<%BANNER%>

Modeling Growth and Composition of Perennial Tropical Forage Grasses


PAGE 1

MODELING GROWTH AND COMPOSITION OF PERENNIAL TROPICAL FORAGE GRASSES By STUART J. RYMPH A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Stuart J. Rymph

PAGE 3

iii ACKNOWLEDGMENTS I wish to express my sincere thanks a nd appreciation to Dr. Kenneth Boote (my supervisory committee chair) for introducing me to plant physiology and crop modeling, and especially for the conversations that went a little off-topic and tied the theory back to the field. I would also like to thank my committee memb ers Dr. Lynn Sollenberger for his advice on both agronomy and academia; Dr Charles Staples for our conversations on animal nutrition; a needed break from agr onomy and a reminder of how I enjoy working with dairies; Dr. Jim Jones for making engi neering fun and showing me that there is agriculture outside of the United States; a nd Dr. Tom Sinclair, th e “sounding-board”, for practical discussions and provi ding a more skeptical point of view. I’d also like to recognize the late Dr. Bill Kunkle for our ma ny conversations on farming and ruminant nutrition: it was like going home. Kudos go to Dr. Jean Thomas for the expe rtise, labor, and conversation that made the growth study possible and also enjoyable. Special thanks are offered to Dr. Paul Mislevy at the Florida Range Ca ttle REC in Ona, for providing datasets and also for his friendship, encouragement, and continued effort s to involve me in the practical side of tropical forage production. Special thanks go out to friends for thei r support and encouragement. Dr. John Moore helped to convince me to take on this endeavor. Dr. Kenny Woodard provided practical advice on growing forages, and the vo ice of a farmer, firmly grounded in reality (something that only enhances a modeling project). He and Tony Sweat provided

PAGE 4

iv comradery, lots of humor, and a view of Flor ida that I would otherw ise never have seen and would have been the worse for missing it. Extra special thanks are due my wife, Dr Mary Beth Hall, for her emotional and unending technical support whic h included sharing her exper tise in SAS, also a warm dinner and a break from dissecting bahiagrass ti llers. I also thank her for teaching me to not discard theory if it doesn’t have an imme diately apparent use in the field. Finally, I thank my mother and father, Thelma and Dr. Donald Rymph for instilling in me a healthy curiosity, a strong work ethic, and the belief that practical e xperience can be one of the best teachers.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT.....................................................................................................................xi ii CHAPTER 1 INTRODUCTION........................................................................................................1 2 LITERATURE REVIEW.............................................................................................5 Bahiagrass..................................................................................................................... 5 Perennating Organs: Rhizomes and Stolons.................................................................7 Dormancy.....................................................................................................................8 Photosynthesis............................................................................................................12 The CROPGRO Model...............................................................................................19 Model Evaluation........................................................................................................30 3 BAHIAGRASS GROWTH STUDY..........................................................................35 Introduction.................................................................................................................35 Materials and Methods...............................................................................................36 Results and Discussion...............................................................................................42 Plant Growth........................................................................................................42 Photosynthesis.....................................................................................................47 Conclusions.................................................................................................................50 4 DEVELOPMENT OF CROPGRO SPECIES FILE PARAMETERS FOR BAHIAGRASS...........................................................................................................61 Introduction.................................................................................................................61 Materials and Methods...............................................................................................62 Description of Data Sets Used to Fit Parameters................................................63 Preparation of Datasets........................................................................................65 Results and Discussion...............................................................................................67

PAGE 6

vi Photosynthesis Parameters..................................................................................67 Root Parameters...................................................................................................72 Carbon and Nitrogen Mobilization Parameters...................................................73 Vegetative Partitioning Parameters.....................................................................74 Leaf Growth and Senescence Parameters............................................................75 Phenology Parameters.........................................................................................75 Testing of Literature-Based Parameters..............................................................76 Optimization........................................................................................................79 Testing of Optimized Parameters........................................................................80 Conclusions.................................................................................................................82 5 ADAPTING CROPGRO TO MODEL PERENNIAL TROPICAL GRASSES: STRUCTURAL CHANGES TO THE MODEL........................................................97 Introduction.................................................................................................................97 Materials and Methods...............................................................................................99 Results and Discussion.............................................................................................103 Storage Organ....................................................................................................103 Dormancy..........................................................................................................107 Freeze Damage..................................................................................................114 Photosynthesis...................................................................................................117 Overall Model Performance – Herb age Mass and N Concentration.................123 Conclusions...............................................................................................................127 6 SUMMARY AND CONCLUSIONS.......................................................................151 Bahiagrass Growth Study.........................................................................................152 Development of Species File Parameters for Bahiagrass.........................................152 Adapting CROPGRO to Model Perennial Trop ical Grasses: Structural Changes to the Model.............................................................................................................153 Implications of the Research....................................................................................155 Future Research........................................................................................................156 APPENDIX A CROPGRO CSM PARAMETER CODE DEFINITIONS.......................................158 B SPECIES, CULTIVAR, AND ECOTYPE FILES FOR THE UNMODIFIED CSM VERSION OF CROPGRO.......................................................................................162 C SPECIES, CULTIVAR, AND ECOTYPE FILES FOR THE FORAGE VERSION OF CROPGRO.........................................................................................................183 D NEW PARAMETER CODE DEFINITIONS FOR THE FORAGE VERSION OF CROPGRO...............................................................................................................198 E DORMANCY AND STOR CODES AND DE FINITIONS FOR DATA.CDE FILE203

PAGE 7

vii F CODE ADDITIONS AND CHANGES IN THE FORAGE VERSION OF CROPGRO...............................................................................................................206 LIST OF REFERENCES.................................................................................................307 BIOGRAPHICAL SKETCH...........................................................................................315

PAGE 8

viii LIST OF TABLES Table page 3-1 Schedule of sampling and harvest activities..........................................................52 3-2 Weekly averages of daily temperatures and daily solar radia tion and total weekly rainfall + irrigation water applied to bahiagrass grown at the Irrigation Park, Gainesville, FL 2001...........................................................................................52 3-3 Results of statistical comparison of treatment effects on plant growth and photosynthesis. Period means are least s quares means. Sign ificance determined by ANOVA for Period and orthogonal contrast for Week and Per X Week interaction..............................................................................................................53 4-1 Bahiagrass parameter values for the CRO PGRO species file. Preliminary values were derived from the literature. Optimized values were derived from optimization runs made based on the preliminary values......................................85 4-2 Evaluation of the performance of CROP GRO with literature-based and optimized species files, with and without winter photosynthesis reduction...........................87 5-1 Summary of performance of the forage version of CROPGRO in simulating mass of below-ground plant organs (kg DM ha-1)........................................................129 5-2 Summary of performance of the CS M (unmodified) and forage version of CROPGRO in simulating five experiment s to predict herbage mass, herbage N concentration, and herbage N mass.....................................................................129

PAGE 9

ix LIST OF FIGURES Figure page 2-1 Modular structure a nd summary of model comp onents of the DSSAT-CSM cropping systems model.........................................................................................22 3-1 Sod core as removed from the soil.........................................................................54 3-2 Example of a separated subsample of bahiagrass after removing roots................54 3-3 Total plant mass for established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.......................................55 3-4 Root mass for established bahiagrass grow n at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001................................................55 3-5 Stolon mass for established bahiagrass gr own at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001...........................................56 3-6 Stem mass for established bahiagrass grow n at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001................................................56 3-7 Leaf mass for established bahiagrass grow n at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001................................................57 3-8 V-stage for established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001................................................57 3-9 Leaf area index (LAI) for established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001..........................58 3-10 Specific leaf area (SLA) for established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001..........................58 3-11 Leaf + Stem (green) ar ea index (GrAI) for established bahiagrass grown at Gainesville, FL from18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001........................................................................................................................59 3-12 Leaf photosynthetic rate for established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.......................................59

PAGE 10

x 3-13 Canopy gross photosynthetic rate adjusted to 1500 mol Par m-2 s-1 (P1500) for established bahiagrass grown at Gainesvi lle, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001..................................................................................60 4-1 Observed herbage mass ( ), predicted herbage mass ( ), water stress ( ), and N stress ( ) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, using the literature-based species file and the leaf-level photosynthesis option, with a) No adjustment to winter growth, or b) 70% reduction in photosynthetic rate and partial defoliatio n (frost) over the winter.................................................88 4-2 Observed bahiagrass herbage mass ( ) and predicted bahiagrass herbage mass using the preliminary (literature-based, non-optimized) species file and the leaflevel photosynthesis option ( ), or daily canopy photosynthesis option ( ). ............................................................................................................................... .89 4-3 Observed bahiagrass herbage N concentration ( ) and predicted bahiagrass herbage N concentration using the pre liminary (literature-based, non-optimized) species file and the leaf-level option ( ), or daily canopy option ( ). For bahiagrass grown at Ona, FL with 468 kg N ha-1 yr-1............................................90 4-4 Observed bahiagrass herbage N concentration ( ) and predicted bahiagrass herbage N concentration using the pre liminary (literature-based, non-optimized) species file and the leaf-level option ( ), or daily canopy option ( ). For bahiagrass grown at Eagle Lake, TX with 168 kg N ha-1 year-1............................91 4-5 Observed herbage mass ( ), predicted herbage mass ( ), water stress ( ), and N stress ( ) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, using the optimized species file with a winter defoliation, 70% reduction in winter photosynthetic rate, and a) the leaf -level p hotosynthesis option or b) daily canopy photosynthesis option................................................................................92 4-6 Predicted vs. observed herbage mass of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, and 168 kg N ha-1 yr-1 at Eagle Lake, TX, using a) the leaf-level photosynthesis option, or b) daily canopy photosynthesis option.........................93 4-7 Observed herbage N concentration ( ) and predicted herbage N concentration of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, using the optimized species file and the leaf-level option ( ) or daily canopy photosynthesis option ( ).94 4-8 Observed herbage N concentration ( ) and predicted herbage N concentration of bahiagrass grown with 168 kg N ha-1 yr-1 at Eagle Lake, TX, using the optimized species file and the leaf-level option ( ) or daily canopy photosynthesis option ( ).....................................................................................................................95 4-9 Predicted vs. observed he rbage N concentration (g kg-1) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, and grown with 168 kg N ha-1 yr-1 at Eagle Lake, TX, using a) the leaf-level photos ynthesis option, or b) the daily canopy photosynthesis option.............................................................................................96

PAGE 11

xi 5-1 Schematic of daily partitioning of ne w growth among vegetative tissues for the forage version of CROPGRO..............................................................................130 5-2 Schematic of the calculation of potential mobilization of CH2O from leaf, stem, root and STOR in the forage version of CROPGRO...........................................131 5-3 Schematic of the calculation of potential mobilization of N from leaf, stem, root and STOR in the forage version of CROPGRO..................................................132 5-4 Predicted vs. observed stolon mass for bahiagrass grown in the field in Gainesville, FL in 2001( ), and in temperature and CO2 gradient greenhouses at 360 L CO2 L-1 ( ), and 700 L CO2 L-1 ( )......................................................133 5-5 The a) controlling functions and b) seasonal expression of th e predicted effect of daylength on incremental (increase a bove baseline) partitioning to STOR( ) or mobilization from STOR ( ) in the forage version of CROPGRO..........134 5-6 Mobilization factors in th e forage version of CROPGRO that affect mobilization from STOR as a function of a) vegetative N status and b) LAI..........................135 5-7 Predicted ( ) vs. observed bahiagrass herbage mass for late-season harvests at Ona, FL in the 1993-1994 and the 19 95-1996 growing seasons using the modified leaf-level photosynthesis option in the forage versi on of CROPGRO. 136 5-8 Schematic of freeze damage to leaves and stems and cold-hardening of STOR tissues in the forage version of CROPGRO.........................................................137 5-9 Predicted CO2 compensation point for the CSM ve rsion of CROPGRO ( ) and for two hypothetical C4 species with a Ci/Ca of 0.4 and a CO2 concentrating factor of either 3 ( ) or 10 ( ) predicted using the forage version of CROPGRO..........................................................................................138 5-10 Relative CO2 concentration effect on a) CO2 factor for LFMAX and b) QE for a C3 species in the CSM version ( ) and for a C4 species in the forage version of CROPGRO using a Ci/Ca of 0.4 and CCNEFF of 3 ( )...........................139 5-11 Relative temperature effect on a) CO2 factor for LFMAX and b) QE for a C3 species in the CSM version ( ) and for a C4 species in the forage version of CROPGRO using a Ci/Ca of 0.4 and CCNEFF of 3 ( )................................140 5-12 Predicted growth of bahiagrass components under 350 L CO2 L-1 atmospheric CO2 stolon ( ), root ( ), and herbage ( ) relative to predicted growth under 700 L CO2 L-1 atmospheric CO2 stolon ( ), root ( ), and herbage ( ) using the forage version of CROPGRO...................................................141 5-13 Observed bahiagrass herbage mass ( ), predicted stolon ( ), root ( ), and herbage mass ( ) for bahiagrass grown in a) Ona, FL with 468 kg N ha-1 yr-1or b) Eagle Lake, TX with 168 kg N ha-1 yr-1..........................................................142

PAGE 12

xii 5-14 Predicted stolon ( ), root ( ), leaf ( ), and stem ( ) growth for bahiagrass grown with 468 kg N ha-1 yr-1 in Ona, FL in 1997 using the forage version of CROPGRO..........................................................................................143 5-15 Observed herbage mass ( ), predicted herbage mass ( ), water stress ( ), and N stress ( ) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL. Predicted using the leaf-lev el photosynthesis option in a) the forage version of CROPGRO or b) the unmodified CSM version of CROPGRO..........................144 5-16 Predicted ( ) vs. observed bahiagrass herbage mass for five experiments, using the modified leaf-level photosynthesis op tion in the forage version of CROPGRO. ..............................................................................................................................1 45 5-17 Observed/predicted bahiagrass herbage mass for a) the 0 kg N ha-1 yr-1 ( and ), 468 kg N ha-1 yr-1 ( and ), and 942 kg N ha-1 yr-1 ( and ) treatments at Ona, FL and b) the 0 kg N ha-1 yr-1 ( and ), 168 kg N ha-1 yr-1 ( and ), and 336 kg N ha-1 yr-1 ( and ) treatments at Eagle Lake, TX. ..............................................................................................................................1 46 5-18 Observed bahiagrass herbage N concentration ( ) and predicted bahiagrass herbage N concentration using the forage version of CROPGRO and the leaf-level option ( ), or daily canopy option ( ). For a) bahiagrass grown at Ona, FL with 468 kg N ha-1 yr-1, or b) bahiagrass grown at Eagle Lake, TX with 168 kg N ha-1 year-1.............................................................................................................147 5-19 Predicted ( ) vs. observed bahiagrass herbage N concentration for five experiments, using the modified leaf-lev el photosynthesis option in the forage version of CROPGRO..........................................................................................148 5-20 Observed/predicted bahiagrass herbage N concentration for a) the 0 kg N ha-1 yr-1 ( and ), 468 kg N ha-1 yr-1 ( and ), and 942 kg N ha-1 yr-1 ( and ) treatments at Ona, FL and b) the 0 kg N ha-1 yr-1 ( and ), 168 kg N ha-1 yr-1 ( and ), and 336 kg N ha-1 yr-1 ( and ) treatments at Eagle Lake, TX..............................................................................................................149 5-21 Predicted ( ) vs. observed bahiagrass herbage N mass for five experiments, using the modified leaf-level photosynthesis op tion in the forage version of CROPGRO. ..............................................................................................................................1 50

PAGE 13

xiii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MODELING GROWTH AND COMPOSITION OF PERENNIAL TROPICAL FORAGE GRASSES By Stuart J. Rymph December 2004 Chair: Kenneth J. Boote Major Department: Agronomy In addition to their role as feedstuffs, pe rennial tropical forage grasses such as bahiagrass ( Paspalum notatum Flgge) can play a major ro le in nutrient management on livestock farms; recycling N from fertiliz ers and manure to produ ce feed and reduce the importation of other feeds, while lowering po tential N leaching. Balancing feed quality, feed quantity, and nutrient recovery can be difficult. A computer model capable of simulating forage growth, composition, and Ndynamics could be a useful management tool. Farmers and consultants could test ma nagement practices virtually, then implement those showing the most promise. Our objec tive was to develop a tool to predict the growth and composition of bahiagrass that re sponds to environmental and management inputs. Bahiagrass sod cores were dug weekly for tw o 8-week regrowth periods (18 July to 12 September, and 12 September to 7 November). Plants were separated into leaf blades, stem, stolon, and roots. Leaf and canopy photos ynthesis were measured in each period.

PAGE 14

xiv Leaf photosynthetic rate was not different be tween periods. Leaf and stem growth and rate of development of new leaves was re duced in the second period; however, stolon mass increased dramatically st arting in mid-October. This information aided the development of species-specific parameters required for simulating bahiagrass in the model, CROPGRO. In the process, limitations in the model structure that prevented the prediction of realistic growth patterns were identified. Despite the limitations, prediction of herbage mass was good, having an index of agreement of 0.85, with slightly lower accur acy predicting herbage N concentration. To address the model’s limitations, we modified the CROPGRO source code to include a storage organ (STOR) equivalent to a stolon, and added dormancy functions to increase partitioning of growth to STOR a nd reduce mobilization from STOR and roots under short daylengths. The freeze-kill functi on was modified, allowing gradual death of leaves. The Rubisco specificity factor in the leaf-level photo synthesis option was modified for C4 photosynthesis. Model performan ce was improved, predicting realistic seasonal growth patterns. Excessive N stre ss was predicted frequently, but the cause was not identified. The forage version of CROP GRO performs realistically but should be tested under cooler temperatur es and finer-textured soils.

PAGE 15

1 CHAPTER 1 INTRODUCTION Perennial tropical grasses such as bahiagrass ( Paspalum notatum Flugge) and bermudagrass [ Cynodon dactylon (L.) Pers.] have long been the basis of beef production in Florida. As environmental concerns have come to the fore over potential movement of nitrate into Florida groundwater from dairy farms, there is a growing interest in use of these grasses as part of the nutrient manageme nt systems on Florida dairies. In both beef and dairy operations, the success of a forage program is measured by several criteria: it must generate enough mass to feed the herd, it may need to utilize a minimum amount of manure nutrients, and the forage produced must meet a significant proportion of the nutritional needs of the animals consuming it. Managing forages to meet all of these goals is a complex trade-off of yield and quality. Emphasizing forage quality generally requires harvesting at young stages of maturit y, with shorter regrow th periods and lower yields, whereas managing for forage mass produces forage with reduced concentrations of digestible nutrients. Current management strategies for perenni al tropical forages re ly on harvesting at regular intervals with no allowance for ch anging weather conditions. A 3 to 5 week harvest interval will nearly always produce high quality forage (Staples, 1995) but often the yields are unacceptably low. Delaying ha rvest when forage growth rates are low increases yield but at what cost to quality? A system that could compensate for changing temperature, rainfall, and soil fertility might allow farmers to harvest forage of more consistent quality in reasonable quantities.

PAGE 16

2 Such a system also could improve N cyc ling on the farm and reduce N leaching to groundwater. The high growth potential and e xpansive root system of tropical grasses such as bahiagrass and bermudagrass allo w these species to not only uptake large quantities of N from the soil but they may a lter the seasonal pattern of N leaching as well (Woodard et al., 2002). Properly managi ng the timing of fertilizer and manure applications as well as the timing of forage harvests can reduce nutrient losses to the environment while producing a dditional feed for the herd. A crop model capable of predicting fora ge yield and composition along with N leaching dynamics could allow evaluation of management practices (including harvest strategies) before implementing them in the field. To provide output with the desired level of detail, the model must have the abil ity to respond to a variety of environmental (temperature, rainfall, daylengt h, soil moisture, etc.) as well as management (fertilizer, irrigation, harvest schedule, etc.) inputs. It would also be advantageous if the model could be easily adapted to simulate other fo rage species to avoid rewriting the model for each new species to be modeled. Th e CROPGRO model meets both of these requirements. CROPGRO is a mechanistic or process-orient ed model. That is, it predicts plant growth by simulating many of the underlying bi ological processes. Some of the plant processes simulated include photosynthesis, transpiration, senes cence/mobilization of plant tissue, and root uptake of water and nutrients. In addition several supporting processes such as water infiltration and evaporation in the soil, soil organic matter and N dynamics (mineralization, nitrification, deni trification, and leaching), and biological N

PAGE 17

3 fixation are also simulated. The structure of the model also allows it to be readily adapted to new plant species. The model code itself is quite generic, including a minimu m of crop-specific relationships in the source code. Rath er, parameters defining the species-unique responses of the various processes are read from a set of species-speci fic input files. By providing alternate parameter files, the behavior of the model can be adapted to simulate other plant species. Parameter files have already been created to allow CROPGRO to simulate several crop species including soybean ( Glycine max L.), peanut ( Arachis hypogaea L.), dry bean ( Phaseolus vulgaris L.), faba bean ( Vicia faba L.), and tomato ( Lycopersicon esculentum Mill.) (Scholberg et al., 1997; Boote et al., 1998a, 1998b, 2002). Modeling perennial tropical grasses poses challenges not encountered when modeling annual grain or fruit crops. Perennial forages must persist from one year to another, regrow each spring, and go dormant each fall. Additionally, forages are not usually allowed to reach physiological maturity and may be harvested several times in a single growing season. The cu rrent CROPGRO structure can accommodate each of these challenges to some degree. However, some code changes are needed to achieve the desired level of model perfor mance for perennial forages. The goal of this project is to adapt the CROPGRO model code a nd species file to simulate the growth and composition of ba hiagrass over multiple, consecutive growing seasons. Parameter values will be obtained from the existing literature when available, directly measured in experiments when practical, and estimated by calibration when necessary. The Cropping Systems Model or CS M version of CROPGRO which is part of the Decision Support System for Agrotec hnology Transfer (DSSAT) version 4 model

PAGE 18

4 will be evaluated to quantify its ability to predict and determine any limitations for modeling perennial grasses. Finally, portions of existing code will be changed and some new code added to create an independent perennial forage version of CSM CROPGRO. The build date for the code used in our project is 11 July 2003. This is a prerelease version, but the relevant code is s ubstantially the same as the final version released as DSSAT v4 (Hoogenboom et al., 2003) To be compatible with the release version, the PESTCP module was updated, ch anging the forage cutting code from MOWE to MOW and changing the stubble mass units from g (leaf+stem) DM m-2 to kg (leaf+stem) ha-1. While efforts will be made to main tain compatibility with the existing CROPGRO model and its input-o utput structure, the perennial forage version will be considered to be a “new” model rather th an a new version of the current model.

PAGE 19

5 CHAPTER 2 LITERATURE REVIEW Bahiagrass Native to South America (Ward and Watson, 1973) and northward to Mexico (Scott, 1920), bahiagrass ( Paspalum notatum Flgge) is a perennial tropical grass that spreads by seed and vegetative stolons. First planted at the Florida Agricultural Experiment Station in May of 1913 (Scott, 1920), bahiagrass use has spread to over one million hectares (2.5 million acres) in Flor ida alone (Chambliss, 2002). Initial reports praised the grass for its palatability to cattle and especially for its persistence and capacity to spread and form a sod even under hea vy grazing pressure (Scott, 1920). This persistence along with it s ability to tolerate a wide va riety of management systems has been key to the popularity of this species. While used primarily for graz ing beef cattle, bahiagrass is also harvested as hay. Annual hay yields of near 8000 kg ha-1 may be expected when harvested every 4 wk (Johnson et al., 2001) and may exceed 11000 kg ha-1 when harvested every 6 to 8 wk (Blue, 1973). Bahiagrass responds we ll to N fertilization, yielding 2700 kg ha-1 yr-1with no fertilizer N (Beaty et al., 1964) and increasi ng to nearly 14000 kg ha-1 yr-1 when fertilized with very high levels of N and cu t to a low stubble height (Pedreira and Brown, 1996a). The most prolific biomass growth occurs below the common cutting height of 5 to 10 cm. The prostrate growth habit of ‘P ensacola’ bahiagrass produces a large amount of leaf low in the canopy, with reports of as much as 5600 kg ha-1 of leaf DM ( Beaty et al., 1964) and a leaf area i ndex (LAI) in excess of 1.6 m2 leaf m-2 land remaining after

PAGE 20

6 harvest (Pedreira and Brown, 1996b) Beaty et al. (1968) report ed that nearly 40% of the total forage produced by Pensacola bahiagra ss was present in the bottom 2.5 cm of the canopy. Depending on the fertility and age of the stand, an additional 9800 kg ha-1 (Beaty et al., 1964) or more of rhizome or stolon (the terms rhizome and stolon are used interchangeably in the literature when referr ing to bahiagrass) mass were present at the soil surface. Bahiagrass also has an extensiv e fibrous root system which may produce as much as 19700 kg ha-1 DM in just the top 15 cm of th e soil (Beaty et al., 1964), although a level near 4500 kg ha-1 root DM (Burton et al., 1954) is more common. Pensacola bahiagrass roots have been reported to ex tend more than two meters below the soil surface (Burton et al., 1954; Chambliss, 2002) Although other cultivars may exhibit a more erect growth habit or lower stubble ma ss in the seedling year than Pensacola, even these “improved” cultivars may still have a harvest index of 10% or less (harvested forage as a proportion of total above-ground bi omass present at harvest) after the field has been established for more than two years (Pedreira and Brown, 1996b). Such an investment in biomass close to the ground and below-ground encourages the persistence and quick sod formation that the species is no ted for. It also can support the crop through periods of stress, allowing it to adapt to a variety of management practices. The most popular bahiagrass cultivar is Pensacola bahiagrass (Chambliss, 2002), a West Indies-type bahiagrass, t hought to have been introduced in ballast offloaded at the Perdido Wharf sometime prior to 1926 a nd promoted after 1939 by Escambia County Cooperative Extension county agent Ed Finl ayson for its aggressive sod formation, persistence, drought tolerance, and palatability to cattle (Finlayson, 1941). Pensacola is distinguished by long, narrow leaves and tall seedheads. Aside from its persistence,

PAGE 21

7 Pensacola bahiagrass is also more cold-tol erant than many other cultivars (Ward and Watson, 1973; Chambliss, 2002). The productive season for bahiagrass is Apri l to November in North Florida. The season starts in early March in South Fl orida but growth sti ll slows in October (Chambliss, 2002), suggesting that the decline in growth may not be solely temperature dependent. Indeed, recent work by Sinclair et al. (2001; 2003) a nd Gates et al. (2001) demonstrate a role of daylengt h in initiating winter dormancy. Perennating Organs: Rhizomes and Stolons While bahiagrass reproduces by seed, the plants spread and form a dense sod through the growth of rhizomes (equiva lent to stolons) (Ward and Watson, 1973; Chambliss, 2002). Stolons allow the plants to spread and occupy more land area with nodes on the stolons providing new growing points for additi onal tiller growth. New tillers form at an axillary bud in response to the flowering of a nearby culm (Sampaio and Beaty, 1976), maintaining tiller density and re juvenating the stand. New tillers may also form in response to changing light (daylength and light quality) with the onset of spring or after a harvest (and the po ssible removal of apical dom inance). This new growth allows the stand to persist from year to year and under grazing or hay management systems. The stolons also promote persistence by acting as a storage vessel for C and N reserves, providing nutrients for growth unde r stressful conditions and promoting rapid regrowth after winter or harv est. As was previously men tioned, the mass of stolons may be several times the mass of leaves, and the reserves may last two to three years (Chambliss, 2002), providing for new growth. Partitioning to, and mobilization from, stolons, may at times account for the majority of the nutrient flows within the plant.

PAGE 22

8 Studies of carbon allocation and movement in bahiagrass using 14CO2 have shown 50 – 60% of the 14C partitioned to stolons, with only 10% partitioned to new leaf growth (Beaty et al., 1974). Patterns of mobilizati on are also affected by harvesting. The same study reported that re-mobilization of 14C from the surviving tissue ceased within 3 to 6 d after a severe defoliation. No more than 20% of the 14C assimilated was re-mobilized, of which only 10% moved to the leaves of the main plant while 20 to 40% went to the stolons. Such isotope studies characterizing patterns of mobilization and partitioning to new growth during spring regrowth have not been conducted, but general studies of fall and spring plant growth suggest that there are regulatory roles for two additional factors: daylength and temperature. Late and early season growth can be slowed by the cooler temperatures associated with the fall, winter, and spring months. This is also the period of the year when daylengt h is near its minimum. Dormancy Forage production from tropical gras ses like bahiagrass and bermudagrass [ Cynodon dactylon (L.) Pers.] drops dramatically in the late summer and fall months in the southeastern United States despite contin ued warm temperatures. Gates et al. (2001) quantified this reduction in yi eld in bahiagrass by measuring forage growth through two successive fall and winter seasons concurre ntly at Ona, FL and Tifton, GA. Mean temperatures were 6C cooler in Tifton where multiple freeze events occurred while there was only one night in the two growing seasons where temperatures dropped below 0C at Ona. The pattern of seasonal forage pr oduction was similar for both sites; however, production at Tifton was lower than for Ona. At Ona daily forage production (leaf and stem mass recovered above the mowi ng height) decreased from 36 kg DM ha-1 d-1 to 8 kg DM ha-1 d-1 between 23 September 1993 and 8 N ovember 1993, a 78% decrease in

PAGE 23

9 growth rate. The growth rate remained between 5 and 15 kg DM ha-1 d-1 through 7 March, then increased to 35 kg DM ha-1 d-1 by 6 April. Cooler temperatures in the second growing season (1994-1995) resulted in a similar pattern of forage growth but lower overall yields and a more rapid increa se in growth rate in the spring. Although there was no difference in growth rates betw een the three cultivars tested (Pensacola, Tifton 9, and RRPS Cycle 18) on dates when minimum growth rates were realized, regrowth of Pensacola in September and Apr il was considerably lower than that of the newer, higher yielding cultivar, RRPS Cycle 18. No information on the cause of the seasonal reduction could be discerned from the study other than the generally lower yields at lower temperatures. Growers in Georgia have attr ibuted the low productivity of tropical grasses in the fall to a lack of rainfall and fertilizer (Burton et al., 1988). Investigating this hypothesis, Burton et al. (1988) compared the forage yi elds of well-fertiliz ed, irrigated and nonirrigated Coastal bermudagrass (another C4 grass that exhibits winter dormancy) harvested every 24 d from 1 April through 27 October for three consecutive years. A seasonal decrease in forage yield was obser ved in all treatments during the fall and winter. The magnitude of the drop was quite dramatic with yields for the September – October harvest being only slightly more th an 1/3 the yield of the May – June harvest period despite slightly higher temperatures in the fall. Correlation analyses of the data showed only a moderate relationship between yield and temperature (r=0.46) or growing degree days (GDD) (r=0.37), t hus, discounting the roles of temperature, rainfall, and fertilizer in the seasonal yield reduction patte rn. Two variables that did show a strong correlation with yield were daylength (r=0.95) and daily solar radiat ion (r=0.93). It was

PAGE 24

10 not possible to separate these tw o effects with this dataset. Stepwise regression was also employed using individual growth period data for all years. The single variable model with the best fit was daylength (r2= 0.64, 0.69, and 0.61 for all yields, irrigated yields only, and non-irrigated yields, respectively). When daylength was excluded from the model, the best fit for a single variable was rainfall + irrigation (r2=0.45). Adding solar radiation to the rainfall model only increased the R2 to 0.57, while the daylength + rainfall gave an R2 of 0.73 for all yields and 0.77 for irri gated yields only. A model with daylength and water deficit (pan evaporation – rainfall) had an R2 of 0.71 when fit to the non-irrigated yield data. Stepwise regres sion did not compensate for the confounded effects of daylength and total solar radiation but did lend mo re weight to the daylength argument. Marousky et al. (1992) attempted to separate daylength effects from solar radiation effects in a study of daylength effects on tu rf-type bermudagrass cultivars. Long days were simulated using the same 9-h daytime as the short-day treatment but the dark period was interrupted with 4 hours of light from incandescent lamps. Interruption of the dark period with light causes the reversal of many red/far-red phytochrome reactions; in effect, simulating a much longer day than the total time exposed to light. Long-day plants exhibited a greater degree of l eaf extension but no change in plant dry weight or number of stolons produced. The authors’ rationale fo r the apparent conflict with the results of Burton et al. (1988) was that cu ltivar differences resulted in different responses between the long-leaved, forage-type Coastal berm udagrass and the fine-leaved, turf-type cultivars.

PAGE 25

11 Another possible explanation might be f ound in work by Britz et al. (1958) who studied short-day (late-season) accumulation of st arch in leaves of another tropical grass, Digitaria decumbens Stent.. This accumulation of star ch was associated with a decrease in translocation of a ssimilate under short days. The aut hors used several methods to elicit a long-day response. Gradual extension of the daylength from 7-h (short-day) to 14 h (long-day) with full-spectrum, fu ll-irradiance light (400-600 mol m-2 s-1 photosynthetic photon flux density) resulted in no decrease in starch accumulation with 9 h daylength but a dramatic drop between 11 and 12 h of light with starch concentr ations after 12 h of light equal to levels at 14-h daylength. In contrast, interr upting the dark period with 0.5 h or 2.0 h of full-spectrum light resulted in a partial (~40% and ~60%, respectively) reduction in starch accumulation relative to a 14 h day. The full long-day response was achieved when the 7 h light period was shifted from the first half of the 24 h period to the second half. These results imp lied that timing of the light pe riods may be more important than total time of exposure. To separate th e effects of daylength and total solar radiation, plants were exposed to 14 h of ambe r light (589 nm), having a good photosynthetic spectra but lacking photomor phogenetically-active wavele ngths. Photosynthate production under long days of amber light was similar to that of the long-day plants under full-spectrum light but starch partitioni ng was similar to that of the short-day plants. This helped to solidify the conc lusion that the starch accumulation was a photoperiodic response but also indicated that it involved a more complex mechanism than a single red/far-red phytochrome response. The different res ponses of the long day and interrupted night treatments may indicate a similarly complex mechanism in the short-day response of bermudagrass.

PAGE 26

12 Sinclair et al. (2003) avoided these complications by extending the natural daylength over field-grown Pe nsacola bahiagrass, ‘Florakirk ’ bermudagrass, ‘Tifton’ 85 bermudagrass ( Cynodon spp.), and ‘Florona’ stargrass ( Cynodon nlemfuensis Vanderyst) to 15 h using halogen lamps during the short daylength months. Light levels under the lamps were equivalent to less than 2% of full sunlight and, thus, not considered to contribute significantly to total daily radiation. While all fo ur species had higher forage production under the extended daylength treat ment, bahiagrass showed the greatest response. Extended daylength yields were frequently more than twice those of the natural daylength treatment with yields six ti mes greater on one harvest date. Despite the increases, mid-winter yields under extended da ylength were still quite low compared to summer yields. Although forage growth increased, below-ground biomass production was not affected by extending the light period. Another f actor potentially related to dormancy, total nonstructural carbohydrate (TNC) concentr ation in the below-ground biomass (root + stolon), decreased during the sh ort-daylength months in both the naturaland extended-daylength treatments. All this paints a picture of fa ll dormancy in tropical grasses being triggered, at le ast in part, by daylength. Th e exact mechanisms have not been described but the relatively low forage production even under extended daylength as well as the continued loss of TNC during th e winter, point to a complex of factors contributing to the response rather than a simple phytochrome-mediated response. Photosynthesis The large DM production of bahiagrass can largely be attributed to it being a tropical grass species expressing the C4 photosynthetic pathway. Temperate grasses express the C3 photosynthetic pathway wherein CO2 is fixed by ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco) in the chlo roplasts to produce 2 molecules of 3-

PAGE 27

13 phosphoglyceric acid (3-PGA; a 3-carbon acid) and, eventually carbohydrate via the Calvin cycle. Rubisco is also an oxygenase, capable of fixing O2 as well as CO2. When O2 is fixed, one molecule of 2-phosphoglycol ate and only one molecule of 3-PGA are produced. Not only is there half as much 3-PGA produced when O2 is fixed, two molecules of glycolate can be metabo lized to release a molecule of CO2, hence the term photorespiration referring to oxygenase activity. Both CO2 and O2 compete for binding sites on Rubisco, with higher relative concentrations of O2 in the chloroplast resulting in higher rates of photorespira tion and reduced photosynthetic efficiency. In C4 photosynthesis, CO2 is fixed in the mesophyll cells (cells near the leaf surface) by phosphoenolpyruvate carboxylase (PEPCase) to form oxaloacetate (OAA; a 4-carbon acid). The OAA is subsequently converted to either malate or aspartate, depending on the type of C4 pathway expressed in the species, and transported to the bundle sheath cells isolated deep within the leaves. In the bundle sheath cells, the malate or aspartate is decarboxylated near the chloroplast where Rubisco re-fixes the CO2 to produce 2 molecules of 3-PGA as in C3 photosynthesis. Energetically, C4 photosynthesis is more expensive with the additional cost of 2 ATPs to fix HCO3 with PEPCase on top of the cost of 3 ATPs and 2 NADP Hs to fix one mole of CO2 in the C3 pathway (Kanai and Edwards, 1999). However, PEPCase has no oxygenase activity and, thus, can be more efficient than Rubisco in fixing CO2 in the presence of O2. Also, by isolating Rubisco in the bundle sheath and “shuttling” the CO2 to it, CO2 is concentrated around the Rubisco and photorespiration is minimized. U nder atmospheric conditions (~21% O2 and 370 ppm CO2), photorespiration in C4 plants may be on the order of 3% of the net rate of CO2 fixation, compared to rates of 54% observed in the C3 species wheat ( Triticum aestivum

PAGE 28

14 L.) (Kanai and Edwards, 1999). Reduced photorespiration not only increases the efficiency of Rubisco carboxylation, it also increases the light le vel at which light saturated photosynthesis occurs, lowers the CO2 compensation point, increases quantum efficiency (QE), changes the temperature se nsitivity of both QE and photosynthesis, and allows high photosynthetic rate s at relatively low concen trations of leaf N or photosynthetic enzymes. The higher carboxylation rate resulting from CO2 saturation of Rubisco and the reduced photorespiration enables C4 plants to potentially at tain higher photosynthetic rates at high light. While C3 photosynthesis becomes light saturated at relatively low light levels, the enhanced capacity of the C4 system can tolerate very high light levels without becoming light saturated. Ludlow and Wilson (1971) compared leaf net photosynthetic rates of tropical grasses (C4) with tropical legumes (C3) over a range of illuminances. Consistent with the capacity to respond to higher light levels, the C4 grasses had net photosynth etic rates (37.8 mol CO2 m-2 s-1 ) that were approximately double those noted for the C3 legumes (17.7 mol CO2 m-2 s-1). The light response curves of the legumes reached a plateau at around 4000 – 5000 foot-candles of light while the curves for the C4 grasses were only approachi ng saturation at the highest illuminance of 10 000 foot-candles. Boote et al. (1999) measured leaf net photosynthesis (Pn) near 28 mol CO2 m-2 s-1 in established bahiagrass under atmospheric conditions compared to approximately 20 mol CO2 m-2 s-1 for the C3 legume rhizoma peanut ( Arachis glabrata Benth.). When the CO2 concentration was doubled from 350 L L-1 to 700 L L-1, bahiagrass responded with only a 20% in crease in Pn, about half of the 36% increase measured for rhizoma peanut (Boote et al., 1999).

PAGE 29

15 Higher photosynthetic efficien cy also allows the CO2 compensation point (CCMP) (atmospheric CO2 concentration where the rate of CO2 uptake by photosynthesis equals the rate of CO2 efflux) of C4 plants to be considerably below that of C3 plants. Bolton and Brown (1980) recorded CO2 compensation points of 4-14 L L-1 for the C4 grass Panicum maximum much lower than the values (47-59 L L-1) measured in tall fescue ( Festuca arundinacea Schreb.), a C3 grass. Several other C4 grasses have been shown to have CCMPs near 0 L L-1 including Vetiveria zizanoides (0 to 5 L L-1), and a variety of Cymbopogon species (0 to 3 L L-1) (Rajendrudu and Das, 1981). Efficiency of light utilization may also improve with reduced photorespiration. Quantum efficiency or quantum yield is th e efficiency of leaf photosynthesis when measured at low light and is generally expressed as mol CO2 mol-1 absorbed photons. This describes the initial slope of the phot osynthetic response to light (light-limited region). A frequently cited QE value for C3 species is 0.05241 mol CO2 mol-1 absorbed photons, the average QE measured at 330 L CO2 L-1 for seven C3 species (Ehleringer and Bjrkma n, 1977). Values for C4 species are generally higher than those for C3 species and may range from 0.046 for Sorghastrum nutans (Monson et al., 1982) to 0.075 for Saccharum spontaneum (Meinzer and Zhu, 1998). Differences in efficiency exist between the three types of C4 pathways (NAD-ME, NADP-ME, and PCK-type) (Ehleringer and Pearcy, 1983) as well. Ehle ringer and Pearcy (1983) found that species expressing the NADP-ME pathway, [e.g. bahiagrass and maize ( Zea mays L.)] exhibited the highest QE with eight monocot species averaging 0.065 mol CO2 mol-1 absorbed photons. Reports of QE for other NADP-ME species range from 0.062 to 0.075 mol CO2 mol-1 absorbed photons for various sugarcane ( Saccharum ) species at 350 ppm

PAGE 30

16 CO2 (Meinzer and Zhu, 1998). Quan tum efficiencies for other C4 pathway types recorded by Ehleringer a nd Pearcy (1983) were 0.064 mol CO2 mol-1 absorbed photons for five monocot species exhi biting the PCK-type pathway, and 0.060 mol CO2 mol-1 absorbed photons for three NAD-ME species It should be noted that both the Ehleringer and Pearcy (1983) and Ehleringer and Bjrkman (1977) studies measured QE at 330 ppm CO2 rather than the customary 350 ppm CO2. The narrow range of values for a given pathway when meas ured across many species suggests that QE may be a characteristic that is highly conserved across species. This would imply that values for a particular pathway type may be generally applied to other species exhibiting that photosynthetic pathway and that th e QE for bahiagrass is near 0.065 mol CO2 mol-1 absorbed photons. While the QE of C4 plants is generally greater than that of C3 plants at higher temperatures, the QE of C3 plants may exceed those of C4 plants at temperatures below 30C (Ehleringer and Bjrkman, 1977; Ku and Edwards, 1978; Monson et al., 1982). The QE of C4 plants is temperature insensitive, d ecreasing very little as temperatures increase while the QE of C3 plants falls dramatically as temperature increases (Ehleringer and Bjrkman, 1977; Ku and Edwards, 1978; Monson et al., 1982; Ehleringer and Pearcy, 1983). The decrease in C3 QE at higher temperatures is attributed to increased photorespiration (Ehleringer and Bjrkman, 1977; Monson et al., 1982). The solubility of CO2 decreases relative to that of O2 as temperature increases, creating a condition where O2 concentration around Rubisco may be enhanced, resulting in increased oxygenase activity relative to carboxylation. This effect alone cannot account for the drop in QE of C3s at high temperatures (Ehler inger and Bjrkman, 1977; Ku and

PAGE 31

17 Edwards, 1978), rather it is likely the combined effects of the changing relative gas solubilities and a changing affinity of Rubisco for CO2 and O2 (Jordan and Ogren, 1984) that favors O2 at higher temperatures. The CO2-concentrating mechanisms of the C4 pathways provide a high concentration of CO2 relative to O2 in the bundle sheath chloroplast such that photore spiration effects are not evid ent (Jordan and Ogren, 1984). Consistent with a higher QE at high te mperatures, the temperature optimum for photosynthesis in C4 plants is generally abou t 10C higher than for C3s (Long, 1999). Conversely, at lower temperatures, C4 photosynthetic rates may be below that of comparable C3 plants. The higher temperature optim um may be explained by the lack of photorespiration in the C4s, but the mechanism behind the decreased performance at lower temperatures has been more elus ive. In an extensive review of C4 photosynthesis at low temperatures, Long (1983) explored the effect of lo w temperatures on several of the steps in the C4 pathway(s). No single step could be shown to be the limiting factor at low temperatures. Two enzyme steps were id entified to be the most likely to limit C4 photosynthetic rate at low temper atures: pyruvate Pi dikinase (PPDK) activity or Rubisco activity. Pyruvate Pi dikinase was suspect ed for its relatively low activity at all temperatures and the dramatic increase in its activation time unde r cold conditions. Rubisco limitation might occur due to th e low quantities of the enzyme in C4 leaves along with a possible inhibition of the CO2 concentrating mechanism under cold temperatures. In a more recent review of responses of C4 photosynthesis to various environmental factors, Long (1999) revisits his investigation of low temperature effects, this time refuting the argument for PPDK limitation of photosynthetic rate and concluding that “there is no inherent flaw in C4 photosynthesis that prevents efficient operation at low

PAGE 32

18 temperatures” (p. 240) and citing Miscanthus as an example of a C4 species capable of persisting at high altitudes/low temperatures. Recent work by Pitterman and Sage (2000) using two ecotypes of Bouteloua gracilis Lag., adapted to high or low elevations implicates Rubisco capacity as limiting photos ynthesis at temperat ures below 17C. They also identify PPDK activity or ribulos e 1,5-bisphosphate (RuBP) regeneration as potentially limiting photosynthesis between 20 C and the optimum temperature (~37C), and PEPCase capacity as potentially limiting photosynthetic rate at temperatures greater than the optimum. That C4 photosynthesis could have a rate lower than the C3 rate at low temperatures if Rubisco was limiting in both pathways was attributed to the low Rubisco content of C4 leaves coupled with the added co st of carboxylation of PEP in the C4 pathway. The low Rubisco content of C4 leaves (Pittermann and Sage, 2000) contributes to a lower leaf N concentration and smaller propor tion of soluble leaf N than is typically found in C3 plants (Slack and Hatch, 1967; Crespo et al., 1979; Sugiyama and Hirayama, 1983; Usuda et al., 1984). While Rubisco may contribute over 40% of the soluble leaf N in a C3 leaf, it represents only about 5-28% of soluble N in C4s (Slack and Hatch, 1967; Sugiyama and Hirayama, 1983). The added N pr esent in PEPCase (4-1 0% of soluble leaf N (Slack and Hatch, 1967)) does not match the relative decrease in Rubisco concentrations, resulting in a lower total quantity of photosyntheti c enzymes (PEPCase plus Rubisco) and lower to tal N concentration in C4 leaves compared to C3s (Slack and Hatch, 1967; Brown, 1978; Raghavendra and RamaDas, 1993). Leaf N concentration can be increased by increasing N supply to the plant. Sugiyama and Hirayama (1983) found that Rubisco, PEPCase, and PPDK quantitie s all increased with increasing N supply.

PAGE 33

19 However, while the concentration of PEPC ase and PPDK increased (as a proportion of DM), the concentration of R ubisco decreased relative to C3 plants even as leaf N concentration increased with N supply. Th e high photosynthetic ra tes at low leaf N concentrations result in a hi gh N-use efficiency, a factor cr itical to the productivity of many tropical grasses in low input systems. Conversely, their low crude protein (CP) concentration coupled with hi gh cell wall concentrations (lar gely a consequence of the extensive vascular system associated w ith the Kranz anatomy and bundle sheath) are primary contributors to the relatively low fora ge quality of most tropical grasses. The CROPGRO Model The goal of this section is to familiarize the reader with the major features of CROPGRO pertinent to adapting the model for perennial tropi cal forages. Discussion will be limited to a review of the pattern of information flow and summaries of some of the plant-related subroutines. More comp lete descriptions of CROPGO have been published by Boote et al. (1998a, 1998b) with an extensive review of the hedgerow photosynthesis approach by B oote and Pickering (1994). One of the primary objectives in devel oping CROPGRO was to have a model that could easily be adapted to simulate the grow th of different plan t species. CROPGRO was created as a way to consolidate th e existing SOYGRO, PNUTGRO, and BEANGRO models into a single program (Boote et al., 1998a). The three programs shared much of their code, so merging the models primarily involved moving the parameters describing species and cultivar traits from the actual code to external input files. This structure allows other species to be modeled by creating new parameter (input) files. CROPGRO was developed in the early 1990s as a stand-alone model but could also be run under the Decision Support System fo r Agrotechnology Tran sfer (DSSAT) shell

PAGE 34

20 (ICASA, 1998), allowing it to be linked to other crop modeling programs as well as graphics programs to automate presentati on of results. After DSSAT version 3.1 was released in 1996, the CROPGRO code was reor ganized into a modular structure. Code for simulating different plant and soil processes was organi zed into individual subroutines for each process. The new subrout ines were designed to be executed in four common steps (initialization, rate calculati on, integration and final/summary) called by the main model. The modular structure was desi gned to allow users to add new code or to link code from other programs by inserting new modules into the CROPGRO structure. This approach has been applied to the DSSA T shell such that CROPGRO is an integral component of the Cropping Systems Mode l (CSM) (Jones et al., 2003). CROPGRO serves as the crop template module (Hooge nboom et al., 2003; Jones et al., 2003); a universal interface for modeling several di fferent species. An overview of DSSAT version 4, which incorporates the CSM ve rsion of CROPGRO, can be found in the DSSAT v4 documentation (Hoogenboom et al ., 2003) and in Jones et al. (2003). The modular version of CROPGRO was the fi rst version to include the soil organic matter (SOM) transformation module based on the CENTURY model (Gijsman et al., 2002). The only SOM transformation option av ailable in earlier versions was an adaptation of the PAPRAN model (Godwin a nd Jones, 1991). Both options address mineralization from SOM as well as immobiliz ation, nitrification, a nd denitrification of N. The CENTURY option adds the capability of simulating decay of surface residues and the movement of those nutrients into th e soil profile. Pastur e and other perennial forage systems are often low-input systems re lying heavily on recycled nutrients from plant residues for continued productivity. Omission of surface litter nutrient pools from

PAGE 35

21 the PAPRAN option may dramatically re duce the available soil N and C pools, potentially resulting in chronic underpredicti on of plant growth especially over the multiyear simulation periods typical for perenni al forage systems. Thus, the CENTURY option is better suited to m odeling perennial forages. CROPGRO and CSM incorporate several modules for simulating environmental and management responses, soil N transformati ons, soil water availability, etc. (Figure 21). Individual modules depict ing different processes are exec uted once within each of the four steps of the modular structure. This modular structure should enhance the future development of these programs as it allows ot hers to add new features by inserting their own modules. Also, by separating the rate and in tegration steps, the or der of execution of the modules within each step is generally less critical, again, facilitating further development. In the initialization step, parameter valu es and simulation control information are read from various input files and initial values set for state variables (variables representing the state of the system at the end of the day, variable s that have a quantity such as mass of roots or number of leaves ). This step is run once per simulation, although there are provisions to re-run each simulation for multiple years where the initialization step is r un once per repetition. Several input files are used to set the parameters for a species and control the execution of the simulation. Three files cont ain the plant parameters: a species file, a cultivar file, and an ecotype file. The experi ment file, or “X-file” controls the simulation and is supplemented by weather files, a soil in formation file, and a pest file. Comparison of predicted data to measured experimental results is automated, with the program

PAGE 36

22 Figure 2-1. Modular structure and summary of model component s of the DSSAT-CSM cropping sy stems model (Jones et al., 2003, p. 239). Primary Modules Weather CROPGRO Plant Template Plant Management Soil Soil Plant Atmosphere Pest & Disease Damage Environmental Modifications Plant Modules CERES Maize CERES Wheat SUBSTOR Potato Other crops CERES Rice Planting Harvesting Irrigation Residue Placement Fertilizer Application Soil Temperature Soil Water Soil Dynamics Soil Nitrogen & Carbon Template Crop Models Soybean Tomato Dry bean Peanut Other crops Main Program Start Run Initialization Seasonal Initialization Rate Calculations Integration Output Summary EndSeasonal loop Daily loop Land Unit Module(The Land Unit Module is called by the Main Program to perform each step of processing and in turn calls each of the Primary Modules.)

PAGE 37

23 reading the measured data from a time-course fi eld data file or “T-f ile” or a field average observational data file or “file A”. The species file contains species-specific parameters describing the response of the crop to the environment as well as paramete rs describing growth and photosynthesis. These parameters are set during model develo pment and are not generally altered by the user. Some parameters may be adjusted to reflect differences in behavior of different cultivars and ecotypes via addi tional parameters in the cultivar and ecotype files. Parameters in these files include factors such as differences in physiological time between growth stages, relative differences in photosynthetic rate, and leaf size, among others. Other files, such as the Xand T-files, must be created by the user. A file X contains information describing the simulation, including what weather files to use, what soil type is present (along with site-speci fic soil profile information), location of the experiment site, management information (p lanting date, fertiliz ation and irrigation schedules, harvest date, etc.), and simula tion controls such as which photosynthesis option to use, whether to pred ict potential growth (assume no stresses), or water-limited growth, or waterand N-limited growth, and other options that determine modules of the model to be used. If the simulation results are to be compared to an actual experi ment, measured data for each treatment can be listed in a time-course field data file or “T-file” or the field average observational data file (file A). Data from this file can be read by a graphics program to plot the predicted variables against the measured data. The T-file also can be used to list amounts of pest damage and the dates on which the damage is to be imposed.

PAGE 38

24 Based on the pest codes used in the T-file, the Pest file dete rmines what type of damage to impose. The Pest file is generally wr itten during model deve lopment, tailoring the codes to the specific crop being modeled. Becau se the “harvest” date listed in the X-file signals the termination of simulation in CRO PGRO, periodic forage harvests must be simulated using the pest code, “MOW” that is entered in the T-file. The user specifies the amount of stubble mass to remain after a harvest or grazing event and CROPGRO reduces the amount of leaf and stem proporti onally to leave that amount of stubble at a reduced V-stage. As with any pest damage, the leaf and stem removed by MOW are lost from the simulated system, with the assumption that they were exported from the field. While annual species may be simulated wit hout enabling the Pest option, all forage simulations must use the Pest option to create multiple harvests. Other site-specific information that must be provided by the user is daily weather information and soil profile information. Da ily solar radiation, rainfall, and minimum and maximum temperature data are listed in the weather files which are named according to their location and the year for which they contain data. The soil file (SOIL.SOL) contains soil profile information on the sp ecified soil types. The user must add information to create the soil for the simulation in a specified format. CROPGRO requires all of the above menti oned files except the T-file and A-file to initialize and run a simulation, unless the simulation involves multiple harvest, in which case the T-file is required as well. Initialization of a simulation begins on the simulation start date, which may be well in advance of the planting date. Planting date can be the day that seed is sown or the day that transplants are placed in the field. The transplant option can be used to initialize the

PAGE 39

25 simulation with an established stand of grass. Sowing date triggers the start of plant growth simulation. Separating the beginning of the simulation from the sowing date allows the model to predict soil conditions on sowing day, which may be helpful if the user has limited information about the actual co nditions existing at the site on that day. The rate step is run once on each day of the simulation and calculates the rate variables (variables representi ng the amount of change in a st ate variable o ccurring over a specified period of time – usually one day or le ss) for the current day of the simulation. For each simulated day, prior to running th e CROPGRO plant template, the date, daily minimum and maximum temperatures, daily rain fall, and total daily solar radiation are read from the weather file, and the Xa nd T-files are checked for any management operations (e.g. planting, irriga tion, harvest, or pest damage ) for the day. The weather and management information is fed into the soil processes module, pr edicting the rates of change of available soil water, NH4 +, and NO3 -. The weather data along with plant and soil information from the end of the prev ious day is used to calculate daily evapotranspiration. Predicting plant phenology or stage of maturity is a key component of CROPGRO. Several parameters such as partitioning of new growth between plant organs (leaf, stem, roots), and daylength sensitivity of developm ent are dependent on or may vary with the stage of maturity. Stage of maturity is a function of accumulated physiological time which is a combination of time and temperat ure. Cardinal temperatures (minimum, optimal, and maximum) for various stages of development of each crop are described in the SPE file. Generally, there is no developm ent at temperatures below the minimum or above the maximum cardinal temperatures wherea s plants mature at a faster rate as air

PAGE 40

26 temperature nears the optimal range. This ra te of change/progression towards the next stage can be altered by daylength and water st ress. Progress during vegetative growth is typified by progressive increase in the num ber of leaves per plant (V-stage) and, beginning with floral induction, by progressi on through a series of reproductive stages (R-stage). The daily rate of pest damage is also calcu lated in the rate step. Information on the extent of damage is listed by the user in the “T-file” and CROPGRO uses the previous day’s plant mass and leaf area to calculate “actual” damage. C odes in the Pest file allow the user to specify the amount (kg DM ha-1) or proportion of leaf and stem to be removed or the reduction in assimilate production due to disease or pest damage. As with all other rate variables, the pest damage rates are not deducted from the existing plant mass until the integration step which signals the end of the current day. This ensures that all new rates are based on the same conditions – the pl ant mass at the end of the previous day. The calculation of photosynthetic rate presents an interesting contrast with animalbased models. Many animal nutrition models are sink-driven, predicting nutrient intake as a function of body weight and animal pe rformance while, in CROPGRO, the supply of assimilate, a function of absorbed solar ra diation and photosynthesis, determines plant weight and performance (source-driven). On e exception is that some plant species may exhibit a “juvenile” period where lim ited seedling demand can feed back on photosynthate production. CROPGRO offers tw o options for predicting photosynthesis: a daily canopy photosynthesis option and an hourly leaf-level phot osynthesis option. The daily canopy photosynthesis option is the simpler of the methods and uses an asymptotic exponential response to daily solar radiation to calculat e the potential daily

PAGE 41

27 photosynthetic rate. This pattern of daily pho tosynthetic response to light is defined by two parameters from the species file; th e maximum canopy photosynthesis rate and the amount of PAR at which photosynthesis is 63% of maximum. Estimated light interception of the canopy is a function of the predicted LAI on the previous day and the canopy light extinction coefficient specified in the SPE file. The hourly leaf-level photosynthesis option is more mechanistic (b ased on processes and stoichiometry) but more mathematically complex. The leaf–level option uses the hedgero w approach described in Boote and Pickering (1994) to estimate potential hourly photosynthetic rates for sunlit and also shaded leaf area portions of the canopy. The hedgerow approach uses parameters describing canopy shape, height and width, leaf angle, row wi dth and direction, latitude of the site, day of year and time of day al ong with the predicted LAI to estimate light absorption by sunlit versus shaded leaves. Hourly distribution of solar radiation and temperature are estimated from the daily valu es provided in the weather file and further divided into direct and diffuse components. Potential hourly leaf photosynthetic rate is calculated using an asymptotic exponential equation where quantum efficiency (initial photosynthetic response at low light) sets the initial slop e of the response and the maximum potential leaf photosynthetic rate is the asymptote. The photosynthetic rates for the sunlit and shaded l eaves are multiplied by their respective LAIs and summed each hour to calculate the hour ly canopy photosynthetic rate. The 24 hourly rates are integrated to yield the daily photosynthetic rate. In both the daily canopy and hourly leaf photosynthesis options, the poten tial photosynthetic rate may be adjusted for cultivar

PAGE 42

28 differences, temperature, leaf N concentrati on, leaf thickness (specifi c leaf weight, SLW), atmospheric CO2 concentration, and an incomp lete canopy (light absorption). In the integration step, updated maturity stages are calculated by computing the daily rate of change for each day and adding this to the previous day’s stage rating. All other rates are processed similarly to calculate the new day’s state variables. Much of the mechanics of the integration step is an exercise in carbohydrate (CH2O) accounting and allocation. Potentially available CH2O from stored reserves is calculated and added to the daily photosynthate production to determine the maximum amount of CH2O available for the day. The day’s maintenan ce respiration costs and any a ssimilate loss due to pest damage are subtracted from the total, the remainder being the amount of CH2O available for nutrient uptake and growth. Potential CH2O demand for seed and shell growth, CH2O cost per g of new vegetative growth (based on proportions of leaf and stem predicted by the new V-stage), as well as potential N dema nd to “refill” N that has been mobilized from old tissue are calculated and subtracted from the remaining available CH2O. The amount of new growth that can be produced from the available CH2O depends on the cost of that new growth which is a function of its composition. Composition of total new growth is, in turn, determined by the partitioning of new growth between the different plant organs (leaf, stem, roots, seeds, etc.) and the protein, carbohydrate (comprised of cell wall and starch), lipid, lig nin, organic acid, and ash concentration of each organ’s new growth. Parameters de scribing partitioning to organs and organ composition are listed in the species file. Coefficients develope d by Penning deVries et al. (1974) describing the cost to assemble each of these components (both direct cost of C

PAGE 43

29 for C-skeletons as well as the energy used in the biochemical pathways to form them) expressed in glucose equivalents are used along with the composition parameters to calculate the glucose cost of new tissues. Cr ude protein or N concen trations of new leaf, stem, and root growth can vary within a ra nge of values set by three parameters; a maximum concentration for new growth, a “normal” growth concentration and a residual concentration left after senescen ce. If there is adequate CH2O available but available N is limiting, new growth can occur at reduced N concentrations. Nitrogen demand can be met by two differe nt sources; “Actual” N uptake by the plant, and if this uptake cannot meet the N demand for new growth, N mobilized from vegetative tissues. Once the balance of N uptake and mobilization are calculated, the CH2O cost of N uptake and any N mobilizati on is subtracted from the remaining available CH2O to update the energy budget. If the crop is a legume, N-fixation is estimated, the N added to the available N pool, and the CH2O cost subtracted from the available CH2O. The remaining CH2O is allocated to new growth; first to seed and shell, then to vegetative growth. The increased root length asso ciated with the day’s pred icted new root growth is calculated. The total root length and its dist ribution are used in de termining nutrient and water uptake on the succeeding day. Losses are also calculated; the day’s predicted senescence of leaves, stems, and roots are estimated as well as any damage due to freezing temperatures. Finally, all of the gains in new growth a nd losses due to senescence, pest damage, and frost are added to the previous day’ s pools of tissue mass and a new total is calculated representing the plant mass present at the end of the current day. This ends the

PAGE 44

30 “day” and after outputs are prin ted to their respective files, the integration step ends and the rate step is repeated the next day. The rate and integration steps are repeated daily until the simulation is terminated on the harvest date specified in the X-file or when a killing freeze occurs. At this time the final step is executed to finish printing th e output files. These are the basic mechanics and information flow within CROPGRO. The structure has been shown to work well with annual grain crops and legumes as we ll as tomato. In adapting the model to simulate the growth of perennial tropical gra sses, accurate parameters must be developed for the species file. Any plant processes uni que to these plants that are not already included in the model may require re-definit ion of some of the parameters or even changes and additions to the model code. Model Evaluation The ultimate measure of a model’s performance is the user’s satisfaction with both the accuracy of predictions and overall utility of the model. Understandably, such a measure is difficult to quantify and is relevant only to the user that generated the rating. Statistical approaches to quan tify the accuracy of model pred ictions provide standardized measures of model performance. Unfort unately, even these methods do not provide completely clear-cut conclusions about the accu racy of model predictions. Use of vague terms like “fairly close” in instructions for in terpreting various measur es impart an air of skepticism around the use of some of these methods. Given these caveats, the use of several different measures of performance to evaluate a model may present a more complete picture of model performance than any single measure and allow the user to weight individual results acco rding to their priorities.

PAGE 45

31 Two measures that are commonly reporte d in the literature are the sample correlation coefficient r and coefficient of determination r2. The correlation coefficient provides a measure of the linear relationship or closeness between predicted and observed values. Interpretation of r is quite general. An r of 1.0 indicates perf ect prediction by the model with positive values of r indicating some level of a positive correlation between the predicted values (Pi) and observed values (Oi). Conversely, an r of 0.0 indicates no correlation of the model to reality whatsoev er and negative values indicate an inverse relationship. The coefficient of determinati on is informally described as the proportion of the variance of the observe d values that can be accounted for by the model. This measure has more utility in that it pres ents an idea of how thoroughly the model represents the system. Statistical analyses demonstrating the le vel of significance of r only proves that a linear relationship w ith a non-zero slope exists between Pi and Oi (Snedecor and Cochran, 1989). Th e validity of this conclusion can come into question if Pi and Oi do not meet the underlying assumptions requi red for the particular analysis used (Willmott, 1981). In spite of their popularit y, these measures provi de little detail to characterize the relationship between Pi and Oi. A simple method of visualizi ng the relationship between Pi and Oi is plotting a scatterplot of Pi (Y-axis) and Oi (X-axis), relative to a line designating a 1:1 relationship. While not quantifiable, some relationships (e.g. consistent underprediction) become apparent. Scatterplots al so provide a common sense check for more sophisticated methods of evaluation. If result s of a test do not appear consis tent with the results of the scatterplot, the test should be re-e valuated. The relationship between Pi and Oi presented in the scatterplot can be quantified using lin ear regression. The sl ope of the regression

PAGE 46

32 line ( a ) and its Y-intercept ( b ) may provide evidence of systematic error in the model, providing quantities that can be compared across models. A slope of 1.0 with a Yintercept equal to 0.0 in dicates perfect fit of the model predictions. These results along with the means ( P and O) and standard deviations of th e predicted values and observed values should be considered for their own me rit as well as their use in calculating other measures when evaluating model performance. Difference measures, derived fr om the fundamental quantity ii(P-O) (Willmott, 1982), build on the statistical measures listed ab ove to quantify bias and average error. Root mean squared error (RMSE) desc ribes the average difference between Pi and Oi. n 2 ii 1(P-O) RMSE ni (Eq. 2-1) Also, RMSE can be readily compared agai nst the mean of the observed values for comparison of relative error. Both RMSE a nd its square (mean square error or MSE) can be subdivided into systematic (RMSEs and MSEs) and unsystematic (RMSEu and MSEu) components (Willmott, 1981): n 2 ii 1(P-O) MSE ni n 2 ii 1 s(P-O) MSE ni n 2 ii 1 u(P-P) MSE ni (Eq. 2-2) where n= the number of pairs of pred icted and observed data, and iiP=O+ab. When the systematic component is minimized, the m odel is predicting at its maximum possible accuracy and the primary source of error is not model-related. An alternative presentation is offered by Roseler et al. (1997) where mean square prediction error (MSPE), which has the same mathematical de finition as MSE (Neter et al., 1990; Roseler

PAGE 47

33 et al., 1997), is considered as the sum of three components: mean bias 2(O-P), line bias2 2 p(1-b) S and random variation around the regression line 2 2 2 o(1-) Sr where 2 p S and 2 oS are the variances of the predicted and observed values. These measures provide insight not just on the magnitude of error but also hi nt at the poten tial sources of error. Willmott (1981; 1982) proposed another measure of model performance that he called an “index of agreement”. This is re ferred to elsewhere as the d-index. The dindex describes the degree to which the obs erved data are accurately estimated by the predicted data. More formally, it specifies the degree to which the deviation of the observed data around O corresponds with the deviati on of the predicted data around O, both in magnitude and sign. n 2 ii i=1 n 2 ii i=1(P-O) d=1(P-O) (Eq. 2-3) where iiP=P-O and iiO=O-O. Potential values of d range from 0 to 1, with 1.0 indicating perfect agreement between predic ted and observed data and 0.0 indicating complete disagreement. The sole assumption is that O is free of error so that all error is contained in iP and iO. The equation can be rewritten as n 2 ii i=1n•MSE d=1(P-O) (Eq. 2-4) for simplified calculation when MSE is known. The innovation of the d-index is that it responds to both differences between predic ted and observed data as well as some

PAGE 48

34 changes in proportionality (Willmott, 1981). The d-index is an improvement on the simple “r”; still, it is not an absolute measure of performance. As with the aforementioned methods, the d-index should be evaluated in the context of knowledge of natural variations in the system being modeled, the capabilities of the model, and an awareness of the amount of poten tial error in the obs erved values used in the comparison. No one of these approaches will be best in all situations, reviewing several of these measures together will provide a more comple te description of model performance. The results should also be viewed in the context of the intended use of the model. If the model is to be used to demonstrate the respons e to a change in the environment to a class of students, a model that predicts a response of the correct directi on but severely underor over-predicts the magnitude may be preferable to a more accurate model if the latter is more difficult for the students to use. Users must decide for themselves what level of performance is acceptable. Likewise, individuals will have their own views of which approach is most appropri ate to their interests.

PAGE 49

35 CHAPTER 3 BAHIAGRASS GROWTH STUDY Introduction There has been a resurgence of interest in the cause of winter dormancy in tropical perennial grasses, particularly bahiagrass (M islevy, 1998; Gates et al ., 2001; Sinclair et al., 2003). To date, the primary emphasis of the research has been to identify the conditions triggering dormancy. Traditionally, dormancy was thought to be related to the cooler temperatures of fall and winter; howev er, daylength has recently been implicated as the triggering condition (G ates et al., 2001; Sinclair et al., 2003). Along with identifying the cause, quantifying the effects of dormancy may help us identify specific characteristics associated with dormanc y to both aid in identifying non-dormant individuals as well as help develop management strategies to promote higher yield and longevity of both current and new, non-dor mant bahiagrass varieties that may be developed. Sinclair et al. (2003) presented growth and composition data at the organ level (leaf, stem, and below-ground material) character izing relative differenc es in growth and composition between plants grown under norma l or extended photoperi ods. Their data were quantified for each harvest, at 4 to 5 week intervals. Information quantifying changes within a regrowth period, however, is scarce. More detailed growth analyses conducted during the transition into dormancy ma y help identify some of the mechanisms involved in the reduction of herbage growth a ssociated with dormanc y. Our objective for this study was to document, in detail, weekly pa tterns of plant growth in late summer and

PAGE 50

36 fall regrowth periods with concurrent meas urements of leaf and canopy photosynthesis. The purpose of this information is to help “fill in” some of the detail missed in other studies, and needed for perennial forage crop model development. Materials and Methods This research was conducted at the Plant and Soil Science Field Teaching Laboratory at the campus of the University of Florida, Gainesville (29 38 N, 82 22 W) on an established bahiagrass sod during the summer and fall of 2001. Based on the age of the stand, fine leaf texture and abunda nt seedhead production in June, the variety of bahiagrass was assumed to be Pens acola. The soil was an Arredondo fine sand (loamy, siliceous, hyperthermic Grossarenic Paleudult). The experimental design was a randomized complete block four replications each being 190-m2 plots (PLOT). Treatments were two 8-wk regrowth peri ods (PER) (18 July – 12 September or 12 September – 7 November). The crop was harvested to a stubble height of 10 cm every 8 wk starting 21 May. The 18 July harvest serv ed as the staging harvest, establishing the initial conditions and base stubble mass for the first growth period. A single sod core sample was taken weekly (WEEK) from each of the four plots beginning on 20 July (Table 3-1). Plant height (distance from th e soil surface to the point where the leaves curved over and began to hang down) was meas ured at six locations within each plot on the same days that the sod cores were sampled. While the bahiagrass had been establishe d for several years, it had not been fertilized or irrigated regularly in recent year s. During our study all plots were fertilized with a commercial blended 16-4-8 fertilizer including trace nutrients and slow-release N at 78 kg N ha-1, 10 kg P ha-1, and 37 kg K ha-1 every 8 wk beginning on 13 April. Irrigation was provided as needed to prevent water limitation of plant growth via portable

PAGE 51

37 impact sprinklers (see Table 3-2 for combined rainfall + irrigation). Weather data (total daily PAR, minimum. maximum, and average air temperature and total daily rainfall) were recorded on an automatic datalogger ( CR10, Campbell Scientific, Inc. Logan, UT) and are summarized in Table 3-2. The sod cores measured 20 cm by 35 cm by 15 cm deep and were dug by hand using a pair of narrow-bladed shovels (Figur e 3-1). Loose soil was shaken by hand from the sod core and any loose bits of roots and other plant material we re recovered. All material was placed in a plastic bag and immedi ately placed in a cooler for transport to the laboratory. At the labor atory, each sample was thor oughly washed with a garden hose over a 2-mm sieve to dislodge soil. Rins ed samples were placed in sealed plastic bags and refrigerated until processed. Roots were trimmed from the stolons usi ng hand clippers, placed in a paper bag and transferred to a 55C forced-air oven. A subsample of the remaining plant material was set aside for detailed analyses. Both the remaining sample and subsample were separated into live leaf, stem, stolon, and d ead leaf components. Live leaves were separated from stems at the ligul e (if there was one) or where the leaf emerged from the stem (if the leaf had no ligule). Using th is separation methodology, the leaf sheath is included in the stem fraction (F igure 3-2). Stems were separa ted from stolons at a point where they naturally broke by hand. Dead leav es were peeled from the tillers and, thus, included dead sheath material. Material in the subsample was analyzed fo r leaf and stem areas (one sided only) using a LI-COR model 3100 leaf area meter (LI-COR Inc., Lincoln, NE). Stems were scanned intact (Figure 3-2), and not dissected into individual immature leaves. The

PAGE 52

38 number of stems (tillers) per stolon was counted and the number of leaves per tiller was recorded as the vegetative stag e (V-stage) of each tiller. The presence of seedheads in the subsamples were noted but as there were very few, no additional measures of reproductive stage were recorded. Upon completion of separation and meas urements, the plant components were dried until reaching a constant weight in a 55C forced-air oven. Care was taken to remove loose sand before weighing. Leaf, stem, stolon, and root dry matter (DM) mass (kg DM ha-1) was calculated from the combined sample and subsample masses of leaf, stem, stolon, and root, respectively, and the land area of each core. Specific leaf area (SLA) (m2 leaf kg-1 leaf) was calculated from the meas ured leaf area and leaf mass for each subsample. Leaf area index (LAI) (m2 leaf m-2 land) was then calculated by multiplying the SLA by the total leaf mass m-2 from each core. A “green area index” (GrAI) representing the total photosynthetic area per area of soil su rface was calculated using the sum of the leaf and stem area indi ces. The V-stage was calculated for each plot by averaging the V-stage of all of the tillers in the subsample for each plot. The net accumulation of each component as we ll as net change in V-stage ( Leaf, Stem, Stolon, and Root mass, V-stage) was calculated for each period by subtracting the WEEK 0 (stubble) mass from the WEEK 8 (final) mass. Statistical analyses of the growth data were performed using the Mixed procedure of SAS (SAS Institute In c., 1987) with the model: Yijk = + Ai + Bj + AiBj + Ck + AiCk + BjCk + eijk where was the population mean, A was PL OT, B was PER, C was WEEK, and e was the residual error for i=4, j=2, k=8. The net accumulation data were analyzed using a

PAGE 53

39 simplified form of the same model; Yij = + Ai + Bj + eij. PLOT, and all of its interactions were assumed to be random effects and therefore appeared in the SAS random statement in the order presented above Growth period (PER ), WEEK, and their interaction (PER WEEK) were fixed effects. Means sepa ration for PER was directly from the ANOVA. Orthogonal c ontrasts were used to qualify significant responses to WEEK and PER WEEK. The =0.10 level was selected as the threshold for determining the significance of all effects and contrasts. Concurrent with the growth measur ements, leaf and canopy photosynthesis measurements were recorded four times dur ing each growth period. Due to weather constraints, these measurements were not evenly allocated throughout each growth period, nor did they occur at the same time during each period. Leaf and canopy photosynthesis measurements were made at midday using a LI-COR LI-6200 portable photosynthesis system (LI-COR Inc., Lincol n, NE). Leaf photosynthetic (or carbon exchange rate CER) measurements were made on fully expanded, healthy leaves under full sun conditions (PAR >1600 mol m-2 s-1) using a 0.25-L chamber. Photosynthesis was measured for three leaves per plot on each sampling date. Two 15-sec measurements of carbon exchange ra te (CER) (mol CO2 m-2 leaf s-1), stomatal conductance (mol m-2 leaf s-1), and internal CO2 concentration (L L-1) were recorded for each leaf. For canopy photosynthesis measurements, the leaf chambe r was placed “open” inside an aluminumframe, clear plastic enclosure. The frame enclosed a land area of 0.56 m2, with a total volume of 0.49 m3. Canopy CER (mol CO2 m-2 land s-1) measurements were made under four levels of light varying from full s un to dark. The light level in the chamber was regulated by placing cloths of varying opaqueness over the chamber. Approximate

PAGE 54

40 light levels were: PAR > 1500 mol m-2 s-1 (full sun), 600-800 mol m-2 s-1, 200-400 mol m-2 s-1, and 0 mol m-2 s-1 (dark). Three 16-second measurements of carbon exchange rate (CER) (mol CO2 m-2 land s-1) were recorded at each light level. The canopy enclosure was opened between each light level measurement to let the humidity and [CO2] in the enclosure equilibrate with the atmosphere. Net CER in full darkness was considered to represent canopy + root + soil (dark) respir ation. Gross canopy photosynthesis (mol CO2 m-2 land s-1) for each light level wa s calculated by adding the absolute value of the dark respiration to the measured net photosynthesis for each light level. We fit the canopy light response data to the asymptotic exponential model (Boote et al., 1985): max(*/) max*[1]QEPARPPPe (Eq. 3-1) using TableCurve 2D v4 software (Ja ndel Scientific Software, 1996), where P = canopy gross photosynthetic rate (mol CO2 m-2 s-1), Pmax = maximum photosynthetic rate in saturating light (mol CO2 m-2 s-1), QE = quantum efficiency or initial slope of the CO2 assimilation : incident PAR response ( mol CO2 mol-1 absorbed photons), and PAR = photosynthetically active radiation (mol photons m-2 s-1). We solved for Pmax and QE and used the resulting values to estimate gross canopy photosyntheti c rate at a light intensity of 1500 mol photons PAR m-2 s-1. By expressing photosynthetic rate for a common light intensity we could compare the different treatments and days without the variation due to changing light levels as the days and seasons progressed. We also attempted to predict a light -saturated leaf photosynthetic rate (A sat ) from the canopy gross photosynthesis data. The measured leaf and stem areas as well as

PAGE 55

41 canopy gross photosynthesis and correspondi ng PAR measurements from the canopy light response measurements were input into the hedgerow photosynthesis model of Boote and Pickering (1994), programmed in SAS, then PROC NLIN in SAS (SAS Institute Inc., 1987) was used to solve for A sat using an asymptotic exponential function and outputs from the hedgerow model. Statistical analyses of the leaf photosynthesis data were performed using the Mixed procedure of SAS (SAS Institu te Inc., 1987) with the model: Yijk = + Ai + Bj + AiBj + C + AiC + BjC + AiBjC + Dk + AiDk + BjDk + AiBjDk + CDk + AiCDk + BjCDk + eijk where was the population mean, A was PLOT, B was PER, C was day of period or regrowth (DAY), D was leaf number (LEAF) th ree leaves were measured in each plot and e was the residual error for i=4, j=2, and k= 3. PLOT and all PLOT interactions were assumed to be random effects and therefore appeared in the SAS random statement, in the order presented above. Growth period (PER), LEAF, DAY, and their interactions were considered to be fixed effects. Since measurements were not taken every week or even on the same day of the week, DAY was tr eated as a continuous variable and entered as a covariate. Because there was only a single value for A sat canopy P1500, and canopy respiration for each plot on each sa mpling day, LEAF was not included in the analysis of these variables and a reduced version of the model was used for these variables: Yijk = + Ai + Bj + AiBj + C + AiC + BjC + eij Least squares means were calculated for PER. Means separation for PER was directly

PAGE 56

42 from the ANOVA. The =0.10 level was selected as the threshold for determining the significance of all effects and contrasts. Results and Discussion Plant Growth Statistically, total plant growth ( Total Plant Mass) was greater for PER 2 than PER 1 (Table 3-3). The dramatic loss of total plant mass in PER 1 masked a net increase in leaf mass, overshadowing the fact that canopy growth was greater in PER 1 than PER 2. The loss of total plant mass was driven al most entirely by a decrease in root mass in PER 1. In PER 2 root mass stabilized (Table 3-3) and a net gain of total plant mass was realized (Figure 3-3). Initial root mass for PER 1 was high at 11 500 kg DM ha-1 but dropped to less than 1/3 of the original mass by WEEK 7 of PER 1 (Figure 3-4). The consistent decline during PER 1 and the relatively constant root mass from the end of PER 1 through PER 2 tends to dismiss random sampling error as the cause of this loss. Alternatively, the Spartan maintenance of the site in past years may have played a role in this behavior. There had been no fertilizer or irrigation applications over se veral years and the extensive root system may have developed to more thoroughly mine the soil for water and nutrients. At the time we started measuring plant growth, irrigati on had been available for three months and a second fertilizer app lication had just been applied. As PER 1 progressed, nutrients may have become availa ble in sufficient concentrations that plant needs could be met with a less extensive root system. Excess root mass may have been mobilized for new shoot growth or merely allo wed to die, resulting in the pattern of loss observed. In contrast to this loss in PER 1, root mass remained relatively constant in PER 2. There was a slight increase in root mass in WEEK 6 (Figure 3-4) which

PAGE 57

43 coincided with a period of increasing stolon mass and, t hus, could be related to a dormancy–induced change in priority of assimilate partitioning. A more obvious signal of approaching dorma ncy may be seen in the pattern of stolon growth. Throughout PER 1 stolon ma ss remained unchanged at approximately 4700 kg DM ha-1 (Figure 3-5). Stolon mass increa sed in PER2, peaking at 8980 kg DM ha-1 on WEEK 7 (Figure 3-5). This change in growth pattern, as evidenced by the significant WEEK and PER WEEK intera ction effects, resulted in greater stolon mass in PER 2 (Table 3-3). This late-seas on shift in partitioni ng of growth towards storage tissue may be part of a dormancy re sponse to shorter daylengths. Increased allocation of growth to stolons may have cont ributed to the lower increment of leaf and stem mass observed in PER 2 (Table 3-3). The combination of increasing stolon mass and stable-to-increasing root mass in PER 2 is in contrast to the observations of Sinclair et al. (2003) who reported st eady or decreasing below-ground plant mass for Pensacola bahiagrass between 22 Sep. 1999 and 1 Dec. 1999 sampling dates at Ona, FL. Our longer, 8-wk, harvest interval may have allowed greater accumulation of stolon mass between harvests than the 4 to 5-week interval of Sinclair et al. (2003). The management history of the two sites could also have ha d an influence, as judged by the exceptional pattern of root growth ob served at our site. Changes not only in leaf and stem mass (Table 3-3), but also changes in the “character” of the canopy were observed. Shoot growth during the fall season (PER 2) was slower than in summer (PER 1) resu lting in less stem a nd leaf dry weight accumulation (Table 3-3, Figures 3-6 & 3-7) as well as fewer new mature leaves (Figure 3-8) at the end of PER 2. Stem weights we re lower for all weeks in PER 2 compared to

PAGE 58

44 PER 1 but the decrease in stem mass was not different between periods (Table 3-3). The linear decline in stem mass in both periods and even the “bump” in stem mass in PER 1 (Figure 3-6) was, at least in part, an artif act of the partitioning scheme employed. Our partitioning strategy grouped developing leaves, s till encased in the sheath, with stems. Once the leaves began elongating and emerged fr om the sheath, the le aves became part of the leaf mass and the fraction of their mass th at had previously been developing in the sheath was lost from the stem mass. Very little stem elongation was observed except for the few tillers that developed seedheads, leav ing little opportunity to increase stem mass as the plant matured. Development of leaves in the fall regrowth period (PER 2) was quite different from that in PER 1. Leaf mass increased in a qua dratic manner (Table 3-3) to peak on WEEK 6 of both periods (Figure 3-7) with average leaf mass and leaf mass slightly lower in PER 2. In contrast, the V-stage in PER 2 was only 3.41 le aves, less than half of the 7.65 leaves added in PER 1 (Table 3-3, Fi gure 3-8). Average LAI followed leaf mass more closely than V-stage, and a quad ratic progression in LAI development was observed in both periods. The LAI in PER 1 wa s higher than that in PER 2 (Table 3-3). Despite the slower development, the initial and final LAI were the same for both periods (Figure 3-9). Thus, the same final LAI was achieved in PER 2 with only half as many mature leaves as were observed in PER 1. It should be noted that on ly leaves with ligules were included in the V-stage count but all l eaf blade material extending from the leaf sheath was included in the LAI measurements. Accordingly, the SLA (leaf ar ea per g of leaf mass) was slightly larger, indicating thinner leaves, in PER 2, although a quadrat ic decrease in SLA was observed in both

PAGE 59

45 regrowth periods (Table 3-3, Figure 3-10) Ghannoum et al. (2001a) also observed seasonal effects on SLA in a variety of perennial C4 grasses grown in summer and winter; however, the response was species-dependent, positive in some cases and negative in others. Although peak daily so lar radiation was not recorded total daily solar radiation was lower in PER 2 (Table 3-2). Our SLA valu es are in accord with the range of values reported by Boote et al (1999) (88 to 108 cm2 g-1) for greenhouse-grown bahiagrass at this location. One might expect these levels to be lower (thicker leaves) as high SLAs are generally associated with shaded leaves and leaves grown under reduced light levels such as in greenhouses while our plants were grow n outdoors in full sun. Our values are quite low compared to SLAs reported for other perennial C4 grasses (Ghannoum et al., 2001a, 2001b) and even C3 grasses (Ryser and Wahl, 2001). The mean SLA reported for 11 NADP-ME-type C4 grasses grown inside a glasshouse in summer, with midday PAR levels averaging 860 mol m-2 s-1, was 314 cm2 g-1, almost five times the level observed in our study (Ghannoum et al., 2001a). However, their plants were harvested onl y 46 d (approx 6.5 wk) after planting, much younger than the average age of th e leaves on our plants. In c ontrast, the harvest interval employed by Boote et al. (1999) was longer than the current 8 wk. The fine-leaf structure of Pensacola bahiagrass may predispose this cu ltivar to have a lower than average SLA; however, the low SLA values merit further measurements on this species and its cultivars grown in other locations. Similarly, the LAI values that we are reporting are lower than those observed by others. At the extreme are LAI values above 8.0 reported for bahiagrass by Agata (1985a; 1985b), which are considerably highe r than our values of 1.75 and 1.67 for

PAGE 60

46 8-week regrowth in PER 1 and PER 2, respectiv ely (Figure 3-9), or even our GrAI values which include stem area as well as leaf area (Figure 3-11) Unfortunately, the methodology used by Agata (1985a, 1985b) to de termine LAI was not clear, hindering any further comparisons. Othe r reports give considerably lower LAI values. Pedreira and Brown (1996b) reported LAI for stubble a nd 13-d regrowth for three populations of bahiagrass grown in the field near Athens, GA. Reported values were averages for two cutting heights (3.5 and 10 cm). This st ubble LAI (comparable to our WEEK 0 values) ranged from 0.42 for selection T14 in August to 1.68 for Pensacola bahiagrass in July while regrowth LAI values ranged from 1.67 for selection T14 in August to 2.30 for Tifton 9 in July. These values compare very favorably to our values of 0.19 for WEEK 0 (stubble) and 1.16 for WEEK 2 in PER 1 (Figure 3-9). Methodology for the Athens study was similar to ours in that leaf area m easurements were based on leaf lamina only; however, their sample size was much sm aller (20 leaves vs. 100-500 leaves per replication in our study). As the stem tissue is also green, leaf + stem or “green” area index (GrAI) may present a more accurate measure of photosynthetic ally active plant area than LAI alone. As with LAI, GrAI was higher for PER 1 than PER 2 at 2.49 and 1.85, respectively (Table 3-3). Orthogonal contrasts showed th e relationship between GrAI and WEEK to be cubic (Table 3-3), this is lik ely an artifact of the variati on in stem mass resulting from our partitioning scheme and may not be a biolog ically relevant pattern. The time-series change in GrAI (Figure 3-11) clearly illustrates, more so than LAI, a considerably slower increase and overall lower photosynthetic ar ea throughout most of PER 2. Like LAI,

PAGE 61

47 though, initial GrAI was the same for both peri ods and final GrAI were also much closer than for the middle of the regrowth periods. The slower leaf growth rate cannot be attr ibuted to differences in initial leaf mass and initial leaf area as neither differed between periods, alth ough the slower growth rate would reinforce itself through lower leaf mass and LAI once regrowth began. The slower development of LAI and GrAI may have decr eased potential photosynthesis during much of PER 2; however, final LAI values were similar or identical for both periods (Figure 3-9). Cooler temperatures and lower solar radi ation levels (Table 3-2) likely were major factors reducing fall growth rates. The increased partitioning of growth to stolon tissue could also have reduced leaf grow th in the second half of PER 2. Photosynthesis Some caveats apply to the photosynthesis resu lts. First, due to a combination of equipment repair and availability issues, photosynthesis measurements for PER 1 were made with a different LI-COR LI-6200 than was used in PER 2. However, both instruments were calibrated using the same reference gas and procedures. Second, the complete set of leaf photosynthetic measuremen ts (two 15-s values for 3 leaves per plot) included a few values that were not physiologically realistic. The leaf data were analyzed to identify and remove outlying data point s. Of the 197 photosynthesis measurements, one was more than 1.5 interquartile ranges from the 75th percentile and one was more than 1.5 interquartile ranges below the 25th percentile; both data points were removed from the analysis These were the only data points removed from the analysis. The values removed were quite extreme (gro ss photosynthesis levels of 70.35 and -20.47 mol CO2 m-2 s-1) and may have been the result of air leaking into the leaf chamber during measurement, beginning measurements before CO2 concentration in the chamber

PAGE 62

48 had begun to drop, or simply due to high sensitiv ity to the very small leaf area in the leaf chamber (approximately 1 cm2) and a correspondingly small drop in [CO2] over the measurement period causing excessive variati on on measurements. Under the reduced dataset (n=195), the covariate, DAY, was signi ficant (Table 3-3) w ith leaf photosynthetic rate being highest during early regrowth (Fi gure 3-12). Despite lo wer temperatures in PER 2 and a positive leaf temperature to le af photosynthesis correlation coefficient of r=0.45, PER did not have a significant effect on leaf photosynthesis. Our measured leaf photosynthesis values of 31.0 and 26.6 mol CO2 m-2 s-1 for PER 1 and PER 2, respectively, fall well within th e range reported by Boote et al. (1999) (24.8 – 35.2 mol CO2 m-2 s-1) and Fritschi et al. ( 1999) (19.0 – 35.4 mol CO2 m-2 s-1) for greenhousegrown bahiagrass at 350 L L-1 CO2 concentration at this site. As expected, predicted leaf A sat was considerably higher than the measured leaf photosynthesis values (Table 3-3). Like meas ured leaf photosynthesis, the predicted Asat values were not different between periods however, unlike the measured data, DAY did not affect A sat This would indicate that the maxi mum potential leaf photosynthetic rate remained the same over the temperature rang e experienced in this study. There is precedent for this. Asat has been shown frequently to decrease rapidly below 20C (reviewed by Long [1983]); however, the lowest temperat ure recorded in the canopy chamber during photosynthesis measurements was 27C, considerably higher, where the impact may be slight and difficult to dis cern. This might also help explain why temperature accounted for such a small proporti on of the variation in leaf photosynthetic rate. Although there is no way to ve rify the accuracy of our predicted A sat using the data

PAGE 63

49 we collected, the values seem within reason as a handful of measured values were at or above the predicted rates. To allow us to compare photosynthetic performance across sampling dates and periods despite the va rying light conditions, the canopy photos ynthesis data were fit to an asymptotic exponential function and the results used to pr edict canopy photosynthesis at 1500 mol PAR m-2 s-1 (P1500). Model fit was good with an r2> 0.99 for most plots and sample days and the lowest r2 for a plot/day was 0.89. An alysis of the adjusted canopy data showed a higher canopy gro ss photosynthetic rate in PER 1 than PER 2 (Table 3-3, Figure 3-13). This is consistent with the greater plant mass (parti cularly leaf and stem) observed in PER 1. Interestingly, when we fitted a regression model using daily maximum temperature (Tmax), leaf + stem mass (Greenkg), GrAI, and SLA to the P1500 data, Tmax had the best fit (r2= 0.48) and the addition of either Greenkg or GrAI did not significantly improve the fit of the model. The PER 2 average of 43.0 mol CO2 m-2 land s-1 was within the range of 31.6 to 47.1 mol CO2 m-2 s-1 reported by Boote et al. (1999) for P1500 in greenhouse-grown bahiagrass ca nopy gross photosynthesis at 350 L CO2 L-1 CO2, whereas our PER 1 rate of 55.9 mol CO2 m-2 s-1 was outside this range but below the highest rate of 60.7 mol CO2 m-2 s-1 reported by Fritschi et al. (1999) for P1500 of greenhouse-grown bahiagrass in the es tablishment year. As our plants were grown under full sun, a higher photosyntheti c rate than for greenhouse-grown plants would be expected. That Fritschi observed hi gher rates may be rela ted to the differences in the age of the stands, if th e rates are different at all. Concurrent with the higher photosynthetic rate in PER 1 were higher canopy + root + soil respiration levels. This measurement was based on the CO2 exchange rate

PAGE 64

50 measured in total darkness. Respiration ra te is dependent on both the amount of tissue respiring as well as the temperature. In re gression analysis usi ng the “Backwards” option in PROC REG of SAS (SAS Institute Inc ., 1987), both Tair (air temperature) and Greenkg contributed significantly to the model and, combined, they could account for 87% of the variation observed in respiration. Canopy + root + soil respiration rates were 24.0 mol CO2 m-2 land s-1 in PER 1 and 13.9 mol CO2 m-2 s-1 in PER 2 (Table 3-3), slightly higher than the levels reported by Boote et al. (1999) but within the range of values reported by Fritschi et al. ( 1999) for bahiagrass grown at 350 L L-1 CO2. Respiration rates also vari ed by day of regrowth (Tab le 3-3) but since DAY was a covariate, orthogonal contrasts could not be used to discern a pa ttern of response. Conclusions Winter dormancy, the seasonal depressi on of canopy growth, in bahiagrass often has been attributed to a decr ease in temperature. More recently, daylength has been identified as having a role in triggeri ng dormancy (Mislevy, 1998; Gates et al., 2001; Sinclair et al., 2003). The objective of our study was not to identify the cause of dormancy but rather to quantify growth a nd photosynthesis during the late summer and fall in more detail than previous studies. Our study points out se veral changes in the pattern of plant growth a nd photosynthetic rate that may illuminate parts of the underlying mechanism of dormancy. Two key observations were the sudden increase in stolon growth half-way through PER 2 and th at the leaf photosynthetic rate was not different between PER 1 and PER 2. The effect of air temperature could explain most of the variation observed in canopy – root – soil respiration, but could explain only 25% of daily variation in leaf photosynthesis or 50% of the variation in canopy photosynthesis (data not shown). The rate of V-stage pr ogression was dramatically lower in the fall

PAGE 65

51 (PER 2), much more so than the reduction in leaf mass. Despite having fewer mature leaves per tiller, initial and final LAI were nearly the same for both 8-wk regrowth periods. Consistent with the lower leaf mass and similar LAI, SLA was higher in PER 2. In the case of shaded leaves, the resulting higher SLA is associated with lower concentrations of photosynthetic apparatus and lower potential phot osynthetic rates, however, our predicted A sat was not different between periods. Likely, growth reduction during winter dor mancy is the culmination of a number of factors; reduction in growth rate due to lo wer temperatures, change in partitioning of assimilate favoring storage tissue over leaf growth, and changes in leaf characteristics related to lower light levels. Use of the re sults of this study to develop parameters for modeling bahiagrass growth testing would a llow exploration of “w hat-if” scenarios and possibly help us better understand how these factors interact to reduce forage production during winter dormancy.

PAGE 66

52 Table 3-1. Schedule of sampling and harvest activities. Week Per 1 Date Per 2 Date Activity 0 7/18/01 9/12/01 Mow to 10-cm stubble height 0 7/19/01 9/13/03 Fertilize plots 0 7/20/01 9/14/01 Sample stubble 1 7/25/01 9/19/01 Sample growth 2 8/1/01 9/26/01 Sample growth 3 8/8/01 10/3/01 Sample growth 4 8/15/01 10/10/01 Sample growth 5 8/22/01 10/17/01 Sample growth 6 8/29/01 10/24/01 Sample growth 7 9/5/01 10/31/01 Sample growth 8 9/12/01 11/7/01 Sample growth Final Table 3-2. Weekly averages of daily temp eratures and daily solar radiation and total weekly rainfall + irrigation water applie d to bahiagrass grown at the Irrigation Park, Gainesville, FL 2001 Daily Temperature Rainfall Solar Radiation Week Average MaximumMinimum+ IrrigationDaily Average Period 1 C C C Total (mm)(MJ m-2 day-1) 1 25.9 31.8 22.7 94.8 17.9 2 26.0 32.8 22.2 72.4 16.6 3 25.6 31.3 22.2 38.9 16.4 4 27.6 33.8 23.2 4.3 20.0 5 28.1 34.2 23.4 56.7 21.0 6 27.4 34.1 22.1 30.5 19.1 7 25.9 33.1 22.4 50.3 15.5 8 25.4 31.3 22.7 16.8 13.9 Period 2 1 22.9 27.9 19.1 77.9 13.3 2 24.0 30.1 20.0 63.3 14.7 3 19.8 25.9 14.5 15.0 17.0 4 22.6 28.5 18.5 1.5 12.9 5 21.1 27.8 16.4 27.5 14.0 6 23.6 29.3 20.2 32.0 11.8 7 16.8 24.7 11.3 26.5 14.4 8 19.1 26.1 13.8 0.0 12.1

PAGE 67

53 Table 3-3. Results of statistical comparis on of treatment effects on plant growth and photosynthesis. Period means are le ast squares means. Significance determined by ANOVA for Period and ort hogonal contrast for Week and Per X Week interaction. Period Mean Statistics Growth Variable 1 2 Period Week Per X Week Leaf Mass (kg DM ha-1 ) 21501700*****/qdr ns Leaf Mass (kg DM ha-1 )a 30152511*----Stem Mass (kg DM ha-1 ) 36901960******/lin ns Stem Mass (kg DM ha-1 ) a -456-1396ns----Stolon Mass (kg DM ha-1 ) 47406870****/lin */lin Stolon Mass (kg DM ha-1 ) a 6122350**----Root Mass (kg DM ha-1 ) 62703800****/lin ***/lin Root Mass (kg DM ha-1 ) a -8009-344***----Total Plant Mass (kg DM ha-1 ) 1684514325*ns ***/lin Total Plant Mass (kg DM ha-1 ) a -48393121**----Canopy Height (cm) 29.822.5******/cub ***/lin V-stage (number of fully emerged leaves tiller-1) 3.521.70******/qdr ***/qdr V-stage (number of fully emerged leaves tiller-1) a 7.653.41***----SLA (cm2 leaf g-1 leaf) 64.873.3****/qdr ns LAI (m2 leaf m-2 land) 1.281.10****/qdr ns LAI (m2 leaf m-2 land) 1.561.52Ns----GrAI (m2 leaf + stem m-2 land) 2.491.85******/cub ns GrAI (m2 leaf + stem m-2 land) 1.841.12*----Photosynthesis Variable Period Day Per X Day Measured Leaf Photosynthesis (mol CO2 m-2 leaf s-1) 31.026.6ns *** ns Predicted Max Leaf Photosynthesis (mol CO2 m-2 leaf s-1) 44.239.9nsns ns Predicted Canopy Gross Photosynthesis at 1500 mol photons (mol CO2 m-2 land s-1) 55.943.0*** ns Canopy – Root – Soil Respiration (mol CO2 m-2 land s-1) 24.013.9*** ns *P<0.10, **P<0.05, ***P<0.01, ns=not significant, --does not apply to this variable. lin, qdr, cub = linear, quadratic or cubic orthogonal contrasts, respectively, are significant a values are net change over period = Week 8 values – Week 0 values.

PAGE 68

54 Figure 3-1. Sod core as removed from the soil. Figure 3-2. Example of a separated subs ample of bahiagrass after removing roots.

PAGE 69

55 0 5000 10000 15000 20000 25000 0123456789Weeks of RegrowthTotal Plant Mass (kg DM ha-1) Figure 3-3. Total plant mass for established ba hiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001. 0 2000 4000 6000 8000 10000 12000 14000 0123456789Weeks of RegrowthRoot Mass (kg DM ha-1) Figure 3-4. Root mass for esta blished bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.

PAGE 70

56 0 2000 4000 6000 8000 10000 12000 0123456789Weeks of RegrowthStolon Mass (kg DM ha-1) Figure 3-5. Stolon mass for established bahiag rass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001. 0 1000 2000 3000 4000 5000 6000 0123456789Weeks of RegrowthStem Mass (kg DM ha-1) Figure 3-6. Stem mass for esta blished bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.

PAGE 71

57 0 500 1000 1500 2000 2500 3000 3500 4000 0123456789Weeks of RegrowthLeaf Mass (kg DM ha-1) Figure 3-7. Leaf mass for established bahiagra ss grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001. 0 1 2 3 4 5 6 7 8 9 0123456789Weeks of RegrowthNumber of new leaves with ligules Figure 3-8. V-stage for estab lished bahiagrass grown at Gaines ville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.

PAGE 72

58 0 0.5 1 1.5 2 2.5 0123456789Weeks of RegrowthLAI (m2 leaf m-2 land) Figure 3-9. Leaf area index (LAI) for esta blished bahiagrass grow n at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001. 0 20 40 60 80 100 120 0123456789Weeks of RegrowthSLA (cm2 leaf g-1 leaf) Figure 3-10. Specific leaf area (SLA) for es tablished bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.

PAGE 73

59 0 0.5 1 1.5 2 2.5 3 3.5 0123456789Weeks of RegrowthGrAI (m2 (leaf+stem) m-2 land) Figure 3-11. Leaf + Stem (green) area inde x (GrAI) for establishe d bahiagrass grown at Gainesville, FL from18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001. 0 5 10 15 20 25 30 35 40 0102030405060Days of RegrowthLeaf Net Photosynthetic Rate (mol CO2 m-2 s-1) Figure 3-12. Leaf photosynthetic rate for esta blished bahiagrass grow n at Gainesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.

PAGE 74

60 0 10 20 30 40 50 60 70 0102030405060Days of RegrowthP1500 (mol CO2 m-2 s-1) Figure 3-13. Canopy gross photosynthetic rate adjusted to 1500 mol Par m-2 s-1 (P1500) for established bahiagrass grown at Ga inesville, FL from 18 July to 12 Sept. ( ) and 12 Sept. to 7 Nov. ( ), 2001.

PAGE 75

61 CHAPTER 4 DEVELOPMENT OF CROPGRO SPE CIES FILE PARAMETERS FOR BAHIAGRASS Introduction CROPGRO is a mechanistic model that pr edicts yield and co mposition of crops based on plant, soil, management, and weather inputs. As such, it appears well suited to the task of modeling forage growth and nutri ent concentration. A dditionally, the ability to simulate soil water and N balances, soil organic matter – residue dynamics, and pest/disease damage increase CROPGRO’s u tility as a tool for evaluating potential environmental consequences of management changes. Its generic, process-oriented design has allowed it to be adapted to mode l a variety of different species including soybean ( Glycine max L.), peanut ( Arachis hypogaea L.), dry bean ( Phaseolus vulgaris L.), faba bean ( Vicia faba L.), and tomato ( Lycopersicon esculentum Mill.) (Scholberg et al., 1997; Boote et al., 1998a, 1998b, 2002). Ad aptation is accomplished by changing a set of parameters and relationships descri bing the species’ respons e to environmental variables. The procedure is desc ribed in Boote et al. (2002). Kelly (1995) previously attempted to ad apt CROPGRO to model the growth of bahiagrass with the objectiv e of using the model as a component of a system for simulating peanut cropping systems. Simula tion results were incorporated into an economic model to predict the sustainability and profitability of the cropping systems. The species, cultivar, and ecotype files deve loped were later rele ased as a “pasture” model in DSSAT v 3.5 (ICASA, 1998). Our appl ication of this model to simulate data

PAGE 76

62 sets of bahiagrass hay production revealed consistent overprediction of DM yields, particularly in the cooler months of the year More rigorous app lications and objectives for the use of the model impose different sta ndards of accuracy and our proposed use as a practical planning and teaching tool requires a more accurate prediction capability and a more faithful representation of the seasonal patterns of growth of bahiagrass. The objective of this work was to develop parameters, from searching the literature, experiments, and calibration, to model ba hiagrass growth and composition with the CROPGRO CSM model. Materials and Methods In deriving model parameters to describe bahiagrass growth and composition, we followed the general adaptation procedure de scribed by Boote et al. (2002). Where possible, parameters describing the basic pr ocesses of photosynthesi s, respiration, N assimilation, and plant development in bahiag rass were derived from the literature. Parameters describing basic bi ochemical processes assumed to be conserved, or similar (e.g.. growth respiration cost per unit of protein), among sp ecies are universal throughout all CROPGRO species files. For some less conserved processes and traits where data were lacking, parameters from the CROPGRO soybean species file were used. Soybean was selected as it is one of the original models used to develop CROPGRO and is arguably the most tested and robust version of the model. Examples of parameters incorporated from the soybean files are lipid, lignin, organic acid, and mineral composition, as well as carbon cost to mob ilize N from senesced proteins (Penning de Vries et al., 1974). Where processes or parameters were belie ved to be divergent from soybean or thought to be unique to perennial forage spec ies, parameter estimates were interpolated

PAGE 77

63 from literature data from other tropical pe rennial grass species or selected through sensitivity analysis of the bahiagrass model. We also developed an “optimized” set of parameters using a custom built optimization program utilizing a “brute-force” optimization strategy. In the program, the user specifies a minimum and maximum accep table value for each parameter and the desired number of “steps” between those limits for up to five parameters. Simulations were run using all possible combinations of the specified parameters. Results from each set of parameters were statistically analyzed for mean of simulated results, slope and intercept of a fitted regression line of predicted and observed data, r2, d-index value (index of agreement [Willmott, 1981]), and r oot mean square error (RMSE). Results from all runs were saved to an output file and the combination with the lowest RMSE was listed at the end. The output file was then exported to a spreadsheet, parsed, and sorted from highest (best fit) to lowest d-index rating. The optimized parameter values were selected using a combination of high d-index ranking, low RMSE value, and most logical fit. The optimized parameter set was subsequently tested for fit, using the same procedure as used for testing the literature-based species file. Description of Data Sets Used to Fit Parameters Two data sets were selected for use in fitting parameters and testing the optimized model. Only a brief listing of growing condi tions will be given here. A more complete description of each data set may be found in the cited articles. The study at Ona, FL was part of a threespecies study of forage protein response to N fertilization and cutting date (Johnson et al., 2001). The experiment was conducted at the Range Cattle Research and Education Cent er (REC) at Ona, FL (27 25’N, 81 55’W; elevation 27.4 m) on a Pomona fine sa nd (sandy siliceous, hyperthermic Ultic

PAGE 78

64 Haplaquod) soil. Pensacola bahiagrass received five fertilizer treatments (0, 39, 78, 118, and 157 kg N ha-1 cutting-1), equivalent to annual applications of 0, 234, 468, 708, and 942 kg N ha-1 supplied as ammonium nitrate. Fert ilizer was applied on 5 May and on the day after each cutting except for the October (last) harvests. Staging harvests marking the beginning of each growing season were made on 5 May 1997 and 4 May 1998 with successive harvests every 28 d until October. Forage yield and crude protein concentration were measured for all but the staging harvests. Daily weather data were acquired from the REC’s weather station. Te mperatures rarely dropped below 0C in the winter. Rainfall totale d 1142 mm for October 1996 through September 1997 and 2110 mm from October 1997 through September 1998. The Eagle Lake, TX (29 35’N, 96 20’W, elevation 46 m) experiment was part of a larger study of N cont ributions of arrowleaf ( Trifolium vesiculosum Savi) and subterranean ( Trifolium subterraneum L.) clovers overseeded on bahiagrass and bermudagrass [ Cynodon dactylon (L.) Pers.] conducte d over the 1979-1981 growing seasons (Evers, 1985). The study was locate d in southeastern Texas at the Texas Agricultural Experiment Sta tion at Eagle Lake on a Crowley fine sandy loam (fine montmorillonitic, thermic, Typic Albaqualfs) soil. The Pensacola bahiagrass treatments included fertilizer applied at annual rates of 0, 84, 168, 252, or 336 kg N ha-1. The fertilizer was split into thr ee equal applications made on or about 1 April, 1 June, and 1 August of each year. All plots were harves ted monthly from May through October. Forage yield and crude protei n data were reported. Daily weather data were acquired from the experiment station’s weather st ation. Freezing temperatures were not uncommon in the winter with minimum temperat ures as low as -9C. Average rainfall

PAGE 79

65 was less than for Ona, with annual preci pitation of 1354 mm, 765 mm, and 1223 mm for 1979, 1980, and 1981 seasons, respectively. The two data sets were split for optimization and testing. The two lowest N treatments and the two highest N treatments fr om each site were used in the optimization process. The middle N treatment from both data sets was reserved for testing the literature-based and optimized species files. The rationale behind the splitting scheme was to maximize the range of N fertilization and number of observ ed data pairs (108) available for the optimization process. At the same time the test data sets would be most indicative of how the model will perform under “normal” or the most frequently encountered conditions. The primary object ive for the optimization process was to minimize RMSE for the prediction of herbage mass (leaf + stem weight) or herbage N concentration (leaf + stem N concentration), depending on the variables being optimized. Preparation of Datasets There were no data available regarding initial plant mass or soil conditions for either experiment, so actual initial crop condi tion could not be input into the model. Instead, we estimated the initial conditions by running each simulation for one full growing season/winter cycle prior to the meas ured seasons. Actual weather data were used for the prior year. The season began with an established plant stand cut to the same stubble height used in the measured years. Fertilization during the prior year was consistent with the medium N fertiliza tion treatment for each site (468 kg N ha-1 yr-1 at Ona, FL, and 168 kg N ha-1 yr-1 at Eagle Lake, TX). To compare simulated and observed growth the two sets of results had to be expressed on a common basis. The field stud ies reported yield as herbage (leaf + stem) mass harvested above a base cutting or stubbl e height while simulation results reported

PAGE 80

66 yield as the total amount of leaf and stem. The difference between the two is the amount of leaf and stem mass in the stubble left af ter each harvest. Usi ng the results of other studies (Beaty et al., 1968; Pedreira and Brown, 1996b; Rymph and Boote, 2002), we developed estimates of post-harvest stubble mass for the different cutting heights used in the Ona and Eagle Lake experiments. Thes e estimated stubble masses were added to the reported harvest yields to approximate total herbage mass observed for these experiments. Estimates for stubble mass le ft under 3.5-cm, 5-cm, 7.5-cm, and 10-cm cutting heights were 1500, 1800, 2400, and 3000 kg DM ha-1. These values may apply only to Pensacola bahiagrass. Newer bahiag rass varieties with more upright growth habits may have considerably less stubble ma ss (Pedreira and Brown, 1996b). This tactic basically creates a consistent season-long offset while actual stubble mass may vary with season and N fertilization. This approach represents a compromise between accuracy and simplicity of implementation/utility. Initial testing of the model revealed some characteristic s of the CROPGRO program code that were not compatible w ith a perennial forage. FREEZ1 and FREEZ2 are parameters describing temperatures where a ll leaves fall off of the crop or the entire crop dies (respectively) due to cold. We found that after a FREEZ1 event occurred, there was no regrowth of new leav es, resulting in the plants exhausting all reserves on maintenance respiration and dying. The problem was related to the strategy used to end photosynthesis of grain legumes after a foliage -killing freeze event. Since we could not alter the code, both FREEZ1 and FREEZ2 were set to -25C, essentially disabling the FREEZ1 function but allowing the simulati on to continue through the winter.

PAGE 81

67 Additionally, we simulated frost damage of leaves by partially defoliating the crop each January using the PEST routine. Results and Discussion This discussion will be confined to parameters that were unique to perennial tropical grasses or required redefinition or al teration in concept. A complete list of parameter values is provided in Table 4-1. Photosynthesis Parameters CROPGRO has two options for predicting daily assimilate production: a daily canopy option and an hourly leaf-level opti on. The daily canopy option is the more simplistic approach, predicting photosynthate pr oduction as an asymptotic light response to total daily solar ra diation levels. The leaf-level photosynthesis option predicts hourly photosynthetic rates for sunlit and shaded leaf area by simulating the dynamics of Rubisco activity and electron transport and in tegrates them within the hourly hedgerow approach to yield a daily ca nopy rate. Both options incl ude adjustments for current temperature, CO2 concentration, and leaf N concentration conditions. All previous efforts to adapt CRO PGRO involved crops using the C3 photosynthesis pathway/mechanism. In contrast, bahiagrass expresses the C4 photosynthetic pathway, more sp ecifically, it is an NADP-ME type species (Hattersley and Watson, 1976), the same pathway that is expressed in maize ( Zea mays L.). Concentration of CO2 in the bundle sheath chloro plasts through the “CO2 concentrating shuttle” contributes several advantages to C4 plants. Since CO2 is concentrated around the CO2-fixing Rubisco enzyme in the bundle sh eath chloroplasts, the relatively lower solubility of CO2 at high temperatures is of little consequence, allowing higher rates of carboxylation and suppressing oxyge nation activity of Rubisco at higher temperatures

PAGE 82

68 and light levels than is generally possible in C3 plants. Likewise, the quantum efficiency (QE) is not temperature sensitive, not decreasing at high temperatures as in C3 plants. As photorespiration is reduced, less Rubisco is required to maintain high carbon exchange rates (CERs), and leaf protein le vels are generally lower for C4 plants. These differences must be reflected in our pred icted patterns of photosynthetic response to light intensity, CO2 concentration, leaf N concentrati on, and temperature in the model. The asymptotic light response curve used to predict daily canopy photosynthesis is defined by two parameters; PARMAX the le vel of photosynthetical ly active radiation (PAR) at which photosynthetic rate is 63% of maximum (moles [quanta PAR]m-2d-1), and PHTMAX, the asymptote (maximum) of daily assimilation rate (g CH2O m-2 d-1) occurring at very high light (a t least three times as high as PARMAX). These values are not generally presented in the literature so pr eliminary values (Table 4-1) were estimated as 150% of the corresponding parameter values used for soybean. The lower rate of photorespiration observed in C4 species results in a lower CO2 compensation point (CCMP) (atmospheric CO2 concentration where the rate of CO2 uptake by photosynthesis equals the rate of CO2 evolution by respiration) in C4 plants. Values of 0 to 14 L L-1 have been reported as the CO2 compensation point for various C4 species (Bolton and Brown, 1980; Rajendrudu and Das, 1981), most on the order of 0 to 5 L L-1(Rajendrudu and Das, 1981). Based on thes e results we selected a value of 5 L L-1 for CCMP which is used for th e daily canopy photosynthesis option. The leaf-level photosynthes is option is a more comp licated system requiring several more parameters than the daily ca nopy option, but the model at the leaf and chloroplast level incorporates several conser ved processes for which parameters may be

PAGE 83

69 directly measured. Leaf quantum efficien cy (QE) is typical of these conserved parameters/processes. Quantum efficiency (p arameter name PGEFF) or quantum yield is broadly defined as the initi al slope of the leaf CO2 assimilation:absorbed PAR response. A value of 0.0541 mol CO2 mol-1 absorbed photons (Ehler inger and Bjrkman, 1977) is typically used in CROPGRO for all C3 species, including soybean. While the same biochemical processes are used in both C3 and C4 photosynthesis, the CO2 concentrating effect of the C4 systems increases their QE. Differen ces in efficiency exist between the three variations of the C4 photosynthetic pathway (NADME, NADP-ME, and PCK-type) with NADP-ME species exhibiting the highe st QE with an average QE of 0.065 mol CO2 mol-1 absorbed photons (Ehleri nger and Pearcy, 1983). We selected this value for the bahiagrass species file (Table 4-1). This value appears to be quite robust as it falls well within the range of QE values predicted from bahiag rass canopy photosynthetic light response data (0.054 – 0.081 mol-1 absorbed photons) (Rymph and Boote, 2002), and reported QE values (0.062 to 0.075 mol CO2 mol-1 absorbed photons) for another NADP-ME species, sugarcane ( Saccharum spp.) (Meinzer and Zhu, 1998). One inconsistency that remains is the rela tionship between temperature and QE. In C3 plants, as temperature increases, the solubility of CO2 decreases relative to the solubility of O2, lowering QE of C3 species at high temperat ures. Because of the high CO2 concentration surrounding Rubisc o in bundle sheath cells of C4 plants, the effect of temperature on QE is negligible. However, the temperature effect on QE is hard-coded into CROPGRO and currently, there are no user parameters to modify that response. The other parameter required is light-s aturated leaf assimilation (LFMAX) for leaves at high N concentration, 30 C, and a given specific leaf we ight. We based our

PAGE 84

70 estimate of LFMAX (and PGREF) on a pred icted maximum leaf photosynthetic rate developed from bahiagrass canopy light res ponse data (Rymph and Boote, 2002) of approximately 40.0 mol CO2 m-2 leaf s-1. Relative differences among cultivars are modeled by changing the ratio of LFMAX (max imum leaf photosynt hetic rate for the cultivar) to PGREF (maximum leaf photosynthetic rate for th e species). As Pensacola was the “reference” cultivar on which the species parameters are based and was the cultivar measured, PGREF=LFMAX=1.760 mg CO2 m-2 s-1. The amount of photosynthetic enzymes in th e leaf affect photosynthetic rate as well. Generally, higher N concentrations in th e leaves are correlated with higher levels of these enzymes and higher photosynthetic capacity. Bahiagrass and other C4 grasses are generally considered to have low concentra tions of N in the leaves, yet maintain high photosynthetic rates. Thus, optimal N concen trations for photosynthesis of bahiagrass are likely to be lower than for soybean. We could find no reports of the minimum N concentration required for photosynthesis [F NPGN(1)], so we defined this lower threshold of the N response function [FNP GN(1)] from the lowest reported leaf N concentration 7.6 g N kg-1 leaf (Beaty and Tan, 1972). Sugimoto and Nikki (1979) observed a curvilinear increas e in bahiagrass leaf photos ynthetic rate as leaf N concentration increased from approximately 20 up to 30 g N kg-1. The rate remained constant from 30 g N kg-1 to almost 40 g N kg-1. Hence, we chose a curvilinear (quadratic) shape to define the response of bahiagrass to leaf N concentration, with optimum photosynthetic rates beginning at 30 g N kg-1 [FNPGN(2)] and no decline in rate at higher N concen tration. The 30 g N kg-1 optimum was also used for LNREF, the N concentration at which PGREF is defined for the species.

PAGE 85

71 The high concentration of CO2 around Rubisco in the bundle sheath chloroplasts permits high photosynthetic rate s at higher temperatures than typically observed in C3 plants. Although the mechanism is not well understood, C4 species generally also have a greater sensitivity threshold for low temper ature reduction of photos ynthetic rate than C3 species (Long, 1983; 1999)). Thus bahiagrass should have a base temperature required for photosynthesis that is higher than soyb ean and it should have higher optimum and maximum (highest temperature at which photosynthesis occurs ) temperatures as well. Several studies have been conducted to quant ify the cardinal temperatures for tropical C4 grasses (Ludlow and Wilson, 1971; Wilson, 1975; Unruh et al., 1996); unfortunately none included bahiagrass. Our interpretation of these results are th at the optimum range for leaf photosynthesis for a tropical grass sp ecies should be between 35 and 45C, with a base temperature around 7C and a maximum critical temperature for zero rate near 55C. The daily canopy calculations us e a daily, rather than hour ly, time step. To find daytime temperature thresholds for the daily canopy option, we compared the threshold temperatures for the leaf-level option to daily minimum and/or maximum temperatures for Gainesville, FL and calculated a corres ponding average daytime temperature. The corresponding temperatures were: base temp erature [FNPGT(1)], 12C, optimum range [FNPGT(2), FNPGT(3)] from 25C to 38C, and maximum temperature of 50C [FNPGT(4)]. Low temperatures may also have a prol onged effect on photosynthesis, affecting photosynthetic rate after temperatures have returned to the optimal range. CROPGRO uses another set of temperature parameters FNPGL(1-4) and TYPPGL, to describe the

PAGE 86

72 effect of minimum night temperature on th e subsequent day’s li ght saturated leaf photosynthetic rate. West (1973) observed that Digitaria decumbens grown at 30C and subjected to just one night at 10C and retu rned to 30C, showed a 40% decrease in photosynthetic rate compared to plants held continuously at 30 C. Based on this, we set the minimum temperature [no photosynthesi s on the day after experiencing this temperature FNPGL(1)] to 7C, optimum ni ght temperature [no effect on subsequent days photosynthesis -FNPGL(2)] to 18C, with a quadratic (curvilinear) response between these points (Table 4-1). Root Parameters Bahiagrass poses an additional challenge to modeling its growth using CROPGRO because a significant proportion of total pl ant mass is represented by stolon mass and CROPGRO does not include a stol on organ in its structure. To include stolons in the stem fraction would have confounded the comp utation of protein/N removed at harvest and further complicated the esti mation of stubble mass. Thus we redefined “roots” in the model to include both stolons and roots. Th is “redefinition” wit hout a code change required considerable modification of the gr owth and senescence parameters relative to those used for other species modeled by CROPG RO. The largest adjustment was for the root length density (RFAC1) parameter (cm of root length per g of root). Stolons are much thicker than roots and may represent mo re plant mass than the roots. Additionally, N uptake per length of stolon (i f any) is likely to be much lower than for roots, further decreasing their “effective” length as a root. Based on the relative proportions of stolons and roots reported by Rymph and Boote (2002), RFAC1 was reduced to 5000 cm g-1, 33% lower than the value us ed for soybean roots.

PAGE 87

73 As stolon mass is routinely mobilized to support new growth, the maximum senescence rate (RTSDF) of the combined orga ns was increased from 0.01 to 0.02 or 2% per day. In preliminary simulations this yielded a maximum predicted root mass of approximately 10 000 kg root dry matter (DM) ha-1, in the range of the combined stolon and root mass observed by Boote et al. (1999) (10 660 to 15 370 kg ha-1) and Rymph and Boote (2002) (7155 to 15 740 kg ha-1). Carbon and Nitrogen Mobilization Parameters Another area where modeling perennial fo rages and annual grains differs is N mobilization. The basic concept is the same but the timing and purpose differ. Nitrogen reserves in annual grain crops are generall y mobilized for filling seed. Although many perennial forages such as Pensacola bahiagra ss may set seed, they are generally harvested at a younger stage of maturity and reserves are used primarily to speed vegetative regrowth after a harvest or in the spring. Since perennial fora ges must be able to do this repeatedly over several growing seasons, the rate and extent of N mobilization may be quite different than that observed in annual grain crops. Reports from Skinner et al. (1999), estimating N and total nonstructural carbohydrate (TNC) mobilization in blue grama grass [ Bouteloua gracilis (H.B.K.) Lag ex Steud] dur ing regrowth after cutting, showed quite high rates of N mobilization. On average, 36% of the available N was mobilized within 7 to 10 d of cutting. Th is translates into approximately 5% d-1 or a maximum available N mobilization fraction (NMOBMX) of 0.05. Total nonstructural carbohydrate mobilization was also reported but the measured TNC concentrations were 2.5 to 3 times the levels found in bahiagrass, and we could not adapt them to our purpose with any confidence. With no TNC data to support changes, the soybean values were used for maximum available CH2O mobilization fraction (CMOBMX) (Table 4-1).

PAGE 88

74 Vegetative Partitioning Parameters During vegetative growth, partitioning of new growth among leaf, stem, and roots is a function of the vegetative stage of the crop (V-stage). This is another area where the concept of V-stage is different between annua l crops and perennial forage crops. Annual crops, as well as seedling forages, progre ss through the sequential increase in leaf numbers in a relatively orderl y fashion. Established perennial forages, however, are periodically “re-staged” by harvests and fros ts, interrupting the orderly pattern. As a seedling, bahiagrass could reach a V-stage of 4 (four fully-expanded leaves) with a relatively small root mass and few, if any, st olons. An established stand of bahiagrass, with a relatively large root and stolon syst em capable of mobilizing significant amounts of N, could also have the same V-stage ra ting of 4 after a harves t. In the CROPGRO model, partitioning of subsequent growth is handled identically in both scenarios. A unifying assumption is that, in both cases, if V-stage is low, the priority for partitioning is towards growing leaf mass/area to establ ish photosynthetic capacity. As V-stage increases, more DM may be partitioned to stolon and root. Additionally, since stolon and root mass were combined, partitioning of new growth between organs required modification from the proportions used for soybean. While partitioning in seedlings may be meas ured by changes in leaf, stem, and root mass over time, the presence of older, senesci ng material in established plants prevents such a simple determination. Assuming that the model would be used most often to predict growth of established stands, we developed the partitioning parameters around observed patterns of regrowth, rather than purely on seedling growth. Parameter values were estimated prior to optimization usi ng growth patterns reported by Rymph (2002) and Boote et al. (1999) and then refined by running simulations and manually adjusting

PAGE 89

75 the parameters to match growth patterns and relative magnitudes of each organ (leaf, stem, or root) (Table 4-1). Leaf Growth and Senescence Parameters Complications caused by repeated re-setting of the V-stage of the crop within a growing season mandated some other modifica tions in addressing leaf growth-related parameters. As the V-stage of the plants is reset after each harvest, there is potential for V-stage to be quite low for a mature plant wi th numerous growing points (stolons). Use of the VSSINK function which allows photosynthe sis and leaf expansion to be limited by sink strength rather than assimilate supply, while potentially appropriate for a small seedling, is not likely to fit the conditions of the older plant. To prevent potential limitations to growth in older plants, th e VSSINK function was disabled by assigning a value of 0.0 to the VSSINK parameter. Senescence parameters were modified ve ry little from Kelly’s (1995) pasture model. The time constant for senescence (TCMP) was set to 25 thermal days based on the weekly counts of dead leaves and weat her from the raw data of Rymph and Boote (2002). The light compensation point trigger for leaf senescence (ICMP), which triggers leaf senescence due to shading of lowe r leaves was set to 0.8 moles photons m-2 d-1, the same as soybean. In a similar vein, the Vstage trigger for senescence (when 12% of the plant’s leaf number is assumed to have been senesced) (XSTAGE) was lowered from 14 leaves for soybean to 9. This was necessary because of the relati vely low number of leaves on a bahiagrass plant compared to a soybean plant. Phenology Parameters The actual resetting of V-stage after a harv est is done in the PE ST routine, using either the MOW function (remove herbag e to a designated residual mass) or a

PAGE 90

76 combination of the HARV and HRVS (rem ove a designated proportion of existing herbage mass and number of leaves) functions and no modification of the species file was required. To implement the MOW function, the user supplies the harvest date(s) and the amount(s) of stubble mass to remain after harvest. On the harvest date, CROPGRO then calculates proportion of canopy mass re moved and leaf mass, stem mass, and V-stage are each reduced by that proporti on. The HARV and HRVS functions work similarly except that the user sets the proportion of herbag e mass to be removed (HARV) separately from the proporti on of V-stage lost (HRVS). The influence of temperature on the rate of phenological devel opment of bahiagrass is not well documented in the literature. Th erefore, we set the cardinal temperatures for base (no new leaves or seeds) optimum (maximum rate of leaf or seed addition), and maximum (upper failure temperature) point s (Table 4-1) based on our experiences growing bahiagrass (K.J. Boote, personal communication). Testing of Literature-Based Parameters Testing of the preliminary, literature-ba sed, species file was encouraging with d-index values of 0.843, 0.605, and 0.925 for accumulated herbage mass, herbage N concentration and accumulated herbage N ma ss, respectively, using the leaf-level photosynthesis option and 0.851, 0.531, and 0.907 for the daily canopy photosynthesis option (Table 4-2). Despite the moderate d-index values for predicted herbage N concentration, the r2 values were quite low at 0.18 and 0.19 for the leaf and canopy models, indicating that the N response was a weak point. The higher d-index value for predicted herbage N mass than for either herbage mass or herbage N concentration represents the effect of offs etting errors (underestimation of yield coincident with an overprediction of N concentration).

PAGE 91

77 Reviewing the predicted pattern of grow th, however, showed excessive rates of winter and spring growth (Figure 4-1a). Water and N demand associated with this excessive growth caused elevated water a nd N stress throughout th e spring and early summer (Figure 4-1a), reducing predicted grow th rates in May and June. The principle cause of this discrepancy was our failure to simulate winter dormancy. Compounding this was the lack of a working freeze damage routine promoting an artificially high LAI and photosynthetic capacity through much of the winter. CROPGRO has no provisions for modeling dormancy so we attempted to duplicate the dormancy effect through other methods. To reduce winter growth rate we used the PEST routine in CROPGRO to reduce daily photosynthesis production by 70% fr om 23 October through 30 March. This approach reduces photosynthetic rate but does not concurrently reduce transpiration. Coincident with this change, the periods of water and N stress were shortened considerably (Figure 4-1b). Statistically, how ever, there were minima l changes in the fit of either leaf or canopy mode ls after this modification (T able 4-2). This approach resulted in slight improvements in d-inde x values for leaf and canopy model predicted herbage mass, leaf model predicted herbage N mass, and slightly lower (worse) d-index values for canopy model predicted herbage N concentration and he rbage N mass (Table 4-2). Late spring regrowth was still cons iderably reduced compared to the observed growth (Figure 4-1b), possibly because th e PEST option reducing photosynthesis does not reduce transpiration except indirectly th rough the lower LAI re sulting from slowed growth.

PAGE 92

78 Further investigation revealed that two m echanisms may have been responsible for the early season water and N stress. In reduc ing photosynthesis in the PEST routine, the normal photosynthetic rate and transpirati on rate were calcula ted and then the photosynthetic rate was reduced by the designa ted percentage. Tran spiration, however, was not reduced so water uptake continued at the normal (now exce ssive) rate, depleting available soil water. The only “reduction” in transpiration was due to the lowered LAI that resulted from the slowed growth. Also root N was mobilized throughout the winter and into the spring to compensate for the reduced assimilate pr oduction, reducing root mass considerably by the end of the winter/ear ly spring period (dat a not shown). While more water and N may have been available in the soil, the diminished root system had a reduced capacity to exploit them, suppressi ng early season growth rates. Despite the failure to statistically improve the fit, we us ed this strategy in all optimization and testing runs as the patterns of N stress were more realistic than before. Performance of the model using the daily canopy option and wint er photosynthetic reduction was quite similar to the leaf-level option performance fo r predicting herbage mass (Figure 4-2). The d-index values for th e fit of the predicted data to herbage mass were identical to those for the leaf-level option. The predicte d pattern of growth was also quite similar with slightly reduced winter growth rates but nearly identical summer growth (Figure 4-2). Fit of predicted herbag e N concentration was slightly poorer for the canopy option, but fit of predicted herbage N mass was similar for both options (Table 4-2). Fit of predicted herbage N concentration was not as good as was herbage mass for either option (Table 4-2). Leaf + stem N concentration was consistently overpredicted

PAGE 93

79 for the Ona, FL site (Figure 4-3). During wi nter regrowth, after th e “simulated frost” defoliation, herbage N concentration exceeded 40 g N kg-1, equivalent to 250 g CP (crude protein) kg-1, higher than the “maximum” leaf N concentration set by PROLFI. This is related to the N allocation problem in the code cited earlier that prohibited leaf regrowth after a FREEZ1 event. As the goal of the pres ent exercise was to cal ibrate the parameters without changing any source code, this problem could not be addressed. For the Eagle Lake, TX data, predicted he rbage N concentration appeared to follow a more accurate pattern despite greater variat ion in the observed values (Figure 4-4). Prediction of herbage N concentration wa s more balanced, being both overand underpredicted. The improved prediction pattern ma y be related to the lower fertilizer levels used at Eagle Lake and the lower yields for that site. Values from the daily canopy photosynthesis option were genera lly higher than for the leaf -level option (Figures 4-3 and 4-4). Optimization Since we were unable to accurately predic t the spring growth pa ttern, some of the early season yield data were excluded from the optimization process. The rationale for “culling” these two data points was that th e model was consistently predicting early season N and water stress when there was none ; thus keeping those data points in the optimization would influence the final paramete r values in order to compensate for the predicted stresses. This left 52 data pairs for calibration. No data were excluded from the datasets used to test the performance of the model. All testing runs used all of the observed data available for the site/fertility combinations used (27 data pairs). The distinction being made here is that we want ed to develop the most accurate parameters for the model through optimization (hence le aving out the early season data) while

PAGE 94

80 presenting a fair evaluation of the performan ce of the model through testing (by including all data). Our strategy was to first optimize the leaf-level option temperature parameters to establish proper general patte rns of growth, then refine the prediction by optimizing parameters that affect the gr owth response to N. After op timizing the leaf-level option, the temperature parameters for the daily canopy photosynthesis option were optimized followed by simultaneous optimization of PARMAX and PHTMAX. All temperature parameters (leaf-l evel and daily canopy options) and PARMAX/PHTMAX optimizations used only the two highest N fertility treatments from each experiment, assuming that N would not be limiting growth for those treatments. The two highest and two lowest N fertilizati on treatments from each study were used in optimizing the N parameters as this pres ented the broadest range of conditions. Testing of Optimized Parameters Optimization improved the predicted winter growth pattern (Figur es 4-5a and 4-5b) but fit of predicted herbage mass during th e growing season was generally unaffected (Table 4-2) with similar dindex values for both optimized and literature-based species files. Winter growth pattern was improve d most for the daily canopy option where there was almost no regrowth thr ough the winter (Figure 4-5b). A wider range of potential parameter values were offered in the daily canopy option optimization as there was less data available to define the ranges. Wi nter regrowth was cu rtailed by increasing FNPGT(1), only allowing growth on days with average temperatures greater than 20C (Table 4-1). Normally, this would affect growth rates well into the spring and fall growing season, but PARMAX and PHTMAX were also boosted in the optimization (Table 4-1). This combination allowed grow th rates to be nearly identical during the

PAGE 95

81 normal growing season (Figures 4-5a and 4-5b). Nitrogen stress was reduced in the optimized simulations (Figures 4-5a and 4-5b); however, wa ter stress was still extensive in the spring, even for the da ily canopy option, resulting in a continued poor prediction of first cutting regrowth at Ona, particularly in the second grow ing season. To compensate for the failure of the model to properly simulate winter dormancy, the optimization process promoted combinations of extreme pa rameter values to improve the fit of the model. Overall, both the leaf-level and daily canopy options tended to overpredict herbage mass at lower yields and underpredict at highe r yields with this phenomena most evident in the daily canopy option results (Figures 4-6a and 4-6b). Despite the improved fit, the optimized values for these three parame ters, FNPGT(1), PARMAX, and PHTMAX, are not realistic and reflect the emphasis on comp ensating for the excessive winter growth pattern, not improving the relevanc y of the parameter value. Fit of herbage N concentration predicti ons improved considerably for both the optimized leaf-level and optimized daily canopy photosynthesis options (Table 4-2, Figures 4-7 and 4-8). The dindex rating for both options improved consid erably after optimization; however, the r2 values remained quite low (Table 4-2). Predicted herbage N concentrations were still consistently ove rpredicted at Ona (Figure 4-7) despite optimization toward lower PROLFI, PROL FG, PROSTI, and PROSTG parameters (Table 4-1). The pattern of predicted herb age N concentration for Eagle Lake remained realistic after optimization (Figure 4-8), how ever, the optimized parameter values were generally lower than the literature-based parameters (Figure 4-4). The difference between predicted and observed values was al so reduced for both the leaf-level (Figure

PAGE 96

82 4-9a) and daily canopy options (Figure 4-9b) i ndicating a more consistent prediction. Both leaf-level and daily options tended to ove rpredict herbage N concentration, but this was more pronounced in the daily canopy option predictions (Figures 4-9a and 4-9b). Despite the improvement in herbage N concentr ation prediction, fit of predicted herbage N mass showed little improvement (Table 4-2) but had been quite high to begin with. Conclusions Performance of the literature-based paramete rs was quite good, especially related to predicting herbage mass and herbage N mass. The prediction of herbage N concentration needed improvement. On review of the result s, there appeared to be some features of CROPGRO that may have made significant contri butions to the errors in predicting both herbage N concentration and herbage mass. The absence of a dormancy routine to control vegetative growth dur ing the winter and spring m onths had a profound effect on early season N and water availability, contributing to low predicted herbage mass throughout the season. The absence of a st orage organ such as a rhizome or stolon contributed to this problem by confounding effects of cha nging proportions of stolon and root mass. Quirks related to modeling of freeze damage and patterns of “refilling” of N in old tissues complicated matters even more. Imposing a 70% reduction in potential daily photosynthesis during the winter months compensated for some of the problems, albeit in an artificial way. Optimization did improve the fit of both the leaf and canopy models compared to simulations using the literature-based pa rameters. Winter growth was slow, and excessive levels and variation in leaf N concentration were controlled using the optimized parameters. However, some of the optimi zed parameters are at the edges of their biological range or beyond as a result of compensa ting for the missing/problem

PAGE 97

83 components in the model code. The optimization was more an exercise in compensating for the model than in divining more accurate parameter values. In order to better mimic the biology of pe rennial, tropical grasses, modifications must be made to the model code itself. Pr imary among these changes is the addition of a dormancy routine. Evidence for this and th e mechanism required is available in the literature. Sinclair et al. (2003), Mislevy et al. (2000), a nd Gates et al. (2001) clearly demonstrate the role of daylength in cont rolling dormancy. Hints for the mechanism involved in reducing leaf and stem growth during dormancy can be found in Rymph and Boote (2002) and Boote et al. (1999) where significant shifts in allocation of new growth from shoots to stolons were observed in th e fall. Adding a mechanism controlled by daylength to reduce partitioning and mobiliza tion to the shoot while increasing the same to the stolon would complement the maturity, temperature, and stress mechanisms already present in CROPGRO. Addition of the storage organ would also allow more realistic prediction of the patterns of accumulation and depletion of roots, avoiding confounding root mass and N uptake parameters to compensate for the presen ce of stolons in the root mass. Providing a storage organ not only provide s a sink to store the excess assimilate that is currently allocated to leaves and stems in the wi nter, it would also supply a source of CH2O and N for regrowth after frosts, in th e spring, and after harvests. This would prevent the current situation of the plants dying after a frost and allow for more rapid regrowth in the spring. Other elements of the model, such as the freeze damage scheme and the partitioning of N to replenish old leaves, likely stem from past approaches to modeling an annual grain crop compared to a perennial forage Situations such as low leaf mass after

PAGE 98

84 a harvest coupled with large amounts of available N from the roots are not generally encountered in the life cycle of maize or soyb ean but are dominant features of the pattern of growth of a perennial gr ass. These differences are better addressed through adapting the model code than by adjusting species parameters. Consideration of these diffe rences notwithstanding, the overall performance of both the literature-based and optimized parameters was good. If used carefully, the optimized leaf and canopy models should perform well. More testing would be in order if these models were to be used extensively. As mentioned earlier, further optimization will only improve our ability to compensate for the model code, not improve the quality of the parameters. Taking steps such as running the simulation for a year prior to the measured growing seasons to establish initial conditions addition of defoliation events to simulate frosts, and addition of photosynthesis reducti on schemes to reduce winter growth will be as critical as changing parameter values in establishing a good fit of model predictions to observed data. The bulk of future efforts shou ld be directed at changing the model code to more accurately reflect the lif e cycle of perennial grasses.

PAGE 99

85 Table 4-1. Bahiagrass parameter values fo r the CROPGRO species file. Preliminary values were derived from the literature. Optimized values were derived from optimization runs made based on the preliminary values. Parameter Name Preliminary value Optimized value PARMAX 60.0 140.0 PHTMAX 90.0 180.0 CCMP 5.0 FNPGN(1-4) 0.75, 3.0, 10.0, 10.0 1.0, 3.0, 10.0, 10.0 TYPPGN QDR FNPGT(1-4) 12.0, 25.0, 38.0, 50.0 20.0, 25.0, 30.0, 50.0 TYPPGT LIN XLMAXT -5.0, 7.0, 35.0, 45.0, 55.0, 60.0 -5.0, 10.0, 26.0, 45.0, 57.0, 60.0 YLMAXT 0.0, 0.0, 1.0, 1.0, 0.0, 0.0 0.0, 0.0, 1.0, 1.0, 0.0, 0.0 FNPGL(1-4) 7.0, 18.0, 45.0, 57.0 7.0, 18.0, 45.0, 57.0 TYPPGL QDR PGEFF 0.065 SLWREF 0.0035 LNREF 3.0 PGREF 1.76 PROLF I, G, and F 0.22, 0.11, 0.05 0.15, 0.05, 0.04 PROST I, G, and F 0.11, 0.07, 0.033 0.125, 0.04, 0.022 PRORT I, G, and F 0.101, 0.040, 0.022 PLIP LF, ST, RT 0.025, 0.020, 0.020 PLIG LF, ST, RT 0.04, 0.06, 0.07 PCAR LF, ST, RT 0.602, 0.697, 0.702 0.672, 0.682, 0.702 CMOBMX 0.025 NMOBMX 0.05 NVSMOB 1.00 XLEAF 0.0, 1.5, 2.0, 3.0, 5.0, 7.0, 30.0 YLEAF 0.45, 0.5, 0.6, 0.4, 0.25, 0.2, 0.2 YSTEM 0.05, 0.05, 0.1, 0.1, 0.05, 0.05, 0.05 FRSTMF 0.05 FRLFF 0.20 FRLFMX 0.60 FINREF 144 SLAREF 285 SIZREF 2.0 VSSINK 0.0 SLAMAX 350 SLAMIN 200 XVGROW 0.0, 5.0, 10.0, 15.0, 20.0, 25.0 YVREF 0.0, 10.0, 20.0, 30.0, 40.0, 50.0 XSLATM -50.0, 00.0, 10.0, 30.0, 60.0

PAGE 100

86 Table 4-1. Continued Parameter Name Preliminary value Optimized value YSLATM 0.25, 0.25, 0.25, 1.00, 1.00 FREEZ1, FREEZ2 -25.0, -25.0 ICMP 0.8 TCMP 25 XSTAGE 0.0, 5.0, 9.0, 50.0 XSENMX 3.0, 5.0, 10.0, 50.0 RTDEPI 20 RFAC1 5000 RTSDF 0.02 RWUEP1 1.5 Vegetative TB, T1, T2, TMax 9.0, 32.0, 40.0, 45.0 Early Reproductive TB, T1, T2, TMax 10.0, 28.0, 32.0, 45.0 Late Reproductive TB, T1, T2, TMax 10.0, 28.0, 32.0, 45.0 XVSHT (1-10) 0.0, 1.0, 4.0, 6.0, 8.0, 10.0, 14.0, 16.0, 20.0, 40.0 YVSHT (1-10) 0.0150, 0.0265, 0.0315, 0.0330, 0.0345, 0.0330, 0.0310, 0.0255, 0.0170, 0.0030 YVSWH (1-10) 0.0150, 0.0255, 0.0310, 0.0320, 0.0330, 0.0315, 0.0295, 0.0230, 0.0125, 0.0005

PAGE 101

87 Table 4-2. Evaluation of the performan ce of CROPGRO with literature-based and optimized species files, with and wi thout winter photosynt hesis reduction. Observed values Mean Herbage Mass 3215 kg DM ha-1 Herbage N Conc. 18.6 g N kg-1 Herbage N Mass 60.1 kg N ha-1 Literature-based species file alone Leaf-level photosynthesis option predictions Mean Slope Intercept d-index r2 RMSE Herbage Mass 3119 0.611 1154 0.843 0.54 557 Herbage N Conc. 20.1 0.603 8.9 0.605 0.18 5.8 Herbage N Mass 65.0 1.005 4.57 0.925 0.77 14.5 Daily canopy photosynthesis option predictions Herbage Mass 2944 0.605 999 0.851 0.68 536 Herbage N Conc. 22.8 0.603 11.54 0.531 0.19 6.9 Herbage N Mass 69.21 0.990 9.7 0.907 0.78 16.13 Literature-based species file with winter photosynthesis reduction Leaf-level photosynthesis option predictions Mean Slope Intercept d-index r2 RMSE Herbage Mass 3119 0.624 1113 0.856 0.58 533 Herbage N Conc. 21.1 0.648 9.0 0.582 0.18 6.3 Herbage N Mass 68.2 1.085 3.0 0.926 0.83 14.9 Daily canopy photosynthesis option predictions Herbage Mass 2944 0.626 931 0.856 0.68 534 Herbage N Conc. 24.1 0.657 11.9 0.482 0.19 8.0 Herbage N Mass 73.4 1.096 7.6 0.890 0.82 18.6 Optimized species file with winter photosynthesis reduction Leaf-level photosynthesis option predictions Mean Slope Intercept d-index r2 RMSE Herbage Mass 3214 0.649 1129 0.860 0.58 528 Herbage N Conc. 18.2 0.426 10.32 0.676 0.19 4.3 Herbage N Mass 60.0 0.774 13.5 0.935 0.78 11.7 Daily canopy photosynthesis option predictions Herbage Mass 2923 0.713 631 0.847 0.62 587 Herbage N Conc. 21.6 0.438 13.4 0.600 0.22 5.1 Herbage N Mass 64.0 0.803 15.8 0.935 0.80 12.0

PAGE 102

88 0 1000 2000 3000 4000 5000 60002 7 -J a n 2 8 Mar 2 7 -May 26 Jul 24-Se p 23-Nov 2 2 -J a n 2 3 M a r 2 2 -May 21-Jul 19-Se pDateHerbage Mass (kg DM ha-1)0 0.5 1 1.5 2 2.5 3Stress Factor (0 1)Forced defoliation to simulate frost damage a 0 1000 2000 3000 4000 5000 60002 7Ja n 28 Mar 27-May 26-Jul 24 S e p 23Nov 2 2Ja n 23 Mar 22-May 21-Jul 19 S e pDateHerbage Mass (kg DM ha-1)0 0.5 1 1.5 2 2.5 3Stress Factor (0 1)b Figure 4-1. Observed herbage mass ( ), predicted herbage mass ( ), water stress ( ), and N stress ( ) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, using the literature-bas ed species file and the leaf-level photosynthesis option, with a) No adjustment to winter growth, or b) 70% reduction in photosynthetic rate and partia l defoliation (frost) over the winter. Predicted stress factors are based on a 0–1 scale with 0= no stress/normal growth rate and 1= severe stress/no growth. The broken horizontal line ( ) denotes the stubble mass left in the field after each harvest.

PAGE 103

89 0 500 1000 1500 2000 2500 3000 3500 4000 4500 500027-Ja n 2 8-Mar 2 7-May 2 6 -Jul 2 4 S e p 23-Nov 22-Jan 2 3-Mar 2 2-May 2 1 -Jul 1 9 S e pDateHerbage Mass (kg DM ha-1) Figure 4-2. Observed bahiagrass herbage mass ( ) and predicted bahiagrass herbage mass using the preliminary (literature-based non-optimized) species file and the leaf-level photos ynthesis option ( ), or daily canopy photosynthesis option ( ). For bahiagrass grown at Ona, FL with 468 kg N ha-1 yr-1. The broken horizontal line ( ) denotes the stubble mass left in the field after each harvest.

PAGE 104

90 0 10 20 30 40 5027-Ja n 2 8Ma r 2 7-May 2 6J u l 2 4-S e p 2 3-N o v 22 J a n 23-Mar 2 2 Ma y 21J ul 19S e pDateHerbage N Conc. (g N kg-1) Figure 4-3. Observed bahiagra ss herbage N concentration ( ) and predicted bahiagrass herbage N concentration using the preliminary (literature-based, non-optimized) species file a nd the leaf-level option ( ), or daily canopy option ( ). For bahiagrass grown at Ona, FL with 468 kg N ha-1 yr-1.

PAGE 105

91 0 10 20 30 40 502 5-Jan 26 M ar 25 May 2 4-Jul 2 2-Sep 21-No v 20-Jan 20-Mar 19-May 18-Jul 1 6-Se p 15-N ov 14-Jan 15-M ar 14 M ay 13-Jul 1 1 S epDateHerbage N Conc. (g N kg-1) Figure 4-4. Observed bahiagra ss herbage N concentration ( ) and predicted bahiagrass herbage N concentration using the preliminary (literature-based, non-optimized) species file a nd the leaf-level option ( ), or daily canopy option ( ). For bahiagrass grown at Eagle Lake, TX with 168 kg N ha-1 year-1.

PAGE 106

92 0 1000 2000 3000 4000 500027 -Ja n 2 8Ma r 2 7-M a y 26-Jul 2 4-Sep 23-Nov 22 -J an 23M ar 2 2Ma y 21 Jul 1 9 -SepDateHerbage Mass (kg DM ha-1)0 0.5 1 1.5 2 2.5Stress Factor (0 1) a 0 1000 2000 3000 4000 500027 Ja n 28-Mar 2 7-May 26 Ju l 2 4Sep 2 3No v 22-Ja n 2 3M a r 22-May 21-Ju l 19-SepDateHerbage Mass (kg DM ha-1)0 0.5 1 1.5 2 2.5Stress Factor (0 1) b Figure 4-5. Observed herbage mass ( ), predicted herbage mass ( ), water stress ( ), and N stress ( ) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, using the optimized species file with a winter defoliation, 70% reduction in winter pho tosynthetic rate, and a) the leaf -level photosynthesis option or b) daily canopy photosynthesis option. Predicted stress factors are on a 0–1 scale with 0= no stress/normal growth rate and 1= severe stress/no growth. The broken horizontal line ( ) denotes the stubble mass left in the field after each harvest.

PAGE 107

93 y = 0.6486x + 1128.8 R2 = 0.5791 y = 0.6236x + 1113.4 R2 = 0.5802 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0500100015002000250030003500400045005000Observed Herbage Mass (kg DM ha-1)Predicted Herbage Mass (kg DM ha-1)a y = 0.7128x + 631.07 R2 = 0.6165 y = 0.6261x + 930.69 R2 = 0.6783 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0500100015002000250030003500400045005000Observed Herbage Mass (kg DM ha-1)Predicted Herbage Mass (kg DM ha-1)b Figure 4-6. Predicted vs. observed herbag e mass of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, and 168 kg N ha-1 yr-1 at Eagle Lake, TX, using a) the leaflevel photosynthesis option, or b) daily canopy photosynthesis option. Both options included a 70% reduction of photosynt hetic rate over th e winter. Plots for both the literature-based ( ) and optimized ( ) species files are presented. Linear regression lines for preliminary ( ) and optimized ( ) species files are presented with thei r corresponding e quations and r2 values along with a line designating a theoretic al 1:1 relationship is ( ).

PAGE 108

94 0 5 10 15 20 25 30 35 4027 Ja n 28-M ar 2 7 M ay 26-Jul 2 4-Se p 23-N ov 22-Jan 2 3-Mar 22 May 2 1-Jul 1 9-SepDateHerbage N Conc. (g N kg-1) Figure 4-7. Observed herbage N concentration ( ) and predicted herbage N concentration of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, using the optimized species file and the leaf-level option ( ) or daily canopy photosynthesis option ( ).

PAGE 109

95 0 5 10 15 20 25 30 35 402 5Ja n 26-M ar 25 May 24 Ju l 22-Sep 21-Nov 20 Ja n 20-M ar 19 May 18 Ju l 16-Sep 15-Nov 14 Ja n 15-M ar 14 M a y 13 Ju l 1 1-SepDateHerbage N Conc. (g N kg-1) Figure 4-8. Observed herbage N concentration ( ) and predicted herbage N concentration of bahiagrass grown with 168 kg N ha-1 yr-1 at Eagle Lake, TX, using the optimized species file and the leaf-level option ( ) or daily canopy photosynthesis option ( ).

PAGE 110

96 y = 0.4258x + 10.316 R2 = 0.1932 y = 0.6481x + 8.9819 R2 = 0.1841 0 5 10 15 20 25 30 35 40 0510152025303540Observed Herbage N Conc. (g N kg-1)Predicted Herbage N Conc. (g N kg-1)a y = 0.4376x + 13.4 R2 = 0.2168 y = 0.657x + 11.846 R2 = 0.1854 0 5 10 15 20 25 30 35 40 0510152025303540Observed Herbage N Conc. (g N kg-1)Predicted Herbage N Conc. (g N kg-1)b Figure 4-9. Predicted vs. observe d herbage N concentration (g kg-1) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL, and grown with 168 kg N ha-1 yr-1 at Eagle Lake, TX, using a) the leaf-level photosynthesis option, or b) the daily canopy photosynthesis option. Both options us ed a 70% reduction in photosynthetic rate over the winter months. Plots fo r both the preliminary (literature-based, non-optimized) ( ) and optimized ( ) species files are presented. Linear regression lines for preliminary ( ) and optimized ( ) species files are presented with their corresponding equations and r2 values along with a line designating a theoretical 1:1 relationship is ( ).

PAGE 111

97 CHAPTER 5 ADAPTING CROPGRO TO MODEL PERENNIAL TROPICAL GRASSES: STRUCTURAL CHANGES TO THE MODEL Introduction A model capable of predicting forage yi eld and composition would be useful in developing harvest management strategies to improve nutrient management on farms and to assist in matching harvested forage co mposition with animal requirements. The current approach of basing harvest schedul es for tropical grasses like bahiagrass ( Paspalum notatum Flgge) and bermudagrass [ Cynodon dactylon (L.) Pers.] on fixed regrowth periods is inflexible and too generalized to be useful in planning or testing new strategies tailored to more specific needs. A crop model that will predict responses of both yield and forage composition to changing environmental conditions and management inputs could be useful for comparing several management systems to determine the best fit for a particular situ ation. Likewise, such a tool could aid researchers and extension pe rsonnel selecting best manage ment practices (BMPs) and developing management recommendations by allowing them to test the practices virtually, and using the knowledge of that expe rience to design better field experiments. In so doing, the development process may be accelerated and costs reduced. If the model proves accurate and consistent enough, it coul d be used to support management decisions and nutrient management plans, allowing managers and planners more freedom in responding to conditions and designing plans th an being limited to an approved list of BMPs.

PAGE 112

98 The CROPGRO model fits these general qua lifications. It is a process-oriented model that uses daily weather and management inputs to predict daily changes in plant composition and growth. Plant composition co mponents that are pr edicted by the model include CP, carbohydrate (cell wall and starch), lipid, lignin, organic acid, and ash. Growth and yield components predicted ar e leaf, stem, seed, and root mass. Additionally, the model includes soil organic matter, N, and water balance simulation, thus providing information on N cycling, upt ake, and losses. CROPGRO has already been adapted to model several di fferent species including soybean ( Glycine max L.), peanut ( Arachis hypogaea L.), dry bean ( Phaseolus vulgaris L.), faba bean ( Vicia faba L.), and tomato ( Lycopersicon esculentum Mill.) (Scholberg et al., 1997; Boote et al., 1998a, 1998b, 2002). Generally, CROPGRO can be adapted to model a species by developing new input files containing species, cultivar, and ecotype parameters appropriate to that species. In a previous chapter we used this approach to adapt CROPGRO to model the C4 perennial grass, bahiagrass, with some degree of success. However, in the adaptation process severa l areas were identified where the CROPGRO model structure either was lacking a critical process or did not adequately describe a process related to perennial tropical or s ubtropical grass growth (Rymph et al., 2003). Addressing these processes required modifica tion of the existing CROPGRO code. Our objective in this research was to improve CROPGRO prediction of bahiagrass growth and composition by adding or modifying the code to add a storage organ (e.g., stolon or rhizome) to the model structure, add a dorma ncy process to alter partitioning of growth and mobilization of N during the winter, alte r the freeze damage routine to allow partial loss of leaves and stems to freezing temper atures, and modify the leaf photosynthesis

PAGE 113

99 mechanism to accommodate the effects of the CO2-concentrating mechanisms in C4 photosynthesis. These modifications will be specific to perennial forage crops and warrant a new designation for the model. Th e modified model will be referred to as the forage version of CROPGRO and treated as a separate entity rather than an incremental improvement of the original model. Materials and Methods The CSM version of CROPGRO, distribute d as part of DSSATv4 (Hoogenboom et al., 2003) served as the base model for thes e modifications. Besides being the most recent version of CROPGRO, this versi on includes a soil organic matter (SOM) decomposition and mineralization option ba sed on the CENTURY SOM model (Gijsman et al., 2002). This option was essential for simulating bahiagrass growth under low input or low N fertilization systems. Previous versions of CROPGRO used an SOM option that did not include decomposition of residue on the soil surface, an important element of the N-cycling dynamics in perennial forage syst ems. The CSM version is also structured to compartmentalize different plant processe s into individual m odules. New processes may be added by writing new modul es and linking them to the rest of the model. The new dormancy process was added in this way. The freeze damage module was modified for progressive freeze damage as well. However, adding the STOR organ was a more comprehensive change, requiring modification of several of the modules. Species-specific parameters for bahiagra ss developed for the CSM version (Rymph et al., 2003) were used in the forage version as well. Where possible, values for any new parameters were derived from measurements re ported in the literature. Where data for a process were not available, parameters were estimated by calibration of model performance to data compiled from several e xperiments. Primary calibration was done at

PAGE 114

100 the process level. New model functions were written and plotted in MathCad (MathSoft, 1998) and the parameters adjusted to match th e response reported in the literature or observed in the test data. The Mathcad equations were rewritten in Fortran and incorporated into CROPGRO. The model wa s then used to simulate each of the experiments that we had assembled. Paramete rs that could not be modeled in Mathcad, such as the fraction of new growth partiti oned to the new storage organ (STOR) and the senescence rate of STOR, were adjusted to fit the CROPGRO predic tions to the observed data or to at least provide realistic pattern s of growth. Model performance was evaluated by comparing the predictions from the simulatio ns to the results of actual experiments. Several measures of model fit were calculate d to aid in the evaluation. These included the mean of all observations, m ean of all corresponding predic ted values, slope, intercept, and r2 of the linear regression of predicted values against observed data, index of agreement (d-index) (Willmott, 1981; 1982), an d root mean square error (RMSE). Data from five experiments conducted at tw o sites in Florida a nd one site in Texas were compiled for use in calibrating the new parameters. Combined, these experiments provided 303 measurements of forage mass a nd 227 measurements of forage CP. Only a brief description of each study will be presente d here as more complete descriptions of each study may be found in their original pub lications (Evers, 1985; Boote et al., 1999; Gates et al., 2001; Johnson et al., 20 01; Rymph and Boote, 2002). Eagle Lake, TX 1979-1981 (Evers, 1985) (29 35’N, Lon 96 20’W) This experiment reported forage yield from Pens acola bahiagrass pastures fertilized with annual rates of 0, 84, 168, 252, or 336 kg N ha-1. All treatments were harvested monthly

PAGE 115

101 from May to October. Forage DM yield and crude protein (CP) concentration were measured at each harvest. Gainesville, FL 1995-1998 (Boote et al., 1999) (29 38 N, 82 22 W) Pensacola bahiagrass was grown in temperature/CO2 gradient greenhouses a nd in outside plots. Temperature and CO2 concentration treatments were: ba seline temperature (near ambient temperature) and baseline temperature + 1. 5, 3.0, or 4.5C under either ambient (360 L CO2 L-1) or double ambient (700 L CO2 L-1) CO2 concentrations. Forage DM yield, forage CP concentration, leaf CP concentra tion, stem+stolon CP concentration, leaf area index, root mass, and stem+stolon ma ss were reported for various harvests. Gainesville, FL 2001 (Rymph and Boote, 2002) (29 38 N, 82 22 W) This was a bahiagrass growth study. Irrigated Pens acola bahiagrass was fertilized at an annual rate of 312 kg N ha-1. There were two 8-wk growth pe riods, 18 July – 12 September and 12 September – 7 November with weekly diggin g of sod cores. Sod cores were separated into leaf, stem, root, and stolon. Leaf, stem stolon, and root mass were reported as well as herbage mass (calculated as the sum of leaf and stem mass). Ona, FL 1993-1995 (Gates et al., 2001) (27 25’N, 81 55’W) Pensacola bahiagrass was harvested between Septembe r and April for two consecutive growing seasons. Treatments were staggered in time with initial harvests starting on 30 September, 15 October, 30 October, 14 November, 14 December, 13 January, 12 February, 27 February, 14 March, 29 March, and 13 April. The 14 December through 27 February treatments had been staged 30 d prio r to harvest while all other harvests were staged 15 d prior to the harvest. The 30 September through 14 November treatments were all harvested a second time on 14 Decem ber while all later treatments were

PAGE 116

102 harvested on 13 April. Thus 10 treatments were harvested twice and the regrowth interval between the first and second harvest varied from 0 to 120 d. The 13 April treatment was harvested only once. Forage DM yield was measured for all winter harvests and forage CP concentration wa s measured for the ha rvests in the 1994-1995 winter season only. Ona, FL 1996-1998 (Johnson et al., 2001). (27 25’N, 81 55’W) Non-irrigated bahiagrass was grown in the field with five N fertilization treatments (0, 39, 78, 118, and 157 kg N ha-1 cutting-1), equivalent to annual applica tions of 0, 234, 468, 708, and 942 kg N ha-1. Forage DM yields and CP concentrations were measured for five harvests each season for two seasons. All treatments were staged by mowing in early May each year and harvested every 28 d until October. To compare the predicted herbage mass (l eaf and stem mass from the surface of the ground to the top of the canopy) values gene rated by CROPGRO to the forage mass (leaf and stem mass from the cutting height to the top of the canop y) reported in most of these studies, we converted the reported forage ma ss to herbage mass. Estimates for stubble mass left under 3.5-, 5-, 7.5-, and 10-cm cutting heights were 1500, 1800, 2400, and 3000 kg DM ha-1 and were determined using data from several studies (Beaty et al., 1968; Pedreira and Brown, 1996b; Rymph and Boote, 2002). The estimated stubble mass was added to the reported forage mass to estimat e total above-ground he rbage mass and this was used as the observed data for comparis on with the model predictions. The following terms will be used in the discussion of the model calibration: herbage mass to indicate the total leaf and stem mass from the soil surface to the top of the canopy, herbage N

PAGE 117

103 concentration indicating the N concentration in the herbage mass, and herbage N mass being the product of herbage mass and herbage N concentration. Results and Discussion Storage Organ In previous attempts to model bahiag rass in CROPGRO, we had grouped stolons with the roots (Rymph et al., 2003) to fit the model structure. Combining the two organs confounded their different patterns of growth and mobilization and required considerable modification of the root length density (cm of roots per g of roots) parameter to control water and N uptake capacity. The prodigi ous amount of stolon mass produced by bahiagrass made this no small problem. Thus new code was added to include a storage organ representing stolons, rhizomes, or crow ns (STOR) to the model structure. As STOR represents stolons in bahiagrass, we will refer to actual plant storage tissue as stolons in this chapter and the model variable name as STOR. For brevity, this discussion will concentrate on the differen ces between the existing CROPGRO model and our forage version rather than the ove rall function of CROPGRO A more thorough description of the structure of the CROPGRO model can be found in works by Boote et al. (1998a, 1998b) and Boote and Pickering ( 1994) as well as in the documentation for DSSAT version 4 (Hoogenboom et al ., 2003) which includes CROPGRO. Although stolons may give rise to new tille rs, their role as a sink and source of N for regrowth of the plant sugge sts they be classified as ve getative organs rather than reproductive. The reproductive role of the storage organ is recognized in the revised freeze-damage routine where the simulation is no t terminated as long as there is storage organ tissue remaining, recognizing the role of the organ in initiati ng regrowth. For the most part, the new code for STOR is a duplic ation of the existing leaf and stem code,

PAGE 118

104 although partitioning of new grow th to STOR is regulated somewhat differently and there are some added provisions for the new dormancy process. The normal rate of partitioning to stolons is defined by the parameter YSTOR, an array containing a series of rates (fractions) corresponding to a progression of V-stage values held in another array (XLEAF). Through use of a lookup function, partitioning will vary with stage of maturity. Once vegetative development ceases, the partitioning fraction to STOR is set to a single value (FRS TRF). All three parameters are read from the species parameter file (species file). Th is file is one of three external input files containing parameter values used in the model to define plant responses to the environment and management (the other file s are the cultivar a nd ecotype files with cultivar and ecotype-specific parameters). Partitioning to, and mobilization from, STOR are not directly affected by N or water stresses; however, there is an indirect effect as those stresses increase partitioning of growth to root at the expense of leaf, stem, and STOR growth. Dormancy may increase parti tioning of new growth to STOR during the short-daylength months. The decision point s and information flow for partitioning are depicted in the flowchart in Figure 5-1 which will be discussed in more detail in the dormancy section that follows. Mobilization of ava ilable carbohydrate (CH2O) and N from leaves and stems is unchanged in the forage version of CROPGRO. M obilization of CH2O and N from STOR relies on a similar approach but is modified to increase mobilization and accelerate regrowth when LAI is low and/or vegetative (leaf, stem, root, a nd STOR) N-status is high. Rather than sharing a common mobiliz ation parameter with leaves, stems, and roots, the “normal” fraction of available CH2O and N that can potentially be mobilized

PAGE 119

105 from STOR is set by independent para meters, CMOBSRN and NMOBSRN, in the species file. If LAI is below 3.0 or if vege tative N-status greater than 0.3, the “actual’ fraction of CH2O and N that can potentially be mobilized will be between the “normal” fraction (CMOBSRN, and NMOBSRN) and th eir maximum fractions (CMOBSRX and NMOBSRX) set in the species file. These fr actions can also be decreased if dormancy has been induced. The decision points and information flow controlling mobilization responses are described in Figures 5-2 (CH2O) and 5-3 (N), while the calculations are discussed in more detail in th e following section on dormancy. Like leaves and stems, STOR also senesces as the plant ages. The senescence rate is set by the species parameter SENSR and is a function of age/temperature (physiological days/day), the same factors that influence senescence of leaves, stems, and roots. Water and N stress do not affect th e senescence rate of STOR. The vegetative stage of maturity (VSTAGE) does not decrease as leaves are senesced. The fraction of new growth allocated to STOR was estimated to vary from 30% for young plants (VSTAGE 3.0) up to 45% for older plants (VSTAGE >7.0) with a maximum partitioning of 90% of new growth during dormancy. These partitioning rates were derived primarily from data of 14C experiments conducted by Beaty et al. (1974) showing 33%-63% of 14C translocated in bahiagrass plan ts was found in the stolons. The 90% maximum partitioning parameter during peak dormancy was determined by running the model and varying the fraction until leaf a nd stem growth were reduced to expected levels during the winter. The targeted level of STOR mass was between 8000 and 16 000 kg DM ha-1. These levels were consistent with data for combined root and stolon mass in Pensacola bahiagrass ranging from 16 500 to 24 000 kg DM ha-1 reported by Impithuksa

PAGE 120

106 and Blue (1985) and 9300 – 15 150 kg DM ha-1 reported by Beaty et al. (1975). These levels are higher than the measured rhizome weights of 2900 – 9800 kg rhizome DM ha-1 of Beaty et al. (1964) for Pens acola bahiagrass; however, they also reported an additional 4200 – 10 600kg DM ha-1 of “miscellaneous” material which likely was primarily senescing and decomposing rhizomes. With the partitioning set as listed above, the STOR senescence rate was set at 1.5% d-1 to produce the targeted levels of STOR. There were two experiments in our comp ilation of experiments that measured stolon mass and, thus, were available for eval uating the model’s pred iction of STOR. Of the two experiments, only Rymph and Boote (20 02) measured solely stolons. These data were used for evaluating the fit of the predic tions (Table 5-1) while the data from Boote et al. (1999) which combined st ems with stolons, were used to provide information on the potential range of stolon mass f ound in the field. Despite ha ving only 19 data pairs in the evaluation, the fit of the pr edicted values was quite good w ith a slope of 0.81 and an r2 of 0.86 for the linear regression of predicted to ob served data. The index of agreement or dindex was low in relation to those parameters at 0.615 (a d-index of 1.0 indicates perfect agreement), likely because the model consistent ly overpredicted stolon mass (Figure 5-4). Increasing the stolon senescence rate or shif ting some of the partitioning from stolon to root would have lessened the overprediction and possibl y improved the prediction of roots (Table 5-1); however, comp ared to the stolon + stem data of Boote et al. (1999), the model consistently underpredicted stolon mass. The extent of this underprediction easily exceeded any anticipated level of stem mass that was included in the measurements. Given those opposing results, the parameters we re left as described and not optimized further.

PAGE 121

107 Dormancy Several studies have reported a seasonal de cline in forage yield of bahiagrass and bermudagrass in the fall and winter months desp ite the relatively mild temperatures of the Southeast (Burton et al., 1988; Gates et al., 2001; Chambliss, 2002; Sinclair et al., 2003). Winter dormancy in Bouteloua gracilis has been modeled as a function of low soil temperatures (Detling et al., 1979); howeve r, studies by Sincla ir et al. (2003) demonstrated a controlling role of daylengt h in triggering the dormancy response. Gates et al. (2001) and Sinclair et al. (2003) reported the greatest reductions in yield to occur in mid-winter, indicating that the degree of dormancy may increase with decreasing daylength. The exact mechanism of dormancy has not been described. There are no reports of significant reductions in photos ynthetic rate beyond the expe cted effects of cooler temperatures. Assuming that herbage grow th is reduced to a greater degree than photosynthetic rate, photosynthate must be diverted elsewher e. DiPaola et al. (1982) reported an increase in root mass of turf-type bermudagrass for a short period in the fall after shoot growth had ceased, then root grow th ceased as well. Sinclair et al. (2003) found no difference in below-ground biomass growth of bahiagrass between natural daylength and extended daylength treatment s but did observe a higher proportion of leaves to total plant mass during the shortdaylength months in the natural daylength treatments. Although Sinclair et al. (2003) did not report specifically on stolon growth, Rymph and Boote (2002) observe d a sharp increase in bahi agrass stolon mass in October and November in Gainesville, FL. Given this information, the most likely sink for bahiagrass photosynthate produc tion during dormancy is the st olons. Thus, the dormancy process model added to CROPGRO increases partitioning to and decreases mobilization

PAGE 122

108 from the stolon. Increasing partitioning to STOR provides a sink for assimilate that results in decreased apparent above-ground shoot production during th e winter and will also serve as a N source after dormancy. Minimizing mobilization of N from STOR during dormancy curbs the regrowth rate of th e shoot. Parameters we re added to allow a dormancy effect on photosynthesis in the event that evidence is presented supporting that phenomenon. For the present time, that featur e has been disabled by assigning a constant value of 1.0 to this factor so it wi ll not modify photosynthetic rate. The dormancy process in the forage ve rsion of CROPGRO has a partitioning component (Fig. 5-1) and mobilization compone nts (Figures 5-2 and 5-3), each of which is modeled independently. These processes have their own daylength thresholds for initiating/ending dormancy and for imposing maximum dormancy, and they have their own parameters defining the pattern of dorma ncy. These parameters are defined in the species file but another eco type-specific parameter can scale the degree of dormancy exhibited, to create differential response s of individual ecotypes if needed. As illustrated in Fig. 5-1, the “normal” fraction of new growth partitioned to STOR (FRSTR) is set through a lookup-table matching the current V-stage to the fraction of new growth to be partitioned to leaf, stem, and STOR. Partitioning to root is then calculated by difference with all fractions summ ing to 1.0. When daylength is less than 12.5 h, dormancy is progressively induced a nd partitioning to STOR is increased above the “normal” fraction to reflect a diversion of growth away from leaf and stem to stolon. When daylength is above 12.5 h, the dormancy partitioning factor (PPTFAC) is set to 0.0 and partitioning to STOR is not affected As daylength drops below 12.5 h and approaches the maximum dormancy threshold of 10.5 h, PPTFAC increases linearly to

PAGE 123

109 1.0 (Figure 5-5), increasing th e partitioning of new growth towards the maximum fraction specified in the species file (FRSTRMX) using the equation: FRSTR = (FRSTRMX FRSTR) PPTFAC + FRSTR (Eq. 5-1) where FRSTR is the fraction of new growth partitioned to STOR (initially set by V-stage), FRSTRMX is the maximum frac tion partitioned to STOR under maximum dormancy (0.90), and PPTFAC is the dormancy adjustment factor for partitioning to STOR. Using the current parameters and the sche me in Fig. 5-1, if the crop is at V-stage 7.0, FRSTR is 0.40 or 40% of the new grow th would normally be partitioned to the STOR organ, 35% would be allocated to roots, 20% to leaves, and 5% to stems. If the daylength is 11.5 h; however, PPTFAC is 0.5, and the adjusted partitioning to STOR would be [(0.90-0.40)*0.5+0.40] or 0.65 and 65% of all growth would be partitioned to STOR. With this increase in FRSTR, partiti oning to the other vegetative organs must be recalculated. The relative propor tions of “normal” allocation to leaf, stem, and root are used to calculate the reduced partitioning fractions to non-STOR organs as follows: TFRLF = FRLF/(FRLF + FRSTM + F RRT) (1-FRSTR) (Eq. 5-2) where TFRLF is a temporary variable holding the new, reduced fraction of new growth partitioned to leaf, FRLF is the “normal” fraction of new growth partitioned to leaf, FRSTM is the “normal” fraction of new grow th partitioned to stem, and FRSTR is the fraction of new growth partitioned to STOR after being adjusted for dormancy. Using

PAGE 124

110 the example listed above, the original 20% fr action of new growth partitioned to leaves becomes [0.20/(0.20+0.05+0.35)*(1-0.65)] or 12%. Stem partitioning is adjusted using a similar equation and root partitioning is calculated by difference so that all fractions add to 1.0 or 100%. Thus, the new stem partitioning fraction would be [0.05 / (0.20+0.05+0.35) (1-0.65)] or 3% and the fraction partitioned to root would be 20%. Once adjusted for dormancy, partitioning amon g organs is only affected by water-stress and N-stress; both of which increase partitioning of new growth to roots, and, consequently, decrease partitioning to l eaves, stems, and STOR (Figure 5-1). Separate flowcharts depict decision poin ts and the flow of information for CH2O mobilization (Figure 5-2) and N mobilization (Figure 5-3). Potential mobilization of CH2O and N is a function of the “normal” fraction mobilized, maximum fraction mobilized, LAI, vegetative N-status, and st age of dormancy or daylength. On each simulation day, mobilization adjustment fact ors are calculated from the current LAI, vegetative N-status (leaf, stem, root, and ST OR), and daylength. These adjustment factors are calculated from separate functi ons defined in the speci es file. Depending on the daily values of these factors, the fraction of CH2O and N that can potentially be mobilized from STOR can range from 10% of the normal potential mobilization rate (CMOBSRN, NMOBSRN) to a maximum rate (CMOBSRX, NMOBSRX) defined in the species file. The calculation of the potential fraction of N mobilized is a three-step process and while the day’s poten tial fraction of mobilizable CH2O and N are calculated independently, they both use the same basi c equations. The N m obilization calculations will be used as an example for both nutrients. The N-status adjustment factor (VNMOBR) is applied as follows:

PAGE 125

111 NMOBSR = (NMOBSRN + (NMOBSRX -NMOBSRN)*VNMOBR) (Eq. 5-3) where NMOBSR is the fraction of available N that can potentially be mobilized from STOR on the current day, NMOBSRN is the “nor mal” fraction of available N that can be mobilized from STOR, NMOBSRX is the ma ximum fraction of available N that is potentially mobilizable, and VNMOBR is the Nstatus factor (ranging from 0.0 to 1.0). Vegetative N-status is calcul ated as the current N mass of the leaves, stems, roots, and STOR relative to the maximum potential N ma ss if all four organs contained their maximum (new growth) N concentrations. Th is factor may have a value ranging from a minimum of 0.0 (no increase in potential N or CH2O mobilization when N-status is 30% of maximum or lower, increasing in a curvilinear pattern to 1.0 (maximum CH2O and N mobilization) as N-status approaches 70% of maximum (Figure 5-6a). Thus, potential mobilization of N from STOR may rema in at NMOBSRN if VNMOBR is 0.0 and increases towards NMOBSRX as VNMOBR approaches 1.0. Next, NMOBSR is updated to reflect the LAI and dormancy adjustment factors: NMOBSR = (NMOBSR + (NMOBSRX -NMOBSR)*LAIMOBR)*PPMFAC (Eq. 5-4) where LAIMOBR is the LAI adjustment factor (ranging from 0.0 to 1.0) and PPMFAC is the dormancy or daylength factor (ranging from 0.1 to 1.0) c ontrolling mobilization. The LAI adjustment factor (LAIMOBR) has a value of 0.0 (no in crease in potential mobilization) when LAI is greater than 3. 0 and increases to 1.0 (maximum potential

PAGE 126

112 mobilization) as LAI decreases to 1.0 or below (Figure 5-6b). If VNMOBR is 1.0, or LAIMOBR is 0.0, then LAIMOBR has no eff ect; however, if VNMOBR is less than 1.0 and LAIMOBR is greater than 0.0, then the fraction of available N potentially mobilized increases from the value of NMOBSR calculated in equation 5-3. For NMOBSR to equal NMOBSRX either VNMOBR or LAIMOBR or both factors must have a value of 1.0. The dormancy factor, PPMFAC, is applied af ter all other adjustments are made and acts as a scalar, decreasing potential mobiliz ation from STOR as daylength decreases below a dormancy threshold. If daylength is greater than 12.5 h, PPMFAC is set to 1.0 and there is no effect of dormancy on mobilization (the adjusted fraction is multiplied by 1.0). As daylength decreases from 12.5 h to 10.5 h, PPMFAC decreases linearly from 1.0 to 0.1. This decreases the potential m obilization fraction proportionally (e.g., if daylength is 11.4 h, PPMFAC will be 0.5, and the fraction of available N that is potentially mobilizable will be 50% of what would be mobilized if only adjusted for LAI and N-status). As daylength decreases below 10.5 h, maximum dormancy is achieved and PPMFAC remains at 0.1. The daylength thresholds for initiation of dormancy and maximum dormancy as well as the mini mum and maximum value for PPMFAC are defined in the species file and, hence, may be adjusted for other species. There is also an ecotype-specific parameter in the ecotype file that allows for differences in various ecotypes’ dormancy responses relati ve to the species response. Potential mobilization of CH2O and N from roots is also reduced by PPMFAC during dormancy to control depletion of root reserves and root mass over the winter. Reducing the fraction of CH2O and N that can potentially be mobilized preserves STOR and root mass for spring regrowth and, some times, reduces leaf and stem growth rates

PAGE 127

113 during dormancy. Mobilization rates of leaf and stem reserves are unaffected by these dormancy factors although the amount of mobilization may increase due to the limitations placed on STOR and root mobiliza tion. It should be stressed that this discussion addresses the fraction of CH2O and N that can potentially be mobilized in a day. The fraction actually mobilized in the model depends on daily photosynthate production, N uptake, and crop N demand as well as the actual pool of CH2O and N that can be mobilized. Running simulations with the forage vers ion, the partitioning and mobilization factor values cycled thro ugh the seasons, initiating dorma ncy on September 1, reaching maximum dormancy on November 22 then lessening the degree dormancy on January 19 until completely breaking dormancy on April 9 (Figure 5-5b). To evaluate the accuracy of herbage mass predictions during dormancy, we simulated the cool season bahiagrass yield experiment from Ona, FL (Gates et al., 2001) and compared model predictions of herbage mass to the measured results. Although the r2 of the linear regression of the predicted data against the obs erved was only 0.29 (Figure 57), the other measures of model fit were quite positive. The slope of the linear regression was nearly 0.8 and the intercept was 494 kg DM ha-1, indicating systematic error was relatively small. The mean of the predicted herbage mass was 2160 kg ha-1 (data not shown) and was quite low compared to summer harvests at other locations and is also very close to the mean of the observed values of 2100 kg ha-1. Other statistical indicato rs were promising with a dindex of 0.68 and an RMSE of 291 kg DM ha-1 using 41 data pairs in the analysis. The dormancy factors were effective in reducing herbage yiel d during the winter, which reduced the water stress predicted in the spring compared to levels predicted by the

PAGE 128

114 previous CSM model. Despite the reduced winter growth, spring water stress was still predicted in experiments wher e there should have been none. Modification of the code for modeling evapotranspiration (ET) may be required to correct this. The current ET model uses a daily energy balance approach (Priestley-Taylor) and is linked to photosynthesis only if evaporative demand exceed s potential root water uptake. There is no feedback from low photosynthesis (in winter months) to transpiration as may actually occur via stomatal regulation effects. While photosynthate production was limited by the reduced leaf growth in the forage versi on, ET still responded to the mild winter temperatures with potentially excessive rates of water loss. Freeze Damage Freezing temperatures can cause two types of damage in the unmodified existing CROPGRO. If temperatures drop to a sub-le thal threshold (FREEZ1), all leaves on the plants are killed, if temperat ures fall to the lethal thre shold (FREEZ2), all plants are killed and the simulation is terminated. Earlier experiences in deriving species parameters for bahiagrass (Rymph et al., 2003) showed that plants would not recover from even a FREEZ1 event. Inspection of the code revealed that while CH2O and N from the roots could be mobilized, there wa s no photosynthesis (no l eaf area after freeze) and the mobilized CH2O was expended on maintenance respiration and N uptake. With no assimilate left, N uptake went to refill N de pleted tissues and no new growth occurred. Eventually the simulated crop exhausted CH2O reserves and died. Adding stolon (STOR) with large pools of carbohydrate and N, as well as decreasing maintenance respiration coefficients, helped solve the pr oblem of allowing regrowth. In addition, peripheral changes were made to correct CH2O and N allocation in the model in general to restrict CH2O use for N demand and refill of N-depleted tissue when total growth is

PAGE 129

115 low as it is after total leaf area loss. Th ese changes are not directly related to the adaptation of the model for bahiagrass and will be addressed in another venue. The freeze damage routine in CROPGRO need ed to be more flexible for modeling forage crops. While we kept the two fr eeze threshold parameters, FREEZ1 and FREEZ2, we added a progressive leaf da mage process as well as a co ld-hardening routine to allow gradual loss of leaves and increased freeze tolera nce with exposure to cold temperatures. For these changes we adapted the approach used by Kanneganti et al. (1998) in the ALFACOLD model. The FREEZ1 parameter was re-defined here as the temperature that triggers the onset of freeze damage to leaves and stems (pre viously it resulted in killing of all foliage, and thus eliminating photosynthesis). Temp eratures below FREEZ1 result in leaf and stem death at a fraction defined by a freezing death-constant parameter (FRZDC) in the species file. The death-constant defines the proportion of leaf and stem killed on a given day for each degree that the minimum daily temperature (TMIN) drops below FREEZ1. Plants will survive for a period with no livi ng leaves if the STOR tissue has not been killed. The FREEZ2 parameter is still the lethal temperature th reshold that kills all plants and ends the simulation. With the addition of the STOR tissue, FREEZ2 now reflects the TMIN required to kill the STOR tissue, which ma y be protected in the top layer of soil. When FREEZ2 is reached, all STOR tissue is killed, currently there is no gradual killing of STOR. However, a cold hardening proce ss for STOR tissue only was added that will allow plant survival below this temperature if the crop exhibits cold-hardening and has been previously exposed to cold temperat ures (Fig. 5-8). The degree of tolerance increases with added exposure to cold temperatures and is reversed by warm

PAGE 130

116 temperatures. A dehardening process begins when daylength begins to increase. Dehardening reduces the de gree of cold-hardening and operates using temperature thresholds defined in the species file separa te of the cold-hardeni ng thresholds. Thus, it is possible for a crop to experience cold harden ing and dehardening at the same time. As there is no evidence of a significant cold hardening effect on bahiagrass, the cold hardening process was disabled by setting a ll related parameter values to -25C, a temperature below that experienced in our calibration datasets. As the cold-hardening function was not used for bahiagrass, discussi on of its function will be limited to that already presented along with a flowchart (Figure 5-8) de picting decision points and information flow related to calculating the de gree of cold hardeni ng and adjustment of FREEZ2. Establishing accurate values for FREEZ1 and FREEZ2 was not possible given the dataset that we had compiled. There were no datasets with measurements of freeze damage and no sites that experienced lethal freezing temperatures. The coldest site was Eagle Lake, TX with an average of only 18 nights below freezing each winter. Freeze damage parameters were set to start killing leaves at FREEZ1 = -5 C with a daily death loss of 5% of the leaves and stems for each 1C that nighttime minimum temperatures drop below -5C. Work by Kimball and Salis bury (1973) favor a threshold temperature closer to 0C as they observed a 90% death lo ss of bahiagrass plants kept at 0C for 3 d and 100% loss at -5C. However, they us ed very young plants (2 leaves, no stolons present) grown in pots, whereas established stands grown in the field should tolerate lower temperatures especially for shorter durations. As there were no reports of plant

PAGE 131

117 death due to freezing, an arbitrary value of -18C was assigned to FREEZ2 to avoid predicting plant death at a ny of the test sites. Applying these parameters predicted freeze damage on 2 to 3 nights per year in Eagle Lake, TX, reduced herbage mass by as much as 22% in a single night yet still allowed for rapid regrowth in the spring. This approach to predicting freeze damage is flexible and easily adjusted for each species, however the parameter values are somewhat arbitrary and calibration of these parameters against measured data on freeze damage is still required for them to be useful at more northern latitudes or at altitude. Photosynthesis Bahiagrass is a C4 plant, wherein CO2 is first fixed in the mesophyll cells by phosphoenolpyruvate carboxylase (PEPCase) then tr ansported to the bundle sheath cells and released near the chloroplasts. This results in much higher concentrations of CO2 in the bundle sheath than in the mesophyll, enha ncing the carboxylation efficiency of the ribulose 1,5-bisphosphate (Rubisco) in the bundle sheath chloroplasts by reducing the competitive inhibition of Rubisco carboxylase activity by O2. This concentrating of CO2 around Rubisco imparts different responses to light, CO2 concentration, and temperature. A typical C4 plant has a high potential photosynthetic rate whic h is not light saturated except at high light levels. Also, C4 photosynthesis is quite sens itive to low temperatures while tolerant of high temp eratures. Additionally, C4 photosynthesis is not as responsive to elevated CO2 concentrations as the C3 photosynthetic pathway exhibited in many temperate grasses and legumes. CROPGRO offers two options for pred icting daily canopy photosynthesis. The simplest is the daily canopy photosynthes is option which estimates daily canopy photosynthetic rate directly from total dail y photosynthetically activ e radiation (PAR).

PAGE 132

118 The leaf-level option predicts hourly leaf photos ynthesis for sunlit and shaded leaves and, through hourly iteration, estimates daily canopy photosynthesis (Boote and Pickering, 1994). Modeling of daily canopy and hourly leaf-level photosynthesis is based on asymptotic exponential representations of ca nopy (for daily canopy optio n) and leaf (for leaf-level option) pho tosynthetic responses to light. Photosynthetic responses to environmental factors are imposed through vari ous modifiers which simulate some of the underlying biochemical proce sses controlling photosynthe sis. Most of the CO2 concentration effects listed above are simula ted in the CSM version of CROPGRO, with no modification of the daily canopy photosynt hetic option required beyond determining appropriate parameter values defining the CO2 response. These values were estimated from the canopy photosynthesis measurements of Boote et al. (1999) The leaf-level photosynthesis option, however, does not accurate ly reflect some of the effects of the CO2-concentrating mechanism on the sensitivi ties of quantum efficiency (QE) and maximum photosynthetic rate to temperature and CO2 concentration, and thus it required modification. The asymptotic exponential equation desc ribing leaf photosynt hetic response to light (PGLF) is -QE*PARLF LFMAXPGLF = LFMAX 1.0 () e (Eq. 5-5) where LFMAX is the potential light-saturated leaf photosynt hetic rate at 30C and 350 L L-1 CO2, and 21% O2, QE is the quantum efficiency of the species defined at the same conditions, and PARLF is the absorbed PAR mol quanta m-2 s-1. The leaf QE, which defines the initial slope of th e light response curve, is spec ified in the species parameter file but is modified by factor s for hourly temperature and CO2 concentration and leaf N

PAGE 133

119 concentration. The cultivar-specific maximum leaf photosynthetic rate (LFMAX), describes the asymptote of the response and is also specified in the species parameter file and subsequently multiplied by factors for SLW, hourly air temperature, leaf N concentration, CO2 concentration, and nighttime chilling. The effects of CO2 concentration and temperature are estimated separately for QE and LFMAX but both originate in a single equation: 28990 -3.949+ 8.314(T+273) =e, (Eq. 5-6) where Tau ( ) is the specificity factor of Rubisco for CO2 relative to O2 or the relative tendency of Rubisco to fix CO2 rather than O2, T is the temperature in C, and 8.314 is the universal gas constant. Tau refl ects the decreasing solubility of CO2 relative to O2 as temperature increases. The effect of on CO2 compensation point ( *)and photosynthetic response (at QE and saturating rate) to CO2 depend further on intercellular CO2 concentration as parameterized by the Ribulose 1,5-bisphosphate (RuBP)-limited photosynthesis algorithm of Fa rquhar and von Caemmerer (1982) To allow adjustment of the ratio to account for the CO2 concentrating effect of C4 photosynthesis, we added a “CO2-concentrating” factor (CCNEFF) to the species file. Tau is multiplied by CCNEFF to predict the relative increase in CO2 concentration at the Rubisco site in the bundle sheath cells. Increasing “ times CCNEFF” in CROPGRO reduced and decreased the sensitivity of both the QE and LFMAX CO2 concentration adjustment factors to higher temperatures and CO2 concentrations. An arbitrarily selected value of 10 for CCNEFF produced a of 5 L CO2 L-1 at 30C (Figure 5-9), consistent with reports of CO2 compensation points ranging from 0.0 to 5.0 for various C4 species (Rajendrudu and Das, 1981). When the CO2 concentration

PAGE 134

120 factors for QE and LFMAX were re-scaled to 1. 0 at 30C (where they are defined to 1.0) to account for the lowered *, the responsiveness of LFMAX to high CO2 concentrations decreased. The new system predicted a CO2 adjustment factor for LFMAX (CO2MAX) for an atmospheric CO2 concentration of 700 L CO2 L-1 giving only a 6% increase in the predicted light-saturated leaf rate when coupled with a ratio of inte rcellular to atmospheric CO2 concentration (Ci:Ca) set to 0.4. In contrast, B oote et al. (1999) measured an 18% increase in photosyntheti c rate under high light when atmospheric CO2 concentration was doubled from ambient (360 L CO2 L-1) to 700 L CO2 L-1. Working backwards from the measured data we arri ved at a value of 3.0 for CCNEFF. This produced a CO2MAX which increased from 1.0 at 350 L CO2 L-1 to 1.18 at 700 L CO2 L-1 (Figure 5-10) yet still lowered to 18 L CO2 L-1 at 30C (Figure 5-9). This is near the CO2 compensation values of 4-14 L CO2 L-1 reported for the C4 grass Panicum maximum by Bolton and Brown (1980) and cons iderably lower than the 55 L CO2 L-1 predicted by the CSM version of the CROPGRO code for C3 species. Prediction of the effect of temperatur e on QE was also improved using CCNEFF set to 3.0. The predicted temperature/CO2 concentration factor (C O2QE) varied only 7% from 1.034 at 0C to 0.967 at 45C (Figure 511), demonstrating the substantial lack of temperature sensitivity of QE in C4 species. This response is quite small compared to the 46% range predicted by the CSM version over th e same temperature range. Using a QE of 0.0541 for a hypothetical C3 species and 0.065 for the C4 bahiagrass, and CCNEFF=3.0, the predicted QE of the C3 species will be higher than that of the C4 only at temperatures below 5C. This crossover temperature is considerably lower than the approximate 20to 23C crossover temperatur es reported by Ku and Edwards (1978) and

PAGE 135

121 Monson et al. (1982). Increasing CCNEFF to 10.0 increases the temperature at which C4 QE surpasses the C3 QE to 14C which presents an argument for a CCNEFF of 10.0. However, as QE has the most influence on l eaf photosynthetic rate at low light levels while LFMAX dominates at higher levels CCNEFF was set to 3.0, favoring the improved CO2MAX predictions. Evaluation of the performance of the modified C4 photosynthesis options was somewhat complicated compared to other asp ects of the model. Differences in growth resulting from changes in photosynthetic ra te are compounded each day, so a simple comparison of changes in plant mass are inadequate. To judge the photosynthetic response to increasing CO2 concentration, the experiment of Johnson et al. (2001) was simulated using 350 and 700 L CO2 L-1 ambient CO2 concentrations. This experiment was chosen as it began with an established crop. Gro ss canopy photosynthesis for the first day of each simulation was compar ed. Using the daily canopy photosynthesis option, the gross canopy photosynthesis at 700 L CO2 L-1 was 11% greater than at 350 L CO2 L-1 (data not shown), verifying that the CO2 concentration factor was producing the intended increase in photosynthetic rate Similarly, when the noontime maximum photosynthetic rates for sunlit l eaves predicted in the leaf-level option were compared, the rate predicte d for the 700 L CO2 L-1 level was 17% greater than for the 350 L CO2 L-1 level, again, the desired response. Despite the increased photosynthetic rates, predicted herbage mass was not increased at the higher CO2 concentrations except at the highest N fertilization treatments regard less of the photosynthesis option used. This lack of a growth response despite an apparent increase in photosynthetic rate is likely due to excessive predicted N-st ress limiting growth. For most of the

PAGE 136

122 experiments, N should have been limiting only in the low N fertility treatments, but there was no mention of N deficiency symptoms in the reports. However, even under normal atmospheric CO2 concentrations (330 to 350 L CO2 L-1), predicted N-stress frequently limited growth in the low and mid-level N treatments. Unlike the early–season N-stress predicted using the unmodified CSM version of CROPGRO, this predicted N-stress recurred throughout the growi ng season. At elevated CO2 concentrations, photosynthetic rate increased but without extra available N, N stress also increased and the extra CH2O produced was stored as mobile carbohydrate in the stems, r oots and, primarily, in the STOR organ. Thus there was a sizeable increase in STOR and root mass with no corresponding increase in leaf mass (Figure 5-12). It is difficult to assess the cause of the predicted N shortage. Potential errors in the prediction of root mass and N uptake capacity are prime candidates. However, there are several other potential c ontributors outside of the crop m odel itself. As there were no actual soil analyses available for any of the experiments, the soil profiles used in the simulations were based on soil surveys with so il water retention pa rameters derived from a database using the k-Nearest Neighbor appr oach (Jagtap et al., 2004). Similarly, the model’s default coefficients were used to set initial propor tions of microbial, intermediate, and passive OM fractions for the SOM processes. Other initial soil conditions such as available water and N, amount and composition of residue, and even initial plant mass were estimated by starting the simulation one year before the actual experiment, letting the system predict the in itial conditions. Inaccur acies in any and all of these estimates contribute to the problem. The predicted N stress was generally worst

PAGE 137

123 in the spring while the majority of the N l eaching occurred in the winter, implying that mineralization rates of SOM may be excessive during the winter. Overall Model Performance – Herbage Mass and N Concentration Plots of the predicted growth from the experiments were encouraging as much for the patterns of growth as for the levels of production (Figure 5-13). Predicted stolon mass and root mass were on the order of reports by Blue (1973) and B eaty et al. (1968). Stolon growth increased during the shortdaylength months and decreased over the summer while root mass increased over the summer and was partially depleted over the winter. Leaf and stem growth were reduced in the fall and winter but still aggressive in the spring (Figure 5-14), especially when viewed on an annual cycle (Fig. 5-13). The performance of the new CROPGRO-forage ve rsion was greatly improved over the CSM model where an artificial defoliation had to be imposed during the winter to slow growth (Figure 5-15). The modifications implemented in the fo rage version of CROPGRO improved the performance of the model over the CSM version in almost every measure. Compared to the unmodified CSM version running the bahiag rass parameters developed in an earlier step of the model adaptation (R ymph et al., 2003), the forage version had a slope closer to 1.0, an intercept closer to 0.0, and a larger r2 of the linear regr ession of predicted against observed values, a larg er d-index and smaller RMSE for predicted herbage mass (Table 5-2). The greatest improvements were a 14% increase in the d-index and a large increase in the r2. A 46% increase in the slope of the linear regression indicated that systematic error was reduced, meaning that le ss of the total error was due to the way the model worked and more was due to random erro r such as errors in estimating the stubble mass for each experiment.

PAGE 138

124 In absolute terms, the prediction of herbage mass by the forage version was encouraging as well. A d-index value of 0.81 was achieved using the leaf-level photosynthesis option with a nearly identical fit for the daily canopy version (0.82) (Table 5-2). The mean predicted values for th e two versions were also nearly identical at 3145 and 3150 kg DM ha-1 for the leaf-level and daily ca nopy versions, respectively, both very close to the observe d mean of 3066 kg DM ha-1. The RMSE with the leaf-level photosynthesis was less at 857 kg DM ha-1 and when expressed as a coefficient of variation, was 28% of the observed mean. Th is was slightly better for the daily canopy option. The greater slopes of 0.694 (leaf-l evel) and 0.62 (daily canopy), indicate that systematic errors may be less than the ra ndom errors. However, the scatterplot of predicted against observed herb age mass (Figure 5-16) indicate s that there is a tendency to overpredict herbage mass at higher levels an d underpredict at lower levels of growth. A tendency to overpredict at higher leve ls of herbage mass production may explain the apparent contradiction be tween the lower predicted mean for all experiments and the overprediction of herbage mass compared to the underprediction of the CSM version for the Ona, FL 1996-1998 experiment (Figure 5-15). The herbage mass production for the Ona 1996-1998 experiment was the highest of any of the experiments in our dataset. If the forage model overpredicts at higher levels of herbag e mass production, it would be most apparent in the Ona 1996-1998 simulati on. In comparing the predicted herbage mass across different N fertilization treatments, the forage model di d not appear to be excessively sensitive to N fertilization, part icularly at Ona in the 1996-1998 experiment. In the Ona experiment, increasing annua l N fertilization from 0 to 468 kg N ha-1 resulted in a large increase in fora ge production but doubling the N application to 942 kg N ha-1

PAGE 139

125 yr-1 had little effect on forage production (J ohnson et al., 2001). Th e forage version of CROPGRO predicted a similar pattern of response of herb age mass to N fertilization (Figure 5-17), with little or no increase in growth at the higher N fertilization levels. At Eagle Lake, TX where lower levels of N fertil ization were applied, the magnitude of both the observed and predicted responses to higher le vels of N fertilizer was greater than for Ona. This, along with the other improvement s noted, suggests that despite the excessive predictions of N stress, the N-response of the new forage model is realistic. Herbage N concentration prediction was also improved for the forage version although not to as great an extent as herbag e mass. While herbage mass predictions using leaf-level and daily canopy photos ynthesis options were similar, the two options diverged more in their predictions of herbage N con centration. The mean predicted herbage N concentration using the leaf -level option (14.6 g N kg-1) was slightly lower than the mean observed value of 15.6 kg N ha-1, while the daily canopy opt ion produced a mean N concentration of 17.6 kg N ha-1, greater than the observed mean (Table 5-2). The seasonal dynamics of herbage N concentra tion was predicted better by the modified model (Figure 5-18), especially for Ona, FL The slope of the linear regression of predicted herbage N concentration was not any closer to 1:1 for the modified model (0.78 and 0.60 for the daily canopy and leaf-level options, respectively); however, the d-index values for herbage N concentration was improved, as evidenced by the improved N predictions at Ona (Figure 5-18). Scatterplo ts of predicted agai nst observed herbage N concentration show no clear pa ttern of overor underpredict ion except that the magnitude of overprediction may be greater than for unde rprediction (Figure 5-19) This is evident when herbage N concentration response is compared across the va rious N-fertilization

PAGE 140

126 treatments. While observed herbage N concentrations increased as N fertilization increased at Ona, FL, the magnitude of the predicted response was much greater (Figure 5-20). Herbage N concentration was underpre dicted by the forage model for the 0 kg N ha-1 yr-1 fertilization treatment but overp redicted for the 468 and 942 kg N ha-1 yr-1 treatments. While no statistical comparison was made, the magnitude of overprediction for the 942 kg N ha-1 yr-1 appears to be greater than for the 468 kg N ha-1 yr-1 treatment. Predicted herbage N concentrations for the Eagle Lake, TX experi ment did not show clear patterns of overpredicti on or underprediction (Figure 5-20). The magnitude of the predicted response to increas ing the annual fertilization rate from 168 to 336 kg N ha-1 appeared to be greater than the response to increasing N fertilization form 0 to 168 kg N ha-1 yr-1. This is consistent with the observed pattern of response to N fertilization, particularly for the last two growing seasons. It appears that at high N fertilization levels the forage version of the CROPGRO model is predicting a higher concentration of N than is observed but this is not reflecte d in increased herbage mass production. The d-index values for herbage N mass (> 0.85 for both leaf-level and daily canopy options) were higher than those for either herbage mass or herbage N concentration, implying a better fit for herbage N mass than for either component used to calculate it (Table 5-2). The other measures of fit also supported this conclusion (Table 5-2, Figure 5-21). The slightly high herbage mass a nd slightly low herbage N concentration predicted using the leaf-level photosynthesis option likely offset each other to produce the improved fit. Some other offsetting errors could be responsible for the good fit of the predictions produced by the daily canopy phot osynthesis option as both herbage mass and N concentration were overpredicted. The one fit parameter that was slightly worse for

PAGE 141

127 the forage model predictions of both herb age N concentration and herbage N mass was their slopes (Table 5-2). Bo th photosynthesis options produc ed linear regression results with greater (albeit only sli ghtly) slopes than were seen with the CSM model, indicating that the error likely lies with the m odel and was not due to random error. Conclusions The addition of a storage organ and addition of dormancy regulation of partitioning/mobilization to the CROPGRO stru cture was an integral part of the new improvements for dormancy and re-growth processes of the forage version of CROPGRO. The two modifications to the model structure allowed seasonal reductions in herbage production by partitioning a greate r proportion of new growth to STOR while reducing mobilization from STOR and roots. This produced a more realistic pattern of bahiagrass growth than was possible with the CSM version of the model. The perennial nature of bahiagrass was also better simu lated by the modification of the freeze damage process to allow gradual loss of leaves and stems. Damage proportional to the temperature can be imposed without terminati ng the simulation. The modification to the leaf-level photosynthesis option successfully simulated the lowered response to CO2 concentration observed in C4 plants. This response was not reflected in the pattern of herbage mass accumulation as N stress, which was already overpredicted, was heightened by the increased photosynthate production. Thus under CO2 enrichment, this resulted in most of the additional photosynthate accumu lating in STOR and roots as mobile CH2O. Additional testing using sites with finer te xtured soils may help differentiate the roles of the crop, soil water, and SOM pro cesses in predicting excessive N stress. Modification of the ET process to include a link to photosynth esis may also improve the soil water balance processes. While the freeze damage parameters may be adequate for

PAGE 142

128 the subtropical areas used here, calibration ag ainst data from cooler regions may improve the accuracy of the dormancy and freeze damage parameters and broaden the model’s utility. The modifications made to CROPGRO to cr eate the forage version predicted more realistic growth patterns for bahiagrass and improved prediction of herbage mass, herbage N concentration, and herbage N mass. While not ready for general release, the performance of the forage version of CROPGRO was quite good and should progress from this calibration stage on to more vigorous testing.

PAGE 143

129 Table 5-1. Summary of perf ormance of the forage version of CROPGRO in simulating mass of below-ground plant organs (kg DM ha-1). Mean ObservedPredictedSlopeInterceptd-index r2 RMSE Stolon (kg DM ha-1) 6463 4268 0.3262160 0.382 0.052613 Root (kg DM ha-1) 5930 8507 0.8143682 0.615 0.862638 Table 5-2. Summary of perf ormance of the CSM (unmodified) and forage version of CROPGRO in simulating five experiments to predict herbage mass, herbage N concentration, and herbage N mass. Observed values Mean Herbage Mass 3066 kg DM ha-1 Herbage N Conc. 15.6 g N kg-1 Herbage N Mass 49.1 kg N ha-1 CSM version Leaf-level photosynthes is option predictions Mean Slope Intercept d-index r2 RMSE Herbage Mass 3165 0.475 1709 0.714 0.24 1003 Herbage N Conc. 18.0 0.671 7.6 0.661 0.24 7.8 Herbage N Mass 59.3 1.133 3.7 0.849 0.66 22.8 Daily canopy photosynthesis option predictions Herbage Mass 3092 0.461 1677 0.734 0.28 930 Herbage N Conc. 19.4 0.806 6.8 0.690 0.35 7.8 Herbage N Mass 61.4 1.108 7.0 0.849 0.68 22.5 Forage version Leaf-level photosynthes is option predictions Mean Slope Intercept d-index r2 RMSE Herbage Mass 3145 0.694 1017 0.814 0.44 857 Herbage N Conc. 14.6 0.600 5.3 0.776 0.36 5.5 Herbage N Mass 50.5 1.270 -12.1 0.892 0.75 19.5 Daily canopy photosynthesis option predictions Herbage Mass 3150 0.616 1261 0.823 0.46 782 Herbage N Conc. 17.6 0.782 5.4 0.767 0.40 6.2 Herbage N Mass 58.6 1.227 -1.9 0.863 0.71 22.1

PAGE 144

130 Calculate partitioning to leaf, stem, STOR, root Daylength? Daylength <10.5 h Daylength 10.5< and <12.5 h Daylength >12.5 h PPTFAC=1.0, Maximum partitioning to STOR PPTFAC 0.0< and <1.0, Intermediate increase in partitioning to STOR PPTFAC=0.0, No increase in partitioning to STOR Re-calculate partitioning to leaf, stem, STOR, root Adjusting for PPTFAC Increase partitioning to root based on the more limiting stress Is there H2Oor N-stress? Final partitioning fractions Yes No Read “normal” V-stage partitioning lookup table and maximum fraction partitioning to STOR from species file. V-Stage? Re-calculate partitioning to leaf, stem, STOR, adjusting for increased root partitioning. Figure 5-1. Schematic of daily partitioni ng of new growth among vegetative tissues for the forage version of CROPGRO. Diamond-shaped boxes indicate decisionpoint variables predicted in the model, rectangle-shaped boxes indicate an action in the model.

PAGE 145

131 Leaf, stem, STOR, and root N concentration? VNSTAT <0.3 VNSTAT 0.3< and <0.7 VNSTAT >0.7 VNMOBR 0.0< and <1.0, Intermediate increase in mobilization from STOR VNMOBR=1.0, Maximum mobilization from STOR VNMOBR=0.0, Mobilization from STOR not increased by N-status Calculate vegetative N-status (VNSTAT) Leaf, stem, STOR, and root mass? Read N-status mobilization parameters (VNMOBR) from species file. Read minimum and maximum mobilization parameters for CH2O (CMOBSRN, CMOBSRX) for STOR from species file Calculate potential fraction of CH2O mobilized from STOR (CMOBSR) for current day. Calculate potential CH2O mobilization from crop (CMINEP) for current day. Calculate potential fraction of CH2O mobilized from leaf and stem for current day. Calculate potential fraction of CH2O mobilized from root for current day. LAI <1.0 LAI 1.0< and <3.0 LAI >3.0 LAIMOBR 0.1< and <1.0, Intermediate increase in mobilization from STOR LAIMOBR=0.0, Mobilization from STOR not increased by low LAI LAIMOBR=1.0, Maximum mobilization from STOR LAI? Read LAI mob ilization parameters (LAIMOBR) from species file. LAI effect Dormancy effect N-status effect Daily potential mobilizable CH2O Start Finish Daylength? Daylength <10.5 h Daylength 10.5< and <12.5 h Daylength >12.5 h PPMFAC=0.0, Minimum mobilization from STOR PPMFAC 0.1< and <1.0, Intermediate reduction in mobilization from STOR PPMFAC=1.0, Mobilization from STOR not decreased by daylength Read dormancy mobilization parameters (PPMFAC) from species file. Set CH2O mobilization fraction from STOR to “normal” values (CMOBSRN). Figure 5-2. Schematic of the calculation of potential mobilization of CH2O from leaf, stem, root and ST OR in the forage version of CROPGRO. Diamond-shaped boxes indicate decision-point variables predicted in the model, rectangle-shaped boxes indicate an action in the model.

PAGE 146

132 Leaf, stem, STOR, and root N concentration? VNSTAT <0.3 VNSTAT 0.3< and <0.7 VNSTAT >0.7 VNMOBR 0.0< and <1.0, Intermediate increase in mobilization from STOR VNMOBR=1.0, Maximum mobilization from STOR VNMOBR=0.0, Mobilization from STOR not increased by N-status Calculate vegetative N-status (VNSTAT) Leaf, stem, STOR, and root mass? Read N-status mob ilization parameters (VNMOBR) from species file. Read minimum and maximum mob ilization parameters for N (NMOBSRN, NMOBSRX) for STOR from species file Calculate potential fraction of N mobilized from STOR (NMOBSR) for current day. Calculate potential N mobilization from crop (NMINEP) for current day. Calculate potential fraction of N mobilized from leaf and stem for current day. Calculate potential fraction of N mobilized from root for current day. LAI <1.0 LAI 1.0< and <3.0 LAI >3.0 LAIMOBR 0.1< and <1.0, Intermediate increase in mobilization from STOR LAIMOBR=0.0, Mobilization from STOR not increased by low LAI LAIMOBR=1.0, Maximum mobilization from STOR LAI? Read LAI mob ilization parameters (LAIMOBR) from species file. LAI effect Dormancy effect N-status effect Daily potential mobilizable N Start Finish Daylength? Daylength <10.5 h Daylength 10.5< and <12.5 h Daylength >12.5 h PPMFAC=0.1, Minimum mobilization from STOR PPMFAC 0.1< and <1.0, Intermediate reduction in mobilization from STOR PPMFAC=1.0, Mobilization from STOR not decreased by daylength Read dormancy mobilization parameters (PPMFAC) from species file. Set N-mobilization fraction from STOR to “normal” values (NMOBSRN). Figure 5-3. Schematic of the calculation of potential mobilization of N from leaf, stem, root and STOR in the forage version o f CROPGRO. Diamond-shaped boxes indicate decision-point variables predicted in the model, rectangle-shaped boxes indicate an action in the model.

PAGE 147

133 y = 0.8137x + 3682.1 r2 = 0.8584 y = 0.137x + 5545.1 r2 = 0.3979 y = 0.1212x + 5153.5 r2 = 0.19310 5000 10000 15000 20000 25000 0500010000150002000025000Observed Stolon Mass (kg DM ha-1)Predicted Stolon Mass (kg DM ha-1)Rymph and Boote, 2002Stolon onlyBoote et al., 1999Stem + Stolon 360 L CO2 L-1700 L CO2 L-1 Figure 5-4. Predicted vs. observed stolon mass for bahiag rass grown in the field in Gainesville, FL in 2001(), and in temperature and CO2 gradient greenhouses at 360 L CO2 L-1 (), and 700 L CO2 L-1 (). Predicted using the modified leaf-level photosynthesis option in the forage vers ion of CROPGRO. Linear regression lines ( ) are presented with their corresponding equations and r2 values along with a line designating a theoretical 1:1 relationship (– –).

PAGE 148

134 0.00 0.25 0.50 0.75 1.00 1.25 0.06.012.018.024.0Daylength (h)Factor Value a 0.0 0.2 0.4 0.6 0.8 1.0 1.21 J a n 2 0 F e b 1 1 A p r 3 1 M a y 2 0 J u l 8 S e p 2 8 O c t 1 7 D e cDateFactor Value b Figure 5-5. The a) controlling functions and b) seasonal expression of the predicted effect of daylength on incremental (i ncrease above baseline) partitioning to STOR( ) or mobilization from STOR ( ) in the forage version of CROPGRO.

PAGE 149

135 0.00 0.25 0.50 0.75 1.00 1.25 0.00.20.40.60.81.01.2N statusVNMOBR a 0.00 0.25 0.50 0.75 1.00 1.25 0123456LAI (m2 leaf m-2 land)LAIMOBR b Figure 5-6. Mobilization f actors in the forage versi on of CROPGRO that affect mobilization from STOR as a function of a) vegetative N status and b) LAI.

PAGE 150

136 y = 0.7976x + 494.23 r2 = 0.28980 500 1000 1500 2000 2500 3000 3500 0500100015002000250030003500Observed Herbage Mass (kg DM ha-1)Predicted Herbage Mass (kg DM ha-1) Figure 5-7. Predicted ( ) vs. observed bahiagrass herbage mass for late-season harvests at Ona, FL in the 1993-1994 and th e 1995-1996 growing seasons using the modified leaf-level photosynthesis option in the forage version of CROPGRO. The linear regression line ( ) is presented with its corresponding equations and r2 values along with a line designating a theoretical 1:1 rela tionship (– –).

PAGE 151

137 TMIN >FREEZ1 TMIN FREEZ2 No Freeze damage. Leaf, stem, and STOR mass unchanged. Calculate proportion of leaf and stem lost to freeze damage. Re-calculate leaf, stem and STOR mass Read freeze damage and cold hardening threshold temperatures from species file Minimum daily temperature (TMIN)? Is daylength increasing? Calculate change in cold-hardening. TMIN
PAGE 152

138 0 25 50 75 100 125 0102030405060Temperature (C)CO2 Compensation Point (L CO2 L-1) Figure 5-9. Predicted CO2 compensation point for the CSM version of CROPGRO ( ) and for two hypothetical C4 species with a Ci/Ca of 0.4 and a CO2 concentrating factor of either 3 ( ) or 10 ( ) predicted using the forage version of CROPGRO.

PAGE 153

139 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0100200300400500600700800Atmospheric [CO2] (L L-1)[CO2] Factor for LFMAX (CO2MAX) a 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0100200300400500600700800Atmospheric [CO2] (L L-1)[CO2] Factor for QE (CO2QE) b Figure 5-10. Relative CO2 concentration effect on a) CO2 factor for LFMAX and b) QE for a C3 species in the CSM version ( ) and for a C4 species in the forage version of CROPGRO using a C i/Ca of 0.4 and CCNEFF of 3 ( ).

PAGE 154

140 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0102030405060Temperature (C)[CO2] Factor for LFMAX (CO2MAX) a 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0102030405060Temperature (C)[CO2] Factor for QE (CO2QE) b Figure 5-11. Relative temperature effect on a) CO2 factor for LFMAX and b) QE for a C3 species in the CSM version ( ) and for a C4 species in the forage version of CROPGRO using a C i/Ca of 0.4 and CCNEFF of 3 ( ).

PAGE 155

141 0 3000 6000 9000 12000 15000 180001 J a n 2 M a r 1 M a y 3 0 J u n 2 9 A u g 2 8 O c t 2 7 D e cDatePredicted Growth (kg DM ha-1) Stolon Herbage Root Figure 5-12. Predicted growth of bahiagrass components under 350 L CO2 L-1 atmospheric CO2 stolon ( ), root ( ), and herbage ( ) relative to predicted growth under 700 L CO2 L-1 atmospheric CO2 stolon ( ), root ( ), and herbage ( ) using the forage version of CROPGRO.

PAGE 156

142 0 3000 6000 9000 12000 15000 180002 7 J a n 28 Mar 27-May 2 6-J u l 2 4 -S e p 2 3-No v 22-Jan 2 3 Ma r 2 2-Ma y 21-Ju l 19-SepDatePredicted Growth (kg DM ha-1)a 0 3000 6000 9000 12000 15000 18000Ja n 7 9 Mar-79 May-79 Ju l 7 9 S e p-79 Nov-79 J a n-80 Mar -80 May-80 Jul-80 Sep-80 N o v-80 Ja n 8 1 Mar-81 May-81 J ul 8 1 S e p-81DatePredicted Growth (kg DM ha-1)b Figure 5-13. Observed bahiagrass herbage mass ( ), predicted stolon ( ), root ( ), and herbage mass ( ) for bahiagrass grown in a) Ona, FL with 468 kg N ha-1 yr-1or b) Eagle Lake, TX with 168 kg N ha-1 yr-1. Predicted values generated by the forage version of CROPGRO.

PAGE 157

143 0 4000 8000 12000 160001 J a n 2 M a r 1 M a y 3 0 J u n 2 9 A u g 2 8 O c t 2 7 D e cDatePredicted Growth (kg DM ha-1) Stolon Stem Leaf Root Figure 5-14. Predicted stolon ( ), root ( ), leaf ( ), and stem ( ) growth for bahiagrass grown with 468 kg N ha-1 yr-1 in Ona, FL in 1997 using the forage version of CROPGRO.

PAGE 158

144 0 1000 2000 3000 4000 5000 600027 Ja n 28-Mar 27-May 26-Jul 24-Sep 23N ov 22 Ja n 23-Mar 2 2Ma y 21-Jul 19-SepDateHerbage Mass (kg DM ha-1)0 0.5 1 1.5 2 2.5 3Stress Factor (0 1) a 0 1000 2000 3000 4000 5000 600027 Ja n 28-Mar 27-May 26-Jul 24-Sep 23N ov 22 Ja n 23-Mar 2 2Ma y 21-Jul 19-SepDateHerbage Mass (kg DM ha-1)0 0.5 1 1.5 2 2.5 3Stress Factor (0 1)b Figure 5-15. Observed herbage mass ( ), predicted herbage mass ( ), water stress ( ), and N stress ( ) of bahiagrass grown with 468 kg N ha-1 yr-1 at Ona, FL. Predicted using the le af-level photosynthesis option in a) the forage version of CROPGRO or b) the unmodified CSM version of CROPGRO. To simulate dormancy in the CSM simulation, photosynthesis was reduced by 70% and partial defoliation was simulated in the winter. Both predicted stress factors are based on a 0–1 scale with 0= no stress/normal growth rate and 1= severe stress/no growth. The broken horizontal line ( ) denotes the stubble mass left in the fi eld after each harvest.

PAGE 159

145 y = 0.6938x + 1017 r2 = 0.4360 2000 4000 6000 8000 02000400060008000Observed Herbage Mass (kg DM ha-1)Predicted Herbage Mass (kg DM ha-1) Figure 5-16. Predicted ( ) vs. observed bahiagrass herbag e mass for five experiments, using the modified leaf-level photosynt hesis option in the forage version of CROPGRO. The linea r regression line ( ) is presented with its corresponding equations and r2 values along with a line designating a theoretical 1:1 relationship (– –).

PAGE 160

146 0 1000 2000 3000 4000 5000 6000 70001 J a n 1 M a y 2 9 A u g 2 7 D e c 2 6 Ap r 2 4 A u gDateHerbage Mass (kg DM ha-1) a 0 1000 2000 3000 4000 5000 60001 J a n 1 M a y 2 9 Au g 2 7 D e c 2 5 A p r 2 3 A u g 2 1 D e c 2 0 A p r 1 8 A u gDateHerbage Mass (kg DM ha-1) b Figure 5-17. Observed/predicted bahiagrass herbage mass for a) the 0 kg N ha-1 yr-1 ( and ), 468 kg N ha-1 yr-1 ( and ), and 942 kg N ha-1 yr-1 ( and ) treatments at Ona, FL and b) the 0 kg N ha-1 yr-1 ( and ), 168 kg N ha-1 yr-1 ( and ), and 336 kg N ha-1 yr-1 ( and ) treatments at Eagle Lake, TX. Predicted values generated by the forage version of CROPGRO.

PAGE 161

147 0 10 20 30 4027-Ja n 2 8-Mar 2 7-May 2 6 -Jul 2 4 S e p 23-Nov 22-Jan 2 3-Mar 2 2-May 2 1 -Jul 1 9 S e pDateHerbage N Conc. (g N kg-1)a 0 10 20 30 402 5-Jan 26 M ar 25 May 2 4-Jul 2 2-Sep 21-No v 20-Jan 20-Mar 19-May 18-Jul 1 6-Se p 15-N ov 14-Jan 15-M ar 14 M ay 13-Jul 1 1 S epDateHerbage N Conc. (g N kg-1)b Figure 5-18. Observed bahiagrass herbage N concentration ( ) and predicted bahiagrass herbage N concentration using the fo rage version of CROPGRO and the leaf-level option ( ), or daily canopy option ( ). For a) bahiagrass grown at Ona, FL with 468 kg N ha-1 yr-1, or b) bahiagrass grown at Eagle Lake, TX with 168 kg N ha-1 year-1.

PAGE 162

148 y = 0.5998x + 5.2568 r2 = 0.36080.0 10.0 20.0 30.0 40.0 0.010.020.030.040.0Observed Herbage N Conc. (g N kg-1)Predicted Herbage N Conc. (g N kg-1) Figure 5-19. Predicted ( ) vs. observed bahiagrass herbag e N concentration for five experiments, using the modified leaf-lev el photosynthesis option in the forage version of CROPGRO. Th e linear regression line ( ) is presented with its corresponding equations and r2 values along with a line designating a theoretical 1:1 relationship (– –).

PAGE 163

149 0 10 20 30 401 J a n 1 M a y 2 9 A u g 2 7 D e c 2 6 Ap r 2 4 A u gDateHerbage N Concentration (g N kg-1) a 0 10 20 30 401 J a n 1 M a y 2 9 A u g 2 7 D e c 2 5 A p r 2 3 A u g 2 1 D e c 2 0 A p r 1 8 A u gDateHerbage N Concentration (g N kg-1) b Figure 5-20. Observed/predicted bahi agrass herbage N concentration for a) the 0 kg N ha-1 yr-1 ( and ), 468 kg N ha-1 yr-1 ( and ), and 942 kg N ha-1 yr-1 ( and ) treatments at Ona, FL and b) the 0 kg N ha-1 yr-1 ( and ), 168 kg N ha-1 yr-1 ( and ), and 336 kg N ha-1 yr-1 ( and ) treatments at Eagle Lake, TX. Predicted values generated by the forage version of CROPGRO.

PAGE 164

150 y = 1.2703x 12.119 r2 = 0.74590 50 100 150 200 050100150200Observed Herbage N Mass (kg N ha-1)Predicted Herbage N Mass (kg N ha-1) Figure 5-21. Predicted ( ) vs. observed bahiagrass herbage N mass for five experiments, using the modified leaf-level photosynt hesis option in the forage version of CROPGRO. The linea r regression line ( ) is presented with its corresponding equations and r2 values along with a line designating a theoretical 1:1 relationship (– –).

PAGE 165

151 CHAPTER 6 SUMMARY AND CONCLUSIONS Perennial tropical forages such as bahiagrass (Paspalum notatum Flgge) fulfill roles as both a feedstuff and a component of nutrient recovery/recycling on livestock farms. These crops may be managed for any combination of yield, feed quality, and nutrient uptake. A crop model that predicts the growth and composition of the forage as well as the N dynamics of the crop-soil system could be a useful tool in finding the proper balance between these objectives. The objective of this research was to de velop a tool to pred ict the growth and composition of bahiagrass that responds to environmental and management inputs. In this effort, a bahiagrass growth study was conducted in the summer and fall of 2001 to examine, in detail, changes in growth as the plant advances through the regrowth cycle and through the end of the grow ing season (Chapter 3). We used information from this experiment to supplement the existing literatur e in an effort to ad apt the CSM version of the CROPGRO model to simulate the growth and composition of bahiagrass (Chapter 4). Our initial testing revealed limitations in the model structure preventing realistic simulation of seasonal growth patterns. To address these limitations, the source code was modified to add a storage organ (stolon) and dormancy process, the freeze damage process was modified to allow gradual loss of leaves and stems in winter, and the leaflevel photosynthesis op tion was modified to better simulate C4 photosynthesis (Chapter 5).

PAGE 166

152 Bahiagrass Growth Study An established stand of bahiagrass was m onitored weekly to quantify changes in growth patterns and photosynthesis during la te summer (18 July – 12 September) and fall (12 September – 7 November) regrowth peri ods. Sod cores, dug weekly, were washed and plant material separated into leaf blade, stem, stolon, and root components. Leaf and canopy photosynthesis were measured periodic ally during both regrowth periods. There was a large decrease in root mass in the first period, thought to be due to previous management of the field. Stem mass decreased during each period, likely due to our reclassification of leaves from stems pr ior to elongation and recl assification as leaf blade after elongation. Leaf growth was lo wer in the second period and only half as many leaves emerged per tiller during the second period as in the first. Stolon mass did not change during the first period but incr eased dramatically during the second. Although canopy photosynthesis was lower in th e second period, consistent with the reduced leaf growth, leaf phot osynthetic rate did not change Thus, leaf growth was reduced in the fall as growth of stolons increased. Development of Species File Parameters for Bahiagrass New species, cultivar, and ecotype files were created to allow simulation of bahiagrass growth and composition with CROPG RO. A set of files was created from data available in the literature and minimal curve fitting. Two experiments were selected from the literature and simulated in CRO PGRO. The measured data from the experiments were split into two sets with the high and low N fertilization treatments used for calibration of the model a nd the medium N fertilization tr eatments used to test the model. The fit of the herbage mass predictions from the literature-based species file were good with an index of agreement (d-index) of 0.85, a slope of 0.611 and an r2 of 0.54

PAGE 167

153 though there was a tendency to underpredict, es pecially at higher amounts of herbage mass. Fit of herbage N concentration was not as accurate with a d-index of 0.605 and an r2 of 0.18 and a tendency to overpredict herb age N concentration. Performance using both the leaf-level and daily canopy photos ynthesis options of the CSM model was similar. Three limitations posed by the model structure were identified that inhibited simulation of realistic growth patterns. Thes e items were 1) a lack of a storage organ (stolon) in the model structure, 2) the fr eeze damage routine was not flexible enough as it would either kill all leaves and compromise the simulation or killed none and allowed excessive leaf growth to pers ist, and 3) lack of a dorma ncy process to slow herbage growth during the winter. An optimized specie s file was developed but it performed little better than the literature-based file. It was concluded that the optimi zation did little more than try to compensate for the limitations of the model and did not represent a better representation of the biology. Adapting CROPGRO to Model Perennial Trop ical Grasses: Structural Changes to the Model The source code for the CSM version of CROPGRO was modified to address the limitations identified while developing new species parameter files. A new organ (STOR) was added to simulate stolons. Much of the new code was a duplication of that already in use for simulating leaves and stems. Partitioning of new growth and mobilization of N were handled differently for STOR. New functions were added to increase mobilization of both N and carbohydrate (CH2O) from STOR when 1) LAI was low (<3), and / or when whol e-plant N status was adequa te to high (>30%). These factors allowed faster regrowth after harvests and in the sp ring. The STOR organ was an integral component of the dormancy proce ss. Daylength both induces and breaks

PAGE 168

154 dormancy. Once induced by a daylength of <12.5 h, dormancy functions increase partitioning to STOR and decrease mobilization of both N and CH2O from both STOR and roots. The degree of dormancy and ex tent of modification of partitioning and mobilization reach a maximum when daylength decreases to 10.5 h. As daylength begins to lengthen above 10.5 h, the degree of dormancy decreases with dormancy breaking when daylength reaches 12.5 h. As part itioning to STOR increases, partitioning to leaves, stems, and roots decreases, resulting in reduced herbage growth over the winter. The freeze damage process was modified with a “death constant” reducing leaf and stem mass 5% for every degree the daily mini mum temperature drops below -5C. Code for a reversible cold-hardeni ng process with dehardening as daylength increases was added for future modeling of more cold-hardy sp ecies but was not enabled for bahiagrass. A “CO2-concentrating factor” was also added to the leaf-level photosynthesis code to allow better simulation of C4 photosynthesis. This reduced the sensitivity of quantum efficiency and maximum leaf photosynt hetic rate to temperature and CO2, thus reflecting the influence of the high CO2 concentrations around the bundl e sheath chloroplasts of C4 plants. Three more experiments were added to our collection for calibrating the new forage version of the model. Most of the new parameters were either selected from the literature or calculated to impart the desired respons e. General parameters such as maximum fraction of new growth partitioned to STOR during dormancy were adjusted while running the model. Performance of the new fo rage version was better than for the CSM version of the model. Based on the larger da taset, the d-index for predicted herbage mass was 0.81 and 0.82 using the leaf-level and da ily canopy options (respectively) in the new

PAGE 169

155 version compared to 0.71 and 0.73 for the CSM version. All other measures of fit improved as well. Prediction of herbage N concentration was not as accurate as prediction of herbage mass, but similarly im proved over the CSM version. The predicted CO2-induced leaf-level photosynthe tic response to elevated CO2 was of the same magnitude as observed in the field but the higher rate was not reflected in increased herbage mass. Frequent N stress was predicte d, resulting in the increased photosynthate production being stored in STOR and roots. Th is was likely an overpre diction of N stress as most treatments simulated were designed so that N would not be limiting growth. The cause of the prediction of excessive N stress wa s not identified but could be due to errors in estimating soil and soil organic matte r parameters or initial conditions. Implications of the Research As the first working version, the modi fied forage version of CROPGRO CSM model marks a significant step in adapting the model to be tter represent the biology and management of tropical perennial forages. Viable mechanisms have been added to simulate some of the unique as pects of both growth and harv est management of perennial tropical forages. Parameters may be easily adjusted as new knowledge becomes available. Adaptation to other perennial fo rages should be much simpler with the new version’s structure. Given these advances additional testing a nd calibration is still needed to improve the robustness of the model for general use. Beyond the added utility of the model, th e performance may be adequate to use under limited conditions. Accurate prediction of single season produc tion may be a ways off, limiting its usefulness to farmers and consultants looking for a tool to make midseason management changes. However, multi-year simulations comparing relative differences between management strategies may produce useful results regarding herbage

PAGE 170

156 production and even herbage N composition. Th us, there is potential utility for use in research and forecasting. Given the excessive N stress frequently predicted by the model, use to evaluate N management strategies may require extensiv e calibration of soil parameters or species N parameters for the give n location prior to running the evaluation. Future Research The most immediate priority to further model development is to identify the cause of the excessive N-stress predictions. On ce that is done, other parameters can be recalibrated. The next step beyond that would be to test the temperature parameters over a wider range of conditions. Addressing those two concerns should improve the usefulness of the model consid erably. Confidence in shorter-term simulation results should increase. After this, use of the mode l in long and short-term testing of nutrient management strategies and use of the results in directing research priorities may be quite viable. The adaptation of the forage vers ion to other forages such as alfalfa (Medicago sativa) and temperate grasses could follow soon. Looking towards future improvements, two additional modifications that would impr ove the utility and accuracy of the model are the development of a “true” C4 photosynthesis option and expans ion to include the ability to predict changes in forage quality. While simulating C4 photosynthetic proce sses down to the molecular level may not be necessary, a dual compartment system that simulated mesophyll and bundle sheath processes separately could allow more flexib ility and accuracy in simulating the response of C4 photosynthesis to CO2 concentration and temper ature. The bundle sheath photosynthesis component would essentially be an extension of the current leaf-level photosynthesis option, hence useful for both C3 and C4 species.

PAGE 171

157 Looking in new directions for improving the model, we should look to the potential users. As cattle are the ultimate consumers of forage, the priorities of the model should extend in that direction. The forage versi on of CROPGRO already pr edicts leaf and stem crude protein, lignin, and carbohydrate com position, and tracks mobile carbohydrate content and physiological age. From these va riables, a new output file could be created to provide information targeted to livestock feeding and perf ormance. Neutral detergent fiber (cell wall, an indicator of energy content and “fill” value) concentration might be predicted by subtracting mobile carbohydrate concentr ation from total carbohydrate concentration. Forage dige stibility might be predicted by correlation with predicted lignin and cell wall concentration, physiol ogical age, and, possi bly, crude protein concentration. These additions are natura l extensions of this adaptation process, initiating new directions for model development.

PAGE 172

158 APPENDIX A CROPGRO CSM PARAMETER CODE DEFINITIONS Variable definitions listed in order of th eir appearance in the species file, followed by those in the ecotype file. PARMAX PAR at which photosynthesis is 63% of the maximum value (PHTMAX). Used in daily canopy photosynthesis calculations (moles[quanta] m-2 d-1). Used to define asymptotic response of daily canopy assimilation to daily photosynthetic photon flux. PHTMAX Maximum amount of CH2O which can be produced if photosynthetically active radiation (PAR) is very high (3 times PARMAX) and all other factors are optimal (g[CH2O]/m-2 d-1). Used to define asymptotic response of daily canopy assimilation to daily phot osynthetic photon flux. CCMP Canopy CO2 compensation point (CO2 at which daily photos ynthesis is 0.0). FNPGN(4) Critical leaf N concentration for function to reduce photosynthesis due to low leaf N levels (4 values for function CURV). This is a two-sided generic curve using only the two leftmost point s to describe both canopy and leaf photosynthesis response to leaf N concentration. Same leaf-derived function is used for leaf and canopy options; however, subroutine PHOTO modifies it for DAILY option with a square root change. TYPPGN Character variable sp ecifying the type of function to use for the relationship between leaf N concentration and photos ynthesis (for use in function subroutine CURV). QDR defines a quadratic curve as photosynthesis increases from zero at the minimum leaf N concentration to its maximum at the maximum leaf N concentration. FNPGT(4) Critical values of temperature for the functions to reduce canopy PG under non-optimal temperatures (in function CU RV). A two-sided generic curve that describes daily canopy assimilation in re sponse to average daytime temperature TYPPGT Character variable sp ecifying the type of function to use for the relationship between temperature and photosynthesis (for use in function s ubroutine CURV). LIN means linear 4-point lookup. XLMAXT (6 VALUES) and YLMAXT (6 VALUE S) This is a 6-point lookup function, that describes relative ra te of photosynthetic electron-transport (YLMAXT) in response to temperature (XLMAXT). Us ed only for Leaf option and affects computed light-saturated leaf photosynthesis.

PAGE 173

159 FNPGL(4), Describes relative effect of minimum night temperature (TMIN) on next day's single leaf light-satur ated photosynthesis rate. Use only the two leftmost values, TYPPGL Character variable sp ecifying the type of function to use for the relationship between TMIN and next day' s single leaf light-saturat ed photosynthesis rate. QDR quadratic/parabolic response fro m 0.0 to 1.0 (no effect) on leaf photosynthesis. PGEFF Quantum efficiency of leaf photosynthesis, defined at 350 ppm CO2, 21% oxygen, and 30 C. Set from published va lues in literatu re, Ehleringer and Bjrkman (1977). SLWREF Specific leaf weight at which LFMAX is defined. LNREF Value of leaf N above which canopy PG is maximum (for standard cultivar). Leaf N concentration at which LFMAX is defined. PGREF Reference value for leaf level phot osynthesis used in canopy light response curve ( mol[CO2] m-2 s-1). Used only for the “Daily” canopy assimilation option. Allows cultivar variation in LFMAX to change daily canopy assimilation. Not used for the leaf-level, he dgerow photosynthesis option. PROLFI, PROLFG, PROLFF "Maximum", "n ormal growth", and "final" protein concentrations of leaf tissue. PROSTI, PROSTG, PROSTF "Maximum", "normal growth", and "final" protein concentrations of stem tissue. PRORTI, PRORTG, PRORTF "Maximum", "normal growth", and "final" protein concentrations of root tissue. PLIPLF, PLIPST, PLIPRT, PLIPSH, PLIPNO Li pid concentration of leaf, stem, root, shell, and nodule tissues, respectively. PLIGLF, PLIGST, PLIGRT, PLIGSH, PLIGSD, PLIGNO Lignin concen tration of leaf, stem, root, shell, seed, a nd nodule tissue, respectively. PCARLF, PCARST, PCARRT, PCARSH, PCAR SD, PCARNO Carbohydrate-cellulose concentration of leaf, stem, root, shel l, seed, and nodule tissues, respectively. CMOBMX Maximum rate of mobilization of carbohydrate from vegetative tissues, fraction of available ca rbohydrate pool per day. NMOBMX Maximum rate of mobilization of protein from vegetative tissues, during reproductive growth, fraction of av ailable protein pool per day. NVSMOB Maximum rate of mobilization of carbohydrate from vegetative tissues, during vegetative growth, fraction of available protein pool per day. XLEAF(8) V-stage at which partitioni ng to leaves is YLEAF(I) (leaf nodes). YLEAF(8) Partitioning fraction to leaf gr owth at V-stage XLEAF(I) ( g[leaf] g[veg. plant]-1).

PAGE 174

160 YSTEM(8) Partitioning fraction to stem gr owth at V-stage XLEAF(I) (g[stem] g[veg. plant]-1) FRSTMF Fraction of daily dr y weight increase in vegeta tive plant parts which goes to stems after the day on which the maximum number of V-stages occurs (NDVSTG). (g[stem] g[veg]-1). FRLFF Fraction of daily increase in vegeta tive dry weight which goes to leaves after the day on which the maximum number of V-stages occurs (NDVSTG). (g[leaf] g[veg]-1). FRLFMX Maximum leaf partitioning (g[leaf] g[veg]-1). FINREF Specific leaf area (S LA) of leaves of standard crop cultivar when plants emerge (cm2[leaf] g[leaf]-1). SLAREF Specific leaf area (S LA) for new leaves during peak vegetative growth for the standard cultivar. (cm2 g-1). SIZREF The size of a normal upper node leaf (nodes 8 – 10) of standard cultivar. (cm2 leaf-1). VSSINK Vegetative stage be yond which sink-limited leaf ar ea expansion can no longer limit photosynthesis or leaf area growth. SLAMAX Maximum specific leaf area for new leaves when grown under low radiation, but optimum water and temperature conditions (cm2 g-1). The (thinnest) leaves can be under low light. SLAMIN Minimum specific leaf area for ne w leaves when grown under high radiation and optimum water and temperature conditions (cm2 g-1). The (thickest) leaves can be under high light. XVGROW(6) V-stage at which maximum leaf area growth per plant since emergence is YVGROW(I). (number of leaf nodes). YVREF(6) Maximum leaf area grown per pl ant at V-stage XVGROW(I), for reference cultivar. (cm2 plant-1). XSLATM(5) Temperature values for functi on that reduces specific leaf area (SLA) (C). YSLATM(5) Array which describes the effect of temperature on specific leaf area. Relative temperature effect on specific leaf area of newly-formed leaves. FREEZ1 Temperature below which all leaves are killed, but stems, pods, and seeds remain alive, development to maturity proceeds, and seeds can grow on mobilized reserves (C). FREEZ2 Temperature below which plant gr owth stops completely, development and crop growth simulation stops (C). ICMP Light compensation point for senescen ce of lower leaves because of excessive self-shading by crop canopy (moles m-2 d-1).

PAGE 175

161 TCMP Time constant for senescence of lowe r leaves because of excessive self-shading by crop canopy (thermal days). XSTAGE(4) V-stage at which SENPOR(I) fract ion of cumulative leaf growth will have been senesced if no water stress occurred (number of leaf nodes). SENPOR(4) The fraction leaf abscission that normally occurs by a given V stage. Proportion of leaf weight grown which w ill have been senesced by a given Vstage (XSTAGE(I)) if no water stress has occurred prior to this V-stage (XSTAGE(I)) -normal vegetative senescence does not occur if prior water stress has already reduced leaf mass. XSENMX(4) V-stage at which maximum frac tion of cumulative leaf growth vulnerable to loss due to water stress is SE NMAX(I) (number of leaf nodes). SENMAX(4) Maximum proportion of total le af weight as a function of V-stage (XSENMX(I)) which can be senesced due to water stress. The maximum allowable leaf abscission due to water st ress that is allowed to occur by a given V stage. RTDEPI Depth of roots on day of plant emergence. (cm). RFAC1 Root length per unit root mass (cm g-1). RTSDF Maximum fraction of r oot length senesced in a gi ven layer per physiological day when water content in a given laye r falls below 25 % of extractable soil water. RWUEP1 The ratio of evapor ative demand:root water uptak e at which the water stress factor, TURFAC, is 1.00 (maximum stre ss), and declines below that ratio. TB, T1, T2, TMAX Base temperature (TB), first optimum (TO1), second optimum (TO2), and maximum temperature (T M). Relative rate of phenological development is zero at TB and TM, and optimum (1.0) at TO1 and TO2. Possible to use linear (LIN) or parabolic (QDR) between these points. XVSHT(10) Node number on main stem for us e in computing height and width growth rates. YVSHT(10) Length of internode (m) vs. position on the main stem defined by XVSHT (m node-1). YVSWH(10) Increase in canopy width per node develope d on the main stem (m node-1).

PAGE 176

162 APPENDIX B SPECIES, CULTIVAR, AND ECOTYPE FI LES FOR THE UNMODIFIED CSM VERSION OF CROPGRO Literature-Based Species file:

PAGE 177

163*BAHIA SPECIES COEFICIENTS !*PHOTOSYNTHESIS PARAMETERS 60.0 90.0 0.55 PARMAX, PHTMAX,KCAN 20.0 1.30 .0090 CCMP,CCMAX,CC EFF; CO2 EFFECT ON PGCAN 0.75 3.0 10 .0 10.0 QDR FNPGN(4),TYPPGN-LF N EFFECT ON PG 12.0 25.0 38.0 50.0 LIN FNPGT(4),TYPPGT-TEMP EFFECT-CANOPY PG -5.0 7.0 35.0 45.0 55.0 60.0 XLMAXT (6 VALUES) 0.0 0.0 1.0 1.0 0.0 0.0 YLMAXT (6 VALUES) 7.0 18.0 45.0 57.0 QDR FNPGL (4),TYPPGL-TMIN EFFECT-LEAF PG .0650 0.20 0.80 2.0 PGEFF SCV KDIF, LFANGB 11/5/02 .0035 .0002 .2000 3.00 1.760 SLWRE F,SLWSLO,NSLOPE,LNREF,PGREF 0.0 .001 .002 .003 .0035 .004 .005 .006 .008 .010 XPGSLW(1-10) .162 .679 .8 67 .966 1.000 1.027 1.069 1.100 1.141 1.167 YPGSLW(1-10) !*RESPIRATION PARAMETERS 0.35E-04 .0029 RES30C,R30C2 2.556 2.556 .360 2.830 RNO3C,RNH4C,RPR O,RFIXN 1.242 3.106 2.174 .929 0.05 1.13 RCH20,RLIP,RLIG,ROA,R MIN,PCH2O !*PLANT COMPOSITION VA LUES .220 .110 .050 .110 .070 .033 PROLFI,PR OLFG,PROLFF,PROSTI,PROSTG,PROSTF .101 .040 .022 .039 .039 .038 PRORTI,PROR TG,PRORTF,PROSHI,PROSHG,PROSHF .115 .115 .300 .0 15 .019 .800 SDPROS,SDPROG ,PRONOD,PROMIN,PROMAX,THETA

PAGE 178

164 .602 .697 .702 .251 .750 .480 PCARLF,PCAR ST,PCARRT,PCARSH,PCARSD,PCARNO .025 .020 .020 .011 .050 PLIPLF,PL IPST,PLIPRT,PLIPSH,PLIPNO .040 .060 .070 .100 .030 .070 PLIGLF,PLIG ST,PLIGRT,PLIGSH,PLIGSD,PLIGNO .050 .050 .050 .0 40 .040 .050 POALF,POAST,P OART,POASH,POASD,POANO .063 .063 .057 .100 .030 .050 PMINLF,PMIN ST,PMINRT,PMINSH,PMINSD,PMINNO !*SEED COMPOSITION VALUES 7.168 23.65 0.908 0.430 LIPTB,LIPOP T,SLOSUM*100,CARMIN !*CARBON AND NITROGEN MINING PARAMETERS 0.025 0.75 .280 0.050 1.00 0.15 CMOBMX,CADSTF, CADPR1,NMOBMX,NVSMOB,NRCVR PD XPODF 0.04 0.08 0.04 0.08 ALPHL,ALPHS ,ALPHR,ALPHSH !*NITROGEN FIXATION PARA METERS .050 .160 0.01 0.0 0.04 0.05 SNACTM,NODRGM,DWNODI,TTFIX,NDTHMX, CNODCR 7.00 28.0 35.0 44 .0 LIN FNNGT(4),TY PNGT-TEMP EFF ON NOD GR OWTH 5.00 23.0 35.0 44.0 LIN FNFXT(4),TYPFXT-TEMP EF F ON N FIX -.15 0.20 1.00 10 .0 LIN FNFXD(4), TYPFXD-REL SW-DRY E FF ON N FIX -.02 .001 1.00 2. 00 LIN FNFXW(4), TYPFXW-REL SW-WET E FF ON N FIX 0.00 0.10 1.00 0.00 INL FNFXA(4), TYPFXA-AGE EFF ON N FIX !*VEGETATIVE PARTITIONING PARAMETERS 0.0 1.5 2.0 3. 0 5.0 7.0 30.0 40.0 XLEAF VALUES 0.60 0.60 0.60 0.40 0.25 0.20 0. 20 0.20 YLEAF VALUES

PAGE 179

165 0.10 0.10 0.10 0.10 0.05 0.05 0.05 0.05 YSTE M VALUES 0.55 0.00 0.05 0. 20 1.00 WT FSD,PORPT,FRSTMF,FRLFF,ATOP 0.60 FRLFMX !*LEAF GROWTH PARAMETERS 144. 285. 2 .0 0.0 00.0 FI NREF,SLAREF,SIZREF,VSSINK,EVMODC 350. 200. -.0 48 1.20 SLAMAX,SLAMIN ,SLAPAR,TURSLA 0.0 5.0 10.0 15.0 20.0 25.0 XVGROW(1-6), VSTAGE VALUES 0.0 10.0 20.0 30.0 4 0.0 50.0 YVREF(1-6), LEAF AREA VALUES,CM2 -50.0 00.0 10.0 30 .0 60.0 XSLATM(1 -5),TEMP VALUES 0.25 0.25 0.25 1.00 1.0 YSLAT M(1-5),EFFECT ON SLA !*LEAF SENESCENCE FACTORS 0.80 0.00 0.05 -25.0 -25.0 SENRTE,SENRT2,SENDAY,FREEZ1,FREEZ2 0.80 25.0 ICMP,TCMP(L ight comp, time constant-senes) .......XSTAGE......... .... ...XSENMX......... 0.0 5.0 9 .0 50.0 3.0 5. 0 10.0 50.0 .......SENPOR......... .... ...SENMAX......... 0.0 0.0 0.12 0. 12 0.0 0.2 0.6 0.6 !*ROOT PARAMETERS 20.0 5000. 0.010 0.1 0 .02 1.50 0.04 RTDEPI,RFAC1,RT SEN,RLDSM,RTSDF, RWUEP1,RWUMX 0.0 2.50 3.0 2. 50 6.0 2.50 30.0 2.50 XRTF AC,YRTFAC

PAGE 180

166 0.006 0.006 0.02 0.10 RTNO3,RTNH4,PORMIN,RTEXF !*SEED AND SHELL GROWTH PAR AMETERS 0.60 0.3 0. 00 100. SETMAX, SRMAX,RFLWAB,XMPAGE 15.0 0.0 0 .0 DSWBAR, XFRMAX,SHLAG 14.0 21.5 26.5 40 .0 QDR FNPDT(1-4),T YPPDT-TEMP EFFECT ON POD SET 6.0 21.0 23.5 41.0 QDR FNSDT(1 -4),TYPSDT-TEMP EFFECT ON SD GRWTH 0.00 10.00 20.00 26.00 32.00 60 .00 XXFTEM(1-6),TEMPERAT URES 1.00 1.00 1.00 1.00 1.00 1.00 YXFTEM(1-6),RE L CHG IN PARTIT 0.00 0.50 1.00 1. 00 XSWFAC(1-4) 0.00 1.00 1.00 1. 00 YSWFAC(1-4) 0.00 0.01 0.25 1.00 1.00 XSWBAR(1-5), REL WATER TOPSOIL 1.00 1.00 1.00 1.00 1.00 YSWBAR(15),EFFECT ON SEED ADDITION 0.00 0.50 0.75 1. 00 XTRFAC(1-4),TU RFAC 0.00 0.00 0.00 0.00 YTRFAC( 1-4),ENHANCE REPROD. GROWTH !*POD LOSS PARAMETERS N 6.0 .3961 -.865 1.00 0.00 DETACH,DWC ,PR1DET,PR2DET,XP1DET,XP2DET !*PHENOLOGY PARAMETERS TB TO1 TO2 TM I 9.0 32.0 40.0 45.0 1 VEGETATIVE DE VELOPMENT 10.0 28.0 32.0 45.0 2 EARLY REPRODUCTIVE DEVELOPM ENT 10.0 28.0 32.0 45.0 3 LATE RE PRODUCTIVE DEVELOPMENT

PAGE 181

167!FOLLOWING LINE: STAGE; REF ST AGE; PHOTOPERIOD FUNCTION; TEMPERATURE FUNCT; !POINTER TO VEGD (1) OR REPDA(2) OR REPDB(3) TEMP SE NS; SENS TO WATER;N; AN D P 1 1 NON LIN 1 -0.50 0.00 0.00 PLANT(STG 1) TO EMERG(STG 2) PHASE 2 2 NON LIN 1 -0.50 0. 00 0.00 EMERG(STG 2) TO V1(STG 3) PHAS E 3 2 NON LIN 1 -0.50 0. 00 0.00 EMERG(STG 2) TO END JV(STG 4) PHAS E 4 4 INL LIN 1 -0.40 0.00 0.00 END JV(STG 4) TO FL IND(STG 5) PHASE 5 5 INL LIN 1 -0.40 0. 00 0.00 FL IND(STG 5) TO 1ST FL(STG 6) PHASE 6 6 INL LIN 1 -0.50 0.00 0.00 1ST FL(STG 6) TO 1ST PEG(STG 7) PHASE 7 6 INL LIN 2 -0.50 0.00 0.00 1ST FL(STG 6) TO 1ST POD(STG 8) PHASE 8 6 INL LIN 2 -0.50 0. 00 0.00 1ST FL(STG 6) TO 1ST SD(STG 9) PHAS E 9 9 INL LIN 3 0.30 0.00 0.00 1ST SD(STG 9) TO LST SD(S TG 10) PHASE 10 9 INL LIN 3 0.30 0.00 0.00 1ST SD(STG 9) TO PH MAT(STG 11) PHASE 11 11 NON NON 1 0.00 0.00 0.00 PH MAT(STG 11) TO H-MAT( STG 12) PHASE 12 6 INL LIN 2 -0.70 0.00 0.00 1ST FL(STG 6) TO LST VST(STG 13) PHASE 13 6 INL LIN 2 -0.70 0.00 0.00 1ST FL(STG 6) TO LST LF(STG 14) PHASE !*CANOPY HEIGHT AND WI DTH GROWTH PARAMETERS VSTAGE, FOLLOWED BY INTERNODE LENGTH PER NODE, THEN CA NOPY WIDTH PER NODE 0.00 1.00 4.00 6. 00 8.00 10.00 14.00 16.00 20.00 40.00 XVSHT(1-10 ) .0150 .0265 .0315 .0330 .0345 .0330 .0310 .0255 .0170 .0030 YV SHT(1-10) .0150 .0255 .0310 .0320 .0330 .0315 .0295 .0230 .0125 .0005 YV SWH(1-10) -50.0 00.0 15.0 30 .0 60.0 XHWTEM(1-5),TE MPERATURES 0.55 0.55 0.55 1.00 1.00 YHWTEM(1-5), RELATIVE EXPAN

PAGE 182

168 0.00 5.00 7.50 10.00 15.00 20.00 30.00 80.00 XHWPAR(1-8),PAR VA LUES 4.00 2.00 1.50 1. 25 1.05 1.00 1.00 1.00 YHWPAR(1-8),RELATIV E EXPAN !*EVAPOTRANSPIRATION 0.70 1.0 KEP, EORATIO

PAGE 183

169 Cultivar File:

PAGE 184

170*BAHIA GENOTYPE COEFFICIENTS ! COEFF DEFINITIONS ===== =========== ECO# Code for the ecotype to whic h this cultivar belongs (see *.eco file) CSDL Critical Short Day Length below which reproductive development progresses with no daylen gth effect (for short day plants) (hour) PPSEN Slope of th e relative response of developm ent to photoperio d with time (positive for shortday plants) (1/hour) EM-FL Time between plant em ergence and flower appearance (R1) (photothermal days) FL-SH Time between first flower and first pod (R3) (photothermal days) FL-SD Time between f irst flower and first seed (R5) (photothermal days) SD-PM Time between first seed (R 5) and physiological maturity (R7) (photothermal days) FL-LF Time between first flower (R1) and end of leaf expansion (photothermal days) LFMAX Maximum leaf phot osynthesis rate at 30 C, 350 vpm CO2, and high light (mg CO2/m2-s) SLAVR Specific leaf area of cultivar under st andard growth conditions (cm2/g) SIZLF Maximum size of f ull leaf (three leaflets) (cm2) XFRT Maximum fracti on of daily growth that is part itioned to seed + shell WTPSD Maximum weight per seed (g) SFDUR Seed filling d uration for pod cohort at st andard growth conditions

PAGE 185

171! (photothermal days) SDPDV Average seed per pod unde r standard growing conditions (#/pod) PODUR Time required for cultivar to reach final pod load under optimal condit ions (photothermal days) @VAR# VRNAME.......... ECO# CSDL PP SEN EM-FL FL-SH FL-SD SD -PM FL-LF LFMAX SLAVR SIZLF XFRT WTPS D SFDUR SDPDV PODUR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 UF0001 PENSACOLA BAHIA G00001 12.00 0.200 99.0 10.0 18.0 33.00 25 .00 1.76 285. 2.0 0.01 0.020 15.0 2.05 20.0

PAGE 186

172 Ecotype File:

PAGE 187

173*BAHIA ECOTYPE COEFFICIENTS ! COEFF DEFINITIONS ===== =========== ECO# Code for the ecotype to which a cultivar belongs (see *.cul file) ECONAME Name of the ecotype, whic h is referenced fr om *.CUL file MG Maturity group number fo r this ecotype, such as maturity group in soybean TM Indicator of temperature adaptation THVAR Minimum rate of reprod uctive development under short days an d optimal temperature PL-EM Time between planting and emergence (V0) (thermal days) EM-V1 Time required from emergence to first true leaf (V1), thermal days V1-JU Time required from first true leaf to end of juvenile phase, thermal days JU-R0 Time required for floral induc tion, equal to the minimum number of days for floral induction un der optimal te mperature and daylen gths, photothermal days PM06 Prop ortion of time between fir st flower and fi rst pod for firs t peg (peanut only) PM09 Prop ortion of time between fir st seed and physiological maturity that the last seed can be formed LNGSH Time required for growth of individu al shells (photothermal

PAGE 188

174! days) R7-R8 Time between p hysiological (R7) and harv est maturity (R8) (days) FL-VS Time f rom first flower to last leaf on main stem (photothermal days) TRIFL Rate of appe arance of leaves on the mainstem (leaves per thermal day) default was 0.10, not getting en ough leaves, ch anged to 0.15 2/21/03 RWDTH Relative width of this ecotype in comp arison to the standard width per node (YVSWH) defined in the species file (*.SPE) RHGHT Relative heig ht of this ecotype in comparison to the standard he ight per node (YVSHT) defin ed in the species file (*.SPE) THRSH The ma ximum ratio of (seed/(se ed+shell)) at maturity. Causes seed to stop growing as their dry weights increase unti l shells are filled in a cohort. (T hreshing percentage). SDPRO Fraction protein in seeds (g(protein)/g(seed)) SDLIP Fraction oil in seeds (g(oil)/g(seed)) R1PPO Increase in daylength sensitivity afte r R1 (CSDVAR and CLDVAR bo th decrease with the same amount) (h) OPTBI Minimum daily temperature above which th ere is no effect on slowing normal de velopment toward flowering (oC) SLOBI Slope of rela tionship reducing progres s toward flowering if TMIN for the day is less than OPTBI

PAGE 189

175@ECO# ECONAME.......... MG TM THVAR PL-EM EM-V1 V1-JU JU-R0 PM06 PM09 LNGSH R7-R8 FLVS TRIFL RWDTH RHGHT THRSH SDP RO SDLIP R1PPO OPTBI SLOBI 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 G00001 PENSACOLA BAHIA 00 01 1.00 0.0 0.0 9999. 9999. 0. 0 0.75 10.0 9999. 9999. 0.15 1.0 1.0 78.0 .115 .035 .000 0.0 .000

PAGE 190

176 Optimized Species File:

PAGE 191

177*BAHIA SPECIES COEFICIENTS !*PHOTOSYNTHESIS PARAMET ERS 106.0 185.0 0.55 PAR MAX,PHTMAX,KCAN 5.0 1.40 .0105 CCMP,CCMAX,CC EFF; CO2 EFFECT ON PGCAN 1.00 3.0 10 .0 10.0 QDR FNPGN(4),TYPPGN-LF N EFFECT ON PG 23.0 25.0 25.5 50.0 LIN FNPGT(4),TYPPGT-TEMP EFFECT-CANOPY PG -5.0 7.0 35.0 45.0 55.0 60.0 XLMAXT (6 VALUES) 0.0 0.0 1.0 1.0 0.0 0.0 YLMAXT (6 VALUES) 7.0 18.0 45.0 57.0 QDR FNPGL (4),TYPPGL-TMIN EFFECT-LEAF PG .0650 0.20 0.80 2.0 PGEFF SCV KDIF, LFANGB 11/5/02 .0035 .0002 .2000 3.00 1.760 SLWRE F,SLWSLO,NSLOPE,LNREF,PGREF 0.0 .001 .002 .003 .0035 .004 .005 .006 .008 .010 XPGSLW(1-10) .162 .679 .8 67 .966 1.000 1.027 1.069 1.100 1.141 1.167 YPGSLW(1-10) !*RESPIRATION PARAMETERS 3.5E-04 .0029 RES30C,R30C2 2.556 2.556 .360 2.830 RNO3C,RNH4C, RPRO,RFIXN 1.242 3.106 2.174 .929 0.05 1.13 RCH20,RLIP,R LIG,ROA,RMIN,PCH2O !*PLANT COMPOSITION VA LUES .170 .080 .050 .090 .060 .033 PROLFI,PR OLFG,PROLFF,PROSTI,PROSTG,PROSTF .101 .040 .022 .039 .039 .038 PRORTI,PROR TG,PRORTF,PROSHI,PROSHG,PROSHF .115 .115 .300 .015 .019 .800 SDPROS,SDPRO G,PRONOD,PROMIN,PROMAX,THETA

PAGE 192

178 .652 .717 .702 .251 .750 .480 PCARLF,PCAR ST,PCARRT,PCARSH,PCARSD,PCARNO .025 .020 .020 .011 .050 PLIPLF,PL IPST,PLIPRT,PLIPSH,PLIPNO .040 .060 .070 .100 .030 .070 PLIGLF,PLIG ST,PLIGRT,PLIGSH,PLIGSD,PLIGNO .050 .050 .050 .0 40 .040 .050 POALF,POAST,P OART,POASH,POASD,POANO .063 .063 .057 .100 .030 .050 PMINLF,PMINS T,PMINRT,PMINSH,PMINSD,PMINNO !*SEED COMPOSITION VALUES 7.168 23.65 0.908 0.430 LIPTB,LIPOP T,SLOSUM*100,CARMIN !*CARBON AND NITROGEN MINING PARAMETERS 0.025 0.75 .280 0.050 1.00 0.15 CMOBMX,CADSTF ,CADPR1,NMOBMX,NVSMOB,NRCVR PD XPODF 0.04 0.08 0.04 0.08 ALPHL,ALPHS ,ALPHR,ALPHSH !*NITROGEN FIXATION PARA METERS .050 .160 0.01 0.0 0.04 0.05 SNACTM,NODRGM,DWNODI,TTFIX,NDTHMX, CNODCR 7.00 28.0 35.0 44 .0 LIN FNNGT(4),TY PNGT-TEMP EFF ON NOD GR OWTH 5.00 23.0 35.0 44.0 LIN FNFXT(4), TYPFXT-TEMP EFF ON N FIX -.15 0.20 1.00 10 .0 LIN FNFXD(4), TYPFXD-REL SW-DRY E FF ON N FIX -.02 .001 1.00 2. 00 LIN FNFXW(4), TYPFXW-REL SW-WET E FF ON N FIX 0.00 0.10 1.00 0.00 INL FNFXA(4),TYPFXA-AGE EFF ON N FIX !*VEGETATIVE PARTITIONING PARAMETERS 0.0 1.5 2.0 3. 0 5.0 7.0 30.0 40.0 XLEAF VALUES 0.60 0.60 0.60 0.40 0.25 0.20 0. 20 0.20 YLEAF VALUES

PAGE 193

179 0.10 0.10 0.10 0.10 0.05 0.05 0.05 0.05 YSTE M VALUES 0.55 0.00 0.05 0. 20 1.00 WT FSD,PORPT,FRSTMF,FRLFF,ATOP 0.60 FRLFMX !*LEAF GROWTH PARAMETERS 144. 285. 2 .0 0.0 00.0 FI NREF,SLAREF,SIZREF,VSSINK,EVMODC 350. 200. -.0 48 1.20 SLAMAX,SLAMIN ,SLAPAR,TURSLA 0.0 5.0 10.0 15.0 20.0 25.0 XVGROW(1-6), VSTAGE VALUES 0.0 10.0 20.0 30.0 4 0.0 50.0 YVREF(1-6), LEAF AREA VALUES,CM2 -50.0 00.0 10.0 30 .0 60.0 XSLATM(1 -5),TEMP VALUES 0.25 0.25 0.25 1.00 1.0 YSLAT M(1-5),EFFECT ON SLA !*LEAF SENESCENCE FACTORS 0.80 0.00 0.05 -25.0 -25.0 SENRTE,SENRT2,SENDAY,FREEZ1,FREEZ2 0.80 25.0 ICMP,TCMP(L ight comp, time constant-senes) .......XSTAGE......... .... ...XSENMX......... 0.0 5.0 9 .0 50.0 3.0 5. 0 10.0 50.0 .......SENPOR......... .... ...SENMAX......... 0.0 0.0 0.12 0. 12 0.0 0.2 0.6 0.6 !*ROOT PARAMETERS 20.0 5000. 0.010 0.1 0 .02 1.50 0.04 RTDEPI,RFAC1,RT SEN,RLDSM,RTSDF, RWUEP1,RWUMX 0.0 2.50 3.0 2.50 6.0 2.50 30.0 2.50 XRTF AC,YRTFAC

PAGE 194

180 0.006 0.006 0.02 0.10 RTNO3,RTNH4,PORMIN,RTEXF !*SEED AND SHELL GROWTH PARAMETERS 0.60 0.3 0. 00 100. SETMAX, SRMAX,RFLWAB,XMPAGE 15.0 0.0 0 .0 DSWBAR, XFRMAX,SHLAG 14.0 21.5 26.5 40 .0 QDR FNPDT(1-4),T YPPDT-TEMP EFFECT ON POD SET 6.0 21.0 23.5 41.0 QDR FNSDT(1 -4),TYPSDT-TEMP EFFECT ON SD GRWTH 0.00 10.00 20.00 26.00 32.00 60 .00 XXFTEM(1-6), TEMPERATURES 1.00 1.00 1.00 1.00 1.00 1.00 YXFTEM(1-6),RE L CHG IN PARTIT 0.00 0.50 1.00 1. 00 XSWFAC(1-4) 0.00 1.00 1.00 1. 00 YSWFAC(1-4) 0.00 0.01 0.25 1.00 1.00 XSWBAR(1-5), REL WATER TOPSOIL 1.00 1.00 1.00 1.00 1.00 YSWBAR(15),EFFECT ON SEED ADDITION 0.00 0.50 0.75 1. 00 XTRFAC(1-4),TURF AC 0.00 0.00 0.00 0.00 YTRFAC( 1-4),ENHANCE REPROD. GROWTH !*POD LOSS PARAMETERS N 6.0 .3961 -.865 1.00 0.00 DETACH,DWC ,PR1DET,PR2DET,XP1DET,XP2DET !*PHENOLOGY PARAMETERS TB TO1 TO2 TM I 9.0 32.0 40.0 45.0 1 VEGETATIVE DE VELOPMENT 10.0 28.0 32.0 45.0 2 EARLY REPRODUCTIVE DEVELOPM ENT 10.0 28.0 32.0 45.0 3 LATE RE PRODUCTIVE DEVELOPMENT

PAGE 195

181!FOLLOWING LINE: STAGE; REF STAGE; PHOTOPERIOD FUNCTION; TEMPERATURE FUNCT; !POINTER TO VE GD(1) OR REPDA(2) OR REPDB(3) TEMP SENS; SENS TO WATE R;N; AND P 1 1 NON LIN 1 -0.50 0. 00 0.00 PLANT(STG 1) TO EMERG(STG 2) PHASE 2 2 NON LIN 1 -0.50 0. 00 0.00 EMERG(STG 2) TO V1(STG 3) PHAS E 3 2 NON LIN 1 -0.50 0. 00 0.00 EMERG(STG 2) TO END JV(STG 4) PHAS E 4 4 INL LIN 1 -0.40 0.00 0.00 END JV(STG 4) TO FL IND(ST G 5) PHASE 5 5 INL LIN 1 -0.40 0. 00 0.00 FL IND(STG 5) TO 1ST FL(STG 6) PHASE 6 6 INL LIN 1 -0.50 0.00 0.00 1ST FL(STG 6) TO 1ST PEG(STG 7) PHASE 7 6 INL LIN 2 -0.50 0.00 0.00 1ST FL(STG 6) TO 1ST POD(STG 8) PHASE 8 6 INL LIN 2 -0.50 0. 00 0.00 1ST FL(STG 6) TO 1ST SD(STG 9) PHAS E 9 9 INL LIN 3 0.30 0.00 0.00 1ST SD(STG 9) TO LST SD(S TG 10) PHASE 10 9 INL LIN 3 0.30 0.00 0.00 1ST SD(STG 9) TO PH MAT(STG 11) PHASE 11 11 NON NON 1 0.00 0.00 0.00 PH MAT(STG 11) TO H-MAT( STG 12) PHASE 12 6 INL LIN 2 -0.70 0.00 0.00 1ST FL(STG 6) TO LST VST(STG 13) PHASE 13 6 INL LIN 2 -0.70 0.00 0.00 1ST FL(STG 6) TO LST LF(STG 14) PHASE !*CANOPY HEIGHT AND WI DTH GROWTH PARAMETERS VSTAGE, FOLLOWED BY INTERNODE LENGTH PER NODE, THEN CA NOPY WIDTH PER NODE 0.00 1.00 4.00 6. 00 8.00 10.00 14.00 16.00 20.00 40.00 XVSHT(1-10) .0150 .0265 .0315 .0330 .0345 .0330 .0310 .0255 .0170 .0030 YV SHT(1-10) .0150 .0255 .0310 .0320 .0330 .0315 .0295 .0230 .0125 .0005 YV SWH(1-10) -50.0 00.0 15.0 30 .0 60.0 XHWTEM(1-5),TE MPERATURES 0.55 0.55 0.55 1.00 1.00 YHWTEM(1-5),R ELATIVE EXPAN

PAGE 196

182 0.00 5.00 7.50 10.00 15.00 20.00 30.00 80.00 XHWPAR (1-8),PAR VALUES 4.00 2.00 1.50 1. 25 1.05 1.00 1.00 1.00 YHWPAR(1-8),RELATIV E EXPAN !*EVAPOTRANSPIRATION 0.70 1.0 KEP, EORATIO

PAGE 197

183 APPENDIX C SPECIES, CULTIVAR, AND ECOTYPE FILES FOR THE FORAGE VERSION OF CROPGRO Species file:

PAGE 198

184*BAHIA SPECIES COEFICIENTS !*PHOTOSYNTHESIS PARAMETERS 60.0 90.0 0.55 PARMAX, PHTMAX,KCAN 11.0 1.23 .0081 CCMP,CCMAX,CC EFF; CO2 EFFECT ON PGCAN 0.75 3.0 10 .0 10.0 QDR FNPGN(4),TYPPGN-LF N EFFECT ON PG 12.0 25.0 38.0 50.0 LIN FNPGT(4),TYPPGT-TEMP EFFECT-CANOPY PG -5.0 7.0 35.0 45.0 55.0 60.0 XLMAXT (6 VALUES) 0.0 0.0 1.0 1.0 0.0 0.0 YLMAXT (6 VALUES) 7.0 18.0 45.0 57.0 QDR FNPGL (4),TYPPGL-TMIN EFFECT-LEAF PG .0650 0.20 0.80 2.0 PGEFF SCV KDIF, LFANGB 11/5/02 .0035 .0002 .2000 3.00 1.760 SLWRE F,SLWSLO,NSLOPE,LNREF,PGREF 0.0 .001 .002 .003 .0035 .004 .005 .006 .008 .010 XPGSLW(1-10) .162 .679 .8 67 .966 1.000 1.027 1.069 1.10 0 1.141 1.167 YPGSLW(1-10) 0.4 3.0 5.650 4.660 C4 CI CA, CCNEFF, CMXSF, CQ ESF, PGPATH 0.7 10.0 4.272 4.191 C4 CICA, CCNEFF, CMXSF, CQESF, PGPATH !*RESPIRATION PARAMETERS 3.5E-05 .0029 RES30C,R30C2 2.556 2.556 .360 2.830 RNO3C,RNH4C, RPRO,RFIXN 1.242 3.106 2.174 .929 0.05 1.13 RCH20,RLIP,R LIG,ROA,RMIN,PCH2O !*PLANT COMPOSITION VA LUES

PAGE 199

185 .220 .110 .050 .110 .070 .033 PROLFI,PR OLFG,PROLFF,PROSTI, PROSTG,PROSTF .101 .040 .022 .039 .039 .038 PRORTI,PROR TG,PRORTF,PROSHI,PROSHG,PROSHF .115 .115 .300 .0 15 .019 .800 SDPROS,SDPRO G,PRONOD,PROMIN,PROMAX,THETA .602 .697 .702 .251 .750 .480 PCARLF,PCAR ST,PCARRT,PCARSH,PCARSD,PCARNO .025 .020 .020 .011 .050 PLIPLF,PL IPST,PLIPRT,PLIPSH,PLIPNO .040 .060 .070 .100 .030 .070 PLIGLF,PLIG ST,PLIGRT,PLIGSH,PLIGSD,PLIGNO .050 .050 .050 .0 40 .040 .050 POALF,POAST,P OART,POASH,POASD,POANO .063 .063 .057 .100 .030 .050 PMINLF,PMIN ST,PMINRT,PMINSH,PMINSD,PMINNO .092 .064 .056 PROSRI,PROSRG,PROSRF .711 .020 .0 70 .050 .057 PCARSR,PLIPSR,PLIGSR,POASR,PMINSR 0.05 KCOLD !*SEED COMPOSITION VALUES 7.168 23.65 0.908 0.430 LIPTB,LIPOPT ,SLOSUM*100,CARMIN !*CARBON AND NITROGEN MINING PARAMETERS 0.025 0.75 .280 0.050 1.00 0.15 CMOBMX,CADSTF ,CADPR1,NMOBMX,NVSMOB,NRCVR PD XPODF 0.04 0.08 0.04 0.08 0.20 A LPHL,ALPHS,ALPHR,ALPHSH, ALPHSR 0.01 0.05 0.65 0. 01 0.08 CMOBSRN, CMOBSRX, CADSRF, NMOBSRN, NMOBSRX 0.00 0.00 1.00 3.00 SIN LRMOB(4), TYPLMOB LAI EFF ON MOBILIZATION 0.30 0.70 10.00 10.0 SIN NRMOB(4), TYPNMO B N-STATUS EFF ON MOBILIZATION !*NITROGEN FIXATION PARA METERS .050 .160 0.01 0.0 0.04 0.05 SNACTM,NODRGM,DWNODI,TTFIX,NDTHMX, CNODCR

PAGE 200

186 7.00 28.0 35.0 44 .0 LIN FNNGT(4),TY PNGT-TEMP EFF ON NOD GR OWTH 5.00 23.0 35.0 44.0 LIN FNFXT(4), TYPFXT-TEMP EFF ON N FIX -.15 0.20 1.00 10 .0 LIN FNFXD(4), TYPFXD-REL SW-DRY E FF ON N FIX -.02 .001 1.00 2. 00 LIN FNFXW(4), TYPFXW-REL SW-WET E FF ON N FIX 0.00 0.10 1.00 0.00 INL FNFXA(4), TYPFXA-AGE EFF ON N FIX !*VEGETATIVE PARTITIONING PARAMETERS 0.0 1.5 2.0 3. 0 5.0 7.0 30.0 40.0 XLEAF VALUES 0.60 0.60 0.60 0.40 0.25 0.20 0. 20 0.20 YLEAF VALUES 0.10 0.10 0.10 0.10 0.05 0.05 0.05 0.05 YSTE M VALUES 0.55 0.00 0.05 0. 20 1.00 0.00 WTFSD,POR PT,FRSTMF,FRLFF ,ATOP,FRCNOD 0.60 FRLFMX 0.10 0.10 0.20 0.30 0.35 0.40 0. 45 0.45 YSTOR VALUES 0.10 0.90 FRSTRF, FRSTRMX 2.00 1.50 1.00 1. 00 PWLF, PWST, PWRT, PWSR !*LEAF GROWTH PARAMETERS 144. 285. 2 .0 0.0 00.0 FI NREF,SLAREF,SIZREF,VSSINK,EVMODC 350. 200. -.0 48 1.20 SLAMAX,SLAMIN,SLAPAR,TURSLA 0.0 5.0 10.0 15.0 20.0 25.0 XVGROW(1-6), VSTAGE VALUES 0.0 10.0 20.0 30.0 4 0.0 50.0 YVREF(1-6), LEAF AREA VALUES,CM2 -50.0 00.0 10.0 30 .0 60.0 XSLATM(1 -5),TEMP VALUES 0.25 0.25 0.25 1.00 1.0 YSLAT M(1-5),EFFECT ON SLA

PAGE 201

187!*LEAF SENESCENCE FACTORS 0.80 0.20 0.06 -5 .0 -18.0 SENRTE,SENRT2,SENDAY,FREEZ1,FREEZ 2 0.80 25.0 ICMP,TCMP(L ight comp, time constant-senes) 0.01 35.0 ICMP,TC MP(Light comp, time constant-senes) .......XSTAGE......... .... ...XSENMX......... 0.0 5.0 9 .0 50.0 3.0 5. 0 10.0 50.0 .......SENPOR......... .... ...SENMAX......... 0.0 0.0 0.12 0. 12 0.0 0.2 0.6 0.6 !*ROOT PARAMETERS 20.0 7500. 0.020 0.1 0.02 1.50 0.04 RT DEPI,RFAC1,RTSEN,RLDSM, RTSDF,RWUEP1,RWUMX 0.0 2.50 3.0 2.50 6.0 2.50 30.0 2.50 XRTF AC,YRTFAC 0.006 0.006 0.02 0.10 RTNO3,RTNH4,PORMIN,RTEXF !*SEED AND SHELL GROWTH PAR AMETERS 0.60 0.3 0. 00 100. SETMAX, SRMAX,RFLWAB,XMPAGE 15.0 0.0 0 .0 DSWBAR, XFRMAX,SHLAG 14.0 21.5 26.5 40 .0 QDR FNPDT(1-4),TYPPDT-TEMP EFFECT ON POD SET 6.0 21.0 23.5 41.0 QDR FNSDT(1 -4),TYPSDT-TEMP EFFECT ON SD GRWTH 0.00 10.00 20.00 26.00 32.00 60 .00 XXFTEM(1-6),TEMPERAT URES 1.00 1.00 1.00 1.00 1.00 1.00 YXFTEM(1-6),RE L CHG IN PARTIT 0.00 0.50 1.00 1. 00 XSWFAC(1-4) 0.00 1.00 1.00 1.00 Y SWFAC(1-4) 0.00 0.01 0.25 1.00 1.00 XSWBAR(1-5), REL WATER TOPSOIL

PAGE 202

188 1.00 1.00 1.00 1.00 1.00 YSWBAR(15),EFFECT ON SEED ADDITION 0.00 0.50 0.75 1. 00 XTRFAC(1-4),TURF AC 0.00 0.00 0.00 0.00 YTRFAC( 1-4),ENHANCE REPROD. GROWTH !*POD LOSS PARAMETERS N 6.0 .3961 -.865 1.00 0.00 DETACH,DWC ,PR1DET,PR2DET,XP1DET,XP2DET !*PHENOLOGY PARAMETERS TB TO1 TO2 TM I 9.0 32.0 40.0 45.0 1 VEGETATIVE DE VELOPMENT 10.0 28.0 32.0 45.0 2 EARLY REPRODUCTIVE DEVELOPM ENT 10.0 28.0 32.0 45.0 3 LATE RE PRODUCTIVE DEVELOPMENT !FOLLOWING LINE: STAGE; REF ST AGE; PHOTOPERIOD FUNCTION; TEMPERATURE FUNCT; !POINTER TO VEGD (1) OR REPDA(2) OR REPDB(3) TEMP SE NS; SENS TO WATER;N; AN D P 1 1 NON LIN 1 -0.50 0. 00 0.00 PLANT(STG 1) TO EMERG(STG 2) PHASE 2 2 NON LIN 1 -0.50 0. 00 0.00 EMERG(STG 2) TO V1(STG 3) PHAS E 3 2 NON LIN 1 -0.50 0. 00 0.00 EMERG(STG 2) TO END JV(STG 4) PHAS E 4 4 INL LIN 1 -0.40 0.00 0.00 END JV(STG 4) TO FL IND(STG 5) PHASE 5 5 INL LIN 1 -0.40 0. 00 0.00 FL IND(STG 5) TO 1ST FL(STG 6) PHASE 6 6 INL LIN 1 -0.50 0.00 0.00 1ST FL(STG 6) TO 1ST PEG(STG 7) PHASE 7 6 INL LIN 2 -0.50 0.00 0.00 1ST FL(STG 6) TO 1ST POD(STG 8) PHASE 8 6 INL LIN 2 -0.50 0. 00 0.00 1ST FL(STG 6) TO 1ST SD(STG 9) PHAS E 9 9 INL LIN 3 0.30 0.00 0.00 1ST SD(STG 9) TO LST SD(S TG 10) PHASE 10 9 INL LIN 3 0.30 0.00 0.00 1ST SD(STG 9) TO PH MAT(STG 11) PHASE

PAGE 203

189 11 11 NON NON 1 0.00 0.00 0.00 PH MAT(STG 11) TO H-MAT( STG 12) PHASE 12 6 INL LIN 2 -0.70 0.00 0.00 1ST FL(STG 6) TO LST VST(STG 13) PHASE 13 6 INL LIN 2 -0.70 0.00 0.00 1ST FL(STG 6) TO LST LF(STG 14) PHASE !*CANOPY HEIGHT AND WI DTH GROWTH PARAMETERS VSTAGE, FOLLOWED BY INTERNODE LENGTH PER NODE, THEN CA NOPY WIDTH PER NODE 0.00 1.00 4.00 6. 00 8.00 10.00 14.00 16.00 20.00 40.00 XVSHT(1-10 ) .0150 .0265 .0315 .0330 .0345 .0330 .0310 .0255 .0170 .0030 YV SHT(1-10) .0150 .0255 .0310 .0320 .0330 .0315 .0295 .0230 .0125 .0005 YV SWH(1-10) -50.0 00.0 15.0 30 .0 60.0 XHWTEM(1-5),TE MPERATURES 0.55 0.55 0.55 1.00 1.00 YHWTEM(1-5), RELATIVE EXPAN 0.00 5.00 7.50 10.00 15.00 20.00 30.00 80.00 XHWPAR (1-8),PAR VALUES 4.00 2.00 1.50 1. 25 1.05 1.00 1.00 1.00 YHWPAR(1-8),RELATIV E EXPAN !*EVAPOTRANSPIRATION 0.55 1.0 KEP, EORATIO !*STORAGE ORGA N PARAMETERS 1.000 0.000 STRSRFL, STRLYR1STOR on soil surface and in soil layer 1 0.015 SENSR rate fo r senescence of st orage organ tissue !*DORMANCY PARAMETERS DAYLENGTH TO INDUCE/DEEPEN /END DORMANCY 0.00 10.5 12.5 0.000 INL FNPTD(4),TYPP TD-DAYLENGTH EFFECT-PARTITIONING

PAGE 204

190 10.5 12.5 0.10 1.0 00 DRD FNPMD(4),TYPPM D-DAYLENGTH EFFECT ON MOBILIZATION 0.00 0.00 0.00 1.000 DRD FMPGD(4),TYP PGD-DAYLENGTH EFFECT ON PG -18.0 -25.0 0.050 HARD1, HARD2, FRZDC Freezing temperatures and death rate -25.0 -25.0 -25.0 0.16 REV FRZHRD(4 ),TYPHRD-COLD HA RDENING RESPONSE -25.0 -25.0 -25.0 0. 80 DHD FRZDHD(4),TY PDHD-COLD DEHARD ENING RESPONSE

PAGE 205

191 Cultivar File:

PAGE 206

192*BAHIA GENOTYPE COEFFICIENTS ! COEFF DEFINITIONS ===== =========== ECO# Code for the e cotype to which this cultivar belongs (see *.eco file) CSDL Critical Short Day Length below which reproductive development progresses with no daylen gth effect (for short day plants) (hour) PPSEN Slope of th e relative response of develop ment to photoperiod with time (positive for shortday plants) (1/hour) EM-FL Time between plant em ergence and flower appearance (R1) (photothermal days) FL-SH Time between first flower and first pod (R3) (photothermal days) FL-SD Time between f irst flower and first seed (R5) (photothermal days) SD-PM Time between first seed (R 5) and physiological maturity (R7) (photothermal days) FL-LF Time between first flower (R1) and end of leaf expansion (photothermal days) LFMAX Maximum leaf p hotosynthesis rate at 30 C, 35 0 vpm CO2, an d high light (mg CO2/m2-s) SLAVR Specific leaf area of cultivar under st andard growth conditions (cm2/g) SIZLF Maximum size of f ull leaf (three leaflets) (cm2) XFRT Maximum fracti on of daily growth that is part itioned to seed + shell WTPSD Maximum weight per seed (g) SFDUR Seed filling d uration for pod cohort at st andard growth conditions

PAGE 207

193! (photothermal days) SDPDV Average seed per pod under standard gr owing conditions (#/pod) PODUR Time required for cultivar to reach final pod load under optimal condit ions (photothermal days) @VAR# VRNAME.......... ECO# CSDL PP SEN EM-FL FL-SH FL-SD SD -PM FL-LF LFMAX SLAVR SIZLF XFRT WTPS D SFDUR SDPDV PODUR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 UF0001 PENSACOLA BAHIA G00001 12.00 0.200 99.0 10.0 18.0 33.00 25 .00 1.76 285. 2.0 0.01 0.020 15.0 2.05 20.0

PAGE 208

194 Ecotype File:

PAGE 209

195*BAHIA ECOTYPE COEFFICIENTS ! COEFF DEFINITIONS ===== =========== ECO# Code for the ecotype to which a cultivar belongs (see *.cul file) ECONAME Name of the ecotype, whic h is referenced fr om *.CUL file MG Maturity group number fo r this ecotype, such as maturity group in soybean TM Indicator of temperature adaptation THVAR Minimum rate of reprod uctive development under short days an d optimal temperature PL-EM Time between planting and emergence (V0) (thermal days) EM-V1 Time required from emergence to first true leaf (V1), thermal days V1-JU Time required from first true leaf to end of juvenile phase, thermal days JU-R0 Time required for fl oral induction, equa l to the minimum number of days for floral induction un der optimal te mperature and daylen gths, photothermal days PM06 Prop ortion of time between fir st flower and fi rst pod for firs t peg (pea nut only) PM09 Prop ortion of time between fir st seed and physiological maturity that the last seed can be formed LNGSH Time required for growth of individu al shells (photothermal

PAGE 210

196! days) R7-R8 Time between p hysiological (R7) and harve st maturity (R8) (days) FL-VS Time f rom first flower to last le af on main stem (photothermal days) TRIFL Rate of appe arance of leaves on the mainstem (leaves per thermal day) default was 0.10, not getting en ough leaves, ch anged to 0.15 2/21/03 RWDTH Relative width of this ecotype in comp arison to the standard width per node (YVSWH) defined in the species file (*.SPE) RHGHT Relative heig ht of this ecotype in comparison to the standard he ight per node (YVSHT) defin ed in the species file (*.SPE) THRSH The ma ximum ratio of (seed/(se ed+shell)) at maturity. Causes seed to stop growing as their dry weights increase unti l shells are filled in a cohort. (Thr eshing percentage). SDPRO Fraction protein in seeds (g(prot ein)/g(seed)) SDLIP Fraction oil in seeds (g(oil)/g(seed)) R1PPO Increase in daylength sensitivity afte r R1 (CSDVAR and CLDVAR bo th decrease with the same amount) (h) OPTBI Minimum dail y temperature above which there is no effect on slowing normal de velopment toward flowering (oC) SLOBI Slope of rela tionship reducing progres s toward flowering if TMIN for the day is less than OPTBI RDRMT Relative dorm ancy sensitivity of this cultivar to daylength part itioning (0-1)

PAGE 211

197! RDRMG Relative dormancy sensi tivity of this cultivar to da ylength photosynthesis (01) RDRMM Relative dormancy sensi tivity of this cultivar to da ylength mobilization (0-1) RCHDP Relative cold hardening potential (0-1) @ECO# ECONAME.......... MG TM THVAR PL-EM EM-V1 V1-JU JU-R0 PM06 PM09 LNGSH R7-R8 FLVS TRIFL RWDTH RHGHT THRSH SDPRO SDLIP R1PPO OPTBI SLOB I RDRMT RDRMG RDRMM RCHDP 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 G00001 PENSACOLA BAHIA 00 01 1.00 0.0 0.0 0.0 9 999. 0.0 0.75 10.0 9999. 9999. 0.15 1.0 1.0 78.0 .115 .035 .000 0.0 .000 1.000 1.000 1.000 1.000

PAGE 212

198 APPENDIX D NEW PARAMETER CODE DEFINITIONS FOR THE FORAGE VERSION OF CROPGRO Variable definitions listed in order of th eir appearance in the species file, followed by those in the ecotype file. CICA – The ratio of intercellular:atmospheric CO2 concentration used to calculate CO2MAX (0.7 for C3 species, suggest 0.4 for C4 species). CCNEFF – CO2-concentrating factor, used for C4 species to determine photosynthetic response to CO2. CMXSF – Factor to scale CO2MAX to 1.0 at 30C and 350 L CO2 L-1 (varies with CICA and CCNEFF). CQESF – Factor to scale CO2Q E to 1.0 at 30C and 350 L CO2 L-1 (varies with CCNEFF). PGPATH – Species’ photosynt hetic pathway. Currently recognizes two codes “C3” and “C4”. If code is “C3” then model uses unmodified CSM code to predict daily canopy and leaf-level photosynthesis. PROSRI – Maximum protein co mposition in storage organ dur ing growth with luxurious supply of N (g[protein] g[storage]-1). PROSRG – Normal growth protein composition in stor age organ during growth (g[protein] g[storage]-1). PROSRF Minimum storage organ protein composition af ter N mining (g[protein] g[storage]-1). PCARSR Carbohydrate-cellulo se concentration of st orage tissue (fraction). PLIPSR Proportion of storage tis sue that is lipid (fraction). PLIGSR Proportion of storage tissu e that is lignin (fraction). POASR Proportion of storage tissue th at is organic acid (fraction). PMINSR – Proportion of storage tiss ue that is mineral (fraction). KCOLD – Curvature factor (K value) for e xponential function limiting refilling of N to older tissues (NDMOLD) when photosynthesis is low. Part of new functionality to prevent expending all assimilate on N uptake and leaving none for new growth. CMOBSRN – “Normal” / minimum fraction of CH2O which can be mobilized from storage organ in a day.

PAGE 213

199 CMOBSRX – Maximum fraction of CH2O which can be mobilized from storage organ in a day. CADSRF – Proportion of CH2O reserves that are added to storage organ (fraction). NMOBSRN – "Normal" / minimum fraction of N which can be mobilized from storage organ in a day NMOBSRX – Maximum fraction of N which can be mobilized from storage organ in a day. LRMOB(1-4)– Critical LAI for functi on to increase mobilization of CH2O and N from STOR due to low leaf area (4 values fo r function CURV). Th is is a two-sided generic curve using only the two rightmost points to describe mobilization response to LAI. LRMOB(3) is the (low ) LAI where mobilization from STOR is maximum. LRMOB(4) is th e (high) LAI where there is no increased mobilization of CH2O or N from STOR. TYPLMOB – Character variab le specifying the type of function to use for the relationship between LAI and mobilization of CH2O and N from STOR (for use in function subroutine CURV). SIN defi nes a sinusoidal curve that decreases from 1.0 at an LAI of LRMOB(3) to its minimum of 0.0 at an LAI of LRMOB(4). NRMOB(1-4) Critical vege tative N-status for function to increase mobilization of CH2O and N from STOR due to high N-stat us (4 values for function CURV). This is a two-sided generi c curve using only the two le ftmost points to describe mobilization response to N-status. NRMOB( 1) is the lowest N-status where there is no increase in mobilization from STOR. LRMOB(2) is the (high) N-status where mobilization of CH2O or N from STOR is increased to maximum. TYPNMOB – Character variable specifying the type of function to use for the relationship between vegetative N-status and mobilization of CH2O and N from STOR (for use in function subroutine CU RV). SIN defines a sinusoidal curve that increases from 0.0 at an N-status of NRMOB(1) to its maximum of 1.0 at an N-status of NRMOB(2). YSTOR(1-8) Partitioning fr action to storage tissue gr owth at V-stage XLEAF(I) (g[STOR] g[vegetation. plant]-1). FRSTRF – Fraction of daily dr y weight increase in vegeta tive plant parts which goes to storage organ after the day on which the maximum number of V-stages occurs (NDVSTG). (g[storage] g[vegetation]-1). FRSTRMX – Maximum storage organ pa rtitioning (g[storage] g[vegetation]-1). Occurs during dormancy. PWLF – Weighting factors for partitioni ng N when refilling old leaf tissues. PWST – Weighting factors for partitioni ng N when refilling old stem tissues. PWRT – Weighting factors for partitioning N when re filling old root tissues. PWSR Weighting factors for partitioni ng N when refilling old STOR tissues. STRSRFL – Proportion of storage organ mass occurring on/above the soil surface.

PAGE 214

200 STRLYR1 Proportions of storage organ mass occurring in soil layer 1. SENSR Constant for senescence of stor age organ tissue (proportion of cumulative STOR mass lost / physiological day). FNPTD(1-4) – Parameters defining the daylen gth effect on partitioning of new growth to STOR. FNPTD(1) is ignored (set value to 0.0). FNPTD(2) is the daylength threshold (h) when the dormancy effect is maximum. FNPTD(3) is the daylength threshold (h) that induces (starts) dorma ncy. FNPTD(4) is the fraction increase (from FRSTR to FRSTRMX) in partiti oning to STOR when the crop is nondormant always set to 0.0 as we wa nt no dormancy effect when crop is not dormant. TYPPTD Character variable sp ecifying the type of function to use for the relationship between dormancy and partitioning of new growth to STOR (for use in function subroutine CURV). INL or SHO used fo r species that are dormant during short daylengths. This is an inverse linear response. The fraction increase (from FRSTR to FRSTRMX) in partitioning to STOR is lowered from 1.0 (100% of FRSTRMX) at a daylength for maximum dormancy of FNPTD(2) to its minimum or non-dormant, fraction of FNPTD(4) (0% increase over FRSTR) at a daylength of FNPTD(3). FNPMD(1-4) Parameters defining the daylength effect on mobilization of CH2O and N from STOR:. FNPMD(1) is the dayle ngth (h) when the dormancy effect is maximum, FNPMD(2) is the daylength (h) when dormancy ceases to affect mobilization, FNPMD(3) is the (low) proportion of the LAIand Nstatus-adjusted fracti on of mobilizable CH2O and N that can potentially be mobilized from STOR at peak dormancy, and FNPMD(4)=1.0 or the proportion (100%) of the LAIand N-status-adj usted fraction of mobilizable CH2O and N that can potentially be mobilized from STOR when non-dormant. TYPPMD Character variable sp ecifying the type of function to use for the relationship between dormancy and mobilization of CH2O and N from STOR (for use in function subroutine CURV). DRD is a new CURV function that is similar to LIN except that response values (proportion of the LAIand N-status-adjusted fraction of mobilizable CH2O and N that can potentially be mobilized from STOR) vary from FNPMD(3) to FNPMD(4) instead of 0 to 1. This is a linear response that increases from FNPMD(3) at a daylengt h of FNPMD(1) (maximum dormancy) to its maximum of FNPMD(4) at a daylength of FNPMD(2) (non-dormant). This is a new CURV function added for th e forage version of CROPGRO. FMPGD(1-4) Parameters defining the daylen gth effect on photosynthe sis currently not in use. TYPPGD Character variable sp ecifying the type of function to use for the relationship between dormancy and photosynthesis (for use in function subroutine CURV) currently not in use. HARD1 – Freezing temperature that will kill storage organ if no cold-hardening has occurred. Should be the same as FREEZ2. If TMIN gets to this temperature, all STOR is killed and the simulation is terminated.

PAGE 215

201 HARD2 Freezing temperature that will ki ll storage organ if cold-hardening is maximum. Should be the lo wer than FREEZ2 if species exhibits cold-hardening, or equal to FREEZ2 if it doesn’t. If TMIN gets to this temperature, all STOR is killed and the simulation is terminated. FRZDC Freezing leaf and stem death constant. Fraction of leaf and stem mass that will be killed or lost in a day for ever y C that TMIN drops below the FREEZ1 threshold. FRZHRD(1-4) Cold hardening response parame ters, not used for bahiagrass. Reduces FREEZ2 from HARD1 to HARD2 with increasing exposure to cold (but nonlethal) temperatures. The function is reversible as warm temperatures will reduce the amount of cold-hardening. Prolonged periods of warm temperatures will increase FREEZ2 to HARD1 (if hard ening has already occurred). TYPHRD Character variable specifying the type of functi on to use for the relationship between TMIN and cold-hardening (lin ear progression of FREEZ2 from HARD1 to HARD2) for use in subroutine CURV. REV is used for species that exhibit reversible cold-hardening. Increases th e rate of progression of FREEZ2 from HARD1 to HARD2 [C drop in FREEZ2 per degree TMIN drops below FRZHD(2)] linearly as TMIN drops fr om FRZHD(2) to FRZHD(1), with a maximum rate of FRZHD(4) when TMIN=F RZHRD(1). If TMIN is greater than FRZHD(2), cold-hardening is reversed (FREEZ2 increases linearly towards HARD1), reaching a rate of [–FRZHD(4) ] when TMIN=FRZHD(3). This is a new CURV function added for the forage version of CROPGRO. FRZDHD(1-4), Non-reversible cold-dehardening response parameters, not used for bahiagrass. Increases FREEZ2 toward s HARD1. Differs from the reverse cold-hardening simulated by FRZHD(1-4) as dehardening can only occur when daylength is increasing and de hardening is not reversible (i.e. cold temperatures will not reduce the amount of dehardening). TYPDHD Character variable sp ecifying the type of function to use for the relationship between TMIN and cold-dehardening (for use in function subroutine CURV). DHD is a new CURV function, essentially a positive linear res ponse from 0.0 to FRZDHD(4) as TMIN increases from FRZDHD(1) to FRZDHD(2). Allows dehardening only when dayle ngth is increasing. Increa ses the rate of progression of FREEZ2 from current value to HA RD1 linearly as TMIN increases from FRZDHD(1) to FRZDHD(2), with a maxi mum rate of FRZDHD(4) [C increase in FREEZ2 per degree TMIN increases above FRZHD(1)]. FRZDHD(3) is ignored. Dehardening cannot raise FR EEZ2 above HARD1. This is a new CURV function added for the forage version of CROPGRO. RDRMT Relative dormancy sensitivity of ecoty pe to daylength for partitioning (0–1). If ecotype exhibits less shift in partiti oning to STOR during short daylengths than the standard species ecotype, RDRMT<1.0. If ecotype exhibits a greater shift in partitioning during short daylengths than the standard species ecotype, RDRMT>1.0.

PAGE 216

202 RDRMG Relative dormancy sensitivity of ecotype to daylength for photosynthesis reduction (0–1). If ecotype exhibits le ss photosynthesis reduction during short daylengths than the standard species ecot ype, RDRMG<1.0. If ecotype exhibits a greater reduction in photosynthesis during short daylengths than the standard species ecotype, RDRMG>1.0. RDRMM Relative dormancy sensitivity of ecotype to daylength for decreased mobilization of CH2O and N from STOR (0–1). If ecotype exhibits less reduction in mobilization of CH2O and N from STOR during s hort daylengths than the standard species ecotype, RDRMM<1.0. If ecotype exhibits a greater reduction in mobilization of CH2O and N during short daylengths than the standard species ecotype, RDRMM>1.0. RCHDP Relative cold hardening potential ( 0–1) of ecotype. If eco type exhibits less cold-hardening than the standard spec ies ecotype, RCHDP<1.0. If ecotype exhibits more cold hardening than th e standard species ecotype, RCHDP>1.0.

PAGE 217

203 APPENDIX E DORMANCY AND STOR CODES AND DE FINITIONS FOR DATA.CDE FILE These lines should be added to the bottom of the DATA.CDE file to allow the GBUILD program to display the definitions of th e new dormancy and STOR output codes along with the variable name. *DORMANCY @CDE LABEL DESCRIPTION OTHER CODE(S) SO SE C%SR %OF C IN STORE % of CH2O reserves added to storage organ O FRZ2 FREEZE KILL TMP Current freeze kill temperature O FSRX MAX STORE PART Max partitioning to storage organ O FZDC FREEZE DEATH% Feezing death coefficient / degree below FREEZ2 O HRD1 SOFT FREEZE Killing freeze temperature before cold hardening OO HRD2 HARD FREEZE Killing freeze temperature after cold hardening O LAIT MOBIL THRESHOLD LAI trigger increased mobil from str organ OO PPGF DORM PG FACTOR Proportion reduction in PG due to dormancy O PPMF DORM MOBIL FAC Proportion reduction in mobilization due to dormancy O PPTF DORM PART FAC Proportion reduction in partitioning due to dormancy O QDSD DORM STATUS Text variable listing dormancy status O RDMG REL DORM PG Relative dormancy effect on PG of cultivar O RDMM REL DORM MOBIL Relative dormancy effect on mobilization of cultivar OO RDMT REL DORM PART Relative dormancy effect on partitioning of cultivar O SC%M MAX STOR C MOBL Maximum % of storage organ C to mobilize OO SNSR STOR SENES % Senescence rate of storage organ /physiological day OO SP%1 STOR PART FAC1 YSTOR1 Storage partitioning factor for XLEAF(1) OO SP%2 STOR PART FAC2 YSTOR2 Storage partitioning factor for XLEAF(2) OO SP%3 STOR PART FAC3 YSTOR3 Storage partitioning factor for XLEAF(3) OO SP%4 STOR PART FAC4 YSTOR4 Storage partitioning factor for XLEAF(4) OO SP%5 STOR PART FAC5 YSTOR5 Storage partitioning factor for XLEAF(5) OO SP%6 STOR PART FAC6 YSTOR6 Storage partitioning factor for XLEAF(6) O SP%7 STOR PART FAC7 YSTOR7 Storage partitioning factor for XLEAF(7) OO SP%8 STOR PART FAC8 YSTOR8 Storage partitioning factor for XLEAF(8) OO SP%F STOR FINAL PROT Minimum storage organ protein % after N mining OO SP%G STOR GROW PROT% Normal growth protein % of storage organ OO SP%I STOR INIT PROT% Protein % of new storage organ growth OO SR%C %STORE MOBIL C % CH2O in New storage organ growth O SRC% STOR %CH2O Proportion of storage organ that is CH2O O SRG% STOR %LIGNIN Proportion of storage organ that is lignin O SRG1 DORM PG DAYL1 Base daylength for dormancy effect on PG O SRG2 DORM PG DAYL2 1st optimum daylength for dormancy effect on PG O SRG3 DORM PG DAYL3 2nd optimum daylength for dormancy effect on PG OO SRG4 DORM PG DAYL4 Max daylength for dormancy effect on PG OO SRL% STOR %LIPID Proportion of storage organ that is lipid O SRL1 STOR LAYER1 Initial/normal % of storage organ below soil surface O SRM% STOR %MINERAL Proportion of storage organ that is mineral OO SRM1 DORM MOB DAYL1 Base daylength for dormancy effect on mobilization OO SRM2 DORM MOB DAYL2 1st optimum daylength for dormancy effect on mobil OO SRM3 DORM MOB DAYL3 2nd optimum daylength for dormancy effect on mobil OO SRM4 DORM MOB DAYL4 Max daylength for dormancy effect on mobilization OO

PAGE 218

204 SRO% STOR %ORG ACID Proportion of storage organ that is organic acid OO SRSF STOR SURFACE Initial/normal % of storage organ above soil surface OO SRT1 DORM PART DAYL1 Base daylength for dormancy effect on partitioning OO SRT2 DORM PART DAYL2 1st optimum daylength for dormancy effect on part OO SRT3 DORM PART DAYL3 2nd optimum daylength for dormancy effect on part OO SRT4 DORM PART DAYL4 Max daylength for dormancy effect on partitioning OO TDHD DEHARD CURV Curve/response type for dehardening OO THRD HARDEN CURV Curve/response type for cold hardening O TSDM DRM MOBIL CURV Curve/response type OF mobilization to dormancy O TSPG DORM PG CURV Curve/response type OF PG to dormancy O TSPT DRM PART CURV Curve/response type OF partitioning to dormancy O TSRD S-TMP SOIL SURF Soil temperature on soil surface (oC) OO ZDH1 DEHARD TEMP1 Dehardening base temperature OO ZDH2 DEHARD TEMP2 Dehardening temperature 1 O ZDH3 DEHARD TEMP3 Dehardening temperature 2 O ZDH4 DEHARD RATE Relative rate of dehardening O ZHD1 HARDEN TEMP1 Cold hardening base temperature (max hardening) O ZHD2 HARDEN TEMP1 Cold hardening temperature 1 (no hardening) O ZHD3 HARDEN TEMP1 Cold hardening temperature 2 (reverse hardening) O ZHD4 HARDEN RATE Relative rate of cold hardening O *STORAGE @CDE LABEL DESCRIPTION OTHER CODE(S) SO SE Q1%D STOR % SOIL Proportion of storage organ mass below soil surface OO QC%D STOR % CH2O Fraction of storage organ tissue that is CH2O OO QC%M STOR C MINE P Potential C mining rate from storage organ OO QCAD STOR CUM GROW Cumulative storage organ growth OO QCAG STOR NET C ADD Net C addition to storage organ O QCDAM STOR DAMAGE Calculated storage organ damage O QCDD STOR CUM DAM Cumulative storage organ damage O QCFD STOR CUM FRZDM Cumulative frozen storage organ tissue OO QCQD STOR CH2O REQ Mass CH2O required for new storage organ growth OO QCRD STOR CH2O RESRV Mass of CH2O reserves in storage organ O QDAD STOR CUM LOSS Cumulative storage organ losses O QDTD STOR DESIRE WT Desired storage organ mass O QEAD STOR SENES Daily senescence of storage organ tissue OO QEWD STOR SENES H2O Daily senescence of storage organ due to H2O stress OO QFAD STOR FRZ LOSS Storage organ weight loss to freezing today O QFD1 STR FRZDAM SOIL Freeze damage to below ground storage organ mass O QFDS STR FRZDAM SURF Freeze damage to above ground storage organ mass OO QHAD STOR CH20 ADD Mass of CH2O added to storage organ OO QL%1 STOR %DAM SOIL Proportion of damaged storage organ below soil surface OO QL%S STOR %DAM SURF Proportion of damaged storage organ above soil surface OO QMAD STOR C MOBIL C mobilized from storage organ in a day O QMAM STOR PEST DAM Weight of storage organ consumed by pests today O QN%D STOR N% Storage organ N concentration (%) O QN%I STOR MIN N% Minimum fraction N for growing storage organ tissue OO QN%N STOR N %OF TOT N content in storage organ (fraction) OO QN%X STOR MAX N% Maximum fraction N for growing storage organ tissue O QNAA STOR N ADD RES N added to storage organ reserves O QNAC STOR CUM N ADD Cumulative N added to storage organ O QNAD STOR N WT Mass of N in storage organ OO QNAG STOR N ADD N added to storage organ today OO QNAL STOR CUM N LOSS Cumulative N lost from storage organ O QNAL STOR N LOSS N loss from storage organ today O QNAM STOR N MOBIL Actual n mobilized from storage organ in a day OO QNAM STOR NET N ADD Net n added to storage organ today OO

PAGE 219

205 QNAR STOR N MINE P Potential N mining rate from storage organ OO QNMD STOR N AVAIL N available for mobilization from storage organ OO QNRN STOR MIN N DEM Minimum N demand for storage organ O QNRX STOR MAX N DEM Maximum N demand for storage organ O QP%N STR CP% NEW GRO Protein fraction of new storage organ growth O QRAD STOR RESP REQ Respitation requirement for new storage growth OO QS%D STOR % SURFACE Proportion of storage organ mass on soil surface OO QT%1 STOR TEMPORARY1 Intemediate to calculate % storage organ below grouND O QT%S STOR TEMPORARYS Intemediate to calculate % storage organ above grouND O QV%D STOR % OF VEG Fraction of vegetative growth that goes to storage organ O QV%T STOR % OF TOT Fraction of growth going to storage organ OO QW%C STR CUM OBS DAM Observed cumulative percentage storage organ mass damage OO QWAD STOR DRY MASS Storage organ weight (kg dm/ha) O QWAI STOR INITIAL WT Initial storage organ weight O QWND STOR NEW GROW New storage tissue growth today OO QWNG STR NET GRO RTE Net storage organ growth rate OO XSTR STOR PART DIFF Diff. in storage organ partitioning between R1 and NDLEAF OO LV%D STOR % OF VEG Fraction of vegetative growth that goes to leaf O SV%D STOR % OF VEG Fraction of vegetative growth that goes to stem O RV%D STOR % OF VEG Fraction of vegetative growth that goes to Root O

PAGE 220

206 APPENDIX F CODE ADDITIONS AND CHANGES IN THE FORAGE VERSION OF CROPGRO Code changes are listed by modul e, alphabetically. Bold, underlined text describes the change or addition and is not part of the computer code. MODULE: CROPGRO.FOR Add dormancy variables to SUBROUTINE statement SUBROUTINE CRO PGRO(CONTROL, ISWITCH, & CO2, DAYL, EOP, YREND, HARVFRAC, NH4, NO3, !Input & PAR, SOILPROP, ST, SW TAVG, TDAY, TGRO, !Input & TGROAV, TMIN, TRWUP, YRPLT, !Input & CANHT, EORATIO, HARVRES, KEP, MDATE, !Output & NSTRES, PORMIN, RLV, RWUM X, SENESCE, !Output & STGDOY, UNH4, UNO3, XHLAI, XLAI) !Output Declare Dormancy and N returned N partitioning variables in CROPGRO C Storage organ parameters for forage model CHARACTER*6 DRMST CHARACTER*3 TYPPGD,TYPPTD, TYPPMD, TYPHRD, TYPDHD REAL AGRSTR, ALPHSR, CADSR, CADSRF, CLAIT, CMOBSR, & CMOBSRN, CMOBSRX, CP FSTR, CRUSSR, CSRFRZ, CSRW, CSTRM, & DSTOR, FNINSR, FN INSRG, FRSTR, FRSTRF, FRSTRM, & FRSTRMX, FRZDC, FRZDL, HARD1, HARD2, NADSR, & NGRSR, NGRSRG, NMOBSR, NMOBSRN, NMOBSRX, NRUSSR, NSRALL, & NSRDOT, NSROFF, NV STSR, PCARSR, PCNSR, PCSTRD, & PLIGSR, PLIPSR, PMINSR, POASR, PPGFAC, PPMFAC, & PPTFAC, PROSRF, PROSRG, PROSRI, PROSRT, & PSRLYR1, PSRLYRD, PSR SRFD, PSRSRFL, RCHDP, RDRMG, & RDRMM, RDRMT, RHOSR, SENSR, SRDAM, SRFTEMP, SRLYRD, & SRSRFD, SSRDOT, SSRN DOT, STRLYR1, STRSRFL, STRWT, & TPSRLYR1, TPSRSRFL, WCRSR, WNRSR, WRCSRDT, WSFDOT, & WSRDOT, WSRDOTN, WS RFDOT, WSRI, WSRIDOT, WTNSR, & WTNSRA, WTNSRO, WTSRO, XSTR REAL FNPGD(4), FNPMD(4), FN PTD(4), FRZDHD(4), FRZHRD(4), & YSTOR(8) REAL CADRT, CADSH, NADSH, NRUSTOT REAL PNMLF, PNMRT, PNMSH, PNMSR, PNMST

PAGE 221

207 REAL CSFRZ REAL NVSTL, NVSTR, NVSTS, NVSTSR REAL NUPNH4(NL), NUPNO3(NL) REAL LAIMOBR, VNMOBR, VEGNCNT, VEGNCMX CHARACTER*3 TYPLMOB, TYPNMOB REAL LRMOB(4), NRMOB(4), VEGNPCT, VEGNPMX, VNSTAT REAL CURV Add CALL statements for DORMANT CALL DORMANT( CONTROL, & DAYL, TMIN, !Input & DRMST, FREEZ2, FRZD C, PPGFAC, PPTFAC, PPMFAC, !Output & FNPGD, FNPMD, FNPTD, FRZDHD, FRZHRD, HARD1, !Output & HARD2, RCHDP, RDRMG, RDRMM, RDRMT, TYPDHD, TYPHRD, !Output & TYPPGD, TYPPMD, TYPPTD) !Output Reduce daily photosynthesis for Dormancy effect C-------------------------------------------------------------------C Reduce daily photosynthesis for Dormancy effect PG = PG PPGFAC Add storage organ to list of N and C mining sources. Default mobilization is 1% per day, modified by LAI and dormancy status. Low LAI, below threshold (LAI0.0), lowering mobilization as dormancy “deepens”. LAIMOBR = CURV(TYPLMOB,LRMOB(1),LRMOB(2),LRMOB(3), & LRMOB(4), MIN(XLAI,LRMOB(4))) C-------------------------------------------------------------------C Increase mobilization from storag e if N status of plant is high. C-------------------------------------------------------------------VEGNCNT = PCNL/100*WTLF + PCNST/100*STMWT + & PCNRT/100*RTWT + PCNSR/100*STRWT VEGNCMX = FNINL*WTLF + FNINS*STMWT + FNINR*RTWT + FNINSR*STRWT VNSTAT = MIN((VEGNCNT / VEGNCMX), 1.0) VNMOBR = CURV(TYPNMOB,NRMOB(1),NRMOB(2) ,NRMOB(3),NRMOB(4),VNSTAT) C--------------------------------------------------------------------

PAGE 222

208 C Set N mobilization rate from storage C Default to NMOBSRN under most conditions C set to NMOBSRX (max rate) after harvest or severe damage C Reduce from either level de pending on degree of dormancy C Mobilization from storage is unaffected by water or N stress C but is accelerated by low N status. C-------------------------------------------------------------------NMOBSR = (NMOBSRN + (N MOBSRX-NMOBSRN)*VNMOBR) NMOBSR = (NMOBSR + (NMOBS RX-NMOBSR)*LAIMOBR)*PPMFAC C-------------------------------------------------------------------C Set C mobilization rate from storage C Default to CMOBSRN under most conditions C set to CMOBSRX (max rate) after harvest or severe damage C Reduce from either level de pending on degree of dormancy C Mobilization from storage is unaffected by water or N stress C-------------------------------------------------------------------CMOBSR = (CMOBSRN + (C MOBSRX-CMOBSRN)*VNMOBR) CMOBSR = (CMOBSR + (CMOBS RX-CMOBSR)*LAIMOBR)*PPMFAC CMINEP=CMOBMX*(DTX+DXR57)*(WCRLF+WCRST+WCRSH)+ & CMOBMX*(DTX+DXR57)*(PPMFAC)*WCRRT + & CMOBSR*WCRSR*(DTX+DXR57) PGAVL = PG + CMINEP Deduct cost of new storage tissue from PGAVL and NAVL C-------------------------------------------------------------------C Deduct cost of new storage tissue from PGAVL and NAVL C-------------------------------------------------------------------PGAVL = PGAVL AGRVG (WLDOTN + WSDOTN + WRDOTN + WSRDOTN) NAVL = NAVL (NGRLF + NGRST + NGRRT + NGRSR) NAVL = NAVL (NADLF + NADST + NADRT + NADSR) PGAVL = PGAVL (CADST + CADLF + CADSR) PCH2O Modify CALL IPPLNT statement to include storage variables CALL IPPLNT(CONTROL, & CADPR1, CMOBMX, CROP, DETA CH, ECONO, EORATIO, !Output & FILECC, FILEGC, FRCNOD FREEZ1, FREEZ2, KCAN, KEP,!Output & NOUTDO, PCARSH, PCH2O, PLIPSH, PLIGSD, PLIG SH, !Output & PMINSD, PMINSH, POASD, P OASH, PORMIN, PROLFI, !Output & PRORTI, PROSHI, PROSTI, R 30C2, RCH2O, RES30C, !Output & RFIXN, RLIG, RLIP, RMIN, R NH4C, RNO3C, ROA, !Output & RPRO, RWUEP1, RWUMX, TTFIX !Output & PROSRI, STRSRFL, STRLYR1) !Output

PAGE 223

209 Initialize values for PSRSRFL and PSRLYR1 PSRSRFL = STRSRFL PSRLYR1 = STRLYR1 Modify CALL PEST statement to include storage variables CALL PEST(CONTROL, ISWITCH, & AREALF, CLW, CSW, LA GSD, LNGPEG, NR2, PGAVL, !Input & PHTIM, PLTPOP, RTWT, SLA, SLDOT, SOILPROP, !Input & SSDOT, STMWT, TOPWT, WLFDOT, WTLF, YRPLT, !Input & RLV, SDNO, SHELN, SWIDOT, !Input/Output & VSTAGE, WSHIDT, WTSD, WTSHE, !Input/Output & ASMDOT, DISLA, NPLTD, PPLTD, !Output & SDDES, WLIDOT, WRIDOT, WSIDOT,SDWT, !Output & CSRW, SSRDOT, STRW T, WSFDOT, WSRFDOT, !Input & WSRIDOT, !Output & CSFRZ, CSRFRZ, CSTR M, DSTOR, SRDAM) !Output Modify CALL PHOTO statement to include dormancy variables CALL PHOTO(CONTROL, & BETN, CO2, DXR57, EXCESS, NR5, PAR, SLPF, !Input & RNITP, SLAAD, SW FAC, TDAY, XHLAI, XPOD, !Input & AGEFAC, PG) !Output Modify CALL DEMAND statement to include storage variables IF (CRO P .NE. 'FA') THEN CALL DEMAND(SEASINIT, & AGRLF, AGRRT, AGRSH2, AGRSTM, CROP, DRPP, DXR57, !Input & FILECC, FILEGC, FILEIO, FNINSH, FRACDN, L AGSD, !Input & LNGPEG, NDLEAF, NSTRES, PAR, PCNL, PCNRT, PCNST, !Input & PGAVL, PUNCSD, PUNCTR, PLTPOP, RPROAV, RTWT, !Input & SDDES, SDNO, SDVAR, SH ELN, SHVAR, STMWT, SWFAC, !Input & TAVG, TDUMX, TDUMX2, TGRO, TURFAC, VSTAGE, WCRLF, !Input & WCRRT, WCRST, WNRLF, WNRRT, WNRSH, WNRST, WTLF, !Input & WTSD, WTSHE, XPOD, YRDOY, NVEG0, NR1, NR2, NR5, !Input & NR7, YRSIM, !Input & AGRSD1, AGRSD2, AGRVG, AGRVG2, CDMREP, F, FNINL, !Output & FNINR, FNIN S, FNINSD, FRLF, FRRT, FR STM, GDMSD, !Output & GRRAT1, NDMNEW, NDMOLD NDMREP, NDMSDR, NDMTOT, !Output & NDMVEG, NMINEP, NMOBR, PH TIM, PNTIM, POTCAR, !Output & POTLIP, SDGR, TURADD, XFRT, !Output & NMOBSR, PPMFAC, PPTFAC, PCNSR, STRWT, !Input & WCRSR, WLIDOT, WNRSR, XLAI, !Input & AGRSTR, FNINSR, FRSTR, !Output

PAGE 224

210 & FRSTRF, FRSTRM, FRSTRMX, LRMOB, NMOBSRN, NMOBSRX, !Output & NRMOB, NVSTL, NVSTR, NVSTS, NVSTSR, TYPLMOB !Output & TYPNMOB, XSTR, YSTOR) !Output Modify CALL NUPTAK statement to incl ude CH2O supply/requirement variables to limit N uptake by PGAVL CALL NUPTAK( & BD, DLAYR, DUL, FILECC, LL, NDMSDR, NDMTOT, NH4, !Input & NO3, NLAYR, PGAVL, RLV, RNH4C, RNO3C, SAT, SW, !Input & NUPNH4, NUPNO3, TRNH4U, TRNO3U, TRNU, UNH4, UNO3, !Output & RUNINIT) !Control Modify CALL INCOMP statement to include storage variables CALL INCOMP( & ECONO, FILECC, FILEGC, FRLF, FRRT, !Input & FRSTM, !Input & AGRLF, AGRNOD, AGRRT, AGRSD1, AGRSD2, !Output & AGRSH1, AGRSH2, AGRSTM, AGRVG, AGRVG2, !Output & SDPROR, !Output & AGRSTR, FRSTR, !Output & DYNAMIC) Modify CALL SENES statement to include storage variables CALL SENES( & FILECC, CLW, DTX, NR7, NRUSLF, PAR, RHOL, !Input & SLAAD, STMWT, SWFAC, VSTAGE, WTLF, XLAI, !Input & YRDOY, YRSIM, !Input & SLDOT, SLNDOT, SSDOT, SSNDOT, !Output & STRWT, !Input & SSRDOT, SSRNDOT, !Output & SENSR, !Output & INTEGR) !Control Modify CALL GROW statement to include storage variables CALL GROW(CONTROL, IS WITCH, RUNINIT, SOILPROP, & AGEFAC, CADLF, CADST, CRUSLF, CRUSRT, CRUSSH, !Input & CRUSST, DISLA, F, FILECC, FILEGC, FRLF, FRSTM, !Input & NADLF, NADRT, NADST, ND TH, NFIXN, NGRLF, NGRRT, !Input & NGRSD, NGRSH, NGRST, NM INEA, NODGR, NOUTDO, !Input & NPLTD, NRUSLF, NRUSRT, NRUSSH, NRUSST, POTCAR, !Input

PAGE 225

211 & POTLIP, PPLTD, SD IDOT, SDPROR, !Input & SENNOD, SENRT, SLDOT, SL NDOT, SRDOT, SSDOT, !Input & SSNDOT, TRNH4U, TRNO3U, TRNU, !Input & TURFAC, WLDOTN, WLIDOT WRDOTN, WRIDOT, WSDDTN, !Input & WSDOTN, WSHDTN, WSIDOT, WTABRT, WTSHMT, YRNR1, !Input & MDATE, YRPLT, !Input & SWIDOT, WLFDOT, WSHIDT, WTNFX, XHLAI, !Input/Output & AREALF, BETN, CANNAA, CA NWAA, CLW, CSW, DWNOD, !Output & DWNODA, GROWTH, GRWRES, LA IMX, PCCSD, PCLSD, !Output & PCNL, PCNRT, PCNSD, PCNSH, PCNST, PLTPOP, !Output & PLIGLF, PLIGNO, PLIGRT, PLIGSD, PLIGSH, PLIGST, !Output & PODWT, PUNCSD, PUNCTR, RHOL RHOS, RNITP, !Output & ROWSPC, RTWT, SDNPL, SDRA TE, SDWT, SDWTAM, !Output & SEEDNI, SEEDNO, SENESCE, SHEL WT, SLA, !Output & SLAAD, STMWT, TOPWT, TO TWT, WCRLF, WCRRT, WCRSH, !Output & WCRST, WNRLF, WNRRT, WNRSH, WNRST, WTCO, !Output & WTLF, WTLO, WTMAIN, WTNCAN, WTNEW, WTNLA, WTNLF, !Output & WTNLO, WTNNA, WTNNAG, WTN NO, WTNNOD, WTNOO, !Output & WTNRA, WTNRO, WTNRT, WTNS A, WTNSD, WTNSDA, !Output & WTNSDO, WTNSH, WTNSHA, WT NSHO, WTNSO, WTNST, !Output & WTNUP, WTRO, WTSDO, WTS HO, WTSO, XLAI, XPOD, !Output & CADRT, CADSH, NADSH, !Input & CADSR, CRUSSR, FRST R, NADSR, NGRSR, NRUSSR, !Input & PSRLYRD, PSRSRFD, PSRS RFL, PSRLYR1, SS RDOT, !Input & SSRNDOT, STRSRFL, ST RLYR1, WSRDOTN, WSRIDOT, !Input & WSFDOT, WSRFDOT, !Input/Output & CSRW, PCNSR, PLIGSR, RHOSR, STRWT, WCRSR, !Output & WNRSR, WTNSR, WTNS RA, WTNSRO, WTSRO, !Output & ALPHSR, PCARSR, PLIPSR PMINSR, POASR, PROSRF, !Output & CPFSTR, NSRALL, NSR DOT, NSROFF, TPSRLYR1, !Output & TPSRSRFL, WRCSRDT, WSRDOT, WSRI, !Output & VSTAGE) !Input/Output Move CALL DEMAND statement in EMERG step in front of CALL GROW to set proper initial partitioning of transplant DM !------------------------------------------------------------------IF (DAS .EQ. NVEG0) THEN !-----------------------------------------------------------------! On day of emergence, initialize: !------------------------------------------------------------------CALL DEMAND(EMERG, & AGRLF, AGRRT, AGRSH2, AGRSTM, CROP, DRPP, DXR57, !Input

PAGE 226

212 & FILECC, FILEGC, FILEIO, FNINSH, FRACDN, L AGSD, !Input & LNGPEG, NDLEAF, NSTRES, PAR, PCNL, PCNRT, PCNST, !Input & PGAVL, PUNCSD, PUNCTR, PLTPOP, RPROAV, RTWT, !Input & SDDES, SDNO, SDVAR, SH ELN, SHVAR, STMWT, SWFAC, !Input & TAVG, TDUMX, TDUMX2, TGRO, TURFAC, VSTAGE, WCRLF, !Input & WCRRT, WCRST, WNRLF, WNRRT, WNRSH, WNRST, WTLF, !Input & WTSD, WTSHE, XPOD, YRDOY, NVEG0, NR1, NR2, NR5, !Input & NR7, YRSIM, !Input & AGRSD1, AGRSD2, AGRVG, AGRVG2, CDMREP, F, FNINL, !Output & FNINR, FNIN S, FNINSD, FRLF, FRRT, FR STM, GDMSD, !Output & GRRAT1, NDMNEW, NDMOLD NDMREP, NDMSDR, NDMTOT, !Output & NDMVEG, NMINEP, NMOBR, PH TIM, PNTIM, POTCAR, !Output & POTLIP, SDGR, TURADD, XFRT, !Output & NMOBSR, PPMFAC, PPTFAC, PCNSR, STRWT, !Input & WCRSR, WLIDOT, WNRSR, XLAI, !Input & AGRSTR, FNINSR, FRSTR, !Output & FRSTRF, FRSTRM, FRSTRMX, LRMOB, NMOBSRN, NMOBSRX, !Output & NRMOB, NVSTL, NVSTR, NVSTS, NVSTSR, TYPLMOB, !Output & TYPNMOB, XSTR, YSTOR) !Output !------------------------------------------------------------------CALL GROW(CONTROL ISWITCH, EMERG, SOILPROP, & AGEFAC, CADLF, CADST, CRUSLF, CRUSRT, CRUSSH, !Input & CRUSST, DISLA, F, FILECC, FILEGC, FRLF, FRSTM, !Input & NADLF, NADRT, NADST, ND TH, NFIXN, NGRLF, NGRRT, !Input & NGRSD, NGRSH, NGRST, NM INEA, NODGR, NOUTDO, !Input & NPLTD, NRUSLF, NRUSRT, NRUSSH, NRUSST, POTCAR, !Input & POTLIP, PPLTD, SD IDOT, SDPROR, !Input & SENNOD, SENRT, SLDOT, SL NDOT, SRDOT, SSDOT, !Input & SSNDOT, TRNH4U, TRNO3U, TRNU, !Input & TURFAC, WLDOTN, WLIDOT WRDOTN, WRIDOT, WSDDTN, !Input & WSDOTN, WSHDTN, WSIDOT, WTABRT, WTSHMT, YRNR1, !Input & MDATE, YRPLT, !Input & SWIDOT, WLFDOT, WSHIDT, WTNFX, XHLAI, !Input/Output & AREALF, BETN, CANNAA, CA NWAA, CLW, CSW, DWNOD, !Output & DWNODA, GROWTH, GRWRES, LA IMX, PCCSD, PCLSD, !Output & PCNL, PCNRT, PCNSD, PCNSH, PCNST, PLTPOP, !Output & PLIGLF, PLIGNO, PLIGRT, PLIGSD, PLIGSH, PLIGST, !Output & PODWT, PUNCSD, PUNCTR, RHOL RHOS, RNITP, !Output & ROWSPC, RTWT, SDNPL, SDRA TE, SDWT, SDWTAM, !Output & SEEDNI, SEEDNO, SENESCE, SHEL WT, SLA, !Output & SLAAD, STMWT, TOPWT, TO TWT, WCRLF, WCRRT, WCRSH, !Output & WCRST, WNRLF, WNRRT, WNRSH, WNRST, WTCO, !Output & WTLF, WTLO, WTMAIN, WTNCAN, WTNEW, WTNLA, WTNLF, !Output

PAGE 227

213 & WTNLO, WTNNA, WTNNAG, WTN NO, WTNNOD, WTNOO, !Output & WTNRA, WTNRO, WTNRT, WTNS A, WTNSD, WTNSDA, !Output & WTNSDO, WTNSH, WTNSHA, WT NSHO, WTNSO, WTNST, !Output & WTNUP, WTRO, WTSDO, WTS HO, WTSO, XLAI, XPOD, !Output & CADRT, CADSH, NADSH, !Input & CADSR, CRUSSR, FRST R, NADSR, NGRSR, NRUSSR, !Input & PSRLYRD, PSRSRFD, PSRS RFL, PSRLYR1, SS RDOT, !Input & SSRNDOT, STRSRFL, ST RLYR1, WSRDOTN, WSRIDOT, !Input & WSFDOT, WSRFDOT, !Input/Output & CSRW, PCNSR, PLIGSR, RHOSR, STRWT, WCRSR, !Output & WNRSR, WTNSR, WTNS RA, WTNSRO, WTSRO, !Output & ALPHSR, PCARSR, PLIPSR PMINSR, POASR, PROSRF, !Output & CPFSTR, NSRALL, NSR DOT, NSROFF, TPSRLYR1, !Output & TPSRSRFL, WRCSRDT, WSRDOT, WSRI, !Output & VSTAGE) !Input/Output Modify CALL MOBIL statement to include storage variables and proportions of N mobilized from each organ CALL MOBIL( & NDMNEW, NMINEP, NMOBR, RP RO, TRNU, !Input & WNRLF, WNRRT, WNRSH, WNRST, !Input & NMINEA, NRUSLF, NRUSRT, NRUSSH, NRUSST, !Output & NMOBSR, PPMFAC, WNRSR, !Input & NRUSSR, PNMLF, PNMST, PNMRT, PNMSR, PNMSH, !Output & SEASINIT) !Control Add stem and storage organ variables to CALL FREEZE statement, then reset freeze damage variables for days without freeze C-------------------------------------------------------------------C Call freeze damage routine if TMIN is less than FREEZ1 deg C C-------------------------------------------------------------------IF (TMIN .LT. FREEZ1 .OR. TMIN .LT. FREEZ2) THEN CALL FREEZE(FILEIO, RUN, & FREEZ1, FREEZ2, IDETO, NOUTDO, NRUSLF, SLDOT, !Input & TMIN, WTLF, YRDOY, YRPLT, !Input & MDATE, !Input/Output & WLFDOT, !Output & FRZDC, NRUSSR, NRU SST, PSRSRFL, PSRLYR1, !Input & SRFTEMP, SSDOT, SSR DOT, ST, STMWT, STRWT, !Input & PSRLYRD, PSRSRFD, WS FDOT, WSRFDOT, !Output

PAGE 228

214 & FRZDL, SRLYRD, SR SRFD, VSTAGE) !Output ELSE WLFDOT = 0.0 WSRFDOT = 0.0 WSFDOT = 0.0 Modify CALL VEGGR statement to include storage variables CALL VEGGR( & AGRLF, AGRRT, AGRSTM, CMINEP, CSAVEV, & DTX, DXR57, ECONO, FILECC, FILEGC, FNINL, & FNINR, FNINS, NAVL, NDMNEW, NDMOLD, & NR1, NVSTL, NVSTR, NVSTS, NVSTSR, & PAR, PCH2O, PCNL, PCNST, PCNRT, PCNSR, PG, !Input & PGAVL, ROWSPC, RT WT, RVSTGE, STMWT, TGRO, & TURFAC, VSTAGE, WCRLF, WCRRT, WCRSH, & WCRST, WTLF, XL AI, YRDOY, YREMRG, YRSIM, & AGRVG, FRLF, FRRT, FRSTM, NMINEA, !Input/Output & NFIXN, TRNU, & CADLF, CADST, CANHT, CANWH, CMINEA, & CRUSLF, CRUSRT, CRUSSH, CRUSST, EXCESS, NADLF, & NADRT, NADST, NGRLF, NGRRT, NGRST, !Output & NSTRES, TNLEAK, WLDOT N, WRDOTN, WSDOTN, & CLAIT, NRUSTOT, !Input & PNMLF, PNMRT, PNMSH, PNMSR,PNMST,RPRO, & CADRT, CADSH, NADSH, !Output & AGRSTR, CMOBSR, FNINSR PPMFAC, STRWT, WCRSR, !Input & FRSTR, !Input/Output & CADSR, CRUSSR, NADSR NGRSR, WSRDOTN, !Output & CADSRF, CMOBSRN, CMOBSRX, !Output & FNINSRG, NGRSRG, PROSRG, PROSRT, !Output & NLAYR, NUPNH4, NUPNO3, PROLFI, PRORTI, !Input & PROSTI, PROSRI, RFIXN, RNH4C, RNO3C, TRNH4U, & TRNO3U, & SEASINIT) !Control Modify CALL PlantNBal statement to include storage variables IF (DYNAMIC .EQ. FINAL) THEN IF ( CROP .NE. 'FA') THEN CALL Plan tNBal (CONTROL, ISWITCH, & SEEDNI, TNLEAK, WTNF X, WTNLA, WTNLF, WT NLO, !Input

PAGE 229

215 & WTNNA, WTNNO, WTNNOD, WTNRA, WTNRO, WTNRT, !Input & WTNSA, WTNSD, WTNS DA, WTNSDO, WTNSH, WTNSHA, !Input & WTNSHO, WTNSO, WTNST, WTNU P, !Input & WTNSR, WTNSRA, WTNSRO) !Input ENDIF Define Dormancy variables at end of subroutine CMOBSRN Minimum fraction of C which can be mobilized from storage organ in a day CMOBSRX Maximum fraction of C which can be mobilized from storage organ in a day DRMST Dormancy status ( NODORM=not dormant, DORM=dormant reversible, FRZDL Todays death loss of storag e tissue/plant population due to freezing (proportion of STRWT and PLNTPOP) NADSH N added to shell N reserves (g[N] / m2 / d) NMOBSRN Minimum/ "normal" fracti on of N which can be mobilized from storage organ in a day NMOBSRX Maximum fraction of N which can be mobilized from storage organ in a day PPGFAC Reduction in photosynth etic rate due to dormancy PPMFAC Reduction in mobili zation rate due to dormancy PNMLF Proportion of actually mobilized N mobilized from leaves in a day PNMST Proportion of actually mobilized N mobilized from stems in a day PNMRT Proportion of actually mobilized N mobilized from roots in a day PNMSR Proportion of actually mobilized N mobilized from storage organ in a day PNMSH Proportion of actually mobilized N mobilized from shells in a day PPTFAC Reduction in partitioning to vegetative tissues during dormancy WSFDOT Stem weight losse s due to freezing (g[stem]/m2-d)

PAGE 230

216 MODULE: DEMAND.FOR Add storage/dormancy variabl es to SUBROUTINE command SUBROUTINE DEMAND(DYNAMIC, & AGRLF, AGRRT, AGRSH2, AGRSTM, CROP, DRPP, DXR57, !Input & FILECC, FILEGC, FILEIO, FNINSH, FRACDN, L AGSD, !Input & LNGPEG, NDLEAF, NSTRES, PAR, PCNL, PCNRT, PCNST, !Input & PGAVL, PUNCSD, PUNCTR, PLTPOP, RPROAV, RTWT, !Input & SDDES, SDNO, SDVAR, SH ELN, SHVAR, STMWT, SWFAC, !Input & TAVG, TDUMX, TDUMX2, TGRO, TURFAC, VSTAGE, WCRLF, !Input & WCRRT, WCRST, WNRLF, WNRRT, WNRSH, WNRST, WTLF, !Input & WTSD, WTSHE, XPOD, YRDOY, NVEG0, NR1, NR2, NR5, !Input & NR7, YRSIM, !Input & AGRSD1, AGRSD2, AGRVG, AGRVG2, CDMREP, F, FNINL, !Output & FNINR, FNIN S, FNINSD, FRLF, FRRT, FR STM, GDMSD, !Output & GRRAT1, NDMNEW, NDMOLD NDMREP, NDMSDR, NDMTOT, !Output & NDMVEG, NMINEP, NMOBR, PH TIM, PNTIM, POTCAR, !Output & POTLIP, SDGR, TURADD, XFRT, !Output & NMOBSR, PPMFAC, PPTFAC, PCNSR, STRWT, !Input & WCRSR, WLIDOT, WNRSR, XLAI, !Input & AGRSTR, FNINSR, FRSTR, !Output & FRSTRF, FRSTRM, FRSTRMX, LRMOB, NMOBSRN, NMOBSRX, !Output & NRMOB, NVSTL, NVSTR, NVSTS, NVSTSR, TYPLMOB, !Output & TYPNMOB, XSTR, YSTOR) !Output Declare Storage organ and dormancy variables REAL SDAGPL !------------------------------------------------------------------C Variables for adding storage organ and dormancy functions !------------------------------------------------------------------REAL AGRSTR, FNINSR, FRSTR, WLIDOT, & FRSTRF, FRSTRM, FRSTRM X, NMOBSR, NMOBSRN, NMOBSRX, & NVSTSR, PCNSR, PPM FAC, PPTFAC, PROSRF, PROSRI, & STRWT, TFRLF, TFRSTM TFRSTR, TFRRT, WCRSR, WNRSR, & XLAI, XSTR, YSTOR(25) Declare new variables for apportioning NDMVEG & NDMOLD as function of PGAVL !------------------------------------------------------------------C Variables for apportioning NDMVEG and NDMOLD !------------------------------------------------------------------REAL CDMOLD, CHOPRO, FROLDA, KCOLD

PAGE 231

217 Declare new variables for modifying mobilization from storage CHARACTER*3 TYPLMOB, TYPNMOB REAL LRMOB(4), NRMOB(4) Add storage, mobilization modifiers, and NDMOLD variables to CALL IPDMND statement CALL IPDMND( & FILECC, FILEGC, FILEIO, !Input & CARMIN, FINREF, FNSDT, FR LFF, FRLFMX, FRRTMX, !Output & FRSTMF, LIPOPT, LIPTB, LNGSH, NMOBMX, !Output & NRCVR, NVSMOB, PLIGSD, PMINSD POASD, !Output & PROLFF, PROLFI, PRORTF, PRORTI, PROSTF, PROSTI, !Output & RCH2O, RLIG, RLIP, RMIN, RNO3C, ROA, !Output & RPRO, SDLIP, SDPRO, SHLAG, SLAMAX, SLAMIN, !Output & SLAPAR, SLAREF, SLAVAR SLOSUM, SIZELF, SIZREF, !Output & SRMAX, THRESH, TURSLA, TYPSDT, VSSINK, XFRM AX, !Output & XFRUIT, XLEAF, XSLATM, XTRFAC, XVGROW, XXFTEM, !Output & YLEAF, YSLATM, YSTEM YTRFAC, YVREF, YXFTEM, !Output & FRSTRF, FRSTRMX, LRMOB, NMOBSRN, NMOBSRX, !Output & NRMOB, PLME, PROSRF, PROSRI, SDAGPL, TYPLMOB, !Output & TYPNMOB, YSTOR, KCOLD ) !Output Initialize FNINSR FNINSR=0.0 Add FRSTR to initial part itioning for seeded crops C-------------------------------------------------------------------C INITIALIZE PARTITI ONING PARAMETERS seedlings C-------------------------------------------------------------------FRLF = TABEX(YLEAF,XLEAF,0.0,8) FRSTM = TABEX(YSTEM,XLEAF,0.0,8) FRSTR = TABEX(YSTOR,XLEAF,0.0,8) FRRT = 1.0 FRLF FRSTM – FRSTR Assign value to FNINSR FNINSR = PROSRI 0.16 Calculate initial partitioning for transplants in EMERG step C-------------------------------------------------------------------C INITIALIZE PARTITIONING PARAM ETERS for transplants adjusted for C SDAGE and growth temperature ) C-------------------------------------------------------------------IF (PLME .EQ. 'T') THEN FRLF = TABEX(YLEAF,XLEAF,VSTAGE,8) FRSTM = TABEX(YSTEM,XLEAF,VSTAGE,8)

PAGE 232

218 FRSTR = TABEX(YSTOR,XLEAF,VSTAGE,8) FRRT = 1.0 FRLF FRSTM FRSTR ENDIF Calculate N available from storage organ – adjust for LAI, N status, and dormancy NMOBR = NVSMOB NMOBMX TDUMX IF (DAS .GT. NR5) THEN NMOBR = NMOBMX TDUMX2 (1.0 + 0.5*(1.0 SWFAC)) & (1.0 + 0.3*(1.0 NSTRES)) (NVSMOB + (1. NVSMOB) & MAX(XPOD,DXR57**2.)) ENDIF Add Storage N to CMINEP C-------------------------------------------------------------------C Add N from storage to N ava ilable for potential mobilization C-------------------------------------------------------------------NMINEP = NMOBR (WNRLF + WNRST + WNRSH) + & NMOBR PPMFAC WNRRT + & NMOBSR WNRSR Include Storage organ in partitioning scheme make behave like leaves and stems. Also allow rate to increase during dormancy !------------------------------------------------------------------C Fraction of growth going to leaves, roots and storage decreases C linearly between R1 and NDLEAF. C-------------------------------------------------------------------FRLFM = TABE X (YLEAF, XLEAF, VSTAGE, 8) FRSTMM = TABEX (YSTEM, XLEAF, VSTAGE, 8) FRSTRM = TABEX (YSTOR, XLEAF, VSTAGE, 8) YY = FRLFM FRLFF XX = FRSTMM FRSTMF XSTR = FRSTRM FRSTRF ENDIF !------------------------------------------------------------------IF (DAS .LT. NR1) THEN C-------------------------------------------------------------------C Calculate Pattern of Vegetati ve Partitioning, a function of V-STAGE C-------------------------------------------------------------------FRLF = TABEX(YLEAF,XLEAF,VSTAGE,8) FRSTM = TABEX(YSTEM,XLEAF,VSTAGE,8)

PAGE 233

219 FRSTR = TABEX (YSTOR, XLEAF, VSTAGE, 8) ELSE C-------------------------------------------------------------------C Partitioning between vegeta tive tissues depends on development C as expressed by FRACDN, the re lative development between R1 and NDLEAF C-------------------------------------------------------------------FRLF = FRLFM YY FRACDN FRSTM = FRSTMM XX FRACDN FRSTR = FRSTRM XSTR FRACDN IF ( DAS .GE. NDLEAF) THEN FRLF = FRLFF FRSTM = FRSTMF FRSTR = FRSTRF ENDIF ENDIF FRRT = 1. FRLF FRSTM -FRSTR IF (PPTFAC .GT. 0.0) THEN FRSTR = (FRSTRMX FRSTR) PPTFAC + FRSTR FRRT = (FRRTMX FRRT) PPTFAC + FRRT TFRLF = FRLF/(FRLF + FRSTM) (1-FRSTR-FRRT) TFRSTM = FRSTM/(FRLF + FRSTM) (1-FRSTR-FRRT) TFRRT = FRRT/(FRLF + FRSTM + FRRT) (1-FRSTR) FRLF=TFRLF FRSTM=TFRSTM FRLF=1.0 (FRSTM + FRRT + FRSTR) FRRT=1.0 (FRLF + FRSTM + FRSTR) ENDIF Continuation of implementati on of storage organ partitioning after adjusting FRLF for F and VSSINK C-------------------------------------------------------------------C Recompute FRSTM, FRSTR, and FRRT based on FRLF C-------------------------------------------------------------------FRSTM = (1. FRLF) FRSTM / (FRSTM + FRRT + FRSTR) FRSTR = (1. FRLF) FRSTR / (FRSTM + FRRT + FRSTR) FRRT = 1. FRLF FRSTM FRSTR C-------------------------------------------------------------------ENDIF C-------------------------------------------------------------------C Compute CH2O cost per g of ti ssue, excluding cost for protein (AGRVG)

PAGE 234

220 C and total CH2O cost per g of veg tissue (AGRVG2) C-------------------------------------------------------------------AGRVG = AGRLF FRLF + AGRRT FRRT + AGRSTM FRSTM & + AGRSTR FRSTR AGRVG2 = AGRVG + (FRLF* PROLFI+FRRT*PRORTI+FRSTM*PROSTI+ & FRSTR*PROSRI)*RPROAV C-------------------------------------------------------------------C Compute N Demand for New Tissue, including reproductive and vegetative C-------------------------------------------------------------------NDMVEG = (CDMVEG/AGRVG2) (FRLF*FNINL+FRSTM*FNINS+ & FRRT*FNINR+FRSTR*FNINSR) NDMNEW = NDMREP + NDMVEG Calculate value for NVSTSR for calculat ion of refilling of storage N during vegetative growth NVSTSR = FNINSR Calculate value for NVSTSR for calculat ion of refilling of storage N during reproductive growth NVSTSR = PROSRF*0.16 + (F NINSR-PROSRF*0.16) FRNLFT Calculate NDMOLD – refilli ng of N in old tissue including storage organ NDMOLD = (WTLF WCRLF) MAX(0.0,(NVSTL PCNL /100.)) & + (STMWT WCRST) MAX(0.0,(NVSTS PCNST/100.)) & + (RTWT WCRRT) MAX(0.0,(NVSTR PCNRT/100.)) & + (STRWT WCRSR) MAX(0.0,(NVSTSR PCNSR/100.)) Recalculate NDMOLD and NDMVEG, limiting to a total CH2O demand=PGAVL C-------------------------------------------------------------------C KJB/SJR New code to minimize/fix situation of overs pending PGAVL on C N demand. Original code uses all PGAVL for CDMREP and CDMVEG C then goes ahead and calculates NDMOLD using CDMVEG again. C Get to N uptake and can potentially take up more N than have CHO C to reduce. Cont ributes to NLEAK. C-------------------------------------------------------------------IF (NDMOLD .GT. 0.0) THEN CHOPRO = NDMOLD 6.25 RNO3C FROLDA = 1-EXP(-KCOLD (CDMVEG / CHOPRO)) CDMOLD = CHOPRO FROLDA NDMOLD = FROLDA NDMOLD CDMVEG = CDMVEG CDMOLD

PAGE 235

221 NDMVEG = (CDMVEG/AGRVG2) (FRLF*FNINL+FRSTM*FNINS+ & FRRT*FNINR+FRSTR*FNINSR) ENDIF C-------------------------------------------------------------------C KJB/SJR End new code. Now CDMTOT=PGAVL, not more. C-------------------------------------------------------------------Add storage organ parameters to SUBROUTINE statement for IPDMND subroutine SUBROUTINE IPDMND( & FILECC, FILEGC, FILEIO, !Input & CARMIN, FINREF, FNSDT, FR LFF, FRLFMX, FRRTMX, !Output & FRSTMF, LIPOPT, LIPTB, LNGSH, NMOBMX, !Output & NRCVR, NVSMOB, PLIGSD, PMINSD POASD, !Output & PROLFF, PROLFI, PRORTF, PRORTI, PROSTF, PROSTI, !Output & RCH2O, RLIG, RLIP, RMIN, RNO3C, ROA, !Output & RPRO, SDLIP, SDPRO, SHLAG, SLAMAX, SLAMIN, !Output & SLAPAR, SLAREF, SLAVAR SLOSUM, SIZELF, SIZREF, !Output & SRMAX, THRESH, TURSLA, TYPSDT, VSSINK, XFRM AX, !Output & XFRUIT, XLEAF, XSLATM, XTRFAC, XVGROW, XXFTEM, !Output & YLEAF, YSLATM, YSTEM YTRFAC, YVREF, YXFTEM, !Output & FRSTRF, FRSTRMX, LRMOB, NMOBSRN, NMOBSRX, !Output & NRMOB, PLME, PROSRF, PROSRI, SDAGPL, TYPLMOB, !Output & TYPNMOB, YSTOR, KCOLD ) !Output Define storage organ parameters as REAL REAL FRSTRF, FRSTRMX, NM OBSRX, PROSRF, PROSRI, & SDAGPL, REAL YSTOR(25) Define NDMOLD curvature (K) factor as REAL REAL KCOLD Declare mobilization rate modifiers CHARACTER*3 TYPLMOB, TYPNMOB REAL LRMOB(4) REAL NRMOB(4), VEGNPCT, VEGNPMX, VNMOBR, VNSTAT Add read statement for storage pr otein concentration parameters CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(F6.0,6X, F6.0)',IOSTAT=ERR) PROSRI,PROSRF IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM)

PAGE 236

222 Add read statement for KCOLD parameter CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'( F6.0)',IOSTAT=ERR) KCOLD IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) Add read statement for storage N mobiliz ation parameter NMOBSRX mobilization modifiers CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(18X,2F6.0 )',IOSTAT=ERR) NMOBSRN, NMOBSRX IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(4(1X ,F5.2),3X,A3)',IOSTAT=ERR) & (LRMOB(II),II=1,4), TYPLMOB IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(4(1X ,F5.2),3X,A3)',IOSTAT=ERR) & (NRMOB(II),II=1,4), TYPNMOB IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) Add READ statement for stor age partitioning parameters CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(8F6. 0)',IOSTAT=ERR)(YSTOR(II),II=1,8) IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(3F6.0)', IOSTAT=ERR) FRSTRF,FRSTRMX,FRRTMX IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) AGRSTR Mass of CH2O require d for new storage organ growth (g[CH2O] / g[storage]) CDMOLD Total CH2O de mand for refilling old tissue N (g[CH2O] / m2 / d) CHOPRO CHO required for uptake and reduction of N to fully refill ol d tissue N (g[CH2O] / m2 / d) CMOBSRX Maximum storage organ C pool mobilization rate (g [CH2O] / m2 / d) DRMRED Intermediate value calc ulated in determining reduction in mobilization due to dormancy FNINSR Maximum fraction of N for growing storage tissue (g[N] / g[storage]) FROLDA Fraction of max potenti al NDMOLD allowed to be met given

PAGE 237

223 today's level of CDMVEG. Prevents refilling old tissue without allowi ng any new growth due to low PG. FRRTMX Maximum proportion of vege tative growth that goes to roots on a day. Occurs only during dormancy. (g[root] / g[veg]) FRSTR Fraction of vegetative tissue growth that goes to storage organ on a day (g[storage] / g[veg]) FRSTRF Fraction of daily dry wei ght increase in vege tative plant parts which goes to storage organ after the day on which the maximum number of V-stages occurs (NDVSTG). (g[storage] / g[veg]) FRSTRM Fraction of growth going to storage organ decreases linearly between R1 and NDLEAF (g[stem] / g[veg]) FRSTRMX Maximum storage organ partitioning (g[storage] / g[veg]) KCOLD Curvature factor (K value) for exponential function limiting NDMOLD when PG is low LAIMOBR Effect of LAI on N & C mobilization. (0= no effect, mobilization at minimum rate, 1.0 = increase mobilization to maximum rate. Increases mobilization after harvest, damage. LRMOB(4) CURV response of mobilization to current LAI NMOBSR Stage-dependent potentia l N mining rate from storage organ expressed as a fraction of the maxi mum rate (NMOBSRX) NMOBSRN Minimum/ "normal" fracti on of N which can be mobilized from storage organ in a day NMOBSRX Maximum fraction of N which can be mobilized from storage organ in a day! NVSTSR N content in storage tissue (fraction) NRMOB(4) CURV response of mobilization to current vegetative N status of plant relative to maximum N concentration PCNSR Percent N in storage organ tissue (100 g[N] / g[storage]) PPMFAC Reduction in mobilization fr om storage organ due to photoperiod induced dormancy PPTFAC Reduction in partitioning to shoot due to photoperiod induced dormancy PROSRF Minimum storage orga n protein composition after N mining (g[protein] / g[storage]) PROSRI Maximum protein compositi on in storage organ during growth with luxurious supply of N (g[protein] / g[storage]) STRWT Dry mass of storag e organ tissue, including C and N (g[storage] / m2[ground) TYPLMOB Shape of CURV re sponse for mobilization to LAI TYPNMOB Shape of CURV response for mobilization to current vegetative N status of plant relati ve to maximum N concentration WCRSR Mass of CH2O reserves in storage organ (g[storage CH2O] / m2[ground]) WNRSR N available for mobiliza tion from storage organ above lower mining (g[N] / m2) XLAI Leaf area (one side) per unit of ground area

PAGE 238

224 XSTR Difference between partiti oning fraction to storage organ at beginning bloom (R1) and at the day on which the maximum number of V-stages occurs (NDLEAF) YSTOR(I) Partitioning factor for st orage organ growth at V-stage XSTOR(I) (g[storage] / g[veg. plant])

PAGE 239

225 MODULE: DORMANCY.FOR Calculate effect of dormancy on partitio ning, photosynthesis, and mobilization for the current day. Also calculate cold-har dening and dehardening for current day. Update FREEZ2 for cold-hardening/deharde ning. Add new .OUT output files for dormancy and STOR-related variables. C======================= ==================== ================== C DORMANCY Subroutine 6/20/03 SJR C Fall Dormancy with cold hardening for perennial grasses and legumes C Separate functions for effects on partitioning, photosynthesis, mobilization C Generate a reduction factor for ea ch for use in appropriate modules C Factor is 0-1 value adjusted for cultivar sensitivity to daylength C Cold hardening lowers the minimum survivable temperature for the crop C with increased exposure to low temperatures. C This subroutine also provides a death rate to allow partial C or total depletion of the stand by a freeze event C------------------------------------------------------------------C Called by: CROPGRO C Calls : None C======================= ==================== ================== SUBROUTINE DORMANCY( CONTROL, & DAYL, TMIN, !Input & DRMST, FREEZ2, FRZD C, PPGFAC, PPTFAC, PPMFAC, !Output & FNPGD, FNPMD, FNPTD, FRZDHD, FRZHRD, HARD1, !Output & HARD2, RCHDP, RDRMG, RDRMM, RDRMT, TYPDHD, TYPHRD, !Output & TYPPGD, TYPPMD, TYPPTD) !Output USE ModuleDefs !Definitio ns of constructed variable types, which contain control information, soil parameters, hourly weather data. C-------------------------------------------------------------------IMPLICIT NONE C-------------------------------------------------------------------CHARACTER*1 BLANK CHARACTER*3 TYPPGD,TYPPTD, TYPPMD, TYPHRD, TYPDHD CHARACTER*6 SECTION, ECOTYP, ECONO, ERRKEY, DRMST CHARACTER*12 FILEIO, FILEC, FILEE CHARACTER*16 ECONAM CHARACTER*80 PATH CR, PATHEC, CHAR CHARACTER*92 FILECC, FILEGC CHARACTER*255 C255

PAGE 240

226 INTEGER RUNINIT, SEASINIT, RATE, EMERG, INTEGR, OUTPUT, FINAL, $ DYNAMIC PARAMETER (RUNINIT = 1, SEASINIT = 2, EMERG = 3, RATE = 3, & INTEGR = 4, OUTPUT=5, FINAL = 6) INTEGER DYNAMIC, II, LUNIO, LUNECO INTEGER LUNCRP, ISECT, ERRNUM, PATHL INTEGER ERR, LINC, LNUM, FOUND PARAMETER (BLANK = ') PARAMETER (ERRKEY = 'DORMAN') PARAMETER (LUNIO=21, LUNCRP=10, LUNECO=10) REAL DAYL, CURV, TMIN REAL PPGFAC, PPTFAC, PPMFAC, RDRMT, RDRMG, RDRMM REAL FNPGD(4), FNPTD(4), FN PMD(4), FRZHRD(4), FRZDHD(4) REAL DAYLY, RCHDP REAL DHARDR, FREEZ2, FR ZDC, HARD1, HARD2, HARDR C The variable "CONTROL" is of type "ControlType". TYPE (ControlType) CONTROL TYPE (SwitchType) ISWITCH C Transfer values from construc ted data types into local variables. DYNAMIC = CONTROL % DYNAMIC FILEIO = CONTROL % FILEIO LUNIO = CONTROL % LUNIO LUNECO = CONTROL % LUNECO C********************************************************************** C Run Initialization Called once per simulation C********************************************************************** IF (DYNAMIC .EQ. RUNINIT) THEN C-------------------------------------------------------------------C Initialize Dormancy variables C-------------------------------------------------------------------PPGFAC = 0.0 PPTFAC = 0.0 PPMFAC = 0.0

PAGE 241

227 DRMST = 'NODORM' DAYLY = 0.0 RDRMT = 1.0 RDRMG = 1.0 RDRMM = 1.0 C-------------------------------------------------------------------C Read in values from temporary file, which were previously input C in Subroutine IPIBS. C-------------------------------------------------------------------OPEN (LUNIO, FILE = FILEIO, STATUS = 'OLD', IOSTAT=ERRNUM) IF (ERRNUM .NE. 0) CALL ERROR(ERRKEY,ERRNUM,FILEIO,0) READ (LUNIO,100) FILEC, PATHCR 100 FORMAT(//////,15X,A12,1X,A80) READ (LUNIO,105) FILEE, PATHEC 105 FORMAT(15X,A12,1X,A80) C-------------------------------------------------------------------C Subroutine FIND finds appropriate SECTION in a file by C searching for the specified 6-character string at beginning C of each line. C-------------------------------------------------------------------SECTION = '*HARVE' CALL FIND(LUNI O, SECTION, LNUM, FOUND) IF (FOUND .EQ. 0) CALL ERROR (ERRKEY,2,FILEIO,LNUM) C-------------------------------------------------------------------C Find and read Cultivar Section C-------------------------------------------------------------------SECTION = '*CULTI' CALL FIND(LUNI O, SECTION, LNUM, FOUND) IF (FOUND .EQ. 0) CALL ERROR (ERRKEY,2,FILEIO,LNUM) READ(LUNIO,1650) ECONO 1650 FORMAT(24X,A6) CLOSE (LUNIO) C-------------------------------------------------------------------C Open FILEE C-------------------------------------------------------------------LNUM = 0 PATHL = INDEX(PATHEC,BLANK)

PAGE 242

228 IF (PATHL .LE. 1) THEN FILEGC = FILEE ELSE FILEGC = PATHEC(1:(PATHL-1)) // FILEE ENDIF C-------------------------------------------------------------------C Read Ecotype Parameter File C-------------------------------------------------------------------C------------------------------------------------------------------C READ FILEE C-------------------------------------------------------------------OPEN (LUNECO,FILE = FI LEGC,STATUS = 'OLD',IOSTAT=ERR) IF (ERRNUM .NE. 0) CALL ERROR(ERRKEY ,ERRNUM,FILEE,0) ECOTYP = ' DO WHILE (ECOTYP .NE. ECONO) CALL IGNORE(LUNECO, LNUM, ISECT, C255) IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILEGC,0) IF ((ISECT .EQ. 1) .AND. (C255(1:1) .NE. '*')) THEN READ (C255,'(A6,1X,A16,121X,4F6.0)',IOSTAT=ERR) & ECOTYP, ECONAM, RDRMT,RDRMG,RDRMM, RCHDP IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILEGC,LNUM) IF (ECOTYP .EQ. ECONO) THEN EXIT ENDIF ELSE IF (ISECT .EQ. 0) THEN IF (ECONO .EQ. 'D FAULT') CALL ERROR(ERRKEY,3,FILEGC,LNUM) ECONO = 'DFAULT' REWIND(LUNECO) ENDIF ENDDO CLOSE (LUNECO) C-------------------------------------------------------------------C Open FILEC C-------------------------------------------------------------------LINC = 0

PAGE 243

229 PATHL = INDEX(PATHCR,BLANK) IF (PATHL .LE. 1) THEN FILECC = FILEC ELSE FILECC = PATHCR(1:(PATHL-1)) // FILEC ENDIF OPEN (LUNCRP,FILE = FI LECC, STATUS = 'OLD',IOSTAT=ERR) IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,0) C-------------------------------------------------------------------C READ DORMANCY PARAMETERS ******************* C-------------------------------------------------------------------SECTION = '!*DORM' CALL FIND(LUN CRP, SECTION, LINC, FOUND) IF (FOUND .EQ. 0) THEN CALL ERRO R(ERRKEY, 1, FILECC, LINC) ELSE CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR,'(4F6.0,3X,A3)',I OSTAT=ERR) (FNPTD(II),II=1,4), TYPPTD IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR,'(4F6.0,3X,A3)',I OSTAT=ERR) (FNPMD(II),II=1,4), TYPPMD IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR,'(4F6.0,3X,A3)',I OSTAT=ERR) (FNPGD(II),II=1,4), TYPPGD IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) ENDIF !------------------------------------------------------------------! Find and read Temperature threshold for cold hardening in !------------------------------------------------------------------CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR,'(3F 6.0)',IOSTAT=ERR)HARD1,HARD2,FRZDC IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR, '(4F6.0,3X,A)',IOSTAT=ERR) & (FRZHRD(II), II=1,4), TYPHRD

PAGE 244

230 IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR, '(4F6.0,3X,A)',IOSTAT=ERR) & (FRZDHD(II), II=1,4), TYPDHD IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) ENDIF CLOSE (LUNCRP) !*********************************************************************** Seasonal initiali zation run once per season !*********************************************************************** !*********************************************************************** Initialize yesterdays dayle ngth to that passe d in from PLANT !*********************************************************************** ELSE IF (DYNAMIC .EQ. SEASINIT) THEN DAYLY=DAYL C-------------------------------------------------------------------C All cultivars share the same freeze-killing temperature before C cold hardening C C Minimum survivable temperature after hardening varies with the C cold hardening potential of the cultivar C-------------------------------------------------------------------FREEZ2=HARD1 HARD2 = HARD1-(HARD1-HARD2)*RCHDP C-------------------------------------------------------------------C C THE FOLLOWING GENERATES HEADINGS FOR NEW OUTPUT FILE DORMANT.OUT C C-------------------------------------------------------------------! OPEN(UNIT = NOUTDT, FI LE = OUTT, STATUS = 'UNKNOWN') ! IF (IDETL .EQ. 'Y') THEN ! C-------------------------------------------------------------------C Variable heading for DORMANT.OUT C--------------------------------------------------------------------

PAGE 245

231 ! WRITE (NOUTDT,2202) NREP,TITLET, & MODEL,CROPD, EXPER,CROP,ENAME,TR TNO,TITLET, ECOTYP, ECONAM 2202 FORMAT (/,'*RUN ',I3,8X,': ',A25,/, & 1X,'MODE L',10X,':',1X,A8 ,' ',A10,/, & 1X,'EXPERIMENT',5X,':',1X,A8,1X,A2,4X,A47,/, & 1X,'TR EATMENT',I3, 3X,':',1X,A25,/, & 1X,'ECOTYPE', 8X,':',1X,A6,1X,A16,/) ! WRITE (NOUTDT,2203) 2203 FORMAT('@DATE', & DAYL DRMST PPGFAC PRTFAC WTLF WCRL F LFDM STMWT', & WCRST STDM STRW T WCRSR SRDM RT WT WCRRT RTDM') ENDIF ELSE IF (DYNAMIC .EQ. EMERG) THEN NONE C********************************************************************** C Daily Rate Calculations C********************************************************************** ELSE IF (DYNAMIC .EQ. RATE) THEN NONE C********************************************************************** C Daily Integration C********************************************************************** ELSE IF (DYNAMIC .EQ. INTEGR) THEN C********************************************************************** C Calculate cold-h ardening status for day C Killing freeze temperature decr eases as cold hardening proceeds. C Cold hardening is reversible while days are getting shorter. C Maximum rate of hardening (degre es C decrease in FREEZ2 per day) C occurs at FRZHRD(1) with fracti onal rates between FRZHRD(1) and FRZHRD(2) C Cold hardening is reversed be tween FRZHRD(2) and FRZHRD(3).

PAGE 246

232 C Dehardening will not occur until daylength begins to increase. C Dehardening is not reversible. C FRZDC is the rate of plant/ti ssue death per degree C below FREEZ2. C This allows gradual killing of the stand with increased rate at C lower temperatures. C Adapted from ALFACOLD model (Kanne ganti et al., 1998, Agron J. 90:687-697) C Note: this routine allows hardeni ng and dehardening to occur on the C same day with reverse hardening and dehardening combining to C accelerate dehardening at higher temperatures. C********************************************************************** IF (DAYL .LE. DAYLY) THEN HARDR = CURV(TYPHRD,FRZHRD(1),FRZHRD(2 ),FRZHRD(3),FRZHRD(4),TMIN) DHARDR = 0.0 ELSE HARDR = CURV(TYPHRD,FRZHRD(1),FRZHRD(2 ),FRZHRD(3),FRZHRD(4),TMIN) DHARDR = CURV(TYPDHD,FRZDHD(1),FRZDHD(2) ,FRZDHD(3),FRZDHD(4),TMIN) ENDIF FREEZ2 = FREEZ2 (HARDR DHARDR)*RCHDP FREEZ2 = MIN(HARD1, FREEZ2) FREEZ2 = MAX(FREEZ2, HARD2) C********************************************************************** C Calculate Partitioning, Pg, and mobilization reduction C factors and dormancy state for day C********************************************************************** PPTFAC=CURV(TYPPTD,FNPTD(1),FNPTD(2),FNPTD(3),FNPTD(4),DAYL) PPTFAC=RDRMT PPTFAC PPTFAC=MIN (PPTFAC,1.0) FNPGD(4) = RD RMG FNPGD(4)

PAGE 247

233 PPGFAC=CURV(TYPPGD,FNPGD(1),FNPGD(2),FNPGD(3),FNPGD(4),DAY L) PPGFAC=MIN (PPGFAC,1.0) FNPMD(4) = RDRMM FNPMD(4) PPMFAC=CURV(TYPPMD,FNPMD(1),FNPMD(2),FNPMD(3),FNPMD(4),DA YL) PPMFAC=MIN (PPMFAC,1.0) IF (PPTFAC .GT. 0.0 .OR. PPGFAC .LT. 1.0 & .OR. PPMFAC .LT. 1.0) THEN DRMST='DORM' ELSE DRMST='NODORM' ENDIF DAYLY=DAYL ELSE IF (DYNAMIC .EQ. OUTPUT) THEN C-------------------------------------------------------------------C Calculate new variables for DORMANT.OUT C-------------------------------------------------------------------C-------------------------------------------------------------------C Sent daily growth and partitioning detail to DORMANT.OUT C-------------------------------------------------------------------! IF (IDETL .EQ. 'Y') THEN C-------------------------------------------------------------------C Print out dormancy parameters TEMPORARY C-------------------------------------------------------------------! WRITE (NOUTDT,401) YRDOY, DAYL, DRMST, PPGFAC, & PPTFAC, WTLF, WCRLF, LFDM, & STMWT, WCRST, STDM, STRWT, WCRSR, SRDM, & RTWT, WCRRT, RTDM 401 FORMAT (1X,I5,1X,F5.2,1X,A6,1X,F5.3,1X,F5.3,1X,F6.1, & 1X,F6.1,1X,F6.1,1X,F7.1,1X,F6.1,1X,F6.1,1X,F7.1, & 1X,F6.1,1X,F6.1,1X,F6.1,1X,F6.1,1X,F6.1) ENDIF

PAGE 248

234 !*********************************************************************** FINAL !*********************************************************************** ELSE IF (DYNAMIC .EQ. FINAL) THEN C-------------------------------------------------------------------C Close DORMANT.OUT C-------------------------------------------------------------------! CLOSE (NOUTDT) END IF RETURN END !SUBROUTINE DORMANT !------------------------------------------------------------------! DORMANT VARIABLES !------------------------------------------------------------------! BLANK ' C255 255 character record DAYL Current daylength (hours) DAYLY Yesterdays daylength (hours) DRMST Dormancy status ( NODORM=not dormant, DORM=dormant reversible, DYNAMIC Controls run sequence: DYNAMIC =RUNINIT, SEASINIT, RATE, EMERG, INTEGR, OUTPUT, or FINAL ECONAM Ecotype name not used ECONO Used to match ECOTYP in .ECO file ECOTYP Ecotype code ERR ERRKEY ERRNUM FILEC, FILEE Filenames for Crop and Species files FILECC, FILEGC File+pathname for Crop and Eco files FILEIO Filename for Input file FNPGD(1) Base daylength for CURV function for daylength effect on Pg for short-day dormancy (daylength when dormancy is maximum) FNPGD(2) Daylength threshol d where dormancy effect begins for daylength effect on Pg (for short-day dormancy) FNPGD(3) Longest daylength threshol d where there is no dormancy effect for daylength effect on Pg (for long-day dormancy) FNPGD(4) Daylength when dormancy effect is maximum for daylength effect on Pg (long-day dormancy) FNPMD(1) Base daylength for CURV f unction for daylength effect on mobilization for short-day dormancy (daylength when dormancy is maximum) FNPMD(2) Daylength threshol d where dormancy effect begins

PAGE 249

235 for daylength effect on mobilization (for short-day dormancy) FNPMD(3) Longest daylength thres hold where there is no dormancy effect for daylength effect on m obilization (for long-day dormancy) FNPMD(4) Daylength when dormancy effect is maximum for daylength effect on mobilization (long-day dormancy) FNPTD(1) Ignored for short-day dormancy FNPTD(2) Daylength threshold where dormancy effect is maximum for daylength effect on partitioning (for short-day dormancy) FNPTD(3) Shortest daylength thres hold where there is no dormancy effect for daylength effect on partitioning (for short-day dormancy) FNPTD(4) Minimum relative effect of dor mancy when crop is non-dormant (set to 0.0) FREEZ2 Temperature below which plant growth stops completely. (C) FRZDC Freezing death coefficient pe rcentage tissue/population death per day per degree below FREEZ2) FRZDHD(1) Minimum temperature at whic h dehardening begins (relative rate=0) FRZDHD(2) Temperature at which dehardening reaches maximum rate (relative rate=1) FRZDHD(3) Not used FRZDHD(4) Maximum (absolute) rate of dehardening (degrees C increase above HARD2 per day)n of STRWT and PLNTPOP) FRZHRD(1) Temperature at which cold ha rdening reaches maximum rate (relative rate=1) FRZHRD(2) Temperature below which cold hardening begins (relative rate=0) FRZHRD(3) Temperature at which hardening is reversed at maximum rate (relative rate=-1) FRZHRD(4) Maximum (absolute) rate of co ld hardening (degrees C decrease towards HARD2 per day) HARD1 Killing low temperature before cold hardening (begins killing storage organ) HARD2 Killing low temperature after cold hardening (begins killing storage organ) ISECT LNUM LUNECO Logical unit number for ECO files LUNIO Input file logical unit no. PATHL PPGFAC Reduction in photosynth etic rate due to dormancy PPMFAC Reduction in mobili zation rate due to dormancy PPTFAC Reduction in partitioning to vegetative tissues during dormancy RCHDP Ecotype relative co ld hardening potential (0-1) RDRMG Relative sensitivity of ecotype to daylength/dormancy effects on Pg RDRMM Relative sensitivity of eco type to daylength/ dormancy effects on mobilization RDRMT Relative sensitivity of ecotype to daylength/dormancy effects on partiti oning to perenniating tissues SECTION Heading name in input files TMIN Daily average temperature

PAGE 250

236 TYPDHD Response type for cold dehardening TYPHRD Response type for cold hardening TYPPGD Type of response curve for effect of daylength/dormancy on Pg TYPPMD Type of response curve for effect of daylength/dormancy on mobilization TYPPTD Type of response curve fo r effect of daylength/dormancy on partitioning to perenniating organ !------------------------------------------------------------------! END SUBROUTINE DORMANT !-------------------------------------------------------------------

PAGE 251

237 MODULE: ETPHOT.FOR Declare photosynthetic pathway parameter variables CHARACTER PGPATH*2 REAL CCNEFF, CICA D, CMXSF, CQESF Add photosynthetic pathway parameters to CALL PGINP statement IF (MEPHO .EQ. 'L .AND. CROP .NE. 'FA') THEN CALL PGINP( & FILEIO, LUNIO, SALB, !Input & AZIR, BETN, FNPGL, FNPGN, LFANGD, LMXREF, !Output & LNREF, NSLOPE, PALBW, QEREF, ROWSPC, !Output & SCVP, SLWREF, SLWSLO TYPPGL, TYPPGN, !Output & XLMAXT, YLMAXT, PHTHRS10, & CCNEFF, CICAD, CMXSF, CQ ESF, PGPATH) !Output Add photosynthetic pathway parameters to SUBROUTINE PGINP statement SUBROUTINE PGINP( & FILEIO, LUNIO, SALBW, !Input & AZIR, BETN, FNPGL, FNPGN, LFANGD, LMXREF, !Output & LNREF, NSLOPE, PALBW, QEREF, ROWSPC, !Output & SCVP, SLWREF, SLWSLO, TYPPGL, TYPPGN, !Output & XLMAXT, YLMAXT, PHTHRS10, & CCNEFF, CICAD, CMXSF, CQESF, PGPATH) !Output Declare photosynthetic pathway parameter variables in PGINP CHARACTER PGPATH*2 REAL CCNEFF, CICA D, CMXSF, CQESF Declare photosynthetic pathway parame ter variables in CANOPG subroutine CHARACTER PGPATH*2 REAL CCNEFF, CICA D, CMXSF, CQESF Read photosynthetic pathway parameters from species file CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,L NUM,ISECT,C80) !12th line READ(C80,'(4F6.0,2X,A)',IOSTAT=ERRNUM) CICAD,CCNEFF, & CMXSF,CQESF,PGPATH IF (ERRNUM .NE. 0) CALL ERROR(ERRKEY,ERRNUM,FILECC,LNUM+12) Add photosynthetic pathway parame ters to CALL ETPHR statement CALL ETPHR( & CANHT, CEC, CEN, CLOUDS, CO2HR, DAYTIM, !Input

PAGE 252

238 & DLAYR2, DULE, FNPG L, FNPGN, FRACSH, FRSHV, !Input & KDIRBL, LAISH, LAISHV, LAISL, LAISLV, LLE, !Input & LMXREF, LNREF, LW IDTH, MEEVP, MEPHO, NLAYR, !Input & NSLOPE, PARSH, PA RSUN, QEREF, RABS, RCUTIC, !Input & REFHT, RHUMHR(H), RNITP, RWUH, SHCA P, SLAAD, !Input & SLWREF, SLWSLO, ST COND, SWE, TAIRHR(H), TA, !Input & TMIN, TYPPGL, TYPP GN, WINDHR(H), XHLAI, !Input & XLMAXT, YLMAXT, !Input & AGEFAC, EHR, LFMX SH, LFMXSL, PCNLSH, PCNLSL, !Output & PGHR, SLWSH, SLWS L, T0HR, TCAN(H), THR, TSHR, !Output & TSURF, !Output & CCNEFF, CICAD, CMXSF, CQ ESF, PGPATH) !Input

PAGE 253

239 MODULE: ETPHR.FOR Add photosynthetic pathway parameters to SUBROUTINE ETPHR statement SUBROUTINE ETPHR( & CANHT, CEC, CEN, CLOUDS, CO2HR, DAYTIM, !Input & DLAYR2, DULE, FNPGL, FNPG N, FRACSH, FRSHV, !Input & KDIRBL, LAISH, LAISHV, LA ISL, LAISLV, LLE, !Input & LMXREF, LNREF, LWIDTH, MEEVP, MEPHO, NLAYR, !Input & NSLOPE, PARSH, PARSUN, QE REF, RABS, RCUTIC, !Input & REFHT, RHUMHR, RNITP, RWUH, SHCAP, SLAAD, !Input & SLWREF, SLWSLO, STCOND, SW E, TAIRHR, TA, !Input & TMIN, TYPPGL, TYPPGN, WINDHR XHLAI, !Input & XLMAXT, YLMAXT, !Input & AGEFAC, EHR, LFMXSH, LFMX SL, PCNLSH, PCNLSL, !Output & PGHR, SLWSH, SLWSL, T0HR, TCAN, THR, TSHR, !Output & TSURF, !Output & CCNEFF, CICAD, CMXS F, CQESF, PGPATH) !Input Declare photosynthetic pathway parameter variables in ETPHR CHARACTER PGPATH*2 REAL CCNEFF, CICA D, CMXSF, CQESF Add photosynthetic pathway parame ters to CALL CANOPG statement CALL CANOPG( & CO2HR, FNPGL, FNPGN, LAISH, LAISL, LMXREF, !Input & LNREF, NSLOPE PARSH, PARSUN, QEREF, RNITP, !Input & SLAAD, SLWSLO TMIN, TSURF, TYPPGL, TYPPGN, !Input & XLMAXT, YLMAXT, !Input & AGEFAC, CONDSH, CONDSL, CSHSTR, CSLSTR, !Output & LFMXSH, LFMXSL, PCNLSH, PCNLSL, PGHR, !Output & SLWREF, SLWSH, SLWSL, STRESS, !Output & CCNEFF, CICAD, CMXSF, CQESF, PGPATH) !Input Add photosynthetic pathway parameters to SUBROUTINE CANOPG statement SUBROUTINE CANOPG( & CO2HR, FNPGL, FNPGN, LAIS H, LAISL, LMXREF, !Input & LNREF, NSLOPE, PARSH, PA RSUN, QEREF, RNITP, !Input & SLAAD, SLWSLO, TMIN, TSUR F, TYPPGL, TYPPGN, !Input & XLMAXT, YLMAXT, !Input & AGEFAC, CONDSH, CONDSL, CSHS TR, CSLSTR, !Output & LFMXSH, LFMXSL, PCNLSH, PCNL SL, PGHR, !Output & SLWREF, SLWSH, SLWSL, STRESS, !Output & CCNEFF, CICAD, CMXS F, CQESF, PGPATH) !Input Declare photosynthetic pathway parameter variables in CANOPG

PAGE 254

240 CHARACTER PGPATH*2 REAL CCNEFF, CICA D, CMXSF, CQESF Add photosynthetic pathway parame ters to CALL PGLFEQ statement CALL PGLFEQ( & CO2HR, FNPGL, FNPGN, LMXR EF, LNREF, QEREF, !Input & PCNLSH, SLWSH, SLWREF, TEMPSH, TMIN, TYPPGL, !Input & TYPPGN, XLMAXT, YLMAXT, !Input & AGMXSH, LFMXSH, QEFFSH, !Output & CCNEFF, CICAD, CMXS F, CQESF, PGPATH) !Input CALL PGLFEQ( & CO2HR, FNPGL, FNPGN, LMXR EF, LNREF, QEREF, !Input & PCNLSL, SLWSL, SLWREF, TEMPSL, TMIN, TYPPGL, !Input & TYPPGN, XLMAXT, YLMAXT, !Input & AGMXSL, LFMXSL, QEFFSL, !Output & CCNEFF, CICAD, CMXS F, CQESF, PGPATH) !Input Add photosynthetic pathway parameters to CALL PGLEAF statement CALL PGLEAF( & CO2HR, LFMXSL, PARSUN(I), QEFFSL, TEMPSL, !Input & CONSUN, PGSUN, !Output & CCNEFF, CICAD, PGPATH) !Input IF (I .EQ. 2) THEN PGSUM = PGSUM + PGSUN*1.6 CONSUM = CONSUM + CONSUN*1.6 ELSE PGSUM = PGSUM + PGSUN CONSUM = CONSUM + CONSUN ENDIF ENDDO PGSL = PGSUM / 3.6 CONDSL = CONSUM / 3.6 C Compute photosynthesis and leaf CO2 conductance for shaded leaves CALL PGLEAF( & CO2HR, LFMXSH, PARSH, QEFFS H, TEMPSH, !Input & CONDSH, PGSH, !Output & CCNEFF, CICAD, PGPATH) !Input Add photosynthetic pathway parameters to SUBROUTINE PGLFEQ statement SUBROUTINE PGLFEQ( & CO2HR, FNPGL, FNPGN, LMXR EF, LNREF, QEREF, !Input

PAGE 255

241 & RNITP, SLW, SLWREF, TEMP HR, TMIN, TYPPGL, !Input & TYPPGN, XLMAXT, YLMAXT, !Input & AGEMXL, LFMAX, QEFF, !Output & CCNEFF, CICAD, CMXS F, CQESF, PGPATH) !Input IMPLICIT NONE Declare photosynthetic pathway parameter variables in PGLFEQ CHARACTER PGPATH*2 REAL CCNEFF, CICA D, CMXSF, CQESF Adjust TAU for CO2 concentrating effect in C4’s IF (RT .GT. 1000.) THEN IF (PGPATH .EQ. "C3" .OR. PGPATH .EQ. "c3") THEN TAU = EXP(-3.949 + 28990.0/RT) ELSE TAU = EXP(-3.949 + 28990.0/RT) CCNEFF ENDIF GAMST = 0.5 O2 / TAU ELSE TAU = 1E10 GAMST = 0.0 ENDIF Rescale CO2MAX for C4’s, including changi ng CICA to 0.4 (set in species file). IF (PGPATH .EQ. "C3" .OR. PGPATH .EQ. "c3") THEN CICA = 0.7 CINT = CICA*CO2HR + (1.0-CICA)*GAMST CINT = MAX(CINT,GAMST) CO2MAX = 7.179 (CINT-GAMST) / (4.0*CINT+8.0*GAMST) ELSE CICA = CICAD CINT = CICA*CO2HR + (1.0-CICA)*GAMST CINT = MAX(CINT,GAMST) CO2MAX = CMXSF (CINT-GAMST) / (4.0*CINT+8.0*GAMST) ENDIF Rescale CO2QE for C4’s, including changi ng CICA to 0.4 (set in species file). CINT = MAX(CO2HR,GAMST) IF (PGPATH .EQ. "C3" .OR. PGPATH .EQ. "c3") THEN CO2QE = 6.225 (CINT-GAMST) / (4.*CINT+8.*GAMST) ELSE CO2QE = CQESF (CIN T-GAMST) / (4.*CINT+8.*GAMST) ENDIF Add photosynthetic pathway parameters to SUBROUTINE PGLEAF statement

PAGE 256

242 SUBROUTINE PGLEAF( & CO2HR, LFMAX, PARLF, QEFF, TEMPHR, !Input & CONDLF, PGLF, !Output & CCNEFF, CICAD, PGPATH) !Input Declare photosynthetic pathway parameter variables in PGLEAF CHARACTER PGPATH*2 REAL CCNEFF, CICAD Adjust TAU for CO 2 concentrating effect in C4’s IF (RT .GT. 1000.) THEN IF (PGPATH .EQ. "C3" .OR. PGPATH .EQ. "c3") THEN TAU = EXP(-3.9489 + 28990.0/RT) ELSE TAU = EXP(-3.9489 + 28990.0/RT) CCNEFF ENDIF GAMST = 1.0E6 0.5 0.21 / TAU ELSE TAU = 1E10 GAMST = 0.0 ENDIF Adjust CICA to 0.4 (set in speci es file) for other than C3’s. C CICA = 0.4+0.6*EXP(-0.002*CO2HR) IF (PGPATH .EQ. "C3" .OR. PGPATH .EQ. "c3") THEN CICA = 0.7 ELSE CICA=CICAD ENDIF CINT = CICA*CO2HR + (1.0-CICA)*GAMST CCO2LF = MAX(PNLF/(CO2HR-CINT),0.0)

PAGE 257

243 MODULE: FREEZE.FOR Add storage parameters to SUBROUTINE FREEZE statement SUBROUTINE FREEZE(FILEIO, RUN, & FREEZ1, FREEZ2, IDETO, NOUTDO, NRUSLF, SLDOT, !Input & TMIN, WTLF, YRDOY, YRPLT, !Input & MDATE, !Input/Output & WLFDOT, !Output & FRZDC, NRUSSR, NRU SST, PSRSRFL, PSRLYR1, !Input & SRFTEMP, SSDOT, SSR DOT, ST, STMWT, STRWT, !Input & PSRLYRD, PSRSRFD, WS FDOT, WSRFDOT, !Output & FRZDL, SRLYRD, SRSRFD VSTAGE) !Output Add USE ModuleDefs statement for access to array size of ST (soil temperature by layer) USE ModuleDefs !Definitio ns of constructed variable types, Declare storage variables REAL FRZDC, FRZDL, NRUSSR, NR USST, PSRSRFL, PSRLYR1, SRSRFD, & PSRLYRD, SRLYRD,SRFTEMP, SRSRFD, SSRDOT, SSDOT, STMWT, & STRWT, VSTAGE, WSFDOT, WSRFDOT REAL, DIMENSION(NL) :: ST Test for and calculate freeze-damage to ab ove and below ground portions of storage organ and sum for total storage organ damage Set MDate to 0.0 if total freeze kill of storage organ. Set storage organ freeze damage variable s to 0.0, part of abandoned strategy to allow partial to total kill of the storage or gan. Plant would live as long as there was storage organ left. May want to revisit this if modeling alfalfa. WSRFDOT = 0.0 PSRSRFD = 0.0 PSRLYRD = 0.0 Calculate freeze damage to above ground pl ant mass. Is a progressive function, the colder it gets, the more of the plan t dies. Reset VSTAGE accordingly. C-------------------------------------------------------------------C Calculate proportion of above -ground plant mass that will C be lost to frost. Function of number of degrees below threshold C temperature multiplied by the proportion lost per degree below C threshold. For total kill (like DSSAT35) set FRZDC=1.0

PAGE 258

244 C-------------------------------------------------------------------FRZDL = (FREEZ1 TMIN) FRZDC FRZDL = MAX(FRZDL,0.0) WLFDOT = (WTLF SLDOT NRUSLF/0.16) FRZDL WLFDOT = MIN ((WTLF SLDOT NRUSLF/0.16), WLFDOT) C-------------------------------------------------------------------C SJR 5/12/04 Moved VSTAGE adjustment to GROW C to be compatible with adjustment for senescence C-------------------------------------------------------------------! VSTAGE = ((WTLF-WLFDOT)/WTLF) VSTAGE WSFDOT = (STMWT SSDOT NRUSST/0.16) FRZDL WSFDOT = MIN ((STMWT SSDOT NRUSST/0.16), WSFDOT) If TMIN < FREEZ2, then kill the plant and end the simulation IF (TMIN .LE. FREEZ2) THEN IF (MDATE .LT. 0) THEN MDATE = YRDOY ENDIF WLFDOT = WTLF SLDOT NRUSLF/0.16 C-------------------------------------------------------------------C SJR 5/12/04 Moved VSTAGE adjustment to GROW C to be compatible with adjustment for senescence C-------------------------------------------------------------------! VSTAGE = 0.0 WSFDOT = STMWT SSDOT NRUSST/0.16 ENDIF List storage variables at end of subroutine FRZDC Freezing death coefficient pe rcentage tissue/population death per day per degree below FREEZ2 FRZDL Todays death loss of storag e tissue/plant population due to freezing (proportion of STRWT and PLNTPOP) NRUSSR N actually mobilized from st orage organ in a day (g[N]/m2-d) NRUSST N actually mobilized fro m stems in a day (g[N]/m2-d) PSRLYRD Proportion of total storage orga n tissue loss from below ground tissue PSRLYR1 Proportion of storage organ tissu e in soil layer 1 (below soil surface) PSRSRFD Proportion of tota l storage organ tissue loss from above ground tissue PSRSRFL Proportion of storage or gan tissue on/above soil surface SRFTEMP Temperature of soil surface (degrees C)

PAGE 259

245 SRLYRD Freeze damage to belo w ground storage organ tissue SRSRFD Freeze damage to a bove ground storage organ tissue SSDOT Stem loss due to daily senescence (g/m2/day) SSRDOT Storage organ loss due to daily senescence (g/m2/day) ST(I) Temperature in soil layer (I) (degrees C) STMWT Dry mass of stem tissue, in cluding C and N (g[stem] / m2[ground) VSTAGE Number of nodes on main stem of plant WSFDOT Stem weight losses due to freezing (g[stem]/m2-d) WSRFDOT Storage organ weight losse s due to freezing (g[storage]/m2-d)

PAGE 260

246 MODULE: GROW.FOR Include storage variables in SUBROUTINE GROW command SUBROUTINE GROW(CONTROL, ISWITCH, DYNAMIC, SOILPROP, & AGEFAC, CADLF, CADST, CRUSLF, CRUSRT, CRUSSH, !Input & CRUSST, DISLA, F, FILECC, FILEGC, FRLF, FRSTM, !Input & NADLF, NADRT, NADST, ND TH, NFIXN, NGRLF, NGRRT, !Input & NGRSD, NGRSH, NGRST, NM INEA, NODGR, NOUTDO, !Input & NPLTD, NRUSLF, NRUSRT, NRUSSH, NRUSST, POTCAR, !Input & POTLIP, PPLTD, SD IDOT, SDPROR, !Input & SENNOD, SENRT, SLDOT, SL NDOT, SRDOT, SSDOT, !Input & SSNDOT, TRNH4U, TRNO3U, TRNU, !Input & TURFAC, WLDOTN, WLIDOT WRDOTN, WRIDOT, WSDDTN, !Input & WSDOTN, WSHDTN, WSIDOT, WTABRT, WTSHMT, YRNR1, !Input & MDATE, YRPLT, !Input & SWIDOT, WLFDOT, WSHIDT, WTNFX, XHLAI, !Input/Output & AREALF, BETN, CANNAA, CA NWAA, CLW, CSW, DWNOD, !Output & DWNODA, GROWTH, GRWRES, LA IMX, PCCSD, PCLSD, !Output & PCNL, PCNRT, PCNSD, PCNSH, PCNST, PLTPOP, !Output & PLIGLF, PLIGNO, PLIGRT, PLIGSD, PLIGSH, PLIGST, !Output & PODWT, PUNCSD, PUNCTR, RHOL RHOS, RNITP, !Output & ROWSPC, RTWT, SDNPL, SDRA TE, SDWT, SDWTAM, !Output & SEEDNI, SEEDNO, SENESCE, SHEL WT, SLA, !Output & SLAAD, STMWT, TOPWT, TO TWT, WCRLF, WCRRT, WCRSH, !Output & WCRST, WNRLF, WNRRT, WNRSH, WNRST, WTCO, !Output & WTLF, WTLO, WTMAIN, WTNCAN, WTNEW, WTNLA, WTNLF, !Output & WTNLO, WTNNA, WTNNAG, WTN NO, WTNNOD, WTNOO, !Output & WTNRA, WTNRO, WTNRT, WTNS A, WTNSD, WTNSDA, !Output & WTNSDO, WTNSH, WTNSHA, WT NSHO, WTNSO, WTNST, !Output & WTNUP, WTRO, WTSDO, WTS HO, WTSO, XLAI, XPOD, !Output & CADRT, CADSH, NADSH, !Input & CADSR, CRUSSR, FRST R, NADSR, NGRSR, NRUSSR, !Input & PSRLYRD, PSRSRFD, PSRS RFL, PSRLYR1, SS RDOT, !Input & SSRNDOT, STRSRFL, ST RLYR1, WSRDOTN, WSRIDOT, !Input & WSFDOT, WSRFDOT, !Input/Output & CSRW, PCNSR, PLIGSR, RHOSR, STRWT, WCRSR, !Output & WNRSR, WTNSR, WTNS RA, WTNSRO, WTSRO, !Output & ALPHSR, PCARSR, PLIPSR PMINSR, POASR, PROSRF, !Output & CPFSTR, NSRALL, NSR DOT, NSROFF, TPSRLYR1, !Output & TPSRSRFL, WRCSRDT, WSRDOT, WSRI, !Output & VSTAGE) !Input/Output Declare new variables as REAL C--------------------------------------------------------------------

PAGE 261

247 C New storage variables for forage model C-------------------------------------------------------------------REAL CADSR, CRUSSR, FRST R, NADSR, NGRSR, NRUSSR, & SSRDOT, SSRNDOT, WSRDOTN, WSRIDOT, & WSRDOT, WSRFDOT, & CSRW, PCNSR, PLIGSR RHOSR, STRWT, WCRSR, & WNRSR, WTNSR, WT NSRA, WTNSRO, WTSRO, & CPFSTR, NSRDOT, ALPHSR, & PCARSR, PLIPSR, PM INSR, POASR, PROSRI, PROSRF, & WSRI, WRCSRDT, NSROFF, NSRALL, & PSRLYRD, PSRSRFD, PSRSRFL, PSRLYR1, STRSRFL, & STRLYR1, TPSRSRFL, TPSRLYR1 C-------------------------------------------------------------------C New shell and root variables for N accounting C-------------------------------------------------------------------REAL CADRT, CADSH, NADSH C-------------------------------------------------------------------C New stem freeze damage variables for forage model C-------------------------------------------------------------------REAL WSFDOT C-------------------------------------------------------------------C VSTAGE for adjustment due to senescence and freeze damage C-------------------------------------------------------------------REAL VSTAGE Add storage variables tot the CALL IPGRO statement CALL IPGROW( & FILEIO, FILECC, FILEGC, CROP, !Input & ALPHL, ALPHR, ALPHS, ALPHSH, !Output & PCARLF, PCARST, PCARRT, PCARSH PCARSD, PCARNO, !Output & PLIGLF, PLIGST, PLIGRT, PLIGSH PLIGSD, PLIGNO, !Output & PLIPLF, PLIPST, PLIPRT, PLIPSH, PLIPNO, !Output & PMINLF, PMINST, PMINRT, PMINSH PMINSD, PMINNO, !Output & POALF, POAST, POART, POAS H, POASD, POANO, !Output & PROLFF, PROSTF, PRORTF, PROSHF, PRONOD, !Output & PROLFI, PROSTI, PRORTI, !Output & PLTPOP, ROWSPC, RMIN, PLME, SDWTPL, !Output & SDLIP, SDPRO, WTFSD, WTPSD, XPODF, !Output & ALPHSR, PCARSR, PLIGSR, PLIPSR, PMINSR, POASR, !Output & PROSRF, PROSRI) !Output Calculate net respiration requirement for net storage organ tissue growth CPFSTR = 30./44.*(PLIPSR*1.720 + PLIGSR*0.659 + POASR*(-0.011)

PAGE 262

248 & + PMINSR*RMIN + PCARSR*0.170) Initialize storage variables CSRW= 0.0 !Cumulative storage organ growth NSRDOT = 0.0 PCNSR = 0.0 RHOSR = 0.0 STRWT = 0.0 WCRSR = 0.0 WNRSR = 0.0 WTNSR = 0.0 WTNSRA = 0.0 WTNSRO = 0.0 WTSRO = 0.0 WSFDOT = 0.0 WSRDOT = 0.0 WSRFDOT= 0.0 Calculate initial storage tissue weight for transplants TOPWT = WTLF + STMWT STRWT = FRSTR TOTWT RTWT = TOTWT TOPWT – STRWT WSRI = STRWT Initialize cumulative storage mass CSRW = STRWT Initialize storage tissue CH2O reserves WCRSR = ALPHSR STRWT Carbohydrate composition of storage tissue (fraction) RHOSR = ALPHSR Initialize net carbon addi tion for storage tissue WRCSRDT= 0.0 Compute N in storage tissue WTNSR = WSRTI PROSRI 0.16 WTNTOT = WTNLF + WTNS T + WTNRT + WTNSH + WTNSD + WTNSR Calculate seed or transplant N at planting SDNPL = WTPSD SDPRO 0.16 0.75 PLTPOP & (WTNLF + WTNST + WTNRT + WTNSR) SDNPL = MAX(SDNPL,0.0) SEEDNI = WTNLF + WTNST + WTNRT + SDNPL + WTNSR

PAGE 263

249 Initialize cumulative N variable for storage tissue WTNSRA = WTNSR Percent N in storage tissue PCNSR = WTNSR / WSRI 100 Begin daily integration steps Add storage dry weight growth rate to total plant growth rate GROWTH = WLDOTN + WS DOTN + WRDOTN + WSHDTN + WSDDTN + NODGR & + WSRDOTN GRWRES = WLDOTN*CPFLF + WSDOTN*CPFSTM + WRDOTN*CPFRT + & NODGR*CPFNOD + WSHDTN*CPFSH1 + WSDDTN*CPFSD1 + & WSRDOTN*CPFSTR + & 30./44*(TRNO3U/0.16 1.798 + TRNH4U/0.16 0.462 + & NFIXN/0.16 1.798) Calculate stem tissue net growth rate including freeze losses C-------------------------------------------------------------------C WSDOT = Net stem growth rate C-------------------------------------------------------------------WSDOT = WSDOTN SSDOT WSIDOT WSFDOT NRUSST / 0.16 CRUSST IF (STMWT .GT. 0.0) THEN WSDOT = WSDOT + (CADST+NADST/0.16) & (1. MIN(1.0, (SSDOT+WSIDOT+WSFDOT)/STMWT)) ENDIF IF (WSDOT .LT. 0.0) THEN WSDOT = MAX(WSDOT, -STMWT) ENDIF Calculate storage tissue net growth rate and new proportions of above and below ground storage organ tissue C-------------------------------------------------------------------C WSRDOT = Net storage tissue growth rate C-------------------------------------------------------------------WSRDOT = WSRDOTN SSRDOT WSRIDOT WSRFDOT NRUSSR / & 0.16 CRUSSR IF (STRWT .GT. 0.0) THEN WSRDOT = WSRDOT + (CADSR+NADSR/0.16) & (1. MIN(1.0,(SSRDOT+WSRIDOT+WSRFDOT)/STRWT)) ENDIF IF (WSRDOT .LT. 0.0) THEN WSRDOT = MAX(WSRDOT, -STRWT) ENDIF

PAGE 264

250 C-------------------------------------------------------------------C Allocate net growth to above and below ground storage organ tissue C using initial partitioning strategy from species file. C Allocate freeze losses according to the proporti ons calculated in FREEZE C Allocate all other losses usi ng current "actual" proportions. C If losses for one portion are greater than available mass C the excess loss is taken from the other portion C NOTE: using TPSRLYR1 and TPSRSRFL to hold intermediate values for C themselves before calcu lating actual new proportions. C-------------------------------------------------------------------TPSRLYR1 = (WSRDOTN STRLYR1) ( SSRDOT + WSRIDOT + NRUSSR / .16 & + CRUSSR) PSRLYR1 WSRFDOT PSRLYRD TPSRSRFL = (WSRDOTN STRSRFL) ( SSRDOT + WSRIDOT + NRUSSR / .16 & + CRUSSR) PSRSRFL WSRFDOT PSRSRFD IF (STRWT .GT. 0.0) THEN WSRDOT = WSRDOT + (CADSR+NADSR/0.16) & (1. MIN(1.0,(SSRDOT+WSRIDOT+WSRFDOT)/STRWT)) TPSRLYR1 = TPSRLYR1 + (CADSR+NADSR/0.16) & (1. MIN(1.0,(SSRDO T+WSRIDOT+WSRFDOT)/STRWT))*STRLYR1 TPSRSRFL = TPSRSRFL + (CADSR+NADSR/0.16) & (1. MIN(1.0,(SSRDO T+WSRIDOT+WSRFDOT)/STRWT))*STRSRFL C-------------------------------------------------------------------C Check to make sure not losing more mass from above or below ground C portions than exist in each. If so, allocate excess loss to the C other portion. If total loss is more than storage organ mass, C limit loss to actual mass of storage organ present C NOTE: must be done even when net gain is positive because freeze C damage could result in a net lo ss of above ground tissue and a net C gain of total and below ground tissue. C-------------------------------------------------------------------IF (-TPSRSRFL .GE. STRWT*PSRSRFL) THEN TPSRLYR1 = TPSRLYR1 + (TPSRSRFL+(STRWT*PSRSRFL))

PAGE 265

251 TPSRSRFL = -STRWT PSRSRFL IF (-TPSRLYR1 .GE. STRWT*PSRLYR1) THEN TPSRLYR1 = -STRWT PSRLYR1 TPSRSRFL = -STRWT PSRSRFL ENDIF ENDIF IF (-TPSRLYR1 .GE. STRWT*PSRLYR1) THEN TPSRSRFL = TPSRSRFL + (TPSRLYR1+(STRWT*PSRLYR1)) TPSRLYR1 = -STRWT PSRLYR1 IF (-TPSRSRFL .GE. STRWT*PSRSRFL) THEN TPSRLYR1 = -STRWT PSRLYR1 TPSRSRFL = -STRWT PSRSRFL ENDIF ENDIF ENDIF IF (WSRDOT .LT. 0.0) THEN WSRDOT = MAX(WSRDOT, -STRWT) ENDIF C-------------------------------------------------------------------C Calculate proportions of be low and above ground storage organ C tissue at end of day with new growth and losses. C Still need current days propor tions for allocating senesced C nutrients to appropriate soil layers for CENTURY C-------------------------------------------------------------------TPSRLYR1 = (TPSRLYR1 + (STR WT*PSRLYR1))/(STRWT+ WSRDOT) TPSRSRFL = (TPSRSRFL + (STR WT*PSRSRFL))/(STRWT+ WSRDOT) Calculate root tissue net growth rate – including returning N/CH2O from excess N mobilization C-------------------------------------------------------------------C Net root growth rate C-------------------------------------------------------------------WRDOT = WRDOTN SRDOT NRUSRT/0.16 CRUSRT WRIDOT IF (RTWT .GT. 0.0) THEN WRDOT = WRDO T + (CADRT + NADRT/0.16) & (1. MIN(1.0,(SRDOT+WRIDOT)/RTWT)) ENDIF IF (WRDOT .LT. 0.0) THEN WRDOT = MAX(WRDOT, -RTWT)

PAGE 266

252 ENDIF Calculate shell tissue net growth rate – including returning N/CH2O from excess n mobilization C-------------------------------------------------------------------C Net shell growth rate C-------------------------------------------------------------------WSHIDT = MIN(WSHIDT,SHELWT) pest damage to shells WSHDOT = WSHDTN WSHIDT WTABRT NRUSSH / 0.16 CRUSSH IF (SHELWT .GT. 0.0) THEN WSHDOT = WS HDOT + (CADSH + NADSH/0.16) & (1. MIN(1.0,WTABRT/RTWT)) ENDIF Calculate total net plant growth rate WDOT = WLDOT + WSDOT + WRDOT + WPDOT + WNDOT + WSRDOT Reduce VSTAGE by proportion of WTLF lost due to freezing C-------------------------------------------------------------------C Integration, Add Today's Net Growth to Existing Weights C-------------------------------------------------------------------C-------------------------------------------------------------------C SJR 5/12/04 reduce VSTAGE for forage model C Lower VSTAGE for leaf lost to freezing C-------------------------------------------------------------------IF ((SLDOT + WLDOT) .GE. WTLF) THEN VSTAGE=0.0 ELSE C SJR 5/19/04 drop leaf loss due to senescence per discussion C with KJB decided that it was a bad idea VSTAGE = ((WTLF-(SLDOT+WLFDOT))/WTLF) VSTAGE VSTAGE = ((WTLF-WLFDOT)/WTLF) VSTAGE ENDIF Integrateadd today’s storage tissue growth to existing mass STRWT = STRWT + WSRDOT Calculate cumulative storage tissue growth CSRW= CSR + WSRDOTN Carbon Reserve accounting – NOTE: Origin ally noticed that WLDOTN was used in calculating WRCLDT but WSDOT, WSHD OT, and WRDOT used for calculating Net C addition to other organs – changed code to use WSDOTN, WSHDTN, and WRDOTN and WSRDOTN changed back after discussions with Dr. Boote (8/7/03) – Plants did not recover from freeze (70 kg N ha-1 ) switched back to using

PAGE 267

253 the “N” values. Also changed SLNDOT, SSNDOT, and SSRNDOT (senescence due to water stress) to SLDOT (daily senescence – includes the “N” senescence). C-------------------------------------------------------------------C Carbon Reserves: Net Growth Rates for Mobile Carbohydrates C-------------------------------------------------------------------C Account for N added to exis ting plant tissue, e.g., NADLF, that C is damaged by insects, freezi ng, or senesced. Otherwise, could C get increase in tissue N composition when tissue is aborted. Need C to account for mass, N and C lost this way in sections below C Original code commented out C-------------------------------------------------------------------! WRCLDT = ALPHL WLDOTN CRUSLF RHOL*(SLNDOT+WLIDOT+WLFDOT) WRCLDT = AL PHL WLDOTN CRUSLF RHOL*(SLDOT+WLIDOT+WLFDOT) IF (WTLF .GT. 0.0) THEN WRCLDT = WRCLDT + CADLF & (1. MIN(1.0,(SLDOT+WLIDOT+WLFDOT)/WTLF)) ENDIF C-------------------------------------------------------------------C Original code commented out C Appears to be double accounting fo r C in new C reserves in stem C Changed WSDOT to WSDOTN to fix this/be consistent with leaf code C C-------------------------------------------------------------------! WRCSDT = ALPHS WSDOT CRUSST RHOS SSNDOT WRCSDT = ALPHS WSDOTN CRUSST RHOS (SSDOT+WSFDOT+WSIDOT) IF (STMWT .GT. 0.0) THEN WRCSDT = WRCSDT + CADST & (1. MIN(1.0,(SSDOT+WSIDOT+WSFDOT)/STMWT)) ENDIF WRCSRDT = ALPHSR WSR DOT CRUSSR RHOSR (SSRNDOT & +WSRIDOT + WSRFDOT) WRCSRDT = ALPHSR WS RDOTN CRUSSR RHOSR (SSRDOT & +WSRIDOT + WSRFDOT) IF (STRWT .GT. 0.0) THEN WRCSRDT = WRCSRDT + CADSR & (1. MIN(1.0,(SSRDOT+WSRIDOT+WSRFDOT)/STRWT)) ENDIF WRCRDT = ALPHR WRDOT CRUSRT RHOR SRDOT

PAGE 268

254 WRCRDT = ALPHR WRDOTN CRUSRT RHOR (SRDOT + WRIDOT) WRCSHD = ALPHSH WSHDOT CRUSSH RHOSH*(WTABRT+WTSHMT+WSHIDT) WRCSHD = ALPHSH WSHDTN CRUSSH RHOSH*(WTABRT+WTSHMT+WSHIDT) Update C storage in storage tissues WCRSR = WCRSR + WCRSRDT IF (WCRSR .LE. 1.0E-30) WCRSR = 0.0 Compute cumulative C lost from st em & storage tissue over the season WTSRO = WTSRO + SSRDOT + WSRIDOT + WSRFDOT WTSO = WTSO + SSDOT + WSIDOT + WSFDOT Compute CH2O fraction of storage tissue IF (STRWT .GT. 0.0001) THEN RHOSR = WCRSR/STRWT ELSE RHOSR = 0.0 ENDIF Change calculation of net growth rate of N in shell tissue to include excess mobilized N returned to shell. C-------------------------------------------------------------------C Net growth rate of nitrogen in shells C-------------------------------------------------------------------NSHDOT = NGRSH NSHOFF NRUSSH IF (SHELWT .GT. 0.0) THEN NSHDOT = NSHDOT + NADSH & (1. MIN(1.0,(WTABRT +WSHIDT) / SHELWT)) ENDIF Calculate stem tissue N balance C-------------------------------------------------------------------C Stem nitrogen senescence and pest damage loss C-------------------------------------------------------------------NSOFF = (SSNDOT+WSIDOT+WSFDOT) (PCNST/100.) + (SSDOT SSNDOT) & PROSTF 0.16 IF (NSOFF .LT. 0.0) THEN NSOFF = 0.0 ENDIF C--------------------------------------------------------------------

PAGE 269

255 C Net growth rate of nitrogen in the stems C-------------------------------------------------------------------NSDOT = NGRST NSOFF NRUSST IF (STMWT .GT. 0.0) THEN NSDOT = NSDOT + NADST & (1. MIN(1.0,(SSDOT+WSIDOT+WSFDOT) / STMWT)) ENDIF Calculate storage tissue N balance C-------------------------------------------------------------------C Storage tissue nitrogen senescence and pest damage loss C Using different strategy St olon not abscised like leaf C therefore, are assuming senes ced tissue lost at average C N concentration C-------------------------------------------------------------------! NSROFF = (SSRNDOT + WS RIDOT + WSRFDOT) (PCNSR/100.) + & (SSR DOT-SSRNDOT) PROSRF 0.16 NSROFF = (SSRDOT + WS RIDOT + WSRFDOT) (PCNSR/100.) IF (NSROFF. LT. 0.0) THEN NSROFF = 0.0 ENDIF C-------------------------------------------------------------------C Net growth rate of nitrogen in the storage tissues C-------------------------------------------------------------------NSRDOT = NGRSR NSROFF NRUSSR IF (STRWT .GT. 0.0) THEN NSRDOT = NSRDOT + NADSR & (1. MIN(1.0,(SSR DOT + WSRIDOT + WS RFDOT) / STRWT)) ENDIF C-------------------------------------------------------------------C Total nitrogen in the storage tissues C-------------------------------------------------------------------IF ((NSRDOT .LT. 0.0) AND. (ABS(NSRDOT) .GT. WTNSR)) THEN NSRDOT = WTNSR ENDIF WTNSR = WTNSR + NSRDOT C-------------------------------------------------------------------C Total nitrogen in the plant C-------------------------------------------------------------------WTNTOT = WTNLF + WTNST + WTNSR + WTNRT + WTNSH + WTNSD + WTNNOD Compute cumulative N added to stem tissues over the season

PAGE 270

256 Stem NSALL = NGRST NRUSST IF (STMWT .GT. 0.0) THEN NSALL = NSALL + NADST & (1. MIN(1.0,(SSDOT+WSIDOT+WSFDOT) / STMWT)) ENDIF Compute cumulative N added to storage tissues over the season Storage tissues NSRALL = NGRSR NRUSSR IF (STRWT .GT. 0.0) THEN NSRALL = NSRALL + NADSR & (1 MIN(1.0,(SSRDOT + WSRIDOT + WSRFDOT) & / STRWT)) ENDIF Compute cumulative N added to shell tissues over the season Shell and seed NSHALL = NGRSH NRUSSH C-------------------------------------------------------------------C Added these lines after adding NADSH variable C-------------------------------------------------------------------IF (SHELWT .GT. 0.0) THEN NSHALL = NSHALL + NADSH & (1. MIN(1.0,(WTABRT +WSHIDT) / SHELWT)) ENDIF Compute cumulative N added to storage tissues over the season WTNSRA = WTNSRA + NSRALL Compute cumulative N loss from storage tissues over the season WTNSRO = WTNSRO + NSROFF Compute percentage N in storage tissues IF ((STRWT .GT. 0.0) .AND. (STRWT .GT. WTNSR) .AND. & (WTNSR .GT. 0.0)) THEN PCNSR = WTNSR / STRWT 100.0 ELSE PCNSR = 0.0 ENDIF Calculate remaining N in storage tissues that can be mined IF ((STRWT WCRSR) .GT. 0.0) THEN WNRSR = MAX (WTNSR PROSRF 0.16 (STRWT-WCRSR), 0.0) ELSE

PAGE 271

257 WNRSR = 0.0 ENDIF Distribute senesced storage tissue C to soil surface and soil layer 1 C pools C-------------------------------------------------------------------C This section added to provi de senescence parameters to the C Soil N routines (senescence and freeze values are added to soil C residue; pest damage components are lost from the system) C-------------------------------------------------------------------C Surface carbon includes all senescence and freeze variables for C leaf (SLDOT, WLFDOT), stem ( SSDOT), storage organ (SSRDOT, WSRFDOT) C and shell (WTABRT). At this time, the model does not senesce seed. C Convert from biomass to C with a factor 0.40. C NOTE storage organ senescence is apportioned to both surface layer C and soil layer1 according to propor tions PSRSRFD and PSRLYRD from FREEZE C-------------------------------------------------------------------! SENCLN(0,1) = (SLDOT + WLFDOT + SSDOT + WTABRT) 0.40 SenWt(0) = (SLDOT + WLF DOT + SSDOT + WSFDOT + WTABRT + & SSRDOT *PSRSRFL) + WSRFDOT PSRSRFD C Convert from g/m2 to kg/ha with a factor of 10. SENCLN(0,1) = AMAX1(SENCLN(0,1), 0.) 10.0 !kg[C]/ha SenWt(0) = AMAX1(SenWt(0), 0.) 10.0 !kg[dry matter]/ha C-------------------------------------------------------------------C Surface nitrogen includes nitrogen losses computed above minus the C pest damage components. C-------------------------------------------------------------------! SENCLN(0,3) = (NLOFF WLIDOT PCNL / 100.) + SenE(0,1) = (NLOFF WL IDOT PCNL / 100.) + !Leaf & (NSOFF WSIDOT PCNST / 100.) + !Stem & (NSHOFFWSHIDT PCNSH / 100.) + !Shell & (NSROFF(WSR IDOT + WSRFDOT) PCNSR / 100.*PSRSRFL) & + WSRFDOT PCNSR / 100 PSRSRFD !Storage C Convert from g/m2 to kg/ha with a factor of 10. SENCLN(0,3) = AMAX1 (SENCLN(0,3),0.) 10.0 !kg[N]/ha SenE(0,1) = AMAX1 (SenE(0,1),0.) 10.0 !kg[N]/ha C-------------------------------------------------------------------C Contribution of lignin to surf ace litter from senesced and frozen C plant matter C-------------------------------------------------------------------! SENCLN(0,2) = (SLDOT + WL FDOT) PLIGLF + SSDOT PLIGST + SenLig(0) = (SLDOT + WLFDOT) PLIGLF + (SSDOT+WSFDOT) PLIGST +

PAGE 272

258 & WTABRT PLIGSH + (SSRDOT*PSRSRFL + WSRFDOT*PSRSRFD)* PLIGSR C Convert from g/m2 to kg/ha with a factor of 10. SENCLN(0,2) = AMAX1 (SENCLN(0, 2),0.) 10.0 !kg[lig]/ha SenLig(0) = AMAX1 (SenLig(0),0.) 10.0 !kg[lig]/ha C-------------------------------------------------------------------C Senescence of roots, nodules, a nd subsurface storage organs (kg/ha) C-------------------------------------------------------------------DO L = 1, NLAYR IF L=1 THEN SENCLN(L,1) = (SENRT (L) + SENNOD(L)) 0.40 !kg[C]/ha SenWt(L) = SENRT(L) + SENNOD(L)+ & (SSRDOT*PSRLYR1 + WSRFDOT*PSRLYRD) !kg[dry matter]/ha SENCLN(L,2) = SENRT(L) PL IGRT + SENNOD(L) PLIGNO !kg[lig]/ha SenLig(L) = SENRT(L) PLIGRT + SENNOD(L) PLIGNO + & (SSRDOT*PSRLYR1 + WSRFDOT*PSRLYRD)* PLIGSR !kg[lig]/ha SENCLN(L,3) = (SENRT(L)* PRORTF + SENNOD(L) PRONOD) 0.16 SenE(L,1) = (SENRT(L)* PRORTF + SENNOD(L) PRONOD) 0.16 + & (NSROFF(WSRIDOT + WSRFDOT) PCNSR / 100.*PSRLYR1) & + WSRFDOT PCNSR / 100 PSRLYRD ELSE SENCLN(L,1) = (SENRT(L ) + SENNOD(L)) 0.40 !kg[C]/ha SenWt(L) = (SENRT(L) + SENNO D(L)) !kg[dry matter]/ha SENCLN(L,2) = SENRT(L) PL IGRT + SENNOD(L) PLIGNO !kg[lig]/ha SenLig(L) = SENRT(L) PL IGRT + SENNOD(L) PLIGNO !kg[lig]/ha SENCLN(L,3) = (SENRT(L)* PRORTF + SENNOD(L) PRONOD) 0.16 SenE(L,1) = (SENRT(L)* PRORTF + SENNOD(L) PRONOD) 0.16 ENDIF ENDDO Update PSRSRFL and PSRLYR1 to ref lect today’s growth and losses C-------------------------------------------------------------------C Update distribution of stor age tissue above and below soil surface C-------------------------------------------------------------------PSRLYR1 = TPSRLYR1 PSRSRFL = TPSRSRFL Add line to prevent AREALF (and XLAI) from going below 0.0 AREALF = AREALF + ALFDOT C-------------------------------------------------------------------C New line for Forage model to prevent negative LAI's in winter

PAGE 273

259 C-------------------------------------------------------------------AREALF = MAX(AREALF, 0.0) Add condition that plant cannot die if there is storage tissue left C-------------------------------------------------------------------C Terminate growth if stress causes extremely low plant weights C-------------------------------------------------------------------IF (STRWT .LT. 0.00001) THEN IF (TOPWT .LT. 0.00001 .OR. STMWT .LT. 0.00001) THEN CALL STRESS(FILEIO, RUN, & AGEFAC, DWNOD, IDETO, IHARI, NOUTDO, PODWT, !Input & RTWT, SDWT, SHELWT, STMWT, TOPWT, !Input & TOTWT, TURF AC, WTLF, YRDOY, YRPLT, !Input & MDATE, !Output & STRWT) !Input RETURN ENDIF IF (IHARI .NE. 'R' .AND. IHARI .NE. 'D') THEN IF (RTWT .LT. 0. 00001 .OR. WTLF .LT. 0.00001) THEN CALL STRESS(FILEIO, RUN, & AGEFAC, DWNOD, ID ETO, IHARI, NOUTDO, PODWT, !Input & RTWT, SDWT, SHELWT, ST MWT, TOPWT, !Input & TOTWT, TURFAC, WTLF, YRDOY, YRPLT, !Input & MDATE, !Output & STRWT) !Input RETURN ENDIF ENDIF ENDIF Add STRWT to the CALL STRESS statement SUBROUTINE STRESS(FILEIO, RUN, & AGEFAC, DWNOD, IDETO, I HARI, NOUTDO, PODWT, !Input & RTWT, SDWT, SHELWT, STMWT, TOPWT, !Input & TOTWT, TURFAC, WTLF, YRDOY, YRPLT, !Input & MDATE, !Output & STRWT) !Input Add STRWT to the SUBROUTINE STRESS statement SUBROUTINE STRESS(FILEIO, RUN, & AGEFAC, DWNOD, IDETO, I HARI, NOUTDO, PODWT, !Input & RTWT, SDWT, SHELWT, STMWT, TOPWT, !Input & TOTWT, TURFAC, WTLF, YRDOY, YRPLT, !Input

PAGE 274

260 & MDATE, !Output & STRWT) !Input Declare STRWT REAL REAL STRWT Prevent reported STRWT from being below 0 STRWT = MAX(0.,STRWT) Add Storage parameters to the IPGROW SUBROUTINE statement SUBROUTINE IPGROW( & FILEIO, FILECC, FILEGC, CROP, !Input & ALPHL, ALPHR, ALPHS, ALPHSH, !Output & PCARLF, PCARST, PCARRT PCARSH, PCARSD, PCARNO, !Output & PLIGLF, PLIGST, PLIGRT PLIGSH, PLIGSD, PLIGNO, !Output & PLIPLF, PLIPST, PLIPRT, PL IPSH, PLIPNO, !Output & PMINLF, PMINST, PMINRT PMINSH, PMINSD, PMINNO, !Output & POALF, POAST, POART, POASH, POASD, POANO, !Output & PROLFF, PROSTF, PRORTF, PR OSHF, PRONOD, !Output & PROLFI, PROSTI, PRORTI, !Output & PLTPOP, ROWSPC, RMIN, PLME, SDWTPL, !Output & SDLIP, SDPRO, WTFSD, WTPSD, XPODF, !Output & ALPHSR, PCARSR, PLIGSR PLIPSR, PMINSR, POASR, !Output & PROSRF, PROSRI) !Output Declare storage parameters REAL REAL ALPHSR, PCARSR, PLIGSR PLIPSR, PMINSR, POASR, & PROSRF, PROSRI Read Storage tissue comp osition parameter values CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(F6.0, 6X, F6.0)',IOSTAT=ERR) & PROSRI, PROSRF IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LNUM) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(5F6.0)',IOSTAT=ERR) & PCARSR, PLIPSR, PLIGSR, POASR, PMINSR IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LNUM) Read ALPHSR value CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(5F6.0)',IOSTAT=ERR) & ALPHL, ALPHS, ALPHR, ALPHSH, ALPHSR IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LNUM) ENDIF

PAGE 275

261 List all storage and dormancy varia bles at end of GROW subroutine ALPHSR Fraction of new storage orga n growth that is mobile C (fraction) CADRT Mass of CH2O added back to root s for CH2O cost of excess mobilized N (never really mobilized) (g[CH2O] / m2 / d) CADSH Mass of CH2O added back to shells for CH2O cost of excess mobilized N (never really mobilized) (g[CH2O] / m2 / d) CADSR Mass of CH2O added to storage organs (g[CH2O] / m2 / d) CPFSTR Respiration requirement for net storage organ growth (g[CH20] / g[tissue]) CRUSSR C mobilized from storage or gans in a day (g[CH2O] / m2 / d) CSRW Cumulative storage growth (g[storage]/m2) FRSTR Fraction of ve getative tissue growth that goes to storage organs on a day (g[storage] / g[veg] )! NADSR N added to storage organ N reserves (g[N] / m2 / d) NADSH N added to shell N reserves (g[N] / m2 / d) NADSR N added to storage or gan N reserves (g[N] / m2 / d) NGRSR Maximum N demand for storage organ growth (g[stem N] / m2[ground] / d) NRUSSR N actually mobilized from storage organs in a day (g[N]/m2-d) NSRALL N added to storage organ today (g[N]/m2-d) NSRDOT Net N addition for stor age organ (g[N] / m2[ground] / d) NSROFF N loss from storag e organ in a day (g[N]/m2-d) PCARSR Proportion of storage ti ssue that is car bohydrate (fraction) PCNSR Percent N in storag e organ (100 g[N] / g[storage]) PLIGSR Proportion of storage or gan that is lignin (fraction) PLIPSR Proportion of storage ti ssue that is lipid (fraction) PMINSR Proportion of storage ti ssue that is mineral (fraction) POASR Proportion of storage tissue that is organic acid (fraction) PROSRF Minimum storage orga n protein composition after N mining (g[protein] / g[storage]) PROSRI Maximum protein compositi on in storage organ during growth with luxurious supply of N (g[protein] / g[storage]) PSRLYRD Proportion of total storage or gan tissue loss from below ground tissue PSRLYR1 Proportion of storage organ ti ssue in soil layer 1 (below soil surface) PSRSRFD Proportion of total storage or gan tissue loss from above ground tissue PSRSRFL Proportion of storag e organ tissue on/above soil surface RHOSR Fraction of storag e tissue which is carbohydrate (g [CH2O] / g[storage]) SSRDOT Daily senescence of st orage organ tissue (g / m2 / d) SSRNDOT Daily senescence of storage organ due to water stress (g/m2/day) STRLYR1 Proportion of storage or gan dry mass in soil layer 1 STRSRFL Proportion of storage organ dry mass on soil surface STRWT Dry mass of storage organ, including C and N (g[storage] / m2[ground) TPSRLYR1 Intermediate value us ed in updating value of PSRLYR1 TPSRSRFL Intermediate value us ed in updating value of PSRSRFL

PAGE 276

262 VSTAGE Number of nodes on main stem of plant WCRSR Mass of CH2O reserves in st orage organ (g[storage CH2O] / m2[ground]) WNRSR N available for mobiliza tion from storage organ above lower limit of mining (g[N] / m2) WRCSRDT Net C addition for st orage organ (g[CH2O] / m2 /d) WSFDOT Stem weight losse s due to freezing (g[stem]/m2-d) WSRI Initial weight of storage organ (g[storage] / m2) WSRDOT Net storage organ gro wth rate (g[storage] / m2 / d) WSRDOTN New storage tissue grow th today (g[storage] / m2 / d) WSRFDOT Storage organ weight losse s due to freezing (g[storage]/m2-d) WSRIDOT Weight of stor age organ consumed by pests today (g[storage]/m2-d) WTNSR Mass of N in storage organ (g[storage N] / m2[ground]) WTNSRA Cumulative N added to storage organ (g[N]/m2-d) WTNSRO Cumulative N loss fr om storage organ (g[N] / m2) WTSRO Cumulative storage organ losses (g[storage] / m2)

PAGE 277

263 MODULE: INCOMP.FOR Add storage terms to SUBROUTINE line and 2 CALL lines in CROPGRO SUBROUTINE INCOMP( & ECONO, FILECC, FILEGC, FRLF, FRRT, !Input & FRSTM, !Input & AGRLF, AGRNOD, AGRRT, AGRSD1, AGRSD 2, !Output & AGRSH1, AGRSH2, AGRSTM, AGRVG, AGR VG2, !Output & SDPROR, !Output & AGRSTR, FRSTR, !Output & DYNAMIC) Declare storage parameters as REAL REAL PROSRI, PLIPSR, PLIGSR POASR, PMINSR, PCARSR, & AGRSTR, AGRSR2, FRSTR Read in storage composition parameters C-------------------------------------------------------------------C Read New Storage Or gan Composition Parameters C-------------------------------------------------------------------CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(F6.0)',IOSTAT=ERR) & PROSRI IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) ENDIF CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(5F6.0)',IOSTAT=ERR) & PCARSR, PLIPSR, PLIGSR, POASR, PMINSR IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) Calculate AGRSTR AGRSTR = PLIPSR*RLIP + PLIGSR*RLIG + POASR*ROA & + PMINSR*RMIN + PCARSR*RCH2O Add AGRSTR to AGRVG and AGRVG2 AGRVG = AGRLF FRLF + AGRRT FRRT + AGRSTM FRSTM & + AGRSTR FRSTR AGRVG2 = AGRVG + (FRLF* PROLFI+FRRT*PRORTI+FRSTM*PROSTI & +FRSTR*PROSRI)*RNO3C Define storage parameters at end of subroutine. AGRSTR Mass of CH2O required for new storage tissue growth FRSTR Fraction of vegetative tissue growth that goes to storage on a day

PAGE 278

264 PCARSR Proportion of storag e tissue that is carbohydrate (fraction)","IPGROW, INCOMP PLIGSR Proportion of stor age tissue that is lignin (fraction)","IPGROW, INCOMP PLIPSR Proportion of stor age tissue that is lipid (fraction)","IPGROW, INCOMP PMINSR Proportion of storag e tissue that is mineral (fraction)","IPGROW, INCOMP POASR Proportion of storage tissue that is organic acid (fraction)","IPGROW, INCOMP PROSRI Maximum protein composition in storage during growth with luxurious supply of N (g[pro tein] / g[stem])","IPPLNT, IPDMND, IPGROW,

PAGE 279

265 MODULE: IPPLNT.FOR Add storage parameters to SUBROUTINE IPPLNT statement SUBROUTINE IPPLNT(CONTROL, & CADPR1, CMOBMX, CROP, DETA CH, ECONO, EORATIO, !Output & FILECC, FILEGC, FRCNOD FREEZ1, FREEZ2, KCAN, KEP,!Output & NOUTDO, PCARSH, PCH2O, PLIPSH, PLIGSD, PLIG SH, !Output & PMINSD, PMINSH, POASD, P OASH, PORMIN, PROLFI, !Output & PRORTI, PROSHI, PROSTI, R 30C2, RCH2O, RES30C, !Output & RFIXN, RLIG, RLIP, RMIN, R NH4C, RNO3C, ROA, !Output & RPRO, RWUEP1, RWUMX, TTFIX, !Output & PROSRI, STRSRFL, STRLYR1) !Output Declare storage parameter types REAL PROSRI, STRSRFL, STRLYR1 Read in value for PROSRI CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR,'(F6.0)',IOSTAT=ERR) PROSRI Read in values for STRSRFL and STRLYR1 C-------------------------------------------------------------------C C ***** READ STORAGE ORG AN PARTITIONING PARAMETERS ***** C C-------------------------------------------------------------------SECTION = '!*STOR' CALL FIND(LUN CRP, SECTION, LNUM, FOUND) IF (FOUND .EQ. 0) THEN CALL ERRO R(ERRKEY, 1, FILECC, LNUM) ELSE CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(2F6.0)',IOSTAT=ERR) STRSRFL, STRLYR1 IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LNUM) ENDIF Define parameters at end of subroutine. PROSRI Maximum protein composition in storage during growth with luxurious supply of N (g[protein] / g[stem])" STRLYR1 Initial proportion of storage organ dry mass in soil layer 1 STRSRFL Initial proportion of st orage organ dry mass on soil surface

PAGE 280

266 MODULE: MOBIL.FOR List Storage variables in th e SUBROUTINE MOBIL statement SUBROUTINE MOBIL( & NDMNEW, NMINEP, NMOBR, RP RO, TRNU, !Input & WNRLF, WNRRT, WNRSH, WNRST, !Input & NMINEA, NRUSLF, NRUSRT, NRUSSH, NRUSST, !Output & NMOBSR, PPMFAC, WNRSR, !Input & NRUSSR, PNMLF, PNMST, PNMRT, PNMSR, PNMSH, !Output & DYNAMIC) !Control Declare variables REAL REAL NRUSSR, WNRSR, NMOBSR REAL NRUSTOT, PNMLF, PNMST, PNMRT, PNMSR, PNMSH Initialize NRUSSR NRUSSR = 0.0 PNMLF=0.0 PNMST=0.0 PNMRT=0.0 PNMSR=0.0 PNMSH=0.0 Adjust NRUSRT for dormancy and calculate NRUSSR NRUSRT = NMINER WNRRT (1-PPMFAC) NRUSSR = NMINEA WNRSR NMOBSR / NMINEP Add code to calculate proportion of mobi lized N coming from each organ. Elected to calculate proportion from shell by difference so sum would always be 1.0 and shell will be 0 or small for forages. NRUSTOT = NRUSLF+NRUSST+NRUSSH+NRUSRT+NRUSSR IF (NRUSTOT .GT. 0.0) THEN PNMLF=NRUSLF/NRUSTOT PNMST=NRUSST/NRUSTOT PNMRT=NRUSRT/NRUSTOT PNMSR=NRUSSR/NRUSTOT PNMSH=1.0-(PNMLF+PNMST+PNMRT+PNMSR) ELSE PNMLF=0.0 PNMST=0.0 PNMRT=0.0 PNMSR=0.0 PNMSH=0.0 ENDIF List storage variables at end of subroutine NMOBSR Stage dependent N mining rate for storage organ

PAGE 281

267 NRUSSR N actually mobilized from storage organ in a day (g[N]/m2-d) NRUSTOT Total N actually mobilized in a day (g[N]/m2-d) PPMFAC Reduction in mobilization from storage organ due to photoperiod induced dormancy PNMLF Proportion of actually mobilized N mobilized from leaves in a day PNMST Proportion of actually mobilized N mobilized from stems in a day PNMRT Proportion of actually mobilized N mobilized from roots in a day PNMSR Proportion of actually mobilized N mobilized from storage organ in a day PNMSH Proportion of actually mobilized N mobilized from shells in a day WNRSR N available for mobilizati on from storage organ above lower limit of mining (g[N] / m2)

PAGE 282

268 MODULE: NUPTAK.FOR Add code to prevent cost of N uptake and reduction from exceeding PGAVL Bring PGAVL, RNH4C, and RNO3C into NUPTAK SUBROUTINE NUPTAK( & BD, DLAYR, DUL, FILECC, LL, NDMSDR, NDMTOT, NH4, !Input & NO3, NLAYR, PGAVL, RLV, RNH4C, RNO3C, SAT, SW, !Input & NUPNH4, NUPNO3, TRNH4U, TRNO3U, TRNU, UNH4, UNO3, !Output & DYNAMIC) !Control Declare new variables as REAL REAL PGAVL, PRSPNH4, PRSPNO 3, PTRNH4U, PTRNO3U, RNH4C, RNO3C REAL NUPNH4(NL), NUPNO3(NL) Initialize new totalizing variables in INTEGR step PTRNO3U = 0.0 PTRNH4U = 0.0 Calculate new variables – tota l potential NO3 & NH4 uptake C-------------------------------------------------------------------C SJR 10/17/03 Calculate potential total NO3 and NH4 uptake C-------------------------------------------------------------------PTRNO3U = PTRNO3U + RNO3U(L) PTRNH4U = PTRNH4U + RNH4U(L) Calculate cost for uptake and reduction of total potential NO3 and NH4 uptake C-------------------------------------------------------------------C SJR 10/17/03 Calculate cost of uptake and reduction of CP C from potential NO3 and NH4 C-------------------------------------------------------------------PRSPNO3 = (PTRNO3U/10)/0.16 RNO3C PRSPNH4 = (PTRNH4U/10)/0.16 RNH4C Check cost of uptake of potential NO3 and NH4 against PGAVL to set the upper limit of N uptake – prevents RSPNO3 + RSPNH4 from exceeding PGAVL in CROPGRO.FOR C-------------------------------------------------------------------C SJR 10/17/03 Check that co st of uptake and reduction of CP C from TRNU does not exceed PGAVL C-------------------------------------------------------------------IF (PGAVL .LT. (PRSPNO3 + PRSPNH4)) THEN TRNU = (PGAVL / (PRSPNO3+PRSPNH4)) TRNU ENDIF

PAGE 283

269 Create new DO loop at end of actual uptake loop to calc ulate the proportion of NO3 and NH4 taken up from each soil layer – used in VEGGR to put NLEAK back. C-------------------------------------------------------------------C Calculate proportion of TRNU coming from NO3 and NH4 in each layer C for use in "returning" NLEAK C-------------------------------------------------------------------DO L=1,NLAYR IF (TRNO3U .GT. 0.0) THEN NUPNO3(L) = UNO3(L) / TRNO3U ENDIF IF (TRNH4U .GT. 0.0) THEN NUPNH4(L) = UNH4(L) / TRNO3U ENDIF ENDDO Define all new variables NUPNO3(L) Proportion of TRNU from a soil layer that is nitrate NUPNH4(L) Proportion of TRNU from a soil layer that is ammonium PGAVL Total available CH2O available for growth & respiration (g[CH2O] / m2) PRSPNO3 Respiration cost to fix a nd reduce all potentially available NO3 g CH2O/M2 to CP PRSPNH4 Respiration cost to fix a nd reduce all potentially available NH4 g CH2O/M2 to CP PTRNO3U Total potential NO3 uptake (kg [N] / ha) PTRNH4U Total potential NH4 uptake (kg [N] / ha) RNH4C CH2O required for protei n synthesis when source of N is ammonium uptake (g[CH2O] / g[protein]) RNO3C Respiration requi red for reducing NO3 to protein (g[CH2O] / g[protein])

PAGE 284

270 MODULE: OPGROW.FOR Add dormancy and storage variables to SUBROUTINE OPGROW statement SUBROUTINE OP GROW(CONTROL, ISWITCH, & CADLF, CADST, CA NHT, CANWH, CMINEA, DWNOD, & GROWTH, GRWRES, MAINR, MDATE, NFIXN, NSTRES, & PCLSD, PCCSD, PCNL PCNRT, PCNSD, PCNSH, PCNST, & PG, PODNO, PODWT, PODWTD, RHOL, RHOS, RLV, RSTAGE, & RTDEP, RTWT, SATFAC, SDWT, SEEDNO, SLA, STMWT, SWFAC, & TGRO, TGROAV, TOPW T, TOTWT, TURFAC VSTAGE, WTCO, & WTLF, WTLO, WTNCAN WTNLF, WTNST, WTNSD, WTNUP, & WTNFX, WTSO, XLAI, YRPLT, & CADSR, PCNSR, PSRSRFD, PSRSRFL, RHOSR, STRWT, & WTNSR, WTSRO, & DRMST, PPGFAC, PPMFAC PPTFAC, SRFTEMP, ST(1), FREEZ2, & AGRSTR, CADSR, CMOBSR CPFSTR, CRUSSR, CSRFRZ, CSRW, & CSTRM, DSTOR, FNINSR, FNINSRG, FRSTR, FRSTRM, NADSR, & NGRSR, NGRSG, NMOBSR, NRUSSR, NSRALL, NSRDOT, NSROFF, & NVRSTR, PCNSR, PCST RD, PROSRT, PSRSRFD, PSRLYRD, & PSRSRFL, PSRLYR1, RHOSR, SRDAM, SRSRFD, SRLYRD, SSRDOT, & SSRNDOT, STRWT, TPSRSR FL, TPSRLYR1, WCRSR, WNRSR, & WCRSRDT, WSRDOT, WSR DOTN, WSRFDOT, WSRI, WSRIDOT, & WTNSR, WTNSRA WTNSRO, WTSRO, XSTR, & PROSRF, PROSRG, PROSRI, PCARSR, PLIGSR, & PLIPSR, POASR, PMINSR, ALPHSR, CMOBSRX, CADSRF, NMOBSRX, CLAIT, & YSTOR, FRSTRF, FRSTRM X, STRSRFL, STRLYR1, SENSR, & FNPTD, TYPPTD, FNPMD, TYPPMD, FNPGD, TYPPGD, HARD1, HARD2, & FRZDC, FRZHRD, TYPHRD, FRZDHD, TYPDHD, RDRMG, RDRMM, RDRMT) Declare dormancy and storage variables CHARACTER*6 DRMST REAL PPGFAC, PPMFAC, PPTFAC, SRFTEMP, ST(1), FREEZ2 REAL AGRSTR, CADSR, CMOBSR, CPFSTR, CRUSSR, CSRFRZ, CSRW, & CSTRM, DSTOR, FNINSR, FNINSRG, FRSTR, FRSTRM, NADSR, & NGRSR, NGRSG, NMOBSR, NRUSSR, NSRALL, NSRDOT, NSROFF, & NVRSTR, PCNSR, PCST RD, PROSRT, PSRSRFD, PSRLYRD, & PSRSRFL, PSRLYR1, RHOSR, SRDAM, SRSRFD, SRLYRD, SSRDOT, & SSRNDOT, STRWT, TPSRSR FL, TPSRLYR1, WCRSR, WNRSR, & WCRSRDT, WSRDOT, WSR DOTN, WSRFDOT, WSRI, WSRIDOT, & WTNSR, WTNSRA WTNSRO, WTSRO, XSTR

PAGE 285

271 REAL PROSRF, PROSRG, PROSRI, PCARSR, PLIGSR, & PLIPSR, POASR, PMINSR, ALPHSR, CMOBSRX, CADSRF, NMOBSRX, CLAIT, & YSTOR, FRSTRF, FRSTRM X, STRSRFL, STRLYR1, SENSR, & FNPTD, TYPPTD, FNPMD, TYPPMD, FNPGD, TYPPGD, HARD1, HARD2, & FRZDC, FRZHRD, TYPHRD, FRZDHD, TYPDHD, RDRMG, RDRMM, RDRMT Add DORMANCY.OUT output file OUTDRM = 'Dormancy.OUT' CALL GETLUN('OUTDRM', NOUTDRM) Add STORAGE.OUT output file OUTSTR = 'Storage.OUT CALL GETLUN('OUTSTOR', NOUTSTOR) Add INSTORAGE.OUT output file OUTINSTR = 'StorSpIn.OUT' CALL GETLUN( 'OUTINSTR', NOUTINSTR) Write headers to new output files !------------------------------------------------------------------! Initialize daily Dormancy output file INQUIRE (FILE = OUTDRM, EXIST = FEXIST) IF (FEXIST) THEN OPEN (UNIT = NOUTDR M, FILE = OUTDRM, STATUS = 'OLD', & IOSTAT = ERRNUM, ACCESS = 'APPEND') FIRST = .FALSE. ELSE OPEN (UNIT = NOUTDR M, FILE = OUTDRM, STATUS = 'NEW', & IOSTAT = ERRNUM) WRITE(NOUTDRM,'( "*PLANT DORMANCY OUTPUT FILE")') FIRST = .TRUE. ENDIF !Write headers CALL HEADER(SEASI NIT, FILEIO, NOUTDRM, RUN) WRITE (NOUTDRM,270) 270 FORMAT('@YEAR DOY DAS DAP', & QDSD PPGF PPMF PPTF TSRD TS1D FRZ2') !------------------------------------------------------------------! Initialize daily Storage output file

PAGE 286

272 INQUIRE (FILE = OUTSTOR, EXIST = FEXIST) IF (FEXIST) THEN OPEN (UNIT = NOUTST OR, FILE = OUTSTOR, STATUS = 'OLD', & IOSTAT = ERRNUM, ACCESS = 'APPEND') FIRST = .FALSE. ELSE OPEN (UNIT = NOUTST OR, FILE = OUTSTOR, STATUS = 'NEW', & IOSTAT = ERRNUM) WRITE(NOUTSTOR,' ("*PLANT STORAGE OUTPUT FILE")') FIRST = .TRUE. ENDIF !Write headers CALL HEADER(SEASI NIT, FILEIO, NOUTSTOR, RUN) WRITE (NOUTSTOR,280) 280 FORMAT('@YEAR DOY DAS DAP TWAD PHAD', QCQD QHAD QC%M QRAD QMAD QCFD QCAD QCDD QDTD', QN%X QN%I QV%D QV%T QNAA QNRX QNRN QNAR QNAM', QNAG QNAN QNAL QN%N QN%D QWAC QP%W QL%S QL%1', QS%D Q1%D QC%D QCAM QFDS QFD1 QEAD QEWD QWAD', QT%S QT%1 QCRD QNMD QCAG QWNG QWND QFAD QWAI', QMAM QNAD QNA C QNLC QDAD XSTR') !------------------------------------------------------------------! Initialize St orage inputs output file INQUIRE (FILE = OUTINSTOR, EXIST = FEXIST) IF (FEXIST) THEN OPEN (UNIT = NOUTINSTR, FILE = OUTSTOR, STATUS = 'OLD', & IOSTAT = ERRNUM, ACCESS = 'APPEND') FIRST = .FALSE. ELSE OPEN (UNIT = NOUTINSTR, FILE = OUTSTOR, STATUS = 'NEW', & IOSTAT = ERRNUM) WRITE(NOUTINSTR,'("* PLANT STORAGE INPUTS OUTPUT FILE")') FIRST = .TRUE. ENDIF !Write headers CALL HEADER(SEASI NIT, FILEIO, NOUTINSTR, RUN)

PAGE 287

273 WRITE (NOUTINSTR,290) 290 FORMAT(' SP%F SP%G SP%I SRC% SRG% SRL% SRO% SRM%', & SR%L SC%M C%SR SMNX LAIT SP%1 SP%2 SP%3 SP%4', & SP%5 SP%6 SP%7 SP%8 FRSR FSRX SRSF SRL1 SNSR', & SRT1 SRT2 SRT3 SRT4 TSPT SRM1 SRM2 SRM3 SRM4', & TSPM SRG1 SRG2 SR G3 SRG4 TSPG HRD1 HRD2 FZDC', & ZHD1 ZHD2 ZHD3 ZH D4 THRD ZDH1 ZDH2 ZDH3 ZDH4', & TDHD RDMG RDMM RDMT') Write daily data to DORMANCY.OU T and STORAGE.OUT output files WRITE (NOUTDRM,610) YEAR, DOY, DAS, DAP, & DRMST, PPGF AC, PPMFAC, PPTFAC, SRFTEMP, ST(1), FREEZ2 610 FORMAT(1X,I4,1X,I3.3,2(1X,I5),A6,1X,F5.1,3(1X,F5.3), & 2(1X,F5.1)) WRITE (NOUTSTOR,710) YEAR, DOY, DAS, DAP, & AGRSTR, CADS R, CMOBSR, CPFSTR, CRUSSR, CSRFRZ, CSRW, & CSTRM, DSTOR, FNIN SR, FNINSRG, FRSTR, FRSTRM, NADSR, & NGRSR, NGRSG, NMOBSR, NRUSSR, NSRALL, NSRDOT, NSROFF, & NVRSTR, PCNSR, PCSTRD, PROSRT, PSRSRFD, PSRLYRD, & PSRSRFL, PSRLYR1 RHOSR, SRDAM, SRSRFD, SRLYRD, SSRDOT, & SSRNDOT, STRWT, T PSRSRFL, TPSRLYR1, WCRSR, WNRSR, & WCRSRDT, WSRDOT, WSRDOTN, WSRFDOT, WSRI, WSRIDOT, & WTNSR, WT NSRA, WTNSRO, WTSRO, XSTR 710 FORMAT(1X,I4,1X,I3.3,2(1X,I5),9(1X,F5.2),4(1X,F5.3), & 9(1X,F5.2),8(1X,F5.3),6(1X,F5.2), 2,(1X,F5.3), & 13(1X,F5.2)) Write data to INSTORAGE.OUT out put file at end of simulation WRITE (NOUTINSTR,810) PR OSRF, PROSRG, PROSRI, PCARSR, PLIGSR, & PLIPSR, POASR, PMINSR, ALPHSR, CMOBSRX, CADSRF, NMOBSRX, CLAIT, & YSTOR(1), YSTOR(2), YSTOR( 3), YSTOR(4), YSTO R(5), YSTOR(6), & YSTOR(7), YSTOR(8), FRSTRF, FRSTRMX, STRSRFL, STRLYR1, SENSR, & FNPTD(1), FNPTD(2), FNPT D(3),FNPTD(4), TYPPTD, FNPMD(1), & FNPMD(2), FNPMD(3), FNPMD( 4), TYPPMD, FNPG D(1), FNPGD(2), & FNPGD(3), FNPGD(4), TYPPGD, HARD1, HARD2, FRZDC, FRZHRD(1), & FRZHRD(2), FRZHRD(3), FRZH RD(4), TYPHRD, FRZDHD(1), FRZDHD(2), & FRZDHD(3), FRZDHD(4), TYPDHD, RDRMG, RDRMM, RDRMT) 810 FORMAT(8(1X,F5.3),15(1X,F5.2),3(1X,F5.3),4(1X,F5.1),3X,A3,

PAGE 288

274 & 4(1X,F5.1),3X,A3,4(1X,F5.1),3X,A3,2(1X,F5.1),1X,F5.3, & 4(1X,F5.1),3X,A3,4(1X,F5.1),3X,A3,3(1X,F5.3))

PAGE 289

275 MODULE: PEST.FOR Add storage parameters to SUBROUTINE PEST statement SUBROUTINE PEST(CONTROL, ISWITCH, & AREALF, CLW, CSW, LA GSD, LNGPEG, NR2, PGAVL, !Input & PHTIM, PLTPOP, RTWT, SLA, SLDOT, SOILPROP, !Input & SSDOT, STMWT, TOPWT, WLFDOT, WTLF, YRPLT, !Input & RLV, SDNO, SHELN, SWIDOT, !Input/Output & VSTAGE, WSHIDT, WTSD, WTSHE, !Input/Output & ASMDOT, DISLA, NPLTD, PPLTD, !Output & SDDES, WLI DOT, WRIDOT, WSIDOT,SDWT, !Output & CSRW, SSRDOT, STRW T, WSFDOT, WSRFDOT, !Input & WSRIDOT, !Output & CSFRZ, CSRFRZ, CSTR M, DSTOR, SRDAM) !Output Declare Stem freeze variable as REAL REAL CSFRZ, WSFDOT Declare storage variables as REAL REAL CSRW, CSTRM, SSRDOT, STRWT, WSRFDOT, WSRIDOT Add storage variables to the CALL VEGDM statement CALL VEGDM( & AREALF, CLW, CSW, PCLMA, PCLMT, !Input & PCSTMD, PDLA, PLFAD, PLFMD, PSTMD, !Input & PVSTGD, SLA, SLDOT, SSDOT, STMWT, !Input & TDLA, VSTGD, WLFDOT, WSTM D, WTLF, !Input & TLFAD, TLFMD, VSTAGE, WLIDOT, !Input/Output & CLAI, CLFM, CSTEM, DISLA, DISLAP, !Output & LAIDOT, WSIDOT, !Output & CSRW, PCSTRD, PSTR D, SSRDOT, STRWT, !Input & WSFDOT, WSRFDOT, WSTRD, !Input & CSTRM, WSRIDOT, !Output & CSRFRZ, DSTOR, SRDAM, !Output & SEASINIT) !Control List storage variables at end of subroutine CSFRZ Cumulative frozen stem tissue (g[stem]/m2) CSRW Cumulative storage organ growth (g[storage]/m2) CSTRM Cumulative storag e organ mass destroyed (g/m2) SSRDOT Daily senescence of storage organ (g / m2 / d) STRWT Dry mass of stor age organ tissue, including C and N (g[storage] / m2[ground)

PAGE 290

276 WSFDOT Stem weight losse s due to freezing (g[stem]/m2-d) WSRFDOT Storage organ weight losses due to freezing (g[storage]/m2-d) WSRIDOT Daily pest damage to storage organ mass (g/m2/day)

PAGE 291

277 MODULE: PESTCP.FOR Add storage variables to SU BROUTINE PESTCP statement SUBROUTINE PESTCP( & PCN, PCPID, PCTID, PDCF1, !Input & PL, PLTPOP, PNO, RTWT, SLA, STMWT, TOPWT, !Input & TSDNOL, TSDNOM, TSDNOS, TSDWTL, TSDWTM, TSDWTS, !Input & TSHNOL, TSHNOM, TSHNOS, TSHWTL, TSHWTM, TSHWTS, !Input & VSTAGE, WTLF, !Input & NSDDL, NSDDM, NSDDS, NS HDL, NSHDM, NSHDS, !Input/Output & PPLTD, TLFAD, TLFMD, TRTLV, !Input/Output & WRTMD, WSDDL, WSDDM, WSDDS, !Input/Output & WSHDL, WSHDM, WSHDS, !Input/Output & CPPLTD, NPLTD, PCLMA, PCLMT, !Output & PCSTMD, PDLA, PLFAD, PLFMD, PPSR, !Output & PRTLF, PRTLV, PRTMD, PSDDL, PSDDM, PSDDS, !Output & PSHDL, PSHDM, PSHDS, PSTMD, PVSTGD, !Output & TDLA, TPSR, TRTLF, VSTGD, WS TMD, !Output & PCSTRD, PSRMD, WSTRMD, !Output & DYNAMIC,WSDD,PSDD,PRLV) Declare storage variables REAL C Storage Variables REAL PCSTRD, PSRMD, WSTRMD Initialize storage variables C -Storage Variables -WSTRD = 0.0 PSTRD = 0.0 PCSTRD = 0.0 Change MOW option to not alter VSTAGE C-------------------------------------------------------------------C Mowing option C-------------------------------------------------------------------IF (INDEX( PC PID(I,J),'TOPWT') .GT. 0.0) THEN C-------------------------------------------------------------------C SJR 3/16/04 Divide PL(K) by 10 to allow MOW input as kg/ha C instead of old g/m2 C-------------------------------------------------------------------IF (TOPWT .G T. 0.0 .AND. TOPWT .GT. PL(K)/10.) THEN C------------------------------------------------------------------

PAGE 292

278 C Determine the fraction of the top weight that will be C removed by mowing. Include an IF statement to protect C against the co ndition that PL=0 (fileT with grazing or C mowing has a se quence of zero damage real damage C zero damage), would otherwise result in 100% damage. C-------------------------------------------------------------------IF (PL(K) .GT. 0.0) THEN DAM = (TOPWT PL(K)/10.) / TOPWT 100. PDCF1(I,J) ELSE DAM = 0.0 ENDIF ELSE DAM = 0.0 ENDIF C------------------------------------------------------------------C Set the top we ight to the desired value after mowing. C This also corrects the LAI in subroutine VEGDM. C-------------------------------------------------------------------PLFMD = PLFMD + DAM PSTMD = PSTMD + DAM PVSTGD = PVSTGD + DAM ENDIF Add new code to allow user to set number of leaves to remain after harvest – uses new PEST code MVS C-------------------------------------------------------------------C SJR 5/19/04 C MVS Leaf Number reduction option used in conjunction with MOW C-------------------------------------------------------------------IF (INDEX( PCPID(I,J),'NOLF') .GT. 0.0) THEN C-------------------------------------------------------------------C SJR 5/19/04 PL(K) input as numbe r of leaves left after mowing C-------------------------------------------------------------------IF (VSTAGE .GT. 0.0 .AND. VSTAGE .GT. PL(K)) THEN C------------------------------------------------------------------C Determine the fraction of leaves that will be C removed by mowing. Include an IF statement to protect C against the co ndition that PL=0 (fileT with grazing or C mowing has a se quence of zero damage real damage C zero damage), would otherwise result in 100% damage. C-------------------------------------------------------------------IF (PL(K) .GT. 0.0) THEN DAM = (VSTAGE PL(K)) / VSTAGE 100. PDCF1(I,J) ELSE DAM = 0.0

PAGE 293

279 ENDIF ELSE DAM = 0.0 ENDIF List storage variables at end of subroutine PCSTRD Observed cumulative pe rcentage storage organ mass damage (%) PSTRD Daily percent storage organ mass damage (%) WSTRD Daily absolute storage mass damage (g/m2/day)

PAGE 294

280 MODULE: PHENOL.FOR Relocate calculation of FT(2) in RATE st ep on emergence day to earlier in the subroutine. C-------------------------------------------------------------------C Transplants C SJR 5/26/04 moved statement to this location must appear before C calculation of today's FT(2). Otherwise sets DTX using ATEMP C instead of TGRO(I) on emergence day results in inaccurate C estimate of today's increase in VSTAGE C-------------------------------------------------------------------IF (PLME .EQ. 'T' .AND. YRPLT .EQ. YRDOY) THEN K = TSELC(2) FT(2) = CURV(CTMP(2) ,TB(K),TO1(K),TO2(K),TM(K),ATEMP) PHZACC(2) = FT(2) SDAGE ENDIF C-------------------------------------------------------------------C Compute dev rates for all other phases, using hourly air temp C-------------------------------------------------------------------DO J = 2,NPHS K = TSELC(J) FT(J) = 0.0 DO I = 1,24 FTHR = CURV(CTMP( J),TB(K),TO1(K),TO2(K),TM(K),TGRO(I)) FT(J) = FT(J) + FTHR/24. ENDDO IF (DAS .LT. NR1) THEN FUDAY(J) = CURV(DLTYP(J),1.0,CSDVAR,CLDVAR,THVAR,DAYL) ELSE FUDAY(J) = CURV (DLTYP(J),1.0,CSDVRR,CLDVRR,THVAR,DAYL) ENDIF FSW(J) = 1. + (1. SWFAC) WSENP(J) FNSTR(J) = 1. + (1. NSTRES) NSENP(J) FPSTR(J) = 1. C-------------------------------------------------------------------C Add FPSTR(J) later for phosphorus effects on development C-------------------------------------------------------------------ENDDO Notation on behavior of model with VSTA GE on day of emergence of transplants – final VSTAGE for day reflects initial VSTAGE plus day’s growth. This VSTAGE is what is used for initial partitioning of tran splant DM. Made no change in code, just noted how can override day’s growth and end day with initial VSTAGE

PAGE 295

281 !------------------------------------------------------------------! V-Stage for transplants SJR note Will end day with initial VSTAGE calculated here plus increase for the emergence day. If don't want any increase in VSTAGE on emergence day move this statement to end of subroutine to re-calculate VSTAGE to this value and ignore any new leaves. !------------------------------------------------------------------IF (PLME .EQ. 'T' .AND. YRPLT .EQ. YRDOY) THEN VSTAGE = 1. + (PHZACC(2) MNEMV1) TRIFOL ENDIF !------------------------------------------------------------------IF (DAS .GE. NVEG 0 .AND. DAS .LE. NDVST) THEN IF (DAS .LT. NVEG1) THEN VSTAGE = PHZACC(2)/MNEMV1 ELSE IF (VSTAGE .LT. ABS(EVMODC) .AND. & ABS(EVMODC) .GT. 0.0001) THEN EVMOD = 1.0 + (ABS(EVMODC)VSTAGE) / EVMODC EVMOD = AMIN1(2.0,EVMOD) EVMOD = AMAX1(0.0,EVMOD) ELSE EVMOD = 1.0 ENDIF VSTAGE = VSTAGE + DTX TRIFOL EVMOD*TURFAC*(1.0-XPOD) ENDIF ENDIF !*********************************************************************** !*********************************************************************** End of DYNAMIC IF construct !***********************************************************************

PAGE 296

282 MODULE: SENES.FOR Add storage variables to the SUBROUTINE SENES statement SUBROUTINE SENES( & FILECC, CLW, DTX, NR7, NRUSLF, PAR, RHOL, !Input & SLAAD, STMWT, SWFAC, VSTAGE, WTLF, XLAI, !Input & YRDOY, YRSIM, !Input & SLDOT, SLNDOT, SSDOT, SSNDOT, !Output & STRWT, !Input & SSRDOT, SSRNDOT, !Output & SENSR, !Output & DYNAMIC) !Control Declare variables as REAL REAL SENSR, STRWT, SSRDOT, SSRNDOT READ SENSR from species file C-------------------------------------------------------------------C C ***** READ Storage organ senescence parameters ***** C C-------------------------------------------------------------------SECTION = '!*STOR' CALL FIND(LUN CRP, SECTION, LINC, FOUND) IF (FOUND .EQ. 0) THEN CALL ERRO R(ERRKEY, 1, FILECC, LINC) ELSE CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) CALL IGNORE(LUNCRP,LINC,ISECT,CHAR) READ(CHAR,'(F6.0)',IOSTAT=ERR) SENES IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LINC) ENDIF Initialize SSRDOT & SSRNDOT SSRDOT = 0.0 SSRNDOT = 0.0 Calculate daily senescence from storage organ C-------------------------------------------------------------------C This section calculates natu ral senescence of storage organ tissue C Thought about moving this below the IF...Then line but did not C Don't want to senesce if seedli ng but do want to senesce mature C stand after it has been cut or frozen back C------------------------------------------------------------------SSRDOT = STRWT SENSR DTX SSRDOT = MIN(SSRDOT,STRWT)

PAGE 297

283 Fix calculation of senescence due to low light in lower canopy so SLDOT never > WTLF SLDOT = SLDOT + LTSEN 10000. / SLAAD SLDOT = MIN(WTLF,SLDOT) Fix calculation of senescence due to water stress so SLDOT never > WTLF SLDOT = SLDOT + SLNDOT SLDOT = MIN(WTLF,SLDOT) Add “hook” to allow future manipulation of senescence as function of water stress SSRDOT = SSRDOT + SSRNDOT SSRDOT = MIN(STRWT,SSRDOT) List storage variables at end of subroutine SENSR Constant for senesc ence of storage organ tissue (proportion of cumulative storage weight lost / physiological day) SSRDOT Daily senescence of st orage organ tissue (g / m2 / d) SSRNDOT Storage organ senescen ce due to water stress (g/m2/day) STRWT Dry mass of storag e organ tissue, including C and N

PAGE 298

284 UTILS.FOR Add new output files to SUBROUTINE GETLUN CASE ('OUTDRM'); LUN = 47 !Dormancy.OUT CASE ('OUTSTOR'); LUN = 48 !Storage.OUT CASE ('OUTINSTR'); LUN = 74 !StorS pIn.OUT List of storage inputs Add new curve types for cold hardenin g and dehardening to SUBROUTINE CURV C--------------------------------------------------------------------------C Curve type REV Reversible process used for cold hardening C Rate of cold hardening increases as TMIN decreases from X1 to XB C Cold hardening reverses at an increasing rate as TMIN increases from X1 to X2 C Process at maximum rate at or below XB C Rate decreases linearly to 0 at X1 C Process reverses at a line ar rate from X1 to X2 C XM is the maximum absolute rate C--------------------------------------------------------------------------IF(CTYPE .EQ. 'REV' .OR. CTYPE .EQ. 'rev') THEN CURV = 1. IF(X .GT. XB .AND. X .LT. X1)CURV = 1.0-((X-XB)/(X1-XB)) IF(X .GE. X1 .AND. X .LE. X2)CURV = 0. 0-((X-X1)/(X2-X1)) IF(X .GT. X2 )CURV = -1.0 CURV = MAX(CURV,-1.0) CURV = MIN(CURV,1.0) CURV = CURV XM ENDIF C--------------------------------------------------------------------------C Curve type DHD used for cold dehardening in spring C No cold dehardening below XB (rate=0) C Rate of cold dehardening increases as TMIN increases from XB to X1 C Process at maximu m rate at or above X1 C XM is the maximum absolute rate C--------------------------------------------------------------------------IF(CTYPE .EQ. 'DHD .OR. CTYPE .EQ. 'dhd') THEN CURV = 0. IF(X .GT. XB .AND. X .LT. X1)CURV = (X-XB)/(X1-XB) IF(X .GE. X1 .AND. X .LE. X2)CURV = 1 IF(X .GT. X2 )CURV = 1 CURV = MAX(CURV,0.0) CURV = MIN(CURV,1.0) CURV = CURV XM ENDIF C--------------------------------------------------------------------------C Curve type DRD used for reduci ng rates of processes as dormancy advances C Multiply rates by this factor to reduce them on short days,

PAGE 299

285 C no effect on long days C XM is the maximum reduction factor at full dormancy (daylength=XB) C Less reduction as daylength gets longer C Process at maximu m rate at or above X1 C X2 is not used C--------------------------------------------------------------------------IF(CTYPE .EQ. 'DRD' .OR. CTYPE .EQ. 'drd') THEN CURV = X2 IF(X .GT. XB .AND. X .LT. X1) & CURV = X2+(XM-X2)*(X-XB)/(X1-XB) IF(X .GE. X1 )CURV = XM CURV = MAX(CURV,X2) CURV = MIN(CURV,XM) ENDIF C--------------------------------------------------------------------------C Curve type CDD used for reducing rates of processes as dormancy advances C Multiply rates by this factor to reduce them on short days, C Long day effect depends on value of XM C X2 is the maximum reduction fact or at full dormancy (daylength=XB) C Less reduction as daylength gets longer C Process at maximu m rate at or above X1 C Curvilinear version of DRD C--------------------------------------------------------------------------IF(CTYPE .EQ. 'CDD .OR. CTYPE .EQ. 'cdd') THEN CURV = X2 IF(X .GT. XB .AND. X .LT. X1) & CURV = XM-((X M-X2)*((X1-X)/(X1-XB))**2) IF(X .GE. X1)CURV = XM CURV = MAX(CURV,X2) CURV = MIN(CURV,XM) ENDIF Assign file init numbers to new output files via GETLUN C======================= ==================== =================== C GETLUN, Subroutine, C. H. Porter C-------------------------------------------------------------------C Assigns unique output file unit numbers to input and output files C based on file variable name. If valid file variable name is not C specified, unit numbers are assigned incrementally starting with C unit 90. C-------------------------------------------------------------------C REVISION HISTORY C 10/17/2001 CHP Written. C--------------------------------------------------------------------

PAGE 300

286 Called by: IRRIG, OP WBAL, OPGROW, . Calls: None C======================= ==================== =================== SUBROUTINE GETLUN(FileVarName, LUN) !------------------------------------------------------------------IMPLICIT NONE LOGICAL FEXIST, FPRINT(200) INTEGER COUNTER, ERRNUM, Length, I, LUN, OUTLUN, StartLun CHARACTER*(*) FileVarName CHARACTER*30 SaveName(200) DATA StartLun /90/ DATA COUNTER /0/ DATA FPRINT /200*.FALSE./ DATA OUTLUN /83/ !L ist.OUT list of unit assignments !------------------------------------------------------------------! On first call to subroutine, open new file to record input and output file information. INQUIRE (FILE = 'LIST.OUT', EXIST = FEXIST) IF (FEXIST) THEN OPEN (UNIT = OUTLUN, FILE = 'List.OUT', STATUS = 'OLD', & IOSTAT = ERRNUM, ACCESS = 'APPEND') ELSE OPEN (UNIT = OUTLUN, FILE = 'List.OUT', STATUS = 'NEW', & IOSTAT = ERRNUM) WRITE(OUTLUN,10) 10 FORMAT('*Summary of files opened during simulation', & //,'Unit File',/'Num. Variable Name') ENDIF !------------------------------------------------------------------Length = Len(Trim(FileVarName)) SELECT CASE (FileVarName(1:Length)) Input Files (Units 8 through 29): CASE ('FILEA'); LUN = 8 !observed time series data CASE ('FILEC', 'FILEE', 'F INPUT'); LUN = 10 !*.spe, *.eco, miscellaneous input files CASE ('FILEW'); LUN = 11 !*.wth weather files CASE ('FILEP'); LUN = 12 !*.pst pest files CASE ('FILESS'); LUN = 13 !SOILN980.SOL

PAGE 301

287 CASE ('BATCH'); LUN = 14 !Batch run input file CASE ('ERRORX'); LUN = 15 !Model.err CASE ('FILEIO'); LUN = 21 !temporary input file; dssat40.inp CASE ('DTACDE'); LUN = 22 !DATA.CDE CASE ('FILETMP'); L UN = 23 !Tony Hunt temp file Daily Output Files (Units 30 through 49): FName Code: CASE ('OUTM'); LUN = 30 !MgmtOps.OUT CASE ('OUTWTH'); LUN = 31 !Weather.OUT CASE ('OUTG'); LUN = 32 !PlantGro.OUT CASE ('OUTPN'); LUN = 33 !PlantN.OUT CASE ('OUTPC'); LUN = 34 !PlantC.OUT CASE ('OUTD'); LUN = 35 !Pest.OUT CASE ('OUTT'); LUN = 36 !SoilTemp.OUT CASE ('OUTWAT'); LUN = 37 !SoilWat.OUT CASE ('OUTSN'); LUN = 38 !SoilN.OUT CASE ('OUTSC'); LUN = 39 !SoilC.OUT CASE ('OUTSP'); LUN = 40 !SoilP.OUT CASE ('OUTSPAM'); LUN = 41 !SPAM.OUT CASE ('OUTSOM'); LUN = 42 !SOMLIT.OUT CASE ('OUTETP'); LUN = 43 !ETPhot.OUT CASE ('OUTFLD'); LUN = 44 !Flood.OUT CASE ('OUTCH'); LUN = 45 !Chemical.OUT CASE ('FLDN'); LUN = 46 !FloodN.OUT CASE ('OUTDRM'); LUN = 47 !Dormancy.OUT CASE ('OUTSTOR'); LUN = 48 !Storage.OUT Daily Information files: (Units 50 through 59) CASE ('SLDET'); LUN = 50 !Somlit1.OUT CASE ('OUTWARN'); LUN = 51 !Warning.OUT CASE ('WORK.OUT');LUN = 52 !Work.OUT for CSCERES CASE ('ERRORO'); LUN = 53 !Error.OUT echo of screen errors Seasonal output files (Units 60 through 79): CASE ('SOUTM'); LUN = 60 !MgmtOpsSum.OUT CASE ('SOUTWTH'); LUN = 61 !WeatherSum.OUT CASE ('SOUTG'); LUN = 62 !PlantSum.OUT CASE ('SOUTSPAM');LUN = 65 !SPAMSum.OUT CASE ('SOUTR'); LUN = 66 !Operat.OUT CASE ('SOUTE'); LUN = 67 !Environ.OUT CASE ('SEVAL'); LUN = 68 !Evaluate.OUT CASE ('PNBAL'); LUN = 70 !PlantNbal.OUT CASE ('PCBAL'); LUN = 71 !PlantCbal.OUT CASE ('SNBAL'); LUN = 72 !SoilNbal.OUT CASE ('SCBAL'); LUN = 73 !SoilCbal.OUT

PAGE 302

288 CASE ('OUTINSTR'); LUN = 74 !StorS pIn.OUT List of storage inputs Composite output file s (Units 80 through 89): CASE ('OUTO'); LUN = 80 !Overview.OUT CASE ('OUTS'); LUN = 81 !Summary.OUT CASE ('SWBAL'); LUN = 82 !SoilWatbal.OUT !RESERVE UNIT 83 FOR LIST.OUT CASE ('LIST'); LUN = 83 !List.OUT (list of unit assignments) CASE ('OUTBAT'); LUN = 84 !TEMP.BAT file for DOS commands CASE ('TEMP'); LUN = 85 !TEMP file CASE ('OUTLST'); LUN = 86 !OUTPUT.LST list of output files Files not covered above will be assigned numbers incrementally starting with unit number 90. CASE DEFAULT !First check to see if a unit number has already been !assigned to this FileVarName. If so, assign same LUN. DO I = StartLun, StartLun + Counter IF (FileVarNa me .EQ. TRIM(Sav eName(I))) THEN LUN = I EXIT ENDIF ENDDO !Assign a unique unit number to this FileVarName IF (I .GT. StartLun + Counter) THEN LUN = StartLun + COUNTER COUNTER = COUNTER + 1 ENDIF END SELECT Print to 'LIST.OUT' file each file assigned a unit number (only print the first time a unit is assigned) OUTPUT.LST ICASA format headers, etc. Save FileVarName in case it is used again. IF (.NOT. FPRINT(LUN)) THEN WRITE(OUTLUN,'(I4,2X,A)') LUN, FileVarName FPRINT(LUN) = .TRUE. SaveName(LUN) = FileVarName ENDIF CLOSE(OUTLUN) RETURN END SUBROUTINE GETLUN

PAGE 303

289 MODULE: VEGDM.FOR Add storage parameters to SUBROUTINE VEGDM statement SUBROUTINE VEGDM( & AREALF, CLW, CSW, PCLMA, PCLMT, !Input & PCSTMD, PDLA, PLFAD, PLFMD, PSTMD, !Input & PVSTGD, SLA, SLDOT, SSDOT, STMWT, !Input & TDLA, VSTGD, WLFDOT, WSTM D, WTLF, !Input & TLFAD, TLFMD, VSTAGE, WLIDOT, !Input/Output & CLAI, CLFM, CSTEM, DISLA, DISLAP, !Output & LAIDOT, WSIDOT, !Output & CSRW, PCSTRD, PSTR D, SSRDOT, STRWT, !Input & WSFDOT, WSRFDOT, WSTRD, !Input & CSTRM, WSRIDOT, !Output & CSFRZ, CSRFRZ, DS TOR, SRDAM, !Output & DYNAMIC) !Control Declare stem freeze variable as REAL REAL CSFRZ, WSFDOT Declare storage variables as REAL REAL CSRFRZ, CSRW, CSTRM, DSTOR, PCSTRD, PSTRD, SRDAM, & SSRDOT, STRWT, WSRFDOT, WSRIDOT, WSTRD Initialize storage variables CSTRM = 0.0 WSRIDOT = 0.0 CSFRZ = 0.0 CSRFRZ = 0.0 Include freeze damage in calcul ation of observed stem damage IF (PCSTMD .GT. 0.0) THEN C Desired stem mass DSTEM after cumulative damage DSTEM = CSW (1.0 PCSTMD / 100.0) IF ((STMWT SSD OT WSFDOT) .GT. DSTEM) THEN SDAM = ST MWT SSDOT WSFDOT DSTEM ELSE SDAM = 0.0 ENDIF WSIDOT = WSIDOT + SDAM ENDIF Calculate daily storage organ damage C-------------------------------------------------------------------C Desired observed cumu lative storage organ damage

PAGE 304

290 C-------------------------------------------------------------------C When storage organ damage is reported as percent reduction of total C cumulative storage organ mass produced, use this section. The total storage organ mass produced is CSRW. Part of the observed damage comes from senescence and part from pests. C-------------------------------------------------------------------IF (PCSTRD .GT. 0.0) THEN C Desired storage organ mass DSTOR after cumulative damage DSTOR = CSRW (1.0 PCSTRD / 100.0) IF ((STRWT SSR DOT WSRFDOT) .GT. DSTOR) THEN SRDAM = (STR WT SSRDOT WSRFDOT) DSTOR ELSE SRDAM = 0.0 ENDIF WSRIDOT = WSRIDOT + SRDAM ENDIF C-------------------------------------------------------------------C Percent daily storage organ damage C-------------------------------------------------------------------IF (PSTRD .GT. 0.0) THEN SRDAM = PSTRD*STRWT/100.0 WSRIDOT = WSRIDOT + SRDAM ENDIF C-------------------------------------------------------------------C Absolute daily amount of storage organ mass damaged C-------------------------------------------------------------------IF(WSTRD .GT. 0.0) THEN SRDAM = MIN(WSTRD, STRWT) WSRIDOT = WSRIDOT + SRDAM ENDIF WSRIDOT = MAX(0.,WSRIDOT) WSRIDOT = MIN(WSRIDOT, STRWT) Maintain cumulative values for storage organ loss due to freeze IF (WSRFDOT .GT. 0.0) CSRFRZ = CSRFRZ + WSRFDOT Maintain cumulative storage organ damage values CSTRM = CSRM + WSRIDOT List storage organ variab les at end of subroutine CSFRZ Cumulative frozen stem tissue (g[stem]/m2) CSRFRZ Cumulative frozen storag e organ tissue (g[storage]/m2) CSRW Cumulative storage or gan growth (g[storage]/m2)

PAGE 305

291 CSTRM Cumulative storage organ mass destroyed (g/m2) DSTOR Desired storage organ mass (g/m2/d) PCSTRD Observed cumulative percen tage storage organ mass damage (%) PSTRD Daily percent storage organ mass damage (%) SRDAM Calculated stor age organ damage (g/m2/d) SSRDOT Daily senescence of storage organ(g / m2 / d) STRWT Dry mass of storage organ, in cluding C and N (g[storage] / m2[ground) WSFDOT Stem weight losse s due to freezing (g[stem]/m2-d) WSRFDOT Storage organ weight loss es due to freezing (g[storage]/m2-d) WSRIDOT Daily pest damage to storage organ mass (g/m2/day) WSTRD Daily absolute stor age organ damage (g/m2/day)

PAGE 306

292 MODULE: VEGGR.FOR Add storage and dormancy variables to the SUBROUTINE VEGGR statement SUBROUTINE VEGGR( & AGRLF, AGRRT, AGRSTM, CMINEP, CSAVEV, & DTX, DXR57, ECONO, FILECC, FILEGC, FNINL, & FNINR, FNINS, NAVL, NDMNEW, NDMOLD, & NR1, NVSTL, NVSTR, NVSTS, NVSTSR, & PAR, PCH2O, PCNL, PCNST, PCNRT, PCNSR, PG, !Input & PGAVL, ROWSPC, RT WT, RVSTGE, STMWT, TGRO, & TURFAC, VSTAGE, WCRLF, WCRRT, WCRSH, & WCRST, WTLF, XL AI, YRDOY, YREMRG, YRSIM, & AGRVG, FRLF, FRRT, FRSTM, NMINEA, !Input/Output & NFIXN, TRNU, & CADLF, CADST, CANHT, CANWH, CMINEA, & CRUSLF, CRUSRT, CRUSSH, CRUSST, EXCESS, NADLF, & NADRT, NADST, NGRLF, NGRRT, NGRST, !Output & NSTRES, TNLEAK, WLDOT N, WRDOTN, WSDOTN, & CLAIT, NRUSTOT, !Input & PNMLF, PNMRT, PNMSH, PNMSR,PNMST,RPRO, & CADRT, CADSH, NADSH, !Output & AGRSTR, CMOBSR, FNINSR PPMFAC, STRWT, WCRSR, !Input & FRSTR, !Input/Output & CADSR, CRUSSR, NADSR NGRSR, WSRDOTN, !Output & CADSRF, CMOBSRN, CMOBSRX, !Output & FNINSRG, NGRSRG, PROSRG, PROSRT, !Output & NLAYR, NUPNH4, NUPNO3, PROLFI, PRORTI, !Input & PROSTI, PROSRI, RFIXN, RNH4C, RNO3C, TRNH4U, & TRNO3U, & DYNAMIC) !Control Declare variables as REAL REAL AGRSTR, CADSR, CADSRF CLAIT, CMOBSR, CMOBSRN, & CMOBSRX, CRUSSR, FNINSR, FNINSRG, FRSTR, & NADSR, NGRSR, NG RSRG, PPMFAC, PROSRG, & PROSRI, PROSRT STRWT, WCRSR, WSRDOTN REAL NADSH, NRUSTOT, PNMLF, PNMRT, PNMSH, PNMSR, PNMST, & CADRT, CADSH, RPRO

PAGE 307

293 C-------------------------------------------------------------------C New NDMOLD allocation variables for forage model C Used to weight partitioning in favor of one organ over others C-------------------------------------------------------------------REAL NVSTL, NVSTR, NVSTS, NVSTSR, PCNL, PCNST, PCNRT, PCNSR, & PWLF, PW ST, PWRT, PWSR, RTWT C-------------------------------------------------------------------C New NLEAK distribution variables for all models C Used to "put back" and distribute NLEAK to new growth C-------------------------------------------------------------------INTEGER L, NLAYR REAL AGRVGI, AGRVGPI, DNAD RAT, NLKCOST, NLKGROW, & NRFRESP, PNTVG, NUPNH4(NL), NUPNO3(NL), ONDMOLD, & PNUPNH4,PNUPNO3, RFIXN, RNH4C, RNNU, RNO3C, TRNH4U, & TRNO3U, UNH4(NL), UNO3(NL), XTVEGM Read storage organ Protein parameters fr om plant composition section of species file CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(2F6.0) ',IOSTAT=ERR) PROSRI, PROSRG IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) Read mobilization parameters from mining section of species file CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(3F6.0)',IOSTAT=ERR) CMOBSRN, CMOBSRX, & CADSRF IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) Read NDMOLD partitioning weighting factors from species file !------------------------------------------------------------------! Find and Read Partitioning Section !------------------------------------------------------------------SECTION = '!*VEGE' CALL FIND(LUNCRP, SECTION, LNUM, FOUND) IF (FOUND .EQ. 0) THEN

PAGE 308

294 CALL ERROR( ERRKEY, 1, FILECC, LNUM) ELSE DO I=1,4 ISECT = 2 DO WHILE (ISECT .NE. 1) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) ENDDO ENDDO READ(C80,'(24X,F6.0)',IOSTAT=ERR) ATOP IF (ERR .NE. 0) CALL ERROR(ERRKEY ,ERR,FILECC,LNUM) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) CALL IGNORE(LUNCRP,LNUM,ISECT,C80) READ(C80,'(4F6.0) ',IOSTAT=ERR) PWLF,PWST,PWRT,PWSR IF (ERR .NE. 0) CALL ERROR(ERRKEY,ERR,FILECC,LNUM) ENDIF Initialize other variables CADSR = 0.0 CRUSSR = 0.0 FNINSRG = 0.0 NADSR = 0.0 NGRSR = 0.0 WSRDOTN= 0.0 Calculate initial value for FNINSR FNINSRG = PROSRG 0.16 Add storage tissue to the partitioning sc heme – NOTE: Dormancy adjustments were already made in DEMAND C-------------------------------------------------------------------C 0.6 IS A SCALAR, COULD BE LESS, was once 0.8 and 0.7 C 0.7 appears to be too mu ch for peanut, but not for soybean. C-------------------------------------------------------------------FRSTR = (FRSTR/(FRLF+FRSTM+FRSTR))*(1-FRRT) FRLF = (1.0 + 0.6*(1. 0-CUMTUR))*(1.-FRRT-FRSTR)*FRLF/ & (FRLF + FRSTM) FRLF = MIN(FRLF, 0.90*(1. FRRT-FRSTR)) FRSTM = 1.0 FRRT FRLF – FRSTR

PAGE 309

295 C-------------------------------------------------------------------C To prevent negative partitio ning to root and limit leaf plus C stem to a maximum of 98 % of the vegetative partitioning C NOTE was 98% before addi ng storage organ to scheme C-------------------------------------------------------------------FRLF = MIN(FRLF,FRLF *0.98/(MAX(0.001,FRLF+FRSTM+FRSTR))) FRSTM = MIN(FRSTM,FRST M*0.98/(MAX(0.001,FRLF+FRSTM+FRSTR))) FRSTR = MIN(FRSTR,FRST R*0.98/(MAX(0.001,FRLF+FRSTM+FRSTR))) FRRT = 1.0 FRLF FRSTM – FRSTR Add storage tissue to AGRVG ca lculation – otherwise would have to come up with a fixed value for day’s storage growth as done with seed+shell C-------------------------------------------------------------------C Calculate weighted PHI + GR = 1/E = AGRVG for veg. growth C-------------------------------------------------------------------AGRVG = AGRLF FRLF + AGRRT FRRT + AGRSTM FRSTM & AGRSTR FRSTR Calculate new growth rate of storage organ WRDOTN = FRRT VGRDEM Add storage into calculation of Ma x N required for vegetative growth NGRSR = WSRDOTN FNINSR NGRVEG = NGRLF + NGRST + NGRRT + NGRSR Add storage into calculation of minimum N required for tissue growth NGRSRG = WSRDOTN FNINSRG NGRVGG = NGRLFG + NGRSTG + NGRRTG + NGRSRG Add storage into calculations for reducing leaf growth to prevent N conc of new tissue from being below the minimum for growth WSRDOTN = WSRDOTN NRATIO NGRSRT = NGRSRG NRATIO Add storage into calculation to adjust conv ersion costs to account for composition of tissue at lower N concentration AGRVG = AGRLF FRLF (1.0 (PROLFG PROLFI)/(1.0 & PROLFI))+ AGRRT FRRT (1.0 (PRORTG PRORTI)/(1.0 & PRORTI)) + AGRSTM FRSTM (1.0 (PROSTG PROSTI)/ & (1.0 PROSTI)) + AGRSTR FRSTR (1.0 (PROSRG PROSRI) & /(1.0 PROSRI)) Add storage into calculations when leaf expansion occurs as normal, but N concentration is reduced NGRSR = MIN(NAVL NGRSR / NGRVEG, NGRSR)

PAGE 310

296 Compute protein fraction of new storage tissue growth IF (WSRDOTN .GT. 0.0) THEN PROSRT = NGRSR (100./16.)/WSRDOTN ELSE PROSRT = 0.0 Include storage in calculation of respiratio n costs if expansion occurs at low N-conc. To allow N dilution during growth AGRVG = AGRLF FRLF (1.0 (PROLFT PROLFI)/ & (1.0-PROLFI)) + AGR RT FRRT (1.0 (PRORTT PRORTI)/ & (1.0 PRORTI)) + AGRSTM FRSTM (1.0 (PROSTT & PROSTI)/(1.0 PROSTI)) + AGRSTR FRSTR (1.0 (PROSRT & PROSRI)/(1.0-PROSRI)) Compute C and N remaining to add to reserves, including storage organ PGLEFT = MAX(0.0,PGAVL ((WLDOTN + WSDOTN + WRDOTN + WSRDOTN) & AGRVG)) Initialize CADSR and CRUSSR as well as all NADXX variables CADSR = 0.0 CRUSSR = 0.0 Add storage organ to calculation to increase remobilizable C due to N shortage and add to Carbon Pool. Distribute to Le aves and Stems. Want half as much accumulation in stem in veg phase C-------------------------------------------------------------------C Calculate Increase in Remob ilizable C due to N shortage and C add to Carbon Pool. Di stribute to Leaves and Stems. C-------------------------------------------------------------------C Want half as much accumulation in stem in veg phase C-------------------------------------------------------------------! IF (DAS .LT. NR1) THEN LSTR = (1.-0.6*CADSTF)/(0.6*CADSTF) ELSE C-------------------------------------------------------------------C 5/11/04 KJB/SJR Add code to a llocate excess CH2O to Stolon as well C as to leaf and stem. For fora ges chose to ignore different C partitioning for vegetative vs. reproductive stages. C-------------------------------------------------------------------LSTSR = CADSRF/(1-CADSRF) LSTR = (1.-CADSTF)/CADSTF ENDIF IF (STMWT+WTLF .GT. 0.0) THEN LSTSR = LSTSR*STRWT/(STRWT*LSTSR+STMWT)

PAGE 311

297 LSTR = LSTR WTLF/(STMWT+WTLF*LSTR) ENDIF IF (PGLEFT .GE. CMINEP) THEN CADSR = (PGLEFT-CMINEP)/PCH2O LSTSR CADLF = (PGLEFTCMINEP)/PCH2O LSTR*(1-LSTSR) CADST = (PGLEFT-CMIN EP) (1.-LSTR) (1-LSTSR) / PCH2O ELSE C-------------------------------------------------------------------C Calculate actual C used (CMINEA) compute how much is taken C from LF, ST, RT, and SH, which ma y be less than orig calc of CMINEP C C 8/26/97 KJB DTX IN PLACE OF 1 TO SLOW IT DOWN A BIT AT ALL TIMES C AND TO BE SENSITIVE TO TEMPERATURE PRIOR TO R5 STAGE, BUT C STILL WANT THE SPEED-UP CAUS ED BY THE "+ DXR 57" FEATURE AFTER R5. C C-------------------------------------------------------------------C 7/2/03 SJR added (PPMFAC) to limit mobilization from roots C while dormant C-------------------------------------------------------------------IF (CMINEP .GT. 0) THEN CMINEA = CMINEP PGLEFT CRUSLF = CMINEA / CM INEP CMOBMX WCRLF (DTX + DXR57) CRUSST = CMINEA / CM INEP CMOBMX WCRST (DTX + DXR57) CRUSSH = CMINEA / CM INEP CMOBMX WCRSH (DTX + DXR57) CRUSRT = CMINEA / CMINEP CMOBMX PPMFAC WCRRT & (DTX + DXR57) CRUSSR = CMINEA/CMINEP CMOBSR WCRSR (DTX + DXR57) ENDIF ENDIF CADSR = CADSR + CSAVEV/PCH2O LSTSR CADLF = CADLF + CSAVEV/PCH2O LSTR*(1-LSTSR) CADST = CADST + CSAVEV (1. LSTR)*(1-LSTSR)/PCH2O Include storage and shell tissue in increase in remobilizable N due to a C shortage, add to Nitrogen pool NLEFT = MAX(0.0,NAVL (NGRLF + NGRST + NGRRT + NGRSR)) Move TNLEAK calculation to end of subroutine C-------------------------------------------------------------------C Eliminate existing code for NLEAK is a loss to the system C that should not/does not exist C-------------------------------------------------------------------IF (N LEFT .GT. NDMOLD) THEN NLEAK = NLEFT NDMOLD

PAGE 312

298 NLEFT = NLEFT NLEAK ELSE NLEAK = 0.0 ENDIF Eliminate “false” NLEAK values due solely to rounding/precision errors C-------------------------------------------------------------------C SJR 10/20/03 Added code to fix/distribute NLEAK C NLEAK caused by 3 scenarios C I) Rounding/precision errors. C II) Changing FRRT & FRLF in VEGGR.FOR due to H2O or N stress, C changes N demand. C III) Low Pg exhaust PGAVL on N uptake. Have N available but no C CH2O left for growth so N goes to refill old tissue. C Particularly a problem when LFWT is low. Also causes N% to C exceed initial or maximum con centrations set in species file. C-------------------------------------------------------------------C SJR 10/16/03 Eliminate NLEAK values due to rounding errors C see I) above. C-------------------------------------------------------------------IF (NLEAK .LT. 0.0001) NLEAK = 0.0 If there still is NLEAK and NLEFT. Compare NLEFT to NDMOLD calculated when PGAVL is non-limiting (total amount of N required to refill all old tissue). This should be greater than NLEFT (NDM OLD) because of using PGAVL to limit NDMVEG and NDMOLD in DEMAND. If the capacity allows, put NLEAK there. IF (N LEFT .GT. NDMOLD) THEN C-------------------------------------------------------------------C Calculate capacity for refilling old tissue N if PGAVL were not C limiting. This is the original calculation with no modifiers. C-------------------------------------------------------------------ONDMOLD = (WTLF WCRLF) MAX(0.0,(NVSTL PCNL /100.)) & + (STMWT WCRST) MAX(0.0,(NVSTS PCNST/100.)) & + (RTWT WCRRT) MAX(0.0,(NVSTR PCNRT/100.)) & + (STRWT WCRSR) MAX(0.0,(NVSTSR PCNSR/100.)) C-------------------------------------------------------------------C If total capacity for refilling N allows, put "NLEAK" into NLEFT C Otherwise, allocate NLEAK to NL EFT to the amount allowed by ONDMOLD C-------------------------------------------------------------------IF (ONDMOLD .GT. NDMOLD) THEN IF (ONDMOLD NDMOLD .GT. NLEAK) THEN

PAGE 313

299 NLEFT = NLEFT NLEAK = 0.0 ELSE NLEFT = NLEFT + (ONDMOLD NDMOLD) NLEAK = NLEAK (ONDMOLD NDMOLD) ENDIF ENDIF NLEAK = NLEFT NDMOLD NLEFT = NLEFT NLEAK ELSE NLEAK = 0.0 ENDIF Fix proportioning of NDMOLD for refilling old tissue N – DSSAT4 scheme had two problems: Used FRLF, FRSTM, FRRT – proportions of new growth. Can cause excessive N concentration if today’s growth is small. Can have a high proportion of NDMOLD allocated to leaf but have very little leaf mass to “accept” it, resulting in very high leaf N% Address problem – set allocation of NLEFT (NDMOLD) based on present tissue mass (WTLF, etc.) instead of proportions of new growth. Also add weighting factors (PWLF etc.) to allow preferential ly refilling some tissues before others. C-------------------------------------------------------------------C Original code allocated N back to only leaf, stem and root. C Uses FRxx which is the fraction of today's growth going to organ xx. C This is not indicative of the original "source" of the mobilized N being returned. C-------------------------------------------------------------------C-------------------------------------------------------------------C Allocate excess N uptake to refill old tissues C-------------------------------------------------------------------C-------------------------------------------------------------------C sjr 10/16/03 Revised scheme ba sed on existing growth proportions C-------------------------------------------------------------------IF (NLEFT .GT. 0.0) THEN C-------------------------------------------------------------------C Allocate excess N uptake to refill old tissues C-------------------------------------------------------------------DNADRAT = PWLF (WTLF WCRLF) MAX(0.0,(NVSTL PCNL /100.))

PAGE 314

300 & + PWST (STMWT WCRST) MAX(0.0,(NVSTS PCNST/100.)) & + PWRT (RTWT WCRRT) MAX(0.0,(NVSTR PCNRT/100.)) & + PWSR (STRWT WC RSR) MAX(0.0,( NVSTSR PCNSR/100.)) IF (DNADRAT .GT. 0.0) THEN NADRAT = NLEFT / DNADRAT ELSE NADRAT = 0.0 ENDIF NADLF = NADRAT & (WTLF WCRLF) PW LF MAX(0.0,(NVSTL PCNL /100.)) NADST = NADRAT & (STMWT WCRST) PWST MAX(0.0,(NVSTS PCNST/100.)) NADRT = NADRAT & (RTWT WCRRT) PW RT MAX(0.0,(NVSTR PCNRT/100.)) NADSR = NADRAT & (STRWT WCRSR) PW SR MAX(0.0,(NVSTSR PCNSR/100.)) ELSE NADRAT = 0.0 NADLF = 0.0 NADST = 0.0 NADRT = 0.0 NADSR = 0.0 ENDIF If NLEAK exceeds N refill capacity, then put it back in the soil. NOTE: this doesn’t seem to happen. Hopefully we can get rid of this. C-------------------------------------------------------------------C NLEAK caused by 3 scenarios C I) Rounding/precision errors. C II) Changing FRRT & FRLF in VEGGR.FOR due to H2O or N stress, C changes N demand. C III) Low Pg exhaust PGAVL on N uptake. Have N available but no C CH2O left for growth so N goes to refill old tissue. C Particularly a problem when LFWT is low. Also causes N% to C exceed initial or maximum con centrations set in species file. C-------------------------------------------------------------------C SJR 10/20/03 Added code to fix/distribute NLEAK C Fix for II) and III) above. C Basic strategy is to estimate how much growth could have been C achieved if CH2O that was used fo r N mobilization, fixation, or uptake

PAGE 315

301 C was used for growth instead. Then lower NLEAK and the appropriate C source by the amount of N used in and to fuel this growth. C C Considered using PGAVL in this as well but did not because, to date, C NLEAK that would be addressed by th is fix only occurs when PGAVL is C exhausted. If there were more PGAVL, more grow th would have occurred C and there wouldn't be any NLEAK. C C Strategy: C 1) Figure out where it came from (mined, fixed or uptake) C 2) Use appropriate respiration coeffi cient to calculate how much CH2O C was used to liberate that amount of N C 3) Calculate how much new growth could be generated if that amount C of CH2O was used for grow th and liberation of just enough C N for that new growth, no excess. C 4) Recalculate N source (NMINEA, NF IXN, or TRNU). Recalculate NLEAK C 5) In case of TRNU, distribute N to NO3 and NH4 in each soil layer C 6) Recalculate growth rate and N concentration C-------------------------------------------------------------------IF (NLEAK .GT. 0.0) THEN PNTVG=(FRLF*PROLFI+FRSTM*PROSTI+ FRRT*PRORTI+FRSTR*PROSRI)*0.16 AGRVGI=FRLF*AGRLF+FRSTM*AGRSTM+FRRT*AGRRT+FRSTR*AGRSTR AGRVGPI=FRLF*PROLFI+FRSTM*PROSTI+FRRT*PRORTI+ & FRSTR*PROSRI IF (TRNU .GT. 0.0)THEN PNUPNO3 = TRNO3U / TRNU PNUPNH4 = TRNH4U / TRNU RNNU= PNUPNO3 RNO3C + PNUPNH4 RNH4C CALL NLKDIST( & AGRVGI, AGRVGPI, FR LF, FRRT, FRSTM, FRSTR, !Input & NLEAK, RNNU, TRNU, PNTVG, & NGRLF, NGRRT, NGRST, NGRSR, !Input/Output & WLDOTN, WRDOTN, WSDOTN, WSRDOTN, & NLKCOST, NRFRESP, XTVEGM) !Output TRNU=TRNU-NRFRESP

PAGE 316

302 NLEAK=NLEAK-NLKCOST C Allocate N "returned for respirati on" (CH2O) to layers as NO3&NH4 DO L = 1,NLAYR IF (TRNO3U .GT. 0.0) THEN UNO3(L) = UNO3(L) + NRFRESP NUPNO3(L) ENDIF IF (TRNH4U .GT. 0.0) THEN UNH4(L) = UNH4(L) + NRFRESP NUPNH4(L) ENDIF ENDDO ENDIF IF (NLEAK .LT. 0.0001) NLEAK = 0.0 TNLEAK = TNLEAK + NLEAK ENDIF Add NLKDIST subroutine to get rid of NLEAK C-------------------------------------------------------------------C Subroutine NLKDIST for distributing NLEAK C For each potential source of NLEAK: C 2) Use appropriate respiration co efficient (NRSPCST)to calculate C how much CH2O was used to liberate that amount of N C 3) Calculate how much new growth could be generated if that amount C of CH2O was used for grow th and liberation of just enough C N for that new growth, no excess. C-------------------------------------------------------------------SUBROUTINE NLKDIST( & AGRVGI, AGRVGPI, FR LF, FRRT, FRSTM, FRSTR, !Input & NLEAK, NRSPCST, NSOURCE, PNTVG, & NGRLF, NGRRT, NGRST, NGRSR, !Input/Output & WLDOTN, WRDOTN, WSDOTN, WSRDOTN, & NLKCOST, NRFRESP, XTVEGM) !Output REAL AGRVG2I, AGRVGI, AGRVGP I, CH2OCST, FRLF, FRRT, & FRSTM, FRSTR, LKNRET, NLEAK, NRSPCST, NSOURCE, & NGRLF, NGRRT, NGRST, NGRSR, PNTVG, RNNGR, & WLDOTN, WRDOTN, WSDOTN, WSRDOTN, & NLKCOST, NLKGROW, NRFRESP, XTVEGM

PAGE 317

303 AGRVG2I = 0.0 CH2OCST = 0.0 LKNRET = 0.0 NLKCOST = 0.0 NLKGROW = 0.0 NFRRESP = 0.0 RNNGR = 0.0 XTVEGM = 0.0 IF (NLEAK .GT. NSOURCE) THEN Use all NLEAK from NSOURCE for respiration, pull N for new growth from excess NLEAK (from other N sources) LKNRET = NSOURCE AGRVG2I = AGRVGI + AGRVGPI NRSPCST RNNGR = (0.16 / NRSPCST) AGRVG2I XTVEGM = LKNRET / RNNGR CH2OCST = XTVEGM AGRVG2I NRFRESP = (0.16 / NRSPCST) CH2OCST NLKGROW = XTVEGM PNTVG NLKCOST = NRFRESP + NLKGROW IF (NLKCOST .GT. NLEAK) THEN Use all NLEAK from all sources. Allocate to respiration and new growth. Some NLEAK from NSOURCE will be used for new growth LKNRET = NLEAK AGRVG2I = AGRVGI + AGRVGPI NRSPCST RNNGR = (0.16 / NRSPCST) AGRVG2I XTVEGM = LKNRET / (PNTVG + RNNGR) CH2OCST = XTVEGM AGRVG2I NRFRESP = (0.16 / NRSPCST) CH2OCST NLKGROW = XTVEGM PNTVG NLKCOST = NRFRESP + NLKGROW ENDIF ELSE Use all NLEAK. Allocate to respir ation and new growth. Most NLEAK from NSOURCE will be used for respiration, rest for new growth. Thus, will end up with some remaining NSOURCE. LKNRET = NLEAK AGRVG2I = AGRVGI + AGRVGPI NRSPCST RNNGR = (0.16 / NRSPCST) AGRVG2I XTVEGM = LKNRET / (PNTVG + RNNGR)

PAGE 318

304 CH2OCST = XTVEGM AGRVG2I NRFRESP = (0.16 / NRSPCST) CH2OCST NLKGROW = XTVEGM PNTVG NLKCOST = NRFRESP + NLKGROW ENDIF WLDOTN = WLDOTN + (FRLF XTVEGM) WSDOTN = WSDOTN + (FRSTM XTVEGM) WRDOTN = WRDOTN + (FRRT XTVEGM) WSRDOTN = WSRDOTN + (FRSTR XTVEGM) NGRLF = NGRLF + (FRLF XTVEGM) NGRST = NGRST + (FRSTM XTVEGM) NGRRT = NGRRT + (FRRT XTVEGM) NGRSTR = NGRSTR + (FRSTR XTVEGM) RETURN C-------------------------------------------------------------------END SUBROUTINE NLKDIST List variable definitions at end of subroutine AGRSTR Mass of CH2O required for new storage organ growth (g[CH2O] / g[storage]) AGRVG2I CH2O cost for new growth from NLEAK, including cost for liberating and reducing N to CP (g[CH2O] / g[tissue]) varies respiration cost depending on N source AGRVGI CH2O cost for new growth from NLEAK, excludes cost for liberating and reducing N to CP (g[CH2O] / g[tissue]) AGRVGPI Intermediate value used to calculate AGRVG2I CADRT Mass of CH2O added back to root s for CH2O cost of excess mobilized N (never really mobilized) (g[CH2O] / m2 / d) CADSH Mass of CH2O added back to shells for CH2O cost of excess mobilized N (never really mobilized) (g[CH2O] / m2 / d) CADSR Mass of CH2O added to stor age organs (g[CH2O] / m2 / d) CADSRF Proportion of CH2O reserves that are added to storage organ (fraction) CH2OCST CH2O cost for new growth generated from "recovering" NLEAK (g[CH2O] / m2 / d) CLAIT LAI threshold triggering increased mobilization from storage organ tissue represents dramatic loss due to harvest or damage CMOBSR Stage-dependent potentia l C mining rate from storage organ expressed as a fraction of the maximum rate (CMOBSRX) CMOBSRN Minimum fraction of C which can be mobilized from storage organ in a day CMOBSRX Maximum fraction of C which can be mobilized from

PAGE 319

305 storage organ in a day CRUSSR C mobilized from storage orga n tissue in a day (g [CH2O] / m2 / d) DNADRAT Denominator of calculation for NADRAT. Separated out to prevent divide by 0.0 errors. FNINSR Maximum fraction of N for growing storage organ tissue (g[N] / g[storage]) FNINSRG Minimum fraction of N fo r growing storage organ tissue (g[N] / g[storage]) FRSTR Fraction of vegetative tissue growth that goes to storage organs on a day (g[storage] / g[veg]) L A counting variable LKNRET "Generic" variable for amount of NLEAK potentially attributed to an N source (NMINEA, NF IXN, TRNU) (g[N] / m2 / d) NADSH N added back to shell N reserves excess mobilized N (never really mobilized)(g[N] / m2 / d) NADSR N added to storage orga n N reserves (g[N] / m2 / d) NGRSR Maximum N demand for storage organ growth (g[storage N] / m2[ground] / d) NGRSRG Minimum N requirement for storage organ growth (g[storage N] / m2[ground] / d) NL Maximum number of soil layers = 20 NLAYR Actual number of soil layers NLKCOST "Generic" term for the origin al respiration cost for acquiring and reducing all "recovered" NLEAK to CP. (g[CH2O] / m2 / d) NLKCOST Total cost of N for new growth generated from NLEAK. Includes both the actual N in the new tissue but the amount of N "returned" to the s ource to "recover" the CH2O used to acquire it. (g[N] / m2 / d) NLKGROW N "returned" to the source to "recover" the CH2O used to acquire it. (g[N] / m2 / d) NRFRESP N "recovered for respiration" NLEAK "returned to source in exchange for the CH2O used to acquire it (g[N] / m2 / d) NRSPCST "Generic" variable name fo r N respiration cost or CH2O cost to acquire a gram of N from a given source. Assumes values of RPRO, RFIXN, or RNNU depending on N source. (g[CH2O] / g[protein]) NSOURCE "Generic" variable name for N source that NLEAK is being "returned" to. Assumes the values of NMINEA, NFIXN, or TRNU depending on source of N (g[N] / m2 / d) NUPNO3(L) Proportion of TRNU from a soil layer that is nitrate NUPNH4(L) Proportion of TRNU from a soil layer that is ammonium PNTVG Percent N in vegetative tissues at initial or max protein concentration. PNUPNH4 Proportion of TRNU that is NH4 PNUPNO3 Proportion of TRNU that is NO3 PPMFAC Reduction in mobilization from storage organ due to photoperiod induced dormancy PNMLF Proportion of actually mobilized N mobilized from leaves in a day

PAGE 320

306 PNMST Proportion of actually mobilized N mobilized from stems in a day PNMRT Proportion of actually mobilized N mobilized from roots in a day PNMSR Proportion of actually mobilized N mobilized from storage organ in a day PNMSH Proportion of actually mobilized N mobilized from shells in a day PROSRG Normal growth protein com position in storage organ during growth (g[protein] / g[storage]) PROSRI Maximum protein composition in storage organ during growth with luxurious supp ly of N (g[protein] / g[storage]) PROSRT Protein fraction of new storage organ growth (g[protein] / g[storage]) PWLF Weighting factors for partitio ning N when refilling old leaf tissues PWRT Weighting factors for partitio ning N when refilling old root tissues PWSR Weighting factors for partitio ning N when refilling old STOR tissues PWST Weighting factors for partitio ning N when refilling old stem tissues RFIXN CH2O required for biological N fixation (g[CH2O] / g[protein]) RNH4C CH2O required for protein synthesis when source of N is ammonium uptake (g[CH2O] / g[protein]) RNNGR (g[N] / m2 / d) that must be "returned" to "release" enough CH2O to produce 1 gram of new growth RNNU Actual CH2O required for prot ein synthesis when source of N is ammonium + n itrate uptake (g[CH2 O] / g[protein]) RNO3C CH2O required for protein s ynthesis when source of N is nitrate uptake (g[CH2O] / g[protein]) RPRO Respiration required for re -synthesizing protei n from mobilized N (g[CH2O] / g[protein]) STRWT Dry mass of storage organ tissue, including C and N (g[storage] / m2[ground) UNO3 Uptake of NO3 from soil (interim value) (kg N/ha) UNH4 Uptake of NH4 from soil (interim value) (kg N/ha) WCRSR Mass of CH2O reserves in st orage organ (g[storage CH2O] / m2[ground]) WSRDOTN Dry weight growth rate of new storage organ ti ssue including N but not C reserves (g[stem] / m2[ground]-d) XTVEGM "Extra" new growth generated from "returning" N LEAK to its sources in exchange for the CH2O used originally to acquire that amount of N (g[vegetative tissue] / m2-d)

PAGE 321

307 LIST OF REFERENCES Agata, W. 1985a. Studies on dry matter production of bahiagrass (Paspalum notatum) sward: I. Characteristics of dry matter production duri ng the regrowth period. pp 1235-1236. Proceedings of the XV Internati onal Grassland Congress. 24 Aug. 1985. The Science Council of Japan, The Ja panese Society of Grassland Science, Kyoto, Japan. Agata, W. 1985b. Studies on dry matter production of bahiagrass (Paspalum notatum) sward: II. Characteristics of CO2 balance and solar ener gy utilization during the regrowth period. pp 1237-1238. Proceedings of the XV International Grassland Congress. 24 Aug. 1985. The Science Council of Japan, The Japanese Society of Grassland Science, Kyoto, Japan. Beaty, E.R., R.G. Clements, and J.D. Powe ll. 1964. Effects of fertilizing Pensacola bahiagrass with nitrogen. Journal of Soil and Water Conservation 19:194-195. Beaty, E. R., E. V. S. B. Sampaio, D. A. Ashley, and R. H. Brown. 1974. Partitioning and translocation of 14C photosynthate by bahiagrass (Paspalum notatum, Flugge) pp 19-25. In Sectional Papers:"Biological and Physiological Aspects of the Intensification of Grassland Utilization.". June 1974. XII International Grassland Congress Organizing Committee, Moscow. Beaty, E.R., R.L. Stanley, and J. Powell. 1968. Effect of height of cut on yield of Pensacola bahiagrass. Agron. J. 60:356-358. Beaty, E.R., and K.H. Tan. 1972. Organic matter, N, and base accumulation under Pensacola bahiagrass. Journal of Range Management 25:38-40. Beaty, E.R., K.H. Tan, R.A. McCreery, a nd J.B. Jones. 1975. Root-herbage production and nutrient uptake and retention by berm udagrass and bahiagrass. Journal of Range Management 28:385-389. Blue, W.G. 1973. Role of Pensacola bahiagrass stolon-root systems in fertilizer nitrogen utilization on Leon fine sand. Agron. J. 65:88-91. Bolton, J.K., and R.H. Brown. 1980. Photosynt hesis of grass species differing in carbon dioxide fixation pathways V. Response of Panicum maximum, Panicum milioides, and tall fescue (Festuca arundinacea) to nitrogen nutrition. Plant Physiol. 66:97100.

PAGE 322

308 Boote, K. J., J. W. Jones, and J. M. Bennett. 1985. Factors influencing crop canopy CO2 assimilation of soybean. pp 780-788. In Shibles, R. Proceedings of the World Soybean Research Conference III. Westview Press, Boulder, CO. Boote, K.J., J.W. Jones, and G. Hoogenboom. 1998a. Simulation of crop growth: CROPGRO model. p. 651-692. In R.M. Peart, and R.B. Curry (ed.) Agricultural Systems Modeling and Simulation. Marcel Dekker, Inc., New York. Boote, K.J., J.W. Jones, G. Hoogenboom, and N.B. Pickering. 1998b. The CROPGRO model for grain legumes. p. 99-128. In G.Y. Tsuji, G. Hoogenboom, and P.K. Thornton (ed.) Understanding Options fo r Agricultural Production. Kluwer Academic Publishers, Dordrecht, The Netherlands. Boote, K.J., M.I. Mnguez, and F. Sau. 2002. Adapting the CROPGRO legume model to simulate growth of faba bean. Agron. J. 94:743-756. Boote, K.J., and N.B. Pickering. 1994. M odeling Photosynthesis of row crop canopies. HortScience 29:1423-1434. Boote, K. J., L. E. Sollenberger, L. H. Alle n, Jr., and T. R. Sinclair. Carbon balance and growth adaptation of contrasting C3 and C4 perennial forage species to increased CO2 and temperature. Report Number 73. 1999 Tuscaloosa, AL, The University of Alabama. Final Technical Report. South east Regional Center National Institute for Global Environmental Change. Brown, R.H. 1978. A difference in N use efficiency in C3 and C4 plants and its implications in adaptation and evolution. Crop Sci. 18:93-98. Burton, G.W., E.H. DeVane, and R.L. Carter 1954. Root penetration, distribution and activity in Southern grasses measur ed by yields, drought symptoms and P32 uptake. Agron. J. 46:229-233. Burton, G.W., J.E. Hook, J.L. Butler, and R. E. Hellwig. 1988. Effect of temperature, daylength, and solar radiation on producti on of Coastal bermudagrass. Agron. J. 80:557-560. Chambliss, C. G. 2002. Bahi agrass. SS-AGR-36. Institut e of Food and Agricultural Sciences, University of Florida, Ag ronomy Department, Florida Cooperative Extension Service. Crespo, H.M., M. Frean, C.F. Cresswell, and J. Tew. 1979. The occurrence of both C3 and C4 photosynthetic characteristics in a single Zea mays plant. Planta 147:257263. Detling, J.K., W.J. Parton, and H.W. Hunt. 1979. A simulation model of Bouteloua gracilis biomass dynamics on the North Ameri can shortgrass prairie. Oecologia 38:167-191.

PAGE 323

309 DiPaola, J.M., J.B. Beard, and H. Brawa nd. 1982. Key elements in the seasonal root growth of bermudagrass and St. A ugustinegrass. HortScience 17:829-831. Ehleringer, J., and O. Bjrkman. 1977. Quantum yields for CO2 uptake in C3 and C4 plants dependence on temperature, CO2, and O2 concentration. Plant Physiol. 59:86-90. Ehleringer, J., and R.W. Pearcy. 1983. Variation in quantum yield for CO2 uptake among C3 and C4 plants. Plant Physiol. 73:555-559. Evers, G.W. 1985. Forage and nitrogen cont ributions of Arrowleaf and Subterranean clovers overseeded on bermudagrass a nd bahiagrass. Agron. J. 77:960-963. Farquhar, G.D., and S. vonCaemmerer. 1982. Modelling of photosynt hetic response to environmental conditions. p. 549-587. In O.L. Lange, P.S. Nobel, C.B. Osmon, and H. Ziegler (ed.) Physiological Plant Ecology II. Water relations and carbon assimilation. Springer Verlag, Berlin. Finlayson, E. H. Pensacola... A new fine leaved bahia. Southern Seedsman [December], 9-28. 1941. Fritschi, F.B., K.J. Boote, L.E. Sollenberge r, L.H. Allen, Jr., and T.R. Sinclair. 1999. Carbon dioxide and temperature effects on forage establishment: photosynthesis and biomass production. Globa l Change Biology 5:441-453. Gates, R.N., P. Mislevy, and F.G. Ma rtin. 2001. Herbage accumulation of three bahiagrass populations during the co ol season. Agron. J. 93:112-117. Ghannoum, O., S. vonCaemmerer, and J.P. Conroy. 2001a. Carbon and water economy of Australian NAD-ME and NADP-ME C4 grasses. Aust. J. Plant Physiol. 28:213223. Ghannoum, O., S. vonCaemmerer, and J.P. C onroy. 2001b. Plant water use efficiency of 17 Australian NAD-ME and NADP-ME C4 grasses at ambient and elevated CO2 partial pressure. Aust. J. Plant Physiol. 28:1207-1217. Gijsman, A.J., G. Hoogenboom, W.J. Pa rton, and P.C. Kerridge. 2002. Modifying DSSAT crop models for low-input agricu ltural systems using a soil organic matterresidue module from CENTURY. Agron. J. 94:462-474. Godwin, D.C., and C.A. Jones. 1991. Nitrogen dynamics in soil-plant systems. p. 287321. In J. Hanks, and J.T. Ritchie (ed.) M odeling Plant and Soil Systems. ASA, CSSA, SSSA, Madison, WI. Hattersley, P.W., and L. Watson. 1976. C4 grasses: an anatomical criterion for distinguishing between NADP-Malic enzy me species and PCK or NAD-Malic enzyme species. Aust. J. Bot. 24:297-308.

PAGE 324

310 Hoogenboom, G., J.W. Jones, C.H. Porter, K.J. Boote, W.D. Batchelor, L.A. Hunt, A.J. Gijsman, P.W. Wilkens, U. Singh, and W.T. Bowen. 2003. DSSAT v4 cropping system simulation model. In G. Hoogenboom, J.W. Jones, C.H. Porter, P.W. Wilkens, K.J. Boote, W.D. Batchelor, L.A. Hunt, and G.Y. Tsuji (ed.) Decision Support System for Agrotechnology Transfer Version 4.0. Volume 1. University of Hawaii, Honolulu, HI. ICASA. 1998. DSSAT v3.5. Univer sity of Hawaii, Honolulu, HI. Impithuska, V., and W.G. Blue. 1985. Fertilizer nitrogen and nitrog en-15 in three warmseason grasses grown on a Florida spodos ol. Soil Sci. Soc. Am. J. 49:1201-1204. Jagtap, S. S., J. W. Jones, and A. J. Gijsman. 2004. Advances in estimating soil water parameters for soil water balance simulation. pp 42. Biological Systems Simulation Conference. 8 Mar. 2004. BSSG, Gainesville, FL. Jandel Scientific Software. 1996. Table Curv e 2D v4. AISN Software Inc., Mapleton, OR. Johnson, C.R., B.A. Reiling, P. Mislevy, and M.B. Hall. 2001. Effects of nitrogen fertilization and harvest date on yield, digestibility, fibe r, and protein fractions of tropical grasses. J. Anim. Sci. :2439-2448. Jones, J.W., G. Hoogenboom, C. H. Porter, K.J. Boote, W.D. Batchelor, L.A. Hunt, P.W. Wilkens, U. Singh, A.J. Gijsman, and J.T. Ritchie. 2003. The DSSAT cropping systems model. Europ. J. Agronomy 18:235-265. Jordan, D.B., and W.L. Ogren. 1984. The CO2/O2 specificity of ribulose 1,5-bisphosphate carboxylase/oxygenase: Dependence on ribulos ebisphosphate concentration, pH, and temperature. Planta 161:308-313. Kanai, R., and G.E. Edwards. 1999. The biochemistry of C 4 photosynthesis. p. 49-87. In R.F. Sage, and R.K. Monson (ed.) C4 Plant Biology. Academic Press, San Diego, CA. Kanneganti, V.R., C.A. Rotz, and R.P. Wa lgenbach. 1998. Modeling freezing injury in alfalfa to calculate forage yield I: M odel development and sensitivity analysis. Agron. J. 90:687-697. Kelly, T.C. 1995. A Bioeconomic Systems Appro ach to Sustainability Analysis at the Farm Level. Ph.D. thesis, University of Florida. Kimball, S.L., and F.B. Salisbury. 1973. Ultras tructural changes of pl ants exposed to low temperatures. Am. J. Botany 60:1028-1033. Ku, S.B., and G.E. Edwards. 1978. Oxygen i nhibition of photosynthesis III. Temperature dependence of quantum yiel d and its relation to O2/CO2 solubility ratio. Planta 140:1-6.

PAGE 325

311 Long, S.P. 1983. C4 photosynthesis at low temperatures. Plant, Cell and Environment 6:345-363. Long, S.P. 1999. Environmen tal responses. p. 215-250. In R.F. Sage, and R.K. Monson (ed.) C4 Plant Biology. Academic Pr ess, SanDiego, CA, USA. Ludlow, M.M., and G.L. Wilson. 1971. Photos ynthesis of tropical pasture plants I. Illuminance, carbon dioxide c oncentration, leaf temperat ure, and leaf-air vapour pressure difference. Aust J. Biol. Sci. 24:449-470. Marousky, F. J., A. E. Dudeck, L. B. McCarty, and S. F. Anderson. 1992. Influence of daylength and fertility on growth of bermudagrass cultivars. pp 236-238. Proc. Fla. Hort. Soc. 1992. Florida State Hort. Soc., Gainesville, FL MathSoft, I. 1998. Mathcad 8 Professi onal. MathSoft, Inc., Cambridge, MA. Meinzer, F.C., and J. Zhu. 1998. Nitrogen stress reduces the efficiency of the C4 CO2 concentrating system, and therefore quantum yield, in Saccharum (sugarcane) species. J. Exp. Bot. 49:1227-1234. Mislevy, P. Bermudagrass, st argrass, and bahiagrass growth during a frost-free winter. The Florida Cattleman and Livest ock Journal [December]. 1998. Mislevy, P., T. R. Sinclair, and J. Ray. Im proving forage productivity during late fall and early winter. The Florida Cattleman and Livestock Journal [January]. 2000. Monson, R.K., R.O. Littlejohn, Jr., and G.J. Williams, III. 1982. The quantum yield for CO2 uptake in C3 and C4 grasses. Photosynthe sis Research 3:153-159. Neter, J., W. Wasserman, and M.H. Kutner. 1990. Applied Linear Statistical Models: Regression, Analysis of Variance, and Experimental Designs. Richard D. Irwin Inc., Boston, MA. Pedreira, C.G.S., and R.H. Brown. 1996a Physiology, morphology, and growth of individual plants of selected and uns elected bahiagrass populations. Crop Sci. 36:138-142. Pedreira, C.G.S., and R.H. Brown. 1996b. Yi eld of selected and unselected bahiagrass populations at two cutting he ights. Crop Sci. 36:134-137. Penning de Vries, F.W.T., A.H.M. Bruns ting, and H.H. van Laar. 1974. Products, requirements and efficiency of biosynthesi s: A quantitative approach. J. Theor. Biol. 45:339-377. Pittermann, J., and R.F. Sage. 2000. Photosynt hetic performance at low temperature of Bouteloua gracilis Lag., a high-altitude C4 grass from the Rocky Mountains, USA. Plant, Cell and Environment 23:811-823.

PAGE 326

312 Raghavendra, A.S., and V.S. RamaDas. 1993. C4 photosynthesis and C3-C4 intermediacy adaptive strategies for semi arid tropics. p. 317-338. In Y.P. Abrol, P. Mohanty, and Govindjee (ed.) Photosynthesis: Photorea ctions to Plant Productivity. Kluwer Academic Publishers, Boston. Rajendrudu, G., and V.S.R. Das. 1981. C4 photosynthetic carbon metabolism in the leaves of aromatic tropical grasses I. Leaf anatomy, CO2 compensation point and CO2 assimilation. Photosynthesis Research 2:225-233. Roseler, D.K., D.G. Fox, A.N. Pell, and L.E. Chase. 1997. Evaluation of alternative equations for prediction of intake for Hols tein dairy cows. J. Dairy Sci. 80:864-877. Rymph, S. J. and K. J. Boote. 2002. Canopy photosynthesis, resp iration, growth, and partitioning to plant components during regrowth of bahiagrass. In Annual Meeting Abstracts, 2002 Annual Meetings ASA, CSSA, SSSA, Indianapolis, IN. 10-14 Nov.2002. ASA, CSSA, SSSA Madison, WI. (abstr.) Rymph, S. J., K. J. Boote, P. Mislevy, G. W. Evers, and A. Irmak. 2003. Modification of CROPGRO to Simulate Growth and Com position of Perennial Tropical Grasses. In Annual Meeting Abstracts, 2003 Annual Me etings, ASA, CSSA, SSSA, Denver, CO. 2-6 Nov.2003. ASA, CSSA, S SSA. Madison, WI. (abstr.) Ryser, P., and S. Wahl. 2001. Interspecific variation in RGR and the underlying traits among 24 grass species grown in full daylight. Plant biol. 3:426-436. Sampaio, E.V.S.B., and E.R. Beaty. 1976. Mo rphology and growth of bahiagrass at three rates of nitrogen. Agron. J. 68:379-381. SAS Institute Inc. 1987. SAS/STATTM Guide for Personal Computers, Version 6 Edition. SAS Institute Inc., Cary, NC. Scholberg, J.M.S., K.J. Boote, J.W. Jones, and B.L. McNeal. 1997. Adaptation of the CROPGRO model to simulate the growth of field-grown tomato. p. 133-151. In M.J. Kropff, P.S. Teng, P.K. Aggarwal, J. Bouma, B.A.M. Bouman, J.W. Jones, and H.H. van Laar (ed.) Proceedings of the second international symposium on systems approaches for agricultural de velopment. Kluwer Academic Publ., London. Schroder, V. N. 1958. Photosynthetic and so il respiration measurements with various crops. pp 106-110. Proceedings, The Soil and Crop Science Society of Florida. Vol. 18. Gainesville, FL. Dec. 1958. The Soil and Crop Science Society of Florida, Gainesville, FL. Scott, J.M. 1920. Bahia grass. J. Am. Soc. Agron. 12:112-113. Sinclair, T.R., P. Mislevy, and J.D. Ray. 2001. Short photoperiod inhi bits winter growth of subtropical grasses. Planta 213:488-491.

PAGE 327

313 Sinclair, T.R., J.D. Ray, P. Mislevy, a nd L.M. Premazzi. 2003. Growth of subtropical forage grasses under extended photoperi od during short-daylength months. Crop Sci. 43:618-623. Skinner, R.H., J.A. Morgan, and J.D. Hanson. 1999. Carbon and nitrogen reserve remobilization following defoliati on: Nitrogen and elevated CO2 effects. Crop Sci. 39:1749-1756. Slack, C.R., and M.D. Hatch. 1967. Compara tive studies on the ac tivity of carboxylases and other enzymes in relation to the ne w pathway of photosynthetic carbon dioxide fixation in tropical grasse s. Biochem. J. 103:660-665. Snedecor, G.W., and W.G. Cochran. 1989. Stat istical Methods. Iowa State University Press, Ames, IA. Staples, C. R. 1995. Bermudagr ass: Growing, storing, and feeding for dairy animals. Circular 1140. Institute of Food and Agricultural Sciences University of Florida, Florida Cooperative Extension Service. Sugimoto, Y., and I. Nikki. 1979. Studies on th e responses of pasture grasses to nitrogen fertilization. III. Effect of nitrogen ferti lizer rate, leaf nitrogen concentration and chlorophyll content on photosynt hetic activities of some subtropical grass species. J. Japan. Grassl. Sci. 25:121-127. Sugiyama, T., and Y. Hirayama. 1983 Correlation of the activities of phosphoenolpyruvate carboxylase and pyruvate orthophosphate dikinase with biomass in maize seedlings. Plant Cell Physiol. 24:783-787. Unruh, J.B., R.E. Gaussoin, and S.C. Wiest. 1996. Basal growth temperatures and growth rate constants of warm-season turf grass species. Crop Sci. 36:997-999. Usuda, H., M.S.B. Ku, and G.E. Edwards. 1984. Rates of photosynthesis relative to activity of photosynt hetic enzymes, chlorophyll and soluble protein content among ten C4 species. Aust. J. Plant Physiol. 11:509-517. Ward, C.Y., and V.H. Watson. 1973. Ba hiagrass and carpetg rass. p. 314-320. In M.E. Heath, D.S. Metcalfe, and R.F. Barnes (ed.) Forages, the science of grassland agriculture. The Iowa State University Press, Ames, IA. West, S. H. 1973. Carbohydrate metabolism and photosynthesis of tropical grasses subjected to low temperatures. pp 165-168. In Slayter, R. O. Plant Response to Climatic Factors. Proc. Uppsala Symp. UNESCO, Paris. Willmott, C.J. 1981. On the validation of models. Physical Geography 2:184-194. Willmott, C.J. 1982. Some comments on the ev aluation of model performance. Bulletin American Meteorologi cal Society 63:1309-1313.

PAGE 328

314 Wilson, J.R. 1975. Influence of temperatur e and nitrogen on grow th, photosynthesis and accumulation of non-structural carbohydrate in a tropical grass, Panicum maximum var. trichoglume. Neth. J. Agric. Sci. 23:48-61. Woodard, K.R., E.C. French, L.A. Sweat, D.A. Graetz, L.E. Sollenberger, B. Macoon, K.M. Portier, B.L. Wade, S.J. Rymph, G.M. Prine, and H.H. VanHorn. 2002. Nitrogen removal and nitrate leaching for forage systems receiving dairy effluent. J. Environ. Qual. 31:1980-1992.

PAGE 329

315 BIOGRAPHICAL SKETCH Stuart James Rymph was born on June 10, 1961 in Ithaca, NY to Thelma S. and Donald E. Rymph. As his parents shifted car eers from a private veterinary practice to a diversified crop farm, his home environment pr ovided Stuart with br oad experience, an intense interest in animals and plants, and a sense of practicality to complement the academic teachings he received. He received his high school diploma from Greenwich Central School in Greenwich, NY in 1979 a nd began studies at Cornell University, Ithaca, NY. While at Cornel l, Stuart took a variety of courses in agronomy, animal science, agricultural engineering, and agri cultural economics. He graduated with a degree in general agri culture in 1983. On graduation he joined the Wayne Feed Division of Continental Grain Company as a sales trainee. One year later he married Mary Beth Hall and, after a brief honeymoon, moved to Western NY as a distri ct salesman, supporting independent feed dealers in Western NY and Northern PA. In 1988, he left Wayne Feeds to work as a nutritionist for Ag Network, Inc., an i ndependent feed processor/manufacturer and fertilizer blender. Two years later he and his wife returned to the Ithaca area so that Mary Beth could pursue a Ph.D. degree in ruminant nutrition at Cornell. While there, Stuart worked as a nutritionist for a local feed m ill and fertilizer blender, Ward & VanScoy, Inc., and became a Certified Crop Adviser. On completion of her degree in 1996, Mary Be th was hired as an assistant professor by the University of Florida. Stuart seized the opportunity to pursue answers to questions

PAGE 330

316 that arose in his nutrition wo rk and completed a Master of Science degree program in agronomy with a minor in dairy science at UF in 1999. His Ph.D. program was undertaken to expand his experience with fora ge crops and further his qualifications for working in nutrient management with dairy farms.


Permanent Link: http://ufdc.ufl.edu/UFE0006608/00001

Material Information

Title: Modeling Growth and Composition of Perennial Tropical Forage Grasses
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0006608:00001

Permanent Link: http://ufdc.ufl.edu/UFE0006608/00001

Material Information

Title: Modeling Growth and Composition of Perennial Tropical Forage Grasses
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0006608:00001


This item has the following downloads:


Full Text











MODELING GROWTH AND COMPOSITION OF
PERENNIAL TROPICAL FORAGE GRASSES

















By

STUART J. RYMPH


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2004
































Copyright 2004

by

Stuart J. Rymph
















ACKNOWLEDGMENTS

I wish to express my sincere thanks and appreciation to Dr. Kenneth Boote (my

supervisory committee chair) for introducing me to plant physiology and crop modeling,

and especially for the conversations that went a little off-topic and tied the theory back to

the field. I would also like to thank my committee members Dr. Lynn Sollenberger for

his advice on both agronomy and academia; Dr. Charles Staples for our conversations on

animal nutrition; a needed break from agronomy and a reminder of how I enj oy working

with dairies; Dr. Jim Jones for making engineering fun and showing me that there is

agriculture outside of the United States; and Dr. Tom Sinclair, the "sounding-board", for

practical discussions and providing a more skeptical point of view. I'd also like to

recognize the late Dr. Bill Kunkle for our many conversations on farming and ruminant

nutrition: it was like going home.

Kudos go to Dr. Jean Thomas for the expertise, labor, and conversation that made

the growth study possible and also enjoyable. Special thanks are offered to Dr. Paul

Mislevy at the Florida Range Cattle REC in Ona, for providing datasets and also for his

friendship, encouragement, and continued efforts to involve me in the practical side of

tropical forage production.

Special thanks go out to friends for their support and encouragement. Dr. John

Moore helped to convince me to take on this endeavor. Dr. Kenny Woodard provided

practical advice on growing forages, and the voice of a farmer, firmly grounded in reality

(something that only enhances a modeling project). He and Tony Sweat provided










comradery, lots of humor, and a view of Florida that I would otherwise never have seen

and would have been the worse for missing it.

Extra special thanks are due my wife, Dr. Mary Beth Hall, for her emotional and

unending technical support which included sharing her expertise in SAS, also a warm

dinner and a break from dissecting bahiagrass tillers. I also thank her for teaching me to

not discard theory if it doesn't have an immediately apparent use in the field. Finally, I

thank my mother and father, Thelma and Dr. Donald Rymph for instilling in me a healthy

curiosity, a strong work ethic, and the belief that practical experience can be one of the

best teachers.





















TABLE OF CONTENTS


page


ACKNOWLEDGMENT S ............. ......___ .............. iii...


LI ST OF T ABLE S ............. ...... ._ .............. viii...


LIST OF FIGURES .............. .................... ix


AB STRAC T ......__................ ........_._ ........xi


CHAPTER


1 INTRODUCTION ................. ...............1.......... ......


2 LITERATURE REVIEW .............. ...............5.....


Bahiagrass ............... ... .. ......... ............. .............
Perennating Organs: Rhizomes and Stolons ................. ...............7............ ...
Dorm ancy .............. ...............8.....
Photosynthesis .............. ...............12....
T he CROPGRO Model ............. ...... .__ ...............19..
Model Evaluation............... ...............3


3 BAHIAGRAS S GROWTH STUDY .....__.....___ ..........._ ...........3


Introducti on ............. ...... ._ ...............35...
M materials and M ethods .............. ...............36....
Results and Discussion .............. ...............42....
Plant Growth............... ...............42.
Photosynthesi s ................ ...............47.................
Conclusions............... ..............5


4 DEVELOPMENT OF CROPGRO SPECIES FILE PARAMETERS FOR
B AHIAGRA S S ............. ...... ._ ...............61..


Introducti on ............. ...... ._ ...............61...
M materials and M ethods .............. .. .... .......... .. .. ........6

Description of Data Sets Used to Fit Parameters ............. .. ...__...........63
Preparation of Datasets ................. ...............65........... ....
Results and Discussion .............. ...............67....












Photosynthesis Parameters .............. ...............67....
Root Parameters.................... .. ...............7
Carbon and Nitrogen Mobilization Parameters ....._____ .........__ ..............73
Vegetative Partitioning Parameters .............. ...............74....
Leaf Growth and Senescence Parameters............... ...............7
Phenology Parameters .............. ...............75...
Testing of Literature-Based Parameters ............_...... .__ .........._......76
Optimization ............ .... ..__ ...............79....
Testing of Optimized Parameters .............. ...............80....
Conclusions............... ..............8


5 ADAPTING CROPGRO TO MODEL PERENNIAL TROPICAL GRASSES:
STRUCTURAL CHANGES TO THE MODEL ......____ ..... ... ._ ...............97


Introducti on ............ ..... .._ ...............97...
M materials and M ethods .............. ...............99....
Results and Discussion ............ ..... ._ ...............103..
Storage Or gan ............ ..... ._ .............. 103....
Dorm ancy .............. ...............107....
Freeze Damage ............ ..... ._ .............. 114...
Photosynthesis ............ _.. ........_ .... .._ ... .... ... ........ 1
Overall Model Performance Herbage Mass and N Concentration .................1 23
Conclusions............... ..............12


6 SUMMARY AND CONCLUSIONS ................ ...............151...............


Bahiagrass Growth Study .............. .. .... ........ ..........5
Development of Species File Parameters for Bahiagrass .................. .. ............... ...152
Adapting CROPGRO to Model Perennial Tropical Grasses: Structural Changes to
the Model .............. .... ...............153.

Implications of the Research ........._._.. ....... ...............155...
Future Research ........._... ...... ___ ..............._ 156...


APPENDIX


A CROPGRO CSM PARAMETER CODE DEFINITIONS ................ ................. .158


B SPECIES, CULTIVAR, AND ECOTYPE FILES FOR THE UNMODIFIED CSM
VERSION OF CROPGRO ............ ......__ ...............162..


C SPECIES, CULTIVAR, AND ECOTYPE FILES FOR THE FORAGE VERSION
OF CROPGRO .............. ...............183....


D NEW PARAMETER CODE DEFINITIONS FOR THE FORAGE VERSION OF
CROPGRO .............. ...............198....


E DORMANCY AND STOR CODES AND DEFINITIONS FOR DATA.CDE FILE203












F CODE ADDITIONS AND CHANGES IN THE FORAGE VERSION OF
CROPGRO .............. ...............206....


LIST OF REFERENCES ............ ..... ._ ...............307...


BIOGRAPHICAL SKETCH ............ ..... ._ ............... 15...

















LIST OF TABLES


Table pg

3-1 Schedule of sampling and harvest activities. ............. ...............52.....

3-2 Weekly averages of daily temperatures and daily solar radiation and total weekly
rainfall + irrigation water applied to bahiagrass grown at the Irrigation Park,
Gainesville, FL -2001 .............. ...............52....

3-3 Results of statistical comparison of treatment effects on plant growth and
photosynthesis. Period means are least squares means. Signifieance determined
by ANOVA for Period and orthogonal contrast for Week and Per X Week
interaction. ............. ...............53.....

4-1 Bahiagrass parameter values for the CROPGRO species Eile. Preliminary values
were derived from the literature. Optimized values were derived from
optimization runs made based on the preliminary values. .............. .........._....85

4-2 Evaluation of the performance of CROPGRO with literature-based and optimized
species Eiles, with and without winter photosynthesis reduction. ..........................87

5-1 Summary of performance of the forage version of CROPGRO in simulating mass
of below-ground plant organs (kg DM hal. )................... .... ...........12

5-2 Summary of performance of the CSM (unmodified) and forage version of
CROPGRO in simulating five experiments to predict herbage mass, herbage N
concentration, and herbage N mass. ............. ...............129....

















LIST OF FIGURES


Figure pg

2-1 Modular structure and summary of model components of the DSSAT-CSM
cropping systems model............... ...............22.

3-1 Sod core as removed from the soil............... ...............54..

3-2 Example of a separated sub sample of bahiagrass after removing roots. ...............54

3-3 Total plant mass for established bahiagrass grown at Gainesville, FL from 18 July
to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. .................. ...............55

3-4 Root mass for established bahiagrass grown at Gainesville, FL from 18 July to 12
Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. ............. .....................5

3-5 Stolon mass for established bahiagrass grown at Gainesville, FL from 18 July to
12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. ............. .....................5

3-6 Stem mass for established bahiagrass grown at Gainesville, FL from 18 July to 12
Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. ............. .....................5

3-7 Leaf mass for established bahiagrass grown at Gainesville, FL from 18 July to 12
Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. ............. .....................5

3-8 V-stage for established bahiagrass grown at Gainesville, FL from 18 July to 12
Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. ............. .....................5

3-9 Leaf area index (LAI) for established bahiagrass grown at Gainesville, FL from
18 July to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. .........................58

3-10 Specific leaf area (SLA) for established bahiagrass grown at Gainesville, FL from
18 July to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001. .........................58

3-11 Leaf + Stem (green) area index (GrAI) for established bahiagrass grown at
Gainesville, FL froml8 July to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -),
2001 ..............._ ...............59......_......

3-12 Leaf photosynthetic rate for established bahiagrass grown at Gainesville, FL from
18 July to 12 Sept. (m) and 12 Sept. to 7 Nov. (0), 2001............... ..................5










3-13 Canopy gross photosynthetic rate adjusted to 1500 Cpmol Par m-2 S-1 (Pl500) for
established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept. (m) and
12 Sept. to 7 Nov. (0), 2001. ............. ...............60.....

4-1 Observed herbage mass (m), predicted herbage mass (--), water stress (-
-), and N stress (-) of bahiagrass grown with 468 kg N ha-l yr- at Ona, FL,
using the literature-based species file and the leaf-level photosynthesis option,
with a) No adjustment to winter growth, or b) 70% reduction in photosynthetic
rate and partial defoliation (frost) over the winter. ................ ..................8

4-2 Observed bahiagrass herbage mass (m) and predicted bahiagrass herbage mass
using the preliminary (literature-based, non-optimized) species Eile and the leaf-
level photosynthesis option (-), or daily canopy photosynthesis option (-).
.............................89.

4-3 Observed bahiagrass herbage N concentration (m) and predicted bahiagrass
herbage N concentration using the preliminary (literature-based, non-optimized)
species Eile and the leaf-level option (-), or daily canopy option (-). For
bahiagrass grown at Ona, FL with 468 kg N ha-l yr- .............. ..................9

4-4 Observed bahiagrass herbage N concentration (m) and predicted bahiagrass
herbage N concentration using the preliminary (literature-based, non-optimized)
species Eile and the leaf-level option (-), or daily canopy option (-). For
bahiagrass grown at Eagle Lake, TX with 168 kg N ha-l year-. ............ ...............91

4-5 Observed herbage mass (m), predicted herbage mass (--), water stress
(-), and N stress ( ) of bahiagrass grown with 468 kg N ha-l yr- at Ona,
FL, using the optimized species Eile with a winter defoliation, 70% reduction in
winter photosynthetic rate, and a) the leaf -level photosynthesis option or b) daily
canopy photosynthesis option. .............. ...............92....

4-6 Predicted vs. observed herbage mass of bahiagrass grown with 468 kg N ha-l yr-
at Ona, FL, and 168 kg N ha-l yr- at Eagle Lake, TX, using a) the leaf-level
photosynthesis option, or b) daily canopy photosynthesis option. ........................93

4-7 Observed herbage N concentration (m) and predicted herbage N concentration of
bahiagrass grown with 468 kg N ha-l yr- at Ona, FL, using the optimized species
Eile and the leaf-level option (-) or daily canopy photosynthesis option (-).94

4-8 Observed herbage N concentration (m) and predicted herbage N concentration of
bahiagrass grown with 168 kg N ha-l yr- at Eagle Lake, TX, using the optimized
species Eile and the leaf-level option (-) or daily canopy photosynthesis option
( ) .. .. ...............95

4-9 Predicted vs. observed herbage N concentration (g kg- ) of bahiagrass grown with
468 kg N ha-l yr- at Ona, FL, and grown with 168 kg N ha-l yrl at Eagle Lake,
TX, using a) the leaf-level photosynthesis option, or b) the daily canopy
photosynthesis option............... ...............96.










5-1 Schematic of daily partitioning of new growth among vegetative tissues for the
forage version of CROPGRO. ............. .....................130

5-2 Schematic of the calculation of potential mobilization of CH20 from leaf, stem,
root and STOR in the forage version of CROPGRO............. .._.........___....131

5-3 Schematic of the calculation of potential mobilization of N from leaf, stem, root
and STOR in the forage version of CROPGRO. ............. .....................132

5-4 Predicted vs. observed stolon mass for bahiagrass grown in the field in
Gainesville, FL in 2001(), and in temperature and CO2 gradient greenhouses at
360 CLL CO2 L^ (*), and 700 CLL CO2 L^ (0). ............. ...............133....

5-5 The a) controlling functions and b) seasonal expression of the predicted effect of
daylength on incremental (increase above baseline) partitioning to STOR(--)
or mobilization from STOR (- -) in the forage version of CROPGRO. .........134

5-6 Mobilization factors in the forage version of CROPGRO that affect mobilization
from STOR as a function of a) vegetative N status and b) LAI. ......................... 135

5-7 Predicted (m) vs. observed bahiagrass herbage mass for late-season harvests at
Ona, FL in the 1993-1994 and the 1995-1996 growing seasons using the
modified leaf-level photosynthesis option in the forage version of CROPGRO. 136

5-8 Schematic of freeze damage to leaves and stems and cold-hardening of STOR
tissues in the forage version of CROPGRO. ......____ ........__ .................1 37

5-9 Predicted CO2 COmpensation point for the CSM version of CROPGRO (
-) and for two hypothetical C4 Species with a Ci/Ca of 0.4 and a CO2
concentrating factor of either 3 (-) or 10 (-) predicted using the forage
version of CROPGRO .........._ _... .... ._ ...............138..

5-10 Relative CO2 COncentration effect on a) CO2 factor for LFMAX and b) QE for a
C3 Species in the CSM version (- -) and for a C4 Species in the forage version
of CROPGRO using a Ci/Ca of 0.4 and CCNEFF of 3 (--). .......................... 139

5-11 Relative temperature effect on a) CO2 factor for LFMAX and b) QE for a C3
species in the CSM version (- -) and for a C4 Species in the forage version of
CROPGRO using a Ci/Ca of 0.4 and CCNEFF of 3 (-) .............. ..............140

5-12 Predicted growth of bahiagrass components under 350 CLL CO2 L^ atmospheric
CO2 Stolon (- -), root (*****), and herbage (-) relative to predicted growth
under 700 CLL CO2 L^ atmospheric CO2 Stolon (- -), root (*****), and herbage
(-) using the forage version of CROPGRO. ................. ..................141

5-13 Observed bahiagrass herbage mass (m), predicted stolon (- -), root (*****), and
herbage mass ( ) for bahiagrass grown in a) Ona, FL with 468 kg N ha-l yr or
b) Eagle Lake, TX with 168 kg N ha-l yr-. ............ ...............142.....









5-14 Predicted stolon (- -), root (*****), leaf (--), and stem (-) growth for
bahiagrass grown with 468 kg N hal yil in Ona, FL in 1997 using the forage
version of CROPGRO ..........__....... .__ ...............143..

5-15 Observed herbage mass (m), predicted herbage mass (--), water stress ( )
and N stress (-) of bahiagrass grown with 468 kg N hal yil at Ona, FL.
Predicted using the leaf-level photosynthesis option in a) the forage version of
CROPGRO or b) the unmodified CSM version of CROPGRO. .........................144

5-16 Predicted (m) vs. observed bahiagrass herbage mass for five experiments, using
the modified leaf-level photosynthesis option in the forage version of CROPGRO.
............................145.

5-17 Observed/predicted bahiagrass herbage mass for a) the 0 kg N hal yrl (O and
-), 468 kg N hal yil (O and -), and 942 kg N hal yrl (A and *****)
treatments at Ona, FL and b) the 0 kg N ha' yi' (O and -), 168 kg N ha' yr'
(O and -), and 336 kg N hal yr (A and *****) treatments at Eagle Lake, TX.
............................146.

5-18 Observed bahiagrass herbage N concentration (m) and predicted bahiagrass
herbage N concentration using the forage version of CROPGRO and the leaf-level
option (-), or daily canopy option (-). For a) bahiagrass grown at Ona, FL
with 468 kg N hal yr or b) bahiagrass grown at Eagle Lake, TX with 168 kg N
hal yeail. ........... ...............147.....

5-19 Predicted (m) vs. observed bahiagrass herbage N concentration for five
experiments, using the modified leaf-level photosynthesis option in the forage
version of CROPGRO ..........__....... .__ ...............148..

5-20 Observed/predicted bahiagrass herbage N concentration for a) the 0 kg N hal yil
(O and -), 468 kg N hal yil (O and -), and 942 kg N hal yil (A and
*****) treatments at Ona, FL and b) the 0 kg N hal yil (O and -), 168 kg N
hal yrl (O and -), and 336 kg N hal yil (A and *****) treatments at Eagle
Lake, T X .. ............ ...............149.....

5-21 Predicted (m) vs. observed bahiagrass herbage N mass for five experiments, using
the modified leaf-level photosynthesis option in the forage version of CROPGRO.
............................150.
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

MODELING GROWTH AND COMPOSITION OF
PERENNIAL TROPICAL FORAGE GRASSES

By

Stuart J. Rymph

December 2004

Chair: Kenneth J. Boote
Major Department: Agronomy

In addition to their role as feedstuffs, perennial tropical forage grasses such as

bahiagrass (Paspalum notatum Fliigge) can play a maj or role in nutrient management on

livestock farms; recycling N from fertilizers and manure to produce feed and reduce the

importation of other feeds, while lowering potential N leaching. Balancing feed quality,

feed quantity, and nutrient recovery can be difficult. A computer model capable of

simulating forage growth, composition, and N-dynamics could be a useful management

tool. Farmers and consultants could test management practices virtually, then implement

those showing the most promise. Our obj ective was to develop a tool to predict the

growth and composition of bahiagrass that responds to environmental and management

inputs.

Bahiagrass sod cores were dug weekly for two 8-week regrowth periods (18 July to

12 September, and 12 September to 7 November). Plants were separated into leaf blades,

stem, stolon, and roots. Leaf and canopy photosynthesis were measured in each period.









Leaf photosynthetic rate was not different between periods. Leaf and stem growth and

rate of development of new leaves was reduced in the second period; however, stolon

mass increased dramatically starting in mid-October.

This information aided the development of species-specific parameters required for

simulating bahiagrass in the model, CROPGRO. In the process, limitations in the model

structure that prevented the prediction of realistic growth patterns were identified.

Despite the limitations, prediction of herbage mass was good, having an index of

agreement of 0.85, with slightly lower accuracy predicting herbage N concentration.

To address the model's limitations, we modified the CROPGRO source code to

include a storage organ (STOR), equivalent to a stolon, and added dormancy functions to

increase partitioning of growth to STOR and reduce mobilization from STOR and roots

under short daylengths. The freeze-kill function was modified, allowing gradual death of

leaves. The Rubisco specifieity factor in the leaf-level photosynthesis option was

modified for C4 photosynthesis. Model performance was improved, predicting realistic

seasonal growth patterns. Excessive N stress was predicted frequently, but the cause was

not identified. The forage version of CROPGRO performs realistically but should be

tested under cooler temperatures and finer-textured soils.















CHAPTER 1
INTTRODUCTION

Perennial tropical grasses such as bahiagrass (Paspalum notatum Flugge) and

bermudagrass [Cynodon dactylon (L.) Pers.] have long been the basis of beef production

in Florida. As environmental concerns have come to the fore over potential movement of

nitrate into Florida groundwater from dairy farms, there is a growing interest in use of

these grasses as part of the nutrient management systems on Florida dairies. In both beef

and dairy operations, the success of a forage program is measured by several criteria: it

must generate enough mass to feed the herd, it may need to utilize a minimum amount of

manure nutrients, and the forage produced must meet a significant proportion of the

nutritional needs of the animals consuming it. Managing forages to meet all of these

goals is a complex trade-off of yield and quality. Emphasizing forage quality generally

requires harvesting at young stages of maturity, with shorter regrowth periods and lower

yields, whereas managing for forage mass produces forage with reduced concentrations

of digestible nutrients.

Current management strategies for perennial tropical forages rely on harvesting at

regular intervals with no allowance for changing weather conditions. A 3 to 5 week

harvest interval will nearly always produce high quality forage (Staples, 1995) but often

the yields are unacceptably low. Delaying harvest when forage growth rates are low

increases yield but at what cost to quality? A system that could compensate for changing

temperature, rainfall, and soil fertility might allow farmers to harvest forage of more

consistent quality in reasonable quantities.









Such a system also could improve N cycling on the farm and reduce N leaching to

groundwater. The high growth potential and expansive root system of tropical grasses

such as bahiagrass and bermudagrass allow these species to not only uptake large

quantities of N from the soil but they may alter the seasonal pattern of N leaching as well

(Woodard et al., 2002). Properly managing the timing of fertilizer and manure

applications as well as the timing of forage harvests can reduce nutrient losses to the

environment while producing additional feed for the herd.

A crop model capable of predicting forage yield and composition along with N

leaching dynamics could allow evaluation of management practices (including harvest

strategies) before implementing them in the Hield. To provide output with the desired

level of detail, the model must have the ability to respond to a variety of environmental

(temperature, rainfall, daylength, soil moisture, etc.) as well as management (fertilizer,

irrigation, harvest schedule, etc.) inputs. It would also be advantageous if the model

could be easily adapted to simulate other forage species to avoid rewriting the model for

each new species to be modeled. The CROPGRO model meets both of these

requirements.

CROPGRO is a mechanistic or process-oriented model. That is, it predicts plant

growth by simulating many of the underlying biological processes. Some of the plant

processes simulated include photosynthesis, transpiration, senescence/mobilization of

plant tissue, and root uptake of water and nutrients. In addition several supporting

processes such as water infiltration and evaporation in the soil, soil organic matter and N

dynamics mineralizationn, nitrification, denitrification, and leaching), and biological N









Eixation are also simulated. The structure of the model also allows it to be readily

adapted to new plant species.

The model code itself is quite generic, including a minimum of crop-specific

relationships in the source code. Rather, parameters defining the species-unique

responses of the various processes are read from a set of species-specific input files. By

providing alternate parameter files, the behavior of the model can be adapted to simulate

other plant species. Parameter files have already been created to allow CROPGRO to

simulate several crop species including soybean (Glycine max L.), peanut (Arachis

hypogaea L.), dry bean (Pha~seolus vulgaris L.), faba bean (Vicia faba L.), and tomato

(Lycopersicon esculentum Mill.) (Scholberg et al., 1997; Boote et al., 1998a, 1998b,

2002). Modeling perennial tropical grasses poses challenges not encountered when

modeling annual grain or fruit crops. Perennial forages must persist from one year to

another, regrow each spring, and go dormant each fall. Additionally, forages are not

usually allowed to reach physiological maturity and may be harvested several times in a

single growing season. The current CROPGRO structure can accommodate each of these

challenges to some degree. However, some code changes are needed to achieve the

desired level of model performance for perennial forages.

The goal of this proj ect is to adapt the CROPGRO model code and species file to

simulate the growth and composition of bahiagrass over multiple, consecutive growing

seasons. Parameter values will be obtained from the existing literature when available,

directly measured in experiments when practical, and estimated by calibration when

necessary. The Cropping Systems Model or CSM version of CROPGRO which is part of

the Decision Support System for Agrotechnology Transfer (DSSAT) version 4 model









will be evaluated to quantify its ability to predict and determine any limitations for

modeling perennial grasses. Finally, portions of existing code will be changed and some

new code added to create an independent perennial forage version of CSM CROPGRO.

The build date for the code used in our proj ect is 11 July 2003. This is a pre-

release version, but the relevant code is substantially the same as the final version

released as DSSAT v4 (Hoogenboom et al., 2003). To be compatible with the release

version, the PESTCP module was updated, changing the forage cutting code from

MOWE to MOW and changing the stubble mass units from g (leaf+stem) DM m-2 to kg

(leaf+stem) ha- While efforts will be made to maintain compatibility with the existing

CROPGRO model and its input-output structure, the perennial forage version will be

considered to be a "new" model rather than a new version of the current model.














CHAPTER 2
LITERATURE REVIEW

Bahingrass

Native to South America (Ward and Watson, 1973) and northward to Mexico

(Scott, 1920), bahiagrass (Paspalum notatum Flaigge) is a perennial tropical grass that

spreads by seed and vegetative stolons. First planted at the Florida Agricultural

Experiment Station in May of 1913 (Scott, 1920), bahiagrass use has spread to over one

million hectares (2.5 million acres) in Florida alone (Chambliss, 2002). Initial reports

praised the grass for its palatability to cattle and especially for its persistence and capacity

to spread and form a sod even under heavy grazing pressure (Scott, 1920). This

persistence along with its ability to tolerate a wide variety of management systems has

been key to the popularity of this species.

While used primarily for grazing beef cattle, bahiagrass is also harvested as hay.

Annual hay yields of near 8000 kg ha-l may be expected when harvested every 4 wk

(Johnson et al., 2001) and may exceed 11000 kg ha-l when harvested every 6 to 8 wk

(Blue, 1973). Bahiagrass responds well to N fertilization, yielding 2700 kg ha-l yr- with

no fertilizer N (Beaty et al., 1964) and increasing to nearly 14000 kg ha-l yr- when

fertilized with very high levels of N and cut to a low stubble height (Pedreira and Brown,

1996a). The most prolific biomass growth occurs below the common cutting height of 5

to 10 cm. The prostrate growth habit of 'Pensacola' bahiagrass produces a large amount

of leaf low in the canopy, with reports of as much as 5600 kg ha-l of leaf DM ( Beaty et

al., 1964) and a leaf area index (LAI) in excess of 1.6 m2 leaf m-2 land remaining after









harvest (Pedreira and Brown, 1996b). Beaty et al. (1968) reported that nearly 40% of the

total forage produced by Pensacola bahiagrass was present in the bottom 2.5 cm of the

canopy. Depending on the fertility and age of the stand, an additional 9800 kg ha-l

(Beaty et al., 1964) or more of rhizome or stolon (the terms rhizome and stolon are used

interchangeably in the literature when referring to bahiagrass) mass were present at the

soil surface. Bahiagrass also has an extensive fibrous root system which may produce as

much as 19700 kg ha-l DM in just the top 15 cm of the soil (Beaty et al., 1964), although

a level near 4500 kg ha-l root DM (Burton et al., 1954) is more common. Pensacola

bahiagrass roots have been reported to extend more than two meters below the soil

surface (Burton et al., 1954; Chambliss, 2002). Although other cultivars may exhibit a

more erect growth habit or lower stubble mass in the seedling year than Pensacola, even

these "improved" cultivars may still have a harvest index of 10% or less (harvested

forage as a proportion of total above-ground biomass present at harvest) after the field

has been established for more than two years (Pedreira and Brown, 1996b). Such an

investment in biomass close to the ground and below-ground encourages the persistence

and quick sod formation that the species is noted for. It also can support the crop through

periods of stress, allowing it to adapt to a variety of management practices.

The most popular bahiagrass cultivar is Pensacola bahiagrass (Chambliss, 2002), a

West Indies-type bahiagrass, thought to have been introduced in ballast offloaded at the

Perdido Wharf sometime prior to 1926 and promoted after 1939 by Escambia County

Cooperative Extension county agent Ed Finlayson for its aggressive sod formation,

persistence, drought tolerance, and palatability to cattle (Finlayson, 1941). Pensacola is

distinguished by long, narrow leaves and tall seedheads. Aside from its persistence,









Pensacola bahiagrass is also more cold-tolerant than many other cultivars (Ward and

Watson, 1973; Chambliss, 2002).

The productive season for bahiagrass is April to November in North Florida. The

season starts in early March in South Florida but growth still slows in October

(Chambliss, 2002), suggesting that the decline in growth may not be solely temperature

dependent. Indeed, recent work by Sinclair et al. (2001; 2003) and Gates et al. (2001)

demonstrate a role of daylength in initiating winter dormancy.

Perennating Organs: Rhizomes and Stolons

While bahiagrass reproduces by seed, the plants spread and form a dense sod

through the growth of rhizomes (equivalent to stolons) (Ward and Watson, 1973;

Chambliss, 2002). Stolons allow the plants to spread and occupy more land area with

nodes on the stolons providing new growing points for additional tiller growth. New

tillers form at an axillary bud in response to the flowering of a nearby culm (Sampaio and

Beaty, 1976), maintaining tiller density and rejuvenating the stand. New tillers may also

form in response to changing light (daylength and light quality) with the onset of spring

or after a harvest (and the possible removal of apical dominance). This new growth

allows the stand to persist from year to year and under grazing or hay management

sy stem s.

The stolons also promote persistence by acting as a storage vessel for C and N

reserves, providing nutrients for growth under stressful conditions and promoting rapid

regrowth after winter or harvest. As was previously mentioned, the mass of stolons may

be several times the mass of leaves, and the reserves may last two to three years

(Chambliss, 2002), providing for new growth. Partitioning to, and mobilization from,

stolons, may at times account for the maj ority of the nutrient flows within the plant.









Studies of carbon allocation and movement in bahiagrass using 14CO2 have shown 50 -

60% of the 14C partitioned to stolons, with only 10% partitioned to new leaf growth

(Beaty et al., 1974). Patterns of mobilization are also affected by harvesting. The same

study reported that re-mobilization of 14C fTOm the surviving tissue ceased within 3 to 6 d

after a severe defoliation. No more than 20% of the 14C assimilated was re-mobilized, of

which only 10% moved to the leaves of the main plant while 20 to 40% went to the

stolons. Such isotope studies characterizing patterns of mobilization and partitioning to

new growth during spring regrowth have not been conducted, but general studies of fall

and spring plant growth suggest that there are regulatory roles for two additional factors:

daylength and temperature. Late and early season growth can be slowed by the cooler

temperatures associated with the fall, winter, and spring months. This is also the period

of the year when daylength is near its minimum.

Dormancy

Forage production from tropical grasses like bahiagrass and bermudagrass

[Cynodon dactylon (L.) Pers.] drops dramatically in the late summer and fall months in

the southeastern United States despite continued warm temperatures. Gates et al. (2001)

quantified this reduction in yield in bahiagrass by measuring forage growth through two

successive fall and winter seasons concurrently at Ona, FL and Tifton, GA. Mean

temperatures were 60C cooler in Tifton where multiple freeze events occurred while there

was only one night in the two growing seasons where temperatures dropped below 00C at

Ona. The pattern of seasonal forage production was similar for both sites; however,

production at Tifton was lower than for Ona. At Ona daily forage production (leaf and

stem mass recovered above the mowing height) decreased from 36 kg DM ha-l dl to 8 kg

DM ha-l dl between 23 September 1993 and 8 November 1993, a 78% decrease in









growth rate. The growth rate remained between 5 and 15 kg DM ha-l dl through 7

March, then increased to 35 kg DM ha-l dl by 6 April. Cooler temperatures in the

second growing season (1994-1995) resulted in a similar pattern of forage growth but

lower overall yields and a more rapid increase in growth rate in the spring. Although

there was no difference in growth rates between the three cultivars tested (Pensacola,

Tifton 9, and RRPS Cycle 18) on dates when minimum growth rates were realized,

regrowth of Pensacola in September and April was considerably lower than that of the

newer, higher yielding cultivar, RRPS Cycle 18. No information on the cause of the

seasonal reduction could be discerned from the study other than the generally lower

yields at lower temperatures.

Growers in Georgia have attributed the low productivity of tropical grasses in the

fall to a lack of rainfall and fertilizer (Burton et al., 1988). Investigating this hypothesis,

Burton et al. (1988) compared the forage yields of well-fertilized, irrigated and non-

irrigated Coastal bermudagrass (another C4 graSs that exhibits winter dormancy)

harvested every 24 d from 1 April through 27 October for three consecutive years. A

seasonal decrease in forage yield was observed in all treatments during the fall and

winter. The magnitude of the drop was quite dramatic with yields for the September -

October harvest being only slightly more than 1/3 the yield of the May June harvest

period despite slightly higher temperatures in the fall. Correlation analyses of the data

showed only a moderate relationship between yield and temperature (r-0.46) or growing

degree days (GDD) (r=0.37), thus, discounting the roles of temperature, rainfall, and

fertilizer in the seasonal yield reduction pattern. Two variables that did show a strong

correlation with yield were daylength (r=0.95) and daily solar radiation (r=0.93). It was










not possible to separate these two effects with this dataset. Stepwise regression was also

employed using individual growth period data for all years. The single variable model

with the best fit was daylength (r2= 0.64, 0.69, and 0.61 for all yields, irrigated yields

only, and non-irrigated yields, respectively). When daylength was excluded from the

model, the best fit for a single variable was rainfall + irrigation (r2=0.45). Adding solar

radiation to the rainfall model only increased the R2 to 0.57, while the daylength +

rainfall gave an R2 Of 0.73 for all yields and 0.77 for irrigated yields only. A model with

daylength and water deficit (pan evaporation rainfall) had an R2 Of 0.71 when fit to the

non-irrigated yield data. Stepwise regression did not compensate for the confounded

effects of daylength and total solar radiation but did lend more weight to the daylength

argument.

Marousky et al. (1992) attempted to separate daylength effects from solar radiation

effects in a study of daylength effects on turf-type bermudagrass cultivars. Long days

were simulated using the same 9-h daytime as the short-day treatment but the dark period

was interrupted with 4 hours of light from incandescent lamps. Interruption of the dark

period with light causes the reversal of many red/far-red phytochrome reactions; in effect,

simulating a much longer day than the total time exposed to light. Long-day plants

exhibited a greater degree of leaf extension but no change in plant dry weight or number

of stolons produced. The authors' rationale for the apparent conflict with the results of

Burton et al. (1988) was that cultivar differences resulted in different responses between

the long-leaved, forage-type Coastal bermudagrass and the fine-leaved, turf-type

cultivars .









Another possible explanation might be found in work by Britz et al. (1958) who

studied short-day (late-season) accumulation of starch in leaves of another tropical grass,

Digitaria decumbens Stent.. This accumulation of starch was associated with a decrease

in translocation of assimilate under short days. The authors used several methods to elicit

a long-day response. Gradual extension of the daylength from 7-h (short-day) to 14 h

(long-day) with full-spectrum, full-irradiance light (400-600 Cpmol m-2 S-1 photosynthetic

photon flux density) resulted in no decrease in starch accumulation with 9 h daylength

but a dramatic drop between 11 and 12 h of light with starch concentrations after 12 h of

light equal to levels at 14-h daylength. In contrast, interrupting the dark period with 0.5 h

or 2.0 h of full-spectrum light resulted in a partial (~40% and ~60%, respectively)

reduction in starch accumulation relative to a 14 h day. The full long-day response was

achieved when the 7 h light period was shifted from the first half of the 24 h period to the

second half. These results implied that timing of the light periods may be more important

than total time of exposure. To separate the effects of daylength and total solar radiation,

plants were exposed to 14 h of amber light (589 nm), having a good photosynthetic

spectra but lacking photomorphogenetically-active wavelengths. Photosynthate

production under long days of amber light was similar to that of the long-day plants

under full-spectrum light but starch partitioning was similar to that of the short-day

plants. This helped to solidify the conclusion that the starch accumulation was a

photoperiodic response but also indicated that it involved a more complex mechanism

than a single red/far-red phytochrome response. The different responses of the long day

and interrupted night treatments may indicate a similarly complex mechanism in the

short-day response of bermudagrass.









Sinclair et al. (2003) avoided these complications by extending the natural

daylength over field-grown Pensacola bahiagrass, 'Florakirk' bermudagrass, 'Tifton' 85

bermudagrass (Cynodon spp.), and 'Florona' stargrass (Cynodon nlemfuensis Vanderyst)

to 15 h using halogen lamps during the short daylength months. Light levels under the

lamps were equivalent to less than 2% of full sunlight and, thus, not considered to

contribute significantly to total daily radiation. While all four species had higher forage

production under the extended daylength treatment, bahiagrass showed the greatest

response. Extended daylength yields were frequently more than twice those of the

natural daylength treatment with yields six times greater on one harvest date. Despite the

increases, mid-winter yields under extended daylength were still quite low compared to

summer yields. Although forage growth increased, below-ground biomass production

was not affected by extending the light period. Another factor potentially related to

dormancy, total nonstructural carbohydrate (TNC) concentration in the below-ground

biomass (root + stolon), decreased during the short-daylength months in both the natural-

and extended-daylength treatments. All this paints a picture of fall dormancy in tropical

grasses being triggered, at least in part, by daylength. The exact mechanisms have not

been described but the relatively low forage production even under extended daylength as

well as the continued loss of TNC during the winter, point to a complex of factors

contributing to the response rather than a simple phytochrome-mediated response.

Photosynthesis

The large DM production of bahiagrass can largely be attributed to it being a

tropical grass species expressing the C4 photosynthetic pathway. Temperate grasses

express the C3 photosynthetic pathway wherein CO2 is fixed by ribulose 1,5-bisphosphate

carboxylase/oxygenase (Rubisco) in the chloroplasts to produce 2 molecules of 3-










phosphoglyceric acid (3-PGA; a 3-carbon acid) and, eventually carbohydrate via the

Calvin cycle. Rubisco is also an oxygenase, capable of fixing 02 aS well as CO2. When

Ol is fixed, one molecule of 2-phosphoglycolate and only one molecule of 3-PGA are

produced. Not only is there half as much 3-PGA produced when Oz is fixed, two

molecules of glycolate can be metabolized to release a molecule of CO2, hence the term

photorespiration referring to oxygenase activity. Both CO2 and 02 COmpete for binding

sites on Rubisco, with higher relative concentrations of 02 in the chloroplast resulting in

higher rates of photorespiration and reduced photosynthetic efficiency. In C4

photosynthesis, CO2 is fixed in the mesophyll cells (cells near the leaf surface) by

phosphoenolpyruvate carboxylase (PEPCase) to form oxaloacetate (OAA; a 4-carbon

acid). The OAA is subsequently converted to either malate or aspartate, depending on

the type of C4 pathway expressed in the species, and transported to the bundle sheath

cells isolated deep within the leaves. In the bundle sheath cells, the malate or aspartate is

decarboxylated near the chloroplast where Rubisco re-fixes the CO2 to produce 2

molecules of 3-PGA as in C3 photosynthesis. Energetically, C4 photosynthesis is more

expensive with the additional cost of 2 ATPs to fix HCO3- with PEPCase on top of the

cost of 3 ATPs and 2 NADPHs to fix one mole of CO2 in the C3 pathway (Kanai and

Edwards, 1999). However, PEPCase has no oxygenase activity and, thus, can be more

efficient than Rubisco in fixing CO2 in the presence of Oz. Also, by isolating Rubisco in

the bundle sheath and "shuttling" the CO2 to it, CO2 is concentrated around the Rubisco

and photorespiration is minimized. Under atmospheric conditions (~21% 02 and 370

ppm CO2), photorespiration in C4 plants may be on the order of 3% of the net rate of CO2

fixation, compared to rates of 54% observed in the C3 Species wheat (Triticum aestivum









L.) (Kanai and Edwards, 1999). Reduced photorespiration not only increases the

efficiency of Rubisco carboxylation, it also increases the light level at which light

saturated photosynthesis occurs, lowers the CO2 COmpensation point, increases quantum

efficiency (QE), changes the temperature sensitivity of both QE and photosynthesis, and

allows high photosynthetic rates at relatively low concentrations of leaf N or

photosynthetic enzymes.

The higher carboxylation rate resulting from CO2 Saturation of Rubisco and the

reduced photorespiration enables C4 plants to potentially attain higher photosynthetic

rates at high light. While C3 photosynthesis becomes light saturated at relatively low

light levels, the enhanced capacity of the C4 System can tolerate very high light levels

without becoming light saturated. Ludlow and Wilson (1971) compared leaf net

photosynthetic rates of tropical grasses (C4) with tropical legumes (C3) OVer a range of

illuminances. Consistent with the capacity to respond to higher light levels, the C4

grasses had net photosynthetic rates (37.8 Cpmol CO2 m-2 S-1 ) that were approximately

double those noted for the C3 legumes (17.7 Cpmol CO2 m-2 S-1). The light response

curves of the legumes reached a plateau at around 4000 5000 foot-candles of light

while the curves for the C4 graSses were only approaching saturation at the highest

illuminance of 10 000 foot-candles. Boote et al. (1999) measured leaf net photosynthesis

(Pn) near 28 Cpmol CO2 m-2 S-1 in established bahiagrass under atmospheric conditions

compared to approximately 20 Cpmol CO2 m-2 S-1 for the C3 legume rhizoma peanut

(Arachis glabrata Benth.). When the CO2 COncentration was doubled from 350 CLL L^1 to

700 CLL L- bahiagrass responded with only a 20% increase in Pn, about half of the 36%

increase measured for rhizoma peanut (Boote et al., 1999).









Higher photosynthetic efficiency also allows the CO2 COmpensation point (CCMP)

(atmospheric CO2 COncentration where the rate of CO2 uptake by photosynthesis equals

the rate of CO2 OffluX) Of C4 plants to be considerably below that of C3 plants. Bolton

and Brown (1980) recorded CO2 COmpensation points of 4-14 CIL L^1 for the C4 graSS

Panicum maximum, much lower than the values (47-59 CIL L^1) measured in tall fescue

(Festuca arundinacea Schreb.), a C3 graSs. Several other C4 graSses have been shown to

have CCMPs near 0 pIL L^1 including Vetiveria zizanoides (0 to 5 CIL L^)~, and a variety of

Cymbopogon species (0 to 3 CIL L^1) (Rajendrudu and Das, 1981).

Efficiency of light utilization may also improve with reduced photorespiration.

Quantum efficiency or quantum yield is the efficiency of leaf photosynthesis when

measured at low light and is generally expressed as Clmol CO2 ClmOl-1 absorbed photons.

This describes the initial slope of the photosynthetic response to light (light-limited

region). A frequently cited QE value for C3 Species is 0.05241Clmol CO2 ClmOl-1

absorbed photons, the average QE measured at 330 CLL CO2 L^ for seven C3 Species

(Ehleringer and Bjoirkman, 1977). Values for C4 Species are generally higher than those

for C3 Species and may range from 0.046 for Sorgha~strum nutans (Monson et al., 1982)

to 0.075 for Saccharum spontaneum (Meinzer and Zhu, 1998). Differences in efficiency

exist between the three types of C4 pathways (NAD-ME, NADP-ME, and PCK-type)

(Ehleringer and Pearcy, 1983) as well. Ehleringer and Pearcy (1983) found that species

expressing the NADP-ME pathway, [e.g. bahiagrass and maize (Zea mays L.)] exhibited

the highest QE with eight monocot species averaging 0.065 Clmol CO2 ClmOl-1 absorbed

photons. Reports of QE for other NADP-ME species range from 0.062 to 0.075 Clmol

CO2 ClmOl-1 absorbed photons for various sugarcane (Saccharum) species at 350 ppm









CO2 (Meinzer and Zhu, 1998). Quantum efficiencies for other C4 pathway types

recorded by Ehleringer and Pearcy (1983) were 0.064 Clmol CO2 ClmOl-1 absorbed

photons for five monocot species exhibiting the PCK-type pathway, and 0.060 Clmol CO2

Clmol-l absorbed photons for three NAD-ME species. It should be noted that both the

Ehleringer and Pearcy (1983) and Ehleringer and Bjiirkman (1977) studies measured QE

at 330 ppm CO2 rather than the customary 3 50 ppm CO2. The narrow range of values for

a given pathway when measured across many species suggests that QE may be a

characteristic that is highly conserved across species. This would imply that values for a

particular pathway type may be generally applied to other species exhibiting that

photosynthetic pathway and that the QE for bahiagrass is near 0.065 Clmol CO2 ClmOl-1

absorbed photons.

While the QE of C4 plants is generally greater than that of C3 plants at higher

temperatures, the QE of C3 plants may exceed those of C4 plants at temperatures below

300C (Ehleringer and Bjiirkman, 1977; Ku and Edwards, 1978; Monson et al., 1982).

The QE of C4 plants is temperature insensitive, decreasing very little as temperatures

increase while the QE of C3 plants falls dramatically as temperature increases (Ehleringer

and Bjiirkman, 1977; Ku and Edwards, 1978; Monson et al., 1982; Ehleringer and

Pearcy, 1983). The decrease in C3 QE at higher temperatures is attributed to increased

photorespiration (Ehleringer and Bjiirkman, 1977; Monson et al., 1982). The solubility

of CO2 decreases relative to that of 02 aS temperature increases, creating a condition

where Ol COncentration around Rubisco may be enhanced, resulting in increased

oxygenase activity relative to carboxylation. This effect alone cannot account for the

drop in QE of C3S at high temperatures (Ehleringer and Bjiirkman, 1977; Ku and









Edwards, 1978), rather it is likely the combined effects of the changing relative gas

solubilities and a changing affinity of Rubisco for CO2 and 02 (JOrdan and Ogren, 1984)

that favors 02 at higher temperatures. The CO2-COncentrating mechanisms of the C4

pathways provide a high concentration of CO2 relative to 02 in the bundle sheath

chloroplast such that photorespiration effects are not evident (Jordan and Ogren, 1984).

Consistent with a higher QE at high temperatures, the temperature optimum for

photosynthesis in C4 plants is generally about 100C higher than for C3S (Long, 1999).

Conversely, at lower temperatures, C4 photosynthetic rates may be below that of

comparable C3 plants. The higher temperature optimum may be explained by the lack of

photorespiration in the C4S, but the mechanism behind the decreased performance at

lower temperatures has been more elusive. In an extensive review of C4 photosynthesis

at low temperatures, Long (1983) explored the effect of low temperatures on several of

the steps in the C4 pathway(s). No single step could be shown to be the limiting factor at

low temperatures. Two enzyme steps were identified to be the most likely to limit C4

photosynthetic rate at low temperatures: pyruvate Pi dikinase (PPDK) activity or Rubisco

activity. Pyruvate Pi dikinase was suspected for its relatively low activity at all

temperatures and the dramatic increase in its activation time under cold conditions.

Rubisco limitation might occur due to the low quantities of the enzyme in C4 leaVCS along

with a possible inhibition of the CO2 COncentrating mechanism under cold temperatures.

In a more recent review of responses of C4 photosynthesis to various environmental

factors, Long (1999) revisits his investigation of low temperature effects, this time

refuting the argument for PPDK limitation of photosynthetic rate and concluding that

"there is no inherent flaw in C4 photosynthesis that prevents efficient operation at low









temperatures" (p. 240) and citing M~iscanthus as an example of a C4 Species capable of

persisting at high altitudes/low temperatures. Recent work by Pitterman and Sage (2000)

using two ecotypes of Bouteloua gracilis Lag., adapted to high or low elevations

implicates Rubisco capacity as limiting photosynthesis at temperatures below 170C.

They also identify PPDK activity or ribulose 1,5-bisphosphate (RuBP) regeneration as

potentially limiting photosynthesis between 200C and the optimum temperature (~370C),

and PEPCase capacity as potentially limiting photosynthetic rate at temperatures greater

than the optimum. That C4 photosynthesis could have a rate lower than the C3 rate at low

temperatures if Rubisco was limiting in both pathways was attributed to the low Rubisco

content of C4 leaVCS coupled with the added cost of carboxylation of PEP in the C4

pathway .

The low Rubisco content of C4 leaVCS (Pittermann and Sage, 2000) contributes to a

lower leaf N concentration and smaller proportion of soluble leaf N than is typically

found in C3 plants (Slack and Hatch, 1967; Crespo et al., 1979; Sugiyama and Hirayama,

1983; Usuda et al., 1984). While Rubisco may contribute over 40% of the soluble leaf N

in a C3 leaf, it represents only about 5-28% of soluble N in C4S (Slack and Hatch, 1967;

Sugiyama and Hirayama, 1983). The added N present in PEPCase (4-10% of soluble leaf

N (Slack and Hatch, 1967)) does not match the relative decrease in Rubisco

concentrations, resulting in a lower total quantity of photosynthetic enzymes (PEPCase

plus Rubisco) and lower total N concentration in C4 leaVCS compared to C3S (Slack and

Hatch, 1967; Brown, 1978; Raghavendra and RamaDas, 1993). Leaf N concentration can

be increased by increasing N supply to the plant. Sugiyama and Hirayama (1983) found

that Rubisco, PEPCase, and PPDK quantities all increased with increasing N supply.









However, while the concentration of PEPCase and PPDK increased (as a proportion of

DM), the concentration of Rubisco decreased relative to C3 plants even as leaf N

concentration increased with N supply. The high photosynthetic rates at low leaf N

concentrations result in a high N-use efficiency, a factor critical to the productivity of

many tropical grasses in low input systems. Conversely, their low crude protein (CP)

concentration coupled with high cell wall concentrations (largely a consequence of the

extensive vascular system associated with the Kranz anatomy and bundle sheath) are

primary contributors to the relatively low forage quality of most tropical grasses.

The CROPGRO Model

The goal of this section is to familiarize the reader with the maj or features of

CROPGRO pertinent to adapting the model for perennial tropical forages. Discussion

will be limited to a review of the pattern of information flow and summaries of some of

the plant-related subroutines. More complete descriptions of CROPGO have been

published by Boote et al. (1998a, 1998b) with an extensive review of the hedgerow

photosynthesis approach by Boote and Pickering (1994).

One of the primary obj ectives in developing CROPGRO was to have a model that

could easily be adapted to simulate the growth of different plant species. CROPGRO

was created as a way to consolidate the existing SOYGRO, PNUTGRO, and BEANGRO

models into a single program (Boote et al., 1998a). The three programs shared much of

their code, so merging the models primarily involved moving the parameters describing

species and cultivar traits from the actual code to external input files. This structure

allows other species to be modeled by creating new parameter (input) files.

CROPGRO was developed in the early 1990s as a stand-alone model but could also

be run under the Decision Support System for Agrotechnology Transfer (DSSAT) shell









(ICASA, 1998), allowing it to be linked to other crop modeling programs as well as

graphics programs to automate presentation of results. After DSSAT version 3.1 was

released in 1996, the CROPGRO code was reorganized into a modular structure. Code

for simulating different plant and soil processes was organized into individual

subroutines for each process. The new subroutines were designed to be executed in four

common steps initializationn, rate calculation, integration and Einal/summary) called by

the main model. The modular structure was designed to allow users to add new code or to

link code from other programs by inserting new modules into the CROPGRO structure.

This approach has been applied to the DSSAT shell such that CROPGRO is an integral

component of the Cropping Systems Model (CSM) (Jones et al., 2003). CROPGRO

serves as the crop template module (Hoogenboom et al., 2003; Jones et al., 2003); a

universal interface for modeling several different species. An overview of DSSAT

version 4, which incorporates the CSM version of CROPGRO, can be found in the

DSSAT v4 documentation (Hoogenboom et al., 2003) and in Jones et al. (2003).

The modular version of CROPGRO was the first version to include the soil organic

matter (SOM) transformation module based on the CENTURY model (Gij sman et al.,

2002). The only SOM transformation option available in earlier versions was an

adaptation of the PAPRAN model (Godwin and Jones, 1991). Both options address

mineralization from SOM as well as immobilization, nitrification, and denitrifieation of

N. The CENTURY option adds the capability of simulating decay of surface residues

and the movement of those nutrients into the soil profie. Pasture and other perennial

forage systems are often low-input systems relying heavily on recycled nutrients from

plant residues for continued productivity. Omission of surface litter nutrient pools from









the PAPRAN option may dramatically reduce the available soil N and C pools,

potentially resulting in chronic underprediction of plant growth especially over the multi-

year simulation periods typical for perennial forage systems. Thus, the CENTURY

option is better suited to modeling perennial forages.

CROPGRO and CSM incorporate several modules for simulating environmental

and management responses, soil N transformations, soil water availability, etc. (Figure 2-

1). Individual modules depicting different processes are executed once within each of the

four steps of the modular structure. This modular structure should enhance the future

development of these programs as it allows others to add new features by inserting their

own modules. Also, by separating the rate and integration steps, the order of execution of

the modules within each step is generally less critical, again, facilitating further

development.

In the initialization step, parameter values and simulation control information are

read from various input Hiles and initial values set for state variables (variables

representing the state of the system at the end of the day, variables that have a quantity

such as mass of roots or number of leaves). This step is run once per simulation,

although there are provisions to re-run each simulation for multiple years where the

initialization step is run once per repetition.

Several input files are used to set the parameters for a species and control the

execution of the simulation. Three Hiles contain the plant parameters: a species Eile, a

cultivar fie, and an ecotype Eile. The experiment file, or "X-file" controls the simulation

and is supplemented by weather files, a soil information file, and a pest file. Comparison

of predicted data to measured experimental results is automated, with the program





















P rim a ry M od ule s *Environmental Modifications

aPlanting

MIain Weather Havstn
Pro gra m
Irrigation

Start I ILFertilizer Application

Residue Placement


InitiaiRuzation LIod Ini Ma ag me t ol~yamc

ca II I Soll T em p era tu re

Seasonal I IISollWater
Initialization
Soll Nltrogen & Carbon

~Soll :
Rate (TeadntTemplate Crop Models

S Calculations MhoeduMe InsP rda Soyea
to perform each step
of processing and In Peanut
tu rn calls each of th e
Integration I IPrim ary Mo d ule s ) Dyba

Soll Plant -Atmosphe re
Tom ato

Output I IIOther crops


Pest& Disease Damage-
Sum mary
CROPGRO Plant Template Plant Modules
CERES Malze

End CERES Wheat

SUBSTOR Potato

1 Plant CERES Rice

Other crops







Figure 2-1. Modular structure and summary of model components of the DSSAT-CSM cropping systems model (Jones et al., 2003, p.

239).










reading the measured data from a time-course field data file or "T-file" or a field average

observational data file or "file A".

The species file contains species-specific parameters describing the response of the

crop to the environment as well as parameters describing growth and photosynthesis.

These parameters are set during model development and are not generally altered by the

user. Some parameters may be adjusted to reflect differences in behavior of different

cultivars and ecotypes via additional parameters in the cultivar and ecotype files.

Parameters in these files include factors such as differences in physiological time

between growth stages, relative differences in photosynthetic rate, and leaf size, among

others.

Other files, such as the X- and T-files, must be created by the user. A file X

contains information describing the simulation, including what weather files to use, what

soil type is present (along with site-specific soil profile information), location of the

experiment site, management information (planting date, fertilization and irrigation

schedules, harvest date, etc.), and simulation controls such as which photosynthesis

option to use, whether to predict potential growth (assume no stresses), or water-limited

growth, or water- and N-limited growth, and other options that determine modules of the

model to be used.

If the simulation results are to be compared to an actual experiment, measured data

for each treatment can be listed in a time-course field data file or "T-file" or the field

average observational data file (file A). Data from this file can be read by a graphics

program to plot the predicted variables against the measured data. The T-file also can be

used to list amounts of pest damage and the dates on which the damage is to be imposed.









Based on the pest codes used in the T-file, the Pest file determines what type of damage

to impose. The Pest file is generally written during model development, tailoring the

codes to the specific crop being modeled. Because the "harvest" date listed in the X-file

signals the termination of simulation in CROPGRO, periodic forage harvests must be

simulated using the pest code, "MOW" that is entered in the T-file. The user specifies

the amount of stubble mass to remain after a harvest or grazing event and CROPGRO

reduces the amount of leaf and stem proportionally to leave that amount of stubble at a

reduced V-stage. As with any pest damage, the leaf and stem removed by MOW are lost

from the simulated system, with the assumption that they were exported from the field.

While annual species may be simulated without enabling the Pest option, all forage

simulations must use the Pest option to create multiple harvests.

Other site-specific information that must be provided by the user is daily weather

information and soil profile information. Daily solar radiation, rainfall, and minimum

and maximum temperature data are listed in the weather files which are named according

to their location and the year for which they contain data. The soil file (SOIL.SOL)

contains soil profile information on the specified soil types. The user must add

information to create the soil for the simulation in a specified format. CROPGRO

requires all of the above mentioned files except the T-file and A-file to initialize and run

a simulation, unless the simulation involves multiple harvest, in which case the T-file is

required as well.

Initialization of a simulation begins on the simulation start date, which may be well

in advance of the planting date. Planting date can be the day that seed is sown or the day

that transplants are placed in the field. The transplant option can be used to initialize the









simulation with an established stand of grass. Sowing date triggers the start of plant

growth simulation. Separating the beginning of the simulation from the sowing date

allows the model to predict soil conditions on sowing day, which may be helpful if the

user has limited information about the actual conditions existing at the site on that day.

The rate step is run once on each day of the simulation and calculates the rate

variables (variables representing the amount of change in a state variable occurring over a

specified period of time usually one day or less) for the current day of the simulation.

For each simulated day, prior to running the CROPGRO plant template, the date, daily

minimum and maximum temperatures, daily rainfall, and total daily solar radiation are

read from the weather file, and the X- and T-files are checked for any management

operations (e.g. planting, irrigation, harvest, or pest damage) for the day. The weather

and management information is fed into the soil processes module, predicting the rates of

change of available soil water, NH4 and NO3-. The weather data along with plant and

soil information from the end of the previous day is used to calculate daily

evapotranspiration.

Predicting plant phenology or stage of maturity is a key component of CROPGRO.

Several parameters such as partitioning of new growth between plant organs (leaf, stem,

roots), and daylength sensitivity of development are dependent on or may vary with the

stage of maturity. Stage of maturity is a function of accumulated physiological time

which is a combination of time and temperature. Cardinal temperatures (minimum,

optimal, and maximum) for various stages of development of each crop are described in

the SPE file. Generally, there is no development at temperatures below the minimum or

above the maximum cardinal temperatures whereas plants mature at a faster rate as air









temperature nears the optimal range. This rate of change/progression towards the next

stage can be altered by daylength and water stress. Progress during vegetative growth is

typified by progressive increase in the number of leaves per plant (V-stage) and,

beginning with floral induction, by progression through a series of reproductive stages

(R- stage).

The daily rate of pest damage is also calculated in the rate step. Information on the

extent of damage is listed by the user in the "T-file" and CROPGRO uses the previous

day's plant mass and leaf area to calculate "actual" damage. Codes in the Pest fie allow

the user to specify the amount (kg DM ha l) or proportion of leaf and stem to be removed

or the reduction in assimilate production due to disease or pest damage. As with all other

rate variables, the pest damage rates are not deducted from the existing plant mass until

the integration step which signals the end of the current day. This ensures that all new

rates are based on the same conditions the plant mass at the end of the previous day.

The calculation of photosynthetic rate presents an interesting contrast with animal-

based models. Many animal nutrition models are sink-driven, predicting nutrient intake

as a function of body weight and animal performance while, in CROPGRO, the supply of

assimilate, a function of absorbed solar radiation and photosynthesis, determines plant

weight and performance (source-driven). One exception is that some plant species may

exhibit a "juvenile" period where limited seedling demand can feed back on

photosynthate production. CROPGRO offers two options for predicting photosynthesis:

a daily canopy photosynthesis option and an hourly leaf-level photosynthesis option.

The daily canopy photosynthesis option is the simpler of the methods and uses an

asymptotic exponential response to daily solar radiation to calculate the potential daily









photosynthetic rate. This pattern of daily photosynthetic response to light is defined by

two parameters from the species file; the maximum canopy photosynthesis rate and the

amount of PAR at which photosynthesis is 63% of maximum. Estimated light

interception of the canopy is a function of the predicted LAI on the previous day and the

canopy light extinction coefficient specified in the SPE file. The hourly leaf-level

photosynthesis option is more mechanistic (based on processes and stoichiometry) but

more mathematically complex.

The leaf-level option uses the hedgerow approach described in Boote and

Pickering (1994) to estimate potential hourly photosynthetic rates for sunlit and also

shaded leaf area portions of the canopy. The hedgerow approach uses parameters

describing canopy shape, height and width, leaf angle, row width and direction, latitude

of the site, day of year and time of day along with the predicted LAI to estimate light

absorption by sunlit versus shaded leaves. Hourly distribution of solar radiation and

temperature are estimated from the daily values provided in the weather file and further

divided into direct and diffuse components. Potential hourly leaf photosynthetic rate is

calculated using an asymptotic exponential equation where quantum efficiency (initial

photosynthetic response at low light) sets the initial slope of the response and the

maximum potential leaf photosynthetic rate is the asymptote. The photosynthetic rates

for the sunlit and shaded leaves are multiplied by their respective LAls and summed each

hour to calculate the hourly canopy photosynthetic rate. The 24 hourly rates are

integrated to yield the daily photosynthetic rate. In both the daily canopy and hourly leaf

photosynthesis options, the potential photosynthetic rate may be adjusted for cultivar









differences, temperature, leaf N concentration, leaf thickness (specific leaf weight, SLW),

atmospheric CO2 COncentration, and an incomplete canopy (light absorption).

In the integration step, updated maturity stages are calculated by computing the

daily rate of change for each day and adding this to the previous day's stage rating. All

other rates are processed similarly to calculate the new day's state variables. Much of the

mechanics of the integration step is an exercise in carbohydrate (CH20) accounting and

allocation.

Potentially available CH20 from stored reserves is calculated and added to the

daily photosynthate production to determine the maximum amount of CH20 available for

the day. The day's maintenance respiration costs and any assimilate loss due to pest

damage are subtracted from the total, the remainder being the amount of CH20 available

for nutrient uptake and growth. Potential CH20 demand for seed and shell growth, CH20

cost per g of new vegetative growth (based on proportions of leaf and stem predicted by

the new V-stage), as well as potential N demand to "refill" N that has been mobilized

from old tissue are calculated and subtracted from the remaining available CH20.

The amount of new growth that can be produced from the available CH20 depends

on the cost of that new growth which is a function of its composition. Composition of

total new growth is, in turn, determined by the partitioning of new growth between the

different plant organs (leaf, stem, roots, seeds, etc.) and the protein, carbohydrate

(comprised of cell wall and starch), lipid, lignin, organic acid, and ash concentration of

each organ's new growth. Parameters describing partitioning to organs and organ

composition are listed in the species file. Coefficients developed by Penning de~rries et

al. (1974) describing the cost to assemble each of these components (both direct cost of C









for C-skeletons as well as the energy used in the biochemical pathways to form them)

expressed in glucose equivalents are used along with the composition parameters to

calculate the glucose cost of new tissues. Crude protein or N concentrations of new leaf,

stem, and root growth can vary within a range of values set by three parameters; a

maximum concentration for new growth, a "normal" growth concentration and a residual

concentration left after senescence. If there is adequate CH20 available but available N

is limiting, new growth can occur at reduced N concentrations.

Nitrogen demand can be met by two different sources; "Actual" N uptake by the

plant, and if this uptake cannot meet the N demand for new growth, N mobilized from

vegetative tissues. Once the balance of N uptake and mobilization are calculated, the

CH20 cost of N uptake and any N mobilization is subtracted from the remaining

available CH20 to update the energy budget. If the crop is a legume, N-fixation is

estimated, the N added to the available N pool, and the CH20 cost subtracted from the

available CH20. The remaining CH20 is allocated to new growth; first to seed and shell,

then to vegetative growth.

The increased root length associated with the day's predicted new root growth is

calculated. The total root length and its distribution are used in determining nutrient and

water uptake on the succeeding day. Losses are also calculated; the day's predicted

senescence of leaves, stems, and roots are estimated as well as any damage due to

freezing temperatures.

Finally, all of the gains in new growth and losses due to senescence, pest damage,

and frost are added to the previous day's pools of tissue mass and a new total is

calculated representing the plant mass present at the end of the current day. This ends the










"day" and after outputs are printed to their respective files, the integration step ends and

the rate step is repeated the next day.

The rate and integration steps are repeated daily until the simulation is terminated

on the harvest date specified in the X-file or when a killing freeze occurs. At this time

the final step is executed to finish printing the output files. These are the basic mechanics

and information flow within CROPGRO. The structure has been shown to work well

with annual grain crops and legumes as well as tomato. In adapting the model to

simulate the growth of perennial tropical grasses, accurate parameters must be developed

for the species file. Any plant processes unique to these plants that are not already

included in the model may require re-definition of some of the parameters or even

changes and additions to the model code.

Model Evaluation

The ultimate measure of a model's performance is the user' s satisfaction with both

the accuracy of predictions and overall utility of the model. Understandably, such a

measure is difficult to quantify and is relevant only to the user that generated the rating.

Statistical approaches to quantify the accuracy of model predictions provide standardized

measures of model performance. Unfortunately, even these methods do not provide

completely clear-cut conclusions about the accuracy of model predictions. Use of vague

terms like "fairly close" in instructions for interpreting various measures impart an air of

skepticism around the use of some of these methods. Given these caveats, the use of

several different measures of performance to evaluate a model may present a more

complete picture of model performance than any single measure and allow the user to

weight individual results according to their priorities.









Two measures that are commonly reported in the literature are the sample

correlation coefficient r and coefficient of determination T2. The correlation coefficient

provides a measure of the linear relationship or closeness between predicted and observed

values. Interpretation ofr is quite general. An r of 1.0 indicates perfect prediction by the

model with positive values of r indicating some level of a positive correlation between

the predicted values (Pi) and observed values (Oi). Conversely, an r of 0.0 indicates no

correlation of the model to reality whatsoever and negative values indicate an inverse

relationship. The coefficient of determination is informally described as the proportion

of the variance of the observed values that can be accounted for by the model. This

measure has more utility in that it presents an idea of how thoroughly the model

represents the system. Statistical analyses demonstrating the level of significance of r

only proves that a linear relationship with a non-zero slope exists between Pi and Oi

(Snedecor and Cochran, 1989). The validity of this conclusion can come into question if

Pi and Oi do not meet the underlying assumptions required for the particular analysis used

(Willmott, 1981). In spite of their popularity, these measures provide little detail to

characterize the relationship between Pi and Oi.

A simple method of visualizing the relationship between Pi and Oi is plotting a

scatterplot of Pi (Y-axis) and Oi (X-axis), relative to a line designating a 1:1 relationship.

While not quantifiable, some relationships (e.g. consistent underprediction) become

apparent. Scatterplots also provide a common sense check for more sophisticated

methods of evaluation. If results of a test do not appear consistent with the results of the

scatterplot, the test should be re-evaluated. The relationship between Pi and Oi presented

in the scatterplot can be quantified using linear regression. The slope of the regression









line (a) and its Y-intercept (b) may provide evidence of systematic error in the model,

providing quantities that can be compared across models. A slope of 1.0 with a Y-

intercept equal to 0.0 indicates perfect fit of the model predictions. These results along

with the means ( P and O ) and standard deviations of the predicted values and observed

values should be considered for their own merit as well as their use in calculating other

measures when evaluating model performance.

Difference measures, derived from the fundamental quantity (P1-Ox) (Willmott,

1982), build on the statistical measures listed above to quantify bias and average error.

Root mean squared error (RMSE) describes the average difference between Pi and Oi.



RMSE = '1 -- (Eq. 2-1)


Also, RMSE can be readily compared against the mean of the observed values for

comparison of relative error. Both RMSE and its square (mean square error or MSE) can

be subdivided into systematic (RMSEs and MSEs) and unsystematic (RMSE, and MSEu)

components (Willmott, 1981):


f(P,-O,)" 1(Pi-Ox)1 i(P,-Pi)
MSE = MSs= S1 (Eq 2-2)
n n n

where n= the number of pairs of predicted and observed data, and Pi=aOl+b When the

systematic component is minimized, the model is predicting at its maximum possible

accuracy and the primary source of error is not model-related. An alternative

presentation is offered by Roseler et al. (1997) where mean square prediction error

(MSPE), which has the same mathematical definition as MSE (Neter et al., 1990; Roseler









---2
et al., 1997), is considered as the sum of three components: mean bias (O-P) line


bias Sllb~ and random variation around the regression line [S21], where


Sp2 and So2 ar the variances of the predicted and observed values. These measures

provide insight not just on the magnitude of error but also hint at the potential sources of

error.

Willmott (1981; 1982) proposed another measure of model performance that he

called an "index of agreement". This is referred to elsewhere as the d-index. The d-

index describes the degree to which the observed data are accurately estimated by the

predicted data. More formally, it specifies the degree to which the deviation of the

observed data around O corresponds with the deviation of the predicted data around O,

both in magnitude and sign.


1 (P -O1)2
d=1-~ (Eq. 2-3)



where P'1=P1-O and O'1=01-O Potential values of d range from 0 to 1, with 1.0

indicating perfect agreement between predicted and observed data and 0.0 indicating

complete disagreement. The sole assumption is that O is free of error so that all error is

contained in P'1 and O't The equation can be rewritten as

n*MSE
d=1- (Eq. 2-4)



for simplified calculation when MSE is known. The innovation of the d-index is that it

responds to both differences between predicted and observed data as well as some










changes in proportionality (Willmott, 1981). The d-index is an improvement on the

simple "r"; still, it is not an absolute measure of performance. As with the

aforementioned methods, the d-index should be evaluated in the context of knowledge of

natural variations in the system being modeled, the capabilities of the model, and an

awareness of the amount of potential error in the observed values used in the comparison.

No one of these approaches will be best in all situations, reviewing several of these

measures together will provide a more complete description of model performance. The

results should also be viewed in the context of the intended use of the model. If the

model is to be used to demonstrate the response to a change in the environment to a class

of students, a model that predicts a response of the correct direction but severely under-

or over-predicts the magnitude may be preferable to a more accurate model if the latter is

more difficult for the students to use. Users must decide for themselves what level of

performance is acceptable. Likewise, individuals will have their own views of which

approach is most appropriate to their interests.















CHAPTER 3
BAHIAGRASS GROWTH STUDY

Introduction

There has been a resurgence of interest in the cause of winter dormancy in tropical

perennial grasses, particularly bahiagrass (Mislevy, 1998; Gates et al., 2001; Sinclair et

al., 2003). To date, the primary emphasis of the research has been to identify the

conditions triggering dormancy. Traditionally, dormancy was thought to be related to the

cooler temperatures of fall and winter; however, daylength has recently been implicated

as the triggering condition (Gates et al., 2001; Sinclair et al., 2003). Along with

identifying the cause, quantifying the effects of dormancy may help us identify specific

characteristics associated with dormancy to both aid in identifying non-dormant

individuals as well as help develop management strategies to promote higher yield and

longevity of both current and new, non-dormant bahiagrass varieties that may be

developed.

Sinclair et al. (2003) presented growth and composition data at the organ level

(leaf, stem, and below-ground material) characterizing relative differences in growth and

composition between plants grown under normal or extended photoperiods. Their data

were quantified for each harvest, at 4 to 5 week intervals. Information quantifying

changes within a regrowth period, however, is scarce. More detailed growth analyses

conducted during the transition into dormancy may help identify some of the mechanisms

involved in the reduction of herbage growth associated with dormancy. Our obj ective for

this study was to document, in detail, weekly patterns of plant growth in late summer and









fall regrowth periods with concurrent measurements of leaf and canopy photosynthesis.

The purpose of this information is to help "fill in" some of the detail missed in other

studies, and needed for perennial forage crop model development.

Materials and Methods

This research was conducted at the Plant and Soil Science Field Teaching

Laboratory at the campus of the University of Florida, Gainesville (290 38' N, 820 22'

W) on an established bahiagrass sod during the summer and fall of 2001. Based on the

age of the stand, fine leaf texture and abundant seedhead production in June, the variety

ofbahiagrass was assumed to be Pensacola. The soil was an Arredondo fine sand

(loamy, siliceous, hyperthermic Grossarenic Paleudult). The experimental design was a

randomized complete block four replications each being 190-m2 plOts (PLOT).

Treatments were two 8-wk regrowth periods (PER) (18 July 12 September or 12

September 7 November). The crop was harvested to a stubble height of 10 cm every 8

wk starting 21 May. The 18 July harvest served as the staging harvest, establishing the

initial conditions and base stubble mass for the first growth period. A single sod core

sample was taken weekly (WEEK) from each of the four plots beginning on 20 July

(Table 3-1). Plant height (distance from the soil surface to the point where the leaves

curved over and began to hang down) was measured at six locations within each plot on

the same days that the sod cores were sampled.

While the bahiagrass had been established for several years, it had not been

fertilized or irrigated regularly in recent years. During our study all plots were fertilized

with a commercial blended 16-4-8 fertilizer including trace nutrients and slow-release N

at 78 kg N ha- 10 kg P ha- and 37 kg K ha-l every 8 wk beginning on 13 April.

Irrigation was provided as needed to prevent water limitation of plant growth via portable










impact sprinklers (see Table 3-2 for combined rainfall + irrigation). Weather data (total

daily PAR, minimum. maximum, and average air temperature and total daily rainfall)

were recorded on an automatic datalogger (CR10, Campbell Scientific, Inc. Logan, UT)

and are summarized in Table 3-2.

The sod cores measured 20 cm by 35 cm by 15 cm deep and were dug by hand

using a pair of narrow-bladed shovels (Figure 3-1). Loose soil was shaken by hand from

the sod core and any loose bits of roots and other plant material were recovered. All

material was placed in a plastic bag and immediately placed in a cooler for transport to

the laboratory. At the laboratory, each sample was thoroughly washed with a garden

hose over a 2-mm sieve to dislodge soil. Rinsed samples were placed in sealed plastic

bags and refrigerated until processed.

Roots were trimmed from the stolons using hand clippers, placed in a paper bag

and transferred to a 550C forced-air oven. A subsample of the remaining plant material

was set aside for detailed analyses. Both the remaining sample and subsample were

separated into live leaf, stem, stolon, and dead leaf components. Live leaves were

separated from stems at the ligule (if there was one) or where the leaf emerged from the

stem (if the leaf had no ligule). Using this separation methodology, the leaf sheath is

included in the stem fraction (Figure 3-2). Stems were separated from stolons at a point

where they naturally broke by hand. Dead leaves were peeled from the tillers and, thus,

included dead sheath material.

Material in the subsample was analyzed for leaf and stem areas (one sided only)

using a LI-COR model 3100 leaf area meter (LI-COR Inc., Lincoln, NE). Stems were

scanned intact (Figure 3-2), and not dissected into individual immature leaves. The









number of stems (tillers) per stolon was counted and the number of leaves per tiller was

recorded as the vegetative stage (V-stage) of each tiller. The presence of seedheads in

the subsamples were noted but as there were very few, no additional measures of

reproductive stage were recorded.

Upon completion of separation and measurements, the plant components were

dried until reaching a constant weight in a 550C forced-air oven. Care was taken to

remove loose sand before weighing. Leaf, stem, stolon, and root dry matter (DM) mass

(kg DM ha- ) was calculated from the combined sample and sub sample masses of leaf,

stem, stolon, and root, respectively, and the land area of each core. Specific leaf area

(SLA) (m2 leaf kg-l leaf) was calculated from the measured leaf area and leaf mass for

each subsample. Leaf area index (LAI) (m2 leaf m-2 land) was then calculated by

multiplying the SLA by the total leaf mass m-2 fTOm each core. A "green area index"

(GrAI) representing the total photosynthetic area per area of soil surface was calculated

using the sum of the leaf and stem area indices. The V-stage was calculated for each plot

by averaging the V-stage of all of the tillers in the subsample for each plot. The net

accumulation of each component as well as net change in V-stage (A Leaf, Stem, Stolon,

and Root mass, A V-stage) was calculated for each period by subtracting the WEEK 0

(stubble) mass from the WEEK 8 (final) mass.

Statistical analyses of the growth data were performed using the Mixed procedure

of SAS (SAS Institute Inc., 1987) with the model:

Yijk = CI + Ai + Bj + AiBj + Ck + AiCk + BjCk +Cijk

where C1 was the population mean, A was PLOT, B was PER, C was WEEK, and e was

the residual error for i=4, j=2, k=8. The net accumulation data were analyzed using a









simplified form of the same model; Yij = CI + Ai + Bj + eij. PLOT, and all of its

interactions were assumed to be random effects and therefore appeared in the SAS

random statement in the order presented above. Growth period (PER), WEEK, and their

interaction (PER x WEEK) were fixed effects. Means separation for PER was directly

from the ANOVA. Orthogonal contrasts were used to qualify significant responses to

WEEK and PER x WEEK. The oc=0.10 level was selected as the threshold for

determining the significance of all effects and contrasts.

Concurrent with the growth measurements, leaf and canopy photosynthesis

measurements were recorded four times during each growth period. Due to weather

constraints, these measurements were not evenly allocated throughout each growth

period, nor did they occur at the same time during each period. Leaf and canopy

photosynthesis measurements were made at midday using a LI-COR LI-6200 portable

photosynthesis system (LI-COR Inc., Lincoln, NE). Leaf photosynthetic (or carbon

exchange rate CER) measurements were made on fully expanded, healthy leaves under

full sun conditions (PAR >1600 Cpmol m-2 S-1) using a 0.25-L chamber. Photosynthesis

was measured for three leaves per plot on each sampling date. Two 15-sec measurements

of carbon exchange rate (CER) (Cpmol CO2 m-2 leaf s )~, stomatal conductance (mol m-2

leaf s^)~, and internal CO2 COncentration (CLL L^1) were recorded for each leaf. For canopy

photosynthesis measurements, the leaf chamber was placed "open" inside an aluminum-

frame, clear plastic enclosure. The frame enclosed a land area of 0.56 m2, with a total

volume of 0.49 m3. Canopy CER (Cpmol CO2 m-2 land s- ) measurements were made

under four levels of light varying from full sun to dark. The light level in the chamber

was regulated by placing cloths of varying opaqueness over the chamber. Approximate









light levels were: PAR > 1500 Cpmol m-2 S-1 (full sun), 600-800 Cpmol m-2 S-1, 200-400

Cpmol m-2 S-1, and 0 Cpmol m-2 S-1 (dark). Three 16-second measurements of carbon

exchange rate (CER) (Cpmol CO2 m-2 land s- ) were recorded at each light level. The

canopy enclosure was opened between each light level measurement to let the humidity

and [CO2] in the enclosure equilibrate with the atmosphere. Net CER in full darkness

was considered to represent canopy + root + soil (dark) respiration. Gross canopy

photosynthesis (Cpmol CO2 m-2 land s- ) for each light level was calculated by adding the

absolute value of the dark respiration to the measured net photosynthesis for each light

level .

We fit the canopy light response data to the asymptotic exponential model (Boote

et al., 1985):

P = Pmax [1- e(-E*P4R/Pm,)] (Eq. 3-1)

using TableCurve 2D v4 software (Jandel Scientific Software, 1996), where P = canopy

gross photosynthetic rate (Cpmol CO2 m-2 S-1), P;;;a = maximum photosynthetic rate in

saturating light (Cpmol CO2 m-2 S-1), QE = quantum efficiency or initial slope of the CO2

assimilation : incident PAR response (Clmol CO2 ClmOl-1 absorbed photons), and PAR =

photosynthetically active radiation (Cpmol photons m-2 S-1). We solved for Pmax and QE

and used the resulting values to estimate gross canopy photosynthetic rate at a light

intensity of 1500 Cpmol photons PAR m-2 S-1. By expressing photosynthetic rate for a

common light intensity we could compare the different treatments and days without the

variation due to changing light levels as the days and seasons progressed.

We also attempted to predict a light-saturated leaf photosynthetic rate (Asat) from

the canopy gross photosynthesis data. The measured leaf and stem areas as well as










canopy gross photosynthesis and corresponding PAR measurements from the canopy

light response measurements were input into the hedgerow photosynthesis model of

Boote and Pickering (1994), programmed in SAS, then PROC NLIN in SAS (SAS

Institute Inc., 1987) was used to solve for Asat using an asymptotic exponential function

and outputs from the hedgerow model.

Statistical analyses of the leaf photosynthesis data were performed using the Mixed

procedure of SAS (SAS Institute Inc., 1987) with the model:

Yijk = CI + Ai + Bj + AiBj + C + AiC + BJC + AiBjC + Dk + AiDk + BjDk + AiBjDk

CDk + AiCDk + BjCDk + ijk

where C1 was the population mean, A was PLOT, B was PER, C was day of period or

re growth (DAY), D was leaf number (LEAF) three leaves were measured in each plot ,

and e was the residual error for i=4, j=2, and k=3. PLOT and all PLOT interactions were

assumed to be random effects and therefore appeared in the SAS random statement, in

the order presented above. Growth period (PER), LEAF, DAY, and their interactions

were considered to be fixed effects. Since measurements were not taken every week or

even on the same day of the week, DAY was treated as a continuous variable and entered

as a covariate. Because there was only a single value for Asat, canopy Pl500, and

canopy respiration for each plot on each sampling day, LEAF was not included in the

analysis of these variables and a reduced version of the model was used for these

variables:

Yijk = CI + Ai + Bj + AiBj + C + AiC + BJC + eij

Least squares means were calculated for PER. Means separation for PER was directly










from the ANOVA. The oc=0.10 level was selected as the threshold for determining the

significance of all effects and contrasts.

Results and Discussion

Plant Growth

Statistically, total plant growth (A Total Plant Mass) was greater for PER 2 than

PER 1 (Table 3-3). The dramatic loss of total plant mass in PER 1 masked a net increase

in leaf mass, overshadowing the fact that canopy growth was greater in PER 1 than PER

2. The loss of total plant mass was driven almost entirely by a decrease in root mass in

PER 1. In PER 2 root mass stabilized (Table 3-3) and a net gain of total plant mass was

realized (Figure 3-3).

Initial root mass for PER 1 was high at 11 500 kg DM hal but dropped to less than

1/3 of the original mass by WEEK 7 of PER 1 (Figure 3-4). The consistent decline

during PER 1 and the relatively constant root mass from the end of PER 1 through PER 2

tends to dismiss random sampling error as the cause of this loss. Alternatively, the

Spartan maintenance of the site in past years may have played a role in this behavior.

There had been no fertilizer or irrigation applications over several years and the extensive

root system may have developed to more thoroughly mine the soil for water and

nutrients. At the time we started measuring plant growth, irrigation had been available

for three months and a second fertilizer application had just been applied. As PER 1

progressed, nutrients may have become available in sufficient concentrations that plant

needs could be met with a less extensive root system. Excess root mass may have been

mobilized for new shoot growth or merely allowed to die, resulting in the pattern of loss

observed. In contrast to this loss in PER 1, root mass remained relatively constant in

PER 2. There was a slight increase in root mass in WEEK 6 (Figure 3-4) which









coincided with a period of increasing stolon mass and, thus, could be related to a

dormancy-induced change in priority of assimilate partitioning.

A more obvious signal of approaching dormancy may be seen in the pattern of

stolon growth. Throughout PER 1 stolon mass remained unchanged at approximately

4700 kg DM ha-l (Figure 3-5). Stolen mass increased in PER2, peaking at 8980 kg DM

ha-l on WEEK 7 (Figure 3-5). This change in growth pattern, as evidenced by the

significant WEEK and PER x WEEK interaction effects, resulted in greater A stolon

mass in PER 2 (Table 3-3). This late-season shift in partitioning of growth towards

storage tissue may be part of a dormancy response to shorter daylengths. Increased

allocation of growth to stolons may have contributed to the lower increment of leaf and

stem mass observed in PER 2 (Table 3-3). The combination of increasing stolon mass

and stable-to-increasing root mass in PER 2 is in contrast to the observations of Sinclair

et al. (2003) who reported steady or decreasing below-ground plant mass for Pensacola

bahiagrass between 22 Sep. 1999 and 1 Dec. 1999 sampling dates at Ona, FL. Our

longer, 8-wk, harvest interval may have allowed greater accumulation of stolon mass

between harvests than the 4 to 5-week interval of Sinclair et al. (2003). The management

history of the two sites could also have had an influence, as judged by the exceptional

pattern of root growth observed at our site.

Changes not only in leaf and stem mass (Table 3-3), but also changes in the

"character" of the canopy were observed. Shoot growth during the fall season (PER 2)

was slower than in summer (PER 1) resulting in less stem and leaf dry weight

accumulation (Table 3-3, Figures 3-6 & 3-7) as well as fewer new mature leaves (Figure

3-8) at the end of PER 2. Stem weights were lower for all weeks in PER 2 compared to









PER 1 but the decrease in stem mass was not different between periods (Table 3-3). The

linear decline in stem mass in both periods and even the "bump" in stem mass in PER 1

(Figure 3-6) was, at least in part, an artifact of the partitioning scheme employed. Our

partitioning strategy grouped developing leaves, still encased in the sheath, with stems.

Once the leaves began elongating and emerged from the sheath, the leaves became part of

the leaf mass and the fraction of their mass that had previously been developing in the

sheath was lost from the stem mass. Very little stem elongation was observed except for

the few tillers that developed seedheads, leaving little opportunity to increase stem mass

as the plant matured.

Development of leaves in the fall regrowth period (PER 2) was quite different from

that in PER 1. Leaf mass increased in a quadratic manner (Table 3-3) to peak on WEEK

6 of both periods (Figure 3-7), with average leaf mass and A leaf mass slightly lower in

PER 2. In contrast, the A V-stage in PER 2 was only 3.41 leaves, less than half of the

7.65 leaves added in PER 1 (Table 3-3, Figure 3-8). Average LAI followed leaf mass

more closely than V-stage, and a quadratic progression in LAI development was

observed in both periods. The LAI in PER 1 was higher than that in PER 2 (Table 3-3).

Despite the slower development, the initial and final LAI were the same for both periods

(Figure 3-9). Thus, the same final LAI was achieved in PER 2 with only half as many

mature leaves as were observed in PER 1. It should be noted that only leaves with ligules

were included in the V-stage count but all leaf blade material extending from the leaf

sheath was included in the LAI measurements.

Accordingly, the SLA (leaf area per g of leaf mass) was slightly larger, indicating

thinner leaves, in PER 2, although a quadratic decrease in SLA was observed in both









regrowth periods (Table 3-3, Figure 3-10). Ghannoum et al. (2001a) also observed

seasonal effects on SLA in a variety of perennial C4 graSses grown in summer and winter;

however, the response was species-dependent, positive in some cases and negative in

others. Although peak daily solar radiation was not recorded, total daily solar radiation

was lower in PER 2 (Table 3-2). Our SLA values are in accord with the range of values

reported by Boote et al. (1999) (88 to 108 cm2 -1l) for greenhouse-grown bahiagrass at

this location. One might expect these levels to be lower (thicker leaves) as high SLAs are

generally associated with shaded leaves and leaves grown under reduced light levels such

as in greenhouses while our plants were grown outdoors in full sun. Our values are quite

low compared to SLAs reported for other perennial C4 graSses (Ghannoum et al., 2001a,

2001b) and even C3 graSses (Ryser and Wahl, 2001).

The mean SLA reported for 11 NADP-1VE-type C4 graSses grown inside a

glasshouse in summer, with midday PAR levels averaging 860 Cpmol m-2 S-1, was 314 cm2

g- almost five times the level observed in our study (Ghannoum et al., 2001a).

However, their plants were harvested only 46 d (approx 6.5 wk) after planting, much

younger than the average age of the leaves on our plants. In contrast, the harvest interval

employed by Boote et al. (1999) was longer than the current 8 wk. The fine-leaf structure

of Pensacola bahiagrass may predispose this cultivar to have a lower than average SLA;

however, the low SLA values merit further measurements on this species and its cultivars

grown in other locations.

Similarly, the LAI values that we are reporting are lower than those observed by

others. At the extreme are LAI values above 8.0 reported for bahiagrass by Agata

(1985a; 1985b), which are considerably higher than our values of 1.75 and 1.67 for









8-week re growth in PER 1 and PER 2, respectively (Figure 3-9), or even our GrAI values

which include stem area as well as leaf area (Figure 3-1 1) Unfortunately, the

methodology used by Agata (1985a, 1985b) to determine LAI was not clear, hindering

any further comparisons. Other reports give considerably lower LAI values. Pedreira

and Brown (1996b) reported LAI for stubble and 13-d regrowth for three populations of

bahiagrass grown in the field near Athens, GA. Reported values were averages for two

cutting heights (3.5 and 10 cm). This stubble LAI (comparable to our WEEK 0 values)

ranged from 0.42 for selection T14 in August to 1.68 for Pensacola bahiagrass in July

while re growth LAI values ranged from 1.67 for selection T14 in August to 2.30 for

Tifton 9 in July. These values compare very favorably to our values of 0. 19 for WEEK 0

(stubble) and 1.16 for WEEK 2 in PER 1 (Figure 3-9). Methodology for the Athens

study was similar to ours in that leaf area measurements were based on leaf lamina only;

however, their sample size was much smaller (20 leaves vs. 100-500 leaves per

replication in our study).

As the stem tissue is also green, leaf + stem or "green" area index (GrAI) may

present a more accurate measure of photosynthetically active plant area than LAI alone.

As with LAI, GrAI was higher for PER 1 than PER 2 at 2.49 and 1.85, respectively

(Table 3-3). Orthogonal contrasts showed the relationship between GrAI and WEEK to

be cubic (Table 3-3), this is likely an artifact of the variation in stem mass resulting from

our partitioning scheme and may not be a biologically relevant pattern. The time-series

change in GrAI (Figure 3-11) clearly illustrates, more so than LAI, a considerably slower

increase and overall lower photosynthetic area throughout most of PER 2. Like LAI,









though, initial GrAI was the same for both periods and final GrAI were also much closer

than for the middle of the regrowth periods.

The slower leaf growth rate cannot be attributed to differences in initial leaf mass

and initial leaf area as neither differed between periods, although the slower growth rate

would reinforce itself through lower leaf mass and LAI once regrowth began. The slower

development of LAI and GrAI may have decreased potential photosynthesis during much

of PER 2; however, final LAI values were similar or identical for both periods (Figure

3-9). Cooler temperatures and lower solar radiation levels (Table 3-2) likely were maj or

factors reducing fall growth rates. The increased partitioning of growth to stolon tissue

could also have reduced leaf growth in the second half of PER 2.

Photosynthesis

Some caveats apply to the photosynthesis results. First, due to a combination of

equipment repair and availability issues, photosynthesis measurements for PER 1 were

made with a different LI-COR LI-6200 than was used in PER 2. However, both

instruments were calibrated using the same reference gas and procedures. Second, the

complete set of leaf photosynthetic measurements (two 15-s values for 3 leaves per plot)

included a few values that were not physiologically realistic. The leaf data were analyzed

to identify and remove outlying data points. Of the 197 photosynthesis measurements,

one was more than 1.5 interquartile ranges from the 75th percentile and one was more

than 1.5 interquartile ranges below the 25th percentile; both data points were removed

from the analysis These were the only data points removed from the analysis. The

values removed were quite extreme (gross photosynthesis levels of 70.35 and -20.47

Cpmol CO2 m-2 S-1) and may have been the result of air leaking into the leaf chamber

during measurement, beginning measurements before CO2 COncentration in the chamber









had begun to drop, or simply due to high sensitivity to the very small leaf area in the leaf

chamber (approximately 1 cm2) and a correspondingly small drop in [CO2] Over the

measurement period causing excessive variation on measurements. Under the reduced

dataset (n=195), the covariate, DAY, was significant (Table 3-3) with leaf photosynthetic

rate being highest during early regrowth (Figure 3-12). Despite lower temperatures in

PER 2 and a positive leaf temperature to leaf photosynthesis correlation coefficient of

r=0.45, PER did not have a significant effect on leaf photosynthesis. Our measured leaf

photosynthesis values of 31.0 and 26.6 Cpmol CO2 m-2 S-1 for PER 1 and PER 2,

respectively, fall well within the range reported by Boote et al. (1999) (24.8 35.2 Cpmol

CO2 m-2 S-1) and Fritschi et al. (1999) (19.0 35.4 Cpmol CO2 m-2 S-1) for greenhouse-

grown bahiagrass at 350 CLL L1 CO2 COncentration at this site.

As expected, predicted leaf Asat was considerably higher than the measured leaf

photosynthesis values (Table 3-3). Like measured leaf photosynthesis, the predicted Asat

values were not different between periods, however, unlike the measured data, DAY did

not affect Asat. This would indicate that the maximum potential leaf photosynthetic rate

remained the same over the temperature range experienced in this study. There is

precedent for this. Asat has been shown frequently to decrease rapidly below 200C

(reviewed by Long [1983]); however, the lowest temperature recorded in the canopy

chamber during photosynthesis measurements was 270C, considerably higher, where the

impact may be slight and difficult to discern. This might also help explain why

temperature accounted for such a small proportion of the variation in leaf photosynthetic

rate. Although there is no way to verify the accuracy of our predicted Asat using the data









we collected, the values seem within reason as a handful of measured values were at or

above the predicted rates.

To allow us to compare photosynthetic performance across sampling dates and

periods despite the varying light conditions, the canopy photosynthesis data were fit to an

asymptotic exponential function and the results used to predict canopy photosynthesis at

1500 Cpmol PAR m-2 S-1 (Pl500). Model fit was good with an r2> 0.99 for most plots and

sample days and the lowest r2 for a plot/day was 0.89. Analysis of the adjusted canopy

data showed a higher canopy gross photosynthetic rate in PER 1 than PER 2 (Table 3-3,

Figure 3-13). This is consistent with the greater plant mass (particularly leaf and stem)

observed in PER 1. Interestingly, when we fitted a regression model using daily

maximum temperature (Tmax), leaf + stem mass (Greenkg), GrAI, and SLA to the Pl500

data, Tmax had the best fit (r2= 0.48) and the addition of either Greenkg or GrAI did not

significantly improve the fit of the model. The PER 2 average of 43.0 Cpmol CO2 m-2 land

s^l was within the range of 31.6 to 47. 1 Cpmol CO2 m-2 S-1 reported by Boote et al. (1999)

for Pl500 in greenhouse-grown bahiagrass canopy gross photosynthesis at 350 CLL CO2

L CO2, whereas our PER 1 rate of 55.9 Cpmol CO2 m-2 S-1 was outside this range but

below the highest rate of 60.7 Cpmol CO2 m-2 S-1 reported by Fritschi et al. (1999) for

Pl500 of greenhouse-grown bahiagrass in the establishment year. As our plants were

grown under full sun, a higher photosynthetic rate than for greenhouse-grown plants

would be expected. That Fritschi observed higher rates may be related to the differences

in the age of the stands, if the rates are different at all.

Concurrent with the higher photosynthetic rate in PER 1 were higher canopy + root

+ soil respiration levels. This measurement was based on the CO2 exchange rate









measured in total darkness. Respiration rate is dependent on both the amount of tissue

respiring as well as the temperature. In regression analysis using the "Backwards" option

in PROC REG of SAS (SAS Institute Inc., 1987), both Tair (air temperature) and

Greenkg contributed significantly to the model and, combined, they could account for

87% of the variation observed in respiration. Canopy + root + soil respiration rates were

24.0 Cpmol CO2 m-2 land s-l in PER 1 and 13.9 Cpmol CO2 m-2 S-1 in PER 2 (Table 3-3),

slightly higher than the levels reported by Boote et al. (1999) but within the range of

values reported by Fritschi et al. (1999) for bahiagrass grown at 350 CIL L1 CO2.

Respiration rates also varied by day of regrowth (Table 3-3) but since DAY was a

covariate, orthogonal contrasts could not be used to discern a pattern of response.

Conclusions

Winter dormancy, the seasonal depression of canopy growth, in bahiagrass often

has been attributed to a decrease in temperature. More recently, daylength has been

identified as having a role in triggering dormancy (Mislevy, 1998; Gates et al., 2001;

Sinclair et al., 2003). The obj ective of our study was not to identify the cause of

dormancy but rather to quantify growth and photosynthesis during the late summer and

fall in more detail than previous studies. Our study points out several changes in the

pattern of plant growth and photosynthetic rate that may illuminate parts of the

underlying mechanism of dormancy. Two key observations were the sudden increase in

stolon growth half-way through PER 2 and that the leaf photosynthetic rate was not

different between PER 1 and PER 2. The effect of air temperature could explain most of

the variation observed in canopy root soil respiration, but could explain only 25% of

daily variation in leaf photosynthesis or 50% of the variation in canopy photosynthesis

(data not shown). The rate of V-stage progression was dramatically lower in the fall










(PER 2), much more so than the reduction in leaf mass. Despite having fewer mature

leaves per tiller, initial and final LAI were nearly the same for both 8-wk regrowth

periods. Consistent with the lower leaf mass and similar LAI, SLA was higher in PER 2.

In the case of shaded leaves, the resulting higher SLA is associated with lower

concentrations of photosynthetic apparatus and lower potential photosynthetic rates,

however, our predicted Asat was not different between periods.

Likely, growth reduction during winter dormancy is the culmination of a number of

factors; reduction in growth rate due to lower temperatures, change in partitioning of

assimilate favoring storage tissue over leaf growth, and changes in leaf characteristics

related to lower light levels. Use of the results of this study to develop parameters for

modeling bahiagrass growth testing would allow exploration of "what-if scenarios and

possibly help us better understand how these factors interact to reduce forage production

during winter dormancy.





Table 3-1. Schedule of sampling and harvest activities.


Week
0
0
0
1
2
3
4
5
6
7
8


Per 1 Date
7/18/01
7/19/01
7/20/01
7/25/01
8/1/01
8/8/01
8/15/01
8/22/01
8/29/01
9/5/01
9/12/01


Per 2 Date
9/12/01
9/13/03
9/14/01
9/19/01
9/26/01
10/3/01
10/10/01
10/17/01
10/24/01
10/31/01
11/7/01


Activity
Mow to 10-cm stubble height
Fertilize plots
Sample stubble
Sample growth
Sample growth
Sample growth
Sample growth
Sample growth
Sample growth
Sample growth
Sample growth Final


Table 3-2. Weekly averages of daily temperatures and daily solar radiation and total
weekly rainfall + irrigation water applied to bahiagrass grown at the Irrigation
Park, Gainesville, FL -2001
Daily Temperature Rainfall Solar Radiation


Week Average
OC
1 25.9
2 26.0
3 25.6
4 27.6
5 28.1
6 27.4
7 25.9
8 25.4


Maximum
OC
31.8
32.8
31.3
33.8
34.2
34.1
33.1
31.3


Minimum
OC
22.7
22.2
22.2
23.2
23.4
22.1
22.4
22.7


+ Irrigation
Total (mm)
94.8
72.4
38.9
4.3
56.7
30.5
50.3
16.8

77.9
63.3
15.0
1.5
27.5
32.0
26.5
0.0


Daily Average
(MJm-2 day- )
17.9
16.6
16.4
20.0
21.0
19.1
15.5
13.9


Period 1


Period 2


22.9
24.0
19.8
22.6
21.1
23.6
16.8
19.1


27.9
30.1
25.9
28.5
27.8
29.3
24.7
26.1


19.1
20.0
14.5
18.5
16.4
20.2
11.3
13.8


13.3
14.7
17.0
12.9
14.0
11.8
14.4
12.1











Table 3-3. Results of statistical comparison of treatment effects on plant growth and
photosynthesis. Period means are least squares means. Significance
determined by ANOVA for Period and orthogonal contrast for Week and Per
X Week interaction.
Period Mean Statistics
Growth Variable 1 2 Period Week Per X Week
Leaf Mass (kg DM ha-' ) 2150 1700 ** **/qdr ns
A Leaf Mass (kg DM ha-l )a 3015 251 1*----
Stem Mass (kg DM ha-l ) 3690 1960 *** ***/lin ns
A Stem Mass (kg DM hal ) a -456 -1396 ns----
Stolon Mass (kg DM ha-l ) 4740 6870 *** */lin */lin
A Stolen Mass (kg DM ha-l ) a 612 2350 **----
Root Mass (kg DM ha-l ) 6270 3800 ***"/lin ***/lin
A Root Mass (kg DM ha-l ) a -8009 -344 ***----
Total Plant Mass (kg DM ha-l ) 16845 14325 ns ***/lin
A Total Plant Mass (kg DM ha-l ) a -4839 3121 **----
Canopy Height (cm) 29.8 22.5 *** ***/cub ***/lin
V-stage (number of fully emerged
leaves tiller- ) 3.52 1.70 *** ***/qdr ***/qdr
A V-stage (number of fully emerged
leaves tiller ) a 7.65 3.41 ***----
SLA (cm2 -1f ~ leaf) 64.8 73.3 **/qdr ns
LAI (m2 leaf m-2 land) 1.28 1.10 **/qdr ns
A LAI (m2 leaf m-2 land) 1.56 1.52 Ns----
GrAI (m2 leaf + stem m-2 land) 2.49 1.85 *** **/cub ns
A GrAI (m2 leaf + stem m-2 land) 1.84 1.12*----

Photosynthesis Variable Period Day Per X Da
Measured Leaf Photosynthesi s
(ymol CO2 m-2 leaf si) 31.0 26.6 ns *** ns
Predicted Max Leaf Photosynthesis
(Cpmol CO2 m-2 leaf s^l) 44.2 39.9 ns ns ns
Predicted Canopy Gross
Photosynthesis at 1500 Cpmol
photons (Cpmol CO2 m-2 land s^l) 55.9 43.0 ** ns
Canopy Root Soil Respiration
(pol CO2 m-2 land si) 24.0 13.9 ** *ns
*P<0.10, **P<0.05, ***P<0.01, ns=not significant, --- does not apply to this variable.
lin, qdr, cub = linear, quadratic or cubic orthogonal contrasts, respectively, are significant
" A values are net change over period = Week 8 values Week 0 values.

























Figure 3-1. Sod core as removed from the soil.


Stoli


Figure 3-2. Example of a separated subsample of bahiagrass after removing roots.











25000



S20000



m 15000



S10000

0

). 5000


01
0 1 2 3 4 5 6 7 8 9
Weeks of Regrowth


Figure 3-3. Total plant mass for established bahiagrass grown at Gainesville, FL from 18
July to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001.


14000


12000


10000


S8000


~6000


4000


2000 -



0 1 2 3 4 5 6 7 8

Weeks of Reg rowth


Figure 3-4. Root mass for established bahiagrass grown at Gainesville, FL from 18 July
to 12 Sept. (--) and 12 Sept. to 7 Nov. (- -), 2001.












12000


10000


8000





600


2 000


200


0 1 2 3 4 5 6 7 8
Weeks of Reg rowth


Figure 3-5. Stolen mass for established bahiagrass grown at Gainesville, FL from 18 July
to 12 Sept. (--) and 12 Sept. to 7 Nov. (- -), 2001.


6000


5000-


S4000-


3000 t


Ea 2000-

U 1000 -- i -- -





0 1 2 3 4 5 6 7 8

Weeks of Reg rowth


Figure 3-6. Stem mass for established bahiagrass grown at Gainesville, FL from 18 July
to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001.











4000

3500




2000




S1500

S1000 1//



50

0 1 2 3 4 5 6 7 8
Weeks of Reg rowth


Figure 3-7. Leaf mass for established bahiagrass grown at Gainesville, FL from 18 July
to 12 Sept. (--) and 12 Sept. to 7 Nov. (- -), 2001.


6

e5







21
0


i ,g


0 1 2 3 4 5 6 7 8
Weeks of Reg rowth


Figure 3-8. V-stage for established bahiagrass grown at Gainesville, FL from 18 July to
12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001.
















2








0 .5





0.


0 1 2 3 4 5 6 7 8

Weeks of Reg rowth


Figure 3-9. Leaf area index (LAI) for established bahiagrass grown at Gainesville, FL
from 18 July to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -), 2001.

120








400






240


4 5
Weeks of Regrowth


7 8


Figure 3-10. Specific leaf area (SLA) for established bahiagrass grown at Gainesville,
FL from 18 July to 12 Sept. (--) and 12 Sept. to 7 Nov. (- -), 2001.

















'CI
S
2.5
E
E 2
u,

F 1.5


E
1

(3
0.5


0 1 2 3 4 5 6 7 8

Weeks of Reg rowth



Figure 3-11. Leaf + Stem (green) area index (GrAI) for established bahiagrass grown at
Gainesville, FL froml8 July to 12 Sept. (-) and 12 Sept. to 7 Nov. (- -),
2001.


O
O O


30


S25


O 20




10


Days of Reg rowth


Figure 3-12. Leaf photosynthetic rate for established bahiagrass grown at Gainesville, FL
from 18 July to 12 Sept. (m) and 12 Sept. to 7 Nov. (0), 2001.

































30 40
Days of Regrowth


50


t40


a

20


Figure 3-13. Canopy gross photosynthetic rate adjusted to 1500 Cpmol Par m-2 S-1 (Pl500)
for established bahiagrass grown at Gainesville, FL from 18 July to 12 Sept.
(m) and 12 Sept. to 7 Nov. (0), 2001.















CHAPTER 4
DEVELOPMENT OF CROPGRO SPECIES FILE PARAMETERS FOR
BAHIAGRASS

Introduction

CROPGRO is a mechanistic model that predicts yield and composition of crops

based on plant, soil, management, and weather inputs. As such, it appears well suited to

the task of modeling forage growth and nutrient concentration. Additionally, the ability

to simulate soil water and N balances, soil organic matter residue dynamics, and

pest/disease damage increase CROPGRO's utility as a tool for evaluating potential

environmental consequences of management changes. Its generic, process-oriented

design has allowed it to be adapted to model a variety of different species including

soybean (Glycine max L.), peanut (Arachis hypogaea L.), dry bean (Pha~seolus vulgaris

L.), faba bean (Vicia faba L.), and tomato (Lycopersicon esculentum Mill.) (Scholberg et

al., 1997; Boote et al., 1998a, 1998b, 2002). Adaptation is accomplished by changing a

set of parameters and relationships describing the species' response to environmental

variables. The procedure is described in Boote et al. (2002).

Kelly (1995) previously attempted to adapt CROPGRO to model the growth of

bahiagrass with the obj ective of using the model as a component of a system for

simulating peanut cropping systems. Simulation results were incorporated into an

economic model to predict the sustainability and profitability of the cropping systems.

The species, cultivar, and ecotype files developed were later released as a "pasture"

model in DSSAT v 3.5 (ICASA, 1998). Our application of this model to simulate data









sets of bahiagrass hay production revealed consistent overprediction of DM yields,

particularly in the cooler months of the year. More rigorous applications and objectives

for the use of the model impose different standards of accuracy and our proposed use as a

practical planning and teaching tool requires a more accurate prediction capability and a

more faithful representation of the seasonal patterns of growth of bahiagrass. The

obj ective of this work was to develop parameters, from searching the literature,

experiments, and calibration, to model bahiagrass growth and composition with the

CROPGRO CSM model.

Materials and Methods

In deriving model parameters to describe bahiagrass growth and composition, we

followed the general adaptation procedure described by Boote et al. (2002). Where

possible, parameters describing the basic processes of photosynthesis, respiration, N

assimilation, and plant development in bahiagrass were derived from the literature.

Parameters describing basic biochemical processes assumed to be conserved, or similar

(e.g.. growth respiration cost per unit of protein), among species are universal throughout

all CROPGRO species Eiles. For some less conserved processes and traits where data

were lacking, parameters from the CROPGRO soybean species Eile were used. Soybean

was selected as it is one of the original models used to develop CROPGRO and is

arguably the most tested and robust version of the model. Examples of parameters

incorporated from the soybean Eiles are lipid, lignin, organic acid, and mineral

composition, as well as carbon cost to mobilize N from senesced proteins (Penning de

Vries et al., 1974).

Where processes or parameters were believed to be divergent from soybean or

thought to be unique to perennial forage species, parameter estimates were interpolated









from literature data from other tropical perennial grass species or selected through

sensitivity analysis of the bahiagrass model.

We also developed an "optimized" set of parameters using a custom built

optimization program utilizing a "brute-force" optimization strategy. In the program, the

user specifies a minimum and maximum acceptable value for each parameter and the

desired number of "steps" between those limits for up to five parameters. Simulations

were run using all possible combinations of the specified parameters. Results from each

set of parameters were statistically analyzed for mean of simulated results, slope and

intercept of a fitted regression line of predicted and observed data, r2, d-index value

(index of agreement [Willmott, 1981]), and root mean square error (RMSE). Results

from all runs were saved to an output file and the combination with the lowest RMSE

was listed at the end. The output file was then exported to a spreadsheet, parsed, and

sorted from highest (best fit) to lowest d-index rating. The optimized parameter values

were selected using a combination of high d-index ranking, low RMSE value, and most

logical fit. The optimized parameter set was subsequently tested for fit, using the same

procedure as used for testing the literature-based species file.

Description of Data Sets Used to Fit Parameters

Two data sets were selected for use in fitting parameters and testing the optimized

model. Only a brief listing of growing conditions will be given here. A more complete

description of each data set may be found in the cited articles.

The study at Ona, FL was part of a three-species study of forage protein response to

N fertilization and cutting date (Johnson et al., 2001). The experiment was conducted at

the Range Cattle Research and Education Center (REC) at Ona, FL (270 25'N, 810 55'W;

elevation 27.4 m) on a Pomona fine sand (sandy siliceous, hyperthermic Ultic









Haplaquod) soil. Pensacola bahiagrass received five fertilizer treatments (0, 39, 78, 118,

and 157 kg N hal cutting- ), equivalent to annual applications of 0, 234, 468, 708, and

942 kg N ha-l supplied as ammonium nitrate. Fertilizer was applied on 5 May and on the

day after each cutting except for the October (last) harvests. Staging harvests marking

the beginning of each growing season were made on 5 May 1997 and 4 May 1998 with

successive harvests every 28 d until October. Forage yield and crude protein

concentration were measured for all but the staging harvests. Daily weather data were

acquired from the REC's weather station. Temperatures rarely dropped below 00C in the

winter. Rainfall totaled 1142 mm for October 1996 through September 1997 and 2110

mm from October 1997 through September 1998.

The Eagle Lake, TX (290 35'N, 960 20'W, elevation 46 m) experiment was part of

a larger study of N contributions of arrowleaf (Trifolium vesiculosum Savi) and

subterranean (Trifolium subterraneum L.) clovers overseeded on bahiagrass and

bermudagrass [Cynodon dactylon (L.) Pers.] conducted over the 1979-1981 growing

seasons (Evers, 1985). The study was located in southeastern Texas at the Texas

Agricultural Experiment Station at Eagle Lake on a Crowley fine sandy loam (fine

montmorillonitic, thermic, Typic Albaqualfs) soil. The Pensacola bahiagrass treatments

included fertilizer applied at annual rates of 0, 84, 168, 252, or 336 kg N ha- The

fertilizer was split into three equal applications made on or about 1 April, 1 June, and 1

August of each year. All plots were harvested monthly from May through October.

Forage yield and crude protein data were reported. Daily weather data were acquired

from the experiment station's weather station. Freezing temperatures were not

uncommon in the winter with minimum temperatures as low as -90C. Average rainfall









was less than for Ona, with annual precipitation of 1354 mm, 765 mm, and 1223 mm for

1979, 1980, and 1981 seasons, respectively.

The two data sets were split for optimization and testing. The two lowest N

treatments and the two highest N treatments from each site were used in the optimization

process. The middle N treatment from both data sets was reserved for testing the

literature-based and optimized species files. The rationale behind the splitting scheme

was to maximize the range of N fertilization and number of observed data pairs (108)

available for the optimization process. At the same time the test data sets would be most

indicative of how the model will perform under "normal" or the most frequently

encountered conditions. The primary objective for the optimization process was to

minimize RMSE for the prediction of herbage mass (leaf + stem weight) or herbage N

concentration (leaf + stem N concentration), depending on the variables being optimized.

Preparation of Datasets

There were no data available regarding initial plant mass or soil conditions for

either experiment, so actual initial crop condition could not be input into the model.

Instead, we estimated the initial conditions by running each simulation for one full

growing season/winter cycle prior to the measured seasons. Actual weather data were

used for the prior year. The season began with an established plant stand cut to the same

stubble height used in the measured years. Fertilization during the prior year was

consistent with the medium N fertilization treatment for each site (468 kg N ha-l yr- at

Ona, FL, and 168 kg N ha-l yr- at Eagle Lake, TX).

To compare simulated and observed growth, the two sets of results had to be

expressed on a common basis. The field studies reported yield as herbage (leaf + stem)

mass harvested above a base cutting or stubble height while simulation results reported









yield as the total amount of leaf and stem. The difference between the two is the amount

of leaf and stem mass in the stubble left after each harvest. Using the results of other

studies (Beaty et al., 1968; Pedreira and Brown, 1996b; Rymph and Boote, 2002), we

developed estimates of post-harvest stubble mass for the different cutting heights used in

the Ona and Eagle Lake experiments. These estimated stubble masses were added to the

reported harvest yields to approximate total herbage mass observed for these

experiments. Estimates for stubble mass left under 3.5-cm, 5-cm, 7.5-cm, and 10-cm

cutting heights were 1500, 1800, 2400, and 3000 kg DM ha l. These values may apply

only to Pensacola bahiagrass. Newer bahiagrass varieties with more upright growth

habits may have considerably less stubble mass (Pedreira and Brown, 1996b). This tactic

basically creates a consistent season-long offset while actual stubble mass may vary with

season and N fertilization. This approach represents a compromise between accuracy and

simplicity of implementation/utility.

Initial testing of the model revealed some characteristics of the CROPGRO

program code that were not compatible with a perennial forage. FREEZ l and FREEZ2

are parameters describing temperatures where all leaves fall off of the crop or the entire

crop dies (respectively) due to cold. We found that after a FREEZ l event occurred, there

was no regrowth of new leaves, resulting in the plants exhausting all reserves on

maintenance respiration and dying. The problem was related to the strategy used to end

photosynthesis of grain legumes after a foliage-killing freeze event. Since we could not

alter the code, both FREEZ l and FREEZ2 were set to -250C, essentially disabling the

FREEZ l function but allowing the simulation to continue through the winter.









Additionally, we simulated frost damage of leaves by partially defoliating the crop each

January using the PEST routine.

Results and Discussion

This discussion will be confined to parameters that were unique to perennial

tropical grasses or required redefinition or alteration in concept. A complete list of

parameter values is provided in Table 4-1.

Photosynthesis Parameters

CROPGRO has two options for predicting daily assimilate production: a daily

canopy option and an hourly leaf-level option. The daily canopy option is the more

simplistic approach, predicting photosynthate production as an asymptotic light response

to total daily solar radiation levels. The leaf-level photosynthesis option predicts hourly

photosynthetic rates for sunlit and shaded leaf area by simulating the dynamics of

Rubisco activity and electron transport and integrates them within the hourly hedgerow

approach to yield a daily canopy rate. Both options include adjustments for current

temperature, CO2 COncentration, and leaf N concentration conditions.

All previous efforts to adapt CROPGRO involved crops using the C3

photosynthesis pathway/mechanism. In contrast, bahiagrass expresses the C4

photosynthetic pathway, more specifically, it is an NADP-1VE type species (Hattersley

and Watson, 1976), the same pathway that is expressed in maize (Zea mays L.).

Concentration of CO2 in the bundle sheath chloroplasts through the "CO2 COncentrating

shuttle" contributes several advantages to C4 plants. Since CO2 is concentrated around

the CO2-fixing Rubisco enzyme in the bundle sheath chloroplasts, the relatively lower

solubility of CO2 at high temperatures is of little consequence, allowing higher rates of

carboxylation and suppressing oxygenation activity of Rubisco at higher temperatures









and light levels than is generally possible in C3 plants. Likewise, the quantum efficiency

(QE) is not temperature sensitive, not decreasing at high temperatures as in C3 plants. As

photorespiration is reduced, less Rubisco is required to maintain high carbon exchange

rates (CERs), and leaf protein levels are generally lower for C4 plants. These differences

must be reflected in our predicted patterns of photosynthetic response to light intensity,

CO2 COncentration, leaf N concentration, and temperature in the model.

The asymptotic light response curve used to predict daily canopy photosynthesis is

defined by two parameters; PARMAX the level of photosynthetically active radiation

(PAR) at which photosynthetic rate is 63% of maximum (moles [quanta PAR]m-2d- ), and

PHTMAX, the asymptote (maximum) of daily assimilation rate (g CH20 m-2 d- )

occurring at very high light (at least three times as high as PARMAX). These values are

not generally presented in the literature so preliminary values (Table 4-1) were estimated

as 150% of the corresponding parameter values used for soybean.

The lower rate of photorespiration observed in C4 Species results in a lower CO2

compensation point (CCMP) (atmospheric CO2 COncentration where the rate of CO2

uptake by photosynthesis equals the rate of CO2 CVOlution by respiration) in C4 plants.

Values of 0 to 14 CIL L-1 have been reported as the CO2 COmpensation point for various

C4 Species (Bolton and Brown, 1980; Raj endrudu and Das, 1981), most on the order of 0

to 5 CIL L^(~Rajendrudu and Das, 1981). Based on these results we selected a value of 5

CIL L^1 for CCMP which is used for the daily canopy photosynthesis option.

The leaf-level photosynthesis option is a more complicated system requiring

several more parameters than the daily canopy option, but the model at the leaf and

chloroplast level incorporates several conserved processes for which parameters may be









directly measured. Leaf quantum efficiency (QE) is typical of these conserved

parameters/processes. Quantum efficiency (parameter name PGEFF) or quantum yield is

broadly defined as the initial slope of the leaf CO2 aSSimilation:absorbed PAR response.

A value of 0.0541 Clmol CO2 ClmOl-1 absorbed photons (Ehleringer and Bjoirkman, 1977)

is typically used in CROPGRO for all C3 Species, including soybean. While the same

biochemical processes are used in both C3 and C4 photosynthesis, the CO2 COncentrating

effect of the C4 Systems increases their QE. Differences in efficiency exist between the

three variations of the C4 photosynthetic pathway (NAD-ME, NADP-ME, and PCK-type)

with NADP-ME species exhibiting the highest QE with an average QE of 0.065 Clmol

CO2 ClmOl-1 absorbed photons (Ehleringer and Pearcy, 1983). We selected this value for

the bahiagrass species file (Table 4-1). This value appears to be quite robust as it falls

well within the range of QE values predicted from bahiagrass canopy photosynthetic light

response data (0.054 0.081 Clmol-l absorbed photons) (Rymph and Boote, 2002), and

reported QE values (0.062 to 0.075 Clmol CO2 ClmOl-1 absorbed photons) for another

NADP-ME species, sugarcane (Saccharum spp.) (Meinzer and Zhu, 1998).

One inconsistency that remains is the relationship between temperature and QE. In

C3 plants, as temperature increases, the solubility of CO2 decreases relative to the

solubility of Oz, lOwering QE of C3 Species at high temperatures. Because of the high

CO2 COncentration surrounding Rubisco in bundle sheath cells of C4 plants, the effect of

temperature on QE is negligible. However, the temperature effect on QE is hard-coded

into CROPGRO and currently, there are no user parameters to modify that response.

The other parameter required is light-saturated leaf assimilation (LFMAX) for

leaves at high N concentration, 300 C, and a given specific leaf weight. We based our









estimate of LFMAX (and PGREF) on a predicted maximum leaf photosynthetic rate

developed from bahiagrass canopy light response data (Rymph and Boote, 2002) of

approximately 40.0 Cpmol CO2 m-2 leaf s^l. Relative differences among cultivars are

modeled by changing the ratio of LFMAX (maximum leaf photosynthetic rate for the

cultivar) to PGREF (maximum leaf photosynthetic rate for the species). As Pensacola

was the "reference" cultivar on which the species parameters are based and was the

cultivar measured, PGREF=LFMAX=1.760 mg CO2 m-2 S-1

The amount of photosynthetic enzymes in the leaf affect photosynthetic rate as

well. Generally, higher N concentrations in the leaves are correlated with higher levels of

these enzymes and higher photosynthetic capacity. Bahiagrass and other C4 graSses are

generally considered to have low concentrations of N in the leaves, yet maintain high

photosynthetic rates. Thus, optimal N concentrations for photosynthesis of bahiagrass

are likely to be lower than for soybean. We could find no reports of the minimum N

concentration required for photosynthesis [FNPGN(1)], so we defined this lower

threshold of the N response function [FNPGN(1)] from the lowest reported leaf N

concentration 7.6 g N kg-l leaf (Beaty and Tan, 1972). Sugimoto and Nikki (1979)

observed a curvilinear increase in bahiagrass leaf photosynthetic rate as leaf N

concentration increased from approximately 20 up to 30 g N kg- The rate remained

constant from 30 g N kg-l to almost 40 g N kg- Hence, we chose a curvilinear

(quadratic) shape to define the response of bahiagrass to leaf N concentration, with

optimum photosynthetic rates beginning at 30 g N kg-l [FNPGN(2)] and no decline in

rate at higher N concentration. The 30 g N kg-l optimum was also used for LNREF, the

N concentration at which PGREF is defined for the species.









The high concentration of CO2 arOund Rubisco in the bundle sheath chloroplasts

permits high photosynthetic rates at higher temperatures than typically observed in C3

plants. Although the mechanism is not well understood, C4 Species generally also have a

greater sensitivity threshold for low temperature reduction of photosynthetic rate than C3

species (Long, 1983; 1999)). Thus bahiagrass should have a base temperature required

for photosynthesis that is higher than soybean and it should have higher optimum and

maximum (highest temperature at which photosynthesis occurs) temperatures as well.

Several studies have been conducted to quantify the cardinal temperatures for tropical C4

grasses (Ludlow and Wilson, 1971; Wilson, 1975; Unruh et al., 1996); unfortunately

none included bahiagrass. Our interpretation of these results are that the optimum range

for leaf photosynthesis for a tropical grass species should be between 35 and 450C, with a

base temperature around 70C and a maximum critical temperature for zero rate near

550C.

The daily canopy calculations use a daily, rather than hourly, time step. To find

daytime temperature thresholds for the daily canopy option, we compared the threshold

temperatures for the leaf-level option to daily minimum and/or maximum temperatures

for Gainesville, FL and calculated a corresponding average daytime temperature. The

corresponding temperatures were: base temperature [FNPGT(1)], 120C, optimum range

[FNPGT(2), FNPGT(3)] from 250C to 380C, and maximum temperature of 500C

[FNPGT(4)].

Low temperatures may also have a prolonged effect on photosynthesis, affecting

photosynthetic rate after temperatures have returned to the optimal range. CROPGRO

uses another set of temperature parameters, FNPGL(1-4) and TYPPGL, to describe the









effect of minimum night temperature on the subsequent day's light saturated leaf

photosynthetic rate. West (1973) observed that Digitaria decumbens grown at 300C and

subj ected to just one night at 100C and returned to 300C, showed a 40% decrease in

photosynthetic rate compared to plants held continuously at 300C. Based on this, we set

the minimum temperature [no photosynthesis on the day after experiencing this

temperature FNPGL(1)] to 70C, optimum night temperature [no effect on subsequent

days photosynthesis -FNPGL(2)] to 180C, with a quadratic (curvilinear) response

between these points (Table 4-1).

Root Parameters

Bahiagrass poses an additional challenge to modeling its growth using CROPGRO

because a significant proportion of total plant mass is represented by stolon mass and

CROPGRO does not include a stolon organ in its structure. To include stolons in the

stem fraction would have confounded the computation of protein/N removed at harvest

and further complicated the estimation of stubble mass. Thus we redefined "roots" in the

model to include both stolons and roots. This "redefinition" without a code change

required considerable modification of the growth and senescence parameters relative to

those used for other species modeled by CROPGRO. The largest adjustment was for the

root length density (RFAC1) parameter (cm of root length per g of root). Stolons are

much thicker than roots and may represent more plant mass than the roots. Additionally,

N uptake per length of stolon (if any) is likely to be much lower than for roots, further

decreasing their "effective" length as a root. Based on the relative proportions of stolons

and roots reported by Rymph and Boote (2002), RFAC1 was reduced to 5000 cm g l,

33% lower than the value used for soybean roots.









As stolon mass is routinely mobilized to support new growth, the maximum

senescence rate (RTSDF) of the combined organs was increased from 0.01 to 0.02 or 2%

per day. In preliminary simulations this yielded a maximum predicted root mass of

approximately 10 000 kg root dry matter (DM) ha- in the range of the combined stolon

and root mass observed by Boote et al. (1999) (10 660 to 15 370 kg ha- ) and Rymph and

Boote (2002) (7155 to 15 740 kg ha- ).

Carbon and Nitrogen Mobilization Parameters

Another area where modeling perennial forages and annual grains differs is N

mobilization. The basic concept is the same, but the timing and purpose differ. Nitrogen

reserves in annual grain crops are generally mobilized for filling seed. Although many

perennial forages such as Pensacola bahiagrass may set seed, they are generally harvested

at a younger stage of maturity and reserves are used primarily to speed vegetative

regrowth after a harvest or in the spring. Since perennial forages must be able to do this

repeatedly over several growing seasons, the rate and extent of N mobilization may be

quite different than that observed in annual grain crops. Reports from Skinner et al.

(1999), estimating N and total nonstructural carbohydrate (TNC) mobilization in blue

grama grass [Bouteloua gracilis (H.B.K.) Lag ex Steud] during regrowth after cutting,

showed quite high rates of N mobilization. On average, 36% of the available N was

mobilized within 7 to 10 d of cutting. This translates into approximately 5% d-l or a

maximum available N mobilization fraction (NMOBMX) of 0.05. Total nonstructural

carbohydrate mobilization was also reported but the measured TNC concentrations were

2.5 to 3 times the levels found in bahiagrass, and we could not adapt them to our purpose

with any confidence. With no TNC data to support changes, the soybean values were

used for maximum available CH20 mobilization fraction (CMOBMX) (Table 4-1).










Vegetative Partitioning Parameters

During vegetative growth, partitioning of new growth among leaf, stem, and roots

is a function of the vegetative stage of the crop (V-stage). This is another area where the

concept of V-stage is different between annual crops and perennial forage crops. Annual

crops, as well as seedling forages, progress through the sequential increase in leaf

numbers in a relatively orderly fashion. Established perennial forages, however, are

periodically "re-staged" by harvests and frosts, interrupting the orderly pattern. As a

seedling, bahiagrass could reach a V-stage of 4 (four fully-expanded leaves) with a

relatively small root mass and few, if any, stolons. An established stand of bahiagrass,

with a relatively large root and stolon system capable of mobilizing significant amounts

of N, could also have the same V-stage rating of 4 after a harvest. In the CROPGRO

model, partitioning of subsequent growth is handled identically in both scenarios. A

unifying assumption is that, in both cases, if V-stage is low, the priority for partitioning is

towards growing leaf mass/area to establish photosynthetic capacity. As V-stage

increases, more DM may be partitioned to stolon and root. Additionally, since stolon and

root mass were combined, partitioning of new growth between organs required

modification from the proportions used for soybean.

While partitioning in seedlings may be measured by changes in leaf, stem, and root

mass over time, the presence of older, senescing material in established plants prevents

such a simple determination. Assuming that the model would be used most often to

predict growth of established stands, we developed the partitioning parameters around

observed patterns of regrowth, rather than purely on seedling growth. Parameter values

were estimated prior to optimization using growth patterns reported by Rymph (2002)

and Boote et al. (1999) and then refined by running simulations and manually adjusting










the parameters to match growth patterns and relative magnitudes of each organ (leaf,

stem, or root) (Table 4-1).

Leaf Growth and Senescence Parameters

Complications caused by repeated re-setting of the V-stage of the crop within a

growing season mandated some other modifications in addressing leaf growth-related

parameters. As the V-stage of the plants is reset after each harvest, there is potential for

V-stage to be quite low for a mature plant with numerous growing points stolonss). Use

of the VS SINTK function which allows photosynthesis and leaf expansion to be limited by

sink strength rather than assimilate supply, while potentially appropriate for a small

seedling, is not likely to fit the conditions of the older plant. To prevent potential

limitations to growth in older plants, the VS SINK function was disabled by assigning a

value of 0.0 to the VS SINK parameter.

Senescence parameters were modified very little from Kelly's (1995) pasture

model. The time constant for senescence (TCMP) was set to 25 thermal days based on

the weekly counts of dead leaves and weather from the raw data of Rymph and Boote

(2002). The light compensation point trigger for leaf senescence (ICMP), which triggers

leaf senescence due to shading of lower leaves was set to 0.8 moles photons m-2 d- the

same as soybean. In a similar vein, the V-stage trigger for senescence (when 12% of the

plant' s leaf number is assumed to have been senesced) (XSTAGE) was lowered from 14

leaves for soybean to 9. This was necessary because of the relatively low number of

leaves on a bahiagrass plant compared to a soybean plant.

Phenology Parameters

The actual resetting of V-stage after a harvest is done in the PEST routine, using

either the MOW function (remove herbage to a designated residual mass) or a









combination of the HARV and HRVS (remove a designated proportion of existing

herbage mass and number of leaves) functions, and no modification of the species file

was required. To implement the MOW function, the user supplies the harvest date(s) and

the amounts) of stubble mass to remain after harvest. On the harvest date, CROPGRO

then calculates proportion of canopy mass removed and leaf mass, stem mass, and

V-stage are each reduced by that proportion. The HARV and HRVS functions work

similarly except that the user sets the proportion of herbage mass to be removed (HARV)

separately from the proportion of V-stage lost (HRVS).

The influence of temperature on the rate of phenological development of bahiagrass

is not well documented in the literature. Therefore, we set the cardinal temperatures for

base (no new leaves or seeds), optimum (maximum rate of leaf or seed addition), and

maximum (upper failure temperature) points (Table 4-1) based on our experiences

growing bahiagrass (K.J. Boote, personal communication).

Testing of Literature-Based Parameters

Testing of the preliminary, literature-based, species file was encouraging with

d-index values of 0.843, 0.605, and 0.925 for accumulated herbage mass, herbage N

concentration and accumulated herbage N mass, respectively, using the leaf-level

photosynthesis option and 0.851, 0.531, and 0.907 for the daily canopy photosynthesis

option (Table 4-2). Despite the moderate d-index values for predicted herbage N

concentration, the r2 ValUeS were quite low at 0. 18 and 0. 19 for the leaf and canopy

models, indicating that the N response was a weak point. The higher d-index value for

predicted herbage N mass than for either herbage mass or herbage N concentration

represents the effect of offsetting errors (underestimation of yield coincident with an

overprediction of N concentration).









Reviewing the predicted pattern of growth, however, showed excessive rates of

winter and spring growth (Figure 4-la). Water and N demand associated with this

excessive growth caused elevated water and N stress throughout the spring and early

summer (Figure 4-la), reducing predicted growth rates in May and June. The principle

cause of this discrepancy was our failure to simulate winter dormancy. Compounding

this was the lack of a working freeze damage routine promoting an artificially high LAI

and photosynthetic capacity through much of the winter. CROPGRO has no provisions

for modeling dormancy so we attempted to duplicate the dormancy effect through other

methods.

To reduce winter growth rate we used the PEST routine in CROPGRO to reduce

daily photosynthesis production by 70% from 23 October through 30 March. This

approach reduces photosynthetic rate but does not concurrently reduce transpiration.

Coincident with this change, the periods of water and N stress were shortened

considerably (Figure 4-1b). Statistically, however, there were minimal changes in the fit

of either leaf or canopy models after this modification (Table 4-2). This approach

resulted in slight improvements in d-index values for leaf and canopy model predicted

herbage mass, leaf model predicted herbage N mass, and slightly lower (worse) d-index

values for canopy model predicted herbage N concentration and herbage N mass (Table

4-2). Late spring regrowth was still considerably reduced compared to the observed

growth (Figure 4-1b), possibly because the PEST option reducing photosynthesis does

not reduce transpiration except indirectly through the lower LAI resulting from slowed

growth.









Further investigation revealed that two mechanisms may have been responsible for

the early season water and N stress. In reducing photosynthesis in the PEST routine, the

normal photosynthetic rate and transpiration rate were calculated and then the

photosynthetic rate was reduced by the designated percentage. Transpiration, however,

was not reduced so water uptake continued at the normal (now excessive) rate, depleting

available soil water. The only "reduction" in transpiration was due to the lowered LAI

that resulted from the slowed growth. Also, root N was mobilized throughout the winter

and into the spring to compensate for the reduced assimilate production, reducing root

mass considerably by the end of the winter/early spring period (data not shown). While

more water and N may have been available in the soil, the diminished root system had a

reduced capacity to exploit them, suppressing early season growth rates. Despite the

failure to statistically improve the fit, we used this strategy in all optimization and testing

runs as the patterns of N stress were more realistic than before.

Performance of the model using the daily canopy option and winter photosynthetic

reduction was quite similar to the leaf-level option performance for predicting herbage

mass (Figure 4-2). The d-index values for the fit of the predicted data to herbage mass

were identical to those for the leaf-level option. The predicted pattern of growth was also

quite similar with slightly reduced winter growth rates but nearly identical summer

growth (Figure 4-2). Fit of predicted herbage N concentration was slightly poorer for the

canopy option, but fit of predicted herbage N mass was similar for both options (Table

4-2).

Fit of predicted herbage N concentration was not as good as was herbage mass for

either option (Table 4-2). Leaf + stem N concentration was consistently overpredicted









for the Ona, FL site (Figure 4-3). During winter regrowth, after the "simulated frost"

defoliation, herbage N concentration exceeded 40 g N kg- equivalent to 250 g CP (crude

protein) kg- higher than the "maximum" leaf N concentration set by PROLFI. This is

related to the N allocation problem in the code cited earlier that prohibited leaf regrowth

after a FREEZ l event. As the goal of the present exercise was to calibrate the parameters

without changing any source code, this problem could not be addressed.

For the Eagle Lake, TX data, predicted herbage N concentration appeared to follow

a more accurate pattern despite greater variation in the observed values (Figure 4-4).

Prediction of herbage N concentration was more balanced, being both over- and under-

predicted. The improved prediction pattern may be related to the lower fertilizer levels

used at Eagle Lake and the lower yields for that site. Values from the daily canopy

photosynthesis option were generally higher than for the leaf-level option (Figures 4-3

and 4-4).

Optimization

Since we were unable to accurately predict the spring growth pattern, some of the

early season yield data were excluded from the optimization process. The rationale for

"culling" these two data points was that the model was consistently predicting early

season N and water stress when there was none; thus keeping those data points in the

optimization would influence the final parameter values in order to compensate for the

predicted stresses. This left 52 data pairs for calibration. No data were excluded from

the datasets used to test the performance of the model. All testing runs used all of the

observed data available for the site/fertility combinations used (27 data pairs). The

distinction being made here is that we wanted to develop the most accurate parameters

for the model through optimization (hence leaving out the early season data) while









presenting a fair evaluation of the performance of the model through testing (by including

all data).

Our strategy was to first optimize the leaf-level option temperature parameters to

establish proper general patterns of growth, then refine the prediction by optimizing

parameters that affect the growth response to N. After optimizing the leaf-level option,

the temperature parameters for the daily canopy photosynthesis option were optimized

followed by simultaneous optimization of PARMAX and PHTMAX.

All temperature parameters (leaf-level and daily canopy options) and

PARMAX/PHTMAX optimizations used only the two highest N fertility treatments from

each experiment, assuming that N would not be limiting growth for those treatments.

The two highest and two lowest N fertilization treatments from each study were used in

optimizing the N parameters as this presented the broadest range of conditions.

Testing of Optimized Parameters

Optimization improved the predicted winter growth pattern (Figures 4-5a and 4-5b)

but fit of predicted herbage mass during the growing season was generally unaffected

(Table 4-2) with similar d-index values for both optimized and literature-based species

Eiles. Winter growth pattern was improved most for the daily canopy option where there

was almost no regrowth through the winter (Figure 4-5b). A wider range of potential

parameter values were offered in the daily canopy option optimization as there was less

data available to define the ranges. Winter regrowth was curtailed by increasing

FNPGT(1), only allowing growth on days with average temperatures greater than 200C

(Table 4-1). Normally, this would affect growth rates well into the spring and fall

growing season, but PARMAX and PHTMAX were also boosted in the optimization

(Table 4-1). This combination allowed growth rates to be nearly identical during the









normal growing season (Figures 4-5a and 4-5b). Nitrogen stress was reduced in the

optimized simulations (Figures 4-5a and 4-5b); however, water stress was still extensive

in the spring, even for the daily canopy option, resulting in a continued poor prediction of

first cutting regrowth at Ona, particularly in the second growing season. To compensate

for the failure of the model to properly simulate winter dormancy, the optimization

process promoted combinations of extreme parameter values to improve the fit of the

model .

Overall, both the leaf-level and daily canopy options tended to overpredict herbage

mass at lower yields and underpredict at higher yields with this phenomena most evident

in the daily canopy option results (Figures 4-6a and 4-6b). Despite the improved fit, the

optimized values for these three parameters, FNPGT(1), PARMAX, and PHTMAX, are

not realistic and reflect the emphasis on compensating for the excessive winter growth

pattern, not improving the relevancy of the parameter value.

Fit of herbage N concentration predictions improved considerably for both the

optimized leaf-level and optimized daily canopy photosynthesis options (Table 4-2,

Figures 4-7 and 4-8). The d-index rating for both options improved considerably after

optimization; however, the r2 ValUeS remained quite low (Table 4-2). Predicted herbage

N concentrations were still consistently overpredicted at Ona (Figure 4-7) despite

optimization toward lower PROLFI, PROLFG, PROSTI, and PROSTG parameters

(Table 4-1). The pattern of predicted herbage N concentration for Eagle Lake remained

realistic after optimization (Figure 4-8), however, the optimized parameter values were

generally lower than the literature-based parameters (Figure 4-4). The difference

between predicted and observed values was also reduced for both the leaf-level (Figure










4-9a) and daily canopy options (Figure 4-9b) indicating a more consistent prediction.

Both leaf-level and daily options tended to overpredict herbage N concentration, but this

was more pronounced in the daily canopy option predictions (Figures 4-9a and 4-9b).

Despite the improvement in herbage N concentration prediction, fit of predicted herbage

N mass showed little improvement (Table 4-2) but had been quite high to begin with.

Conclusions

Performance of the literature-based parameters was quite good, especially related to

predicting herbage mass and herbage N mass. The prediction of herbage N concentration

needed improvement. On review of the results, there appeared to be some features of

CROPGRO that may have made significant contributions to the errors in predicting both

herbage N concentration and herbage mass. The absence of a dormancy routine to

control vegetative growth during the winter and spring months had a profound effect on

early season N and water availability, contributing to low predicted herbage mass

throughout the season. The absence of a storage organ such as a rhizome or stolon

contributed to this problem by confounding effects of changing proportions of stolon and

root mass. Quirks related to modeling of freeze damage and patterns of "refi11ing" of N

in old tissues complicated matters even more. Imposing a 70% reduction in potential

daily photosynthesis during the winter months compensated for some of the problems,

albeit in an artificial way.

Optimization did improve the fit of both the leaf and canopy models compared to

simulations using the literature-based parameters. Winter growth was slow, and

excessive levels and variation in leaf N concentration were controlled using the optimized

parameters. However, some of the optimized parameters are at the edges of their

biological range or beyond as a result of compensating for the missing/problem










components in the model code. The optimization was more an exercise in compensating

for the model than in divining more accurate parameter values.

In order to better mimic the biology of perennial, tropical grasses, modifications

must be made to the model code itself. Primary among these changes is the addition of a

dormancy routine. Evidence for this and the mechanism required is available in the

literature. Sinclair et al. (2003), Mislevy et al. (2000), and Gates et al. (2001) clearly

demonstrate the role of daylength in controlling dormancy. Hints for the mechanism

involved in reducing leaf and stem growth during dormancy can be found in Rymph and

Boote (2002) and Boote et al. (1999) where significant shifts in allocation of new growth

from shoots to stolons were observed in the fall. Adding a mechanism controlled by

daylength to reduce partitioning and mobilization to the shoot while increasing the same

to the stolon would complement the maturity, temperature, and stress mechanisms

already present in CROPGRO.

Addition of the storage organ would also allow more realistic prediction of the

patterns of accumulation and depletion of roots, avoiding confounding root mass and N

uptake parameters to compensate for the presence of stolons in the root mass. Providing

a storage organ not only provides a sink to store the excess assimilate that is currently

allocated to leaves and stems in the winter, it would also supply a source of CH20 and N

for regrowth after frosts, in the spring, and after harvests. This would prevent the current

situation of the plants dying after a frost and allow for more rapid regrowth in the spring.

Other elements of the model, such as the freeze damage scheme and the

partitioning of N to replenish old leaves, likely stem from past approaches to modeling an

annual grain crop compared to a perennial forage. Situations such as low leaf mass after









a harvest coupled with large amounts of available N from the roots are not generally

encountered in the life cycle of maize or soybean but are dominant features of the pattern

of growth of a perennial grass. These differences are better addressed through adapting

the model code than by adjusting species parameters.

Consideration of these differences notwithstanding, the overall performance of both

the literature-based and optimized parameters was good. If used carefully, the optimized

leaf and canopy models should perform well. More testing would be in order if these

models were to be used extensively. As mentioned earlier, further optimization will only

improve our ability to compensate for the model code, not improve the quality of the

parameters. Taking steps such as running the simulation for a year prior to the measured

growing seasons to establish initial conditions, addition of defoliation events to simulate

frosts, and addition of photosynthesis reduction schemes to reduce winter growth will be

as critical as changing parameter values in establishing a good fit of model predictions to

observed data. The bulk of future efforts should be directed at changing the model code

to more accurately reflect the life cycle of perennial grasses.










Table 4-1. Bahiagrass parameter values for the CROPGRO species file. Preliminary
values were derived from the literature. Optimized values were derived from
optimization runs made based on the preliminary values.
Parameter Name Preliminar value Opiized value
PARMAX 60.0 140.0
PHTMAX 90.0 180.0
CCMP 5.0
FNPGN(1-4) 0.75, 3.0, 10.0, 10.0 1.0, 3.0, 10.0, 10.0
TYPPGN QDR
FNPGT(1-4) 12.0, 25.0, 38.0, 50.0 20.0, 25.0, 30.0, 50.0
TYPPGT LIN
XLMAXT -5.0, 7.0, 35.0, 45.0, 55.0, 60.0 -5.0, 10.0, 26.0, 45.0, 57.0, 60.0
YLMAXT 0.0, 0.0, 1.0, 1.0, 0.0, 0.0 0.0, 0.0, 1.0, 1.0, 0.0, 0.0
FNPGL(1-4) 7.0, 18.0, 45.0, 57.0 7.0, 18.0, 45.0, 57.0
TYPPGL QDR
PGEFF 0.065
SLWREF 0.0035
LNREF 3.0
PGREF 1.76
PROLF I, G, and F 0.22, 0. 11, 0.05 0.15, 0.05, 0.04
PROST I, G, and F 0. 11, 0.07, 0.033 0.125, 0.04, 0.022
PRORT I, G, and F 0.101, 0.040, 0.022
PLIP LF, ST, RT 0.025, 0.020, 0.020
PLIG LF, ST, RT 0.04, 0.06, 0.07
PCAR LF, ST, RT 0.602, 0.697, 0.702 0.672, 0.682, 0.702
CMOBMX 0.025
NMOBMX 0.05
NVSMOB 1.00
XLEAF 0.0, 1.5, 2.0, 3.0, 5.0, 7.0, 30.0
YLEAF 0.45, 0.5, 0.6, 0.4, 0.25, 0.2,
0.2
YSTEM 0.05, 0.05, 0.1, 0.1, 0.05, 0.05,
0.05
FRSTMF 0.05
FRLFF 0.20
FRLFMX 0.60
FINREF 144
SLAREF 285
SIZREF 2.0
VSSINK 0.0
SLAMAX 350
SLAMIN 200
XVGROW 0.0, 5.0, 10.0, 15.0, 20.0, 25.0
YVREF 0.0, 10.0, 20.0, 30.0, 40.0, 50.0
XSLATM -50.0, 00.0, 10.0, 30.0, 60.0











Table 4-1. Continued
Parameter Name Preliminar value Opiized value
YSLATM 0.25, 0.25, 0.25, 1.00, 1.00
FREEZ l, FREEZ2 -25.0, -25.0
ICMP 0.8
TCMP 25
XSTAGE 0.0, 5.0, 9.0, 50.0
XSENMX 3.0, 5.0, 10.0, 50.0
RTDEPI 20
RFAC1 5000
RTSDF 0.02
RWUEP1 1.5
Vegetative
TB, T1, T2, TMax 9.0, 32.0, 40.0, 45.0
Early Reproductive
TB, T1, T2, TMax 10.0, 28.0, 32.0, 45.0
Late Reproductive
TB, T1, T2, TMax 10.0, 28.0, 32.0, 45.0
XVSHT (1-10) 0.0, 1.0, 4.0, 6.0, 8.0, 10.0,
14.0, 16.0, 20.0, 40.0
YVSHT (1-10) 0.0150, 0.0265, 0.0315,
0.0330, 0.0345, 0.0330,
0.0310, 0.0255, 0.0170, 0.0030
YVSWH (1-10) 0.0150, 0.0255, 0.0310,
0.0320, 0.0330, 0.0315,
0.0295, 0.0230, 0.0125, 0.0005