthesis date and which accounts for the accumulated thermal time and the a ssimilation carrying capacity. The growth of the tomato fruit s depends on the amount of assimilates and water that is assigned to the fruit. The assimilates partitioned to the fruits depend,

PAGE 64

64 in turn, on the capacity of the fruit to attract them (sink strength), and on the equilibrium between generative and vegetativ e sinks. In case of indeterminate cultivars growing in greenhouse the strength of the sink depends on the number and the age of the organs. In fact, the current models for tomato do not specifically discriminate between the more determinate cultivars and t he sink source relationships of cultivars such as those growing in open fields as compared with i ndeterminate cultivars grown in greenhouses. The simulation of the fluxes of dry matter and water into the fruits is not common in tomato models which simulate only the carbon influx and then use empirical relationships to convert the accumulated dry matter in fresh weight. The attempts to couple both fluxes and to mechanistically predict both dry and fresh fruit weight is slowly being studied. At present the th eoretical models for tomato developed by Bussieres ( 1993, 1994,), Fishman and Genard (1998) and its recent adaptation for tomato by Liu et al. (2007) are go od examples of those attempts. However, the number of parameters related primarily to the simulation of the resistance path to both fluxes as well as to the wall extensibility coefficients among other parameters may impose restrictions for practical applications of such models. Nevertheless, they seem to be a valuable tool for increasing the knowledge of the complex behavior of tomato fruit growth.

PAGE 65

65 CHAPTER 3 UPDATING P ARAMETERS IN CROPGRO AND TESTING THE MODEL FOR PREDICT ING GROWTH AND DRY MATTE R ACCUMULATION Introduction Calibration is an important task in the crop modeling process and it can follow different paths. Scholberg et al. (1997) adapted the CROPGRO -model for tomato simulation ; however, no further improvements have been introduced A sub model for predicting fresh weight and fruit size from the dry weight of the fruits will be presented in Chapter 5 This sub model starts with the carbon balance and therefore, biases in dry matter predictions can be propagated to fresh weight prediction which in turn will also affect fruit size simulation For this reason the objective of the present chapt er is to ma k e a literature based adaptatio n of parameters of the CROPGRO T omato model related to the simulation of crop development, daily dry matter (DM) production, DM partitioning and DM accumulation in the crop from trans planting to harvest. The adapta tions took advantage of new literature values for species parameters, especially cardinal temperatures that affect tomato phenology. Recent literature reviewed in Chapter 2 suggested that the cardinal temperatures that drive the phenology and plant process es are different from the values used by Scholberg et al (1997) and that are still present in the current version of the tomato model. The new cardinal temperature values are considered to be reliable as they come from recent published data of experiments conducted in controlled temperature environments. The second objective of this chapter is to calibrate the model using field data in response to these changes and evaluate its capability for predicting growth and dry matter accumulation of the tomato cro p through time Material s and Methods Approach: A systematic approach for model improvement as proposed by Boote et al (2002) was followe d and can be summarized in three steps First, an adaptation of the parameters

PAGE 66

66 was performed based on data reported in the literature. The adapted species coefficients were cardinal temperatures for pre anthesis, anth esis, and post anthesis phases. Second life cycle phase durations and other parameters were calibrated against growth data from field experiments. Third, the quality of the model predictions was evaluated using statistics such as RMSE and the Willmott index as evaluation criteria Calibration (second step) was performed according the following order: A) Life cycle was adjusted with the new cardinal temperatur e by comparison against the phase durations observed during experiments carried out during the spring season in Gainesville Florida in 2006 and 2007 and attempts to predict the observed dates for the occurrence of first flowering, first fruit, and maturity This was required because observed data for phenology wer e not available for most of the experiments and therefore the hy pothesis in this effort was that the life cycle duration of the 20062007 experiments should be reproduced since cultivars for fresh market grown in Florida are similar and no important differences in the phase duration among them have been reported. O n the other hand, the simulated phase durations calibrated using Scholbergs field data and the default paramet ers show ed differences as large as 18 days for time to flowering for the same cultivar, growing during the same season and at the same location. The weather records of those experiments do not explains these huge differences in time to flowering. As a consequence we assume d that such discrepancies were att ributed to the model using incorrect cardinal temperatures for the phenology of the specie. Because the phase duration for cultivars used in Florida for fresh market are similar if they are grow n during the same season we assum e d that one standard cultivar can be created for tomato with no differences in genetic parameters at least for experiments carried out in Florida during spring season. This standard cultivar should have a window for flowering between 23 to 28 days after t ransplanting

PAGE 67

67 (depending on the age of the plant at transplanting) and it should reach physiological maturity between 85 to 95 days a fter transplanting. This is supported by our own observations during our 2006 and 2007 experiments and also by literature th at reports a tomato cycle for fresh market cultivars growing in open field of 2 months from anthesis to maturation and three to four weeks from tra nsplanting to anthesis (Jones, 1999; Mossler et a l ., 2005). If the crop is grow n during fall li kely these t imes could be different and appropriat e calibration may be necessary. B) Parameters affecting leaf growth, dry matter (DM) production, and DM partitioning in the crop from the time transplanting to harvest were then calibr ated against the observed data. The model: The CROPGRO model used in this study is included in the DSSATV4.0 software. A complete description can be found in the software documentation, as well as in the paper published by Jones et al (2003). The CROPGRO model simulates development based on seven phases, from emergence to harvest. Phenological development and some growth processes such as leaf expansion and fruit growth depend on the cardinal temperatures (Rinaldi et al ., 2007). Moreover, leaf expansion depends on the new leaf mass produc ed a nd specific leaf area, which is determined by light, temperature, nitrogen, and water status. The carbon bal ance is mainly a function of intercepted light and leaf photosynthesis. The daily growth of tissues results from the gain of photoassimilates t hrough photosynthesis minus respiration and growth maintenance losses. The allocation of assimilates among vegetative organs is driven by partitioning coefficients, but, once set, fruits have first priority sink strength. Although considered semi determina te, the fresh market tomato cultivars grown under field conditions exhibit very low partitioning to vegetative sinks after the fruits begin their rapid growth phase. Thus, determinacy is simulated because the partitioning of assimilates to the reproductive sinks during the reproductive phase effectively causes leaf, stem, and root growth to terminate to the

PAGE 68

68 limits of (1 -XFRUIT), where XFRUIT is a parameter that determines how much of the available dry matter is allocated to the fruits (Boote and Scholberg, 2006). Sub models for nitrogen and water balance, as well as pest damage coupling simulations, are also available and can be switched on or off according to specific applications. The model requires three genetic files: species, ecotype, and cultivar files The species file accounts for the sensitivity of crop processes to environmental factors such as temperature, solar radiation, CO2, and photoperiod. The cultivar parameters define the life phase duration for each cultivar, and coefficients in the ecotype file represent traits that are common to groups of similar cultivars within species (Boote and Scholberg, 2006) Literature search and file modifications : The default model (V4.0) uses the same cardinal temperature values for the three phenological phases (vegetative, early reproductive, and late reproductive phases), and those values are 10 C, 28 C, and 55 C for base, optimum (both Topt1 and Topt2), and Tmax respectively. With the goal of improving temperature -dependent parameters in the CROPGRO -tomat o model and after a literature search of cardinal temperatures for the species, the values published by Adams et al (2001b) and reviewed in Chapter 2 are valuable because the study was carried out under environments in which the temperature was precisely controlled. Therefore, the adaptation of parameter s for this work is based mainly on the data obtained by Adams et al ., although the work of oth er authors is also considered. Table 3.1 summarizes the cardinal temperatures for both the default model and the modified values based on the literature for: phenological development, fruit addition and pollination, and shell and fruit growth of the tomato species Field data for calibration purposes: The growth data for calibration consisted o f a series of eight ex periments conducted between 1991 and 2007 at three locations in Florida: Bradenton,

PAGE 69

69 Gainesville, and Quincy. Three experiments in Bradenton were conducted by McNeal et al. from 1991 to 1994. The treatments selected for calibration were well irrigated and w ell fertilized, and therefore no water or nitrogen stress was present. The soil utilized belongs to t he Eaugallie fine sand family. The datasets for Bradenton 1995, Gainesville 1996, and Quincy 1995 were derived from field experiments conducted by Scholberg (1995, 1996) using irrigated tomatoes and, excluding the Quincy studies, all experiments were carried out during the spring season. For Quincy, one dataset corresponded to the spring season and the other to the fall season. The Gainesville 2007 datase t was obtained from a field grown plastic mulched fresh-market tomato experiment conducted in the spring of 2007, and the treatment selected for calibration was a well -fertilized and well irrigated treatment. The soil consisted of fine sandy soil of the Mi llhopper series for Gainesville in 1996. For Gainesville in 2007, the soils were fine Candler sand and Tavares sand (Buster, 1979; Dukes et al. 2005). The Quincy soils belonged to the Orangeburg series. The treatment was drip irrigated, and the N applicati on rate was 200 kg ha1 at Gainesville, 180 kg ha1 at Quincy (spring), and 200 kg ha1 at Quincy (fall). For the simulations, the irrigation and N rates are not highly relevant because the N and water balance were turned off for all simulations. The geno types were Sunny for Bradenton, Agriset for Gainesville (1996), Florida 47 for Gainesville (2007), and Agriset (spring) and Solarset (fall) for Quincy However because similarities among cultivar s Sunny which will b e called standard hereafter, was calibr ated for all the experiments ex cept for Quincy during fall when the cultivar S olarset was grown a nd a different set of genetic parameters was calibrated for this cultivar. Results and Discussion Life phase duration: Decreasing the base temperature fr om 10C to 7C, as compared with the default values (Table 3 1), without changing the cultivar phase durations accelerated the

PAGE 70

70 plant development rate and markedly shortened the life phase durations, as expe cted (data not shown). Tomato fruit growth, as sup ported by several authors, has a period of very slow growth which last s about two weeks after anthesis ( Mapelly et al ., 1978; Varga and Bruisma, 1986; Boher and Banghert, 1988; Gillaspi et al ., 1993; Heuvelink, 2005) This lag in fruit growth is not considered in the default version. In our calibration we account for this slow growth and modified one specie parameter called shell lag rate (SHLAG) changing its value f ro m 0 to 0.1. The interpretation of this value is that during the days that immediately follow anthesis a single tomato fruit grows at a rate that is 10 % of the rapid growth phase (Dorais et al 2001). This modification will be show n in the next chapters and was needed in or der to appropriately model the growth of individual tomato fruits tagged at different dates. Table 3 2 shows the results for the calibration of parameters in the ecotype file needed to readjust the life cycle ( according our observations ) and carbon balance after changing the cardinal temperatures in the specie file. The main changes in ecotype files are related to an increase in the required thermal time to complete vegetative phases. Furthermore, the rate of truss appearance was slightly reduced from 0.52 in the default version to 0.5 according to the De Koning (2000) da ta. The ecotype PM06 coefficient was also calibrated. In CROPGRO models, this parameter was developed for peanut. It represents a fraction of the time between fir st peg and first seed in the fruit during which the fruit grows slowly If the PMO6 in the eco type file is equal to 0.0, the first peg and the first fruit occur at the same time. If PMO6 is > 0.0, a slow growth phase occurs prior to the reported first fruit In addition the age of the plant s at the time of transplanting was adjuste d for each exper iment (Table 3 3 ). For the simulation, plants were generally 1 to 3 days younge r at the time of transplanting than used in the default version. The real age of plants for those experiments are not available and therefore the criteria for this

PAGE 71

71 adjustment wa s a better phase duration adjustment when cardinal temperatures were updated considering that usually plant ages at transplanting range between three and four weeks in Florida production systems. Tables 3 4 (standard cultivar) and 3 5 ( Solarset cultivar ), show the results of the parameter calibration performed to adjust the life cycle after entering the new cardinal tem peratures in the specie file. The main changes in the cultivar file were the parameters that determine the phase durations. First, the ti ming of flower and fruit appearance was adjusted. The cultivar parameter EM FL (days between plant emergence and flowering) was calibrated to a greater value, (it was increased by 11 %) in order to predict the anthesis date. Next to match the timing from f lowering until the start of slow fruit growth, the cultivar coefficient FL-SH (thermal days between flowering and shell growth) was calibrated to a smaller value. The criteria for this calibration was that the model had to correctly predict the timing of f lowering in order to simulate the development and growth over time of three cohorts whose development and growth began at different dates (separated by one week) based on the tagging dates Our experiments during 2006 and 2007 indicate d that first fruit we re set 26 and 28 days after transplanting respectively. As a consequence, our calibration process, in addition to the changes in cardinal temperatures, needed to calibrate to these observed fruit set ting dates Fruit set (small, but slow growing fruits) o ccurs within few days (FL SH=3) after flowering. The ability of the mode l to correctly predict the time for first fruit is relevant for our purposes of improving the whole model and the fundamentals related to tagged individual fruit growth will be fully e xplai ned in Chapter 5. In addition, t he cultivar coefficient FL -SD (time from flower to fir st seed) was increased by 18%. In order to improve the growth of fruits and seeds, the coefficients PODUR (duration of pod addition) and SFDUR (duration of seed fill ing) required calibration. These parameters were

PAGE 72

72 increased by 18% and 4 %, respectively. In addition, in order to give more priority to shell growth over seed growth, the threshing perc entage was slightly decreased. The coefficient SL A VAR (specific leaf ar ea under standard growing conditions) was also slightly increased from 350 to 360. After calibration, the model was able to reproduce the dates for anthesis, first fruits, and physiological maturity in our observations This also agree s with literature sho wing that ant he sis day occurs between 3 and 4 weeks after transplanting and that the whole life cycle from transplanting to harvest take s about 90 days. In addition the individual cohorts tagged at different dates during 2006 and 2007 seasons were adequat ely tracked by the model which was useful for later studies that we performed with the model (Chapter 5). The only exception was the 1995 Quincy experiment for the fall season in which the Solarset cultivar was grown It was not possible to accurately cal ibrate this experiment because phenology data were not available Th erefore, for this cultivar a preliminary evaluation of the ability of the model to predict growth and yield data during fall season was done and the cultivar coefficients used are shown in Table 3 5. However, more plant phenology data will be necessary to adequately calibrate the Solarset cultivar to confirm our coefficient values Leaf area index : For most of the experiments, the progression of leaf growth was well simulated by the updated version ( Figure 3 1 to 3 4 ). In general, the updated parameters demonstrated better performance for simulating LAI compared with the default version (Tables 3 6 and 3 7) Using the updated parameters the average error for all the experiments was 34% lower than using default values. The Willmot t d index was higher using updated parameter s than using the default ones (better correspondence between simulated and observed values). Nevertheless, the results still show two weaknesses in the model: 1) in both ver sions and for

PAGE 73

73 several experiments, the maximum LAI was overestimated, and the only exceptions were Bradenton 1992 and 1994, for which LAI was underestimated. Because these differences it is hard to obtain L AI simulation s that satisfy the observations for all the experiments. In those years when L AI was overestimated the model failed to reproduce the change in growth after the maximum LAI was attained. Therefore, the trend for early growth was relatively well simulated, although it displayed a small overes timation in the maximum LAI attained. Later in the season, the model, regardless of the version, was not able to efficiently reproduce the decay in leaf area associated with s enescence and N remobilization. This may be fixed by adjusting the remobilization functions in the model, such as the maximum rate of remobilization of protein (NMOBMX) and carbohydrates (CMOBMX) from vegetative to reproductive tissues. These functions in CROPGRO determine how much N and C are remobilized according the age of the leaf, so if, for instance, the rates of remobilization are increased, leaf mass decline may occur naturally as a result of senescence. These changes were made (data not shown) but not used because while it improved LAI simulations for some experiments it made t hem worse in others O verall the simulation s were better leaving the remobilization functions unchanged. Simulations of stem or leaf weight improved using the updated parameters (data not shown) The specific leaf area ( SLA ) was not very well simulated by the model, regardless of the version used. In addition, the Willmott d values for SLA were low, indicating that the variability in the mean SLA throughout the season was not well captured by the model One possible explanation is that the model SLA depends basically on radiation and temperature. However, ontogenetic changes, nutrition, stresses are factors that may affect SLA a fact that is disregarded by the model.

PAGE 74

74 Aboveground biomass and fruit yield: T he total ab ove biomass and fruit yield was well simul ated using the updated parameters ( Figures 3 5 to 3 12) On average, simulations using the updated parameters produced a n RMSE that was 12 % lowe r for total biomass yield and for fruit yield as compar ed with the default version (Tables 3 6 and 3 7). Becaus e the comparisons were made using time series data the Willmott d index is a better indicator of model improvement. Overall, the d index was higher for total dry weight and fruit dry weight using updated parameters as compared with default parameters. On average, the d index for the updated version was 0.99 and 0.96 for total above biomass and fruit dry weight respectively while the default version produced an average d value equal to 0.98 for total above biomass a nd 0.91 for total fruit weight. Fruit nu mber : The fruit number was well simulated by the model when updated parameters wer e used (Figures 3 13 to 3 15). The default version produced poor simulation of fruit number always overestimating this variable with an average error equal to 33 fruits per square meter and a d value equal to 0.77 (Table 3 6). The error averaged over all experiments using the updated values was equal to 16 fruits per square meter and the average d value was 0.93 ( Table 3 7). Averaged for all experiments the RMSE was 58% lowe r using the updated values comparing with the default version. The reason for the large PODUR (pod addition duration) value, is that the model computes this based on seed growth rate (demand) relative to assimilate supply. This feature is unrealistic for c rops with low THRESH (ratio of seed to fruit plus seed). Conclusions After changing the cardinal temperature s for plant phenology using recently published data, it was necessary to increase the required thermal time for completing each phase in order to

PAGE 75

75 pr edict the life cycle of the plant. The da ta obtained for transplant age are frequently absent or not r ecorded when field data are collected for calibration purposes. However, differences of two or three days for plant age at the time of transplant ing are common in real practice and can significantly modify the phase duration in tomato, especially the thermal time accumulated from planting to anthesis, which, if calculated incorrectly, can propagate error to the other phases. Therefore, it is highly recomme nded that this information be documented when possible in order t o obtain a correct calibration. The results obtained for the LAI simulations suggested that the model still requires adjustment s to correct the overestimation or underestimation of LAI depending on the experiment. In addition, in order to better adjust leaf growth, the model should be tested with cultivars that have both determinate and indeterminate growth. Until now, the model has only been calibrated with field data using near determinate cultivars growing in open field, and therefore assumptions regarding the behavior of the model in greenhouse conditions cannot be made until an appropriate evaluation is performed. It is anticipated that the way in which the model simulates the number and growth of stems should be studied and likely modified along with the gro wth of other sinks in the plant Overall, the ability of the CROPGRO tomato model to simulate leaf growth, total biomass, and fruit dry weight yield based on the RMSE and Willmott d i ndex values were improved using the updated version compared with the default version. Therefore, we recommend using the newly calibrated values of the genetic coefficients for the next release version of the CROPGRO T omato model. The proposed adaptation based on reliable values of cardinal temperatures should not only improve the accuracy of the tomato biomass and dry weight yield predictions but also reduce efforts in future model calibrations and evaluations

PAGE 76

76 Table 3 1. Parameters represent temperature depen dence of tomato phenology, fruit and seed growth, default V4.0 values and updated values in specie file. Parameter Rat e of leaf or truss appearance (vegetative development ) Default value (C) Updated value (C) T b C 10 7 T op1 C 28 22 T op2 C 28 28 T max C 55 48 Rate of progress to anthesis /truss appearance (e arly reproductive growth) Default Updated T b C 10 7.2 T op1 C 28 22 T op2 C 28 28 T max C 55 48 Rate of fru it development and maturation (l ate reproductive growth) Default Updated T b C 10 5.7 T op1 C 28 26 T op2 C 28 28 T max C 55 48 R elative effect of t emperature on rate of f ruit addition and pollination. Default Updated T b C 6 7.2 T op1 C 8 22 T op2 C 28 25.5 T max C 30 32 Relative effect of t emperature on rate o f individual seed/fruit growth. Default Updated T b C 6 6 T op1 C 8 22 T op2 C 25.5 25 T max C 32 32

PAGE 77

77 Table 3 2 Additional modified parameters in ecotype file, d efault V 4 0 values and calibrated values Parameter Default value Calibrated value Time between planting and emergence PL EM (Td)** 5 6 Time between emergence and first true leaf EM V1 (Td) 20 22 Tim e required for growth of individual fruit LNGSH (Td) 35 39 Time between first flower and last leaf in main stem FL VS (Td) 18 23 PMO6, if PM06 is 0.0, first peg and first pod occur at the same time. If PMO6 > 0.0 there is a slow growth phase 0.0 0.6 Rate of leaf appearance on main stem, TRIFL 0.52 0.5 Maximum ratio of (seed/(seed+shell)) at maturity THRSH (%) 9.2 8.5 **Td=thermal days Table 3 3. Tomato plant age at transplanting in CROPGRO default V4.0 values and calibrated values in experimental f ile (X) Experiment Default Calibrated Bradenton 1991 28 25 Bradenton 1992 28 25 Bradenton 1994 28 26 Bradenton 1995 28 25 Quincy 1995 (spring) 28 26 Quincy 1995 (fall) 28 25 Gainesville (1996 ) 28 26 Gainesville (2007) 28 28

PAGE 78

78 Table 3 4 Cultivar coefficients in CROPGRO Tomato model, default and calibrated values after updating the card inal temperatures in SPE file. Standard cultivar. Parameter Definition Default Calibrated EM FL Time between plant emergence and flower appearance (R1) ( thermal days) 23 2 5 FL SH Time between first flower and first pod (R3) (thermal days) 8 3.0 FL SD Time between first flower and fi rst seed (R 5 ) (thermal days) 17 20 SD PM Time between first seed (R5) and physiological maturity (R7) 50 50 FL LF Time bet ween first flower (R1) and end of leaf expansion (thermal days) 50 50 LFMAX Maximum leaf photosynthesis rate at 30 C, 350 vpm CO2, and high light (mg CO2/m2 -s) 1.36 1.36 SLAVR Specific leaf area of cultivar under standard growth conditions (cm2/g) 350 36 0 SIZLF Maximum size of full leaf (three leaflets) (cm2) 300 300 XFRUT Maximum fraction of daily growth that is partitioned to seed + shell 0.75 0.75 WTPSD Maximum weight per seed (g) 0.0040 0.0040 SFDUR Seed filling duration for pod cohort at standa rd growth conditions (thermal days) 25 26 SDPDV Average seed per pod under standard growing conditions (#/pod) 300 300 PODUR Time required for cultivar to reach final pod load under optimal conditions (thermal days) 42 55

PAGE 79

79 Table 3 5 Cultivar coefficients in CROPGRO Tomato model, default and calibrated values after updating the card inal temperatures in SPE file. Cultivar: Solarset Parameter Definition Default Calibrated EM FL Time between plant emergence and flower appearance (R1) (thermal da ys) 23 32 FL SH Time between first flower and first pod (R3) (thermal days) 8 5.0 FL SD Time between first flower and first seed (R5 ) (thermal days) 17 22 SD PM Time between first seed (R5) and physiological maturity (R7) 50 52 FL LF Time between first flower (R1) and end of leaf expansion (thermal days) 50 52 LFMAX Maximum leaf photosynthesis rate at 30 C, 350 vpm CO2, and high light (mg CO2/m2 -s) 1.36 1.36 SLAVR Specific leaf area of cultivar under standard growth conditions (cm2/g) 350 360 SIZLF Maximum size of full leaf (three leaflets) (cm2) 300 300 XFRUT Maximum fraction of daily growth that is partitioned to seed + shell 0.75 0.75 WTPSD Maximum weight per seed (g) 0.004 0.004 SFDUR Seed filling duration for pod cohort at standard growth co nditions (thermal days) 25 26 SDPDV Average seed per pod under standard growing conditions (#/pod) 300 300 PODUR Time required for cultivar to reach final pod load under optimal conditions (thermal days) 42 60 Table 3 6 Root mean square error and Wil l mot t d index for simulations using default genetic parameter values Experiment Leaf Area Index Total Above Biomass Total Fruit Weight Fruit Number RMSE d RMSE d RMSE d RMSE d BR1991 0.91 0.9 447 0.99 678 0.95 42 0.70 BR1992 0.81 0.93 449 0.99 282 0.99 34 0.88 BR1994 1.10 0.92 1040 0.98 8 74 0.97 51 0.75 BR1995 0.88 0.89 658 0.99 772 0.87 nd QU1995 (S) 0.97 0.95 886 0.97 629 0.97 nd QU1995 (F) 0.36 0. 96 485 0.98 697 0.93 nd GA1996 1.14 0.87 445 0.99 519 0.86 31 0.76 GA2007 1.66 0.79 656 0.99 667 0.79 37 0.78 Average 0.98 0.90 633 0.98 639 0.91 39 0.77

PAGE 80

80 Table 3 7 Root mean square error and Wil l mot t d index for simulations using updated genetic parameter values Ex periment Leaf Area Index Total Above Biomass Total Fruit Weight Fruit Number RMSE d RMSE d RMSE d RMSE d BR1991 0.45 0.98 645 0.99 856 0.94 12 0.85 BR1992 1.0 0.85 686 0.99 576 0.99 12.6 0.97 BR1994 1 0.88 720 0.99 712 0.985 23 0.93 BR1995 0.76 0.91 440 0.99 516 0.92 nd QU1995 (S) 0.32 0.99 391 0.99 572 0.97 nd QU1995 (F) 0.69 0.93 633 0.99 583 0.95 nd GA1996 0.5 0.96 604 0.99 510 0.96 9 0.96 GA2007 0.43 0.97 330 0.99 178 0.99 23 0.91 Average 0. 65 0.93 556 0.99 562 0.96 16 0.92

PAGE 81

81 0 1 2 3 4 5 0 20 40 60 80 100 Days After Planting Leaf Area Index (m2 m-2) LAI (Simulated) BR91 LAI (Simulated) BR92 LAI (Measured) BR91 LAI (Measured) BR92 Figure 3 1. Simulated and observed Leaf Area Index for tomato standard cultivar in Bradenton, Fl during spring of 1991 and 1992 using u pdated parameters (Table 3 4) 0 1 2 3 4 5 6 0 20 40 60 80 100 Days After Planting Leaf Area Index (m2 m-2) LAI (Simulated) BR94 LAI (Simulated) BR95 LAI (Measured) BR94 LAI (Measured) BR95 Figure 3 2. Simulated and observed Leaf Area Index for tomato standard cultivar in Bradenton, Fl during spring of 1994 and 1995 using updated parameters (Table 3 4).

PAGE 82

82 0 1 2 3 4 0 20 40 60 80 100 Days After Planting Leaf Area Index (m2m-2) LAI (Simulated) GA96 LAI (Simulated) GA07 LAI (Measured) GA96 LAI (Measured) GA07 Figure 3 3. Simulated and observed Leaf Area Index for tomato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using updated parameters (Table 3 4). 0 1 2 3 4 5 0 20 40 60 80 100 Days After Planting Leaf Area Index (m2 m-2) LAI (Simulated) QU95(spring) LAI (Simulated) QU95(fall) LAI (Measured) QU95(spring) LAI (Measured) QU95(fall) Figure 3 4. Simulated and observed Leaf Area Index for t omato in Quincy, Fl during 1995 standard cultivar (spring) and S olar set cultivar (fall) using updated parameters (Table 3 4)

PAGE 83

83 0 2000 4000 6000 8000 10000 12000 0 20 40 60 80 100 120 Days After Planting Total Above Biomass (kg ha-1) Tops Wt. (Simulated) BR91 Tops Wt. (Simulated) BR92 Tops Wt. (Measured) BR91 Tops Wt. (Measured) BR92 Figure 3 5 Simulated and observed total above biomass for t omato standard cultivar in Bradenton Fl, during spring of 1991 and 1992 using updated parameters (Table 3 4) 0 2000 4000 6000 8000 10000 12000 14000 0 20 40 60 80 100 120 Days After Planting Total Above Biomass (kg ha-1) Tops Wt. (Simulated) BR94 Tops Wt. (Simulated) BR95 Tops Wt. (Measured) BR94 Tops Wt. (Measured) BR95 Figure 3 6. Simulated an d observed total above biomass for t omato standard cultivar in Bradenton, Fl during spring of 1994 and 1995 using updated parameters (Table 3 4)

PAGE 84

84 0 2000 4000 6000 8000 10000 0 20 40 60 80 100 Days After Planting Total Above Biomass (kg ha-1) Tops Wt. (Simulated) GA96 Tops Wt. (Simulated) GA07 Tops Wt. (Measured) GA96 Tops Wt. (Measured) GA07 Figure 3 7. Simulated and observed total above biomass for t omato standard cultivar in Gainesville, Fl durin g spring of 1996 and 2007 using updated parameters (Table 3 4). 0 2000 4000 6000 8000 0 20 40 60 80 100 Days After Planting Total Above Biomass (kg ha-1) Tops Wt. (Simulated) QU95 (spring) Tops Wt. (Simulated) QU95 (fall) Tops Wt. (Measured) QU95 (spring) Tops Wt. (Measured) QU95 (fall) Figure 3 8. Simulated and observed total above biomass for t omato in Quincy, Fl during 1995 standard cultivar (spring) and S olarset cultivar (fall) using updated parameters (Table 3 4 and 3 5 ).

PAGE 85

85 0 2000 4000 6000 8000 0 20 40 60 80 100 120 Days After Planting Fruit Dry Weight (kg ha-1) Fruit Dry Wt. (Simulated) BR91 Fruit Dry Wt. (Simulated) BR92 Fruit Dry Wt. (Measured) BR91 Fruit Dry Wt. (Measured) BR92 Figure 3 9. Simulated and observed total fruit dry weight for t omato standard cultivar in Bradenton, Fl during spring of 1991 and 1992 using updated parameters (Table 3 4) 0 2000 4000 6000 8000 10000 0 20 40 60 80 100 120 Days After Planting Fruit Dry Weight (kg ha-1) Fruit Dry Wt. (Simulated) BR94 Fruit Dry Wt. (Simulated) BR95 Fruit Dry Wt. (Measured) BR94 Fruit Dry Wt. (Measured) BR95 Figure 3 10. Simulated and observed total fruit dry weight for t omato st andard cultivar in Bradenton, Fl during spring of 1994 and 1995 using updated parameters ( Table 3 4)

PAGE 86

86 0 2000 4000 6000 0 20 40 60 80 100 Days After Planting Fruit Dry Weight (kg ha-1) Fruit Dry Wt. (Simulated) GA96 Fruit Dry Wt. (Simulated) GA07 Fruit Dry Wt. (Measured) GA96 Fruit Dry Wt. (Measured) GA07 Figure 3 11. Simulated and observed total fruit dry weight for t omato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using updated parameters (Table 3 4) 0 2000 4000 6000 0 20 40 60 80 100 Days After Planting Fruit Dry Weight (kg ha-1) Fruit Dry Wt. (Simulated) QU95 (spring) Fruit Dry Wt. (Simulated) QU95 (fall) Fruit Dry Wt. (Measured) QU95 (spring) Fruit Dry Wt. (Measured) QU95 (fall) Figure 3 12. Simulated and observed total fruit weight for t omato in Quincy, Fl during 1995 standard cultivar (spring) and S olarset cultivar (fall) using updated parameters (Table 3 4 and 3 5)

PAGE 87

87 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 Days After Planting Number of fruits (fruit m-2) Fruit number (Simulated) BR91 Fruit number (Simulated) BR92 Fruit number(Measured) BR91 Fruit number (Measured) BR92 Figure 3 13. Simula ted and observed fruit number for tomato standard cultivar in Bradenton, Fl during spring of 1991 and 1992 using updated parameters (Table 3 4) 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 Days After Planting Number of fruits (fruit m-2) Fruit number (Simulated) BR94 Fruit number(Measured) BR94 Figure 3 14. Simulated and observed fruit number for tomato standard cultivar in Bradenton, Fl during spring of 1994 using updated parameters (Table 3 4).

PAGE 88

88 0 20 40 60 80 100 120 140 0 20 40 60 80 100 Days After Planting Number Fruits (fruit m-2) Fruit number (Simulated) GA96 Fruit number (Simulated) GA07 Fruit number(Measured) GA96 Fruit number (Measured) GA07 Figure 3 15. Simulated and observed fruit number for tomato standard cultivar in Gainesville, Fl during spring of 1996 and 2007 using updated parameters (Table 3 4).

PAGE 89

89 CHAPTER 4 EQUATIONS FOR GROWTH OF TOMATO FRUITS UNDER NON -LIMITING CONDITIONS Introduction Several authors have reported that the potential growth of individual tomato fruit exhibits a simple sigmoid or S shaped curve (Gustafson, 1926; Nitsh 1953; Coombe, 1976; Monselise et al 1978; Wolf an d Rudich, 1988; Gillaspi et al., 1993; Chamarro, 1995, Ho and Hewitt, 1986; Heuvelink, 2005). This curve is characterized by three phases. The growth is slow immediately after fertilization, increases gradually to a maximum rate, and finally decreases as t he fruit approaches maturation. During the first stage of fruit development, cell division and enlargement result in slow growth. Following fertilization in the tomato, cell division is activated in the ovary and continues for about 2 weeks (Mapelly et al 1978; Varga and Bruisma, 1986; Boher and Banghert, 1988; Gillaspi et al ., 1993; Heuvelink, 2005). After ~2 3 weeks of slow growth, rapid growth begins. During this time, the cells continue to enlarge by cell expansion. Rapid growth continues for ~3 5 wee ks, culminating in the mature green stage. At this point, the tomato has accumulated the majority of its final weight (Ho and Hewitt, 1986). Finally, when the biochemical changes related to ripening b egin, growth becomes slow again Approximately ten days before the first break of color, growth ceases completely. This occurs because an abscission layer is formed between the calyx and the fruit, which becomes a barrier to the transport of water and assimilates t o the fruit (Heuvelink, 2005). Theoretical grow th functions have been widely used to study plant growth. The construction of growth curves is a valuable tool for dynamic analysis of fruit growth behavior, and growth curves may provide a knowledge base for modeling purposes. Any growth variable, e.g. dr y weight, fresh weight or fruit diameter, can be plotted as a function of time. Additionally, if growth is analyzed as a function of the physiological age of the fruit rather than of the calendar

PAGE 90

90 time, then temperature is incorporated into the analysis, en hancing the predict ive value of the growth curves. Regression analysis is a mathematical tool to establish the relationship between two variables (Steel and Torrie, 1992). For the tomato, nonlinear regression has been used to develop models of fruit growth The two adjusted models that are often reported to successfully fit the sigmoid-shaped growth of tomato fruit are the logistic model (Gustafson, 1927; Calado et al., 1980; Monteiro, 1983; He and Zhang, 2006) and the Gompertz function (De Koning, 1994; Be rtin, 1995; Adams et al ., 2001; Enriquez et al ., 2005; Bertin et al ., 2008). According to Bertin et al. (2008), the Gompertz function is more appropriate than the logistic model to describe the growth of tomato fruit, because the Gompertz function better a ccounts for the slow increase in size at the start of the growth period. The Gompertz function is suitable for a sigmoid growth curve in which growth is frequently not symmetrical about an inflection point, i.e., when the relative growth rate (RGR) decreas es with time (Erickson, 1976), as has been shown for tomato by Monselise et al (1978). A sigmoid growth function ideally originates at (0,0), with a point of inflection occurring early in the growth period, and either asymptotically approaches a maximum value or peaks and falls in the senescent phase (Khamis et al ., 2005). Several criteria are used to compare alternative models of nonlinear regression. In general, the preferred model is the one that has a lower mean square error (MAE), a smaller number of parameters (simplicity principle), and a lower standard error (SE) of the estimates (Infostat, 2002). Examples of mathematical functions used to analyze fruit growth of horticultural crops can be found in Marcelis et al. (1998). In addition to time, enviro nmental factors that have a strong influence on growth may be used to model growth curves. Lakso et al (1995) have tested the use of the expolinear model of Goudriaan and

PAGE 91

91 Monteith (1990), to simulate the growth of apple fruits (Opara, 1999). This model ha s three parameters: the maximum absolute growth rate (weight gain per day reached in the linear phase), the relative growth rate (weight gain per unit weight per day), and the xaxis intercept of the linear growth phase, called the "lost time". The expolin ear model provides a good fit to the growth patterns of two apple cultivars that differ in their rates of growth during the exponential phase due to differences in crop load (Lasko et al., 1995). This fact could be very useful in tomato, because tomato gro wth is characterized by considerable variation in sink strength among fruits (De Koning, 1989; Bertin, 1995; Bruchon and Genard, (1999). Ho (1992) found that the dry matter distribution in tomato fruits is strongly linked to the date of fruit initiation. D e Koning (1989) has observed that individual tomato fruit growth is delayed when the competition for assimilates is high, and that delayed fruits started to grow again when the first trusses reached maturity. To explain this phenomenon, Bertin (1995) used the term fruit delay or fruit latency to describe fruits that have been set but whose growth and development are de layed during their early stage. T hese fruits reassume their growth after a delay of 10 to 50 days, when the first fruits of the earlier i nflorescences have ripened. Moreover, although these delayed fruits come to maturity, their final size is smaller than that of the earlier fruits. The results of Bertin (1995) agree with those of other authors who have also observed this fruit growth latency caused by competition for assimilates (Ho, 1984; Picken, 1984; Wolf and Rudich 1988; De Koning, 1989). In addition, Bertin (1995) has experimentally verified that the delay is not produced by parthenocarpy, but is due to competition for assimilates among fruits in a truss and especially among trusses, with earlier fruits having higher sink strength than later fruits. In addition, Monteiro (1983) has reported that the growth rates of tomato fruits of different ages are significantly different.

PAGE 92

92 The object ive o f this chapter is to examine the growth of individual tomato fruits whose development began on different dates. Material s and Methods Field data: The growth data for evaluation were obtained from experiments conducted in field -grown, plastic -mulched f resh -market tomato plots between April and July of 2006 and repeated in 2007. The experimental area was located at the University of Florida Plant Science Research and Education Unit at Citra, Florida (29 25 N, 82 10 W). The treatment selected for eval uation of fruit growth was well irrigated and fertilized; therefore, no water or nitrogen stress was present at any time during the growing season. The soil at this site is classified as Candler fine sand and as Tavares sand (Buster, 1979; Dukes et al. 2005). These soils contain 97% sand sized particles and have a field water holding capacity of 5.0% to 7.5% by volume in the upper 100 cm of the profile (Carlisle et al., 1988; Dukes et al., 2005). The cultivar evaluated was Florida 47, a midto late -season hybrid whos e fruits are deep globe shaped. Eac h replicate plot consisted of four 15.2 m -long rows in completely mulched, fumigated beds, into which 4 -wk old plants were transplanted on April 04 in 2006 and April 07 in 2007. Row spacing was 1.83 m and plant spacing was 0.45 m, making a total of 11,960 plants ha1. Pre -planting fertilizer applications were 112 kg P2O5 ha1 and 45 kg K2O ha1. Treatments were replicated four times using a randomized complete block design. Water application was based on calcu lated crop evapotranspiration (ET). Nitrogen application rates were based on IFAS recommendations, consisting of 224 kg N ha1, which was applied with the irrigation water. Irrigation was applied with a drip tape system; emitters were spaced 20 cm apart. C limatic data, including temperature, solar radiation, rainfall, wind and humidity, were collected by an automatic weather station located within 1 km of the experimental area.

PAGE 93

93 Growth analysis of individual f ruits : Three sets of flowers separated by one week in age were tagged at anthesis. For each cohort date, at least 60 flowers were tagged in each replicate (modified from Heuvelink, 1995). Starting 3 days after anthesis, and two times per week, two tagged fruits in each plot were randomly sampled at 08.00 h in the morning. Therefore, samples were collected at 3, 7, 11, 14, 18, 22, 26, 30, 33, 37, 41, 45, 49 and 53 days after tagging. For each sampled fruit, the fruit diameter was measured, the fresh weight (FW) was recorded, and the dry weight (DW) was det ermined. The experimental data on fruit growth (DW, FW, and fruit diameter ) were fitted to three parameters of the Gompertz function in order to analyze the fruit growth as a function of time (Bertin, 1995). The data on fruit dry matter concentration (DMC) were fitted to a four parameter Gompertz function with displacement for analy sis of DMC variation over time. Results and Discussion Fruit dry weight and fresh weight growth: Figures 4 1, 4 2, 43 and 4 4 present the experimental data for individual tomato fruits (DW, FW, dry matter concentration (DMC) and fruit diameter) measured during the year 2007 and show their fit to parameters of the Gompertz functions given in Table 4 1 The growth of tomato fruits in DW, FW and diameter was well des cribed by the Go mpertz function The best relationship was for cohort 1. Cohorts 2 and 3 have a similar mean square error (MAE) and standard error (SE) to those of cohort 1 f or the ment of cohorts 2 and 3 (Table 4 1 The parameter followed the final fruit dry weight, fresh w eight or diameter. Tomato fruits showed the classical sigmoid growth curve, albeit with a nearly linear mi ddle phase (Figures 4 1, 4 2 and 4 3). Growth began with a phase of slow growth for a few days after anthesis, then increased to an almost linear slop e for about 4 to 5 weeks, and then slowed again close to maturation. As shown,

PAGE 94

94 the curves are largely convex but have a lag period lasting between 9 and 14 days depending on the cohort. The lag periods were shorter for earlier than for later cohorts. Simi lar results were obtained when the 2006 data was fitted to a Gompertz function (data not shown). The DMC data of 2007 were fit better by a four parameter Gompertz function with displacement than by a simple Gompertz function. Th es e results are shown in Tab le 4 1 and Figure 4 4 Measured fruit growth : The observed differences in the final fresh weights, dry weights and fruit diameters of fruits whose development began on differen t dates are presented in Table 4 2 The total dry and fresh weight differed sign ificantly among cohorts. Fruits that developed from flowers tagged earlier (first cohort) achieved the highest mass (both dry and fresh). Fruits from the second cohort achieved lower final mass than those of the first cohort, but higher final mass than tho se of the third cohort. In addition, the difference between cohorts 1 and 2 was much smaller than the difference between cohorts 1 and 3, indicating very slow growth of the latest setting fruits. These differences among cohorts in the final fresh and dry w eights are partly attributable to differences in the length of the fruit growth period, which was 53 days for cohort 1, 46 days for coh ort 2 and 37 days for cohort 3. In addition, there were differences among cohorts in the rate of growth. The rate of dry mass accumulation estimated from the first derivative of the Gompertz function shows that the three cohorts did not hav e the same growth rate (Figure 4 5). The maximum growth rate was reached relatively earlier in cohort 1, occurring on day 22, which repre sents about 4 0% of the total growth period. Cohort 2 also reached the maximum growth rate on day 22, representing about 45% of the total fruit growth period. Similarly, Bertin (1995) and Dorais et al (2001) have reported that the maximum growth rate in to mato fruits is reached by day 21 to 25 after anthesis. A larger lag was seen in cohort 3. For this cohort, the maximum growth rate occurred on day 26,

PAGE 95

95 after almost 63% of the total fruit growth period. Figure 4 5 shows how the latest set fruits, after over coming their long initial lag phase, reached a relatively high maximum growth rate. This indicates that growth accelerated during the phase of rapid growth in this cohort. Values of tomato fruit maximum growth rate reported in the literature range from 0.2 g dry matter d1 (Jones et al. 1991) to 0.37 g dry matter d1 (Dorais et al, 2001). However, Bertin (1993) has reported a value as high as 1.04 g dry matter d1. Variation in growth rate among cohorts has also been reported by Monteiro (1983), Ruddich and Wolf (1988) and Bertin (1995), who found that the rate of dry matter accumulation was significantly different among fruit s initiated on different dates. Fruit dry matter concentration: The final fruit dry matter concentration was similar for the three coh orts, and no significant difference s were found among them (Table 4 2 ). However, the fact that a modified Gompertz function (with displacement) was needed for a better fit indicates that the pattern of dry matter concentration varied among cohorts, and tha t this variation was related to the timing of fruit initiation. Fruit size: The mean final fruit size differed significantly am ong cohorts, as shown in Table 4 2 Fruits that developed from flowers tagged earlier achieved the greatest diameter. The fruits that developed from the second cohort achieved smaller diameters than those of the first cohort, but were much larger than fruits from the third cohort. These differences might be attributable to the same factors proposed to explain the differences among c ohorts in final fresh and dry weights. In addition, during the 2007 growing season, fruits from cohort 3 did not reach commercial size. Conclusion The growth of individual tomato fruits expressed in terms of dry weight, fresh mass or fruit size follows a sigmoid curve, and is well represented by a three -parameter Gompertz function.

PAGE 96

96 This function is adequate to reproduce the lag in the growth of tomato fruits early in the cycle (between one and two weeks after anthe sis, depending on the cohort). Differences in growth are mainly related to the timing of initiation of the fruits, which determines the duration of the growth period. The sink strength of earlier -set fruits appears to be stronger than that of later -set fruits. The dry matter concentration is bette r represented by a four -parameter Gompertz function than by the classical three -parameter function. Howev er, the four -parameter in the function show s greater uncertainty, as evidenced by the higher S.E. of the estimates for the second and third cohorts.

PAGE 97

97 Table 4 1 Estimated coefficients solved by fitting the 2007 data to a Gompertz function to predict DW, FW and fruit diameter. Parameter Fruit Dry Weight (g) = alpha*exp( beta*exp( gamma*DAT)) Cohort 1 S.E Cohort 2 S.E Cohort 3 S.E 15 .31 0.69 9.22 0.42 6 .2 0.39 5.98 1.00 124 13.7 99 75 0.08 0.01 0.23 0.07 0.17 5.43 Parameter Fruit Fresh Weight (g) = alpha*exp( beta*exp( gamma*DAT)) Cohort 1 S.E Cohort 2 S.E Cohort 3 S.E 313 11.13 183 6.13 87.0 5.39 8.13 1.48 289 290 257 200 0.09 0.01 0.25 0.04 0.20 0.03 Parameter Fruit Diameter (cm) = alpha*exp( beta*exp( gamma*DAT)) Cohort 1 S.E Cohort 2 S.E Cohort 3 S.E 11.47 0.41 7.65 0.19 4.39 0.61 8.54 0.09 1.46 0.01 68.5 0.20 33.0 0.02 39.94 0.09 27.5 0.01 Parameter Fruit DMC (%) = alpha*exp( beta*exp( gamma*DAT))+delta Cohort 1 S.E Cohort 2 S.E Cohort 3 S.E 108 0.84 121 0.71 47 14.7 1.15 0.01 1.97 0.01 1.36 0.02 1.36 0.02 0.03 0.02 2.16 6.8 0.09 0.09 0.14 35.4 0.19 33 Table 4 2 Means comparisons among cohorts for dry weight, fresh weight, dry matter concentration and fruit diameter of individual tomato fruits, measured during the spring of 2006 and 2007 at Gainesville, FL. Cohort Measured data, 2006 Measured data, 2007 DW g fruit 1 DW g fruit 1 1 10.8 a 13.6 a 2 7.9 b 9.1 b 3 6.3 c 3.6 c F W g fruit 1 FW g fruit 1 1 224 a 279 a 2 173 b 172 b 3 154 c 72 c DMC (%) DMC (%) 1 4.8 a 5.0 a 2 4.7 a 4.9 a 3 4.9 a 4.8 a Fruit Diameter (cm) Fruit Diameter (cm) 1 9.0 a 10.4 a 2 7.2 b 7.5 b 3 3.9 c 3.3 c letters indicate significant differences among cohorts (p<= 0.05). Each value is a mean of four fruits.

PAGE 98

98 Figure 4 1. Individual fruit growth curves (g dry matter per fruit) over time in 2007, fitted to a three parameter Gompertz function ( Cohort 1, Cohort 2, Cohort 3). Each point is a mean of four fruits Figure 4 2. Individual fruit growth curves (g fresh mass per fruit) over time in 2007, fitted to a three parameter Gompertz function ( Cohort 1, Cohort 2, Cohort 3). Each point is a mean of four fruits

PAGE 99

99 Figure 4 3. Individual fruit growth curves (fruit diameter in cm) over time in 2007, fitted to a three parameter Gompertz function ( Cohort 1, Cohort 2, Cohort 3). Each point is a mean of four fruits 1 14 28 42 56 Days After Tagging 0 2 4 6 8 11 13Fruit Dry Matter Concentration (%) Figure 4 4. Individual fruit growth curves (dry matter concentration) over time in 2007, fitted to a four parameter Gompertz function with displacement ( Cohort 1, Cohort 2 Cohort 3). Each point is a mean of four fruits

PAGE 100

100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 10 20 30 40 50 60 Days After Tagging Fruit Growth Rate (g dry matter d-1) CH#1 CH#2 CH#3 Figure 4 5. Individual fruit growth rate (g dry ma tter per day) over time, obtained by fitting the 2007 data to a three-parameter Gompertz function and calculating the first derivative of the function

PAGE 101

101 CHAPTER 5 IMPROVING A TOMATO GROWTH MODEL TO PREDI CT FRESH WEIGHT AND SIZE OF INDIVIDUAL FRUIT S I ntroduction A number of tomato crop models that simulate fruit dry weight yield have been developed. Tomato growers, however, would be more interested in models that can predict the weekly pattern of yield on a fresh weight basis, including variation s in t he marketable size. Such a model could assist growers in planning their management and commercialization strategies in order to optimize fruit yield, harvest timing, and fruit quality. These three goals are not easy to attain simultaneously for tomato beca use the traits that enhance yield are often antagonists of those that increase quality. For example, larger fruits tend to have a lower dry matter concentration (Foolad, 2007). In this sense, simulating the growth of individual fruits may represent an effi cient path toward increasing our understanding of the processes involved in the formation of those fruit variables. Modeling of individual fruit behavior over the course of time could be used as a tool to analyze fruit function, as well as to integrate our knowledge about the responses of fruit growth to internal and environmental factors. Researchers study fruit growth in order to answer several important questions in horticulture, such as: what management practice provides the best range of fruit size, sh ape, yield, and marketable quality? Which fruit cultivar is most suited to a particular growing region? What are the most critical stages during fruit development that may be related to the incidence of physiological defects and disorders ? (Opara, 1999). Horticultural crops like tomato are characterized by a low fruit dry matter concentration that varies between 5 to 6.5 % (De Koning, 2000; Heuvelink, 2005). Photosynthesis -based models simulate dry matter production, but few models have attempted to simula te the dry matter concentration (DMC) or water concentration of fruits. As a consequence, a fixed DMC value is frequently assumed to translate dry weight into fresh weight. However, an error in

PAGE 102

102 predicting fresh weight of more than 25 % may arise when a fix ed DMC value is assumed (Marcelis et al ., 1998). Both plant developmental stages as well as the potential gain in carbon are closely linked to temperature; therefore, to mechanistically simulate fruit growth, it would seem feasible to develop a model that relates carbon accumulation with the physiological or thermal age of the frui ts, from anthesis to maturity. Since the amount of a ccumulated dry matter per fruit will determine how much of the growth will be devoted to fresh mass, it follows that after the crop model predicts the dry mass per fruit and fruit thermal age, an equation for converting the dry weight of fruits to fresh weight must be derived. This approach may be more efficient than trying to track fruit growth over time based on changes in diame ter simply because fruit diameter is strongly cultivar dependent, and therefore the inherent generic nature of CROPGRO could be lost while trying to parameterize the equations for fruit radi i. In addition, early and late set fruits vary in size, which br ings up the difficulty associated with predicting the whole dynamics of fruit growth beginning with fruit size. On the other hand, once the fresh mass is simulated, the size of the fruits is relatively simple to model using empirical relationships, which h ave been established by Bussieres (1993) and verified by Scholberg et al. (1997). Based on the conceptual relationships described above, a subroutine was added in the DSSATV4.5 version of the CROPGRO Tomato model to predict the dry matter concentration, f resh weight, and radial size of individual fruits over time on a cohort basis. The objective of this chapter is to evaluate the capabilities of this sub -model in predicting the fresh weight and size of individual fruits and to use data from field experimen ts to calibrate some parameters as well as to assess the quality of the simulations Material s and Methods Field data: The growth data for evaluation were obtained from experiments conducted in field -grown plastic mulched fresh-market tomato plots between April and July of 2006 and

PAGE 103

103 repeated during the next season in 2007. The data from the 2007 season were used to calibrate the genetic parameters, as well as the non -linear adjustment of the equations; therefore, the true independent dataset for the evaluat ions were collected in 2006. The experimental area was located at the University of Florida Plant Science Research and Education Unit at Citra, Florida (29 was well irrigated and well fertilized, so no water or nitrogen stress was present throughout the entire growing season. The soil has been classified as a fine Candler sand and Tavares sand (Buster, 1979; Dukes et al. 2005). These soils contain 97% sand-sized particles and have a field water holdi ng capacity of 5.0% to 7.5% by volume in the upper 100 cm of the profile (Carlisle et al ., 1988; Dukes et al., 2005). The cultivar evaluated was Florida 47, a mid to late -season hybrid whose fruits are deep globe -shaped. Each replicate plot consisted of 15.2 m long rows of 4 -wk old plants established on April 04 in 2006 and April 07 in 2007 in completely mulched fumigated beds. Row spacing was 1.83 m and plant spacing was 0.45 m, which translated to a plant population of 11,960 plants ha1. Pre -plant fert ilizer applications were 112 kg P2O5 ha1 and 45 kg ha1 of K2O. Treatments were replicated four times using a randomized complete block design. Water application was based on the calculated crop ET, and N rates were based on IFAS recommendations consistin g of 224 kg N ha1, which was applied with the irrigation water. Irrigation was applied via the drip tape system, and emitters were spaced 20 cm apart. Climatic data, including temperature, solar radiation, rainfall, wind, and humidity were collected by an automatic weather station located within 1 km of the experimental area. Growth a nalysis: Periodically (every 14 days), four plants per treatment were destructively measured. Stems, removed leaves (separated into blades and petioles), and picked

PAGE 104

104 fruits wer e dried in a ventilated oven at 60 C for at least one week, and leaf area was measured with a LICOR Model 3100 Area Meter (Lincoln, Nebraska, USA). Growth analysis of individual fruits: At the time of anthesis, three sets of flowers separated by one week in age were tagged. For each cohort date, at least 60 flowers were tagged in each replicate (modified from Heuvelink, 1995). Beginning 3 days after anthesis and two times per week, two tagged fruits in each plot were randomly sampled at 0800 h in the morni ng. Therefore, samples were collected at 3, 7, 11, 14, 18, 22, 26, 30, 33, 37, 41, 45, 49, and 53 DAA (days after tagging). During sampling, the fruit diameter was measured, the fresh weight was recorded, and the dry weight was destructively de termined. To tal yield: A total of 32 plants were selected for the final yield. At harvest complete plants (8 per replicate) were harvested separately T he number of fruit was recorded for each replicate and classified according to fruit size. Fruit size designation w as in agreement with the United States standards for grades of fresh tomatoes The fresh weight of fruits was recorded and 1 kg of fruit for each size designation was sampled for dry weight determination. The rest of each harvested plant was chopped and dr y weight was determined in order to obtain the fruit harvest index at harvest The model: The CROPGRO model used in this study is not yet the official version of the DSSAT software. It has the characteristics of version V4.5, w hich will be released in 2009 with fresh weight and the genetic files of the updated parameters explained in Chapter 3. The general description was presented in Chapter 3, and more detailed information may be found in the references mentioned therein. In this section, only characteris tics of the model related to fruit growth are described. The CROPGRO model explicitly predicts fruit addition and fruit growth rates over tim e for the cohorts added daily. The number of fruits, seeds, and flowers that are

PAGE 105

105 initiated each day are stored and updated dynamically, allowing us to know not only the number of fruits, seeds, and flowers but also the real time ages of the fruits as well as their physiological ages. This capability of the model is a benefit over other models for which the number of fr uits is often an input. As each individual cohort of fruits is added, individual fruits possess their own inherent dry matter accumulation rate based on the genetic potential growth rate and temperature (Boote and Scholberg, 2006). Additionally, a proporti on of fruits are allowed to abort based on C availability, N availability, and a water stress variable. The cultivar -specific parameter XFRUIT determines how much of the maximum fraction of the daily available C is partitioned to fruits. The XFRUIT paramet er can be equal to 1.0 for totally determinate cultivars and less than 1.0 for semi and indeterminate cultivars. Seeds (the number of seeds per shell) are set inside the shell after a certain period of time (FLSD FLSH) after shell growth begins. A growth duration (LNGSH) also exists for each shell, as well as for the seeds (SFDUR). The CROPGRO model predicts the physiological or thermal age of each fruit cohort, and fruit shells are divided into age classes. When shells are less than LNG PEG thermal days in age, they are assumed to grow slowly in mass. How slowly they grow depends on a species parameter called SHLAG, which represents a fraction of the rapid shell growth rate. Cohorts are checked by the model to determine if their physiological age is younger than LNGPEG, between LNGPEG and LNGSH, or older than LNGSH thermal days. After fruits are LNGPEG thermal days in age, they grow at a constant rapid rate. This rapid growth continues until fruits are LNGSH thermal days old, after which growth is no longer allowed (Boote et al 2000). In Chapter 3 the species parameter SHLAG was calibrated (approximately 0.1) to allow a slow growth rate during the first two weeks, in agreement with the normal behavior of tomato fruits, which, during that early phase, displa y intense cell division rather than a high growth in mass (Mapelli et al 1978; Varga and

PAGE 106

106 Bruinsma, 1986). Dorais et al (2001) reported that during the first weeks after anthesis, tomato fruits grow at a rate that is 10 % of the rapid growth rate, and thi s slow phase continues for 14 to 21 days depending on the cultivar, the position of the fruit (age), and environmental conditions. The ecotype PM06 c oefficient was also calibrated as described in Chapter 3. As explained in that chapter in CROPGRO models, t his parameter typically has meaning only for peanut. It represents a fraction of the time between first peg and first seed in the pod. If the PMO6 in the ecotype file is equal to 0.0, the first peg and the first pod occur at the same time. If PMO6 is > 0.0 a slow growth phase occurs prior to the reported first pod. In addition, in order to improve the timing of flower and fruit appearance, several other cultivar p arameters were also calibrated. Modeling individual fruit growth: The submodel included in CRO PGRO for fresh mass and size simulations begins with the C balance model predicting the dry matter gain per fruit (plus seeds) over calendar time, along with the prediction of the physiological age of the fruits in thermal time. The function that simulates DW is linear (constant potential rate if assimilate is available but reduced if assimilate is deficient) The fruit dry matter concentration (DMC) is then predicted as a function of fruit thermal age (TT in Equation 5 1). Paramete max in Equation 5 1 were adjusted to the independent experimental data of the 2007 season using nonlinear regression and fitted values will be presented later. These parameters are in some way mechanistic (Scholberg, 1996). The first para maximum fruit dry matter concentration (DMCmax), which occurs at the time of anthesis. For cultivated tomatoes DMCmax is approximately 12 to 14 %, and the minimum concentration at maturity (DMCmin) is considere environmental conditions, especially solar radiation. Under field conditions, considered as fairly stable during the season; however, in the greenhouse, likely needs to be calibrated

PAGE 107

107 accord equation 1 to account for the fact that, for fruit growth simulation, the model starts the accumulation of thermal time on a fruit start basis (FL SH days after flowering ), but in the experiments, timing started when flowers were tagged. In the model, fruit growth ends when PAGE is > LNGSH (physiological age is greater than length of shell growth). Equation 5 1 accounts for this ending stabilizing the DMC toward the end of growth when fruits reach a reference fruit thermal age (TTmax in Equation 5 1). A reference sum of degree days (Cd) could be used as well. Scholberg (1996), for instance, found that this value corresponded to a sum of 750 oCd. In addition, Scholberg and Boote (2006) proposed a maximum thermal age equal to 40 physiological days using 10 C as the basal temperature. Adams et al (2001) reported that 812 oCd were accumulated by tomato fruits in order to reach maturity, while De Koning (2000) reported 940 oCd. The discrepancies are likely related to the base temperature used by each author. Scholberg (1996) used a basal temperature equal to 10 C (fruit progression from anthesis to maturity), De Koning used 4 C, and the version of the CROPGRO tomato model us ed in this paper employed a literature based value eq ual to 5.7 C (see Chapter 3). The physiological age values used to optimize the parameters in E quation 5 1 were calculated by CROPGRO using a basal temperature equal to 5.7 C. ) / ) ( ( *max minTT TT EXP DMC DMC (5 1 ) DMC as a function of the thermal age plus the model predicted dry mass per fruit (DW, g per fruit) permits computation of fresh mass per fruit (FW, g per fruit) according to Equation 5 2. DMC DW FW (5 2 )

PAGE 108

108 The relationship between fruit diameter versus fresh mass, as described by Equation 5 3, is used to predict fruit diameter. To ob 5 3, experimental data from the 2007 season were fitted using an exponential function. The fruit shape factor (FS) was 0.95 for round fruits, 0.90 for flat fruits, and 1.05 for oblong fruits (Bussieres, 1993; Scholberg e t al ., 1997). 3 / 1* 4 / FS FW Diam (5 3 ) Genetic parameters calibration: Genetic parameters of the determinate tomato cultivar Florida 47 were calibrated with the data from 2 007 as explained in Chapter 3. Model validation : Model validation was performed using independent experimental data collected dur ing the spring season in 2006. The root mean square error (RMSE) and the Willmott index (d) were used as evaluation criteria. T he RMSE is a measure of the difference between the simulated and observed data (Rinaldi et al 2007). Simulation results are considered excellent when RMSE is lower than 10% of the mean, good if between 10 and 20 %, fair if between 20 and 30%, and poor if values are greater than 30% of the mean (Jamieson et al 1991; Rinaldi et al ., 2007). The Willmott or d index is useful to evaluate model predictions of a variable over time. To calculate the d value, the numerator is the MSE and the denominator is relat ed to the variability in the measured and simulated values (Wallach, 2006). The d value varies between 0 and 1, and a value closer to one indicates better performance of the model (Willmott, 1981). Results and Discussion C alibration R esults (Year 2007) Gen etic parameters: Several genetic parameters were previously calibrated in order to adjust the development, total biomass, and fruit yield for cultivar FL 47 to the experimental data measured for 2007, as presented in Chapter 3. As explained in that chapter some additional

PAGE 109

109 parameters were calibrated to adjust the growth of individual fruits on a cohort basis to the experimental data measured during the 2007 growth season. First, the timing of flower and fruit appearance w as adjusted. The cultivar parameter E M -FL (days between plant emergence and flowering) was calibrated to 25 photothermal days in order to improve the prediction for anthesis date. In addition, the model starts to grow fruit after a certain amount of thermal time is accumulated after beginning peg, but in the experiments, flowers were tagged and time began at flowering. As a consequence, to match the timing between flowers and the start of slow fruit growth, the cultivar coefficient FL -SH (thermal days between flowering and shell growth) was c alibrated to a smaller value (3 photothermal days) In the same way, the time between first seed and fruit maturity was also calibrated. In order to improve the initial growth of fruits and seeds, the coefficients PODUR (duration of pod addition) and SFDU R (duration of seed filling) required calibration. In addition, in order to give some priority to shell growth over seed growth, the threshing percentage was slightly decreased (f rom 9.2 to 8.5 photothermal days) To track the growth of individual cohorts, the species parameter SHLAG was calibrated to a value equal to 0.1 to allow a slow growth rate during the first 2 weeks, in agreement with the observed growth of tomato fruits. Moreover, to set the duration of this slow fruit growth phase, coefficient PM0 6 (ecotype file) was calibrated to a value of 0.6 for the three cohorts. The criteria for thes e calibration processes were primarily that the model had to correctly predict the timing of flowering in order to simulate the development and growth over time of three cohorts whose development and growth began at different dates (separated by one week) based on the tagging dates. At the same time, growth and yield at the whole plant level was calibrated as well. For the present chapter all the genetic coefficien ts (species, cultivar and ecotype parameters) correspond

PAGE 110

110 to those of Chapter 3 against having been solved plant growth, fruit yield and growth of tagged cohorts from 2007 experiment al data and no further calibration was done for this chapter. Fruit dry we ight calibration results : The observed and simulated dry weight accumulation in individual fruits from flowers that had been tagged at three st ages is presented in Figures 5 1, 5 2, and 5 3 and Table 5 1 for 2007. There was a very good correspondence betwe en the simulated and observed data with regard to cohorts 1 and 2. In these cohorts, the Willmott agreement index (d) was equal to 0.997 and 0.988 for cohorts 1 and 2, respectively. The high d value indicates that the variability in the observations around the mean was very well tracked by the predictions. The average RMSE for both cohorts was low, on average, approximately 0.78 g, which represents an error lower than 6% of the final fruit dry weight for these cohorts. The less satisfactory, although still acceptable, response of the model was observed for cohort 3, which produced a d of 0.96 and RMSE of 0.51g, representing an error of 14% for the final DW of this cohort Optimization of Equation 5 -1: The CROPGRO model does not use a Gompertz function as we used in Chapter 4 to simulate t he growth of individual fruits. Instead, as stated in the Materials and Methods section, CROPGRO predicts a constant genetic potential growth rate (linear phase) after a lag phase of 10% of the genetic potential (Dorais et a l 2001). A set of equations (Eq.51 to Eq.5 3) is then used to convert the dry matter per fruit to fresh weight and size. Experimental data for the fruit dry matter concentration obtained in 2007 were fit to Equat ion 5 1 using nonlinear regression in order to optimize the para me ters of the equation (Figure 5 4 ). Fitted parameters for the th re e cohorts are shown in Table 5 2 The model proposed to estimate DMC exhibited a low MAE, indicating good adjustment. All estimated parameters were significant (p <0. 0001), indicating that four parameters are needed to explain the model.

PAGE 111

111 The SE of the four estimated parameters was low, which denotes adequate estimation of the see parameter was added to Eq. 5 1 to account for the fact that fruit growth simulation in the model initiates the accumulation of thermal time (X axis in Eq. 5 1) on a fruit start basis, but in the experiments, timing began when the flowers were tagged CROPGRO uses only one equation to predict the DMC of the three cohorts. Therefore, for simulation, the optimized parameters for co hort 1 were used in Equation 51 for all coh orts. For cohort 1, t between the maximum fruit dry matter concentration and the minimum concentration at 7.1. In the model, the fruits terminate t heir growth when they reach a reference f ruit thermal age (TTmax in Eq. 5 1). This maximal thermal age was solved to be 55 thermal days in the adjustment for cohort 1. For this cohort, the intercept Fruit dry matter concentration calib ration results : The observed and simulated dry matter concentration of individual fruits from flowers that had been tagged at three stage s are presented in Figures 5 5, 56, and 5 7 and Table 5 1 for 2007. Simulated and measured data agreed reasonably well with RMSE s of 0.3 % and 0.9 % for cohorts 1 and 2, respectively, which represent a model error of 6 % and 18 % for the fruit DMC of cohorts 1and 2 and a higher MSE (1.2 %) for cohort 3 representing an error of 24 % for the fruit DMC. The Willmott index was high (0.996 and 0.97 for cohorts 1 and 2, respectively), but it was lower (0.94) for cohort 3. Fruit fresh weight calibration results : The observed and simulated fresh weight accumulation in individual fruits from flowers that had been tagged at three s tages are presented

PAGE 112

112 in Figures 5 8, 59, and 5 10 and Table 5 1 for 2007. A very good correspondence was observed between the simulated and observed data in cohorts 1 and 2. In these cohorts, the Willmott agreement index (d) was equal to 0.99 for both co horts. The average RMSE for both cohorts was low (15 g), which represents an error of only 5.4 % of the single final fresh fruit weight. Similar to the dry weight prediction, acceptable though less satisfactory results were detected for cohort 3, which d emonstrated a d value of 0.95 and a higher RMSE (12.9 g), representing an error of 16.7 % for the final fresh weight of fruits in this cohort Optimization of Equation 5 -3: The relationship between fruit diameter versus fresh mass as described by Eq. 5 3 is used to predict fruit diameter. Parameters Eq. 5 3 by fitting the experimental data for fruit diameter measured during 2007. The results are shown i n Table 5 2 The MAE of the model was low, indicating a good model. The optim ized estimations was low, indicating that the parameters were adequately estimated. Moreover, both parameters are important for explaining the model (p < 0.0001). It appears that the cohort effects w ere not important for Equation 5 3 because the parameters were similar across cohorts. The fruit shape factor (FS) is 0.95 for round fruits and the Florida 47 cultivar is a round-shaped fruit. Therefore, the expression in t he denominator of Equation 5 3 is a constant equal to 3.8. Fruit size calibration results : The observed and simulated fruit diameter of individual fruits from flowers that had been tagged at three s tages are presented in Figures 5 11, 5 12, and 5 13 and Ta ble 5 1 for 2007. Comparisons among cohorts showed that the fruit sizes of cohorts 1 and 2 were very well simulated by the model, with an average RMSE of 0.6 cm for both cohorts, which represents an error of only 6.9 % for fruit diameter in both cohorts. T he average d value was 0.99, indicating that the variability in the experiment was well captured by the predictions.

PAGE 113

113 Fruit size simulation for cohort 3 was less satisfactory, with an RMSE of 0.74 cm, representing an error close to 22% for the fruit diamete r in this cohort, and a d index (0.95) that was acceptable although much lower than that observed for earlier cohorts V alidation Results (Year 2006 ) Validation of the model was performed using independent experimental data collected during the spring sea son of 2006. Fruit dry weight : The observed and simulated dry weight accumulation in individual fruits from flowers that had been tagged at three stages during 2006 is presented in Figures 5 1 4, 5 15, and 5 16 and Table 5 3 There was a very good corres pondence between the simulated and observed data for cohorts 1 and 2. In these cohorts, the Willmott agreement index (d) was equal to 0.99 and 0.97 for cohorts 1 and 2, respectively. The average RMSE for both cohorts was also acceptable (1g) and represents a model error of 13.8 % for the total fruit dry weight in these cohorts. A less satisfactory model response was observed for cohort 3, which exhibited an RMSE of 1.12 g, representing a model error equal to 17% of the final DW in this cohort. The d value o btained for cohort 3 was acceptable (0.96). Upon initial inspection, the delayed growth of cohort 3 was likely a result of the priority exerted by older cohorts and was reasonably simulated by the model. However, the simulated delay was too short, since the measured data showed that cohort 3 resumed growth and entered a phase of fast growth later in the growth period. The opposite was observed in 2007 when the model behaved in an opposite fashion, indicating that the delay was too long for cohort 3. However those fruits did not attain a high mass, likely because this period of fast growth was too short. Such an effect mostly occurred when the fast phase of growth was terminated in cohorts 1 and 2. Fruit dry matter concentration: The observed and simulated d ry matter concentrations in individual fruits from flowers that had been tagged at three s tag es are presented in Figures 5 17,

PAGE 114

114 5 18, and 5 19, and Table 5 3 for 2006. The simulated and measured data agreed reasonably well with a RMSE in average for both c oh orts of 1.2 % which represent a model error of 16.4% of the mean fruit DMC for these cohorts. The RMSE for cohort 3 was 1.5%, representing a model error of 19 % for the mean DMC in cohor t 3. The Willmott index was 0.96 and 0.94 for cohorts 1 and 2, respe ctively, although it was much lower (0.93) for cohort 3. This last cohort maintained small fruits due to a longer lag phase in growth, as previously noted, which resulted in a higher concentration of dry matter compared with fruits of the same age (calendar days) in the other cohorts. This may explain the higher error observed for this cohort compa red with the other two cohorts. Fruit f resh weight: The observed and simulated fresh weight accumulations in individual fruits from flowers that had been tagged a t three stag es are presented in Figures 5 20, 5 21, and 5 22 and Table 5 3 for 2006. The simulated and measured data agreed reasonably well with a RMSE in average for both cohorts of 18g, which represents a model error of 9.25% of the fruit fresh weight for these cohorts. The RMSE for cohort 3 (18.3g) represents a model error of 12 % for the single final fresh fruit weight in cohort 3. The Willmott index was 0.99 and 0.98 for cohorts 1 and 2, respectively, but it was lower although still acceptable (0.96) f or cohort 3. This last cohort maintained small fruits due to a longer lag phase in growth, as previously noted, resulting in a higher concentration of dry matter as compared to fruits of the same age (calendar days) in the other cohorts. This may explain t he higher error observed for this cohort compa red with the other two cohorts. Fruit size: The observed and simulated fruit diameter of individual fruits from flowers that had been tagged at three s tag es are presented in Fi gures 5 23, 5 24, an d 525 and Tab le 5 3 for 2006. Comparisons among cohorts showed that the fruit size in cohorts 1 and 2 was well

PAGE 115

115 simulated by the model, with an average RMSE of 0.57 cm, representing a model error of 7 % and an average d value equal to 0.985. The fruit size simulation fo r cohort 3 was less satisfactory, with an RMSE equal to 0.75 cm, which represents a model error of 19 % with regard to fruit diameter. The d value was 0.97. This is not surprising since fruit mass in cohort 3 was also poorly simulated, and because of the d ependency of the size equation on FW, the error was propagated as expected Ability of the Model to Predict Total Yield and Size Distribution. Total yield for 2006 and 2007: Overall, the calibration done in Chapter 3 of some additional parameters (SHLAG a nd PMO6) as well as the reduction of time between first flower and first fruit from 8 to 3 required to predict the growth of individual fruits, did not affect the simu lation of total fruit dry weight over time For years 2006 and 2007, total fruit dry weig ht over time was in good agreement with measurements for both seasons (Figure 5 26). For years 2006 and 2007, simulated total fresh weight exceeded the measured growth. This is especially seen toward the end of the season (Figure 527). Size Distribution: Figure 5 28 shows the simulated size distribution for year 2007. According to the simulated values, most of the production (50 Mg ha1) ended in the large category (6.3 to 7.3 cm) which is a good size distribution in the fresh market. The simulated values also explain the overestimation in FW (Figure 5 27) due to the high proportion of fruits in cull class size (smaller than 4.8 cm). Those fruits count for the FW although they do not contribute to the commercial yield of tomato for fresh market. According t o the simulation the medium class (5.8 to 6.3 cm) which has commercial value also contributed to total yield. At harvest the model predicted that almost all the marketable yield (47 % of the total) corresponded to the large class (6.3 to 7.3 cm). In addit ion, 28 % of yield was non marketable cull class. Therefore, simulated number of fruit is not an indicator of high productivity since the model

PAGE 116

116 predicts that the plant is carrying a large number of fruits which contribute to the FW. However, those fruits do not fit the market requirement for fruit size. According to the model, tomato fruits did not reached extra large or maximum large categories likely due an excessive load considering the high contribution of culls class. Figure 5 29 shows simulated and observed size distribution at harvest. The analysis of production by fruit size category indicated that during 2007 40.5%, 19.9% and 20.3% of the observed fresh production corresponded to categories large, extra large and maximum large, respectively. In addi tion, 11.4%, 5%, and 1.89% corresponded to medium, small and extra small class, respectively. The observed percentage of culls was only 0.6%. The model on the other hand, predicted no fruits in extra large and maximum large categories while most of the yie ld at harvest was in large and medium classes. For fresh market, the commercial production can be represented by the sum of the categories medium, large, extra large and maximum large. Therefore, during 2007, 92 % of the observed total yield reached commer cial size. With the model, the large (46.5%) and medium (18.5%) classes contributed only 65.4% of the total yield in commercial size. However, simulated and observed data agree on the fact that the higher contribution to total marketable yield is from the large class. The simulated and observed percentage also agrees reasonably well through the season (Figure 529). Summary and Conclusion A simple set of equations was included in CROPGRO to simulate the dry matter concentration, fresh mass, and size of indi vidual fruits over time. Parameters were calibrated from 2007, and the simulated results agreed reasonably well with the measured independent data obtained for 2006. The dynamics of dry matter accumulation, fresh mass changes, and size increases simulated by the model demonstrate its ability to explain and predict fruit growth. Overall, predictions for earlier fruits were more successful than the latest cohort, although even the last cohort generally demonstrated d values above 0.9 and prediction errors low er than 20%.

PAGE 117

117 In addition, the contribution of the latest setting fruits to total fruit yield was almost negligible, and, therefore the submodel can be used in studies of tomato fruit potential growth as well as for accurate predictions of potential yield and fruit size for tomato without significant errors. In addition, the calibration of the model to simulate the growth of individual fruits did not affect the capabilities of the model to simulate growth at a whole plant level and total fruit yield. Impr ovements should be made regarding simulation of the timing at which earlier cohorts truly exert priority in attracting assimilates over the latest fruit set. The model does give priority to older fruits and delays the last cohort, but the delay is not enti rely satisfactory. To improve the simulation of priority rules among cohorts may be especially necessary when plants face severe stress and the priority of older fruits may be accentuated, which affects not only the fruits added last but also the intermed iate cohorts. In addition, more indeterminate cultivars for which vegetative growth continues during reproductive growth may increase the sink competition for assimilates in plants in response to stress, reducing the accuracy of the predictions. In general CROPGRO deals with stress by allowing later -set fruits to grow more slowly to a point at which abortion of fruits occurs if the assimilate availability is dramatically reduced. This approach needs to be tested for confirmation and is examined fur ther in the following chapters. The size distribution simulated by the model had reasonable agreement with observations for the large marketable class size that contributed most to the commercial yield of tomato in our experiment. There was also good agreement for the medium category. However, the model failed in simulating fruits in the extra large and maximum large sizes Therefore, further research needs to be done to optimize the simulation of size distribution. Determinacy and ability of the model to add late set fruits need s to be tested as the model predict ed more late -set culls than were observed. Still the model was able to produce reasonable outputs showing size distribution over

PAGE 118

118 time a fact that makes CROPGRO tomato model a tool with a high potential for pract ical applications.

PAGE 119

119 Table 5 1 Observed vs. CROPGRO simulated final dry weight, fresh weight, dry matter concentration and fruit size of individual tomato fruits, tagged at three different dates. Observed data came from calibrat ion experiment conducted at Gainesville Fl during spring in 2007. Cohort # Observed Simulated RMSE Willmott index (d ) DW g fruit 1 DW g fruit 1 1 13.6 12.98 0.58 0.99 2 3 9.1 3.6 10.00 3.55 0.98 0.5 1 0.98 0.96 Fruit DMC (%) Fruit DMC (%) 1 5.00 5.0 0.3 0.99 2 4.98 5.0 0.9 0.97 3 4.85 5.1 1.2 0.94 FW g fruit 1 FW g fruit 1 1 279 259 13 0.99 2 3 172 72 198 70 17 12 0.99 0.86 Fruit diameter (cm) Fruit diameter (cm) 1 10.4 10.50 0.20 0.99 2 7.5 7.96 0.77 0.93 3 3.3 2.80 0.74 0.95 Table 5 2 E stimated coefficients solved by fitting data of the experiment 2007 to equation 1 to predict fruit DMC over time for the three cohorts and to equation 3 to predict fruit diameter, (In fostat software) Parameter Model DMC (%) = max ) MAE = 0.03 (average for the three cohorts) Cohort # 1 S.E Cohort # 2 S.E Cohort # 3 S.E 10.06 0.60 11.35 0.19 12.71 0.21 7.10 0.38 6.80 0.33 6.03 0.76 1.60 0.21 2.20 1.19 5.83 4.10 TTmax 55.00 0.79 55.0 0 0.98 55.00 1.61 Parameter Model Fruit Diameter (cm) = MAE=0.11 (average for the three cohorts) Cohort # 1 S.E Cohort # 2 S.E Cohort # 3 S.E 20.05 0.19 20.07 0.33 20.81 0.160 0.020 0.0012 0.021 0.006 0.023 0.092

PAGE 120

120 Table 5 3 Observed vs. CROPGRO simulated dry weight, fresh weight, dry matter concentration and fruit size of individual tomato fruits, tagged at three different dat es. Observed data came from validation experiment conducted at Gaines ville Fl during s pring in 2006. Cohort # Observed Simulated RMSE Willmott index (d ) DW g fruit 1 DW g fruit 1 1 10.8 10.58 0.75 0.99 2 7.9 10.11 1.4 0 0.98 3 6.3 8.17 1.12 0.97 Fruit DMC (%) Fruit DMC (%) 1 4. 8 5.0 1.20 0.93 2 4.7 5.0 1.30 0.94 3 4.9 5.1 1.50 0.92 FW g fruit 1 FW g fruit 1 1 224 210 19.0 0.99 2 173 201 17.6 0.99 3 154 161 18.0 0.98 Fruit diameter (cm) Fruit diameter (cm) 1 9.0 8.5 0.77 0.98 2 7.2 8.1 0.52 0.99 3 3.9 5.6 0.78 0.97

PAGE 121

121 RMSE=0.58 d=0.997 0 2 4 6 8 10 12 14 16 1 6 11 16 21 26 31 36 41 46 51 56 61 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated) Fruit DW,g/f (Observed) Figure 5 1 Predicted vs. observed dry weight of individual fruits of cohort # 1 in Gai nesville Fl, during spring 2007 (e ach point is a mean of four fruits). RMSE=0.98 d=0.98 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated) Fruit DW,g/f (Observed) Figure 5 2 Predicted vs. observed dry weight of individual fruits of cohort # 2 in Gai nesville Fl, during spring 2007 (e ach point is a mean of four fruits).

PAGE 122

122 RMSE=0.52 d=0.96 0 1 2 3 4 0 5 10 15 20 25 30 35 40 45 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated) Fruit DW,g/f (Observed) Figure 5 3 Predicted vs observed dry weight of i ndividual fruits of cohort # 3 in Gainesville Fl during sprin g 2007 (e ach point is a mean of four fruits). Figure 5 4. Fruit dry matter concentration over time fitted to a four parameter CROPGRO function in 2007 ( Cohort #1, Cohort #2, Cohort #3), each point is a mean of four fruits

PAGE 123

123 RMSE=0.3 d=0.99 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated ) Percent DM (Observed) Figure 5 5 Predicted vs observed dry matter concentration of individual fruits, cohort # 1 (e ach point is a mean of four fruits) in Gainesville Fl, during spring 2007. RMSE=0.9 d=0.97 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated) Percent DM (Observed) Figure 5 6 Predicted vs observed dry matter concentration of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 (e ach point is a mean of four fruits).

PAGE 124

124 RMSE=1.2 d=0.94 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated) Percent DM (Observed) Figure 5 7 Predicted vs. observed dry matter concentration of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 (e ach point is a mean of four fruits). RMSE=13.8 d=0.996 0 50 100 150 200 250 300 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated) Fruit FW,g/f (Observed) Figure 5 8 Predicted vs. observed fresh weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 (e ach point is a mean of four fruits).

PAGE 125

125 RMSE=17 d=0.99 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated) fruit FW,g/f (Observed) Figure 5 9 Predicted vs observed fresh weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 (e ach point is a mean of four fruits). RMSE=12.9 d=0.945 0 10 20 30 40 50 60 70 80 90 0 5 10 15 20 25 30 35 40 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated) Fruit FW,g/f (Observed) Figure 5 10. Predicted vs. observed fresh weight of individual fruits of cohort # 3 in Gainesville Fl, d uring spring 2007 (e ach point is a mean of four fruits).

PAGE 126

126 RMSE=0.2 d=0.999 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 55 Days After Tagging Fruit Diameter (cm) Simulated Measured Figure 5 11. Predicted vs. observed fruit diameter of individual fruits cohort # 1 in Gainesville Fl, during spring 2007 (e ach point is a mean of four fruits). RMSE=0.77 d=0.93 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Diameter (cm) Measured Simulated Figure 5 12. Predicted vs observed fruit diameter of individual fruits cohort # 2 in Gainesville Fl, during spring 2007 (e ach point is a mean of four fruits).

PAGE 127

127 RMSE=0.74 d=0.95 0 1 2 3 4 0 5 10 15 20 25 30 35 40 Days After Tagging Fruit Diameter (cm) Measured Simulated Figure 5 1 3 Predicted vs observed fruit diameter of individual fruits cohort # 3 in Gainesville Fl, during spring 2007 (e ach po int is a mean of four fruits). RMSE=0.75 d=0.99 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated) Fruit DW,g/f (Observed) Figure 5 14. Predicted vs observed dry weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits).

PAGE 128

128 RMSE=1.4 d=0.97 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 55 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated) Fruit DW,g/f (Observed) Figure 5 15. Predicted vs observed dry weight of individual f ruits of cohort # 2 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits). RMSE=1.12 d=0.97 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated) Fruit DW,g/f (Observed) Figure 5 16. Predicted vs observed dry weight of individual fruits of cohort # 3 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits)

PAGE 129

129 RMSE=1.2 d=0.93 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated) Percent DM (Observed) Figure 5 17. Predicted vs observed dry matter concentration of individual fruits of cohort # 1 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits). RMSE=1.3 d=0.94 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 55 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated) Percent DM (Observed) Figure 5 18. Predicted vs. observed dry matter concentration of individu al fruits of cohort # 2 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits.

PAGE 130

130 RMSE=1.5 d=0.92 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated) Percent DM (Observed) Figure 5 19. Predicted vs observed dry matter concentration of individual fruits of cohort # 3 in Gainesville Fl, during spring 2006 ( e ach point is a mea n of four fruits). RMSE=19 d=0.99 0 50 100 150 200 250 0 10 20 30 40 50 60 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated) Fruit FW,g/f (Observed) Figure 5 20. Predicted vs observed fresh weight of individual fruits of cohort #1 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits).

PAGE 131

131 RMSE=17 d=0.99 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 Days After Tagging Fruit Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated) Fruit FW,g/f (Observed) Figure 5 21. Predicted vs observed fresh weight of individual fruits of cohort #2 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits). RMSE=18 d=0.98 0 20 40 60 80 100 120 140 160 180 0 5 10 15 20 25 30 35 40 45 50 Days After Planting Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated) Fruit FW,g/f (Observed) Figure 5 22. Predicted vs observed fresh weight of individual fruits of cohort #3 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits).

PAGE 132

132 RMSE= 0.77 d=0.99 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 55 Days After Tagging Fruit Diameter (cm) Measured Simulated Figur e 5 23. Predicted vs observed fruit diameter of individual fruits cohort # 1 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits). RMSE=0.6 d=0.99 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 55 Days After Tagging Fruit Diameter (cm) Measured Simulated Figure 5 24. Predicted vs observed fruit diameter of individual fruits cohort # 2 in Gainesville F l, during spring 2006 (e ach point is a mean of four fruits).

PAGE 133

133 RMSE=0.78 d=0.97 0 2 4 6 0 5 10 15 20 25 30 35 40 45 Days After Tagging Fruit Diameter (cm) Measured Simulated Figure 5 25. Predic ted vs. observed fruit diameter of individual fruits cohort # 3 in Gainesville Fl, during spring 2006 (e ach point is a mean of four fruits) RMSE=178 d=0.99 RMSE=249 d=0.99 0 1000 2000 3000 4000 5000 6000 0 20 40 60 80 100 Days After Planting Fruit Dry Weight (kg ha-1) Fruit wt kg/ha (Control Simulated 2006) Fruit wt kg/ha (Control Simulated 2007) Fruit wt kg/ha (Control Observed 2006) Fruit wt kg/ha (Control Observed 2007) Figure 5 26. Predicted vs. ob served total fruit dry weight in Gainesville Fl, during spring 2006 and 2007.

PAGE 134

134 RMSE=10618 d=0.96 RMSE=15214 d=0.95 0 20000 40000 60000 80000 100000 120000 0 20 40 60 80 100 Days After Planting Total Fruit Fresh Weight (kg ha-1) FreshFruitWt k/h (Control 2006 Simulated) FreshFruit Wt k/h (Control 2007 Simulated) FreshFruitWt k/h (Control Observed 2006) FreshFruitWt k/h (Control Observed 2007) Figure 5 27. Predicted vs. observed total fruit fresh weight in Gainesville Fl, during spring 2006 and 2007. 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 0 10 20 30 40 50 60 70 80 90 100 Days After Planting Fresh weight by class size (kg ha-1) CULLS(FreshWt) SIZE1 Extra Small (FreshWt) SIZE2 Small(FreshWt) SIZE3 Medium (FreshWt) SIZE4 Large (FreshWt) SIZE5 Extra Large (FreshWt) SIZE6 Maximum Large (FreshWt) Figure 5 28. Prediction over time of fruit fresh weight as contri buted by each marketable size class of tomatoes in Gainesville Fl, during spring 2007.

PAGE 135

135 0 5 10 15 20 25 30 35 40 45 50 CULLS SIZE1 (Extra Small) SIZE2 (Small) SIZE3 (Medium) SIZE4 (Large) SIZE5 (Extra Large) SIZE6 (Maximum Large) Size Class Precentage of Fresh Weight at Harvest (kg ha-1) Simulated % of total FW at harvest Observed % of total FW at harvest Figure 5 29. Predicted vs. observed total tomato fruit fresh weight contributed by each marketable size class at harvest in Gainesville Fl, during spring of 2007.

PAGE 136

136 CHAP TER 6 EFFECTS OF WATER AND NITROGEN STRESS ON F RUIT GROWTH AND TOMA TO YIELD Introduction Water and nitrogen (N) fertilizer are important factors for achieving a high tomato yield (Erdal et al. 2006). Several studies have show n that the growth and yield of tomato es are reduced when water and N are deficient (Sweeney et al ., 1987; Scholberg 1996; Scholberg et al.; 2000). T omato crop s are highly responsive to N fertilizer This is particularly true when it is grown i n sandy low organic matter soils (Taber, 2 005; Erdal et al., 2006). During the growing season, tomato es can uptake 150 to 300 kg ha1 of N (Larouche et al., 1989). During rapid growth N uptake rates may exceed 4.3 kg ha1 d1 (Dumas, 1990; Rinaldi et al., 2007). According to Warner et al. (2004), the N requirements of tomatoes are higher during vegetative rather than reproductive growth T heir results agree with Scholberg (1996) and Scholberg et al (2000), who showed that severe N stress significantly reduced leaf area index (LAI) biomass and fr uit yield compared to well -fertilized plants. T he concentration of N in vegetative organs such as leaves and stems was reduced under N stress T he fruits maintained a relatively stable N concentration. Under N stress, the radiation use efficiency ( RUE ) w as reduced by about 30% Scholberg et al. (2000) related this reduction to a lower specific leaf N This in turn affected the photosynthesis rate. The leaf photosynthesis rate, like RUE, was only reduced by 30% compared to control plants This occurred bec ause the leaves were thicker (high SLW) under N stress However, the leaf N concentration was dramatically reduced (40 to 15 mg g1). H igh production, sufficiently fertilized plants should have 3 % to 4% N in their leaves (Lorenz and Tyler, 1983; Mills and Jones, 1987; Hochmuth, 1994). Optimum N must be available early D eficiency symptoms do not appear until the leaf N concentration drops below 2%, a level at which the yield is reduced (Taber, 2005). Huett (1986) and Huett and Dettmann (1988) reported an

PAGE 137

137 in creased yield of semi -determinate tomatoes in response to increasing N This finding was associated with higher ratios of dry fruit matter to total plant dry matter. In tomato crops where the fruits are the dominant sinks, the N level influences the dry m atter in the fruits. Hebbar et al. (2004) reported a 33% higher fruit yield in N fertigated tomatoes compared to plants with low N treatment. They attributed this yield increment to higher LAI, number of fruit and dry matter production. A water deficit oc curs in plants when the rate of transpiration exceeds absorption Deficiency is accompanied by inhibition of growth and development (Wolf and Rudich, 1988). Cleary et al (1996) considered tomato es to be in the group of plants having a low tolerance to wat er stress. They suggested that a water potential in the plant leaves in the range of 0.1 to 0.5 MPa indicates a good supply of water to the plants. More than 90 % of a ripe tomato fruit is water and only 5% to 8 % of the weight is dry matter (Davies and Ho bson, 1981). Consequently, factors affecting water accumulation may determine the size and quality of the tomato fruit (Grange et al 1987). Ho et al. (1987) estimated that the water supply to tomato fruits is predominantly supplied via the phloem. On ave rage, 80% to 85% of the water supply to tomato fruit s occurs via this pathway (Ho et al., 1987; Guichard et al., 2005). The studies on water stress have focused on fruit size, which is largely a measure of the fresh weight (Berman and Dejong, 1996). Resear ch on tomato es suggests that water stress limits fleshy fruit water accumulation but does not affect carbon partitioning to the fruits (Ehret and Ho, 1986). Scholberg (1996) evaluated the effects of water supply on tomato fruit grown under field conditions His experimental data showed that fresh fruit weight was more strongly affected by water stress than dry fruit weight T he decrease in yield under stress was attributed to smaller and fewer fruits compared with the control s In

PAGE 138

138 addition, Garcia et al. (2 004) found that one important response of tomato plants to early season water stress was a significant reduction in the number of floral buds T here was also a delay in the appearance of floral buds. Ehret and Ho (1986), Mitchell et al. (1991), and van Iep eren et al. (2005) showed that water stress during the growth of tomatoes reduced the fruit size by 30%. In addition, many investigations have reported that irrigating at a rate of 120 or 140% of evapotranspirative demand increased the fresh weight yield i n tomato crops (Ortega Farias, 2003). Wolf and Rudich (1988) studied the effects of water stress on dry weight accumulation and the growth of individual fruits. The rate of dry matter accumulation was higher in early fruits than late fruits Dry matter wa s unaffected by the water regime. Water stress, however, shortened the duration of fruit growth and accelerated the ripening. Under water stress ( 80 kPa of water tension in soil compared with 20 kPa in the control treatment), the fruit from the first week of flowering contributed more than 40 % to the final dry weight yield. T he effect of N and water stress on the yield and growth of tomato es ha s been well studied (Nyabundi and Hsiao, 1989; Scholberg, 1996, Garcia et al 2004, Hebbar et al., 2004) However, little information is available about the effect on the rate of growth of individual fruits. Therefore, the objective of this study was to investigate the fresh and dry weight accumulation of individual tomato fruits under water and nitrogen limitations. T he results should increase our understanding of the dynamics of tomato fruit growth under stress. Such information may be used to improve the CROPGRO tomato model to simulate the growth of fruit when water and N bec o me limiting factors Material s and Me thods Four week old tomato seedlings ( cv. Florida 47) were transplanted to the field on April 10, 2007. The experimental area was located at the University of Florida Plant Science Research and

PAGE 139

139 Education Unit in Citra, Florida (29 oil is classified as a fine Candler sand and Tavares sand (Buster, 1979; Dukes et al. 2005). These soils contain 97% sand-sized particles The soil has a field water -holding capacity of 5.0% to 7.5% by volume in the upper 100 cm of the profile (Carlisle et al ., 1988; Dukes et al., 2005). The p lants were grown in open field s on plastic mulched and fumigated beds with 0.45 m spacing in a row and 1.83 m between the rows. Cultural and sanitation management were done as needed according to commercial practices. Pre plant fertilizer applications were 112 kg P2O5 ha1 and 45 kg ha1 of K2O. Irrigation was applied with a drip tape system and emitters were spaced 20 cm apart. Climatic data, including temperature, solar radiation, wind speed and relative humidity was collected by an automatic weather station located within 1 km of the experimental area Treatments: The experiment was set up as a completely randomized block design. Three blocks of four rows (rep licates) were selected. Four rows (replicates) received th e control treatment (CT) which consisted of plants well irrigated and fertilized during the whole cycle. Water was supplied according to calculated daily crop evapotranspiration (Et). Nitrogen was applied following IFAS recommendations (224 kg N ha1) alo ng with the irrigation water. Four other rows (replicates) received the water stress treatment (WS). The water stress was imposed by completely withholding irrigation for several days during the experiment. The duration of the water withholding was variabl e and recorded. The water stressed plants were re -watered when severe leaf wilting was observed and the wilting symptoms persisted during the night (Rahman et al ., 1999). Periods of water withholding were repeated four times during this experimental treatm ent. The N fertilization remained the same as in the control. Another four rows (replicates) received the nitrogen stress treatment (NS) In these plants, nitrogen applications were

PAGE 140

140 c ompletely suspended two weeks after transplanting T he plants remained fu lly irrigated like the control treatment Measurements: Soil water content measurements were made using time domain reflectometry (TDRs) probes. Probes were buried at 20 and 40 cm in the beds and 5 cm away from the drip tube. The readings were averaged eve ry 20 minutes for each day and stored in a data logger (C R10X, Campbell Sci. Logan UT). Leaf water potential was measured with a Scholander pressure chamber technique according to Tyree and Hammel (1972). Measurements were taken on two leaves per plant and two plants per replicate at midday (from 12h to 13h). Determinations were made at 3 day intervals during the experiment. After fruit setting was complete, net leaf photosynthesis, leaf conductance and transpiration flux were measured on two leaves per pl ant and two plants per replicate from a canopy that was well -exposed to sun (above 1000 mol m2 s1). Measurements were taken on 06/05, 06/08, 06/11, 06/14, 06/18, 06/26, and 07/05. Measurements were always taken from 11.00 h to 13.00 h using the LI 6200 portable photosynthesis system model (LICOR Inc., USA). Nitrogen determinations: The N concentration was determined on stems, leaf blades, petioles, and fr uits collected every two weeks. Determinations were done using a modified Kjeldhal meth od (Bremer and Mulvaney, 1982). Growt h a nalysis : E very 14 days, four plants per treatment were measured destructively. Stems, removed leaves (separated into blades and petioles) and picked fruits were dried in a ventilated oven at 60 C for at least one week. The leaf a rea was measured with a Leaf Area Meter model LI 3100 (LICOR Lincoln, NE, USA). Fr uit growth a nalysis : Starting at anthesis three sets of flowers separated by one week in age were tagged. For each cohort, at least 60 flowers were tagged in each replica te (modified

PAGE 141

141 from Heuvelink, 1995). Starting three days after anthesis and two times per week, two tagged fruits in each plot were randomly sampled at 8 00 h in the morning. Therefore, samples were collected at 3, 7, 11, 14, 18, 22, 26, 30, 33, 37, 41, 45, 49 and 53 DAA (days after anthesis). During sampling, the fruit diameter was measured, the fresh weight was recorded, and the dry weigh t was destructively determined. Total y ield : A total of 32 plants per treatment were selected for the final yield. At ha rvest complete plants (8 per replicate) were harvested separately T he number of fruit was recorded for each replicate and classified according to fruit size Fruit size designation was in agreement with the United States standards for grades of fresh tom atoes (Table 6 1) The fresh weight of fruits was recorded and 1 kg of fruit for each size designation was sampled for dry weight determination. The rest of each harvested plant was chopped and dry weight was determined in order to obtain the fruit harvest index at harvest. Data analysis: Treatment comparisons were done using INFOSTAT statistical software and a Duncan means separation test at a significance of p < 0.05. Results Water Stress Leaf water potential: Under water stress (WS) treatment, midday leaf water potential L) decreased to between 1.2 and 1.6 MPa In well -watered plants undergoing control treatment (CT) it re mained around 0.8 MPa (Figure 61). After re -L in plants under WS increased A pproximately three days after re -wa tering it reached similar although slightly lower values compared with plants under CT. In addition, the relative water content of leaves (RWC) from water stressed plants was 86% This value was lower than the leaves from well -irrigated plants The well irrigated plants had an average RWC of 93%. The soil water con tent in WS conditions (Figures 62 and 6 3) was lower than the content of the soil under CT.

PAGE 142

142 On average, CT soil maintained water content values between 0.10 and 0.14 cm3/cm3. T he water content of the soil under WS reached values as low as 0.02 cm3/cm3. E ven after re -watering the content increased only to 0.08 cm3/cm3. Gas exchange: The photosynthetic rate was 13% less in water stressed plants than well watered plants. Stomata resistance was 32% higher in water stressed plants than CT plants (Ta ble 6 2 and Figures 6 4 and 6 5). Under water stress, the stomata l resistance was the most noticeable response of the leaves. The photosynthesis rate was less affected. Total crop yield: In Table 6 3 an d Figure s 6 6 6 7, 6 8 and 69 the results of the tomato plant growth and total fruit yield are presented for the different treatments. The total fresh fruit yield was significantly greater in CT than in WS. On average, water stressed plants produced 27% less (57.7 Mg ha1) than well watered plants (79.2 Mg ha1). The total dry fruit weight was also significantly different between WS and CT. However, the differences were less than th ose found in the fresh weight. The total above ground biomass and fruit dry weight was reduced by 22% and 24%, respectively. The maximum LAI was 15.5% lower in WS plants than in CT plants. The fruit number (measured as the fruit that reached commercial size) was reduced by about 20% in water stressed plants compared to CT plant s. The harvest index was not significantly different between CT and WS treatments. This finding suggests that water stressed plants maintain relatively stable partitioning of dry matter to the fruits in spite of the stress The contribution of each class s ize to the total yield was different according treatments. Thus during 2007 81 % of total fruit yield in CT corresponded to large, extra large and maximum large fruit classes comparing with 63% of contribution from these classes in WS treatment (Table 6 5 ).

PAGE 143

143 Nitrogen Stress Gas exchange: Photosynthetic rates were significantly reduced by N stress (Table 6 2 ; Figures 6 4 and 6 5). The photosynthesis rate was reduced by 19.5% in N stressed plants compared with CT plants. The N stress increased the stomatal r esistance by 23%. The photosynthesis reduction under N stress was less than reported by Scholberg et al. (2000). This report found a reduction of 30% in leaf photosynthesis in plants under severe N stress. A probable explanation is that even when the leaf N concentration in NS was significantly different from the CT, the N concentration was still above the critical level of 2%. Therefore, the leaf photosynthesis rate was expected (Taber, 2005). In addition, the residual fertility of the soil was not measur ed at the start of the experiment. Since the tomatoes were cultivated in the same field during previous year, the leaf N concentration might have been slightly higher than expected. Total crop weight: In Table 6 3 and Figure s 6 6 6 7, 68 and 6 9, the to mato plant growth and total fruit yield are shown for the different treatments. Total fruit fresh yield was significantly less in NS than in CT. On average, N stressed plants produced 41% less fruit (45 Mg ha1) than N fertilized plants in CT (79.2 Mg ha1). Total fruit dry weight was also significantly different between NS and CT. The total biomass and the fruit dry wei ght were reduced by 46% and 59%, respectively by N stress The average maximum LAI was 43% lower in the NS plants than the CT plants. This reduction in LAI had a lag of about four weeks in response to the N stress. In addition, the specific leaf area (SLA) was reduced by almost 50% in N stressed plants comparing with CT (Figure 6 7). The reduction in growth was not noticeable until about one month after the N stress was started (Figure 6 8). The fruit number was reduced by about 35% in N stressed plants compared to the CT. The harvest index was significantly different between the treatments. This finding suggests that nitrogen stressed plants reduced the partitioning of dry matter to the fruits under stress conditions.

PAGE 144

144 The contribution of each class size to the total yield was different according treatments. Thus, during 2007 81 % of total fruit yield in CT corresponded to large, extr a large an d maximum large fruit classes comparing with 65% of contribution from these classes in NS treatment (Table 6 5). Nitrogen concentration: The N treatment significantly changed the leaf blade N concentration. It did not change the N concentration in the peti oles and stems. On average, the blade N concentration of N stressed plants was 20% lower than the N concentration in blades from well fertilized plants. Overall, the difference in the N concentration of the fruit in the varying treatments was less than in the bla des and stems (Figures 6 10 to 6 -13). Growth of Individual Fruits under Water Stress Fruit dry weight and fresh weight: The re were observed differences in the means of the final fresh and final dry weight s of the individual fruits The data concerni ng with growth and development of fruits that beg a n at different dates in the two treatmen t groups (WS and CT) are shown in Table 6 4 The fresh and dry weight per fruit differed significantly between treatments for the three cohorts. Fruits developed on W S plants achieved the lowest dry mass and fresh mass. The final dry mass of cohort number 1 in the WS treatment group was 17% lower than the same cohort of the CT plants (Figure 6 14). The final fresh weight of cohort number 1 from the WS fruits was 24% lo wer than the same cohort of CT (Figure 6 15). In cohort 2 the final DW achieved by WS fruits was 20% lower than the DW achie ved by fruits under CT (Figure 616). The FW of cohort 2 in WS was 24% lower than the FW achieved by th e same cohort under CT (Figu re 6 17). Similar but more accentuated differences were observed in cohort 3 (Figures 6 18 and 6 19). For these late setting fruits, the mean final DW of the fruits from the WS was 24% lower than the fruits from the CT treatment T he fresh mass was 26% lo wer in the WS plants than the same cohort of CT

PAGE 145

145 Fruit dry matter concentration: The final fruit dry matter concentration was stable for the three coho rts across treatments (Figures 6 20, 6 21 and 6 22). On average, the DMC of the fruits from water stresse d plants was higher than the fruits from CT conditions. Statistically, the differences were only significant for cohort 1 ( not for cohort 2 and 3) (Table 64 ). Fruit size: The final fruit size differed significantly between treatment s for the three cohorts (Table 6 4 ). Quantitatively, fruit diameter was the variable that varied most between the treatments. Fruits developed from plants grown under WS achieved a smaller diameter (Figures 6 23, 6 24 and 6 25). On average, the final fruit diameter of fruits fro m WS plants was 27% lower than CT plants for cohorts 1 and 3. The same variable was 19% lower for cohort 2. These results agree with Ehret and Ho (1986), Mitchell et al (1991) and Van Ieperen et al. (2005), who reported that water stress during growth of fruit reduced fruit size by 30%. Akaki et al. (2004) also demonstrated that water deficit resulted in smaller fruits with higher sugar concentrations. Growth of Individual Fruits under Nitrogen Stress Fruit dry weight and fresh weight : The observed differe nces in the means of the final fresh and final dry weight s of individual fruits are presented in Table 6 4 Growth and d evelopment data of fruit that beg an at different dates for the t wo treatments (NS and CT) is presented in Table 6 4 The dry weight per fruit differed significantly between the two cohorts. Fruits developed from NS plants achieved the l owest dry mass and fresh mass. Statistically, only the differences in DW were significant In contrast, the differences in FW were not. The final dry mass o f cohort 1 in the NS treatment group was 23% lower than the same cohort in CT. The final fresh weight of cohort 1 in the NS group was only 5% lower than the same cohort in the CT group (Figures 6 14 and 6 15). In the case of cohort 2, the final DW achieved by fruits of NS was 26% lower than the DW achieved by fruits under CT. The FW of cohort 2 in NS was 7.5% lower than the FW achieved by the sa me cohort under CT (Figures 616 and 617). The latest

PAGE 146

146 setting fruits remained small in plants under NS N o appare nt growth was observed in those fruits. Apparently, the N stress affected the normal development of the latest fruits A lthough no flower abortion was observed, no further growth after setting was observed. Fruit dry matter concentration: The final fruit d ry matter concentration was quite stable for the cohor ts in both treatments (Figures 6 18 and 6 19). On average, the DMC of fruits from N stressed plants was lower than the DMC of fruits under CT conditions S tatistically the differences were significant for both cohorts (T able 6 4 ). NS fruits showed a DMC 25% lower than CT fruits. The FW was not significantly different between treatments T his is related to the lower DMC of fruit from N stressed plants The DW plants sho wed significant differences. Fruit size: The final fruit size differed between treatments for the two cohorts. Fruits developed from the N stressed plants ac hieved lower diameter (Figures 6 23 and 6 24). On average, fruits in NS were 7.5% smaller than CT fruits for cohort 1. The fruits were 9% smaller for cohort 2. Although the difference was significant for cohort 1, it was not for cohort 2 (Table 6 4 ). Discussion Fruit growth: Water stress caused a significant decrease in fresh fruit weight and fruit size Water stress also increased the dry matter concentration. Since the size of tomato fruits depends largely on water accumulation, it is not surprising that the size of the fruits was affected by water stress. Ehret and Ho (1986), Mitchell et al. (1991), Scholberg, (1996) and van Ieperen e t al. (2005) reported similar results. W ater stress had less effect on the dry matter accumulated by the fruits than the fresh weight. As a result of the reductions in size and fresh mass, the DMC of fruits increased in water stressed plants. Reduced fruit fresh weight as result of water stress is expected. Ho et al. (1987) explained that water stress produces an increase in fruit dry matter concentration because it increases the phloem sap concentration and decrease s its flux. The

PAGE 147

147 phloem flux is responsibl e for the increase in size of fruits via both water and carbon M ore than 88% of fruit water acquisition is via phloem (Guichard et al., 2001). Under water stress the decrease in this flux produced smaller fruits that were higher in dry matter concentrati on This finding was shown in our results. In addition, Guichard et al. (1999) showed that under water stress, the xylem flux to the fruits is reduced by 25% T he fruit transpiration is increased by 27%. A s a result the fruits reduce the net water accumul ation via xylem by 30%. Akaki et al. (2004), published similar results. According to their results, in normally watered plants the water accumulation of tomato fruits is 60% via phloem and 40% via xylem. Under water stress the contribution of each pathwa y changes dramatically In fact, 85% of the fruit water is acquired via phloem sap and only 15% via xylem. U nder water stress the water accumulated by toma to fruits is depressed by 36%. Why is the flux altered under water stress? For tomato fruit to grow there must be a total water potential gradient between the fruit and the rest of the pla nt (Grange and Andrews, 1994). Guichard et al. (1999) showed a close relationship between the fruit water potential and the daily flow of water toward the fruit. Johnson et al. (1992) also showed that the water potential gradient between the fruits and the plant controls the phloem -driven growth of the tomato fruits. The water potential of fruits is relatively unrespons ive to the environment (Johnson et al., 1992; Guicha rd et al., 1999) T herefore the flux depends largely on the stem water potential C ontrary to the fruit, the stem water potential is highly sensitive to water stress (Johnson et al., 1992; Guichard et al., 1999). In this study, we assume d that the reducti on in the leaf xylem water potential in the WS wa s an indication that the stem water potential was lowered It is possible that the driving force for water flow into the fruit may be reduced under WS T his may be the reason the FW and the size of individua l fruits of water stressed plants were reduced. In our study, we observed that the fruits under WS ripened before the CT fruits

PAGE 148

148 T he observed reduction in final dry weight may have been caused by a shorter growth period rather than a lower slope of growth rate The results of Wolf and Rudich (1988) support this last assertion. The effect of N stress on the growth of individual fruits was opposite to that of water stress. While the fresh weight and the volumetric size of fruits were slightly affected by N st atus, the dry weight and the DMC was significantly reduced in fruits of N stressed plants compared to N -supplied plants. However, t he reduction in leaf photosynthesis rate was less than proportional to the reduction in dry weight. T he LAI was largely impac ted by N stress because leaf SLA was reduced This change resulted in a decrease in LAI and no change in SLN. The reduced harvest index in N stressed plants may indicate that at least a portion of the reduction in dry weight may be explained by a lower pa rtitioning of dry matter to fruits. Crop total yield: This study showed that the impact of N stress on the growth and final yield of a tomato crop s was greater that the impact of water stress. However, it is important to note that water stress was rather m oderate and the N stress was severe Therefore no conclusions about the relative impacts of each stress should be made. The lower leaf area index, decreased biomass accumulation and yield reduction under N stress agree s with the results of Scholberg et al (2000) This study showed that severe N stress significantly reduced LAI, biomass, and fruit yield of tomatoes grown in open field s The reduction in fruit yield may be due a double effect of N stress T he impact on the vegetative growth (leaf area) in t urn reduce d the accumulated biomass T he number of fruits that reached commercial size was reduced under NS compared with the control plants. The leaf photosynthesis rate of N stressed plants was moderately impacted. However the reduced SLA limited the LAI and canopy assimilation. T his is likely the reason the number of fruits of

PAGE 149

149 commercial size varied with the final yield. In addition the harvest index was lower in N stressed plants than in well fertilized plants This finding suggests that the partitioning of dry matter to the fruits was impacted by N stress. Studies by Huett and Dettmann (1988, 1989, 1991) and by Fujita et al (2004) support that N stress affects the dry matter partitioning of vegetable crops including the tomato T he FW and size of the fruits were relatively un affected by N status T herefore these two variables affected the final yield less In the case of WS, our study showed that it affected final yield through a reduction i n the size and FW of the fruits It had less effect on the number of fruits than the NS. In addition, plant growth and leaf photosynthesis rate s were less affected than in N stressed plants Th ese two variables did not largely affect the final yield. In our experiments, the le L to 1.6 MPa In other studies, it lowered it to 2.6 MPa or lower. This may be why the moderate and successively imposed drought during our experiment did not dramatically reduce the canopy growth and the leaf photosynthesis r ate. Summary and Conclusion O ur results are generally consistent with other authors concerning the effects of water and N stress on the growth of individual fruits tota l plant growth and fruit yield. According to the results discussed above, a way to simu late water stress and N stress (indirectly) would be to reduce the stem water potential (to decrease the dry weight) and the phloem carbon concentration (to decrease the fresh weight ). This approach seems feasible for hypothetical models that specifically simulate water and carbon fluxes into individual fruits. This is not the case for the CROPGRO T omato, which simulates the carbon balance at the whole plant level This type of tomato model lacks phloem transport of carbon and water into the fruits. The fr esh mass and fruit size are predicted from the single fruit DW This prediction of DMC was described in Chapter 5. Consequently, in CROPGRO, it may be more feasible to simulate reductions in the single fruit growth indirectly This could be achieved through reductions in the whole plant, the number of

PAGE 150

150 fruits, the C partitioning or a combination of these factors. This model already uses reducing functions at the whole plant level These functions include TURFAC / SWFAC (for water stress) and NSTRES (for N stre ss). While these functions have been well evaluated in relationship to plant growth and total yield, we need to explore how they may work in simulating th e growth of individual fruits. In addition, empirical modifications of the relatively simple equations that simulate the DW, FW, DMC, and size of individual fruits need to be explored. These formulas may serve to mimic the dynamics of fruit growth under N and water stress.

PAGE 151

151 Table 6 1. Standard size classes for fresh tomatoes in the U.S Size class Diameter Diameter Diameter Diameter Minimum a Maximum b Minimum a Maximum b --mm ----mm ----in ----in --Extra small 48 54 1 28/32 2 4/32 Small 54 58 2 4/32 2 9/32 Medium 58 64 2 9/32 2 17/32 Large 64 73 2 17/32 2 28/32 Extra large 73 88 2 28/32 3 14/32 Max. large 88 3 15/32 Source: United States standards for fresh tomato fruit, in Tomato Plant Culture Jones Jr. (1999). a Will not pass through a round opening of the designated dia meter when the tomato is placed with the greatest transverse diameter across the opening. b Will pass through a round opening of the designated diameter in any position. Table 6 2 Means comparison among three treatments for tomato leaf gas exchange and N leaf, stem and fruit concentration in Gainesville Fl, during spring of 2007. Variable Units Control Water Stress Nitrogen Stress Leaf photosynthesis mol m 2 sec 2 29.9 a 26.0 b 23.8 b Stomata l resistance sec 1 cm 0.32 b 0.47 a 0.42 a Leaf bl ade N conc. % 3.13 a 3.02 a 2.51 b Leaf petiole N conc. % 1.73 a 1.67 a 1.48 a Stem N conc. % 1.85 a 1.78 a 1.67 a Fruit N conc. % 3.93 a 3.69 ab 3.28 b Test: Duncan Alpha:=0.05 Different letters indicate significant difference between t reatments (p<= 0.05) Table 6 3 Means comparison among three treatments for tomat o total above biomass, total fruit dry and fresh fruit yield, harvest index, fruit number and maximum LAI at harvest date on Jun 25 (NS), Jun 28 (WT) and Jul 05 (CT), in Ga inesville, Fl, during spring of 2007. Variable Units Control (CT) Water Stress (WS) Nitrogen Stress (NS) Total above biomass kg ha 1 7137 a 5586 b 3810 c Total fruit dry weight kg ha 1 4874 a 3694 b 1997 c Total fruit fresh weight kg ha 1 79262 a 57715 b 46946 c Maximum LAI m 2 m 2 3.57 a 3.02 a 2.04 b Fruit number units 57 a 46 b 37 c Harvest index kg kg 1 0.69 a 0.67 a 0.52 b Test: Duncan Alpha:=0.05 Different letters indicate significant difference between treatments (p<= 0.05)

PAGE 152

152 Table 6 4 Means comparison among three treatments of growth variables of individual tomato fruit at harvest in Gainesville Fl, during spring of 2007. Treatment Cohort #1 Cohort #2 Cohort #3 DW g fruit 1 Control (CT) 13.6 a 9.08 a 3.58 a Water Stress (WS) 11.3 b 7.25 b 2.71 b Nitrogen Stress (NS) 10.5 b 6.70 b nd FW g fruit 1 Control (CT) 279 a 172 a 72 a Water Stress (WS) 211 b 130 b 53 b Nitrogen Stress (NS) 266 a 159 a nd Fruit DMC (%) Control (CT) 4.8 b 5.0 a 4.6 a Water Stress (WS) 5.3 a 5.5 a 5.1 b Nitrogen Stress (NS) 4.0 c 4.2 b nd Fruit Diam eter (cm) Control (CT) 10.4 a 7.5 a 3.3 a Water Stress (WS) 7.6 c 6.1 b 2.4 b Nitrogen Stress (NS) 9.6 b 6.8 a b nd T est: Duncan Alpha:=0.05 Different letter s indicate significant difference between treatments (p<= 0.05)

PAGE 153

153 Table 6 5 Contribution of each commercial fruit size class to total yield fruit at harvest in Gainesville Fl, during spring of 2007. Treatment Fruit Size Class Fresh Wt. kg ha1 Dry Wt. kg ha1 % of Total fresh yield Number fruits as (%) of each class Control C ull 479 19 0.6 7 Extra small 1494 55 1.89 8.8 Small 3967 138 5 8.8 Medium 9020 337 11.4 10.5 Large 31980 596 40.5 26.3 Extra lar ge 15680 1892 19.9 29.87 Maximum large 15993 678 20.3 9 Water Stress C ull 573 30 1 17 Extra small 1823 87 3.2 15.2 Small 5018 234 8.7 8.7 Medium 8070 371 14 19 .4 Large 17547 1330 30.5 21.7 Extra large 15460 1200 26.8 17.4 Maximum lar ge 3393 548 5.9 2 .4 Nitrogen Stress C ull 917 38 1.96 16.2 Extra small 1500 84 3.19 13.5 Small 5025 175 10.7 8.1 Medium 5823 207 12.4 10.8 Large 14029 607 20.9 18.9 Extra large 12707 539 27 19 Maximum large 7962 255 16.9 10.8

PAGE 154

154 -1.60 -1.40 -1.20 -1.00 -0.80 -0.60 -0.40 -0.20 0.00 6/2 6/5 6/8 6/11 6/14 6/17 6/20 6/23 6/26 6/29 7/2 7/5 7/8 Date Leaf Water Potential (MPa) Contol water on N Stress Figu re 6 1. Leaf water potential of mature tomato leaves under three treatmen ts in Gainesville Fl, during spring o f 2007. 0 0.02 0.04 0.06 0.08 0.1 0.12 6/4 6/8 6/12 6/16 6/20 6/24 6/28 7/2 7/6 Date Daily Soil Water Content (cm3cm-3) Control water on Figure 6 2. Daily soil water content based on 20 minute values from TDR probes buried at 20 cm depth, in experiments conducted in Gaines ville Fl, during spring of 2007.

PAGE 155

155 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 6/4 6/8 6/12 6/16 6/20 6/24 6/28 7/2 7/6 Date Daily Soil Water Content at 40 cm (cm3cm-3) Control water on Figure 6 3. Daily soil water content based on 20 minute values from TDR probes buried at 40 cm depth, in experiments conducted in Gainesville Fl, during spring of 2007. 0 5 10 15 20 25 30 35 6/2 6/7 6/12 6/17 6/22 6/27 7/2 7/7 7/12 Date Leaf Photosynthesis (umol m-2 s-1) Control Water Stress N Stress Figure 6 4. Photosynthesis rate of mature tomato leaves under three treatments in Gainesville Fl, during spring of 2007; symbols represent means of 16 leaves per treatment.

PAGE 156

156 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 6/2 6/7 6/12 6/17 6/22 6/27 7/2 7/7 7/12 Date Stomatal Resistance (s cm-1) Control Water Stress N Stress Figure 6 5. Stomata l resistance of mature tomato leaves under three treatments in Gainesville Fl, during spring of 2007 (symbols represent means of 16 leaves per treatment. 0 1 2 3 4 4/17 4/27 5/7 5/17 5/27 6/6 6/16 6/26 7/6 7/16 Date Leaf Area Index (m2m-2) Control Water Stress N Stress (off from 04/17) Figure 6 6. Effects of water and nitrogen on Leaf Area Index (LAI) over time of tomato plants in Gainesville Fl, during spring of 2007.

PAGE 157

157 0 100 200 300 400 500 4/17 4/27 5/7 5/17 5/27 6/6 6/16 6/26 7/6 7/16 Date Specific Leaf Area (cm2 g-1) Control Water Stress N Stress (off from 04/17) Figure 6 7. Effects of water and nitrogen stress on Specific Leaf Area (SLA) over time of tomato plants in Gainesville Fl, during spring of 2007. 0 2000 4000 6000 8000 10000 4/17 4/27 5/7 5/17 5/27 6/6 6/16 6/26 7/6 7/16 Date Total Above Biomass (kg ha-1) Control Water Stress N Stress (off from 04/17) Figure 6 8. Effects of water and nitrogen stress on total above biomass over time of tomato plants in Gainesville, Fl, during spring of 2007.

PAGE 158

158 0 1000 2000 3000 4000 5000 6000 4/17 4/27 5/7 5/17 5/27 6/6 6/16 6/26 7/6 7/16 Date Total Fruit Dry Weight (kg ha-1) Control Water Stress N Stress (off from 04/17) Figure 6 9. Effects of w ater and nitrogen stress on total fruit dry weight of tomato plants over time in Gainesville, Fl, during spring of 2007. 0 10000 20000 30000 40000 50000 60000 70000 80000 4/17 4/27 5/7 5/17 5/27 6/6 6/16 6/26 7/6 7/16 Date Total Fruit Fresh Weight (kg ha-1) Control Water Stress N Stress (off from 04/17) Figure 6 10. Effects of water and nitrogen stress on total fresh yield of tomato plants over time in Gainesville, Fl, during spring o f 2007.

PAGE 159

159 4/17 N off 0 1 2 3 4 4/17 5/7 5/27 6/16 7/6 7/26 Date Leaf Blade N (%) Control (Blades) N Stress (Blades) W Stress (Blades) Figure 6 1 1. Leaf blades N concentration over tim e of tomato plants under N and water stress grown in Gainesville, Fl, during spring of 2007. 4/17 N off 0.0 1.0 2.0 3.0 4.0 4/17 5/7 5/27 6/16 7/6 7/26 Date Stem N (%) Control N Stress W Stress Figure 6 1 2. Stem N concentration over time of t omato plants under N and water stress grown in Gainesville, Fl, during spring of 2007.

PAGE 160

160 4/17 N 0ff 0.0 1.0 2.0 3.0 4.0 5/17 5/27 6/6 6/16 6/26 7/6 7/16 Date Fruit N (%) Control N Stress W Stress Figure 6 1 3. Fruit N concentration over time of tomato plants under N and water stress grown in Gainesville, Fl, during spring of 2007. 0 2 4 6 8 10 12 14 16 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Dry Weight (g fruit-1) Control Water Stress N Stress Figure 6 14. Effects of water and nitrogen stress on dry we ight ac cumulation over time in individual tomato fruits for cohort number 1 in Gainesville, Fl, during spring of 2007.

PAGE 161

161 0 50 100 150 200 250 300 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fresh Weight (g fruit-1) Control Water Stress N Stress Figure 6 1 5. Effects of water and nitrogen stress on fresh weight accumulation over time in individual tomato fruits for cohort number 1 in Ga inesville, Fl, during spring of 2007. 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Dry Weight (g fruit-1) Control Water Stress N Stress Figure 6 1 6. Effects of water and nitrogen stress on dry weight accumulation over time in individual tomato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007.

PAGE 162

162 0 50 100 150 200 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fresh Weight (g fruit-1) Control Water Stress N Stress Figure 6 1 7. Effects of water and ni trogen stress on fresh weight accumulation over time in individual tomato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007. 0 1 2 3 4 0 5 10 15 20 25 30 35 40 Days After Tagging Dry Weight (g fruit-1) Control Water Stress Figure 6 1 8. Effects of water stress on dry weight accumulation over time in individual tomato fruits for coh ort numb er 3 in Gainesville, Fl, during spring of 2007.

PAGE 163

163 0 20 40 60 80 100 0 5 10 15 20 25 30 35 40 Days After Tagging Fresh Weight (g fruit-1) Control Water Stress Figure 6 1 9. Effects of water stress on fresh weight accumulation over time in individual tomato fruits for cohort numb er 3 in Gainesville, Fl, during spring of 2007. 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fruit Dry Matter Concentration (%) Control Water Stress N Stress Figure 6 20. Effects of water and nitrogen stress on dry mat ter concentration over time in individual tomato fruits fro cohort number 1 in Gainesville, Fl, during spring of 2007.

PAGE 164

164 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Dry Matter Concentration (%) Control Water Stress N Stress Figure 6 2 1 Effects of water and nitrogen stress on dry matter concentration over time in individual to mato fruits for cohort number 2 in Gainesville, Fl, during spring of 2007. 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0 5 10 15 20 25 30 35 40 Days After Tagging Fruit Dry Matter Concentration (%) Control Water Stress Figure 6 22. Effects of water stress on dry matter concentration over time in individual tomato fruits for cohort number 3 in Gainesville, Fl, during spring of 2007.

PAGE 165

165 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fruit Diameter (cm) Control Water Stress N Stress Figure 6 23. Effects of water and nitrogen stress on the di ameter over time of individual tomato fruits for cohort number 1 in Gainesville, Fl, during spring of 2007. 0 1 2 3 4 5 6 7 8 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Diameter (cm) Control Water Stress N Stress Figure 6 24. Effects of water and nitrogen stress on the dia meter over time of individual tomato f ruits for cohort number 2 in Gainesville, Fl, during spring of 2007.

PAGE 166

166 0 1 2 3 4 0 5 10 15 20 25 30 35 40 Days After Tagging Fruit Diameter (cm) Control Water Stress Figure 6 25. Effects of water stress on the diameter over ti me of individual tomato fruits for cohort number 3 in Gainesville, Fl, during spring of 2007.

PAGE 167

167 CHAPTER 7 CROPGRO TOMATO SIMULATION OF INDIVI DUAL FRUIT GROWTH UNDER WATER AND NITROGEN LIMITAT IONS Introduction Size, water, and concentration of carbon compounds are the main criteria for assessing the quality of fresh fruits (Liu et al ., 2007). For fresh market tomato, those va riables also determine the production price, which largely depends on the size of the fruits. Tomato requires adequate water and nitrogen fertilization in order to obtain high yield and fruit quality. However, this goal is frequently not attained. In deve loped countries, because of the relatively inexpensive price of land inputs (a fact that may rapidly change according the economic crisis worldwide), leads the farmers to overuse both water and nitrogen resulting in leaching and often poor fruit quality be cause of fruit cracking and associated problems. In developing countries, on the contrary, because fertilizers and suitable irrigation systems are expensive, farmers tend to grow plants under varying levels of water and N limitation. In either situation, t he result may be losses of fruit yield and quality. Suitable crop growth models may be valuable tools to understand the impact of water and N fertilization management on tomato production, and once validated, the results may be used to optimize resources a nd profits. However, quality traits are seldom subject to modeling, probably because they are the result of a poorly understood chain of processes (Struik et al., 2005). Rinaldi et al (2007) have successfully utilized CROPGRO tomato as decision support r egarding N and irrigation strategies for processing tomato production in southern Italy. In a similar manner, CROPGRO has been used to study soybean and other legumes in the USA as well as in other countries. These applications have evaluated the ability o f CROPGRO to simulate the response of plant growth to water and N stress in terms of total yield and biomass production. The growth of individual fruits in response to water and N

PAGE 168

168 limitations, however, has been not studied although such a simulation may po tentially contribute to optimize management of yiel d and quality of tomato fruits T he results obtained in Chapter 6 in dicated general responses of fruit growth to water and N stress compared with well -watered and fertilized plants. Based on those results, to adequately simulate the response to stress for individual fruit growth, a model must be able to: 1) create lag (slower fruit development) at the start of the growth period of the fruits under assimilate stress, especially for late -set fruits, 2) give priority for assimilates to older, first -formed fruits as compared with younger fruits on a cohort basis if stress is present, 3) reduce the fresh weight and size and increase the dry matter concentration of fruits under water stress and reduce the dry weight of fruits when plants are under N stress. In addition, the model should reduce the total fruit number and total fruit mass per plant under stress. In addition, the leaf expansion (leaf area index) as well as the specific leaf area should be also decreased under water and N stress. As result, a reduction of total biomass and total fruit dry weight is expected Currently, the CROPGRO model accounts for the effects of water deficit on the plant through several reducing factors proposed by Ritchie (19 98) and called SWFAC (for effects on processes such as photosynthesis) and TURFAC (for effects of water deficit on expansive processes such as leaf expansion). Every day, the model computes the ratio between the root water uptake (supply) and the evapotranspirative demand. The ratio needs to be at least 1.0 in order to maintain photosynthesis (dry matter accumulation) a t a normal rate, and 1.5 for expansi ve growth to be at its potential. A similar approach is used in CROPGRO to compute daily N stress. In th is case, the reducing factor is called NSTRES, the model internally computes the daily ratio between the N uptake by the roots (and N -fixation if a legume) and the N demand for dry matter growth. When the ratio between N uptake and N demand falls below a certain

PAGE 169

169 threshold to ensure a target N concentration in tissues grown today, an NSTRES value less than 1.0 results, and tissue N concentration is reduce d and the growth is decreased. At a first look, this factor combined with reduced partitioning to fruits, may be useful to indirectly mimic the reduction in dry weight that fruits experienced under N stress. Indirectly the fruit number also may be affected because effects on fruit abortion may occur when the carrying capacity of the plants is limited due to a low N supply and also low assimilation rate. The impact on water accumulation of fruits under water stress, however, may be more difficult to simulate because there is currently no connection between CROPGROs simulation of water stress and actual water a ccumulation in individual fruits. Water accumulation in fruits is not directly simulated by the model but rather inferred from the dry weight and the computed dry matter concentration. It may be feasible to use the SWFAC or TURFAC as factors/signals to af fect dry matter concentration, and as a result alter the fresh mass and size. For these signals to work adequately, they need to affect also the timing of reproductive events under stress. Thus, it may be necessary to increase the lag at the start of the g rowth period for later -set fruits, but then to accelerate the maturation process resulting in a similar or even shorter total fruit growth period. It may be possible to modify the equations for dry matter concentration, fresh mass and fruit size (presented in the Chapter 5 ), using these water and N stress factors to mimic stress effects. This approach will be followed in this chapter, however, additional experiments may be needed in order to define the proportionality factors to improve accuracy because there are many possible scenarios of stress es The objective of the present Chapter is to analyze the capability of the CROPGRO Tomato model to simulate the growth of individual fruits when plants face water or N stress, and if needed to study possible mechanisms for overcoming the limitations of the model.

PAGE 170

170 Material s and Methods Experimental data: The data used for model evaluation were derived from field experiments carried out at the University of Florida Plant Science Research and Education Unit in Citra, Florida (29 (CT), water stress (WS) and nitrogen stress (NS), as well as the experimental conditions and the measurements were fully described in Chapter 6 In addition, in Table 7 1 the characteristics of the soil series Tavares which correspond to the experimental site are presented. The model : The version of CROPGRO Tomato model as described in Chapter 5 was used in this study. No new calibration of genetic coefficients was neede d because the same cultivar, season, location and experimental conditions as described in Chapters 5 and 6 were evaluated. In order to study the response of the model to water and N stress, the model was first run without any modification of parameters usi ng the observed weather, irrigation, N application, soil water holding and N supplying traits of each experimental treatment. The outputs of the model were compared with observed data for a preliminary assessment of the model Model modification: Following the above comparison, several strategies were tested to improve the ability of the model to reproduce the effects of water deficit and N limitation on dry weight growth, fresh mass growth, and size of individual fruits. The following modifications we re made to the model: 1) In the DEMAND routine of CROPGRO, which calculates the carbon and N demand for vegetative and reproductive growth, the maximum shell (fruit without seed) growth rate was allowed to be reduced by SWFAC and NST RES modifiers to reduce the potential growth rate under water or N deficit according to the following equations: TMPFAC SHVAR GRRAT 1 (7 1)

PAGE 171

171 2 0 1 1 4 0 1 1 1 NSTRES SWFAC TMPFAC SHVARGRRAT (7 2)1 Where Equation 7 1 is the default eq uation in CROPGRO and Equation 72 is the one modified by stress signals. For the default model, this potential fruit growth rate (GRRAT1) is only dependent on temperature, although subsequently, actual fruit growth rate can be limited if C or N supply is insufficient to grow that cohort today. According to equation 7 2 when plants face water stress or N stress the maximum growth rate of fruits will be reduced proportionally to the values of SWFAC or NSTRES. The modifiers 0.4 and 0.2 are empirical and were solved through sensitivity analysis comparing outpu ts of the model with real data. 2) The physiological age of the fruits is related to stress by computing thermal time accumulators that are differently affected by water or N stress, and calling them if needed. In the default model, only one thermal time accumulator ( PHTIM) is used ( Eq. 7 3) By allowing different thermal time accumulators to operate, it is possible to affect the physiological age of the fruits in such a way that the developmental age may be accelerated or slowed as a function of SWFAC or N STRES (Eq. 7 4). This last modification was connected to the function for computing the dry matter concentration of individual fruits as a function of physiological age (PAGE) in the FRESH WEIGHT subrou tine of the model (see Chapter 5 ). The physiological age of the f ruits is the independent variable of the equation and can potentially be different depending on water or N stress. 1 GRRAT1 is the maximum growth per individual shell (g/shell/day) SHVAR is the genetic potential shell growth rate during its rapid growth phase, with all conditions optimal, (g/shell/day) TMPFAC is a modifier of genetic maximum growth rate for shell depending on temperature NSTRES is the N stress factor, computed in the CROPGRO subroutine using total vegeta tive and reproductive N demand, (1.00=no stress, 0.0=max stress) SWFAC is the effect of soil water stress on photosynthesis, (1.0=no stress, 0.0=max stress)

PAGE 172

172 TDUMX NR DAS PHTIM NR DAS PHTIM 2 1 2 (7 3) 1 0 1 1 3 0 1 1* 2 1 1 2 1 CNSTRES SWFAC TDUMX NR DAS PHTIM NR DAS PHTIM (7 4)2 Where Equation 7 3 is the def ault eq uation in CROPGRO and Equation 7 4 is the one modified by stress signals. The modifiers 0.3 and 0.1 are empirical and were solved through sensitivity analysis comparing outputs of the model with real data 3) Finally, the dry matter concentration wa s allowed to respond to the SWFAC and NSTRES factors by including SWFAC and NSTRES in the equation for computing dry matter concentration in such a way that the dry matter concentration is increased under water stress and reduced under N stress in agr eemen t with measured data (Eq. 7 5 and 76 ). 55 6 1 1 7 10 5 PAGE EXP DMC (7 5) 55 6 1 1 1 7 10 1 5 1 1 1 5 1 5 PAGE EXP NSTRES PAGE PAGE DMC (7 6) 2 PHTIM is cumulative photothermal time in ages of seeds and shells the default accumulator PHTIM1 is cumulative photothermal time in ages of seeds and shells modified by water and N stress (updated accumulator) DAS are the d ays after start of simulation, NR2 is days when 50 % of plants have one fruit (fruit ) (d) TDUMX is the photothermal time that occurs in a real day based on early reproductive development temperature function CNSTRES is the N stress factor computed a s an 8 days moving average

PAGE 173

173 Where Equation 7 5 is the default equation in CROPGRO and Equation 7 6 is that modified by the stress signals The PAGE and PAG E1 are the photothermal age s of each cohort and are defined by PAGE= PHTIM(NR2TIM+1) (PHTIM (NPP) (7 7) PAGE1= PHTIM1(NR2TIM+1) (PHTIM1(NPP) (7 8) **NPP is the cohort number used as an index in loops The value 1.5 in equation 7 6 is an estimated value b ased in our observations of real data for increases in DMC of fruits under water stress where the ratio PAGE1/PAGE carries the cumulative water stress (SWFAC) from Equation 7 4 into Equation 7 6 and the 1.5 value causes decreases under N stress. To te st these strategies, comparisons of simulated and observed growth of individual fruits as well as growth of the whole plant and total fruit yield were evaluated using RMSE and Willmott index as criteria of evaluation Results and Discussion The results of the control treatment were discussed in Chapter 5 and are not discussed again in this section. However, they are presented in the figures for comparative reference with stress treatments and to verify that the model modifications based on SWFAC and NSTRES factors did not affect simulated shell growth, fruit age and dry matter concentration of the control treatment, but did affect simulated variables as intended to mimic the observed data of the water and N stress treatments. Water Stress Whole plant level: The CROPGRO Tomato model was able to reproduce the reduction in growth observed in tomato plants under water stress as compared with well -irrigated plants. As shown in Fi gure 7 1, the model was able to reproduce accurately the effect of water stress on l eaf

PAGE 174

174 expansion and LAI compared to measured data over time. In addition, the variation of LAI over time was well simulated by the model as revealed by the high d index and the low RMSE of the predictions. Similarly, the total above biomass (Figure 7 2) and total fruit dry weight (Figure 7 3) during the season were very well predicted by the model with a d index close to the optimal value 1 and low R MSE for both variables. Figure 7 4 shows the results for predicted total fruit fresh weight s howing that simulated total production of fruit fresh weight corresponded well to th at measure d on plants under the water stress treatment. The total fruit fresh weight is a result of (summed over) the number and fres h weight of individual fruits. Therefore it is possible that th is response is not only to the water deficit at the whole plant level but also to the newly added modifiers for affecting the growth of individual fruit s and dry matter concentration. Individual fruit dry weight: Observed and simulated dry weight accumulation in individual fruits from flowers that had been tagged at three weekly dates in plants under water stress tre atment are presented in Figures 7 5, 7 6, and 7 7. There was a good correspondence between simulated and observed data for cohort numb er s 1 and 2. In these cohorts, the Willmott agreement index (d) was close to an optimal value 1 The high d value indicates that the variability in the observations around the mean was well tracked by the predictions and therefore the model was able to rep roduce the timing of when the fruit growth was affected by water stress. The average RMSE for cohorts 1 and 2 was low and represents a model error of 9.5 % of the final fruit dry weight for these cohorts. A less satisfactory response of the model was for c ohort number 3, which gave an unacceptable d value (less than 0.9) and a RMSE that represents an error of 26 % of the final DW for this cohort. The poor simulation for this cohort has only minor implications for predictions of total fruit yield because the contribution of the late -set fruits to the total yield is small and represented less than 4 % of the total yield in the control

PAGE 175

175 treatment during 2007 and even less in the water stress treatment (data not shown). The results show that the model is able to give priority to earlier cohorts over later -formed cohorts when carbon supply is compromised by wa ter deficit. This is noticeable because cohort number 3 started to grow very late in the cycle after the earlier cohorts reached their constant final mass es The results for simulation of dry weight gain of individual fruits under water stress is basically a result of reduction of C availability at a whole plant level combined with minor effects on the maximum shell growth rate in the DEMAND routine, for which the later rate was reduced as a function of SWFAC. Fruit dry matter concentration : Observed and simulated dry matter concentration in individual fruits from flowers that had been tagged at three weekly dates in plants grown under water stress are presente d in Figures 7 8, 7 9 and 7 10. The model was able to mimic the increase in dry matter concentration observed in fruits under water stress, as was show n in the experimental results. Simulated and measured data agreed re asonably well with a RMSE of 0.6 % f or cohorts 1 and 2, respectively, whi ch represents a model error of 12 % of the final fruit DMC for these coh orts. The higher error RMSE (1.0 %) was for cohort number 3 and this error represents 20 % of the final fruit DMC for these fruits The Willmott in dex was high for all cohorts. The main factor improving the prediction of dry matter concentration was the addition of a SWFAC dependent modifier to the e quation 7 6 for computing the dry matter conce ntration of individual fruits. The SWFAC modifier used cumulative SWFAC over the fruit growth cycle to affect dry matter concentration over time, which gave an improved prediction and higher dry matter con centration under water stress. An alternative function using the current daily value of SWFAC did not work well, as it gave intermittent or bumpy instantaneous increases of DMC following daily variations in SWFAC (e quation not shown) These simulations confirm the

PAGE 176

176 concept of cumulative water deficit (memory) effe cts on DMC increase over time. The memory effect was created by carrying different thermal time accumulators into the FreshWt subroutine, one with the stress signal incorpor ated with the thermal accumulator and one without the stress signal (the default PHTIM), and dividing the first by the latter (Eq. 7 6) This approach worked well to reproduce cumulative effects of stress rather than instantaneous effects Fruit fresh weight: Observed and simulated fresh weight accumulation in individual fruits from flowers that had been tagged at three weekly dates in plants under water stress are p resented in Figures 7 11, 7 12, and 713. Measured data show that fresh mass and fruit size of individual fruits were the variables most affected by water stress. There was a very good correspondence between simulated and observed data for cohorts 1 and 2. In these cohorts, the Willmott agree ment index (d) was equal to 0.97 The average R MSE for both cohorts was low (12 g ) which represents an error of 7.4 % of the single fruit final fresh weight. Similar to dry weight pred iction, less satisfactory result s occurred for cohort number 3. The good performance of the model in reducing the fresh weight of individual fruits under water stress was relate d to a combination of steps 1 to 3 described in Material and Methods section. F irst and most important is related to the increased dry matter concentration, created by linkage to the cumulative SWFAC signal (Eq 7 6) However, that was not enough to produce the desired magnitude in reduction of fresh mass growth over time, as dry matt er g rowth was also slightly reduced which was linked to reduction in the maximum fruit (shell) dry matter growth rate (GRRAT1) under water stress (Eq 7 2). Fruit size: Observed and simulated individual fruit diameter from flowers that had been tagged at th ree weekly dates in water stressed treatments are presented in Figures 714, 7 15 and 7 16. Comparisons among cohorts showed that the fruit size s of cohorts 1 and 2 w ere well

PAGE 177

177 simulated by the model although the fruit diameter was slightly underestimated. T he average RMSE was 0.9 2 cm for both cohorts, whic h represents a model error of 14 % for the fruit diameter for both cohort s. The average d value was 0.97 indicating that the variability in the experiment was well captured by the predictions. The fruit size simulation of the cohort number 3 was poor as result of the poor simulation of fresh weight for this cohort. Nitrogen Stress Whole plant level: The N stress factors in CROPGRO were suitable to reproduce the reduction in total biomass and total fruit growt h observed in tomato plants under N stress as compared with well N fertiliz ed plants. As shown in Figure 7 17, the model was able to reproduce the effect of N deficit on leaf expansion and simulated LAI over time although LAI was slightly ove restimated fo r the control In addition, the variation of LAI over time was well simulated by the model as revealed by the high d index and the low RMSE of the predictions. T he total above biomass (Figure 718) was over estimated by the model, in part, because under N stress the model accumulates large amounts of carbohydrate in stems when insufficient N is ava ilable for new tissue growth. On the other hand, the model gives fruits first priority for N and assimilate and the total fruit dry weight (Figure 7 19) during t he season was very well predicted by the model with a d index close to the optima l value and a low RMSE. Figure 7 20 shows the results for predicted total fresh weight. The total production of fruit fresh weight was reasonably well predicted under the N st ress treatment. Fruit fresh weight in the model is calculated from the dry weight growth and the dry matter concentration for individual fruits and then total fruit fresh weight in the model comes from summing fresh weights over all fruit cohorts/numbers p resent. A good response of the model was the simulated reduction in carrying capacity of the crop as the simulated number of fruits under N stress was 62 % less than in control plants, which reduction was close to experiment al results (data not shown). The reduction in total fruit fresh weight in

PAGE 178

178 plants under N stress is a direct consequence of 1) the reduction of canopy expansion and photosynthesis under N deficiency which affected the carbon availability and the N supply itself for adding the next fruit. In addition, NSTRES was used as a factor to reduce the potential growth rate of individual fruit in the DEMAND routine. The N deficit did not appear to affect fru it physiological age or timing. Therefore, the reduced growth under N stress is less related to a different growth timing of fruits as compared with fruits under water stress conditions Fruit dry weight : Observed and simulated dry weight accumulation in individual fruits from flowers that had been tagged at three weekly dates in plants grown und er N stress tre atment is presented in Figures 7 21 and 7 22. There was a good correspondence between simulated and observed data for cohort number 1. For this cohort the Willmott agreement index (d) was 0.98. The RMSE was low and represents a model error of 9 % of the final fruit dry weight for cohort 1. The response of the model for cohort number 2 was poor, the model was able to reproduce stress, but the N stress response was too strong in the mid -cycle of fruit growth, resulting in a large under -predict ion of growth of this cohor t. Apparently, the model gave first priority for assimilate to the first cohort to the detriment of the second cohort and later -set fruits. This simulated behavior was mostly confirmed because the simulated third cohort failed to start, in agreement with the observed failure of the third cohort to begin growth under N stress, even though the se fruits were set and tagged. These late-set tiny fruits remained on the plant with no apparent growth until harvest. Fruit dry matter concen tration: Observed and simulated dry matter concentration in individual fruits from flowers which had been tagged at three weekly dates in plants grown under N stress are presented in Figures 7 23 and 724. The model was able to mimic the decrease in dry m atter concentration observed in fruits under N stress, compared to experimental results.

PAGE 179

179 Simulated and measured data agreed re asonably well with a RMSE of 0.5 % and 0.9 % for cohorts 1 and 2, respectively. The desired variation in dry matter concentration was created by adding an NSTRES -dependent modifier to the equation that calculates the dry matter concentration for individual fruits (Eq. 7 6). Because the NSTRES signal itself is instantaneous (from todays N supply/N demand ratio ), we used in equation 6 4 an 8 day running average (CNSTRES), which is a more cumulative memory type of variable, which worked well in this experiment For this particular experiment, the N cutoff was permanent and no later N was applied, so the memory effect was not as critic al. However it would be appropriate to have a more cumulative signal (as was done for SWFAC) or running average to better account for the real response which is cumulative and to prevent sudden recovery in the event of late N fertilization for example. Fr uit fresh weight: Observed and simulated fresh weight accumulation in individual fruits from flowers that had been tagged at three weekly dates in plants under N stress is presented in Figur es 7 24 and 7 25. There was good correspondence between simulated and observed data in case of cohort number 1. In these cohorts, the Willmott agreement index (d) was equal to 0.99 and the RMSE was 21 g which represents an error of 8 % of the single fruit final fresh weight. However the model underestimated the fresh ma ss toward the end of the season. Similar to dry weight prediction, a less satisfactory result occurred for cohort number 2. Cohort number 3 failed i n the field and the crop model predicted failure for this cohort too. Fruit size: Observed and simulated f ruit diameter of individual fruits from flowers that had been tagged at three weekly dates in N stressed trea tments is presented in Figures 7 27 and 7 28. Comparisons among cohorts showed that the fruit size of cohorts number 1 was well simulated by th e mo del although size was underestimated t oward the end of the season as a result

PAGE 180

180 of the underestimation of fresh weight. The fruit size of cohort number 2 was poorly simulated because of the dependence of this variable on fresh mass and dry mass which as expl ained was poorly predicted for this cohort. To better simulate priorities among cohorts e specially under N stress, there may be a need to modify code where the model currently gives a too strong (absolute) priority for th e first cohort and lets the later o nes fail. An alternative growth function could use Gompertz or logistic functions instead of the constant linear growth used by CROPGRO which would allow somewhat slower growth of cohorts number 1 and 2 late in the ir growth cycle, at a time when cohort nu mber 3 could then begin growth sooner particularly if the growth demand of that cohort is also in its slow growth phase (as represented in the Gompertz and logistics equations). These equations with slower growth demand early and late, may allow assimila te available to initiate grow th of cohort 3 (later fruits). Another possibility is to allow partitioning to rise above XFRUIT late in the li fe cycle to carry the latest fruits, but these possibil ities need be evaluated. Conclusion The CROPGRO Tomato model was tested as a tool for predicting the dynamic growth of individual fruits and the whole plant when the crop was grown under water and N limitation. The effect of N supply, low leaf N concentration on photosynthesis, and the reducing factors SWFAC, TURFAC and NSTRES inside the CROPGRO model were suitable to reproduce efficiently reductions of whole plant growth and total fruit dry matter growth under N deficit. However, in this work, t he same factors were then strategically connected to the addition and dry matter growth rate of individual fruits and the fru its thermal time accumulator. In addition, these factors were used to modify the function that calculates the fruit dry matter concentration. The result was an improved ability of the model to mimic the dry weight and fresh weight growth and dry matter concentration behavior of individual fruits under stress. The results are promising and

PAGE 181

181 suggest that the CROPGRO Tomato model can be a useful research and application tool for tomato fruit yield and qua lity predictions. However, stress physiology in tomato fruits is complex and further experimental work is recommended in order to optimize the magnitude and timing of fruit growth when using SWFAC, TURFAC, and NSTRES as signals to affect the dynamic s of to mato fruits under water and N stresses in the simulations.

PAGE 182

182 Table 7 1. Physical, c hemical and water holding capacity characteristics of soil series Tavares, (uncoated sand soil). S ource Carlisle et al. (1988) Layer cm Lower limit Upper limit Soil Sat. Root growth factor K cm h1 Bulk density g cm 3 Org. carbon % pH w ater Clay % Silt % 5 0.072 0.142 0.364 1 21 1.6 1.41 4.2 1.5 1.5 15 0.072 0.142 0.364 1 21 1.6 1.41 4.2 1.5 1.5 30 0.051 0.102 0.371 0.638 21 1.6 0. 66 4.3 1.5 1.2 45 0.044 0.091 0.371 0.472 21 1.6 0.44 4.3 1.5 1.5 60 0.045 0.092 0.347 0.350 21 1.6 0.18 4.2 1.3 1.5 90 0.038 0.078 0.35 0.223 21 1.67 0.18 4.2 1.5 1.5 120 0.046 0.086 0.364 0.122 21 1.63 0.08 4.3 3 .0 1.2 150 0.048 0.087 0.365 0.067 21 1.63 0.08 4.4 4 .0 1.5 180 0.052 0.091 0.358 0.037 21 1.65 0.05 5 .0 5 .0 1.5

PAGE 183

183 RMSE=0.32 d=0.98 RMSE=0.49 d=0.97 0 1 2 3 4 5 0 20 40 60 80 100 Days After Planting Leaf Area Index (m2 m-2) LAI (Simulated WS) LAI (Simulated C) LAI (Obserevd WS) LAI (Observed C) Figure 7 1. CROPGRO simulated (lines) and measured (points) leaf area index in Gainesville, FL during spring of 2007 comparing well irrigated with water stressed tom ato plants. RMSE=312 d=0.99 RMSE=330 d=0.99 0 2000 4000 6000 8000 10000 12000 0 20 40 60 80 100 Days After Planting Total Above Biomass (kg ha-1) Tops wt kg/ha (Simulated WS) Tops wt kg/ha (Simulated C) Tops wt kg/ha (Observed WS) Tops wt kg/ha (Observed C) Figure 7 2. CROPGRO simulated (lines) and measured (points) total above biomass in Gainesville, FL during spring of 2007 comparing well irrigated with water stressed tomato plants.

PAGE 184

184 RMSE=283 d=0.99 RMSE=178 d=0.99 0 2000 4000 6000 0 20 40 60 80 100 Days After Planting Totla Fruit Dry Weight (kg ha-1) Fruit wt kg/ha (Simulated WS) Fruit wt kg/ha (Simulated C) Fruit wt kg/ha (Observed WS) Fruit wt kg/ha (Observed C) Figure 7 3. CROPGRO simulated (lines) and measured ( points ) total fruit dry weight in Gainesville, FL during spring of 2007 comparing well irrigated with water stressed tomato plants. RMSE=10008 d=0.94 RMSE=15061 d=0.95 0 20000 40000 60000 80000 100000 120000 0 20 40 60 80 100 Days After Planting Total Fruit Fresh Weight (kg ha-1) Fruit FW kg/ha (Simulated WS) Fruit FW kg/ha (Simulated C) Fruit FW kg/ha (Observed WS) Fruit FW kg/ha (Observed C) Figure 7 4. CROPGRO simulated (lines) and measured ( points ) total fruit fresh weight in Gainesville, FL during spring of 2007 com paring well irrigated with water stressed tomato plants

PAGE 185

185 RMSE=0.77 d=0.99 RMSE=0.68 d=0.99 0 2 4 6 8 10 12 14 16 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Dry Weight (g fruit-1) Frui DW,g/f (Simulated WS) Fruit DW,g/f (Simulated C) Fruit DW,g/f (Observed WS) Fruit DW,g/f (Observed C) Figure 7 5. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato p lants. RMSE=0.81 d=0.98 RMSE=0.98 d=0.98 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Dry Weight (g fruit-1) Frui DW,g/f (Simulated WS) Fruit DW,g/f (Simulated C) Fruit DW,g/f (Observed WS) Fruit DW,g/f (Observed C) Figure 7 6. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants.

PAGE 186

186 RMSE=0.87 d=0.85 RMSE=0.62 d=0.89 0 1 2 3 4 0 5 10 15 20 25 30 35 40 45 Days After Tagging Dry Weight (g fruit-1) Frui DW,g/f (Simulated WS) Fruit DW,g/f (Simulated C) Fruit DW,g/f (Observed WS) Fruit DW,g/f (Observed C) Figure 7 7. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. RMSE=0.3 d=0.997 RMSE=0.6 d=0.97 0 2 4 6 8 10 12 14 16 18 0 10 20 30 40 50 60 Dyas After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated WS) Percent DM (Simulated C) Percent DM (Observed WS) Percent DM (Observed C) Figure 7 8. CROPGRO simulated (line) vs. observed (points) dry matter concentration of ind ividual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants.

PAGE 187

187 RMSE=0.9 d=0.97 RMSE=0.9 d=0.97 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Dry Matter Concentration (%) Percent DM (Simulated WS) Percent DM (Simulated C) Percent DM (Observed WS) Percent DM (Observed C) Figure 7 9. CROPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 2 in Gai nesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. RMSE=1 d=0.93 RMSE=1.2 d=0.94 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated WS) Percent DM (Simulated C) Percent DM (Observed WS) Percent DM (Observed C) Figure 7 10. CROPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants.

PAGE 188

188 RMSE=12 d=0.99 RMSE=17 d=0.99 0 50 100 150 200 250 300 0 10 20 30 40 50 60 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated WS) FruitFW,g/f (Simulated C) Fruit FW,g/f (Observed WS) Fruit FW,g/f (Observed C) Figure 7 11. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed toma to plants. RMSE=12 d=0.99 RMSE=16 d=0.99 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated WS) FruitFW,g/f (Simulated C) Fruit FW,g/f (Observed WS) Fruit FW,g/f (Observed C) Figure 7 12. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed plants.

PAGE 189

189 RMSE=15 d=0.82 RMSE=13 d=0.94 0 20 40 60 80 100 0 5 10 15 20 25 30 35 40 45 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated WS) FruitFW,g/f (Simulated C) Fruit FW,g/f (Observed WS) Fruit FW,g/f (Observed C) Figure 7 13. CROPGRO simulated (line) vs observed (points) fresh weight of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. RMSE=0.2 d=0.99 RMSE=0.98 d=0.97 0 2 4 6 8 10 12 0 20 40 60 Days After Tagging Fruit Diameter (cm) FD, cm (Simulated WS) FD, cm (Simulated C) FD cm (Measured WS) FD cm (Measured C) Figure 7 14. CROPGRO simulated (line) vs. observed (points) diameter of individual frui ts of cohort # 1 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants.

PAGE 190

190 RMSE=0.87 d=0.97 RMSE=0.77 d=0.93 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Diameter (cm) FD cm (Simulated WS) FD cm (Simulated C) FD cm (Measured WS) FD cm (Measured C) Figure 7 15. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed tomato plants. RMSE=0.87 d=0.62 RMSE=0.74 d=0.95 0 1 2 3 4 0 5 10 15 20 25 30 35 40 45 Days After Tagging Fruit Diameter (cm) FD, cm (Simulated WS) FD, cm (Simulated C) FD cm (Measured WS) FD cm (Measured C) Figure 7 16. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 3 in Gainesville Fl, during spring 2007 comparing well irrigated with water stressed t omato plants.

PAGE 191

191 RMSE=0.49 d=0.97 RMSE=0.15 d=0.98 0 1 2 3 4 5 0 20 40 60 80 100 Days After Planting Leaf Area Index (m2 m-2) LAI (Simulated NS) LAI (Simulated C) LAI (LAI Observed NS) LAI (LAI Observed C) F igure 7 17. CROPGRO simulated (lines) and measured (points) leaf area index in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants. RMSE=1199 d=0.89 RMSE=330 d=0.98 0 2000 4000 6000 8000 10000 12000 0 20 40 60 80 100 Days After Planting Total Above Biomass (kg ha-1) Tops wt kg/ha (Simulated NS) Tops wt kg/ha (Simulated C) Tops wt kg/ha (Observed NS) Tops wt kg/ha (Observed C) Figure 7 18. CROPGRO simulated (lines) and measured (points) above ground biomass in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants.

PAGE 192

192 RMSE=282 d=0.95 RMSE=178 d=0.99 0 1000 2000 3000 4000 5000 6000 0 20 40 60 80 100 Days After Planting Total Fruit Dry Weight (kg ha-1) Fruit wt kg/ha (Simulated NS) Fruit wt kg/ha (Observed C) Fruit wt kg/ha (Observed NS) Fruit wt kg/ha (Observed C) Figure 7 19. CROPGRO simulated (lines) and measured (points) total fruit dry weight in Gainesville, FL during spring of 2007 comparing w ell fertilized with N stressed tomato plants RMSE=9710 d=0.9 RMSE=15061 d=0.95 0 20000 40000 60000 80000 100000 120000 0 20 40 60 80 100 Days After Planting Total Fruit Fresh Weight (kg ha-1) Fruit FW k/ha (Simulated NS) Fruit FWt k/ha (Simulated C) Fruit FW k/ha (Observed NS) Fruit FW k/ha (Observed C) Figure 7 20. CROPGRO simulated (lines) and measured (points) total fruit fresh weight in Gainesville, FL during spring of 2007 comparing well fertilized with N stressed tomato plants

PAGE 193

193 RMSE=0.68 d=0.997 RMSE=0.56 b d=0.98 0 2 4 6 8 10 12 14 16 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated NS) Fruit DW,g/f (Simulated C) Fruit DW,g/f (Observed NS) Fruit DW,g/f (Observed C) Figure 7 21. CROPGRO simulated (line) vs. observed (points) dry weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants RMSE=1.9 d=0.83 RMSE=0.98 d=0.98 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Dry Weight (g fruit-1) Fruit DW,g/f (Simulated NS) Fruit DW,g/f (Simulated C) Fruit DW,g/f (Observed NS) Fruit DW,g/f (Observed C) Figure 7 22. CROPGRO simulated (line) vs. observed (points) dry weight o f individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants

PAGE 194

194 RMSE=0.3 d=0.996 RMSE=0.5 d=0.98 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated NS) Percent DM (Simulated C) Percent DM (Observed NS) Percent DM (Observed C) Figure 7 23. CROPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants. RMSE= 0.9 d=0.97 RMSE=1 d=0.95 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Dry Matter Concentration (%) Percent DM (Simulated NS) Percent DM (Simulated C) Percent DM (Observed NS) Percent DM (Observed C) Figure 7 24. CROPGRO simulated (line) vs. observed (points) dry matter concentration of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants

PAGE 195

195 RMSE=21 d=0.97 RMSE=13.8 d=0.996 0 50 100 150 200 250 300 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated NS) Fruit FW,g/f (Simulated C) Fruit FW,g/f (Observed NS) Fruit FW,g/f (Observed C) Figure 7 25. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato p lants. RMSE=44 d=83 RMSE=17 d=0.99 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fresh Weight (g fruit-1) Fruit FW,g/f (Simulated NS) Fruit FW,g/f (Simulated C) Fruit FW,g/f (Observed NS) Fruit FW,g/f (Observed C) Figure 7 26. CROPGRO simulated (line) vs. observed (points) fresh weight of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed tomato plants.

PAGE 196

196 RMSE=0.78 d= 0.99 RMSE=0.25 d=0.99 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 50 55 60 Days After Tagging Fruit Diameter (cm) FD cm (Simulated NS) FD cm (Simulated C) FD cm (Observed NS) FD cm (Observed C) Figure 7 27. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 1 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed plants RMSE=2 d=0.86 RMSE=077 d=0.93 0 2 4 6 8 10 0 5 10 15 20 25 30 35 40 45 50 Days After Tagging Fruit Diameter (cm) FD, cm (Simulated NS) FD, cm (Simulated C) FD cm (Observed NS) FD cm (Observed C) Figure 7 28. CROPGRO simulated (line) vs. observed (points) diameter of individual fruits of cohort # 2 in Gainesville Fl, during spring 2007 comparing well fertilized with N stressed plants

PAGE 197

197 CHAPTER 8 CONCLUSIONS One important feature of CROPGRO is the generic nature of the model that allows many applications around the world where model users work with different cultivars at different locations and during different seasons. That is possible because the main processes and functions are common for most the species, a fact that is implicit in the modular structure of CROPGRO. Therefore, CROPGRO simulat ions are good as long as the model is correctly calibrated according to genotypes, and soils and climates are accurately input The calibration processes are more easy and efficient if the genetics parameters of the model are the correct for the specie. CROPGRO -tomato was developed more than 10 years ago and until now has not been modified. We found literature, published after the model was developed, that suggests that the values of the cardinal temperature that control the plant phenology might need adjus tments. Therefore, in this work we updated the cardinal temperatures for tomato phenology (vegetative development, early reproductive growth and late reproductive growth). Cardinal temperatures for fruit set and pollination and fruit growth were updated as well. Most of the time the cardinal temperatures were reduced compared to the prior default version of the model by 2C t o 4C depending on the plant phe nological phase. We re -calibrated the model according to these updates using nine experimental data se t s. Overall the performance of the model for simulating total biomass and fruit dry weight yield was improved after the updates and re -calibration. Becaus e the new values of temperature coefficients came from experiments c onducted in environments with cont rolled temperature, we think that the cardinal temperatures for phenology, fruit setting and grow th of tomato should be updated in the current version of CROPGRO toma to. When tomato is produced for fresh market, the ability to control fruit fresh weight a nd size during the life cycle may have economic benefits because these are the variables that affect most

PAGE 198

198 the yield and price of the production. Crop models are valuable tools to understand the main factors causing variability on these attributes as well a s for predicting the harvest window. Often the later goal is the larger concern of growers who want to plan efficiently their crop according to when they will have the best chance to obtain better prices. However, most of models developed for tomato lack t he ability to predict fruit growth in terms of fresh mass and marketable size, and are limited to predictions of yi eld as total fruit dry weight. The most recent version of CROPGRO tomato (not yet released) includes a subroutine for dynamically simulating the fruit growth in terms of fresh weight, dry matter concentration, and fruit size distribution. The most important part of this work is that we re -calibrated this new feature of the model and successfully evaluated the quality of the simulations with ind ependent data. This is the first report of validation of single fruit fresh weight and size and we found that the model was able to simulate reasonably well the fruit growth in terms of dry weight and fresh weight, dry matter concentration and size o f indi vidual fruits initiated on different dates ( three fruit ages spaced one week apart ). In agreement with the observations and supported by literature the model revealed that the growth of the fruit is strongly related to the initiation date, with earlier se t fruits having higher sink strength than the later set fruits. The model error averaged for all cohorts and variables was less than 20 % and t he values of d index always was above 0.95, showing the suitability of the model for prediction over time. In order to assess whether stress conditions produce variations in the growth of individual fruits, we grew tomato plants under water and nitrogen stress conditions and evaluated the growth of single fruits over time. We found that both stresses caused differ ences in the fruit growth behavior. N stress affected mainly the dry matter accumulation while fresh weight and fruit size were quantitatively less affected. The water stress treatment, on other hand, produced

PAGE 199

1 99 smaller fruits that accumulated less fresh mas s than well irrigated fruits. Contrary to N results, the dry weight of fruits under water stress was less affected and because of the reduction in size and fresh mass, the dry matter concentration of these fruits increased. Leaf photosynthesis rate was red uced and the stomata l resistance was increased in plants under stress, thus contributing to the mass reduction. The partitioning of dry matter toward fruits was not affected under water stress but it was reduced under N stress. CROPGRO -tomato has several f unctions that allow simulating the effect of different stre sses at the whole plant level. After analyzing how t he growth of individual fruit was affected by water and N stress, the next step was to evaluate if the stress functions in CROPGRO -tomato model c ould predict the growth of individual fruits under stress. S tress factors such as SWFAC and TURFAC (for water stress) and NSTRES (for N stress), which are efficient to mimic stress at a whole plant level, were strategically connected to functions of fruit growth in the code in order to affect the timing (thermal time accumulator), the rate of single fruit growth as well as the dry matter concentration. We obtained a good response of the model in simulating water stress effects but more work need s to be done in order to make the model better predict N stress where simulated effects were stronger than we observed in our experiments.

PAGE 200

200 LIST OF REFERENCES Abreu P, Meneses JF, Gary C. 2000 Tompousse, a model of yield prediction for tomato crops: Calibration study for unheated plastic greenhouse. Acta Horticulturae 519, 141149. Acock B, Edwards CDA, Fitter DJ, Hand DW, Ludwing LJ, Warren WJ, Withers AC 1978. The contribution of leaves from different levels within a tomato crop to canopy net photosynthesis: An experimental examination of two canopy models. Journal of Experimental Botany 29, 815827. Acock B, Reynolds JF. 1989. The rationale for adopting a modular generic structure for crop simulators. Acta Horticulturae 284, 391396. Acock B. 1991. Modellin g canopy photosynthesis response to carbon dioxide, light interception, temperature and leaf traits. In : Boote KJ, Loomis RS, eds. Modeling c rop p hotosynthesis from biochemistry to c anopy American Society of Agronomy, Madison, WI, pp 41 55. Adams P. 1986 Mineral Nutrition. In: Atherton, JG, Rudich, J, eds. The tomato crop. A scientific basis for improvement London : Chapman & Hall, pp 281334. Adams SR, Valdez VM, Cave CRJ, Fenlon JS 2001a. The impact of changing light levels and fruit load on the patt ern of tomato yields. Journal of Horticultural Science and Biotechnology 76, 368373. Adams SR, Cockshull KE, Cave CRJ 2001b Effect of temperature on the growth and development of tomato fruits. Annals of Botany 88, 869 877. Adams SR, Valdez VM 2002. T he effect of periods of high temperature and manipulating fruit load on the p attern of tomato yields. Journal of Horticultural Science 77 461 466. Aikman DP. 1989. Potential increase in photosynthetic efficiency from the redistribution of solar radiation in a crop. Journal of Experimental Botany 40, 855864. Aikman DP 1996. A procedure for optimizing carbon dioxide enrichment of a glasshouse tomato crop Journal of Agricultural Engineering 63, 171 184. Akaki T, Eguchi T, Wajima T, Yoshida S, Kitano M 20 04. Dynamic analysis of growth, water balance and sap fluxes through phloem and xylem in a to m ato fruit: short term effect of water stress. Environment Control in Biology 42(3), 225240. A larcon JJ, Sanchez Blanco MJ, Bolarin MC, Torrecillas A 1994. Growt h and osmotic adju stment of two tomato cultivars during and af ter saline stress. Plant and Soil 166, 75 82. Alian A, Altman A, Heuer B 2000. Genotypic differences in salinity and water stress tolerance of fresh market tomato cultivars. Plant Science 152, 59 65.

PAGE 201

201 Atherton JG, Rudich J 1986. The tomato c ro p. A scientific basis for improvement London: Chapman and Hall. 221 234. Atherton JG, Harris GP 1986. Flowering. In : Atherton JG, Rudich J, eds. The tomato crop. A scientific basis for improvement New York : Chapman & Hall, pp 167200. Aung LH 1978. Temperature regulation of growth and development of tomato during ontogeny. In: Cowell R, ed. Proceedings of the 1st International Symposium on Tropical Tomato. Asian Vegetable Research and Development Cen ter Publication no. 7859, Shanhua, Taiwan, Republic of China, 79 93. Bailey BJ, Hunter A 1988. Plant response and energy use in five high thermal resistance greenhouses. Acta Horticuturae 229, 165171. Bangerth F, Ho LC 1984. Fruit position and fruit set sequence in a truss as factors determining final size of tomato fruits. Annals of Botany 53, 315319. Barker AV, Mills HA 1980. Ammonium and nitrate nutrition of horticultural crops. In: Janick J, ed. Horticultural Reviews Westport, CT ; AVI Publishin g, pp 395423. Bar -Tal A, Keinam M, Fishman S, Aloni B, Oserovitz Y, Genard M 1999. Simulation of environmental effects on Ca content in pepper fruit. Acta Horticulturae 507 253262. Benedictos PJr, Yavari N 2000. Optimum sowing date in relation to fl ower drop reduction in t omato. Acta Horticulturae 533, 351357. Berman ME, Dejong TM 1996. Water stress and crop load effects on fruit fresh and dry weight in peach ( Prunus persica ). Tree Physiology 16, 859864. Bernacchi D, Beck -Bunn T, Eshed Y, Lopez Y Petiard V, Uhlig J, Zamir D, Tanksley S 1998. Advances backcros s QTL analysis in t omato. Identification of QTLs for traits of agronomic importance from Lycopersicon hirsutum Theoretical and Applied Genetic 97, 381397. Bertin N, Gary C 1992. Tomato fruit set competition for assimilates, during the early production period. Acta Horticulturae 303, 121126. Bertin N 1993. Environnement climatique competition pour les assimilats et modlisation de la nouaison de la tomate en cul ture sous serre Thse ( Doctorat), Institut National Agronomique Paris Grignon, Paris. Bertin N, Heuvelink E 1994. Dry matter production in a tomato crop: comparison of two simulaton models. Journal of Horticul tural Science 68 9951011. Bertin N 1995. Competition for assimilat es and fruit position affect fruit set in indeterminate greenhouse tomato. Annals of Botany 75, 55 65.

PAGE 202

202 Bertin N, Gary C 1998. Short and long term fluctuations of the leaf mass per area of tomato plants. Implications for growth models. Annals of Botany 82, 71 81. Bertin N, Gary C, Tchamitchian M, Vassiere BE 1998. Influence of cultivar, fruit position and seed content on tomato fruit weight during crop cycle under low and high competition for assimilates. Journal of Horticultural Science and Biotechnology 73(4), 541 548. Bertin N, Buret M, Gary C 2001. Insights into formation of tomato quality during fruit development. Journal of Horticultural Science and Biotechnology 76 (6), 786 792. Bertin N, Genard M, Fishman S 2003. A model for an early stage of tom ato fruit development: cell multiplication and cessation of the cell proliferative activity. Annals of Botany 92, 67 72. Bertin N 2005. Analysis of the tomato fruit g row th response to temperature and plant fruit load in relation to cell division, cell e xp ansion an d DNA endoreduplication. Annals of Botany 95, 439447. Bertin N, Causse M, Brunel B, Tricon D, Genard M 2008. Identification of growth processes involved in QTLs for tomato fruit size and composition. Journal of Experimental Botany Advance Acces s published November 2008. 1 12. Bohner J, Bangerth F 1988. Effects of fruit set sequence and defoliation on cell number, cell size and hormone levels of tomato fruits ( Lycopersicon eculentum Mill.) within a truss. Journal of Plant Growth Regulation 7 14 1 155. Bohner J, Hedden P, Bora -Haber E, Bangerth F 1988. Identification and quantification of gibberellins in fruits of Lycopersicon esculentum and their relationship to fruit size in L. esculentum and L. pimpinellifolium Physiologia Plantarum 73, 348353. Boote KJ, Pickering N B. 1994. Modeling photosynthesis of row crop canopies. Hortscience 29, 14231434. Boote KJ, Jones JW, Hoogenboom G, Pickering NB 1998. The CROPGRO model for grain legumes. In: Tsuji GY, Hoogenboom G, Thornton PK, eds. Understan ding Options for Agricultural Production. Dordrecht: Kluwer Academic Publishers pp 99 128. Boote KJ, Jones JW, Hoogenboom G, Batchelor WD, Porter CH 2000. CROPGRO plant growth and partitioning module. In DSSAT v. 4.0. Documentation and source code listin g. Research Report Number 2000-1204. University of Florida, Gainesville Fl Boote, KJ, Minguez MI, Sau F 2002. Adapting the CROPGRO legume model to simulate growth of faba bean. Agronomy Journal 94, 743 756. Boote, KJ, Scholberg JMS 2006. Developing, pa rameterizing and testing of dynamic crop growth models for horticultural crops. Acta Horticu l turae 718, 23 34.

PAGE 203

203 Bournival BL, Vallejos CE Scott JW 1990. Genetic analysis of resistances to races 1 and 2 of Fusarium oxisporum f. sp. Lycopersici from the wi ld tomato Lycopersicon pennellii Theoretical and Applied Genetic 79, 641 645. Bremer JM, Mulvaney CS 1982. Nitrogentotal. In: Black ed. Methods of soil analysis (2nd ed.), Agronomy Monograph vol. 9 Part 2. Madison, WI, pp. 595 622. Bruchou C, Genard M 1999. A space time model of carbon translocation along a shoot bearing fruits. Annals of Botany 84 565 576. Bruggink GT, Heuvelink E 1987. Influence of light on the growth of young tomato, cucumber and sweet pepper plants in the greenhouse: effects on relative growth rate, net assimilation rate and leaf area ratio. Scientia Horticulturae 31, 161174. Bussieres P 1993. Potential dry matter and water i mpor t rates in the tomato f ruit i n relation to fruit size. Annals of Botany 72, 6372. Bussieres P 1994. Water import rate in tomato fruit: A model of resistance. Annals of Botany 73, 75-82. Buster TP 1979. Soil survey of Marion County, Florida. Soil Conservation Service Washington, D.C. 1979. Calado AM, Portas CAM 1980. Growth and development of toma to Cal j. Acta Horticulturae 100, 159 168. Calado AM, Portas CAM 1987. Base temperature and date of planting in processing tomatoes. Acta Horticulturae 200, 185193. Calvert A 1957. Effect of the early environment on the development of flowering in t he tomato. I. Temperature. Journal of Horticultural Science 32 9 17. Castrillo M, Fernandez D, Calcagno AM, Trujillo I, Guenni L 2001. R esponse of Ribulose 1,5 bisphosphate carboxylase, protein content, and s tomatal conductance to water deficit in maize tomato, and bean. Photosynthetica 39 (2), 221226. Carlisle VW Sodek F III, Collins ME, Hammond LC Harris WG 1988. Characterization data for s elected Florida s oils University of Florida Institute of Food and Agricultural Sciences, Soil Science Rese arch Report 88-1 Gainesville, FL. 28249. Causse M, Saliba Colombani V, Lecomte L, Duffe P, Rouselle P, Buret M 2002. QTL analysis of fruit quality in fresh market tomato: a few chromosome regions control the variation of sensory and instrumental traits. Journal of Experimental Botany 377 (53), 20892098. Cleary B, Laerr J, Hamel J. 1996. Gu i delines for measuring plant moisture stress with a pressure chamber. PMS Instrument Co. 480 SW, Airport Avenue Corvallis, Albany OR, USA.

PAGE 204

204 Cockshull KE. 1988. The inte gration of plant physiology with physical changes in the greenhouse climate. Acta Horticulturae 229, 113123. Cockshull KE, Ho LC. 1995. Regulation of tomato fruit size by plant density and truss thinning. Journal of Horticultural Science 70, 395407. Co ckshull KE, Graves CJ, Cave CRJ. 1992. The influence of shading on yield of glasshouse tomatoes. Journal of Horticultural Science 67, 11 24. Coombe B 1976. The develo pment of fleshy fruits. Annual Review of Plant Physiology 27, 2072 28. Cuartero J, Ferna ndez Munoz R 1999. Tomato and salinity. Scientia Horticulturae 78, 83 125. Challa H, Heuvelink E, Van Meeteren U 1995. Crop growth and development In: Bakker JC, Le Bot GPA, Challa H,Van de Braak NJ, eds. Greenhouse climate control an i ntegral a pproach. Wageningen: Wageningen Press, pp 62 84. Challa H, Bakker J. 1998. Potential production within the gr eenhouse environment. In: Enoch Z, Stanhill G, eds. Ecosystems of the world. T he greenhouse ecosystem Amsterdan: Elsevier, pp 333348. Chamarro JC 1995. Anatomia y fisiologia de la planta. In: Nuez F, ed El c ultivo del Tomate Madrid: Mundi Prensa pp 201239. Charles WB, Harris, RE 1972. Tomato fruit -set at high and low temperatures. Canadian Journal of Plant Science 52, 497506. Davies JN, Hobson GE 1981. The constituents of tomato fruits the influence of the environment, nutrition and genotype. CRC Critical Reviews in Food Science and Nutrition 15, 205280. De Koning ANM 1989a Development and growth of a commercially grown tomato crop. Acta Horticulturae 260, 267 270. De Koning ANM 1989b. The effect of the temperature on fruit growth and fruit load of tomato. Acta Horticulturae 248, 329336. De Koning ANM 1990. Long term temperature integration of tomato. Growth and development under alt ernating temperature regimes. Scientia Horticulturae 45 117127. De Koning ANM, De Ruiter HW 1991. Effect of temperature, plant density and fruit thinning on flower/fruit abortion and dry matter partitioning of tomato. Annual Report 1990. Glasshouse Crop s Research Station, Naaldwijk, The Netherlands, 29. De Koning ANM 1993. Growth of tomato crop: measurements for crop model validation. Acta Horticulturae 328, 141 146.

PAGE 205

205 De Koning ANM 1994. Development and dry matter distribution in glasshouse tomato: a quantitative approach. Doctoral dissertation Wageningen Agricultural University, Wageningen, The Netherlands. D e Koning ANM 1996. Quantifying the responses to temperature of different plant processes involved in growth and development of glasshouse tom ato. Acta Horticulturae 406, 99 104. De Koning ANM 2000. The effect of temperature, fruit load and salinity on development rate of tomato fruit. Acta Horticulturae 519, 85 94. Demetriades -Sha TH, Fuchs M, Kanemasu ET, Filocroft I 1992. A note of caution concerning the relation between cumulated intercepted solar radiation and crop growth. Agricultural Forestry Meteorology 58, 193207. Dewar RC. 1996. The correlation between plant growth and intercepted radiation: An interpretation in terms of opt imal plant nitrogen content. Annals of Botany 78, 125136. Dieleman JA. Heuvelink E. 1992. Factors affecting the number of leaves preceding the first inflorescence in the tomato. Journal of Horticultural Science 67 1 10. Dinar M, Stevens MA 1981. The rel ationship between starch accumulation and soluble solids content in tomato fruits. Journal of the American Society for Horticultural Science 106 415418. Dorais M, Demers DA, Papadopoulos A, Ieperen, W van 2001 In : Janick J. ed. Greenhouse tomato fruit c uticle cracking. Horticultural Reviews 30, 163168. New York : Wiley and Sons, Inc. Dukes MD, Scholberg JM, Hanselman TA 2005. Quantification of nitrogen movment under drip irrigated vegetable production. Transactions of American Society of Agricultural Engineers (ASAE ) paper number 052240, Anual International Meeting Tampa Fl, 17 20 July 2005. Dumas Y. 1990. Tomatoes for processing in 90s: nutrition and crop fertilization. Acta Horticulturae 227, 115 266. Duvick DN 1986. Plant breeding past achievem ents and expectations for the future. Economic Botany 40(3), 289297. Ehret DL, Ho LC. 1986 Effect of salinity on dry matter partitioning and growth in tomato grown in nutrient film culture. Journal of Horticultural Science 61, 361 367. Ehret DL, Ho LC 1986. Effects of osmotic potential in nutrient solution on diurnal growth of tomato fruit. Journal of Experimental Botany 37, 12941302. Enriquez LJ, Mercado Silva E, Castano Tostado E 2005. Effect of two greenhouse covers on growth, dry matter content and tomato fruit quality. Acta Horticulturae 697, 481484.

PAGE 206

206 Erdal I, Ertek A, Sengygit U, Yilmaz HI 2006. Effects of different irrigation programs and nitrogen levels on nitrogen concentration, uptake and utilization of in proccesing tomatoes (Ly copersi con esculentum ). Australian Journal of Experimental Agriculture 46, 16531660. Erickson RO 1976. Modeling of plant growth. Annual Review of Plant Physiology 27, 407434. Eshed Y, Zamir D 1995. An introgression line population of Lycopersicon pennellii in the cultivated tomato enables the identification and fine mapping of yield associated QTL. Genetics 141(3), 11471162. Evans JR 1989. Photosynthesis and nitrogen relationships in leaves of C3 plants. Oecologia 78, 9 19. Farrar J F 1988. Temperature and the partitioning and translocation of carbon. In: Long SP, Woodward FI, eds. Plants and t emperature 42th Symposium of the society for Experimental Biology. Cambridge: Company of Biologists pp 203 235. Fernan dez Munoz R, Cuartero J 1991. Effects of temperature and irradiance on stigma exertion, ovule viability and embryo development in tomato. Journal of Horticultural Science 66 395401. Ferreira ME, Abreu JP, De Bianco VV, Monteiro A 1997. Predicting phasic development of green beans for proce ssing using a model with high temperature reduction of thermal time accumulation. Scientia Horticulturae 69, 123133. Fishman S, Genard M 1998. A biophysical model of fruit growth: simulation of seasonal and diurnal dynamics of mass. Plant cell and Envir onment 21, 739752. Fleisher DH, Logendra LS, Moraru C, Both AJ, Cavazzoni J, Gianfagna T, Lee T, Janes HW 2006. Effect of temperature perturbations on tomato (Lycopersicon esculentum Mill.) quality and production scheduling. Journal of Horticultural Sc ience and Biotechnology 81(1), 125131. Folker E. 1979. Commercial Production. In Folker E, ed. The tomato. Study of the plant and the commercial production. Buenos Aire s, Argentina: Hemisferio Sur, pp 3790. Foolad MR. 2005. Breeding for abiotic stress tolerances in tomato. In: Ashraf M, Harris PJC, eds. Abiotic Stresses: Plant Resistance through Breeding and Molecular Approaches New York: The Haworth Press 613684. Foolad MR 2007. Genome mapping and molecular breeding of tomato. Int ernational Journal of Plant Genomics Vol 2007, 1 52. Frary A, Nesbitt TC, Grandillo S, Van der Knaap E, Cong B, Liu J, Meller J, Elber R, Alpert KB 2000. fw 2.2: a quantitative trait locus key to the evolution of tomato fruit size. Science 289 8588.

PAGE 207

207 Fridman E, Pleban T, Zamir D 2000. A recombination hotspot delimits a wild species quantitative trait locus for tomato sugar content to 484 bp within an invertase gene. Proceeding of the National Academy of Sciences of the United States of America 97(9), 47184723. Fujit a K, Lei K, Saneoka H, Moghaieb REA, Mohapatra PK 2004. Effect of N -deficiency on photoassimilate partitioning and rhytmic changes in fruit and stem diameter of tomato (Lycopersicon esculentum ) during fruit growth. Proceedings of the 4th International C rop Science Co ngress. Brisbane, Australia, Sep 26Oct 1 2004. Garcia DM, Petzall CK, Castrillo M 2004. R espuestas al deficit hidrico en las variedades de tomate Rio Grande y Pera Quibor. Agronomy Tropical 54 (4), 1 35. Gardner FP 1985. Photosynthesis. In: Gardner FP, Pearce RB Mitchell RL, eds Physiology of c rop p lants Iowa State University Press Ames pp 3 29. Garnier E, Berger A 1986. Effects of water stress on stem diameter changes of peach growing in the field. Journal of Applied Ecology 23, 193199. Gary C, Baille A, Navarrete M, Epanet R. 1997. TOMPOUSSE, un modele simplifie de prevision du rendement et du calib er de la tomate. In: Baille A, ed Actes du seminaire de IAIP intersectoriel le Serres, INRA, Avignon, pp 100109. Gary C, Jon es JW, Longuenesse JJ 1993. Modelling daily changes in specific leaf area of tomato: the contribution of the leaf assimilate pool. Acta Horticulturae 328, 205210. Gary C, Tchamitchian M, Bertin N, Boulard T, Baille A, Charasse L, Rebillard A, Cardi JP, Marcelis LFM 1998a SIMULSERRE : education software simulating the greenhouse -crop systems. Acta Horticulturae 456, 451458. Gary C, Jones JW, Tchamitchian M 1998b. Crop modelling in horticulture: state of the art. Scientia Horticulturae 74, 3 20. Gary C, Barezi JF, Bertin N, Tchamitchian M 1995. Simulation of individual organ growth and development on a tomato plant: a model a user friendly interface. Acta Horticulturae 399, 199205. Genard M, Huguet JG 1996. Modeling the response of peach fruit to wa ter stress. Tree Physiology 16, 407415. Georgelis N, Scott JW, Baldwin EA 2004. Relationship of tomato sugar concentration with physical and chemical traits and linkage of RAPD markers. Journal of the American Society for Horticultural Science 129 (6), 8 39 845. Gijsman AJ, Hoogenboom G, Parton WJ, Kerridge PC 2002. Modifying DSSAT crop models for low input agricultural systems using a soil organic matter residue module from CENTURY. Agronomy Journal 94, 462 474.

PAGE 208

208 Gijzen H, Goudriaan J 1989. A flexible and explanatory model of light distribution and photosynthesis in row crops. Agricultural Forestry Meteorology 48, 1 20. Gijzen H. 1992. Simulation of photosynthesis and dry matter production of greenhouse crops. Simultion Reports CABO TT 28, Wageningen, The Netherlands. Gijzen H. 1995. Interaction between CO2 and water loss. In: Baker JC, Bot GPA, Challa H, Van de Braak NJ, eds. Greenhouse climate control, an integrated approach The N etherlands: Wageningen Press, pp 51 62. Gijzen H, Heuvelink E, Challa H, Marcelis LFM, Dayan E, Cohen S, Fuchs M 1998. Hortsim: A model for greenhouse crops and greenhouse climate. Acta Horticulturae 456 441450 Gillaspy G, Ben -David H, Gruissem W 1993. Fruits : A developmenta l perspective. The Plant Cell 5 14391451. Gim enez C, Mitchell VJ, Lawlor DW 1992. Regulation of photosynthetic rate of two sunflower hybrids under water stress. Plant Physiology 98, 516524. Goudriaan J, Monteith JL 1990. A mathematical function for crop growth based on light interception and leaf area expansion. Annals of Botany 66, 695701. Grange RI, Andrews J 1993. Growth rates of glasshouse tomato fruit in relation to final size. Journal of Horticultural Science 68, 747754. Grandillo S, Tanksley SD 1996. QTL analysis of horticultural tr aits differentiating the cultivated tomato from the closely related species Licopersicon pimpinellifolium Theoretical and Applied Genetic 92(8), 935 951. Grandillo S, Ku HM, Tanksley SD 1999. Identifying the loci responsible for natural variation in fr uit size and shape in tomato. Theoretical and Applied Genetic 99, 978987. Grierson D, Kader AA 1986. Fruit ripening and quality. In: Atherton JG, Rudich J, eds. The Tomato Crop: A Scientific Basis for improvment London: Chapman & Hall, pp 241280. G uichard S, Gary C, Longuenesse JJ, Leonardi C. 1999. Water fluxes and growth of greenhouse toomato fruits under summer conditions. Acta Horticulturae 507, 223230. Guichard S, Bertin N, Leonardi C, Gary C 2001. Tomato fruit quality in relation to water a nd carbon fluxes. Agronomie 21, 385392. Guichard S, Gary C, Leonardi C, Bertin N 2005. Analysis of growth and water relations of tomato fruits in relation to air vapor pressure deficit and plant fruit load. Journal of Plant Growth Regulation 24, 201213. Grindlay DJC 1997. Towards an explanation of crop nitrogen demand based on the optimization of l eaf nitrogen per unit leaf area. Journal Agricultural Science 128, 377396.

PAGE 209

209 Gustafson FG 1 926. Growth studies on fruits. Plant Physiology 1 (3), 265-172. Gus tafson FG 1927. Growth studies on fruits an explanation of the shape of the growth curve. Plant Physiology 2 (2), 153161. Garcia DM, Petzall CK, Castrillo M 2004. Respuestas al deficit hidrico en las variedades de tomate Rio Grande y Pera Quibor. Agronomy Tropical 54 (4), 1 35. Haeder HE, Mengel K 1972. Translocation and respiration of assimilates in tomato plants as influenced by K nutrition. Z Pflanz. Boden 131, 139 147. Harper LA, Pallas JE Jr., Bruce RR. Jones JB Jr. 1979. Greenhouse microclimate fo r tomatoes in the southeast. Journal of the American Society for Horticultural Science 104 659663. Hartman PL, Mills HA, Jones JB Jr 1986. The influence of nitrate: ammonium ratios on growth, fruit development, and element concentration in Floradel to mato plants. J ournal of the American Society for Horticultural Science 111 (4), 487490. He C, Zhang Z 2006. Modeling the relationship between tomato fruit growth and the effective accumulated temperature in solar greenhouse. Acta Horticulturae 718, 5815 87. Hebbar SS, Ramachandrappa BK, Nanjappa HV, Prabhakar M 2004. Studies of drip fertigation in field grown tomato ( Lycopersicon esculentum Mill.). European Journal of Agronomy 21, 117 127. Heuvelink, E, Marcelis, LFM 1989. Dry matter distribution in tom ato and cucumber. Acta Horticulturae 260, 149 157. Heuvelink E, Bertin N 1994. Dry Matter partitioning in a tomato crop: comparison of two simulation models. Journal of Horticultural Science 69, 885 903. Heuvelink E 1995. Growth, development and yield of a tomato crop: periodic destructive measurements in a greenhouse. Scientia Horticulturae 61, 77 99. Heuvelink E, Buiskool RPM 1995. Influence of source -sink interaction on dry matter production in tomato. Annals of Botany 75, 381389. Heuvelink E 1996. Tomato growth and yield: quantitative analysis and synthesis Doctoral d issertation, Wageningen Agricultural University. The Netherlands Heuvelink, E 1997. Effect of fruit load on dry matter partitioning in tomato. Scientia Horticulturae 69, 51 59. Heu velink, E 1999. Evaluation of a dynamic simulation model for tomato crop growth and development. Annals of Botany 83, 413422.

PAGE 210

210 Heuvelink E, Bakker M, Stanghellini C 2003. Salinity effects on fruit yield in vegetable crops: a simulation study. Acta Hortic ulturae 609, 133140. Heuvelink E, Tijskens P, Kang MZ 2004. Modelling product quality in horticulture: an o verview. Acta Horticulturae 654, 1925. Heuvelink E. 2005. Developmental processes. In: Heuvelink E ed. Crop p roduction science in horticulture, t omatoes. Cambridge: CABI Publishing, pp 53 83. Ho LC, Sjut V, Hoad GV 1982. The effect of assimilate supply on fruit growth and hormone levels in tomato plants. Journal of Plant Growth Regulation 1 155171. Ho LC, Shaw AW, Hammond JBW, Burton KS. 1983 Source sink relationships and carbon metabolism in tomato leaves. Annals of Botany 52, 365372. Ho LC, Sjut V, Hoad GV 1983. The effect of assimilate supply on fruit growth and hormone levels in tomato plants. Journal of Plant Growth Regulation 1 155171. Ho LC 1984. Partitioning of assimilates in fruiting tomato plants. Journal of Plant Growth Regulation 2 275277. Ho LC, Hewitt JD 1986. Fruit development. In: Atherton JG, Rudich J, eds. The tomato crop. a scientific b asis for i mprovement. London: C hapman and Hall, pp 201239. Ho LC, Grange RI, Picken AJ 1987. An analysis of the accumulation of water and dry matter in tomato fruit Plant, Cell and Environment 10, 157162. Ho LC 1988. Metabolism and compartmentation of imported sugars in sink organs in relation to sink strength. Ann ual Review of Plant Physiology a nd Molecular Biology 39 355378. Ho LC, Grange RI, Shaw AF 1989. Source/sink regulation. In: Baker DA, Milburn J, eds. Transport of photoassimilate London pp 306343. Ho LC 1992. Fruit growth and sink strength. In: Marshall, C, Grace J, eds. Fruit and seeds production. Aspects of development, environmental physiology and ecology. Cambridge University Press. SEB seminar series 47, 121124. Ho L C 1996. Tomato. In: Zamski E, Schaffer A, e ds. Photoassimilate distribution in plants and c rops. Source -sink r elationships New York: Dekker M Inc, pp 709728. Hochmuth GJ 1994. Efficiency ranges for nitrate -nitrogen and potassium for vegetable petiole sap quick test. HorTechnology 4 218222. Hou ghtaling HB 1935. A developmental analysis of size and shape in tomato fruits. Bull. Torrey Botany Club 62, 243252. Huett DO 1986. Response to nitrogen and potassium of tomatoes grown in sand culture. Australian Journal of Experimental Agriculture 26, 133138.

PAGE 211

211 Huett DO, Dettmann EB. 1988. Effect of nitrogen on growth, fruit quality and nutrient uptake of tomatoes grown in sand culture. Journal of Experimental Agriculture 28, 391399. Huett DO, Dettmann EB. 1991. Nitrogen response surface models of zucch ini squash, haed lettuce and potato. Plant and Soil 134, 243254. Hunt R 1982. Plant Growth Curves: the functional approach to plant growth analysis. London: Edward Arnold ed. 248 pp. Hurd RG 1973. Long -day effects on growth and flower initiation of toma to plants in low light. Annals of Applied Biology 73, 2218. Hurd RG, Gay AP, Mountifield AC 1979. The effect of partial flower removal on the relation between root, shoot and fruit growth in the indeterminate tomato. Annals of Applied Biology 93, 7789. Hurd R G, Thornley JHM 1974. An analysis of the growth of young tomato plants in water culture at different light integrals and CO2 concentrations. I. Physiological aspects. Annals of Botany 38, 375388. Hurd RG, Graves CJ 1984. The influence of differen t temperature patterns having the same integral in the earliness and yield of tomato. Acta Horticulturae 148, 547554. Hussey G 1963a. Growth and development in the young tomato. I. The effect of temperature and light intensity on growth of the shoot apex and leaf primordia. Journal of Experimental Botany 14, 31625. Infostat. 2002. Infostat version 1.1 Users Guide. Nacional University of Cordoba, Cordoba A rgentina. Jamieson PD, Porte JR, Wilson DR 1991. A test of the computer simulation model ARC WHEAT 1 on wheat crops grown in New Zeeland Field Crop Research 27, 337350. Janes HW, McAvoy RJ. 1991. Environmental control of a single cluster green house tomato crop. HortTechnology 1 110114. Johnson R, Dixon MA, Lee DR 1992. Water relations of the toma to during fruit gro wth. Plant, Cell and Environment 15, 947953. Jones JB Jr 1999. Tomato Plant Culture. In : Jones JB Jr, ed. Tomato plant c ulture: in the field, greenhouse and home garden. Boca Raton Forida: CRC Press, pp 5172. Jones HG, Tardieu F 1998. Modelling water relations of horticultural crops: a review. Scientia Horticulturae 74, 21 46. Jones JW, Dayan E, Allen LH, Van Keulen H, Challa H 1991. A dynamic tomat o growth and yield model (TOMGRO ). Transactions of American Society of Agricultural Engineers (ASAE) 34, 663672.

PAGE 212

212 Jones JW, Hoogenboom G, Porter CH, Boote KJ, Batchelor WD, Hunt LA, Wilkens PW, Singh U, Gijsman AJ. Ritchie JT 2003. The DSSAT cropping system model. European Journal of Agronomy 18, 2 3 5 265. Khan AA, Sagar GR 1969. Alter ation of the pattern of distribution of photosynthetic products in the tomato by manipulation of the plant. Annals of Botany 33, 753752. Kawabata S, Sasaki H, Sakiyama R 2005. Role of transpiration from fruits in phloem transport and fruit growth in toma to fruits. Physiologia Plantarum 124, 371380. Khamis A, Ismail Z, Haron K, Mohamed AT 2005. Nonlinear growth models for modeling oil palm yield growth. Journal of Mathematics and Statistics 1(3 ), 225233. Khayat E, Ravad D, Zieslin N 1985. The effects of various night -temperature regimes on vegetative growth and fruit production of tomato plants. Scientia Horticulturae 27, 9 13. Kinet JM, Hurdebise D, Parmentier A Stainier R 1978. Promotion of inflorescence development by growth substance treatment t o tomato plant growth in insufficient light conditions. Journal of the American Society for Horticultural Science 103, 724 727. Kinet JM 1987. Inflorescence development in tomato. Control by light, growth regulators and apical dominance. Plant Physiology (Life Science Advance) 6 121127. Kinet JM, Zime V, Linotte A, Jaacqmond BG 1986. Resumption of cellular activity induced by cytokinin and gibberellin in treatments in tomato flowers targeted for abortion in unfavorable light conditions. Physiologia Pla nt arum 64, 67 73. Lohar DP, Peat WE 1998. Floral characteristics of heat tolerant and heat -sensitive tomato (Lycopersicon esculentum Mill.) cultivars at high temperature. Scientia Horticulturae 73, 5360. Lakso AN, Corelli Grappadelli L, Barbad J, Goffin et MC 1995. An expolinear model of fruit growth pattern of the apple fruit. Journal of Horticultural Science 70, 389 394. Langton FA, Cockshull KE 1997. Is stem extension determined by DIF or by absolute day and night temperatures? Scientia Horticutura e 69, 229237. Larouche R, Gosselin A, Vezna LP 1989. Nitrogen concentration and photosynthetic photon flux in greenhouse tomato production: I. Growth and development Journal of American Society for Horticulture Science 114 458461. Lawlor DW 1995. The effects of water deficit on photosynthesis. In: Smirnoff N, ed. Environment and plant metabolism. Flexibility and acclimation Oxford: pp 129160 Lawlor DW 2002. Carbon and nitrogen assimilation in relation to yield: mechanisms are the key to underst anding production systems Journal of Experimental Botany 53 773787.

PAGE 213

213 Le Bot J, Andamowicz S, Robin P 1998. Modeling plant nutrition of horticultural crops: a review. Scientia Horticulturae 74, 47 82. Lee DR, Dixon MA, Johnson RW 1989. Simultaneous meas urements of tomato fruit and stem water potential using in situ stem hygrometers. Canadian Journal of Botany 6 7 23522355. Lee DR 1990. A unidirectional water flux model of fruit growth. Canadian Journal of Botany 68, 12861290. Leonardi C, Baille A, Guichard S. 2000. Predicting transpiration of shaded and non shaded tomato fruits under greenhouse environments. Scientia Horticulturae 84, 2 97307. Lescurret F, Genard M 2005. A virtual peach fruit model simulating changes in fruit quality during the fin al stage of fruit growth. Tree Physiology 25, 13031315. Li SH, Huguet JG, Schoch PG, Orlando P 1989. Response of peach tree growth and cropping to soil water deficit at various phenological stages of fruit development. Journal of Horticultural Science 64, 541 552. Liu J, Cong B, Tanksley SD 2003. Generation and analysis of an artificial gene dosage series in tomato to study the mechanisms by which the cloned quantitative trait locus fw2.2 controls fruit size. Plant Physiology 132, 292299. Liu HF, Gen ard M, Guichard S, Bertin N 2007. Model assisted analysis of tomato fruit growth in relation to carbon and water fluxes. Journal of Experimental Botany 58 (13), 35673580. Longuenesse JJ, Gary C, Tchamitchian M 1993. Modeling CO2 exchanges of greenhouse crops: a matter of scales and boundaries. Acta Horticulturae 328, 33 47. Lorenz OA, Tyler KB 1983. Plant tissue analysis of vegetable crops. In: Reisenauer HM, ed. Soil and plant tissue testing in California. Div of Agr. Sci., Univ. o f California. Berkely: Bul 1879, 2124. Maher MJ. 1976. Growth and nutrient content of a glasshouse tomato crop grown in peat. Scientia Horticulturae 4 23 26. MacArthur JW, 1934 Linkage groups in the tomato. Journal of Genetics 29, 123133. MacArthur JW, Butler L 1938. Size inheritance and geometric growth processes in the tomato fruit. Genetics 23, 253268. Mapelli S, Froba C, Torti G, Zoressi G 1978. Relationship between set, development and activities of growth regulators in tomato fruits. Plant Cell Physiology 19, 12811288. Marcelis LFM 1992. The dynamics of growth and dry matter distribution in cucumber. Annals of Botany 69, 487492.

PAGE 214

214 Marcelis LFM 1993. Fruit g rowth and biomass alloc ation to the fruits in cucumber 1.Effect of fruit load and temperature. Sc ientia Horticulturae 54, 107121. Marcelis LFM. Heuvelink E, Goudriaan J 1998. Modelling biomass production and yield of horticultural crops: a review, Scientia Horticulturae. 74, 83111 Marcelis LFM, Gijzen H 1998. Evaluation under commercial conditions of a model of prediction of yield and quality of cucumber fruits. Scientia Horticulturae 76 171181. Menary RC, van Staden J 1976. Effect of phospho rus nutrition and cytokinins on flowering in t he tomato Lycopersicon esculentum Mill. Australian Journal of Plant Physiology 3 201203. Mills HA, Jones JB Jr 1996. Plant analysis h andbook II. Athens, GA: MicroMacro Publ. Inc., 422 p p Mitchell JP, Shennan C, Grattan SR 1991. Developmental changes in tomato fruit composition in response to water defi cit and salinity. Physiologia Plantarum 83, 117185. Monselise SP, Varga A, Bruinsma J 1978. Gr owth analysis of tomato fruits, Lycopersicon esculentum Mill. Annals of Botany 42, 12451247. Monteiro AA 1983. Tomato fruit growth in relation to methods of improving fruit setting. Acta Horticulturae 137, 307 314. Monteith JL 1977. Climate and the efficiency of crop production in Britain. London: Phil. Trans. Royal Soc., 281, 277294. Monteith JL 1994. Validity of the correlation between intercepted radiation and biomass. Agriculture and Forestry Meteorology 68, 213220. Moore EL, Thomas W 1952. Some effects of shading and parachloro-phenoxy acetic acid on fruitfulness of tomatoes. Proc. American Society Horticultural Science 60, 289 294. Mossler M, Aerts MJ, Nesheim N 2005. Florida crops. Pest management profile s. Tomatoes. University of Florida IFAS Extension CIR 1238, 1 19. Nesbitt TC, Tanksley SD 2001. fw 2.2 directly affects the size of developing tomato fruit, with secondary effects on fr uit number and photosynthate distribution. Plant Physiology 127, 575583. Nitsch JP 1953. The physiology of fruit growth. Annual Review Plant Physiology and Molecular Biology 4 199236. Nuez F 1995. Cultivo del tomate. Nuez F, ed, Madrid, Spain: Mu ndi -prensa, 796 pp.

PAGE 215

215 Nyabundi JO, Hsiao TC 1989. Effects of water stress on growth and yield of field grown tomatoes. I. Canopy development and biomass accumulation. East Africa Agriculture and Forestry Journal 55, 17 26. Opara L 1999. Fruit growth mea surement and a nalysis In: Janick J, ed. Horticultural Reviews 24, 480pp. Ortega Farias S, Ben -Hur L, Valdez H, Paillan H 2003. Effects of f our water applications on yield and quality greenhouse tomatoes produced in spring summer. Technical Agriculture 63 (4 ) 36 42. Papadopoulos AP 1985. Plant spacing effects on photosynthetically active radiation interception and utilization by greenhouse tomatoes. Doctoral dissertation University of Guelph, Ontario, Canada, 428 pp. Papadopoulos AP, Ormrod DP 1988a. Plant spacing effects on light interception by greenhouse tomatoes. Canadian Journal of Plant Science 68, 11971208. Papadopoulos AP, Ormrod, DP 1988b. Plant spacing effects on photosynthesis and transpiration of the greenhouse tomato. Canadian Journal of Plant Science 68, 12091218. Papadopoulos A P, Ormrod DP 1990. Plant spacing effects on yield of the greenhouse tomato. Canadian Journal of Plant Science 70, 565573. Papadopoulos AP, Ormrod DP 1991. Plant spacing effects on growth and development of the greenhouse tomato. Canadian Journal of Plant Science 71, 297304. Papadopoulos AP, Parajasingham S. 1997. The influence of plant spacing on light interception and use in greenhouse tomato ( Lycopersicon esculentum Mill.). A review. Scientia Horticulturae 69, 1 29. Pearce BD, Grange RI, Hardwick K 1993. The growth of young tomato fruit. I. Effects of temperature and irradiance on fruit grown in controlled environment. Journal of Horticultural Science 68 1 11. Perez Alfoncea F, Estan MP, Caro M, Gue rrier G 1993. Osmotic adjustment in Lycopersicon esculentum and L. penellii under NaCl and polyethylengly col 6000 isoosmotic stresses. Physiologia Plantarum 87, 493498. Perry KB, Wu Y, Sanders DC, Garrett JT, Decoteau DR, Nagata RT, Dufault RJ, Batal KD Granberry DM, Mclaurin WJ 1997. Heat units to predict tomato harvest in southeast USA. Agriculture and Forest Meteorology 84, 249254. Picken AJF. 1984. A review of pollination and fruit set in the tomato ( Lycopersicon esculentum Mill.) Journal of Horti cultural Science 59, 1 13.

PAGE 216

216 Picken AJF, Hurd RG, Vince -Prue D 1985. Lycopersicon esculentum. In: Hale vy AH, ed, CRC Handbook of f lowering Vol. 3 Boca Raton, FL: CRC Press pp 330 346. Picken AJF, Stewart K Klapwijk D 1986. Germination and vegetati ve development. In: Atherton JG, Rudich J, eds, The tomato crop. A scientific basis for i mprovement. London: Chapman and Hall pp 111 166. Pill WG, Lambeth VN 1977. Effects of NH+ and NO3 nutrition with an d without pH adjustment on tomato growth, ion composition and water relations. Jou rnal of the American Society for Horticultural Science 102 78 81. Pill WG, Lambeth VN, Hinckley TM 1978. Effect of N form and level on ion concentration, water stress and blossom end rot incidence in tomato. Jou rnal of the American Society for Horticultural Science 10 3 265283. Popper KR 1959. The logic of the scientific discovery. New York: Basic Books, Inc, 480 pp. Pressman E, Peet MM, Pharr DM 2002. The effect of heat stress on tomato pollen characteristics is associated with changes in carbohydrate concentration in the developing anthers. Annals of Botany 5 631636. Rhaman SM, Nawata E, Sakuratani T 1999. Effect of water stress on growth, yield and eco physiologic al responses of four tomato (Lycopersicon esculentum Mill.) cultivars. Journal of the Japanese Society for Horticultural Science 68 (3), 499504. Rinaldi M, Ventrella D, Gagliano C 2007. Comparison of nitrogen and irrigation strategies in tomato using CR OPGRO model. A case study from Southern Italy. Agricultural Water Managment 87, 91 105. Ritchie JT. 1998. Soil water balance and plant water stress. In: Tsuji GY, Hoogenboom G, Thornton PK, eds. Understa nding options for agricultural production. Dordrecht, The Netherlands: Kluwer Academic Publishers, pp 4153 Rylsky I 1979. Fruit set and development of seeded and seedless tomato fruits under diverse regimes of temperature and pollination. Journal of the American Society for Horticultural Science 104 8358 38. Sachs RM, Hackett WP 1969. Control of vegetative and reproductive development in seed plants. HortScience 4 103107. Sandri MA, Andriolo JL, Witter M. Dal Roos T 2003. Effect of shading on tomato plan ts grown under greenhouse Horticu l ture Brasileira 21 (4), p.642645. Sauviller C, Baets W, Pien H, Lemeur R 2002. Simultom: a diagnostic tool for greenhouse tomato production. Acta Horticulturae 593, 219226.

PAGE 217

217 Saw AF, Grange RI, Ho LC 1986. The regulation of source leaf assimilate compartmentation. In: Cronshaw J, Lucas WJ, Guiaquinta RT, eds, Plant b iology phloem t ransport Vol. 1 New York : AR Liss, pp 391398. Sawhney VK 1983. The role of temperature and its relationship with gibberellic acid in the development of floral organs of tomato (Lycopersicon esculentum). Canadian Journal of Botany 61, 12581265. Sawhney VK, Dabbs DH 1978. Gibberellic acid induced multilocular fruits in tomato and the role of Iocule number and seed number in fruit size. Canadian Journal of Botany 56, 28312835. Sawhne y VK, Polowick PL 1984. Fruit development in tomato: the role of the temperature. Canadian Journal of Botany 63, 10311035. Scott JW, Bryan HH, Ramos LJ 1997. High temperature fruit setting ability of large -fruited, jointless pedicel tomato hybrids with various combinations of heat tolerance. Proceedings of the Florida State Horticultural Society 110, 281284. Scott JW, Francis DM, Miller SA, Somodi GC, Jones JB 2003. Tomato bacterial spot resistance derived from PI114490; inheritance of resistance to ra ce T2 and relationship across three pathogens races Journal of the American Society for Horticultural Science 128 (5), 698703. Scholberg JMS. 1996. Ada ptive use of crop growth models to simulate the growth of field grown t omato. Doctoral dissertation, G raduate School of the University of Florida. Gainesville Fl. Scholberg JMS, Boote KJ, Jones JW, McNeal BL 1997. Adaptation of the CROPGR O model to simulate the growth o f field -grown tomato. In: Kropff MJ, Teng PS, Aggarwal PK, Bouma J, Bouman BAM, Jone s JW, van Laar HH, eds. Systems approaches for sustainable agricultural development: Aplications of systems approaches at the f ield Level Dordrecht, The Netherlands: Kluwer Academis Publishers, pp 133151. Scholberg JMS, McNeal BL, Boote KJ, Jones JW, Locasio SJ, Olson SM 2000. Nitrogen stress effects on growth and nitrogen accumulation by field-grown tomato. Agronomy Journal 92, 159167. Schuarz D, Klar ing HP, van Iersel MW, Ingram K T. 2002. Growth and photosynthetic response of tomato to nutrient s olution concentration at two light levels. Journal of American Society for Horticultural Science 127 (6), 984990. Seeney DW, Graetz DA, Bottcher AB, Locascio SJ, Campbell KL 1987. Tomato yield and nitrogen recovery as inflenced by irrigation method, ni trogen source, and mulch. HortScience 22 27 29. Szymkowiak EJ, Sussex IM 1992. The internal meristem layer (L3) determines floral meristem size and carpel number in tomato periclinal chimeras. Plant and Cell 4 10891100.

PAGE 218

218 Steel R, Torrie J 1992. Biosta tistic: Principles and Procedures. Steel R, Torrie J. ed. New York: McGraw -Hill, 633 pp. Struik PC, Yin X, de Visser P 2005. Complex quality traits: now time to model. Trends in Plant Science 10(11) 513516. Taber HG 2001. Petiole sap nitrate sufficien cy values for fresh market tomato production. Journal of Plant Nutrition 24, 945959. Tanaka A, Fuyita K, Kikushi K .1974. Nutrio -physiological studies on the tomato plant. III. Photosynthetic rate of individuals leaves in relation to the dry matter product ion of plant. Soil Science and Plant Nutrition Soil 20, 173183. Tanaka A, Fuyita K. 1974. Nutrio-Physiological studies on the tomato plant. IV. Source -sink relationship and structure of the source sink unit. Soil Science and Plant Nutrition 20, 305315. Thornley JHM 2006. Agriculture Modelling: a possible road map. Journal of Agricultural Science 144 446449. Torrecillas A, Guillaume C, Alarcon JJ, Ruiz Sanchez MG 1995. Water r elations of two tomato species u nder water stress and recovery. Plant Scie nce 105, 169 176. Tyree MT, Hammel HT 1972. The measurement of the turgor pressure and the water relation of plants by the pressure bomb technique. Journal of Experimental Botany 23 267282. Uzun S 2006. The quantitative effects of temperature and li ght on the number of leaves preceding the first fruiting inflorescence on the stem of tomato ( Lycopersicon esculentum Mill.) and aubergine ( Solanum melogena L.). Scientia Horticulturae 109, 142 146. Vallejos CE 1991. Low night temperatures have a differe ntial effect on the diurnal cycling of gene expression in cold sensitive and tolerant tomatoes. Plant, Cell Environment 14, 105112. Van Der Ploeg A, Heuvelink E 2005. Influence of sub-optimal temperature on tomato growth and yield: a review. Journal of Horticultural Science and Biotechnology 80 (6), 652659. Van Ieperen W, Van Meeteren U, Oosterkamp J, Trouwborst G 2005. Macro and m icroscopic aspects of fruit water relations influencing growth and q uality in Tomato Acta Horticulturae 682 (1) 501 505. Varga A, Bruinsma J 1986. Tomato. In CRC Handbook of fruit set and development, Mon selise SP, ed. Boca Raton, Fl: CRC Press, pp 461480. Verkerk K 1955. Temperature, light and the tomato. Meded elingen van de Landbouwhogeschool. Wagenegein, 55, 175 224.

PAGE 219

219 Walker AJ. Ho LC 1977. Carbon translocation in the tomato carbon import and fruit growth. Annals of Botany 41, 813823. Wallach D 2006. Evaluating crop m odels. In: Wallach D, Makowski D, Jones JW, eds. Working with dynamic crop m odels Amste rdam, The Netherlands : Elsevier pp 11 53. Wang F, Sanz A, Brenner ML, Smith A 1993. Sucrose s ynthase, starch accumulation, and tomato f ruit sink strength. Plant Physiology 101, 321327. Warner J, Zhang TQ, Hao X 2004. Effects of nitrogen fertiliza tion on fruit yield and quality of processing tomatoes Canadian Journal of Plant Science 84, 865871. Warnock SJ, Isaacs RL 1969. A linear heat unit system for tomatoes in California. HortScience 5 670 671. Warnock SJ. 1973. Tomato development in Cal ifornia in relation to heat unit accumulation. HortScience 8 (6), 487488. Warren GF. 1998. Spectacular increases in crop yields in the United States in the twentieth century. Weed Technology 12(4), 752760. Went FW. 1945. II Thermoperiodicity in growth and fruit of tomato. American Journal of Botany 31, 135150. Willmott CJ 1981. On the validation of models. Physical Geography 2 184 194. Wilson W J, Hand DW, Hannah MA 1992. Light interception and photosynthetic efficiency in some glasshouse crops. Journal of Experimental Botany 43, 363373. Winsor G, Adams P 1987. Glasshouse Crops Vol. 3 119125. Witt w er SH Aung LH 1969. Lycopersicon esculentum Mill. In: Evans LT, ed The induction of Flowering Some Case Histories. Melbourne, Australia : Macmillan pp 409. Wolf S, Rudich J 1988. The growth rate of fruits on different part of the tomato plant and the effect of water stress on dry weight accumulation. Scientia Horticulturae 34 1 11. Ya Ling L, Marcelis LFM, Stanghellini C 2004. Plant w ater relation s as affected by osmotic potential of the nutrient solution and potential transpiration in tomato ( Lycopersicon esculentum Mill.). Journal of Horticultural Science and Biotechnology 79 (2) 211218. Yeager AF 1927. Determinate growth in tomato. Journal of Heredity 18(6), 263265. Yoshioka H Takahasi K 1981. Studies on the translocation and accumulation of photosynthates in fruit vegetables. V. Translocation of photosynthates in a day, and effects of light conditions and night temperature on translocat ion and distribution of 14C photosynthates in tomato plants. Bull. Vegetables Crops Sta. Japan: A9 63-81.

PAGE 220

220 Young PA, MacArthur JW 1947. Horticultural characters of tomato. Texas Agric. Expt. Sta. Bul. 698, 1 61. Zeng ZB, Houle D, Cockreham CC 1990. H ow informative is Wrights estimator of the number of genes affecting a quantitative character? Genetics 26, 235248.

PAGE 221

221 BIOGRAPHICAL SKETCH M. Raquel Rybak was born in Leandro N. Alem, Misiones, Argentina. She is the daughter of Veronica Kuyesen and Jose Rybak. Raquel graduated from the National University of the Northeast in Corrientes, Argentina, in 1995 with the Agricultural Engineer degree. In 1996 she began to work in the National Institute of Agricultural Technology (INTA). In 1998 she was awarded a fellowship to pursue graduate studies in the University of Concepcion, Chillan, Chile. She graduat ed with the masters degree in agricultural e ngineering with specialization in Irrigation and Drainage in 2001. In 2005 she was awarded a fellowship from IN TA to pursue graduate studies in the United States Raquel enrolled at the University of Florida in summer 2005 with Dr. Kenneth Boote as her adviser. She received her Ph.D. from the University of Florida in the spring of 2009. At the end of her studies she reassumed her position in INTA, Argentina. Raquel is a member of the Professional Council of Agricultural Engineers (CPIA), American Society of Agronomy and Crop Science Society of America She is also a member of the honor Society of Agriculture Gamma S igma Delta.