Economically Optimal Management of Huanglongbing in Florida Citrus

MISSING IMAGE

Material Information

Title:
Economically Optimal Management of Huanglongbing in Florida Citrus
Physical Description:
1 online resource (99 p.)
Language:
english
Creator:
Salifu, Abdul-Wahab
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Food and Resource Economics
Committee Chair:
Spreen, Thomas H
Committee Members:
Valderrama, Diego
Grogan, Kelly A
Roka, Fritz Michael
Hogsette, Jerome A, Jr

Subjects

Subjects / Keywords:
citrus -- economically -- florida -- huanglongbing -- management -- optimal
Food and Resource Economics -- Dissertations, Academic -- UF
Genre:
Food and Resource Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Following the declaration of the endemic status of huanglongbing (HLB) in Florida in 2005 with no formal control policy for the disease, it is natural that an empirical examination and justification of the management protocols implemented at the farm-level to control HLB be made. We develop farm level decision rules to judge when it is economically justified to implement a particular control strategy. Models are developed that allow economic assessment of each strategy and determine the scenarios for which each strategy is optimal or yield a positive net present value, considering average grove age at first detection, and rates of infection at first detection. Our results justify the heterogeneous decisions of growers regarding their choice among control strategies, in a way that optimizes each grower’s utility. As hypothesized, the superiority of either strategy depends upon the level of infection at the time when the disease is first found in a particular block, the rate of spread of the disease, the average age of the grove at first infection, expectations of future fruit prices, and the latency period. Our research identifies important efficacy targets that must be achieved for the long-term economic viability of a citrus grove.  Our results provide a recommendation of the optimal control strategy for a given set of conditions such as the age of the planting and initial rate of infection.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Abdul-Wahab Salifu.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Spreen, Thomas H.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-11-30

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045273:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 ECONOMICALLY OPTIMAL MANAGEMENT OF HUANGLONGBING IN FLORIDA CITRUS By ABDUL WAHAB SALIFU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEG REE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Abdul Wahab Salifu

PAGE 3

3

PAGE 4

4 ACKNOWLEDGMENTS All thanks and praises are due to God. On th is note I would like to first acknowledge that this research was supported by the Citrus Initiative Grant. I wish to express my heart felt appreciation and gratitude to the chair of my committee, Dr. Thomas Spreen, for his role as a professional fath er throughout my career as a Ph.D. student. I especially want to thank him for offering me this research opportunity in spite of my shortcomings in recognizing his generosity from the onset of my Ph.D. career. I also wish to extend my deepest gratitude to my advisory committee: Dr. Jerome Hogsette, Dr. Kelly Grogan, Dr. Fritz Roka, and Dr. Diego Valderrama. I have been very fortunate to have their expertise available to me. Their kindness is very much appreciated. Dr. Grogan has been especially instrumenta l in the model development for this research, and I am so grateful to her and Dr. Roka for the timely intervention in sourcing for the much needed funding for the completion of this research and degree. I would like to acknowledge Drs. Eunice Bonsi, Conra d Bonsi, and Robert University in 2007. I would also like to acknowledge Dr. Nii Tackie for his guidance at The collective and indiv idual contributi ons and support of all my class mates, FRED and UF staff and faculty, and the entire gator nation towards this noble achievement is herein acknowledged and much appreciated. Long live the spirit of the gator nation. I especially wish to th ank Jessica Herman for all the administrative support and encouragement I got from her at FRED.

PAGE 5

5 I am also very blessed to have a superb family. To my wife, Abiba Wumbei, who is selfless in supporting me and our two equally gracious daughters, Thalma and T haida Wahab. I am most gracious to my mom, Memunatu Sumani, whom, in spite of her illiteracy insisted and ensured that I get circular education. May God bless her for me. My late dad, Salifu Alidu has also been equally inspirational and supportive to me th roughout his life, may God have mercy on his soul.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ ........ 10 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 13 Background ................................ ................................ ................................ ............. 13 Problem Statement ................................ ................................ ................................ 14 Strategies of Control ................................ ................................ ............................... 16 Objectives ................................ ................................ ................................ ............... 20 Scope of Research ................................ ................................ ................................ 20 2 LITERATURE REVIEW ................................ ................................ .......................... 22 HLB Disease Incidence, Latency, and Spread ................................ ........................ 22 The Impact of HLB ................................ ................................ ................................ .. 25 HLB Control ................................ ................................ ................................ ............ 26 Social Consequences of HLB Persistence ................................ .............................. 29 Effects of HLB on Yield and Cost of Production ................................ ...................... 30 Economics of Disease Control Strategies ................................ ............................... 33 Bioeconomic Models of Disease Control (with Incorporated Discount Rates) ........ 36 3 BIOECONOMIC ESTIMATION ................................ ................................ ............... 38 Optimal Investme nt Theory ................................ ................................ ..................... 38 Overview ................................ ................................ ................................ .......... 38 Optimal Capital Investment Model ................................ ................................ .... 39 The Economic Model ................................ ................................ .............................. 40 The Biological Model ................................ ................................ ............................... 40 4 MODEL RESULTS ................................ ................................ ................................ 44 Mo del Estimation Assumptions ................................ ................................ ............... 44 Empirical Results of Model ................................ ................................ ............... 45 Conclusions ................................ ................................ ................................ ...... 49 5 SENSITIVITY ANALYSIS ................................ ................................ ....................... 63

PAGE 7

7 The Effects of a Price Decline ................................ ................................ ................. 63 The Effects of a Price Increase ................................ ................................ ............... 64 The Effects of a Lower Annual Rate of Spread ................................ ....................... 65 The Effects of an Increased Annual Rate of Spread ................................ ............... 66 The Effects of a Lowered Latency Period ................................ ............................... 67 6 CONCLUSIONS, RECOMMENDATIONS AND LIMITATIONS .............................. 86 LIST OF REFERENCES ................................ ................................ ............................... 90 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 99

PAGE 8

8 LIST OF TABLES Table page 4 1 Baseline Parameter Values ................................ ................................ ................ 50 4 2 Non Valencia Orange Yield Estimated Boxes per Tree, by Age Group in Florida, 2004 2005 through 2008 2009 ................................ .............................. 51 4 3 NPV 1 for S trategy 1 (Do Nothing) ................................ ................................ ....... 52 4 4 NPV 1 for Strategy 2 (Symptomatic Tree Removal) ................................ ............. 53 4 5 NPV 1 for Strategy 3 (Enhanced Foliar Nutriti onal Program) ............................... 54 4 6 NPV 1 for the Three Strategies for Age Classes 0 and 3 ................................ ..... 55 4 7 NPV 1 for the Three Strategies for Age Classes 6 and 10 ................................ ... 56 4 8 NPV 1 for the Three Strategies for Age Classes 14 and 17 ................................ 57 4 9 NPV 1 for the Three Strategies for Age Classes 0 a nd 3 at Different Yield Penalty 2 Levels for Strategy 3 ................................ ................................ ............ 58 4 10 NPV 1 for the Three Strategies for Age Classes 6 and 10 at Different Yield Penalty 2 Levels for Strategy 3 ................................ ................................ ............ 59 4 11 NPV 1 for the Three Strategies for Age Classes 14 and 17 at Different Yield Penalty 2 Levels for Strategy 3 ................................ ................................ ............ 60 5 1 NPV 1 for the Three Strategies f or Age Classes 0 and 3 from a Price Decline 2 ... 69 5 2 NPV 1 for the Three Strategies for Age Classes 6 and 10 from a Price Decline 2 ................................ ................................ ................................ .............. 70 5 3 NPV 1 for the Three Strategies for Age Classes 14 and 17 from a Price Decline 2 ................................ ................................ ................................ .............. 71 5 4 NPV 1 for the Three Strategies for Age Classes 0 and 3 from a Price Increase 2 ................................ ................................ ................................ ............ 72 5 5 NPV 1 for the Three Strategies for Age Classes 6 and 10 from a Price Increase 2 ................................ ................................ ................................ ............ 73 5 6 NPV 1 for the Three Strategies for Age Classes 14 and 17 fr om a Price Increase 2 ................................ ................................ ................................ ............ 74 5 7 NPV 1 for the Three Strategies for Age Classes 0 and 3 from a Decline in Beta 2 ................................ ................................ ................................ ................... 75

PAGE 9

9 5 8 NPV 1 for the Th ree Strategies for Age Classes 6 and 10 from a Decline in Beta 2 ................................ ................................ ................................ ................... 76 5 9 NPV 1 for the Three Strategies for Age Classes 14 and 17 from a Decline in Beta 2 ................................ ................................ ................................ ................... 77 5 10 NPV 1 for the Three Strategies for Age Classes 0 and 3 from an Increase in Beta 2 ................................ ................................ ................................ ................... 78 5 11 NPV 1 for the Three Strategies for Age Classes 6 and 10 from an Increase i n Beta 2 ................................ ................................ ................................ ................... 79 5 12 NPV 1 for the Three Strategies for Age Classes 14 and 17 from an Increase in Beta 2 ................................ ................................ ................................ ................... 80 5 13 NPV 1 for the Three Str ategies for Age Classes 0 and 3 from a Lowered Latency Period 2 ................................ ................................ ................................ .. 81 5 14 NPV 1 for the Three Strategies for Age Classes 6 and 10 from a Lowered Latency Period 2 ................................ ................................ ................................ .. 82 5 15 NPV 1 for the Three Strategies for Age Classes 14 and 17 from a Shortened Latency Period 2 ................................ ................................ ................................ .. 83

PAGE 10

10 LIST OF FIGURES Figure page 4 1 Net Present Value per Acre as a Function of Disease Incidence and Average Age (Years) of Trees at First Detection with Contour Lines for the Do Nothing Strategy ................................ ................................ ................................ .............. 61 4 2 Net Prese nt Value per Acre as a Function of Disease Incidence and Average Age (Years) of Trees at First Detection with Contour Lines for Strategy 2 ......... 61 4 3 Net Present Value per Acre as a Function of Di sease Incidence and Average Age (Years) of Trees at First Detection with Contour Lines for Strategy 3 (30% Yield Penalty) ................................ ................................ ............................ 62 4 4 Dominant Strategy Given Disease Incidence at First Detection an d Average Grove Age (Price = $1.50/pound solid, 30% yield penalty for strategy 3) ........... 62 5 1 Dominant Strategy Given Disease Incidence at First Detection and Average Grove Age from a Change in Pri ce: Top Subplot is Baseline, Middle and Bottom Subplots Shows Price Decline (fr om $1.50 to $1.20) and Increase ( from $1.50 to $1.80) respectively ................................ ................................ ..... 84 5 2 Dominant Strategy Given Disease Incide nce at First Detection and Average Grove Age from a Change in Beta: Top Subplot is Baseline, Middle and Bottom Subplots Shows Beta Decline and Increase, respectively ...................... 84 5 3 Dominant Strategy G iven Disease Incidence at First Detection and Average Grove Age from a Change in Latency: Top Subplot is Baseline, Bottom Subplot Shows Decline in Latency from 1 year to 6 Months for Groves with Average Age of 0 and 3 while the Latency for Groves 6 Years o r Larger Remain at 2 Years ................................ ................................ .............................. 85

PAGE 11

11 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 ECONO MICALLY OPTIMAL MANAGEMENT OF HUANGLONGBING IN FLORIDA CITRUS By Abdul Wahab Salifu M ay 2013 Chair: Thomas H. Spreen Major: Food and Resource Economics Following the declaration of the endemic status of H uanglongbing ( HLB ) in Florida in 2005 with no f ormal control policy for the disease, it is natural that an empirical examination and justification of the management protocols implemented at the farm level to control HLB be made. We develop farm level decision rules to judge when it is economically just ified to implement a particular control strategy. Models are developed that allow economic assessment of each strategy and determine the scenarios for which each strategy is optimal or yield a positive net present value, considering average grove age at fi rst detection, and rates of infection at first detection. Our results justify the heterogeneous decisions of growers regarding their choice among superiority of either strategy depends upon the level of infection at the time when the disease is first found in a particular block, the rate of spread of the disease, the average age of the grove at first infection, expectations of future fruit prices, and the latency period. Our research identifies important efficacy targets that must be achieved for the long term economic viability of a citrus grove. Our results provide a recommendation of

PAGE 12

12 the optimal control strategy for a given set of conditions such as the age of the pla nting and initial rate of infection

PAGE 13

13 CHAPTER 1 INTRODUCTION Background Huanglongbing (HLB) is a bacterial disease that affects all varieties of citrus. It is commonly referred to as citrus greening. HLB was first discovered in Florida in 2005 and is no w found in all counties where commercial citrus is produced (Manjunath et al. 2008). It is spread by a smal l leaf feeding insect, the Asiatic citrus psyllid (ACP). The ACP was first found in June 1998 in Delray Beach and it is noted for its short range ma neuverabilit y and long range drift by wind, which facilitates its ability to spread HLB far and wide. HLB acts to disrupt the phloem of the tree thereby limiting its ability to uptake nutrients. Initially this leads to yellowing of leaves, promotion of pre mature fruit drop, and production of small, misshapen fruit that contain bitter juice with no economic value. As the disease spreads through the tree, the amount of usable fruit produced diminishes until eventually the tree is of no economic value (Brlansk y et al. 2011). Worldwide, three different bacteria are known to cause HLB: Candidatus Liberibacter asiaticus (LAS), Candidatus Liberibacter africanus (LAF), and Candidatus Liberibacter americanus (LAM). The most prevalent of these is LAS, which is found worldwide, includ ing the United States. Asiatic HLB is caused by LAS, an d it is transmitted by the Asian citrus psyllid (ACP), D iaphorina citri While LAM is found to be prevalent in Brazil and China, the African HLB caused by LAF, can be found in Africa, Saudi Arabia, and the South Asia and is spread by its vector, the African citrus psyllid ( Trioza erytreae ) (Gottwald 2010). HLB is the single most vicious and debilitating citrus disease responsible for the destruction of almost 100 million trees in major citrus growing areas of the world where

PAGE 14

14 the disease has become endemic (Aubert et al. 1985, Bov 1986). This is partly due to its elusiveness to various regionally specific management prescriptions. At the present time, the only known way to effectively c ombat the disease is through early detection and a strict eradication program of infected trees. The standard control strategy adopted by HLB affected regions of the world is an integrated control program that involves psyllid control, symptomatic tree rem oval, restricted movement of citrus propagation materials, and distribution of disease free seedlings and budwood (Got t wald et al. 2012 Aubert 1990 ). Problem Statement Florida is the leading citrus producing state in the United States, with nearly 600,000 acres devoted to commercial production. HLB poses as the most serious stry (National Research Council 2010), which suppo rts almost 80,000 jobs. In its eight year presence in Florida, it is estimated th at over 10 million of the 60 million orange trees are currently infected with HLB (Irey et al. 2011), and $1.3 billion in citrus revenue have been lost ( Hodges and Spreen 2012 ; Bolton 2012 ). To appreciate the devastating impact of HLB on Florida citrus, i t is said to cause far worse tree damage than citrus canker which was responsible for the destruction of over 4 million trees. Tree removal due to HLB infection produ ction, and a 40 percent increase in production costs (Irey et al. 2008) HLB has already been implicated for loss in land acres allocated to citrus in the state since 2006, and soaring grower costs in terms of tree eradication, psyllid control, inspection s, and replanting costs (TBO 2008). Hodges and Spreen (2012) estimated that within the last five years, Florida has lost 8,257 jobs, total revenue of $4.541 billion comprised of

PAGE 15

15 indirect revenue of $2.717 billion, due to HLB. A more important longer term c on sequence has been the fact that HLB has created huge uncertainty among Florida citrus growers with respect to future investment/planting. HLB is a disease with two important characteristics. First, the rate of spread is strongly affected by tree age bec ause psyllids prefer new growth (Brlansky et al. 2008). Young trees, which are more vigorous as compared to mature trees, produce more flushes and thereby are more susceptible to psyllid feeding and disease transmission. In the case of mature trees, the di sease spreads more slowly (Gottwald 2010). Consequently, an infected mature tree is capable of producing usable fruit for several years while at the same time serving as a source of infection for other healthy trees. Other factors that affect the rate of s pread of HLB are the ACP population and initial level of infection at first find of the disease The density of the ACP population is the single most important factor because theoretically, if the ACP population is reduced to zero, spread of HLB will stop with immediate effect Second, control through tree eradication is complicated by a latency period between the time a tree first becomes infected and when it expresses visual symptoms. Once a mature tree is infected, it may not begin to exhibit symptoms of the disease for up to two years ( Gottwald 2010) If the rate of infection in a particular grove is relatively high at the time the disease is first discovered, a policy of eradication of symptomatic trees may result in destruction of the entire grove. Ju st a few months after the discovery of HLB in Florida, the citrus canker eradication program was terminated following the sweeping spread of canker over most southern Florida groves by a s eries of hurricanes that blew over the citrus belt in 2004

PAGE 16

16 and 2005. Later in 2005, an interdisciplinary team of USDA HLB experts declared HLB endemic to Florida, with no chances of eradication (Gottwald and Dixon, 2006). So far, it is even more troubling to note that neither the citrus industry nor the state or USDA has p ut in place a clear cut and decisive procedure for control of HLB, unlike in the case of the aborted citrus canker control program. Strategies of Control At this time, there are three distinct strategies being employed to deal with greening. Strategy 1, re no measures to slow its spread including controlling psyllid populations or mitigating tactics are not m odified. Per acre revenues, however, are gradually affected as the disease spreads and the number of healthy fruit that can be harvested and utilized gradually declines. At some point, per acre revenues will not cover per acre grove maintenance costs and a t that point, the grove is no longer economically viable. The disease spreads faster in younger groves, so younger groves cease to be economically viable at a faster rate compared to an older grove with the same initial level of infection. Strategy 2 foll ows the standard plant pathology disease control model and is the only internationally accepted c ontrol strategy for HLB (Aubert 1990). Under Strategy 2, an aggressive psyllid control program is also put into place to su ppress psyllid populations. In addit ion between four and twelve inspections are conducted annually to identify symptomatic trees. Once found, symptomatic trees are immediately eradicated (Brlansky et al. 2008). The logic behind Strategy 2 is that by eradicating symptomatic trees, the level of inoculum in a particular citrus grove gradually will be reduced. Eventually the incidence of the disease will be reduced to a point where it can be

PAGE 17

17 economically tolerated. Muraro (2010) has estimated that in Florida, Strategy 2 increased pesticides cost s by about $450 per acre. Overall production cost have increased from $800 (2004, pre HLB) to $1,500 (2009, post HLB + canker). There are five problems associated with Strategy 2. First, plant pathologists have yet to characterize the key parameters that w ould significantly define the timeline by which to control HLB through eradication of symptomatic trees. These parameters include a controllable base level of HLB infection, the number of years it would take to achieve that base level, and the probability that young tree resets will survive to productive maturity. Second, the latency period of the disease implies that not all diseased trees will be removed in a timely manner, and these asymptomatic trees will serve as a reservoir of the disease inoculum. Th ird, if a grove is already at a high level of known infection and given that more trees are infected but not yet symptomatic, it may not be possible to effectively reduce inoculum levels in a particular grove without eradicating the entire grove. The proba bility of this outcome is related to the age of the grove and the level of infection when the first positive tree is found. Fourth, eradication or suppression of the disease to a tolerable level in one grove may not be possible if neighboring growers are n ot adequately suppressing the disease in their groves. Neighboring groves will serve as sources of the inoculum, and the disease may be continually re introduced into the groves of the grower following Strategy 2. Fifth, relying on visual detection of HLB infected trees by scouting is estimated to be about 50% 60% effective in finding all the symptomatic trees in a single survey ( Futch et al. 2009; Spann et al. 2010) One other factor that also impacts the effectiveness of this strategy

PAGE 18

18 LB management behavior. If psyllid control or tree removal is not coordinated with neighbors of a grove, inoculum builds up in the local vicinity. Strategy 3 is an approach first developed in southwest Florida and is, in part, a response to the Achilles he el of Strategy 2, namely if Strategy 2 is initiated too late, the entire grove may be eradicated before the disease can be suppressed. While an initial high rate of disease incidence is one possible motivation to adopt Strategy 3, it is also possible that under some conditions, Strategy 3 may yield a higher return than Strategy 2 even though Strategy 2 could successfully reduce HL B inoculums to a manageable level. Strategy 3 proposes to treat the symptoms of HLB through foliar application of micro and macro resulting damage to the root system inhibits the ability of the tree to uptake nutrients from the ground. In t he foliar feeding method, a portion of the nutritional needs of the tree is applied through foliar sprays including both macro and micro nutrients (Spann et al. 2010). Formulation of the enhanced nutritional program depends on the program, but generally th e active ingredients include standard essential micronutrients, and phosphite, and salicylate salts (Gottwald et al. 2012). Symptomatic trees are not removed and scouting for the disease is discontinued. As with Strategy 2, a strong psyllid control program is practiced. Roka, et al. (2010) have estimated that the additional nutrient applications increase production costs between $200 to $600 per acre depending on the type and amount of foliar nutritionals a grower decides to apply. The primary concern amo ng plant pathologists with Strategy 3 is that HLB inoculum is left unchecked. The economic implications of Strategy 3 include whether it

PAGE 19

19 is feasible for young trees (ages 3 8) to reach their productive maturity, whether planting the next generation of citr us trees is economically viable, and whether the presence of a grove following Strategy 3 while other growers follow Strategy 2 will cause increased suggests that spr ead between citrus blocks is a more significant portion of disease spread than the spread of the disease within a citrus block (Gottwald et al. 2008). This suggests that heterogeneous control methods may reduce the viability of Strategy 2. This study addre sses the economic consequences of the three strategies. In other words, how does a grower determine which strategy is in her/his best inte rests (given average grove age and initial infection rate)? Strategy 1 needs to be considered as a baseline to referen ce Strategies 2 and 3. Growers make heterogeneous decisions regarding their cho ice among control strategies. Models are developed that allow economic assessment of each strategy and determine the scenarios for which each strategy is optimal or yield a pos itive net present value, considering tree age at first detection, and rates of infection at first detection. Since the optimal strategy may vary due to tree age at first detection and the rate of infection at first detection, the optimal strategy may vary across growers located nearby. Currently, the long term net present value of the control strategies is unknown because of uncertainty in the efficacy of the strategies. Our research identifies important efficacy targets that must be achieved for the long term economic viability of a citrus grove. Our results provide a recommendation of the optimal control strateg y for a given set of conditions. It is hypothesized that the superiority of any one strategy depend s upon the level of infection at the time whe n the disease is first found in a particular

PAGE 20

20 block, the rate of spread of the disease, the average age of th e grove at first infection, expectations of future fruit prices and the latency period The rate of spread is a function of psyllid populations an d the efficacy of psyllid control measures. Objectives The primary objective of this study is to determine the optimal economic management strategies of citrus greening in Florida. This is accomplished through the following specific objectives: 1. Identify g rove age and level of initial disease incidence at which each strategy yield s positive economic returns. 2. Determine the ranges of initial grove age and initial disease incidence for which a given control method is economically preferred over other availabl e methods. Scope of Research The study implements a n et present value analysis of the control strategies adapted by Florida citrus growers following the advent of HLB in the state. This is essential to the determination of which strategy is economically su perior, from the the tree eradication policy to the grower, as no compensation is paid for removed trees. The impact of HLB on citrus yield is first modeled th rough a disease spread function; a discrete logistic function approximated from a Gompertz function. Since the spread rate of HLB is dependent on the average grove age, the logistic function is approximated for three average age classes of 0, 3, and 6 or older D ue to lack of available data for the estimation of model parameters for Florida, we obtain parameter estimates from a corresponding region of HLB spread. Given this logistic function, disease spread in an infected grove with a tree density of 150 per acre is simulated for given parameter values for each age class while varying the initial level of infection. The logistic curves

PAGE 21

21 thus incorporate both asympt omatic and symptomatic trees expressed in a ratio involving total diseased trees Total diseased is th e sum of the asymptomatic (latently infected trees without visual symptoms) and symptomatic tree categories. From this, a spread funct ion is generated and fed into HLB tree and grove severity functions for the calculation of the relative yield due to HLB p resence in the affected grove. The net present value is then estimated from the corresponding relative yield estimates given the yield from a healthy grove unaffected by HLB, obtained as estimated boxes of fruit per tree by age group for non Valencia orang es from the Florida agricultural statistics service (Florida citrus statistics 2008 2009). Fruit prices are expressed as delivered in (to the processing plant) $/pound solids ($1.50/pound solid is the baseline price ) with pound solids per box values depend ent on tree age. The model described above is the infected tree eradication model and the enhanced foliar nutrition model. These models are unique in the sense that they include a latency period of HLB infection, as well as take into account the average grove age, the natural variation in disease incidence at first detection across groves in a region and periodic removal of symptomatic trees ( specific to the tree eradicat ion model) The robustness of each model is tested by a sensitivity analysis conducted for the main model parameters.

PAGE 22

22 CHAPTER 2 LITERATURE REVIEW Responses to stem the devastating effects of HLB or plant diseases in general especially in academia have been enormous. This chapter reviews relevant literature in all aspects of related disciplines including HLB epidemiology, the variety of control methods experimented to date, the impact of HLB across the globe and in Florida as well as its social ramificat ions if left unchecked. In addition, the review includes work on HLB effects on production and yield costs, and general economic and bio economic models of disease control. HLB Disease Incidence, Latency, and Spread Disease incidence has been estimated us ing a variety of approaches. Gottwald et al. (2010) determined disease incidence via a logistic spread rate per year calculated by linear regression of transform ed 1 disease incidence in Florida. HLB incidence in Florida has also been found in similar studi es to increase within 10 months from 0.2 % to as much as 39 % (Gottwald et al. 2007b, 2008; Irey et al. 2008). Spatiotemporal spread models have also been used to characterize HLB in Florida where simultaneous within and across grove spread were common ( Go ttwald et al. 2008). Other studies have been conducted such as in Vietnam where HLB incidence is found to vary depending on the management stra tegy employed (Gatineau et al. 2006) or in Brazil where incidence has been shown to depend on proximity to HLB i nfected citrus groves Gatine a u et al. 2006; 1 The disease incidence data was first transformed via a logistic linear function given by

PAGE 23

23 Gottwald et al. 2007a; 2007b). Albrecht et al. (2012) showed in a Florida study that HLB disease incidence is unaffected by the type of rootstock used in propagation. Disease latency refers to the time between when infection by a pathogen occurs and the onset of symptoms. HLB latency has also been demonstrated in some studies where for every symptomatic tree in a given grove, 13 (range 2 to 56) HLB posit ive but asymptomatic trees existed in its neighborhood which expressed symptoms in subsequent assessments (Bassanezi et al. 2006). Irey et al. (2006) use PCR techniques to test for the presence of the bacteria that causes HLB ( Candidatus Liberibacter asia ticus) in plots of about 190 trees and found that 60 percent more asymptomatic trees existed in addition to the symptomatic trees that were found (Irey et al. 2006). High correlation (R 2 = 0.89) between infected trees and total number of infected trees amo ng the plots suggests natural disease transition from asymptomatic trees to symptomatic trees. In some instances, high bacteria titer was found with PCR in som e asymptomatic trees, suggesting the need for roguing asymptomatic trees as well (National Resear ch Council 2010; Irey et al. 2006). The presence of a high percent (80%) of infected trees within 25 m of a symptomatic tree also signifies short distance spread of HLB (Irey et al. 2006). HLB progression in a grove has also been determined to depend on t he vector population and inoculum levels as well as average grove age at first detection. HLB progression in Reunion Island, China, and the Philippines is reported to follow a sigmoid curve, with clustering of dise ased trees (Gottwald and Aubert 1991; Gott wald et al. 1989, 1991). In Reunion Island more aggregation towards the direction of prevailing wind was observed, sugg esting that psyllids are dispersed by the wind. Aggregation in

PAGE 24

24 China was facilitated by closer tree spacing. Logistic growth rates are mo re plausible for both growth of an infested area in space and population density growth than consta nt growth rates (Kompas and Che 2009). This suggests that an infected area initially grows exponentially, slows down and finally stops as the potential range of the species is attained. Disease progression can reach asymptotic levels faster in young groves than older groves ( Gottwald et al. 2007, 2007a). The dispersal distance for HLB infected psyllids have been estimated to range from 0.88 to 1.61 km with a m edian of 1.58, which may imply that groves more than 2km apart are unlikely to directly affect each other with HLB ( Gottwald et al. 2007b, Gottwald et al. 2008 ) Thus HLB spread is spatially continuous and simultaneous, primarily via psyllid feeding behavi or between groves and secondarily through within grove feeding of the psyllids, necessitating the need for landscape management practices (neighbors HLB management practices should be compatible) for effective control. Manjunath et al. (2008) in a study to detect HLB bacteria from a sample of over 1,200 psyllid adults and nymphs in Florida found that the bacteria spread in an area may be detected one to several years before symptom development in plants. Raphael et al. (2012) developed a deterministic mathe matical model that involve susceptible citrus, infectious but asymptomatic citrus, symptomatic citrus, non infective adult ACP, and infective adult ACP that acquired HLB in the adult and nymph stages to study the dynamics of HLB in a citrus grove. Results show that a ll trees in the grove are infected after 5 years even after removal of symptomatic trees with 47% detection efficiency. They concluded that the best control strategy is the reduction of the vector populations. Chiyaka et al. (2012) used a mathem atical model of HLB transmission to indicate the importance of ACP for initial

PAGE 25

25 HLB infection. Their work also underscores the importance of flush production and latency period in influencing HLB development. The Impact of HLB southern China in 1919 and spread widely and devastated citrus establishments in the Philippines, Indonesia, Thailand, and S outh Africa between 1960 1980. U ntil recently (2004/5), symptoms of HLB were found i n two countries in the Americas; specifically in So Paulo St ate in Brazil, in which nearly three million HLB infected trees w ere removed in subsequent years, and in Florida, USA (Bov 2006; National Research Council, 2010). HLB now occur in other North American areas, such as Cuba, Georgia, Louisiana, South Carolina, Nayarit (Mexico), California, Texas, Costa Rica, and Belize. destructive abilities are unwavering no matter the mode of propagation; reducing yield significantly through fruit drop, dieback and stunted growth, in addition to causing poor quality of un harvested fruits (National Research Council, 2010). Depending on the psyllid vector population, bacteri a titer, and age cohort of the grove at first detection, HLB can take over an entire grove in 3 13 years following the expression of first symptoms (Catling and Atkinson 1974; Aubert et al. 1984; Gottwald et al. 1991; Gatineau et al. 2006; Gottwald et al 2007a; Gottwald et al. 2009). Symptoms can become very severe within one to five years from onset of the disease, depending upon tree age at time of infection and the range of infection (Lin 1963; Schwarz et al. 1973; Aubert 1992). The progression of HLB severity in a grove results in yield reduction, rendering the grove uneconomical within 7 10 years after planting. (Aubert et al. 1984; Aubert 1990; Gottwald et al. 1991; Roistacher 1996).

PAGE 26

26 Worldwide, nearly 100 million trees are estimated to be affect ed by HLB. In some parts of Thailand in 1981, close to 100 percent of trees were affected. Between 1961 and 1970 in the Philippines, citrus acreage was reduced by 60 percent, which represents the fallout from the infection of an estimated seven million tre es in 1962 ( Altamirano et al. 1976; Martinez and Wallace 1969 ). Three million trees were removed in Java and Sumatra within that same period, and a loss of 3.6 million trees were reported in Bali within four years from 1984 to 1987. The HLB havoc extended to southwestern Saudi Arabia, where most sweet orange and mandarin trees were killed by 1983. In the 1960s, the entire citrus industry in Reunion Island was devoured by HLB ( Altamirano et al. 1976; Martinez and Wallace 1969 ). Since its arrival in S o Paul State, Brazil in early 2004, three million HLB affected sweet orange trees have been removed as part of measures taken to control HLB (National Research Council 2010). In the wake of the panic from the first reports of HLB in Florida citrus in 2005, no pu blic policy has emerged to handle HLB, as a result of which growers evolved their own private stop gap management strategies, rendering the citrus industry to be labeled as an endangered industry. Before effective control of the African psyllid with system ic insecticides was discovered in the late 1980s, HLB devastated the South Africa n citrus industry across the length and bre adth of the country, affecting four million out of the 11 million trees in South Africa, during the mid 1 970s (National Research Cou ncil 2010). HLB Control This section outlines the various recommended control measures for a pre and post HLB presence in a given region. These include quarantine, roguing, psyllid control,

PAGE 27

27 and use of healthy nursery propagation materials. The effective ness of some of these measures as gleaned from the literature is also presented. So far, the first line of control of HLB is by adoption of quarantine measures to prevent disease introduction. If however HLB is found in a hitherto HLB free region, a se ries of coordinated actions known as preventative control measures could be taken to control the disease. Affected areas are mapped through surveys to identify infected trees which are later remove d to prevent re infection A rigorous psyllid control prog ram sh ould also be put in place. To avoid infection through plant propagation practices, production of healthy citrus seedlings should be ensured especially if rese tting is required after symptomatic tree s are removed This is because without control, it t akes on average eight years for a grove to reach 100% infection (Bov 2006). Control by ro guing is effective through well timed and carefully repeated surveys to identify all affected trees as much as possible. The latency period of HLB, which can be up t o two or more years (Gottwald 2010) reduces the effectiveness of roguing as a control measure; hence the need for repeated surveys. The quality of roguing is also affected by the presence of uncontrolled psyllids in the grove in which infected tree remova l is practiced. Roguing must therefore be accompanied with a rigorous psyllid control regime. Detection of HL B in Florida and So Paulo is done b y mounting platforms that allow for inspection of the tops of mature trees as it is reported that many affected trees start showing symptoms first on the upper part of the ca nopy (National Research Council 2010). Brlansky et al. (2009) recommend four inspections per year, even though some growers carry out two to three inspections per year. Futch et al. (2009) ind icated that no scouting method is 100% accurate in detecting HLB

PAGE 28

28 symptomatic trees. This re emphasizes the need for multiple inspections within the year. Irrespective of age or severity of infection, all symptomatic trees should be removed (Ayres et al 20 05), and prior to removal these symptomatic trees sh ould be sprayed with a con tact insecticide (Rogers et al. 2010). Unlike citrus canker in which infected trees as well as all surrounding trees at 15 m radii are removed, this practice is not feasible for HLB (Bassanezi 2005), as the psyllid vector feeds randomly acr oss a given grove, and can disperse farther to other groves by wind, hurricanes or storms. The proportion of the infected trees removed depends on the initial disease incidence and hence the ent ire grove can be eradicated at very high rates of initial disease incidence. For instance, a grove with 10% symptomatic trees implies 20% infected trees, and groves with 20%, 30% and 50% symptomatic trees give rise to 36%, 50% and 70% infected trees respec tively, due to latency and hence the presence of asymptomatic trees (Bov 2006). Recently, Bov (2012) has been discounting the latency period saying that it is just incomplete inspections, while Futch et al. (2009) indicates that the latency period of HLB is unknown within a tree. Resetting can be done with healthy seedlings, afte r infected trees are removed Application of contact and systemic insecticides as well as use of biological agents reduces psyllid populations and HLB spread, depending on the sp ecies of the psyllid. Biological control is reported to have been successful in Reunion Island (Aubert and Bov 1980; Aubert et al. 1980), mainly due to the fact that there were no hyperparasitoids on the introduced Tamarixia radiata and T. dryi parasitoid s to hamper their effectiveness (Aubert and Quilici 1984). Predat ors such as spiders, lacewings, ladybugs, minute pirate bugs, and some wasp parasitoids attack the Asian citrus psyllid.

PAGE 29

29 However, the most effective natural enemy is reported to be the coccin ellid lady beetles Olla v nigram and Harmonia axyridis (Michaud, 2004). In Florida, attempts have been made to establish the biological agent T. radiata to control psyllids (Bov 2006) with little effect on the citrus psyllid population Major reasons for this failure include presence of hyperparasites and inadequate number of alternative hosts for the parasitoids (Halbert and Manjunath 2004, National Research Council, 2010) In Brazil, encouraging results were obtained in the use of tree removal and inse cticides against psyllids to control HLB. HLB incidence decreased from 7% to 0.03% in the 10 th survey of a grove with 71,000 trees (Ayres et al 2005). The African version of HLB in South Africa was effectively managed for some period by the adoption of di sease free nursery stock, intensive psyllid control coupled with rou g ing of symptomatic trees. I n China, however, the Asian HLB has proven more difficult to handle with preventative control measures. Aggressive implementation of similar measures brought so me success in Brazil, and gave rise to the identification of factors that affect HLB preventative control effectiveness. These factors include farm size, age cohort of grove, HLB incidence frequency in the area of the grove, neighbors HLB management behavi or, HLB incidence at first inspection, date of first scouting, number of scouting for affected trees, and frequency of insecticide ap plication (Belasque, Jr. et al. 2009). In Florida, it is also been observed that large grove size with well maintained grov es in the same area that have low bacteria titer lowers HLB incidence by reducing the spread across groves. Social Consequences of HLB Persistence The $9.3 billion citrus industry in Florida supports almost 80,000 jobs (grove employees, seasonal pickers, h aulers, processors, and packers). With total annual

PAGE 30

30 wages of $2.7 billion, these workers earn roughly 1.5 percent of (Norberg 2008). Inefficiency in managing HLB in the state will affect not only these full time equivalent job employ ees of the industry, but also the general public will be deprived of health benefits derived from citrus products, albeit still enjoyed at a high er cost from imported juice and fruit Growers who cherish citrus production as a way of life will be affected. The worldwide recognition of Florida as a citrus producing state whose s retirees, and consumers will also be compromised. To augment shortfalls in both fresh and processed citrus demand domestically, imports have to rise, putting further strains on the economy. E ffects of HLB on Yield and Cost of Production Effective management of HLB implies a dramatic increase in production costs through adopt ion of various control measures such as use of disease free nursery stock, scouting and roguing symptomatic trees, and psyllid vector control. Other reasons for reduced profit include declining yield and fruit quality of affected trees, production of healthy nursery trees, costs of tree replacement and care, and value of income/production losses from replaced trees. Yield effects of HLB depend on tree/grove age and severity of infection. Young trees/groves become unproductive faster than mature trees/groves. Mature trees/groves remain productive for several years with less severe infection, and productive life could be reduced to as low as two years with severe infections on a tree/grove (National Research Council 2010).Y ield reduction is high (19%) for younger infected groves (1 5 years olds) 2 4 ye ars after the onset of infection compared to older groves (over 5 years olds) where high yield reduction occurs only after 5 10 years of first symptomatic tree onset ( Bassanezi et al.

PAGE 31

31 2011; Bassanezi and Bassanezi 2008). Optimal control policy for a pest h as been shown to depend significantly on the costs of pest damage per unit of infected area (Carrasco et al. 2009; Sharov 2004). Stringent requirements for raising disease free nursery trees in screened houses have resulted in an increase from $4.5 0 to $9 .0 0 in the cost per nursery tree Given the recommended four inspections per year for symptomatic trees and an estimated cost of between $25 30/acre per inspection annual inspection cost s could add as much as $100 to $120 to production costs (Morris et al 2008). Assuming six trees are detected for removal each year, tree removal costs add another $34 per acre per year to production expenses (Muraro 2008b). Morris et al. (2008) suggested little economic difference between controlling HLB with roguing and doing nothing to ameliorate the impact of the disease until the grove becomes economically useless. Psyllid control is accomplished either with soil application of recommended insecticides or application of foliar insecticides such as zeta cypermethrin c arbaryl, dimethoate, imidacloprid, chlorpyrifos, malathion, phosmet, spinetoram, spirotetramat, fenpropathrin, and petroleum oil. A combination of soil insecticide and three foliar insecticide applications is required for psyllid control of a mature grove at an estimated cost of $288/acre/year (Morris et al. 2008). Citrus production costs and returns depend on the variety, intended use (fresh fruit market or processed juice market), and yield quantities. Fruits for the processed juice market are sold on po unds solids basis in Florida, which relates to the juice content of the fruit. This is estimated to be about 6.5 pounds of solids /90 pound box of fruit, on average. Given production costs for Valencia oranges without HLB or citrus

PAGE 32

32 canker at $1 657 per acre harvesting, delivery and assessment costs at $1 226 per acre, yield of between 300 to 600 90 lb boxes per acre and assuming no resetting of removed trees, the break even price has been calculated to range from $0.80 to $1.19 per pound solids and with HL B or canker, the break even price would be $0.89 to $1.38 (Muraro 2008a). Fresh fruit is sold on a per box basis and for white grapefruit in the Indian River area; total production cost is estimated at $3,195 per acre without canker or HLB and $3,600 when both canker and HLB are present. This results in break even prices of between $8.37 to $5.82 per box for yields of between 350 to 650 boxes per acre without HLB and canker and $10.03 to $6.71 with both HLB and canker (Muraro 2008a). As suming prices of poun d solids range from $1.25 to $1.50, Morris et al. (2008) deduce that processed citrus production will remai n profitable in spite of the 41% increase in production cost due to the presence of HLB, which can even be offset with a suggested increase in planti ng densities. Growers have been shown to benefit significantly in terms of the yield increase, improved quality of produce and labor productivity, as well as the reduced control costs, following their participation in a landscape management program agains t fruit flies in Hawaii (Mau et al. 2007). Expected expenses needed for the control of an established invasion of pests such as pesticides, labor, and equipment have been shown to depend on the distribution, numbers, and rate of spread of the pests (Stohlg ren and Schnase 2006). Likewise, a credible economic assessment of pest control requires an understanding of the biology of the pest, the region of invasion, and temporal data of the pest in the region under inves tigation (Stohlgren and Schnase 2006). Furt her, grower

PAGE 33

33 losses must include not only reduced yield following a pest infestation, but also the adjusted price effects if the infestation has the potential of affecting the supply/demand mechanisms for the commodity in the locality, depending on its pric e elasticity (Myers et al., 1998). Economics of Disease Control Strategies Economic models have been developed to determine the optimal control strategies for some plant diseases. Such models range from barrier effectiven ess (Brown et al. 2002; Sharov 2004 ; Huffaker et al. 1992), infection risk and insurance premium rates assessments (Goodwin and Piggott 2009) to models of optimal control of invasive species such as effects of the environment, discount rate, marginal damages of invasion and marginal costs o f control on optimal control choices (Olson and Roy 2002). Some models study the effects of imperfect information about the degree of infestation from an invasive species on optimal control policy (Haight and Polasky 2010). Models on optimal control of inv asive species management using a logistic growth function to express the growth of the invasive species have also been demonstrated (Eiswerth and Johnson 2002). It has been shown that the optimal control strategy involving pest eradication, reducing pest s pread rate or doing nothing is a function of the size of the area infested, the pest damage per unit area and the rate of discount used in the net benefit c alculation (Sharov and Liebhold 1998; Sharov 2004). Yet, others have demonstrated the effectiveness of using disease free plants and biological control of psyllid vectors in survey studies ( Aubert et al. 1996 ) Chan and Jeger (1994) develop ed a dynamic mathematical model to assess among other things, the effects of disease control by tree removal and pla nting resets. They found that at low infection rates, symptomatic tree removal alone is sufficient to eradicate the disease. At high infection

PAGE 34

34 rates, removal of asymptomatic but infected as well as symptomatic plants is advisable. Jeger and Chan (1995) exa mined the relevance of theoretical models to strategic disease management decisions and concluded that it is the interplay of theory, relevant epidemiological data and predicting the likely effects of control that offers a useful means of refining tactical disease control decisions. Fishman et al. (1983) develop ed bioeconomic models for c itrus tristeza virus (CTV) infection and spread to assess the cost effectiveness of roguing as a disease eradication policy in Israel. They simulated the model to estimat e the net present values to reflect private and social gains from the two policies of eradication or do nothing and found that roguing is more cost effective than do nothing. Kobori et al. (2011) developed an Individual Based Model (IBM) that simulated the disease spread dynamics of HLB and suggested that delaying the latency period and roguing are two effective ways of reducing the spread of HLB in a grove. Their preliminary results showed that regional control may be effective in reducing HLB bacteria tit er in the field. Fishman and Marcus (1984) provide a deterministic model of infectious disease spread within and across rows of plants with periodic roguing Improving the detection method with other parameters and conditions held constant results in time of infection reduced in some rows, or the number of infected plants increases initially, attains a maximum and declines afterwards. Pierre et al. (2006) considers the optimal combination of monitoring (minimum cost of establishing, maintaining and monitori ng traps for fruit flies) and the cost of control once fruit flie s are detected. They applied a Bayesian decision process to solve the optimization problem of choosing between detection expenses (cost of traps set around entry points) and eradication expen ses

PAGE 35

35 (cost of spraying with insecticides, release of sterile male flies, or quarantine measures). They found optimal trapping density for two entry locations (Miami and Tampa) to be higher than the actual number of traps deployed in practice, suggesting the need for additional modifications to the model. Batabyal and Nijkamp (2008) use renewal theory to construct and analyze a dynamic and stochastic model of optimal control for invasive species in an orchard. They also derive the long run expected cost ( LREC ) for the orchard per unit time and show that the optimal roguing and resetting densities solves the derived LREC minimization problem Their model treated commercial orchards as entities whose growth and output are by nature dynamic and stochastic under c onstant threats from a variety of invasive plant or animal species. Lominac and Batabyal (2009) focused on a representative tree in a grove and used discrete time Markov chains to ed model is used to define the trees one step transition probabilities, determine the time the tree is affected in the long run and the long run schedule of replacements for the representative tree when it dies. Morris and Muraro (2008) perform an economi c analysis of greening management of tree removal with/without different densities of resetting, versus do nothing. They concluded that resetting is preferable to do nothing if resets reach maturity in a mature grove. Replanting at higher density is best f or a grove that is unproductive due to HLB. et al. (2012) present a bio economic model for optimization of disease control with latency, using the network/individual based methodology. Depending on total cost of disease incidence ( consisting of costs of treating infected individuals and prevention of infection ), three optimal strategies are

PAGE 36

36 identified These include total population preventive treatment, local treatment within a neighborhood of certain size, and only treatment of diagnosed cases with no prevention Some epidemiological factors do affect the optimal strategy of local treatment. Bioeconomic Models of Disease Control (with Incorporated Discount Rates) Bioeconomic approaches to optimal management of perennial biological resources involvi ng complex patterns or transitioning processes such as pests/disease incidence require economic assessment that incorporates discount rates into cost benefit estimates involving present value analysis. The net present value of an asset is the weighted expo nential function of the time at which net expected revenues of the asset are obtained; where N(t) is the net revenue at time t, r is the discount rate, and T is the time period (Clark 1976). Only Sharov and Liebhold (1998) use present value for optimization of long term pest management options involving barrier zones. Their theoretical model shows that pest control mechanisms that slow pest population spread is a feasible strat egy unlike strategies that stop population spread, which re quire natural barriers to be optimal. Enkerlin and Mumford (1997) estim ated the net present value for three improved management options to control the Mediterranean fruit fly across three countries (Israel, Palestinian Terri tories, and Jordan). Given the n ine year analysis, the predominant control method is the sterile male suppression option whereas over a longer time period, the sterile male eradication option predominates. Odom et al. (2003) develop ed and applied a deterministic dynamic programming model in a case study to derive optimal control rules for the management of an environmental weed (scotch broom) in a national park. Model results show the need for biological control as a viable option. In a similar study, Chalak Haghighi et al.

PAGE 37

37 (2008) utilized ecological and economic information to construct a dynamic bio economic optimization model to evaluate the net benefits of a range of possible control options for Californian thistle (Cirsium arvense) weed in New Zealand pa sture. Factors considered in the maximization of the net benefit include the costs and effectiveness of control options, and the revenue from animal production. Their results suggest that the optimal strategy is a mix of a bio control agent with one or mor e integrated weed management strategy, especially when the initial density of the thistle population exceeds 1.0 shoot m 2

PAGE 38

38 CHAPTER 3 BIOECONOMIC ESTIMATION This c hapter contains an overview of optimal investment theory, which provides the theoretical fr amework for the use of net present value analysis in the study. Next is an expatiation on the income method for asset valuation and its appropriateness for this type of study, herein known as our economic model. Finally, a biological model is presented an d it spans the Gompertz and logistic functions, the tree and grove severity functions, and the negative exponential function for relative yield. Optimal Investment Theory Overview T wo possibilities exist in designing a framework for the theory of optimal investment, namely the neoclassical theory of optimal capital accumulation and utility maximization theory. We will adopt the neoclassical theory of the firm for our theoretical framework since it is a more powerful theory than the utility maximizing theo ry ( Jorgenson 1967) N eoclassical theory assumes that capital growth depend s on utility maximization of a consumption stream. Essentially, a given firm maximizes consumption utility subject to a given production function at fixed current and future prices and interest rates for both input and output flows. A production plan is then chosen to maximize the present value of the returns from the investment. M aximizing the present value of the firm is the only criterion consistent with utility maximization theor y. The resulting theory of optimal capital accumulation broadly includes special cases of econometric models of investment actions ( Jorgenson 1967) When an investment involves expense and income flows into the future, such as investments in perennial cro p production, it is necessary to estimate the present value

PAGE 39

39 of the series of cash flows projected from the proposed investment Net present value (NPV) calculation attempts to do just that, and gives a measure of how much the investor gains today for inves ting in the project. In NPV calculation, future cash flows are discount ed at a specified discount rate that depends on the time value of money, the interest from an alternative guaranteed investment, and the degree of risk compensation that is being accept ed in the project. Optimal Capital Investment Model The present value of a firm that is assumed to produce a single output from a single variable and capital input is given by the expression: w here r is the d iscount rate, I is net income, Q C and R represent output, variable input and fixed inputs respectively, and p w and q are their corresponding prices ( Jorgenson 1967 ) N et present value ( NPV ) is maximized subject to the constraints that the rate of ch ange of the flow of capital services ( ) is related to the flow of net investment ( ), where is a constant of proportionality and denotes depreciation of capital and K is capital services. The second constraint states that the production function is constrained by output ( Q ) and input ( C K ) level s. The maximization problem is s olved for the optimal variable input ( C* ), output ( Q* ), and capital services ( R *) and the ma rginal productivity for variable input and capital as well as the shadow price for capital services. Thus the complete optimal capital accumulation model consists of the production function, the two marginal productivity conditions, and the function for th e shadow price for capital services ( Jorgenson 1967 )

PAGE 40

40 The Economic Model A citrus grove is an asset. We estimate the economic impact of HLB through its effect on the value of a particular citrus grove. There are a variety of approaches in asset valuation, but the most appropriate approach in this application is the income method. In the income method, future costs and revenues are estimated to give per annum net revenue. Future net revenue is discounted to the present to give net present value (NPV) using the formula, w here is price in time period t is yield in time period t are costs in time period t and r is the discount rate HLB affects the NPV of an infected grove by increasing costs if control is implemented, and decreasing future fruit production, thereby reducing future revenues. Since the rate of spread depends in part upon the tree age at first infection, we compute NP V as a function of tree age as well as the level of infection at first detection. Since the NPV of a particular grove de pends upon several factors, which are subject to random variation, stochastic dominance is an appropriate method to identify the superio r strategy. At this time, however, knowledge of the underlying probability distributions of those random factors is not available, so our economic assessment is done in a deterministic framework. The Biological Model Our original idea to depict HLB spread was motivated by a Gompertz function as proposed by Bassanezi and Bassanezi (2008). This function specifies that the disease incidence, y at time t is:

PAGE 41

41 w here y 0 is the disease incidence at first detection and is the annual rate of spread of the disease. However, the Gompertz function always converges to 100% infection, which does not allow us to analyze control strategies that prevent 100% disease infection. A logistic function has the advantage of being more flexible and allows for a steady state level of disease infection that is less than 100%. In this case we estimate the parameters of the logistic function that approximate the Gompertz function, and use those parameters to esti mate the impact of S trategies 2 and 3. To do this, we use parameter values for y 0 and for each age class from Bassanezi and Bassanezi (2008) to simulate Gompertz spread from low to high incidence until field incidence reaches 100%. Using nonlinear regression, the simulated Gompertz data for each age class are use d to estimate the corresponding logistic Our logistic function is derived from the deterministic differential equation: w here Y is the proportion of diseased trees at time t is the change in the proportion of diseased trees and is the annual rate of spread of the disease The result of this procedure yielded our logistic estimates to be 1.5148125 0.8450625 and 0.4440625 of their Gompertz counterparts of 1.3, 0.65, and 0.325 obtained from Bassanezi and Bassanezi (2008) for each corresponding age class consisting of average grove age of 0, 3, and 6 ( Table 4 1 ) The logistic curves are then generated according to Equation 3 3 :

PAGE 42

42 For strategy 1, include s both symptomatic disease incidence, as well as asymptomatic disease inc idence, The assumption on latency period in the baseline model (Strategy 1) is 1 year for groves of average age of 0 and 3, and 2 years for groves with average age of 6 or larger (Gottwald 2010) In the sensitivity anal ysis, latency period is one of the parameters we alter to check model robustness. For S trategy 2, if the assumption on latency is 1 year for instance, then trees remain asymptomatic for one year, implying that Further, we assume tha t all symptomatic trees are immediately removed once the tree exhibits symptoms, implying that in Equation 3 3 equals Since the disease moves both across trees in the grove and across canopy in a given infected tree, we need to model the spread of the disease in canopy area as well to determi ne the yield effect of HLB for S trategies 1 and 3. It is worthwhile to mention here that HLB is spread only by psyllids hence vector control will h ave significant effects on disease spread. As a result, one of the factors for sensitivity analysis addressed in Chapter 5 is latency period, which we assume to be the proxy for psyllid control. We estimate the yield impact of HLB ( ) as a function of symptomatic grove canopy area or disease severity and yield of a healthy grove ( R t average boxes per tree) for S trategy 1 using the negative exponential model:

PAGE 43

43 w here R t equals 1, denotes the full yield of a healthy grove ( average boxes per tree ) is the percent of healthy yield obtained for a given level of disease severity for strategy 1, b is t he rate of yield reduction as a function of HLB severity X t is total grove severity at time t x is the fraction of HLB symptomatic tree canopy area at time t x 0 is the fraction of HLB symptomatic tree canopy area at first detection, and is the annual rate of disease severity progress in an affected tree. For S trategy 2, all symptomatic trees are removed, so the spread of yield losses through the canopy does not occur. For S trategy 3, the yield effect is assumed to be in between the yield effect for st rategy 1 and a healthy grove. Since the reduction in yield relative to a healthy grove is With all three strategies modeled, we determine the scenarios for which each strategy would be optimal, considering all possible strategy efficacies and tree age and rates of infection at first detection.

PAGE 44

44 CHAPTER 4 MODEL RESULTS In this c hapter, the baseline model results are presented. Key parameters such as the annual rate of spread of HLB ( ), price per pound solids, length of latency, and yield penalties are fixed at specified values according to relevant literature and secondary data sources (Table 4 1). The c hapter begins with an expo sition of the assumptions of the model, after which empirical results of the model are presented followed with some concluding remarks. Model Estimation Assumptions We create disease spread curves using values described in Chapt er 3 and use those parameters to estimate the NPV of Strategies 1, 2, and 3. Historical data on boxes of fruit per tree by age group for non Valencia oranges from the Florida Agricultural Statistics Service (Florida Citrus Statistics 2008 2009) are used to establish yield curves by variety. Next logistic curves of disease spread are interacted with the investment or NPV model as specified above to estimate the impact of HLB on growe r earnings based on tree age and first detection of the disease. Fruit pric es are expressed in $/pound solids delivered in ($1.50/pound solid) with pound solids per box values dependent on tree age. The estimates are made on a per acre basis for a grower with 150 trees per acre and 100% original tree acreage remaining. We use a 1 0% discount rate for calculation of net present values. Operating and production costs for a mature grove include herbicide, pesticide, and fertilizer applications, irrigation, and pruning, but do not include HLB foliar nutritional sprays or pesticide appl ications in the baseline calculations. Since we assume no resetting (replacing trees lost in the citrus grove), the

PAGE 45

45 adjusted reset grove costs by tree age are assumed to be zero 2 as well as the establishment costs/acre for new solid set, the cost of tree removal and planting reset replacement trees, reset frequency, and reset yield adjustments. Yield loss due to freeze or other diseases is assumed to be zero. We calculate net present value of a stylized citrus grove using a 15 year time horizon. Beyond 1 5 years, the net present value per year approaches zero. We calculate the net present value for groves with an initial average age ranging from 0 to 17. Beyond 17 years of age, tree yields no longer increase, so calculations for groves of this age repres ent our net present value upper bound. Empirical Results of Model Under Strategy 1 (do nothing) all groves with an average tree age of 0 and 3 years yield a negative net present value at any initial disease incidence rate. Groves that contain younger trees at first detection also experience a faster spread of the disease. Consequently, young groves that become infected with HLB are unable to produce a sufficient volume of fruit to recover investment costs. Irrespective of the disease incidence rate at first detection, all groves with an average age of 6 years and over yield a positive net present value under Strategy 1. In Table 4 3 the net present values for groves with initial rates of disease incidence varying from 0.1% to 50% and for average initial grov e ages of 0, 3, 6, 10, 14, and 17 years are reported under Strategy 1. A plot of the net present values as a function of disease incidence and average age at first detection is shown in Figure 4 1 A lso shown are contour lines, with 2 The assumption of no resetting greatly simplifies the calculation of disease spread and the accompanying reduction in fruit production per acre. The assumption clearly is a limitation on the derived results.

PAGE 46

46 the green contour line marking the ages and disease rates at which the net present value is $0.00. Under tree removal (Strategy 2), groves with average age of 0 display negative net present values whereas groves with an average age of 3 years show negative net present value w hen the initial disease incidence is 20% and larger. Groves with an average age of 6 show positive net present value for initial disease incidence ranging from 0.1% to 30%, but shows negative net present value for initial disease incidence of 40% and 50%. All other age categories show a positive net present value, no matter the initial rate of disease incidence (Table 4 4 ). In Figure 4 2, the green contour line marks the ages and disease rates at which the net present value is $0.00 for strategy 2. An enha nced foliar nutritional program (Strategy 3) is expected to boost yield of an HLB affected grove, but will be lower compared to a disease free grove. This analysis assumes a yield penalty of 30% compared to a healthy grove under Strategy 3. The estimated N PVs associated Strategy 3 is presented in Table 4 5 As before, groves with average age of 0 show negative net present value at all levels of initial disease incidence. For this strategy, the ages and disease rates at which the net present value is $0.00 are indicated by the green contour line of Figure 4 3. For ease of comparison, Tables 4 6 through 4 8 juxtapose the net present value for the three strategies for each age class. Bolded values indicate the superior strategy at a particular age of first det ection and initial rate of infection. For groves whose average age is 0 at first detection, the net present values are all negative For trees with average age of 3 years, Strategy 2 is better than Strategies 1 and 3 when disease incidence ranges from 0.1 % to 7.0%, and thereafter (incidence of 8.0% to 50%),

PAGE 47

47 Strategy 3 is better than both Strategies 1 and 2. For trees with average age of 6 and 10, Strategy 1 is better than Strategies 2 and 3 at lower rates of initial disease incidence (0.1% to 2.0%), after which Strategy 2 becomes superior to Strategies 1 and 3 when the disease incidence ranges between 3.0% and 10.0%. At the highest initial disease incidence of between 20% and 50%, Strategy 3 is superior to Strategy 2 and 1 in net present value. For trees wi th average age of 14 and 17, Strategy 1 outperforms the other two strategies at the low rates of disease incidence (0.1% to 2.0%), and for the middle rates of disease incidence of between 3.0% and 8.0%, Strategy 2 is better than the other two strategies. A t the highest rates of initial disease incidence (10% to 50%), Strategy 3 becomes superior to Strategies 1 and 2. Th e results presented in Table 4 6 through 4 8 provide several interesting implications. While it may be surprising that Strategy 1 is ever identified as the superior strategy, the results suggest that at very low levels of initial infection, the costs associated with both Strategy 2 and Strategy 3 exceed gains to be realized in the future by mitigating the effects of the disease. One could d escribe this result as the superior strategy because of the large number of trees that must be removed if Strategy 2 is followed. The results herein support that ar gument that once the disease becomes well via Strategy 3. These results also point out a major limitation of the methodology employed here. The neighbor effects are ignor ed. Adoption of any of the three strategies implies neighbor effects. Since both Strategy 1 and 3 entail non removal of symptomatic trees,

PAGE 48

48 the level of disease inoculum in a particular grove is not being diminished. Strategy 2 calls for the removal of s ymptomatic trees and its primary intent is to reduce the level of inoculum. If one grower pursues Strategy 2 and his/her neighbor pursues Strategy 3, the actions of the second neighbor will adversely affect the first neighbor because the neighboring grove will continue to serve as a source of inoculum. Figure 4 4 delineates the ranges of initial grove age and initial disease incidence for which each strategy maximizes net present v alue after adjusting for age classes in which all strategies post negative net present values (age class of 0 and sometimes 3) S trategy 3 dominates at incidence levels of 8% 50% for groves of almost all ages. Strategy 2 dominates for groves with average age of 0 and 3 years at low initial disease incidence of 0.1% t o 2 % and al so at disease incidence levels of 3% to 8% for all groves with average age of 6, 10, 14, and 17. Strategy 1 dominates for all groves with average age of 6, 10, 14, and 17 only when disease incidence is 0.1% to 2% Therefore, for almost all groves at almost all initial HLB disease incidence, there is the likelihood that the enhanced nutritional program will generate a higher NPV for the grower than if one were to follow a do nothing or tree eradication management strategy. For groves with average age of 6 17 years at initial HLB disease incidence of 3% to 8% tree eradication program management strategy will yield higher returns to the grower than do nothing or implementation of the enhanced nutritional program. For groves at 6 17 years at ver y low HLB in cidence (0.1% to 2%) the grower will be better off by doing nothing than either implementing the tree eradication or enhanced nutritional program. For a new solid set grove at any level of initial disease incidence, the enhanced nutritional program is lik ely to give the grower the best earnings on his/her investment

PAGE 49

49 than any other strategy. No matter how high the initial rate of disease incidence, each strategy remains positive in net present value for mature groves (groves with average age of 6 or larger) Strategy 3 performs even better especially for mature trees at almost all rates of disease incidence when the assumption on yield penalt y of a healthy grove is 5%, 10% or even 20% instead of the 30% yield penalty (Tables 4 9 to 4 11 ) used in the compari son. For all age classes, cost eventually exceeds revenue, especially for mature groves at high rat es of initial disease incidence. Conclusions Which strategy is superior to the other(s) depends on the age of trees at first detection and the initial rate of disease incidence at first detection. Each strategy has its initial level of infection and the average age of the grove. Growers with grove s of all ages at 20% or more initial incidence may be better off implementing the enhanced nutritional program (S trategy 3) For growers whose groves are three years or older in average age with initial HLB infection rate at 3% to 8%, (or growers with newly established groves at 0.1% 2% HLB incidence), the best strategy is S trategy 2 (infected tree removal). Strategy 1 is the least optimal strategy and it is only optimal when incidence is very low (0.1% to 2%) for groves with average age of 6 or larger.

PAGE 50

50 Table 4 1 Baseline Param eter Values Parameter Trees Age Class 0 3 6 1.3000000 0.65000000 0.32500000 1.5148125 0.84506250 0.44406250 Price/pound solid ($) 1.5000000 1.50000000 1.50000000 Latency Period (years) 1.0000000 1.00000000 2.00000000 Tree Se 3.6800000 1.84000000 0.92000000 Yield Reduction Rate of HLB ( b ) 1.8000000 1.80000000 1.80000000 Initial Severity (x 0 ) 0.2000000 0.10000000 0.05000000 Source: Bassanezi and Bassanezi (2008) ; Bassanezi et al. (2011)

PAGE 51

51 Table 4 2. Non Valencia Orange Yield Estimated Boxes per Tree, by Age Group in Florida, 2004 2005 through 2008 2009 Tree Age Average 1 Yield (2004/5 2008/9) Yield 2 (boxes/tree) 1 1.2 0 2 0 3 1 4 1.2 5 1.4 6 1.8 1.7 7 1.8 8 1.9 9 2.26 2 10 2.1 11 2.3 12 2.4 13 2.5 14 3.05 2.6 15 2.7 16 2.8 17 2.9 18 3 19 3.1 20 3.2 21 3.3 22 3.4 23 3.5 Sources: Florida Citrus Statistics 2008 2009. FASS 1 Average yields of 1.2, 1.8, 1.26, and 3.05 boxes/tree are for groves of ages 3 5, 6 8 9 13, and 14 23 years respectively 2 Yield for each tree age is derived from the 5 year average yield in column two

PAGE 52

52 Table 4 3 NPV 1 for Strategy 1 (Do Nothing ) Disease Incidence at First Detection Average A ge (Years) of Trees at First Detection 0 3 6 10 14 17 0.001 2,614 3,843 11,463 14,551 16,487 17,101 0.010 4,142 927 9,539 12,562 14,488 15,102 0.020 4,532 17 8,442 11,407 13,322 13,935 0.030 4,696 662 7,686 10,601 12,505 13,118 0.040 4,779 9 61 7,213 10,084 11,978 12,591 0.050 4,942 1,182 6,673 9,505 11,389 12,002 0.060 5,004 1,599 6,360 9,157 11,032 11,644 0.070 5,052 1,754 5,893 8,656 10,521 11,133 0.080 5,089 1,886 5,659 8,393 10,250 10,861 0.100 5,140 2,097 5,265 7,947 9,786 10,396 0.200 5,338 2,960 3,555 6,032 7,799 8,405 0.300 5,369 3,531 2,563 4,897 6,604 7,207 0.400 5,462 3,988 1,463 3,634 5,278 5,877 0.500 5,482 4,164 1,077 3,176 4,779 5,375 1 Cumulative 15 year NPV ($/ac). Beta (1) = 1.5148125 for the 0 Ag e Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 53

53 Table 4 4 NPV 1 for Strategy 2 (Symptomatic Tree Removal ) Disease Incidence at First Detection Average Age (Years) of Trees at Firs t Detection 0 3 6 10 14 17 0.001 645 4,830 8,441 11,534 13,470 14,084 0.010 4,050 4,322 8,207 11,276 13,204 13,818 0.020 5,478 3,790 7,949 10,993 12,910 13,525 0.030 6,302 3,287 7,694 10,712 12,620 13,235 0.040 6,871 2,813 7,442 10,435 12,333 1 2,947 0.050 7,297 2,363 7,193 10,160 12,049 12,663 0.060 7,639 1,936 6,946 9,888 11,768 12,382 0.070 7,916 1,531 6,701 9,619 11,489 12,103 0.080 8,152 1,144 6,460 9,353 11,213 11,828 0.100 8,529 423 5,983 8,828 10,670 11,284 0.200 9,569 2,411 3,745 6,359 8,111 8,725 0.300 10,043 4,421 1,721 4,124 5,790 6,404 0.400 10,295 5,937 106 2,101 3,686 4,300 0.500 10,433 7,114 1,752 276 1,784 2,399 1 Cumulative 15 year NPV ($/ac). Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.845062 5 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 54

54 Table 4 5 NPV 1 for Strategy 3 (Enhanced Foliar Nutritional Program ) Disease Incidence at First Detection Average Age (Years) of Trees at First Detection 0 3 6 10 14 17 0.001 2,170 3,030 7,822 10,915 12,852 13,466 0.010 2,610 2,211 7,245 10,318 12,252 12,866 0.020 2,727 1,872 6,916 9,972 11,902 12,516 0.030 2,776 1,679 6,689 9,730 11,657 12,271 0.040 2,826 1,589 6,547 9,575 11,499 12,113 0.050 2,850 1,52 3 6,385 9,401 11,322 11,936 0.060 2,868 1,398 6,291 9,297 11,215 11,829 0.070 2,883 1,351 6,151 9,146 11,062 11,675 0.080 2,894 1,312 6,081 9,067 10,981 11,594 0.100 2,909 1,248 5,962 8,933 10,841 11,454 0.200 2,968 989 5,449 8,359 10,245 10,857 0.300 2,978 818 5,152 8,019 9,887 10,498 0.400 3,006 681 4,822 7,640 9,489 10,099 0.500 3,011 628 4,706 7,502 9,339 9,948 1 Cumulative 15 year NPV ($/ac). for strategy 3 Beta (1) = 1.5148 125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 55

55 Table 4 6 NPV 1 for the Three Strategies for Age Classes 0 and 3 Disease Incidence at First Detection Average Age (Years) of Trees at First Detection 0 3 Strategy Strategy 1 2 3 1 2 3 0.001 2,614 645 2,170 3,843 4,830 3,030 0.010 4,142 4,050 2,610 927 4,322 2,211 0.020 4,532 5,478 2,727 17 3,790 1,872 0.030 4,696 6,302 2,776 662 3,287 1,679 0.040 4,779 6,871 2,826 961 2,813 1,589 0.050 4,942 7,297 2,850 1,182 2,363 1,523 0.060 5,004 7,639 2,868 1,599 1,936 1,398 0.070 5,052 7,916 2,883 1,754 1,531 1,351 0.080 5,089 8,152 2,894 1,886 1,144 1,312 0.100 5,140 8,529 2,909 2,097 4 23 1,248 0.200 5,338 9,569 2,968 2,960 2,411 989 0.300 5,369 10,043 2,978 3,531 4,421 818 0.400 5,462 10,295 3,006 3,988 5,937 681 0.500 5,482 10,433 3,011 4,164 7,114 628 1 Cumulative 15 year NPV ($/ac). Yield from HLB infected t for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 56

56 Table 4 7 NPV 1 for the Three Strategies for Age Classes 6 and 10 Disease Incidence at First Detection Average Age (Years) of Trees at First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 0.001 11,463 8,441 7,822 14,551 11,534 10,915 0.010 9,539 8,207 7,245 12,562 11,276 10,318 0.020 8,442 7,949 6,916 11,4 07 10,993 9,972 0.030 7,686 7,694 6,689 10,601 10,712 9,730 0.040 7,213 7,442 6,547 10,084 10,435 9,575 0.050 6,673 7,193 6,385 9,505 10,160 9,401 0.060 6,360 6,946 6,291 9,157 9,888 9,297 0.070 5,893 6,701 6,151 8,656 9,619 9,146 0.080 5,659 6,460 6 ,081 8,393 9,353 9,067 0.100 5,265 5,983 5,962 7,947 8,828 8,933 0.200 3,555 3,745 5,449 6,032 6,359 8,359 0.300 2,563 1,721 5,152 4,897 4,124 8,019 0.400 1,463 106 4,822 3,634 2,101 7,640 0.500 1,077 1,752 4,706 3,176 276 7,502 1 Cumulative 15 yea r NPV ($/ac). for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 57

57 Table 4 8 NPV 1 fo r the Three Strategies for Age Classes 14 and 17 Disease Incidence at First Detection Average Age (Years) of Trees at First Detection 14 17 Strategy Strategy 1 2 3 1 2 3 0.001 16,487 13,470 12,852 17,101 14,084 13,466 0.010 14,488 13,204 12,252 15 ,102 13,818 12,866 0.020 13,322 12,910 11,902 13,935 13,525 12,516 0.030 12,505 12,620 11,657 13,118 13,235 12,271 0.040 11,978 12,333 11,499 12,591 12,947 12,113 0.050 11,389 12,049 11,322 12,002 12,663 11,936 0.060 11,032 11,768 11,215 11,644 12,382 11,829 0.070 10,521 11,489 11,062 11,133 12,103 11,675 0.080 10,250 11,213 10,981 10,861 11,828 11,594 0.100 9,786 10,670 10,841 10,396 11,284 11,454 0.200 7,799 8,111 10,245 8,405 8,725 10,857 0.300 6,604 5,790 9,887 7,207 6,404 10,498 0.400 5,278 3,686 9,489 5,877 4,300 10,099 0.500 4,779 1,784 9,339 5,375 2,399 9,948 1 Cumulative 15 year NPV ($/ac). for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 58

58 Table 4 9 NPV 1 for the Three Strategies for Age Classes 0 and 3 at Different Yield Penalty 2 Levels for Strategy 3 Disease Incidence at First Detection Average Age (Years) of Trees at Fi rst Detection 0 3 Strategy Strategy 1 2 3 1 2 3 20% 10% 5% 20% 10% 5% 0.001 2,614 645 1,535 900 582 3,843 4,830 3,477 3,924 4,147 0.010 4,142 4,050 1,828 1,046 655 927 4,322 2,894 3,632 4,002 0.020 4,532 5,47 8 1,906 1,085 675 17 3,790 2,705 3,538 3,954 0.030 4,696 6,302 1,939 1,101 683 662 3,287 2,576 3,473 3,922 0.040 4,779 6,871 1,955 1,110 687 961 2,813 2,516 3,444 3,907 0.050 4,942 7,297 1,988 1,126 695 1,182 2,363 2,513 3,421 3,89 6 0.060 5,004 7,639 2,000 1,132 698 1,599 1,936 2,389 3,380 3,875 0.070 5,052 7,916 2,010 1,137 701 1,754 1,531 2,358 3,364 3,868 0.080 5,089 8,152 2,017 1,141 702 1,886 1,144 2,331 3,351 3,861 0.100 5,140 8,529 2,027 1,146 705 2,097 423 2,289 3,330 3,850 0.200 5,338 9,569 2,067 1,160 715 2,960 2,411 2,117 3,244 3,807 0.300 5,369 10,043 2,073 1,169 716 3,531 4,421 2,002 3,187 3,779 0.400 5,462 10,295 2,092 1,178 721 3,988 5,937 1,911 3,141 3,756 0.500 5, 482 10,433 2,096 1,180 722 4,164 7,114 1,876 3,123 3,747 1 Cumulative 15 year NPV ($/ac). 2 for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for ag e Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 59

59 Table 4 10 NPV 1 for the Three Strategies for Age Classes 6 and 10 at Diff erent Yield Penalty 2 Levels for Strategy 3 Disease Incidence at First Detection Average Age (Years) of Trees a t First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 20% 10% 5% 20% 10% 5% 0.001 11,463 8,441 7,865 7,907 7,929 14,551 11,534 10,958 11,002 11,023 0.010 9,539 8,207 7,480 7,715 7,832 12,562 11,276 10,560 10,803 10,924 0. 020 8,442 7,949 7,260 7,605 7,778 11,407 10,993 10,329 10,687 10,866 0.030 7,686 7,694 7,109 7,529 7,740 10,601 10,712 10,168 10,607 10,826 0.040 7,213 7,442 7,036 7,482 7,716 10,084 10,435 10,087 10,555 10,800 0.050 6,673 7,193 6,906 7,428 7,689 9,505 10,160 9,949 10,497 10,771 0.060 6,360 6,946 6,844 7,397 7,673 9,157 9,888 9,879 10,462 10,754 0.070 5,893 6,701 6,750 7,350 7,660 8,656 9,619 9,779 10,412 10,739 0.080 5,659 6,460 6,704 7,327 7,638 8,393 9,353 9,727 10,386 10,715 0.100 5,265 5,983 6,6 25 7,287 7,619 7,947 8,828 9,637 10,341 10,693 0.200 3,555 3,745 6,283 7,116 7,533 6,032 6,359 9,254 10,150 10,597 0.300 2,563 1,721 6,085 7,017 7,484 4,897 4,124 9,027 10,036 10,541 0.400 1,463 106 5,864 6,907 7,429 3,634 2,101 8,775 9,910 10,478 0.5 00 1,077 1,752 5,787 6,869 7,409 3,176 276 8,683 9,864 10,455 1 Cumulative 15 year NPV ($/ac). 2 for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for ag e Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 60

60 Table 4 11 NPV 1 for the Three Strategies for Age Classes 14 and 17 at Different Yield Penalty 2 Levels for Strategy 3 Disease Incidence at First Detection Average Age (Years) of Trees at First Detection 14 17 Strategy Strategy 1 2 3 1 2 3 20% 10% 5% 20% 10% 5% 0.001 16,487 13,470 12,895 12,939 12,961 17,101 14,084 13,509 13,553 13,575 0.010 14,488 13,204 12,495 12,739 12,861 15,102 13,818 13,110 13,353 13,475 0.020 13,322 12,910 12,262 12,622 12,802 13,935 13,525 12,876 13,236 13,417 0.030 12,505 12,620 12,099 12,541 12,762 13,118 13,235 12,713 13,155 13,376 0.040 11,978 12,333 12,015 12,488 12,735 12,591 12,947 12,629 13,102 13,349 0.050 11,389 12,0 49 11,876 12,429 12,706 12,002 12,663 12,490 13,043 13,320 0.060 11,032 11,768 11,804 12,393 12,688 11,644 12,382 12,418 13,007 13,302 0.070 10,521 11,489 11,702 12,342 12,672 11,133 12,103 12,316 12,956 13,287 0.080 10,250 11,213 11,648 12,315 12,649 1 0,861 11,828 12,261 12,929 13,263 0.100 9,786 10,670 11,555 12,269 12,626 10,396 11,284 12,168 12,883 13,240 0.200 7,799 8,111 11,158 12,070 12,526 8,405 8,725 11,770 12,683 13,140 0.300 6,604 5,790 10,919 11,951 12,466 7,207 6,404 11,531 12,564 13,080 0.400 5,278 3,686 10,653 11,818 12,400 5,877 4,300 11,265 12,431 13,014 0.500 4,779 1,784 10,554 11,768 12,375 5,375 2,399 11,164 12,380 12,988 1 Cumulative 15 year NPV ($/ac). 2 fo r strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 61

61 Figure 4 1. Net Present Value per Acre as a Function of Disease Incidence and Average Age (Years) of Trees at First Detection with Contour Lines for the Do Nothing Strategy Figure 4 2. Net Present Value per Acre as a Function of Disease Incidence and Average Age (Years) of Trees at First Detection with Contour Lines for Strategy 2

PAGE 62

62 Figure 4 3. Net P resent Value per Acre as a Function of Disease Incidence and Average Age (Years) of Trees at First Detection with Contour Lines for Strategy 3 (30% Yield Penalty) Figure 4 4 D ominant Strategy Given Disease Incidence at First Detection and Average Grove Age (Price = $1.50/p ound s olid 30% yield penalty for strategy 3)

PAGE 63

63 CHAPTER 5 SENSITIVITY ANALYSIS In this c hapter, the robustness of the model conclusions is tested by performing sensitivity analysis to determine how chang es the main parameters of the model affect the optimal s trategy mix. Changes in the age dependent rate of spread ( the disease spread, which in turn alte r fruit yield and net returns with a resulting impact on the optimal strategy. Changes in the p rice per pound solids directly impacts the net present value estimates. Other parameters that affect the optimal choice of the model include the period of latency and fruit yield. This c hapter considers the effects of changing prices, betas (rate of spread) and period of latency on optimal strategy. First, the impact of a price decrease from $1.50/pound solids to $1. 20 followed by a price increase from $1.50 to $1.80/pound solids are examined for each of the three age cohorts. Next, the age dependent rate of spread and the latency periods are also adjusted to observe their effect on model results The Effects of a Pr ice Decline Tables 5 1 through 5 3 presents the net present values for the three strategies for a delivered in price from $1.50 to $1.20 per pound solid for each o f the three age categories. In T able 5 1, irrespective of the strategy or disease incidence a t first detection, the age cohorts of 0 and 3 produces negative net present values when price falls. This trend is reversed for the mature groves with average ages of 6, 10, 14, and 17 where net present values are positive except at high incidence values of 30% to 50%, where some strategies still yield negative net present values. In Tables 5 2 and 5 3, at l ow disease incidence of 0.1% 10.0% ( at high incidence of 20% 50%) Strategy 1 (Strategy 3) is the superior strategy for groves with av erage ages of 6, 10, 14, and 17

PAGE 64

64 Overall, t he lowered price results in lower net present value for all groves at all levels of disease incidence A fall in price favors strategy 1 as it completely replaces strategies 2 and 3 at the lower levels of incidence of 0.1% 1 0%. The Effects of a Price Increase When price is increase from $1.50 to $1.80 per pound solid, the net present value is still negative for almost all levels of incidence for groves with average age of 0, but now positive for groves with average ages of 3 or more except at high incidence of 8% to 50% (30% to 50%) in which Strategy 1 (Strategy 2) posts negative net present values (Table 5 4). In T able 5 4, at low initia l disease incidence (0.1% to 7%), Strategy 2 is better than S trategies 1 and 3 for grove s with average age of 3. Thereafter, at initial disease incidence of 8% to 50%, Strategy 3 overtakes S trategy 2 as the best strategy in net present value. This result again confirms that at higher rates of infection, Strategy 3 is preferred over Strategy 2 because of the high tree removal rates associated with Strategy 2 For groves with average age of 6 or more, S trat egy 1 is only dominant at 0.1% to 1% level of initial incidence, wher eas S trategy 2 is dominant for all initial i ncidence rates ranging from 2% to 8%. When the initial inci dence rate exceeds 10% Strategy 3 takes over from S trategy 2 (Tables 5 5 and 5 6). I ncreased price results in higher net present value for all groves at all levels of disease inc idence The switch point (7%) between Strate gy 2 and 3 for groves of average age 3 do not change when price increase. This may be attributed to the fact that as price increase ; net presen t value of existing fruits increases making it more expensive to remove trees In F igure 5 1 (middle subplot) t he ranges of initial grove age and initial disease incidence for which each strategy maximizes net present valu e shows that when price falls, S trategy 1 (Strategy 3) is the optimal strategy for all groves when disease

PAGE 65

65 incidence ranges between 0.1 % to 10 % (20% to 50%). Therefore, when price is lowered, Strategy 1 replace s Strategy 2 (and some part of S trategy 3) as the optimal strategy for all groves when disease i nc idence ranges from 0.1% to 10 % In F igure 5 1 (bottom subplot) when price inc rease, the opt imal strategy is S trategy 2 for initial inc idence of 2 % to 8 % for all groves and for groves of 6 years or larger, Strategy 1 is opti mal at incidence of 0.1% to 1% For all groves at 10% to 50% incidence, Strategy 3 is the best strategy t 2% initial disease incidence for groves over 6 years is taken over by Strategy 2, and S incidence for groves older than 14 years is taken over by S trategy 3 The Effects of a Lower Annual Rate of Spread Tables 5 7 through 5 9 presents results for a lower annual rate of spread of HLB from Beta (1) = 1.5148125 0.3= 1.2148125; Beta (2) = 0.8450625 0.3 = 0.5450625; Beta (3) = 0.4440625 0.3 = 0.1440625, for the respective age cohorts. A lower rate of spread could be attributed to several factors, but most notably, if a psyllid control program proves effect ive its primary consequence would be a reduction in the rate of spread. When the betas are lowered, groves with average age of 0 show negative net present value s irrespective of the initial disease incidence (T able 5 7). For groves with average age of 3, S trategy 2 is superior from disease incid ence of 1% to 10%, after which S trategy 3 becomes superior from diseas e incidence of 20% to 50%. In Tables 5 8 and 5 9, S trategy 1 dominates from 0.1% to 10% init ial disease incidence, whereas S trategy 3 dominates thereafter from 20% to 50% initial disease incidence, for all matured groves with average ages of 6 or more. The decreased betas have resulted in higher net pres ent value for all groves at all levels of disease incidence.

PAGE 66

66 The Effects of an Increase d Annual Rate of Spread An increase in the annual rates of spread from Beta (1) = 1.5148125+ 0.3 = 1.8148125; Beta (2) = 0.8450625 + 0.3 = 1.1450625; Beta (3) = 0.44406 25+ 0.3 = 0.7440625; for the respective average age category, gives similar results for groves with average age of 0 and 3, in which the former presents negative net present values, and the later shows do minance for S trategy 2 at initial incidence of betwe en 0.1% to 2.0%, and thereafter from 3% to 50%, S trategy 3 is the best strategy. However, for grov es with average age of 6 or larger, S trat egy 1 does bes t for incidences of 0.1% only, S trategy 2 does be st for incidences of 1.0% to 6%, and S trateg y 3 does b est for incidences 7 % to 50%. The increased betas have resulted in smaller net present value for all groves at all levels of disease inc idence A reduction in the betas also affects the optimal strategy mix (F igure 5 2 ). Strategy 1 (Strategy 3) is the opt ima l strategy for groves 6 years or larger when initial disease incidence is 0.1% to 10% (20% to 50%) For groves with average age of 0 and 3 (0 or larger, i.e. all groves) S trategy 2 (Strategy 3 ) is optimal at incidence of 1% to 10% (20% to 50%) As a re sult of the reduction in the betas, S trategy 1 replaces Strategy 2 as the dominant strategy for groves older than 6 years at d isease incidence of 3% to 10% For an increase in the betas (Figure 5 2 bottom subplot ), S trategy 1 has the smallest area of opti mality, which occurs only at the lowest level of initial incidence of 0.1% for groves 6 years or larger. For all groves at incidence of 1% 6% (7% 50%), S trategy 2 (Strategy 3) is the best strategy Strat egy 2 replaces strategy 1 as dominant strategy fo r groves older than 6 years when disease incidence is between 1.0% and 2.0%, whereas Strategy 3 replaces S trategy 2 for groves older than 6 years wh en disease incidence is from 6% to 8 %

PAGE 67

67 The primary consequence of decreased rate of spread is to make Strat egy 2 more attractive. This result makes sense as the goal of Strategy 2 is to suppress the level of disease inoculum A lower rate of disease spread gives a grower more time to initiate Strategy 2 and thereby enjoy its benefits. Lower disease spread ra te also means that fewer trees are being removed early in the treatment period. This results in a smaller decrease in fruit revenue. The contrary effect emerges when the rate of spread is increased. Faster spread of the disease means the growers have a shorter window of opportunity to implement Strategy 2; Strategy 3 is preferred at younger ages of first detection and smaller levels of initial incidences at first detection. The Effects of a Shortened Latency Period The latency period refers to the in terval between the time a tree first becomes infected and when is expresses symptoms. The existence of the latency period is one of the most vexing dimensions of the disease in that a tree removal policy fails to eliminate all diseased trees. Bov (2012) has recently argued that the latency period may be shorter than that suggested in earlier literature on HLB. In this section we investigate the impact of a shorter latency period on the optimal strategy. Another dimension to this analysis is the efficacy o f scouting in detecting the disease Futch et al. (2009) argues that one pass through a grove where HLB is present will result in 50% of sym ptomatic trees being detected. Bov (2012) argument regarding latency is based upon the observation that symptomat ic trees may be present, but sco uts are unable to detect them. Therefore, improved detection techniques could reduce the latency period. Tables 5 13 through 5 15 presents results for the scenario when the latency period is reduced such that groves with age s of 0 and 3 now are assumed to have a

PAGE 68

68 latency period o f 6 months instead of 1 year, while the latency period of groves 6 years or larger remain unchanged at 2 years. Results show that all cases in which age of first detection is 0 display negative net pre sent values. Groves with average age of 3 displays net pr esent values in which Strategy 2 is best when disease incidence is 0.1% to 10.0%; Strategy 3 is best when disease incidence is 20.0% to 50.0%. For groves 6 years or larger, S trategy 1 does best from disease incidence of 0.1% to 1.0 %, aft er which Strategy 2 is best at 2.0% to 1 0 .0 % disease incidence followed by Strategy 3, which is optimal at incidence of 20.0% to 50.0% In F igure 5 3 the redu ction in latency period favors Strategy 2 more compared to Strategy 1 (and 3) for groves older than 6 years whe n disease incidence is 2% (10%) Strategy 3 is dominant at disease incidence of 2 0% to 50% for all groves. The change in latency has resulted in lower net present value for all groves at all levels of d isease inc idence Comparison of the results in Table 4 7 and Table 5 13, however, suggest that shortening the latency period doe s impact the optimal strategy. Under the baseline latency period, for groves of three years of age at first detection, Strategy 3 is superior for initial infection rates of 9% and higher, but with a shortened latency period, superiority of Strategy 3 shifts to initial infec tions rates of 10% and higher. While this is a small change, it does indicate that superior detection methods that could reduce the period of latency would benefit Strategy 2.

PAGE 69

69 Table 5 1. NPV 1 for the Three Strategies for Age Classes 0 and 3 from a Price Decline 2 Disease Incidence at First Detection Average Age of Trees at First Detection 0 3 Strategy Str ategy 1 2 3 1 2 3 0.001 3,986 3,397 4,655 521 347 1,115 0.010 5,043 5,902 4,972 1,626 26 1,759 0.020 5,321 6,950 5,056 2,320 418 1,967 0.030 5,436 7,553 5,090 2,794 788 2,109 0.040 5,494 7,969 5,108 3,013 1,137 2,175 0.0 50 5,610 8,280 5,142 3,174 1,468 2,223 0.060 5,653 8,529 5,155 3,480 1,782 2,315 0.070 5,686 8,730 5,165 3,594 2,080 2,349 0.080 5,712 8,901 5,173 3,690 2,365 2,378 0.100 5,747 9,174 5,183 3,843 2,895 2,424 0.200 5,884 9,922 5,225 4,472 4,979 2,613 0.300 5,905 10,259 5,231 4,887 6,456 2,737 0.400 5,970 10,436 5,250 5,218 7,568 2,837 0.500 5,983 10,532 5,254 5,345 8,433 2,875 1 Cumulative 15 year NPV ($/ac). 2 Price per pound soli d is reduced fr om $1.50 to $1.20 for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 70

70 Table 5 2. NPV 1 for the Three Strategies for Age Classes 6 and 10 from a Price Decline 2 Disease Incidence at First Detection Average Age of Trees at First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 0.001 5,980 2,852 2,260 8,149 5,025 4,433 0.010 4,562 2,680 1, 834 6,684 4,835 3,993 0.020 3,754 2,490 1,592 5,832 4,626 3,738 0.030 3,196 2,302 1,425 5,238 4,420 3,560 0.040 2,848 2,116 1,320 4,858 4,215 3,445 0.050 2,450 1,932 1,201 4,431 4,013 3,317 0.060 2,219 1,750 1,131 4,175 3,812 3,240 0.070 1,875 1,570 1,028 3,805 3,614 3,130 0.080 1,703 1,392 977 3,612 3,418 3,072 0.100 1,412 1,041 889 3,283 3,031 2,973 0.200 152 609 511 1,872 1,212 2,550 0.300 578 2,099 292 1,036 435 2,299 0.400 1,389 3,446 49 105 1,925 2,020 0.500 1,673 4,659 36 232 3,270 1,918 1 Cumulative 15 year NPV ($/ac). 2 Price per pound soli d is reduced from $1.50 to $1.20 for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 71

71 Table 5 3. NPV 1 for the Three Strategies for Age Classes 14 and 17 from a Price Decline 2 Disease Incidence at First Detection Average Age of Trees at First Detection 14 17 Strategy St rategy 1 2 3 1 2 3 0.001 9,575 6,452 5,860 10,028 6,904 6,312 0.010 8,103 6,256 5,418 8,555 6,708 5,871 0.020 7,243 6,039 5,160 7,695 6,492 5,613 0.030 6,642 5,826 4,980 7,093 6,278 5,432 0.040 6,253 5,614 4,863 6,705 6,067 5,316 0.050 5,819 5,405 4,733 6,271 5,857 5,185 0.060 5,556 5,197 4,654 6,007 5,650 5,106 0.070 5,180 4,992 4,541 5,630 5,445 4,993 0.080 4,980 4,789 4,481 5,430 5,242 4,933 0.100 4,638 4,389 4,379 5,088 4,841 4,830 0.200 3,174 2,503 3,939 3,620 2,955 4,390 0.300 2,293 793 3,675 2,738 1,245 4,125 0.400 1,316 757 3,382 1,758 305 3,831 0.500 948 2,159 3,272 1,387 1,706 3,720 1 Cumulative 15 year NPV ($/ac). 2 Price per pound soli d is reduced from $1.50 to $1.20 ld for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 72

72 Table 5 4. N PV 1 for the Three Strategies for Age Classes 0 and 3 from a Price Increase 2 Dis ease Incidence at First Detection Average Age of Trees at First Detection 0 3 Strategy Strategy 1 2 3 1 2 3 0.001 1,371 2,108 314 7,165 9,313 7,176 0.010 3,241 2,197 247 3,480 8,671 6,140 0.020 3,744 4,006 398 2,286 7,998 5,712 0.030 3,95 6 5,050 461 1,469 7,363 5,467 0.040 4,065 5,773 526 1,090 6,762 5,353 0.050 4,275 6,315 557 810 6,194 5,269 0.060 4,355 6,749 581 282 5,655 5,111 0.070 4,418 7,102 600 85 5,142 5,052 0.080 4,467 7,402 615 83 4,653 5,001 0.100 4,533 7,883 635 351 3,741 4,921 0.200 4,791 9,216 712 1,449 157 4,592 0.300 4,832 9,828 724 2,175 2,386 4,374 0.400 4,954 10,155 761 2,758 4,305 4,199 0.500 4,980 10,334 769 2,983 5,796 4,131 1 Cumulative 15 year NPV ($/ac). 2 Price p er pound solid is increased from $1.50 to $1.80. for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Lar ger

PAGE 73

73 Table 5 5. N PV 1 for the Three Strategies for Age Classes 6 and 10 from a Price Increase 2 Disease Incidence at First Detection Average Age of Trees at First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 0.001 16,947 14,030 13,384 20,952 18,043 17,397 0.010 14,516 13,734 12,655 18,441 17,717 16,643 0.020 13,131 13,409 12,239 16,982 17,359 16,205 0.030 12,176 13,087 11,953 15,964 17,005 15,900 0.040 11,578 12,768 11,773 15,311 16,654 15,704 0.050 10,896 12,453 11,569 14,579 16,307 15,4 85 0.060 10,500 12,141 11,450 14,140 15,964 15,353 0.070 9,911 11,832 11,273 13,507 15,624 15,163 0.080 9,615 11,527 11,185 13,175 15,287 15,063 0.100 9,117 10,926 11,035 12,610 14,625 14,894 0.200 6,957 8,098 10,387 10,192 11,506 14,169 0.300 5,705 5,542 10,011 8,759 8,683 13,739 0.400 4,314 3,234 9,594 7,163 6,128 13,260 0.500 3,827 1,154 9,448 6,585 3,822 13,086 1 Cumulative 15 year NPV ($/ac). 2 Price per pound solid is increased from $1.50 to $1.80. Yield from HLB infected trees reduced 30% on for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 74

74 Table 5 6. N PV 1 for the Three Strategies for Age Classes 14 and 17 from a Pric e Increase 2 Disease Incidence at First Detection Average Age of Trees at First Detection 14 17 Strategy Strategy 1 2 3 1 2 3 0.001 23,398 20,488 19,843 24,174 21,264 20,619 0.010 20,873 20,152 19,086 21,649 20,928 19,862 0.020 19,401 19,781 18,644 20,175 20,557 19,420 0.030 18,369 19,415 18,335 19,143 20,191 19,110 0.040 17,703 19,052 18,135 18,477 19,828 18,910 0.050 16,959 18,693 17,912 17,733 19,469 18,687 0.060 16,508 18,338 17,776 17,281 19,114 18,551 0.070 15,863 17,986 17,583 16,635 18, 762 18,358 0.080 15,520 17,638 17,480 16,292 18,414 18,254 0.100 14,934 16,952 17,304 15,705 17,728 18,078 0.200 12,424 13,719 16,551 13,190 14,494 17,324 0.300 10,915 10,787 16,098 11,676 11,563 16,870 0.400 9,239 8,130 15,596 9,996 8,906 16,366 0.5 00 8,609 5,727 15,407 9,362 6,503 16,175 1 Cumulative 15 year NPV ($/ac). 2 Price per pound solid is increased from $1.50 to $1.80. for strategy 3 Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger

PAGE 75

75 Table 5 7. N PV 1 for the Three Strategies for Age Classes 0 and 3 from a Decline in Beta 2 Disease Incidence at First Detection Average Age of Trees at First Detection 0 3 Strategy Strategy 1 2 3 1 2 3 0.001 2,065 46 1,986 5,872 4,859 3,639 0.010 3,872 1,462 2,529 2,498 4,595 2,627 0.020 4,228 2,650 2,635 1,394 4,308 2,296 0.030 4,526 3,540 2,725 637 4,026 2,068 0.040 4,649 4,243 2,762 274 3,750 1,960 0.050 4,731 4,817 2,786 262 3,479 1,799 0.060 4,893 5,298 2,802 489 3,213 1,731 0.070 4,946 5,710 2,851 677 2,951 1,674 0.080 4,991 6,067 2,864 1,107 2,695 1,546 0.100 5,059 6,663 2,885 1,373 2,195 1,465 0.200 5,2 94 8,417 2,955 2,422 62 1,151 0.300 5,346 9,330 2,971 3,106 1,987 946 0.400 5,449 9,909 3,002 3,655 3,649 781 0.500 5,470 10,303 3,008 4,112 5,095 644 1 Cumulative 15 year NPV ($/ac re ). 2 Beta (1) =1.5148125 0.3 = 1.2148125 for th e 0 Age Class ; Beta (2) = 0.8450625 0.3= 0.5450625 for age Class of 3 ; Beta (3) = 0.4440625 0.3= 0.1440625 for Age Classes of 6 or Larger for strategy 3

PAGE 76

76 Table 5 8. N PV 1 for the Thre e Strategies for Age Classes 6 and 10 from a Decline in Beta 2 Disease Incidence at First Detection Average Age of Trees at First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 0.001 11,826 8,449 7,931 14,918 11,542 11,025 0.010 11,288 8,285 7,769 14, 351 11,361 10,855 0.020 10,759 8,104 7,611 13,791 11,160 10,687 0.030 10,289 7,922 7,469 13,291 10,959 10,537 0.040 9,866 7,742 7,343 12,840 10,759 10,402 0.050 9,483 7,561 7,228 12,430 10,560 10,278 0.060 9,132 7,381 7,123 12,053 10,361 10,165 0.070 8,810 7,202 7,026 11,706 10,162 10,061 0.080 8,513 7,023 6,937 11,383 9,964 9,965 0.100 7,978 6,667 6,776 10,802 9,570 9,790 0.200 6,081 4,914 6,207 8,710 7,629 9,163 0.300 4,827 3,209 5,842 7,306 5,740 8,752 0.400 3,106 1,550 5,393 5,454 3,901 8,266 0.500 1,303 63 4,774 3,441 2,113 7,582 1 Cumulative 15 year NPV ($/ac re ). 2 Beta (1) =1.5148125 0.3 = 1.2148125 for the 0 Age Class ; Beta (2) = 0.8450625 0.3= 0.5450625 for age Class of 3 ; Beta (3) = 0.4440625 0.3= 0.1440625 for Age Classes of 6 or Larger for strategy 3

PAGE 77

77 Table 5 9. N PV 1 for the Three Strategies for Age Classes 14 and 17 f r om a Decline in Beta 2 Disease Incidence at First Detection Average Age of Trees at First D etection 14 17 Strategy Strategy 1 2 3 1 2 3 0.001 16,854 13,479 12,962 17,469 14,093 13,576 0.010 16,280 13,290 12,790 16,894 13,904 13,404 0.020 15,712 13,081 12,619 16,326 13,695 13,233 0.030 15,205 12,872 12,467 15,818 13,487 13,081 0.040 14 ,746 12,664 12,329 15,358 13,278 12,943 0.050 14,327 12,457 12,204 14,940 13,071 12,817 0.060 13,943 12,249 12,089 14,555 12,864 12,702 0.070 13,588 12,043 11,982 14,200 12,657 12,596 0.080 13,259 11,837 11,883 13,870 12,451 12,497 0.100 12,663 11,427 11,705 13,274 12,041 12,318 0.200 10,505 9,407 11,057 11,112 10,021 11,669 0.300 9,043 7,440 10,629 9,646 8,054 11,240 0.400 7,138 5,525 10,127 7,738 6,139 10,737 0.500 5,062 3,661 9,424 5,658 4,276 10,033 1 Cumulative 15 year NPV ($/ac re ). 2 Beta ( 1) =1.5148125 0.3 = 1.2148125 for the 0 Age Class ; Beta (2) = 0.8450625 0.3= 0.5450625 for age Class of 3 ; Beta (3) = 0.4440625 0.3= 0.1440625 for Age Classes of 6 or Larger for strategy 3

PAGE 78

78 Table 5 10. N PV 1 for the Three Strategies for Age Classes 0 and 3 from an Increase in Beta 2 Disease Incidence at First Detection Average Age of Trees at First Detection 0 3 Strategy Strategy 1 2 3 1 2 3 0.001 3,104 2,466 2,298 2,514 4,7 12 2,632 0.010 4,388 6,016 2,683 60 3,327 1,896 0.020 4,683 7,247 2,772 784 2,110 1,642 0.030 4,789 7,836 2,804 1,337 1,123 1,476 0.040 4,966 8,237 2,857 1,444 297 1,395 0.050 5,039 8,535 2,879 1,810 409 1,335 0.060 5,093 8,771 2,895 1,963 1,023 1,289 0.070 5,131 8,964 2,906 2,081 1,564 1,253 0.080 5,157 9,126 2,914 2,405 2,047 1,156 0.100 5,181 9,387 2,921 2,602 2,877 1,097 0.200 5,374 10,110 2,979 3,341 5,532 875 0.300 5,378 10,468 2,980 3,640 7 ,058 786 0.400 5,473 10,691 3,009 4,042 8,090 665 0.500 5,492 10,479 3,015 4,214 8,832 613 1 Cumulative 15 year NPV ($/ac re ). 2 Beta (1) = 1.5148125 + 0.3 = 1.8148125 for age class of 0 ; Beta (2) = 0.8450625 + 0.3= 1.1450625 for age class of 3 ; Beta (3) = 0.4440625 + 0.3= 0.7440625 for age classes 6 or larger for strategy 3

PAGE 79

79 Table 5 11. N PV 1 for the Three Strategies for Age Classes 6 and 10 f r om an Increase in Beta 2 Disease Incidence at First Detection Average Age of Trees at First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 0.001 10,474 8,423 7,525 13,551 11,514 10,615 0.010 7,698 8,025 6,692 10,658 11,082 9,747 0.020 6,637 7,597 6,374 9,514 10,617 9,404 0.030 6, 085 7,183 6,208 8,901 10,166 9,220 0.040 5,505 6,781 6,034 8,267 9,728 9,030 0.050 5,187 6,391 5,939 7,909 9,303 8,922 0.060 4,929 6,012 5,862 7,615 8,890 8,834 0.070 4,512 5,644 5,737 7,156 8,489 8,696 0.080 4,317 5,286 5,678 6,932 8,098 8,629 0.100 3,992 4,600 5,580 6,558 7,347 8,517 0.200 2,767 1,641 5,213 5,148 4,100 8,094 0.300 1,926 718 4,961 4,174 1,498 7,802 0.400 1,237 2,635 4,754 3,370 629 7,560 0.500 875 4,197 4,645 2,940 2,369 7,432 1 Cumulative 15 year NPV ($/ac re ). 2 Beta (1) = 1.5148125 + 0.3 = 1.8148125 for age class of 0 ; Beta (2) = 0.8450625 + 0.3= 1.1450625 for age class of 3 ; Beta (3) = 0.4440625 + 0.3= 0.7440625 for age classes 6 or larger for strategy 3

PAGE 80

80 Table 5 12. N PV 1 for the Three Strategies for Age Classes 14 and 17 f r om an Increase in Beta 2 Disease Incidence at First Detection Average Age of Trees at First Detection 14 17 Strategy Strategy 1 2 3 1 2 3 0.001 15,487 13,450 12,552 16,101 14, 064 13,166 0.010 12,580 13,008 11,679 13,193 13,622 12,294 0.020 11,421 12,531 11,332 12,034 13,145 11,946 0.030 10,794 12,068 11,144 11,407 12,682 11,758 0.040 10,147 11,619 10,950 10,760 12,233 11,564 0.050 9,777 11,183 10,839 10,389 11,797 11,452 0.060 9,472 10,758 10,747 10,084 11,372 11,361 0.070 9,002 10,345 10,606 9,613 10,960 11,220 0.080 8,769 9,944 10,536 9,379 10,558 11,149 0.100 8,375 9,171 10,418 8,985 9,785 11,031 0.200 6,889 5,820 9,972 7,495 6,435 10,584 0.300 5,854 3,123 9,662 6, 456 3,737 10,272 0.400 4,995 911 9,404 5,594 1,525 10,014 0.500 4,526 907 9,263 5,121 293 9,872 1 Cumulative 15 year NPV ($/ac re ). 2 Beta (1) = 1.5148125 + 0.3 = 1.8148125 for age class of 0 ; Beta (2) = 0.8450625 + 0.3= 1.1450625 for age class of 3 ; Beta (3) = 0.4440625 + 0.3= 0.7440625 for age classes 6 or larger for strategy 3

PAGE 81

81 Table 5 13. NPV 1 for the Three Strategies for Age Classes 0 and 3 from a Lowered Latency Period 2 Diseas e Incidence at First Detection Average Age of Trees at First Detection 0 3 Strategy Strategy 1 2 3 1 2 3 0.001 4,195 192 2,625 1,623 4,861 2,364 0.010 4,898 334 2,836 940 4,614 1,595 0.020 5,152 871 2,913 1,417 4,344 1,452 0.030 5,213 1,365 2,931 2,014 4,077 1,273 0.040 5,244 1,821 2,940 2,199 3,815 1,218 0.050 5,259 2,243 2,945 2,340 3,556 1,176 0.060 5,380 2,634 2,981 2,451 3,301 1,142 0.070 5,392 2,999 2,985 2,542 3,050 1,115 0.080 5,402 3,340 2,988 2,950 2,802 993 0.100 5,417 3,958 2,992 3,075 2,316 955 0.200 5,437 6,188 2,998 3,751 71 752 0.300 5,498 7,590 3,016 3,928 1,907 699 0.400 5,500 8,558 3,017 4,353 3,665 571 0.500 5,501 9,258 3,017 4,440 5,234 546 1 Cumulative 15 year NPV ($/ac). 2 Latency is now 6 months for ages 0 and 3, and 2 year s for a ges of 6. Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger Yield from HLB infected trees redu for strategy 3

PAGE 82

82 Table 5 14. N PV 1 for the Three Strategies for Age Classes 6 and 10 f r om a Lowered Latency Period 2 Disease Incidence at First Detection Average Age of Trees at First Detection 6 10 Strategy Strategy 1 2 3 1 2 3 0.001 11,287 8,447 7,769 14,372 11,540 10,861 0.010 9,181 8,264 7,137 12,193 11,337 10,207 0.020 8,045 8,061 6,796 10,994 11,112 9,848 0.030 7,457 7,859 6,620 10,355 10,888 9,656 0.040 6,849 7,658 6,438 9,703 10,665 9,460 0.050 6,504 7,457 6,334 9,320 10,443 9,345 0.060 6,006 7,258 6,185 8,785 10,223 9,185 0.070 5,761 7,060 6,111 8,509 10,003 9,102 0.080 5,547 6,863 6,047 8,267 9,784 9,030 0.100 4,944 6,471 5,866 7,608 9,350 8,832 0.200 3,533 4,567 5,443 6,006 7,240 8,351 0.300 2,254 2,754 5,059 4,554 5,230 7,916 0.400 1,453 1,031 4,819 3,623 3,320 7,636 0.500 1,075 603 4,705 3,174 1,507 7,502 1 Cumulative 15 year NPV ($/ac). 2 Latency is now 6 months for ages 0 and 3, and 2 year s for a ges of 6. Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.4440625 for Age Classes of 6 or Larger for strategy 3

PAGE 83

83 Table 5 15. NPV 1 for the Three Strategies for Age Classes 14 and 17 f r om a Shortened Latency Period 2 Disease Incidence at First Detection Average Age of Trees at First Detection 14 17 Strategy Strategy 1 2 3 1 2 3 0.001 16,308 13,476 12,798 16,923 14,091 13,412 0.010 14,118 13,265 12,141 14,732 13,879 12,755 0.020 12,907 13,030 11,778 13,521 13,644 12,392 0.030 12,257 12,797 11,583 12,870 13,411 12,197 0.040 11,594 12,565 11,384 12,207 13,179 11,998 0.050 11,202 12,334 11,266 11,814 12,948 11,880 0.060 10,657 12,104 11,103 11,268 12,718 11,716 0.070 10,372 11,8 75 11,017 10,983 12,489 11,631 0.080 10,120 11,647 10,942 10,731 12,262 11,555 0.100 9,444 11,195 10,739 10,054 11,809 11,352 0.200 7,770 8,997 10,237 8,376 9,611 10,848 0.300 6,255 6,903 9,782 6,857 7,517 10,393 0.400 5,265 4,912 9,485 5,864 5,526 10 ,095 0.500 4,777 3,023 9,339 5,372 3,637 9,947 1 Cumulative 15 year NPV ($/ac). 2 Latency is now 6 months for ages 0 and 3, and 2 year s for a ges of 6. Beta (1) = 1.5148125 for the 0 Age Class ; Beta (2) = 0.8450625 for age Class of 3 ; Beta (3) = 0.444062 5 for Age Classes of 6 or Larger for strategy 3

PAGE 84

84 Figure 5 1. D ominant Strategy Given Disease Incidence at First Detection and Average Grove Age from a Change in Price : Top Subplot i s Bas eline, Middle and Bottom Subplots Shows Price Decline (fr om $1.50 to $1.20 ) and Increase ( from $1.50 to $1.80) respectively Figure 5 2. D ominant Strategy Given Disease Incidence at First Detection and Average Grove Age from a Change in Beta : Top Subplot is Baseline, Midd le and Bottom Subplots Shows Beta Decline and Increase, r espectively

PAGE 85

85 Figure 5 3. D ominant Strategy Given Disease Incidence at First Detection and Average Grove Age from a Change in Latency: Top Subplot is Baseline, Bottom Su bplot Shows Decline in Latency from 1 year to 6 Months for Groves with Average Age of 0 and 3 while the Latency for Groves 6 Years or Larger Remain at 2 Years

PAGE 86

86 CHAPTER 6 CONCLUSIONS, RECOMMENDATIONS AND LIMITATIONS The preceding c hapters have sought to de velop a management strategy for HLB at the grove level for citrus producers in Florida. A baseline model was developed to as a benchmark for the modeling of the infected tree removal strategy as well as the enhanced foliar nutritional strategy. The adoption of a particular strategy by a grower is seen to be a function of certain grove characteristics. This research has identified the various zones of op timality for each st rategy given HLB infection rate. This research attempted to integrate the intricate biological realities of HLB into an economic decision making framework for producers. The basis of the biological model is provided by th e works of Bassanezi and Bassanezi (2008) and Bassanezi et al. (2011). This research demonstrate that the complex biological features of HLB can be transformed into an economic decision maki ng process for citrus growers. The effect of latency on control ef fective ness especially when employing S trategy 2 for example is addressed in this analysis. The most important contribution of this research is the incorporation of both symptomatic and asymptomatic trees in the logistic spread curves used in the analysis. This ensures that even if sympto matic trees are removed (as in S trategy 2) ; spread through asymptomatic trees is still accounted for in the model. The knitted into e ach of the three models of control strategies in this analysis. Additionally, i n varying the level of initial infection in the analysis, this research also demonstrates the heterogeneous effects of HLB to NPV at the landscape level, whereby optimal control

PAGE 87

87 decisions varies across neighbors. The significance of this research lies in its ability to address these characteristics and formulate optimal control policy for effective decision making. In summary, we find that g roves that contain younger trees at fi rst detection have low or negative net present value due to the faster spread of the disease in younger groves in addition to low production from young groves For S trategy 1 all groves with an average age of 6 years and larger will yield a positive net p resent value irrespective of the initial level of infection For S trategy 2, except when initial incidence is 40% to 50%, all groves with an average age of 6 or larger yields a p ositive net present value. For S trategy 3, all groves with an average age of 3 or larger at all initial incidence levels yields a positive net present value. Whether cost exceeds revenue in production is a function of disease incidence and average grove age. The higher the initial incidence level (the larger the average grove age) the more likely (less likely) that cost of production exceeds revenue from production. Finally, we find that the optimal strategy to adopt by a grower depends on the average grove age at first detection and the initial rate of disease incidence at first detection. Irrespective of the average grove age, once initial incidence is 20% or larger in a grove, S trategy 3 should be implemented. The intuition of this recommendation is that it is better to incur the extra costs of nutritional supplements than remov e 20% or more of productive but sick trees or even do nothing The marginal revenue from this action is more than its marginal cost. A t higher rates of infection, Strategy 3 is preferred over Strategy 2 because of the high tree removal rates associated wit h Strategy 2 Implementation o f S trategy 2 requires that initial incidence should be 3% to 8% for

PAGE 88

88 groves 6 years or larger or 0.1% 2% for groves of average age 0 or 3 Here, the intuition is that removing 3% to 8% of infected mature trees or 0.1% 2% of newly established trees is more cost effective compared to spraying such trees with nutritional supplements or doing nothing. When there is virtually no infection (0.1% to 1%) in a grove of age 6 or larger, doing nothing is in the best interest of the pro ducer. The relationship between the net present value and model parameters such as delivered in price, the rate of spread of HLB, and latency has been established through the sensitivity analysis. It is found that net present value is positively related to price but negatively related to the rate of spread and the latency period. Changes in these parameters also results in changes in the optimal strategy mix. In particular, results indicate that superior detection methods that could reduce the period of lat ency would benefit Strategy 2. Results also suggest that the primary consequence of decreased /increased rate of spread (our proxy for psyllid control) is to make Strategy 2 /Strategy 3 more attractive. The rate of spread of HLB is related to three factor s including average grove age, initial incidence at first detection, and the psyllid population Even though our model did not directly incorporate psyllid control into the analysis reduction in the annual rate of HLB spread investigated via the sensitivi ty analysis can serve as proxy for psyllid control because total elimination of psyllids notably terminates HLB spread E ffective control of plant diseases that involves spread by vectors and other weather factors such as HLB requires a model that recogn izes landscape management characteristics, which is missing in this analysis for now. Neighbor effects negatively affects heterogeneous management protocols by adjacent growers since buildup of

PAGE 89

89 bacteria titer in a grove practicing Strategy 1 or 3 could dim inish the inoculum reduction objective of a neighbor practicing Strategy 2. One other drawback is the lack of HLB spread data from Florida required to estimate the model paramete rs. Although the parameters used may not be representative of the HLB situatio n in Flori da, the results derived here do serve as a guide and reference point for growers and policy makers in the industry. The assumption of no resetting greatly simplifies the calculation of disease spread and the accompanying reduction in fruit produc tion per acre. However, this assumption clearly is a limitation on the derived results. The lack of knowledge on the underlying distribution of the key variables that affect net present value in the presence of HLB has forced us to proceed with the analysi s in a det erministic framework. C lear y, stochastic dominance would have been the best estimator of superiority. It is hoped that future research can be greatly enhanced when most of these limitations are addressed.

PAGE 90

90 LIST OF REFERENCES Albrecht, U., T. G. on Huanglongbing Disease Development in Field grown Sweet Orange (Citrus Scientia Horticulturae 138:210 220. Altamirano, D. M., C. I. Gonzales, and R. C. Vinas. of Leaf mottling (Greening) Disease of Citrus and its Control Program in the Proc. Conf. Int. Org. Citrus Viral. 7 th 22 26. Integrated Activities for the Control of H uanglungbin G reening and its V ector Diaphorina citri 144 in: B. Aubert, S. Tontyaporn, D. Buangsuwon,, eds., Rehabilitation of Citrus Industry in the Asia Pacific Region. Proc. of the Asia Pacific Intl. Conf. on Citriculture, Chiang Mai, Thailand 4 10 February 1990. UNDP FAO, Rome. Proceedings of the 7th International Citrus Congress 817 820. cillin or Tetracycline Injections of Citrus Proceedings of 8th Conference of IOCV Riverside 103 108. Aubert B., J. M. Bov and J. Etienne. 1980. La Lutte Contre la Maladie du G reening Rsultats et Perspectives Fruits 35: 605 624. Region: Monitoring Flight Activity of Diaphorina citri on Citrus and Murraya C anopies. In Proceedings of the 4th Asia Pacific International Conference on Citriculture eds. A. Bernard, T. Sanchai and B. D., pp. 181 187, February 4 10, at Chiang Mai, Thailand. FAO UNDP, Rome. Aubert, B., M. Garnier, D. Guillaumin, B. Herbagyandodo, L Setiobudi, and F. Nurhadi. Future prospects of integrated control. Fruits 40:549 563. Study of 278 In J. V. da Graa, P. Moreno, and R. K. Yokomi [eds.], Proc. 13th Conference of the International Organization of Citrus Virologists (IOCV). University of California, Riverside.

PAGE 91

91 Aubert, B., Psyllids (Homoptera: Psylloidea), through Eulophid and Encyrtid Parasites In Proceedings of the 9th Conference of International Organ ization of Citrus Virologists eds. S.M. Garnsey, L.W. Timmer and J.A. Dodds, pp. 100 108, May 9 13, 1983, at Argentina. Riverside, CA: IOCV. Available online at http://www.ivia.es/ iocv/archivos/proceedingsIX/9th100_108 .pdf. Disease in Reunion Island before and after the Biological Control of the African In Proceedings of the 5th Meeting of the International Society of Citriculture, 2: 440 442, n.d., at So Paulo, Brazil. Gainesville: University of Florida. Ayres A.J., C. A Massari, S. A. Lopes, R. B. Bassanezi, Junior J. Belasque, P. T. Yamamoto, D. C. Teixeira, N. A. Wulff, N. Gimenes Fernandes, and O. A. Be rgamaschi. 2005. Proceedings of 2nd International Citrus Canker and Huanglongbing Research Workshop, Florida Citrus Mutual, Orlando 2005, H 4. Bassanezi R. B., L. A. Busato, A. Bergamin Filho, L. Amorim, and T. R. Gottwald. 2005. Proc. 16th Conf. Intern. Org. Citrus Virol. 341 55. IOCV, Univ. Calif., Riverside, CA. Bassanezi R. B., A. Bergamin Filho, L. Amorim, and T. R. Gottwald. 2006. Proceedings of Huanglongbing Greening International Workshop Ribeiro Preto. p.37. International Research Conference On Huanglongbing (IRC HLB) Proceedings Orlando, Florida Dec. 2008. Bassanezi, R. B., L. H. Montesino, M. C. G. Gasparoto, A. Bergamin Filho, and L. European Journal of Plant Pathology 130:577 86. Lett. Spat. Resour. Sci. 1 : 107 115. Belasque, Jr., J., N. G. Fernandes, Summa Phytopathologica 35(2):91 92. rdpress.com/2012/04/06/citrus under siege/

PAGE 92

92 FAO Plant Prot. Bull 34:7 14. old Dis ease Journal of Plant Pathology, 88: 7 37. Bov J. M. ade at a meeting at the Southwest Florida Research and Education Center, Immokalee, FL Brlansky, R.H., K. R. Chung and M. Huanglongbing (Citrus Greening). University of Florida IFAS. PP 225. http://edis.ifas.ufl.edu Management Guide: Huanglongbing (Citru 225. Gainesville: Institute of Food and Agricultural Sciences, University of Florida. Available online at http://edis.ifas.ufl.edu/cg086 Brlansky, R.H., M.M. Dewdney, M.E. Rogers, and K. R. Chung. 2009. Huanglongbing (C itrus G reening). 2009 Florida Citrus Pest Management Guide. Publication #PP 225. Gainesville: Institute of Food and Agricultural Sciences, University of Florida. Available online at http://edis.ifas.ufl.edu/CG086 Amer. J. Agr. Econ 84(2): 279 291. Carrasco, L. R., A. MacLeod, J. D. Knight, R. Baker, and J. D. Mu Optimal Control of Spreading Biological Invasions: For how long should We Apply the Paper Presented at the 83rd Annual Conference of the Agricultural Economics Society, 30 March 1 April, Dublin, Ireland. Catling, H.D., and P.R. Atki Trioza erytreae (Del In Proceedings of the 6th Conference of International Organization of Citrus Virologists, eds. L.G. Weathers and M. Cohen, pp. 33 39, August 21 28, at Swaziland. Riverside, CA: IOCV. Journal of Applied Ecology 31 (3): 413 427. Chalak Haghighi, M., E. C. Van Ierland, G. W. Bourdt & D. Leathwick (2008): Managemen t Strategies for an Invasive Weed: A Dynamic Programming Approach for Californian Thistle in New Zealand, New Zealand Journal of Agricultural Research, 51:4, 409 424. Experi ment Station of Lingnan University, PRC. China 3:169 191.

PAGE 93

93 Chiyaka, C., B. H. Singer, S. E. Halbert, J. G. Morris Jr., and A H. C. van Bruggen. 2012. ransmission Within a Citrus T Proceedings of the National Academy of Sci ences of the United States of America 2012. In Environmental and Resource Economics 23(3): 319 342. Control Methods of the Mediterranean Fruit Fly (Diptera: Tephritidae) in Israel, Palestine Journal of Economic Entomology 90:1066 1072. Journal of Mathematical Biology 21:149 158. Fishman, S., R. Marcus, H. Talpaz, M. Bar Joseph, Y. Oren R. Salomon, and M. Zohar. Phytoparasitica 11:39 49. Futch. S., S. Weingarten, and M. Irey. 2009. sing Multiple Survey Methods in Florida C itrus. In: Proceedings Florida State Horticultural Society (FSHS), 122, pp. 152 158. Gatineau, F., T. H. Loc, N. D. Tuyen, T. M. Tuan, N. T. Hien, and N. T. N. Truc. 2006. Diaphorina Citri Proceedings of Huanglongbing Greening International Workshop Ribeiro Preto, Brazil. p.110. Canker Risks: I Amer. J. Agr. Econ. 91(4) (November 2009): 1038 1055. Copyright 2009 Agricultural and Applied Economics Association DOI: 10.1111/j.1467 8276.2009.01321.x Gottwald TR, Aubert B, Huang KL. 1991. Spa tial Pattern Analysis of Citrus G reening in Shantou, China. Proc. 11th I.O.C.V., 421 427. Gottwald, T. R., M. S. Irey, T. Gast, S. R. Parnell, E. L. Taylor and M. E. Hilf 2010. Spatio temporal Analysis of an HLB Epidemic in Florida and Implications for S pread Proceedings, 17th Conference IOCV, 2010 Insect Transmitted Procaryotes Annual Review of Phytopathology. 48:119 39.

PAGE 94

94 Gottwald, T., M. Irey, T. Gast, A. Bergamin Filho, R. Bassanezi, and C. A. Gilligan. Proc. Int. Res. Conf. Huanglongbing 285 90. Gottwald, T. R., M. Irey, T. Gast, S. Parnell, E. Taylor, and temporal Analysis of an HLB Epidemic in Florida and Implications for Future In Proceedings of the 17th Conference of the International Organization of Citrus Virologists, p. 139, October 22 26, 2007, at Adana, Turkey. IOC V, Riverside, CA Plant Health Progress, doi: 10.1094/PHP 2007 0906 01 RV. Gottwald, T.R., M. Irey, T. Gast, S. Parnell, E. Taylor, and M. E. H Spatio temporal A nalysis of an HLB Epidemic in Florida and Implications for Future S In: Proceedings of the 17th Conference of the International Organization Citrus Virologists Univ. California, Riverside (in press). Gottwald, T.R., B. In Proceedings of the 11th Conference of the International Organization of Citrus Virologists eds. R.H. Brlansky, R.F. Lee and L.W. Timmer, pp. 421 427, Novembe r 6 10, 1989, at Orlando, FL. Riverside, CA: IOCV. Available online at http://www.ivia.es/iocv/archivos/ Proceedings of the International Workshop on Citrus Greening, July 14 21, Rib eiro Preto, Brazil. p.67 68. Gottwald, T. R., J. H. Graham, M. S. Irey, T. G. McCollum, and B. W. Wood. 2012. Crop Protection 36: 73 82. Progress of Citrus Greening Epidemics in the People 's Republic of China and Phytopathology 79:687 93. Gottwald, T. R., C. I. Gonzales, a Proc. Conf. Int. Org. Citrus Virol. 11th. In press P syllidae) and greening disease in citrus: a literature review and assessment of Fla. Entomol. 87:330 354. Diffusing Nuisance Natural Resource Modeling 6(1): 71 97.

PAGE 95

95 Resource and Energy Economics, 32:519 533. Hodges, A. W. and T. H. Spreen omic Impacts of Citrus Greening (HLB) in Florida, 2006/07 Resource Economics Department, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida, Gainesville. Replant Young Groves in an Area with Endemic HLB: a Hierarchical Sampling Proc. Int. Res. Conf. Huanglongbing pp. 116 17. Ir ey, M.S., T. Gast, and T.R. Gottwald. 2006. Comparison of Visual Assessment and P olymerase C hai n Reaction Assay Testing to Estimate the Incidence of the Huanglongbing P athogen in Commercial Florida C itrus. Proceedings of the Florida State Horticultural Soc iety 119(2006):89 93. Irey, M., R. A Morris, and M. Estes. 2011. Survey to Estimate the Rate of HLB I nfection in Florida Citrus G roves. plantmanagementnetwork.org/proceedings/irchlb/2011/. p. 73. Theoretical Aspects of E pidemics: Uses of Canadian Journal of Plant Pathology 17(2):109 114. Determinants of Investment Behavior, R. Ferber, UMI, ISBN: 0 87 014 309 3, pp. 129 188. Based Simulation Model for the Spread of Citrus Greening Disease by the Vector Insect JIRCAS Working Report No. 72. Kompas, T., and T. N. A Practical Optimal Surveillance Measure: The Case of Papaya Fruitfly in Australia Environmental Economics, Crawford School of Economics and Government, Canberra, ACT. Acta Phytophylactica Sinica, 2:243 251. Lett Spat Resour Sci 2:123 131. DOI 10.1007/s12076 009 002 9 5

PAGE 96

96 Manjunath, K. L., S. E. Halbert, C. Ramadugu, S. Webb, and R. F. Lee. 2008. Candidatus Diaphorina Citri and its Phytopathology, 98:387 396. Mart Proc. 1st Int. Citrus Symp 3: 1427 31. wide Fruit Fly Pest Management Program: Influence of Partnership s and a Good Education Area wide Control of Insects: From research to field implementation J. B. Vreysen, A. S. Robinson and J. Hendrichs, (eds.), IAEA, Vienna, Austria. pp 671 683. Michaud, J. P. syllid, Diaphorina Citri Entomological News, 113(3):216 222. ication of the Food and Resource Economics Department, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. t Tree Paper presented at the Southern Agricultural Economics Association Annual Meeting February 2 6, in Dallas, TX. Available online at http://purl.umn.edu/6309. M Presented at 2010 Citrus Expo. Ft. Meyers, FL. May 19, 2010. 2008 Citrus Budget for the Indian River a, Institute of Food and Agricultural Sciences. Available online at http://www.crec.ifas.ufl.edu/extension/ economics/pdf/IR_Budget_Summ_2007 2008.pdf. 2008 Citrus Budget for the Central Florida and Agricultural Sciences. http://www.crec.if as.ufl.edu/extension/economics/pdf/CF_Budget_Summ_2007_2 008.pdf Annual Review of Entomology 43(1): 471 491.

PAGE 97

97 n Strategic Planning for Strategic Planning for the Florida Citrus Industry: Addressing Citrus Greening Washington, DC: The National Academies Press, 2010. ISBN: 0 309 15208 9. Norberg, R.P U.S. Department of Agriculture Pierce, FL. Florida Citrus Mutual, Inc. Available online at http://www.flcitrusmutual.com/files/e47fe5d8 ef81 4c15 9.pdf Odom, D. I. S., O. J. Cacho, J. A. Sinden, and G. R. Griffith. 2003. Policies for the Management of Weeds in Natural Ecosystems: The Case of Scotch Broom (Cytisus scop arius, L.) in an Australian National Park. Ecological Economics 44: 119 135. 7(5):e36026. doi:10.1371/journal.pone.0036026. a Stochastic Biological American Journal of Agricultural Economics, 84 (5), Proceedings Issue (Dec.,2002), pp. 1311 1316. Cost of Monitoring and Controlling J. Agric. Appl. Econ 38 :337 343. General Company. Available at http://www2.tbo.com/content/2008/dec/21/bz citrus greening/news money/ (February 2011). Raphael G. d'A. Vilamiu, Sonia Ternes, Guilherme A. Braga, and Francisco F. Laranjeira odel for Huanglongbin g Spread Between Citrus Plants Includin g D elay Times and Human I ntervention AIP Conf erence Proc eedings 1479, 2315 (2012); doi: 10.1063/1.4756657 Rogers, M. E., P. A. Stansly, and L. L. Stelinski. 2010. 2010 Florida Citrus Pest Management Guide: Asian Citrus Psyllid and Citrus L eafminer. Flo rida Citrus Pest Management Guide. November 2009. ENY 734. Gainesville: University of Florida/IFAS. Available online at http://edis.ifas.ufl.edu/IN686 rus Diseases: Huanglongbing Proceedings of the 13th Conference of the International Organization of Citrus Virologists eds. P. Moreno, J.V. da Graa and L.W. Timmer, pp. 279 285, November 16 23, 1995, at Fouzhou, China. Riversi de, CA: IOCV.

PAGE 98

98 International Citrus Economics Conference, Orlando, FL. Oct. 2010. us Greening FAO Plant Protection Bulletin 21:132 138. Risk Analysis, 24(4):879 892. Ecological Applications 3(8):833 845. Spann, T. M., R. A. Atwood, M. M. Dewdney, R. C. Ebel, R. Ehsani, G. England, S. H. Futch, T. Gaver, T. Hurner, C. Oswalt, M. E. Rogers, F. M. Roka, M. A. Ritenour, M. Zekri, B. J. Boman, K. Chung, M. D. Danyluk, R. Goodrich Schneider, K. T. Morgan, R. A. Morris, R. P. Muraro, P. Roberts, R. E. Rouse, A. W. Schumann, P. A. Stansly, and L. L. Stelin and Agricultural Sciences, University of Florida. Available online at http://edis.ifas.uf l.edu/hs1165 Risk Analysis 26(1):163 173.

PAGE 99

99 BIOGRAPHICAL SKETCH By the decree of God Abdul Wahab Salifu was born to a noble couple in a polygamous home in the heart of Tamale in the nort hern region of Ghana on a blessed day He started school at the Methodist primary school in Tamale and proceeded after 6 years to the Bagabaga Demonstration junior secondary school also in Tama le. He obtained his high school certificate at Ghana secondary school in Tamale in 1990. From here, h e proceeded to the Ohawu Agricultural College in the Volta region of Ghana where he obtained the general certificate of agriculture in 1993. He started his work career as an agricultural extension agent at Zabzugu district in the northern region of Ghana. In 1999 and 2003, he obtained his national diploma of agriculture and B.Sc. degree respectively at the University of Ghana. Having worked as assistant regi onal MIS officer at the Ministry of Agriculture in the northern region of Ghana from 2003 to 2004, h e started his research career as an assistant research officer at CSIR Savanna agricultural research institute in Tamale from 2004 to 2006. In the s pring of 2009 he obtained his MS degree from Tuskegee University in Alabama. He received his Ph.D. from the University of Florida in the spring of 2013.