THE HETEROPTERAN PEST COMPLEX IN FLORIDA PEACHES AND THE USE OF PYRIPROXYFEN FOR DIAPAUSE INTERFERENCE AND PARASITOID REARING
By
CORY PENCA
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2019
© 2019 Cory Penca
To my parents, Gary and Janet Penca, who taught me everything I needed to know
ACKNOWLEDGMENTS
I must start by thanking my incredible advisor, Dr. Amanda Hodges, who has made this all possible. Her belief in my potential and her support in helping me reach it has changed my life. I would also like to thank my committee members, Norm Leppla,
Peter Andersen, Ted Cottrell, Joe Eger and Trevor Smith, whose collective wisdom and guidance gave me the confidence to push forward in my research. I thank former committee member Mercy Olmstead, whose early involvement connected me with the peach industry and made this work possible. I also thank Russell Mizzell, professor emeritus, who offered his guidance and expertise on stink bug trapping and provided thousands of specimens for my experiments, this work benefited immensely from his generosity and experience.
I owe so much to the support of my fellow students in the Entomology
Department and Doctor of Plant Medicine Program. I am especially grateful for the members of Biosecurity Research and Extension Lab, especially Arjun Khadka, Morgan
Pinkerton, Sage Thompson, Arianne McCorquodale and our esteemed lab manager
Jenny Carr, for all their help, and for making the past four years fun. Being a part of the
UF Entomology and Nematology Department has been an incredible honor and I thank
Blair Siegfried for steering this magnificent ship, along with the help of the amazing people in the front office, for whom I have deep appreciation.
And, finally, I thank my incredible family and my loving girlfriend Sarah. I’ve felt your support, each and every day.
4
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS ...... 4
LIST OF TABLES ...... 8
LIST OF FIGURES ...... 9
ABSTRACT ...... 11
CHAPTER
1 INTRODUCTION ...... 13
2 THE CATFACING PEST COMPLEX IN FLORIDA PEACHES: SPECIES COMPOSITION AND REGIONAL DIFFERENCES ...... 19
Materials and Methods...... 22 Site Selection ...... 22 Visual and Trap-Based Survey Methods ...... 23 Determination of H. halys Establishment ...... 25 Data Analysis ...... 25 Results ...... 26 Pentatomid Species Composition ...... 26 Incidence of Catfacing Coreidae and Miridae ...... 27 Abundance ...... 27 Evidence of H. halys Establishment ...... 28 Discussion ...... 28
3 TRAP-BASED ECONOMIC INJURY LEVELS AND THRESHOLDS FOR STINKBUGS IN FLORIDA PEACHES ...... 39
Methods ...... 42 Site Selection and Trapping ...... 42 Damage Data ...... 44 EIL Calculations ...... 45 Economic Threshold Calculations ...... 46 Results ...... 47 Relationship between trap capture and fruit injury at harvest ...... 47 EIL Calculation ...... 50 Predicted Rate of Increase – Model Validation ...... 50 Economic Thresholds ...... 51 Discussion ...... 51
5
4 DEVELOPMENT AND THRESHOLD-INDEPENDANT EVALUATION OF SEQUENTIAL SAMPLING PLANS FOR STINK BUG INJURY IN FLORIDA PEACHES ...... 67
Methods ...... 69 Data Collection ...... 69 Estimates of Taylor’s Power Law Parameters ...... 70 Development of Sequential Sampling Plans ...... 71 Fixed Precision Sampling for Damage Estimates ...... 71 Simulation-Based Evaluation of Sampling Procedures ...... 72 Results ...... 74 Discussion ...... 76
5 PYRIPROXYFEN TREATMENT TERMINATES HALYOMORPHA HALYS REPRODUCTIVE DIAPAUSE, WITH AN INDIRECT EFFECT ON ITS EGG PARASITOID TRISSOLCUS JAPONICUS ...... 85
Materials and Methods...... 87 Specimen Sources and Rearing ...... 87 Pyriproxyfen Bioassay ...... 87 Oviposition and Oocyte Development Dose Response ...... 88 H. halys Hatchability and T. japonicus Viability ...... 89 Results ...... 90 Oviposition Response ...... 90 Oocyte Development Dose Response ...... 90 H. halys Hatchability and T. japonicus Viability ...... 91 Discussion ...... 91
6 USE OF PYRIPROXIFEN INTO INCREASE PARASITOID YIELD IN A BIOLOGICAL CONTROL REARING PROGRAM ...... 99
Materials and Methods...... 102 Specimen Collection ...... 102 Treatments and Rearing Conditions ...... 102 Mortality and Egg Data Collection ...... 103 Parasitoid Exposure ...... 104 Data Analysis ...... 105 Results ...... 106 Mortality ...... 106 Egg Production ...... 107 Host and Parasitoid Viability ...... 108 Parasitoid Yield ...... 108 Discussion ...... 108
7 CONCLUSIONS ...... 118
LIST OF REFERENCES ...... 122
6
BIOGRAPHICAL SKETCH ...... 136
7
LIST OF TABLES
Table Page
2-1 Size, trap number, visual survey count, and region name for each sample site used during the course of the survey...... 34
2-2 Pentatomid species incidence by sampling method for each region...... 35
2-3 Observed species richness, estimates of species richness, effective number of species (ENS) and sampling completeness ...... 36
2-4 Individuals per trap for the 5 major species during 2017 and 2018, divided by region...... 37
2-5 Record of H. halys detection, stage and reproductive status during three years of surveillance in Florida peach orchards...... 38
3-1 Estimates of the rate of change, by period, as determined by linear regression with the previous period trap capture as the predictor variable...... 64
3-2 Review of treatment efficacy studies for reduction of catfacing injury in peaches...... 65
4-1 Sample site location, orchard block size, sampling units and sample dates. A sampling unit consisted of 30 fruit per tree, visually inspected for injury...... 80
4-2 Coefficients of binomial-logistic regression of the error rates resulting from simulated sampling...... 83
5-1 H. halys hatchability and T. japonicus viability on pyriproxyfen-induced eggs (Mean ± SE). For H. halys % hatchability ...... 97
6-1 Pairwise comparisons of mortality using the Log-Rank test. Experiment 1...... 113
6-2 Pairwise comparisons of mortality using the Log-Rank test. Experiment 2...... 114
6-3 Estimated M. cribaria hatch probability, as determined by binomial regression. Results are from eggs produced from experiment 2. Rows with the same letters are not significantly different at p<0.05...... 115
6-4 Estimated P. saccharalis hatch probability, as determined by negative binomial regression. Results are from eggs produced from experiment 2. Rows with the same letters are not significantly different at p<0.05...... 116
6-5 Estimated parasitoid yield, as determined by generalized linear regression. Rows without common letters are significantly different at p<0.05 (Least Squares Means)...... 116
8
LIST OF FIGURES
Figure Page
2-1 Peach orchard locations for sampling during 2017 and 2018 peach seasons. ... 33
2-2 Yellow pyramid / Tedder’s trap. The pyramid base stands approximately 1.2 meters in height. Photograph courtesy of the author...... 33
2-3 Arrangement of pyramid traps (yellow triangles) and visual sample trees (red boxes)...... 34
2-4 Coverage based species accumulation curves for each region sampled. Shaded area denotes 95% confidence intervals...... 36
3-1 Population trends for adult E. servus during the 2017 season. Vertical bars represent the trapping date nearest to the commencement of fruit harvest...... 57
3-2 Population trends for adult E. servus during the 2018 season. Vertical bars represent the trapping date nearest to the commencement of fruit harvest...... 58
3-3 Description of the linear relationship between stink bug trap capture and percent injury over the entire injury interval (top) and restricted to the range of observed values (bottom)...... 59
3-4 EIL estimates based on a logistic relationship between E. servus population density and probability of fruit injury...... 60
3-5 EIL estimates based on a linear relationship between E. servus population density and probability of fruit injury...... 61
3-6 Predicted and observed E. servus trap capture, by trapping period for the 2017 season...... 62
3-7 Predicted and observed E. servus trap capture, by trapping period for the 2018 season...... 63
3-8 Evaluation of an economic thresholds based on 2017 (red) and 2018 (blue) trap capture at an EIL of 5.53 bugs per trap, derived from the linear injury relationship ...... 64
4-1 Sequential sampling plans for assessing density of stink bug injury to Florida peaches. Threshold of 2%, 4%, and 6% correspond with 0.6,1.2 and 1.8 injured fruit per sample, respectively...... 81
4-2 Optimum sample size for estimation of fruit injury density at a fixed level of precision...... 81
9
4-3 Approximation of the probability of erroneous classification using either a sequential or simple random sample (SRS) method...... 82
4-4 Average sample number by distance from threshold based on simulations of 27 empirical data sets. Sampling began with three samples and the maximum sample number was constrained to 40 samples...... 83
4-5 Loess regression approximation of the average sample number as a function of distance to threshold...... 84
5-1 Mean ± SE of cumulative eggs produced per female. “cd” = control dark, reared under 10L:14D. “cl” = control light, reared under 16L:8D...... 95
5-2 Percent diapause termination as a response to pyriproxyfen treatment in female H. halys...... 96
5-3 Dissected H. halys ovaries representing various stages of oocyte formation. .... 97
5-4 Trissolcus japonicus pupa. When reared on pyriproxyfen-induced H. halys eggs, T. japonicus ceased development at the pupal stage, with no pupal to adult emergence observed...... 98
6-1 Survival probability for experiment 1 (red) and experiment 2 (blue) with all treatments combined...... 111
6-2 Survivorship curves of M. cribraria from experiment 1, separated by photoperiod, for the three chemical treatments (Control = Acetone, High = 1.0% Pyriproxyfen, Low=0.1% Pyriproxyfen)...... 112
6-3 Survivorship curves of M. cribraria from experiment 2, separated by photoperiod, for the three chemical treatments (Control = Acetone, High = 1.0% Pyriproxyfen , Low=0.1% Pyriproxyfen)...... 113
6-4 Egg production for experiment 1, separated by photoperiod and pyriproxyfen treatment...... 114
6-5 Egg production for experiment 2, separated by photoperiod and pyriproxyfen treatment...... 115
6-6 Display of density estimates produced from simulated (n=2000) estimates of parasitoid yield. The area under all curves sum to 1...... 117
10
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
THE HETEROPTERAN PEST COMPLEX IN FLORIDA PEACHES AND THE USE OF PYRIPROXYFEN FOR DIAPAUSE INTERFERENCE AND PARASITOID REARING
By
Cory Penca
May 2019
Chair: Amanda Hodges Major: Entomology and Nematology
The catfacing pest complex can cause significant economic losses to peach producers in the southeastern United States. While this pest complex has been well- studied in the southeastern United States, there is currently no Florida-specific information appropriate for the nascent subtropical peach industry in the southern half of the state. The goal of this work was to characterize the catfacing pest complex in
Florida peaches in order to provide data to be used in the development of integrated pest management practices that accurately reflect the pest situation in Florida.
We began by identifying the key species responsible for cat-facing injury in
Florida. This was achieved through a multi-year survey at several orchards representing the primary peach growing areas in the state. Surveys employed both trap-based and visual methods, and provided insights into the species composition, relative abundances, population dynamics, and regional differences.
A survey of catfacing injury was conducted at multiple time points throughout the season to identify the spatiotemporal character of catfacing injury in Florida peach orchards. The relationship between mean and variance was used to develop sequential
11
sampling plans based on hypothetical thresholds. The correlation between pentatomid populations and peach injury was used to estimate the expected injury and calculate the economic injury level for catfacing pests in Florida.
Lastly, as reproductive diapause is a critical component of Heteropteran biology and thus is highly relevant to the catfacing pest complex, we evaluated the potential of the juvenile hormone pyriproxyfen to interfere with diapause in two invasive
Heteropteran species. We then provided eggs from pyriproxyfen treated pests to their respective hymenopteran egg parasitoids and evaluated parasitoid viability. Results from this component of the dissertation provide insight into a novel method of control through diapause interference as well as a technique for increasing egg parasitoid production in a biological control rearing program.
12
CHAPTER 1 INTRODUCTION
The peach, Prunus persica (L.) is an important tree-fruit crop in the United
States, with utilized production valued at $605 million (NASS 2016). California is the largest peach-producing state in the nation, with total production in 2015 of 607,600 tons, followed by South Carolina (68,900 tons) and Georgia. (40,600 tons) (NASS
2016). Florida’s peach acreage was as high as 4,000 acres in the 1960’s and 1970’s, largely in the northern half of the state, but by 2006 was as low as 500 acres due to a series of hard freezes in the 1980’s and poor market conditions (Ferguson et al. 2006).
Current estimates for peach production in Florida range from 1500 to 2000 tons on
1,231 acres (Morgan and Olmstead 2013a).
Historically, the ability to produce peaches early in the season was limited by physiological chilling requirements. Peach trees planted in warmer areas did not receive enough “chilling units” (defined as the cumulative hours of temperatures below 7.2 °C), during the season for a healthy dormant period to occur, leading to reduced foliation, flowering and fruiting (Sharpe 1969). In 1952 at the University of Florida, Gainesville,
FL, a breeding program was initiated to develop peach varieties suitable for production in subtropical regions, with chilling-unit requirements in the 100-250 hour range
(Sherman et al. 1996). Trees with low chilling requirements planted in southern Florida flower and set fruit with a reduced risk of frost damage, and are productive early enough to reach the profitable market-window for early season peaches (Olmstead et al. 2013).
The economic viability of early-season peach production is contingent upon reaching the market before the larger production areas of Georgia and California enter the market. As such, the harvest of peaches in Florida begins in April and ends in May, with
13
budbreak and fruit development beginning in January (Olmstead et al. 2013). The primary production areas for subtropical peaches in Florida are Polk, St. Lucie and
Indian River Counties.
Because peach production of any significant scale in central to southern Florida is relatively new, documentation of seasonal pest pressures has not occurred. Of specific concern to peach production is the occurrence of heteropteran pests, which cause a form of injury, referred to as “cat-facing”, where the area of insect feeding develops abnormally (Rings 1957). Incidence of cat-facing injury varies, ranging from 5-
10% in managed peach orchards in North Carolina (Killian and Meyer 1984), 18-36% in managed orchards in Ohio (Rings 1957) and as high as 80% in mid-Atlantic orchards with outbreak populations of the brown marmorated stink bug Halyomorpha halys Stål
(Leskey, Short, et al. 2012).
In the Southeast the most significant cat-facing heteropteran pests include members of the family Pentatomidae, including Euschistus servus (Say), Chinavia hilaris (Say) and Nezara viridula (L.), members of the family Miridae, namely Lygus lineolaris (Palisot de Beauvois), and members of the family Coreidae, namely
Leptoglossus phyllopus (L.) (Foshee et al. 2008). In addition, since 2010 the invasive pentatomid, Halyomorpha halys has emerged as the most serious pentatomid pests of tree-fruit in the mid-Atlantic region and has expanded its range as far south as Alabama and Georgia, with detections frequently occurring in Florida (Leskey, Hamilton, et al.
2012, Medal et al. 2013)
As a generally polyphagous pest group capable of dispersing over long distances in search of feeding sites and having the ability to reproduce away from the crop, cat-
14
facing insects have proven to be a particularly difficult pest to control (McPherson and
McPherson 2000). A single instance of feeding or probing on a developing peach fruit may result in unmarketable fruit; as such the economic impact of even low numbers of stink bugs within an orchard can be significant.
By far the most common method for management of cat-facing pests in peaches is the use of pesticides, applied by an air blast sprayer, with the first application occurring at petal fall, and with additional applications made every 7-10 days (Horton et al. 2016). A lack of knowledge regarding cat-facing pest ecology in Florida peach orchards and a lack of a reliable sampling methodology has led to a situation where the current regime of pesticide applications may not be economically warranted. Monitoring stink bug populations allows growers to make better informed decisions as to when to apply pesticides. Various sampling methods have been used to monitor stink bug populations, including sweep nets and beat sheets on field crops such as soybean, cotton and rice and utilizing a limb-jarring method in tree-fruit orchards (Pedigo and
Buntin 1993, Reay-Jones et al. 2009). Monitoring via trapping has also been attempted, with several different trap designs proposed, including a cone trap, sticky paper and pyramid traps (Leskey and Hogmire 2005). The identification of male-produced aggregation pheromones and subsequent use as a lure added to traps has significantly improved trap effectiveness (Aldrich et al. 1991, Mizell and Tedders 1995, Cottrell and
Horton 2011). Establishing a reliable relationship between the numbers of stink bugs captured in traps and the level of stink bug damage in the field is a necessary component if trap counts are to be used for determining management actions based on economic injury levels (Cullen and Zalom 2000).
15
In an effort to improve the effectiveness of pesticides and to reduce the amount of material sprayed on crops, the use of trap crops and a “trap and kill” approach has been proposed, in which a plant that is attractive to the pest species is planted near the crop plant, reducing feeding pressure on the crop, and allowing for a focused pesticide application in a trap and kill scenario (Bundy and McPherson 2000, Mizell et al. 2008).
The discovery and commercial production of stink bug pheromone lures also lends itself to this approach, as specific trees within the orchard can be baited with pheromone lures, concentrating the population in the orchard to a limited area. Pesticide applications can then be restricted to the vicinity of the baited tree, with the goal of providing equivalent or improved control with a significantly reduced pesticide application (Morrison, Lee, et al. 2016).
Cultural practices can also have a significant effect on controlling populations of cat-facing insects in peach orchards. Management of broadleaf weeds has been shown to reduce the abundance of stink bugs and tarnished plant bugs, leading to a reduction in overall fruit injury (Killian and Meyer 1984). Cat-facing insects do not reproduce within the peach orchard to a significant degree, thus removing non-crop reproductive hosts from the orchard environment can prevent the buildup of populations of cat-facing insects (Woodside 1947). In southern Florida, where peach is produced commercially as a single, early season crop, the buildup of pest populations on weeds after harvest will only lead to economic injury if significant numbers overwinter at the site and persist into the next growing season. Destruction of overwintering sites has been proposed for managing stink bug populations (McPherson and McPherson 2000); however, in subtropical peach production the practice of destroying overwintering sites may fail to
16
be effective in reducing stink bug populations if insufficiently cold weather eliminates the need for overwintering sites.
An additional method for controlling cat-facing insects is the use of natural enemies, either by reliance on the natural enemies already present in the orchard, by the augmentation of natural enemy populations, or by the release of imported biological control agents. Classical biological control has been attempted on several stink bug species, primarily focusing on the use of hymenopteran egg parasitoids from the family
Platygastridae, a classical example of which is the parasitoid, Trissolcus basalis
(Wollaston), used for control of Nezara viridula in California and Hawaii (Mills 2009).
Classical biological control of Halyomorpha halys by the hymenopteran parasitoid
Trissolcus japonicus Ashmead is currently under investigation (Yang et al. 2009).
Recently T. japonicus has been detected in the landscape at several geographically distinct locations including Maryland and the Pacific Northwest (Talamas et al. 2015,
Milnes et al. 2016). Egg predation by a number of arthropod species, including members of the families Tettigoniidae, Carabidae and Gryllidae may also contribute significantly to egg mortality, reducing stink bug populations (Morrison, Mathews, et al.
2016). Nymphal and adult stink bugs are preferentially targeted by parasitoids, including tachinids (Diptera: Tachinidae) (McPherson et al. 1982), and sand wasps of the genus
Astata (Hymenoptera: Crabronidae) (Evans 1957). Generalist predators including the red imported fire ant Solenopsis invicta Buren (Krispyn and Todd 1982, Stam et al.
1987) and several species of spiders have also been observed feeding on stink bugs in agricultural systems (Stam et al. 1987). Cultural practices that support predator and
17
parasitoid populations are a crucial component of an integrated pest management program for the control of cat-facing insects (Orr 2009).
The proposed research aims to investigate the stink bug pest complex and potential impacts on subtropical peach production in Florida. The approach will be multifaceted, centering on a trapping regime in several peach orchards during the 2015-
2017 growing seasons. Trap data and field observations of stink bug damage will provide an understanding of the seasonality, abundance and impact of pentatomids in
Florida’s peach orchards. In addition to trapping, experiments will be conducted to better understand the manifestation of stink bug damage on the varieties of peach commonly grown in Florida. These experiments will be used to accurately identify heteropteran feeding damage in a peach orchard, with the frequency and location of fruit injury being recorded over time so that it can be mapped to describe any patterns in the distribution of feeding activity within the orchard. The total economic impact of stink bugs and other cat-facing insects under the currently employed management strategies will be determined via a statewide, multi-site estimate of fruit injury prior to harvest. Laboratory studies on the effects of insect growth regulators (IGRs) on stink bug diapause and egg production, and the ability of hymenopteran egg parasitoids to utilize IGR-induced eggs will be conducted as a peripheral component of this work, as it will add to our understanding of pentatomid diapause, of parasitoid/pentatomid interactions, and outline the potential of a novel avenue of control for pentatomid pests.
18
CHAPTER 2 THE CATFACING PEST COMPLEX IN FLORIDA PEACHES: SPECIES COMPOSITION AND REGIONAL DIFFERENCES*1
The catfacing pest complex in peaches consists of a diversity of insect pests which feed directly on the fruit, usually through a piercing-sucking mode of feeding characteristic of phytophagous heteroptera (Woodside 1950, Chandler 1955, Foshee et al. 2008). Developing fruit do not expand evenly at the site of feeding injury, leading to a pinched appearance referred to as “catfacing”. Catfacing injury produces unmarketable fruit and can lead to considerable economic losses for peach growers.
As commercial peach production in the southern half of Florida is a relatively new industry, basic information on the catfacing pest complex in Florida peaches has not been developed. The current approach to management of catfacing pests in Florida is to treat catfacing pests as one would in other parts of the Southeast, with broad spectrum sprays targeting Euschistus servus (Say) being made early in the season followed by regular applications up to harvest as permitted by the pre-harvest interval of the materials being applied (Blaauw et al. 2018). While management practices for catfacing pests in the Southeast can achieve acceptable results, there are several reasons why managing catfacing pests in Florida peaches may require a different approach: 1) In the southeastern USA the climate changes from subtropical to tropical as one moves southward into the southern half of peninsular Florida, and changes in species composition and relative abundances may occur along this climatic/latitudinal gradient. 2) The early season nature of the Florida peach industry also raises the question of which species may be most likely to coincide temporally with the crop when
*Portions of this chapter reprinted with the permission of the Florida Entomologist
19
it is at its most vulnerable stage, and at what abundance these species occur. If catfacing pests are able to persist and remain active through the winter at high population levels it is possible that early season peaches in Florida may be exposed to pest attack immediately; alternatively, should stinkbug populations suffer significant decline even in the Florida’s mild winter, there is the possibility that the first part of the growing season may escape insect attack as populations rebound and overwintering individuals resume feeding activity. 3) A third reason for potential differences in the composition of the catfacing pest complex in Florida peaches is the potential for adventive species. An estimated 7.6 percent of insect species in Florida are thought to be adventive-immigrant species, significantly higher than the 1.7 percent estimate given to the rest of the United States (excluding Hawaii) (Frank and McCoy 1995). Invasive
Heteroptera have risen in prominence in recent years, due to the recent establishment of Halyomorpha halys (Stål), Bagrada hilaris (Burmeister) and Megacopta cribraria
(Fabricius) in the US.
The catfacing pest complex is well studied in traditional peach growing areas of the southeast. The major species which comprise this pest complex include the pentatomids from the genus Euschistus, especially E. servus (Say), as well as Nezara viridula (L.), Chinavia hilaris (Say) and Thyanta custator (Fabricius) (Chandler 1955,
Johnson et al. 2002, Foshee et al. 2008, Cottrell and Horton 2011). The coreid
Leptoglossus phyllopus (L.) and the mirid Lygus lineolaris (Palisot de Beauvois) have also been associated with catfacing injury in the Southeast (Killian and Meyer 1984,
Foshee et al. 2008). Halyomorpha halys (Stål) is a recent addition to the southeastern peach catfacing pest complex, having become established in region sometime between
20
2010 to 2015. Though H. halys is among the major catfacing pests in the mid-Atlantic and other parts of its invasive range, it has not yet become a significant part of the catfacing pest complex in the southeastern US.
Answering the question of whether the invasive species H. halys is a member of the catfacing pest complex in Florida peaches is a crucial part of this work.
Halyomorpha halys has been observed in peaches in Georgia but had not been reported to have established in Florida prior to the commencement of this study (see
Penca & Hodges (2018)). In Florida, H. halys is frequently intercepted on goods moving from northern states where H. halys is widespread. Because of the propensity of this species to “hitchhike” into Florida, and the possibility that large numbers of H. halys are entering the state and dispersing in the environment, it is necessary to determine if detections of adult H. halys represent established populations or if detections are of recently introduced individuals or temporary populations that experience rapid extirpation without reproduction. Evidence that would lend support to the possibility of an established population include: 1) repeated trap capture across multiple years; 2) observations of nymphal stages, and; 3) observations of gravid females. The visual and trap surveys deployed during this study provide an opportunity to observe occurrence/non-occurrence of (1) and (2) which would support/undermine a case for H. halys establishment. Evidence of (3) would indicate the biological capacity for reproduction in Florida during the survey period. This information is particularly relevant for H. halys as life history studies suggest a long day photoperiod is necessary for diapause termination and reproduction in this species(Watanabe 1979). A conservative estimate of a photoperiod greater than 12.7 hours was estimated by Nielsen et al.
21
(2017). In the northernmost study area this would equate to mid-April as being the earliest date at which H. halys could begin vitellogenesis. Most orchards in South-
Central Florida will be nearing the end of harvest around this time. If assumptions on H. halys reproductive biology hold true for individuals in Florida’s peach orchard, it is possible that Florida peaches may be at reduced risk of H. halys pest pressure due to photoperiod driven limitations. Understanding if H. halys is truly absent from Florida, or simply unreported, and its capacity to reproduce during the peach season will be crucial as peaches are a preferred host of H. halys.
The primary aims of this chapter are to characterize the species composition of the catfacing pest complex in Florida peaches, identify any key differences in species composition between regions within Florida, and determine if the invasive H. halys is present in Florida. The similarities/differences between the Florida catfacing pest complex and other peach growing areas in the Southeast, along with implications for management, will be discussed.
Materials and Methods
Site Selection
Early season peaches in Florida are grown commercially in the southern half of the state, roughly overlapping with traditional citrus growing areas. Survey sites were selected in Lake, Polk and St. Lucie Counties to provide a geographic representation of
Florida peach production (Figure 2-1). A total of 5 sites were surveyed during 2016-
2018, with 2 sites in Lake County, 1 site in Polk County and 2 sites in St. Lucie County.
Sites in the same county were combined for regional analysis.
22
Visual and Trap-Based Survey Methods
Visual surveys were conducted at four samples dates, approximately 14 days apart, beginning at the time of fruit thinning (ca. February) and ending at harvest (ca.
April). Orchards were subsampled as blocks ranging in size from 1.17 to 1.63 hectares
(Table 2-1). Peach trees were sampled in a grid-like fashion, with sampling effort standardized by limiting search effort to a visual observation of 30 fruits per tree.
Phytophagous Heteroptera observed on any part of the tree during the 30-fruit inspection were recorded and identified to the species level.
Trap based surveys utilized a yellow pyramid trap (Figure 2-2). The trap base was constructed of tempered hardboard (trade name Eucaboard®, Eucatex, Sao Paulo,
Brazil) painted with industrial-safety-yellow exterior oil-based paint (Rust-Oleum, Vernon
Hills, Ill). The trap base consisted of two interlocking trapezoids, with a base width of
55.9 cm, upper width of 5.1 cm and overall height of 121.9 cm. The complementary pieces of the base were interlocked via a 63.5 cm slit made at the center-bottom of the base upwards, and from the center-top downwards, of the complementary piece. The trap top consisted of a 1-gallon plastic jar (ULine, Pleasant Prairie, WI) with ventilation holes 4 cm in width cut into the sides and top of the jar and covered with aluminum insect screening. The entrance cone (constructed from aluminum pet screening) was inserted into the removed bottom of the jar. The opening of the entrance cone into the capture jar was 1.6-cm in diameter. Trap specifications were intended to follow the design of Hogmire and Leskey (2006).
A two-component lure containing the Plautia stali Scott pheromone (Methyl
(E,E,Z)-2,4,6-decatrienoate) and H. halys pheromone (a mixture of (3S,6S,7R,10S)-
10,11-epoxy-1-bisabolen-3-ol and (3R,6S,7R,10S)-10,11-epoxy-1-bisabolen-3-ol)
23
(ChemTica International, SA, Heredia, Costa Rica) was placed in the trap capture jar and changed at every trap servicing. This combination is known to be attractive to several species, with reports in the literature indicating attraction to multiple species of
Euschistus and Thyanta, as well as Nezara viridula, Chinavia hilaris, Murgantia histrionica and H. halys (Morrison, Lee, et al. 2016, Weber et al. 2017).
Traps were deployed along 3 sides of each block, with interior traps placed within the block and in-line with the exterior traps. Traps were deployed in early January and serviced every 13-14 days until the first week of June. A schematic of the trapping and visual sampling plan is provided in figure 2-3.
The visual cue of the yellow trap provides moderate attraction and is capable of trapping species not attracted to the pheromone lure; however, the trap design and the use of a pheromone lure introduces bias into the estimates of relative species abundance. The addition of a visual survey provides a way to address the trap-bias, as visual surveys offer a relatively unbiased estimate of the relative abundance of species which feed on peaches during sampling hours. A combination of visual and trap-based methods provides a better approximation of the relevant species present in Florida peach orchards than either method alone. Sampling using these methods is not expected to provide a complete picture of what species are present, and at what relative abundances; however, we believe the most abundant, and thus most impactful, catfacing Heteroptera will be well represented by the methods used. While rare species may be missed, their scarcity reduces their importance as members of the catfacing pest complex.
24
Determination of H. halys Establishment
The previously mentioned trap and visual surveys were used to provide evidence of presence/absence of H. halys in Florida peach orchards. The capacity of H. halys to reproduce in Florida peach orchards was determined by direct observation of nymphs as well as by dissection of adult females to determine reproductive status. Female collected from traps were dissected and rated based on the condition of their ovaries following the methods of Penca & Hodges (2017). Specimens with regressed ovaries, in which a single oocyte is present in each ovariole, and where the widest point of the ovariole is the nutritive tip, were rated as being regressed, indicative of diapause or the previtellogenic period. Specimens where the oocyte has expanded such that the widest point of the ovariole is below the nutritive tip, were rated as being post-diapause, as vitellogenesis is evident (Davey 1997). Specimens with mature, chorionated eggs, were rated as gravid.
Data Analysis
Composition of catfacing insects was evaluated on a regional level using data from trap capture and visual observations. Species richness was calculated using a sampling-unit based estimator of species richness. An individual sampling unit consisted of the recorded trap capture for each trap after 2 weeks of deployment in the field. Estimates were generated using the R function iNEXT from the package iNEXT
(Hsieh et al. 2016). This method allows for an estimation of unseen species, providing an estimation of species richness at levels of sampling beyond what occurred in the study period, while also giving an estimate of sampling completeness (Chao 1984,
Smith and van Belle 1984, Palmer 1990). The effective number of species (ENS) was determined for each site using the transformed Shannon’s diversity index (Jost 2006,
25
Chao et al. 2014). Differences in beta diversity between regions were compared via the homogeneity dispersion test (betadisper function in the R package Vegan (Oksanen et al. 2013), with dissimilarity calculated using Sorenson’s dissimilarity index. The significance of the predictor variable “region” on dissimilarity (taken as the distance to centroid) was test via ANOVA, followed by Tukey’s HSD for post-hoc analysis of contrasts.
Abundance for key species was estimated for each site and region by fitting observed counts from trap data to a negative binomial model. This approach allows for mean-comparison when violations of normality and heteroscedasticity are violated, as is common in count data (Ver Hoef and Boveng 2007). Contrasts were evaluated by comparing least-squares means with Tukey’s adjustment for multiple comparisons.
Regional differences in abundances were compared for the five most abundant pentatomids, with separate analysis preformed for 2017 and 2018.
Results
Pentatomid Species Composition
A total of 3,271 adult pentatomids, representing 16 species, were collected from traps in 2017 and 2018. The number of species detected by trap and visual methods differed between observation methods and regions (Table 2-2). Visual observation was only able to detect three pentatomid species, Euschistus servus, E. quadrator (Rolston) and E. obscuras (Palisot), during the observation period. Additionally, one region, Polk
Co., had no visual observations of pentatomids during both survey years. In St. Lucie
Co. the only species detected by visual observations was E. servus, which was also the most frequently trapped species at all sites.
26
Coverage based estimates of pentatomid species richness at a regional level were determined to be similar, with overlapping 95% confidence intervals at the extrapolated range (Figure 2-4). The accumulation of species was markedly more rapid at the Lake Co. organic sites when compared to both conventional sites. The values for estimated richness, as well as the effective number of species (transformed Shannon diversity index) and sampling completeness for each region are provided in table 2-3.
While the estimated species richness was highest for St. Lucie Co., with 28.94 species predicted, the estimated effective number of species for St. Lucie was the lowest of the three regions (3.145 ± 0.300), while the organic Lake Co. site had the highest estimated effective number of species (5.844 ± 0.382). Beta diversity was significantly different between regions (df =2,66, F =4.48, p = 0.015), with significant differences occurring between the Lake Co. Region and the St. Lucie County region, (p=0.0107, Tukey’s
HSD), while Polk Co. was not significantly different from either Lake Co. (p=0.2840) or
St. Lucie County (p=0.3398).
Incidence of Catfacing Coreidae and Miridae
Visual surveys suggest that Leptoglossus phyllopus is the second most encountered species; however, this species was rarely encountered in the trap-based survey. The tarnished plant bug, Lygus lineolaris did not occur in a single instance from trap data, nor was it seen on peach fruit during visual observations.
Abundance
A significant difference in the abundance of major species between regions, as well as differences between species within regions, was apparent (Table 2-4).
Euschistus servus was the most abundant species in all regions during both survey years; however, significantly fewer E. servus were captured in traps in Polk Co than in
27
either St. Lucie or Lake Co. Thyanta perditor was the second most frequently encountered pentatomid species at all sites, with significantly higher abundance in the
St. Lucie region when compared to both the Polk Co. and Lake Co. regions (Table 2-4).
Euschistus ictericus (L.), Nezara viridula and Chinavia hilaris abundance did not appear to be significant during the study period, as evidenced by their low trap capture and lack of visual observations on fruit.
Evidence of H. halys Establishment
The survey detected adult H. halys in all 3 yr (2016–2018) (Table 2-5). The discovery of adult H. halys in peach orchards represents the first detection of this species from an agricultural setting in Florida. Of the 11 female H. halys dissected, 6 were gravid, 4 had clear oocyte development indicative of a post-diapause status, and 1 had regressed ovaries indicating diapause or a pre-vitellogenic state (Table 2-5). Gravid females were detected as early as mid-February, and continued to be observed into
June, suggesting a large window for potential reproduction in Florida. In 2018, the detection of 3 nymphs and 2 egg masses at the Lake Co. 2 field site represented the first observation of reproduction in Florida. These nymphal specimens and egg masses were detected by visual survey within the vicinity of a pheromone baited yellow pyramid trap.
Discussion
Euschistus servus appears to be the most important cat-facing pest in Florida peaches. This result does not differ markedly from what has been observed in traditional peach growing regions of the Southeast, such as Georgia and South Carolina
(Woodside 1950, Leskey and Hogmire 2005, Cottrell and Horton 2011). Trap bias favoring E. servus and other Euschistus species may be responsible for some of the
28
difference in abundance between these and other pentatomid pests observed.
However, visual observations of pest presence supported the inference that these were among the most common pests in the orchard system. The contribution of Leptoglossus phyllopus to the catfacing pest complex in Florida peaches is less-clear. Trap results indicate low abundances of L. phyllopus at all sites; however, we did not expect efficient capture of L. phyllopus with trap and pheromone combination used in this study.
Several components of the trap design may influence the observed capture of L. phyllopus, including the attractiveness of the yellow-paneling and the size of the collection jar cone opening, which may repel the larger bodied and longer-limbed L. phyllopus. Visual observations place L phyllopus as the second most encountered cat- facing Heteropteran in Florida, with abundance near those observed for E. servus.
Determining the true impact of L phyllopus on fruit injury will require the development of species-specific monitoring methods. Lastly, Lygus lineolaris has been reported as an important catfacing pest in other peach growing regions of the Southeast, and is established in all surveyed areas, yet it was not part of the incidental trap capture, nor was it seen feeding on peaches during the visual portion of the survey.
Widely used metrics of diversity, including the effective species number (a transformation of Shannon’s index) and the Sorenson dissimilarity index, provide insight into regional differences in the composition of the pentatomid pest complex. The Lake
Co. and the St. Lucie Co. regions showed the greatest differences in both effective species number and Sorenson dissimilarity. Multiple factors may influence the observed differences. Both Lake County sites were certified organic and had very few pesticide applications, whereas both St. Lucie sites were larger conventional operations with
29
frequent applications of broad-spectrum insecticides, including zeta-cypermethrin and phosmet. Additionally, these sites were separated by the greatest distance, with approximately 176 km between the nearest sites. When only North-South distance was measured, the Polk Co. was approximately 59 km due south of the southernmost Lake
Co. site and 62 km due north of the northernmost St. Lucie Co. site, placing roughly in the latitudinal middle of the other two regions. The implications of these observations for pest management is complex; it appears a more diverse pentatomid species complex is present in the Lake Co. sites, whether this is a result of broad regional differences or differences in site specific management practices is an area in need of future research.
The well-studied ecological trend of the latitudinal diversity gradient provides an expectation of increased diversity at lower latitudes when compared to temperate or subtropical latitudes, though the strength of this trend is diminished in smaller organisms at lower trophic levels (Hillebrand 2004). At the local level (comparisons between regions within Florida), we were not able to detect a strong signal of increased diversity in the cat-facing pest complex, possibly because the effects of climate and other latitudinal differences were swamped by differences in site characteristics and pest management strategies. At the regional level we did not observe several catfacing
Heteropterans in Florida that are present in other parts of the Southeast, namely Lygus lineolaris and Euschistus tristigmus (Say). These species are known to occur in the study areas; however, their absence from trap capture or visual surveys indicates low prevalence. Thyanta perditor (Fabricius) was one of the most prevalent stink bug species during the study period and is not known to occur in peach growing areas of Georgia and South Carolina (Panizzi and Herzog 1984, Rider and Chapin 1992), thus it
30
represents the primary difference in species composition between Florida and the rest of the Southeast.
Based on our results, management of catfacing pests in Florida peaches should focus on control of E. servus and L. phyllopus, with secondary attention paid to T. perditor, and, to a lesser extent, N. viridula, and E. ictericus. The importance of Thyanta perditor in the St. Lucie Co. region will be dependent on the propensity of this species to feed on peaches. The literature on this species’ feeding habit does not indicate peach as a common host plant; however, T. perditor is not widespread in peach producing areas and has been reported on a diverse range of host plants (Rider and Chapin 1992,
Laumann et al. 2011). One of its primary host plants, Bidens alba (Asteraceae) (L.) is widespread in Florida and was most abundant in the St. Lucie Co. peach orchards, though it was observed at all study sites (Panizzi and Herzog 1984). Bidens alba is also noted as a host of E. servus and was observed flowering early in the season. The influence of this weed on pest dynamics in peach orchards is an area deserving of further study.
One of the most significant findings from this work is the repeated detection of H. halys In Florida peach orchards. Evidence of reproduction at the Lake Co. sites suggests that this species may be established in Florida and may become a part of the catfacing pest complex in Florida peaches. Halyomorpha halys is a major pest in parts of the mid-Atlantic, causing significant economic losses to tree fruit growers in areas where it has established (Leskey, Short, et al. 2012). Within the state of Florida, the
Florida Department of Agriculture and Consumer Services, Division of Plant Industry is the plant protection agency with the regulatory authority to determine pest status. The
31
Florida Department of Agriculture and Consumer Services, Division of Plant Industry has declared previous reports of adult H. halys as interceptions per the International
Plant Protection Convention glossary of phytosanitary terms (FAO 2006). The repeated detection of adult H. halys in 2016, 2017, and 2018, and the detection of reproductive stages in 2018 suggests H. halys is likely established in the vicinity of the Lake County
2 site. As such, the Lake County population of H. halys has been declared a pest incursion of limited distribution by the Florida Department of Agriculture and Consumer
Services, Division of Plant Industry.
The impact of H. halys may be mitigated by the early season nature of Florida peaches, which could allow for temporal escape from peak H. halys abundance. The status of H. halys in Florida peaches, and other at-risk crops, should be continuously monitored going forward. If establishment of H. halys is widespread and economic injury is occurring, the release of the classical biological control agent Trissolcus japonicus may be justified, pending regulatory approval.
32
Figure 2-1. Peach orchard locations for sampling during 2017 and 2018 peach seasons.
Figure 2-2. Yellow pyramid / Tedder’s trap. The pyramid base stands approximately 1.2 meters in height. Photograph courtesy of the author.
33
Figure 2-3. Arrangement of pyramid traps (yellow triangles) and visual sample trees (red boxes).
Table 2-1. Size, trap number, visual survey count, and region name for each sample site used during the course of the survey. Site Polk 3 was eliminated after 2016 season. Block size # Traps # Visual surveys Site Region (hectares) (2016/2017/2018) (2017/2018) Lake 1 1.17 5 / 7 / 7 4 / 4 Lake co. Lake 2 0.34 3 / 3 / 3 - Lake co. Polk 1 1.32 - / 5 / 5 4 / 4 Polk co. Polk 2 1.42 - / 5 / 5 - Polk co. Polk 3 1.27 5 / x / x / - Polk co. St. Lucie 1 1.63 5 / 7 / 7 4 / 4 St. Lucie co. St. Lucie 2 1.44 5 / 7 / 7 - St. Lucie co.
34
Table 2-2. Pentatomid species incidence by sampling method for each region. Lake Polk St. Lucie All
Species Trap Visual Trap Visual Trap Visual Trap Visual
Banasa euchlora Stål + +
Chinavia hilaris + + +
Euschistus ictericus + + + +
E. obscuras + + + + +
E. quadrator + + + + + +
E. servus + + + + + + +
Halyomorpha halys + + + +
Murgantia histrionica (Hahn) + + + +
Menecles insertus (Say) + +
Mormidea lugens (Fabricius) + +
Nezara viridula + + + +
Oebalus pugnax (Fabricius) + + + +
Proxys punctulatus (Palisot) + + +
Piezodorus guildinii + + + (Westwood)
Thyanta custator + + + +
T. perditor + + + +
Total 15 3 12 0 14 1 16 3
The (+) symbol denotes the observation of the species using the denoted sampling method. Observed species richness is presented as the sum of the species detected for each sampling method.
35
Figure 2-4. Coverage based species accumulation curves for each region sampled. Shaded area denotes 95% confidence intervals.
Table 2-3. Observed species richness, estimates of species richness, effective number of species (ENS) and sampling completeness generated using Chao type-1 coverage-based estimation. Observed Estimated Sampling Region ENS (±SE) richness richness (±SE) completeness
Lake 15 19.48 ± 7.16 5.91 ± 0.32 99.07%
Polk 12 17.96 ± 11.61 4.52 ± 0.52 96.39%
St. Lucie 13 22.97 ± 10.19 3.61 ± 0.20 98.66%
36
Table 2-4. Individuals per trap for the 5 major species during 2017 and 2018, divided by region. Year Species Region Mean est. LCL UCL 2017 C. hilari Lake Co. 0.0857 0.0434 - 0.1692 ab Polk Co. 0.0000 0.0000 - 0.0000 - St. Lucie Co. 0.0071 0.0010 - 0.0511 a
E. ictericus Lake Co. 0.1905 0.1182 - 0.3070 abc Polk Co. 0.1026 0.0567 - 0.1857 ab St. Lucie Co. 0.1571 0.1003 - 0.2462 ab
E. servus Lake Co. 5.4667 4.4471 - 6.7200 gh Polk Co. 0.5641 0.4176 - 0.7620 def St. Lucie Co. 7.5857 6.3692 - 9.0346 h
N. viridula Lake Co. 0.0667 0.0310 - 0.1432 ab Polk Co. 0.0085 0.0012 - 0.0612 ab St. Lucie Co. 0.0000 0.0000 - 0.0000 -
T. perditor Lake Co. 0.1524 0.0901 - 0.2577 ab Polk Co. 0.0427 0.0175 - 0.1046 a St. Lucie Co. 0.9714 0.7681 - 1.2287 ef
2018 C. hilari Lake Co. 0.0472 0.0192 - 0.1156 a Polk Co. 0.0000 0.0000 - 0.0000 - St. Lucie Co. 0.0000 0.0000 - 0.0000 -
E. ictericus Lake Co. 0.0472 0.0192 - 0.1156 a Polk Co. 0.0667 0.0310 - 0.1432 ab St. Lucie Co. 0.1806 0.1190 - 0.2740 ab
E. servus Lake Co. 3.5377 2.8560 - 4.3822 g Polk Co. 0.5238 0.3784 - 0.7251 cde St. Lucie Co. 3.2153 2.6703 - 3.8714 g
N. viridula Lake Co. 0.2736 0.1816 - 0.4122 abcd Polk Co. 0.0667 0.0310 - 0.1432 ab St. Lucie Co. 0.0069 0.0010 - 0.0496 a
T. perditor Lake Co. 0.2170 0.1383 - 0.3403 abcd Polk Co. 0.3333 0.2276 - 0.4882 bcd St. Lucie Co. 1.1528 0.9232 - 1.4395 f LCL and UCL refer to lower and upper 95% confidence levels, respectively. Rows with common letters are not significantly different at alpha=0.05. P values were adjusted using the Tukey method for a family of 30 estimates.
37
Table 2-5. Record of H. halys detection, stage and reproductive status during three years of surveillance in Florida peach orchards. Date Location Stage # Reproductive status 3/31/2016 Lake 1 Adult 1 – 4/14/2016 Lake 1 Adult 1 – 2/19/2017 Lake 1 Adult 1 1/1 gravid 2/19/2017 Lake 2 Adult 1 – 2/19/2017 Polk Adult 1 1/1 gravid 3/5/2017 Polk Adult 1 – 4/2/2017 Lake 1 Adult 1 1/1 gravid 4/14/2017 Lake 2 Adult 1 1/1 non- vitellogenic 4/29/2017 Polk Adult 1 – 3/1/2018 Lake 2 Adult 4 1/1 gravid 3/15/2018 Lake 2 Adult 3 2/2 vitellogenic 4/12/2018 Lake 2 Adult 1 1/1 vitellogenic 4/25/2018 Polk Adult 1 1/1 vitellogenic 5/9/2018 Lake 2 Nymphs 4 – 5/9/2018 Lake 2 Eggs 2 (masses) – 5/23/2018 Lake 2 Adult 1 1/1 gravid 6/7/2018 Lake 1 Adult 1 – 6/7/2018 Lake 2 Adult 1 1/1 gravid 6/7/2018 Polk Adult 1 – 6/7/2018 St. Lucie 1 Adult 1 –
38
CHAPTER 3 TRAP-BASED ECONOMIC INJURY LEVELS AND THRESHOLDS FOR STINKBUGS IN FLORIDA PEACHES
Integrated pest management (IPM) is an approach to pest management which relies on the use of a variety of methods and tactics to reduce pest injury (Pedigo et al.
1986, Ehler 2006). Integrated pest management presents an alternative to pest management strategies which are entirely chemical based, with pesticide applications being made under a calendar schedule or without prior pest monitoring to determine pest levels. IPM programs treat pesticides as one option among many and promote their use in a way that is compatible with other management strategies (Barzman et al.
2015). For instance, the benefits of natural control provided by predators and parasitoids in the farm-scape is severely limited when broad spectrum or non-selective pesticides are applied; IPM programs incorporate the benefits of an enhanced natural enemy community in the overall management strategy and evaluate pesticide usage in light of the harm caused to natural enemies (Michaud and Grant 2003). Integrated pest management has been viewed as being synonymous with economic pest management; by which judicious use of pesticides is obtained through rational evaluation of the costs and benefits of their application. While only one aspect of IPM, economic theory in pest management has been a significant driver of IPM uptake in several agricultural systems
(Gent et al. 2011).
There are several economic models applicable to decision making for pest management, most of which involve a real-time assessment of the field situation before arriving at a decision on what the optimal action may be (Mumford and Norton 1984).
The economic-threshold model explicitly utilizes the relationship between pest-density and expected yield, such that control actions will only be made when yield losses are
39
greater than the cost of control (Stern et al. 1959, Mumford and Norton 1984, Pedigo et al. 1986) Decision making in this context generally involves 3 steps: 1) Identifying the relationship between pest density and yield loss, 2) Determining the economic injury level (EIL), represented as the pest density at which the value of yield gained from control is equivalent to the cost of control, and 3) Determining the economic threshold
(ET), taken as the pest density at which control should be implemented such that populations are held below the EIL due to control actions, and which otherwise would have exceeded the EIL (Stern et al. 1959). The economic threshold is the bridge that connects pest monitoring to action and is a key component of an IPM program.
Economic thresholds can be developed in a variety of ways, and their deployment can take on a variety of forms. Poston et al. (1983) outlined four general approaches to threshold-based pest management decision making;
1. Non-threshold approach. Used when pest populations are consistently above
economic injury levels, or when the true threshold is below a level which can
be detected using available monitoring techniques.
2. Nominal/subjective thresholds. Based on experience and prior success in
managing the target pest. Nominal thresholds are not derived from hard data;
however, they provide a useable point of reference for determining when
control is justified.
3. Simple thresholds. Based on a relationship between pest density, yield loss,
control costs and control efficacy, as specified by the EIL. Simple thresholds
are supported by data and represent a generalization of the dynamics of pest
injury and obtained control.
40
4. Comprehensive thresholds. Proposed by Poston et al. (1983) as a necessary
advancement over simple thresholds. Comprehensive thresholds include
interactions between multiple pests and incorporate site-specific attributes.
In southeastern peach production the management recommendation for cat facing pests involves pesticide applications being made from shuck split to harvest with a spray interval of 7 to 14 days depending on pest pressure (Horton et al. 2016). In the threshold classification schema of Poston et al. (1983) this approach falls between the prophylactic non-threshold approach (control actions are routine) and the experience- based nominal threshold approach (control actions can be scaled based on perceived pest pressure). A nominal threshold, based on sampling via limb-jarring, of 1 stink bug/limb jarring has been proposed for peaches, as treatment at this threshold resulted in losses below 1%, a level deemed to be acceptable for fresh market peach producers
(Johnson et al. 2002). When yellow pyramid traps baited with 50ul of methyl (2E, 4Z)- decadienoate impregnated into a rubber septum were used in lieu of the limb-jarring method, a value of 60 E. servus per trap per week was observed to be equivalent to the
1 stink bug / limb jarring threshold used previously (Johnson et al. 2002). A direct relationship between pest density, as determined by a sampling method, and injury at harvest, has not been developed for southeastern peaches; nor has this relationship been combined with population dynamics to produce a simple threshold for catfacing pests of southeastern peaches.
It is possible that the current reliance on a nominal threshold approach to management of catfacing pests in southeastern peaches is appropriate considering the high crop value, persistent damage potential, and adequate efficacy of available control
41
methods for catfacing pests. Nominal thresholds are the result of experience, having been developed through years of trial and error. However, there are several reasons why reliance on nominal thresholds may not be appropriate for Florida, and possibly for the southeastern generally, and which would justify advancing the current practice from a nominal threshold to a simple threshold. As peach production in southern Florida is a relatively new phenomenon, there is limited prior experience to draw from, a requisite in establishing nominal thresholds. Additionally, increased restrictions on pesticide usage and availability have limited the options for chemical control, forcing growers to move towards less familiar pesticides. Lastly, the arrival of Halyomorpha halys in the southeast, including the detection of a limited population in Florida, suggests the very composition of the catfacing pest complex in southeastern peaches is subject to change
(Penca and Hodges 2018). The combination of these factors indicates a new paradigm for cat facing pest management may developing. The purpose of this chapter is to use a combination of trap-based monitoring and intensive injury sampling to better understand the interaction between stink bug populations and fruit injury, and to put this knowledge to use in the development of an EIL and ETs for stink bugs in Florida peach orchards.
Methods
Site Selection and Trapping
Estimates of E. servus population density were based on capture in yellow pyramid traps baited with a multispecies pheromone lure formulated as a two component lure containing the Plautia stali pheromone (Methyl (E,E,Z)-2,4,6- decatrienoate) and H. halys pheromone (a mixture of (3S,6S,7R,10S)-10,11-epoxy-1- bisabolen-3-ol and (3R,6S,7R,10S)-10,11-epoxy-1-bisabolen-3-ol) (ChemTica
International, SA, Heredia, Costa Rica) (Yonce and Mizell 1997, Hogmire and Leskey
42
2006). Traps were deployed at 5 orchards during the 2017 and 2018 growing seasons, beginning in mid-January and serviced every 13-14 days until mid-June. Sites were selected to represent the major growing areas for subtropical peaches in Florida. Two sites were selected in Lake County, Florida, hereafter referred to as “Lake 1” and “Lake
2”. The trapped area of the Lake 1 site encompassed 1.7 hectares, bordered on the south by citrus and a mixed pine/hardwood forest, with the remaining adjacent areas consisting of mowed grass in a rural-residential setting. A total of 7 traps were placed at
Lake 1, with 5 placed along the orchard edges, and 2 placed in the interior of the block.
Lake 2 was the smallest site sampled, at 0.45 hectares. Lake 2 was bordered by citrus on the northern side, mixed pine/hardwoods on the southern side, and fallow, unmaintained areas on the east and west. The general area surrounding Lake 2 was comprised of rural-residential property and citrus plantings. A total of 3 traps were placed at Lake 2 due to its small size. A single site was selected in Polk County, Florida
(referred to as Polk 1). The total size of the orchard was approximately 6.9 hectares and was divided into two separate blocks (north and south). Polk 1 was in a developed suburban area, with a lake adjacent to the east side of the orchard, and a second, larger lake approximately 200 meters to the west. Citrus production is widespread in the vicinity of the Polk 1 site, with the nearest citrus grove 100 meters to the north. A total of
10 traps were deployed at Polk 1 (5 at each of the two trapped blocks), with 6 traps placed on the block edges, and 4 placed in the block interior. The remaining two sites were in St. Lucie County in southeastern Florida (referred to as SE 1 and SE 2). SE 1 was the largest orchard in this study, with over 49 hectares under cultivation in the immediate area. A 1.63 hectare block was selected for trapping. SE 1 was partially
43
bordered by a fallow field with sporadic hardwood trees on the northern and eastern side, and by additional peach blocks on the remaining borders. An irrigation canal ran the length of the southern side of SE 1, with weeds growing on the bank of the canal.
SE2 was approximately 3.6 hectares in size. The trapping area of SE 2 consisted of 1.5 hectares of the western half of the block. The western border of SE 2 was a mixture of unmaintained weeds, shrubs and mixed-hardwoods. The southern border of SE 2 included an irrigation canal with broadleaf weeds present on the canal banks. A large citrus orchard occupied the area beyond the irrigation canal to the south of SE 2. A second irrigation canal lined the northern border of SE 2, with a fallow field located beyond the irrigation canal. The eastern border of SE 2 contained a pasture area, with a small herd of cattle. A total of 7 traps each were used at SE 1 and SE 2, with 5 traps placed on the edge rows and 2 traps placed in the orchard interiors.
At each visit the captured stink bugs were identified to species, enumerated, and removed from the traps and the pheromone lure was replaced. Trap capture data was censored such that only capture dates from bloom to harvest were used to model the relationship between E. servus population density and fruit injury. Each 14-day trapping period was referred to as a “period”, with each period referencing a specific phenological stage. When converted to E. servus / day, by dividing the average trap capture by the duration of the capture period, the resulting value represents the arrival rate of stink bugs to the trap. Both the unconverted mean trap capture, and the arrival rate serve as an index of E. servus population densities.
Damage Data
To determine the extent of injury resulting from stink bug feeding a damage survey was conducted, with sites Lake 1, Polk 1, SE1 and SE2 surveyed at the
44
commencement of harvest. Fruit were visually inspected for injury, with 30 fruit per tree inspected from approximately 5% of the trees in the sampling block. Trees were selected in a gridded fashion, with sampled trees within the same row separated by three unsampled trees, and sampled rows separated by two unsampled rows. Damage was classified as either catfacing or sap-type injury. Probability of injury at harvest was taken as the total injured fruit (cat facing + sap type injury) divided by the total number of fruits inspected.
EIL Calculations
Economic injury level estimates were determined using the formula of Pedigo et al. (1986):
EIL = C/VIDK (3-1)
Where C is the per-acre cost of a single pesticide application, V is the per-acre value of the crop, I is the relationship between injury units and pest density, D is the damage per injury unit (set at 1 for direct injury wherein the value of the injured fruit is reduced to zero, as is the case in stink bug feeding), and K is the proportional reduction in pest density resulting from the management action.
The relationship between the probability of injury and pest density was estimated by linear and logistic regression, with the response variable (probability of injury at harvest) regressed against the predictor variable, the cumulative rate of E. servus arrivals, equivalent to area under the curve (AUC) of the arrival rate across the growing season. Utilization of the AUC produces a resulting coefficient from the regression equation where each unit change in the response variable (percent injury) represents an arrival rate persisting over a period of one day. This approach allows for prediction of
45
probability of injury for a pest density persisting any specified number of days, a necessary feature for determining economic injury levels and economic thresholds over discrete treatment windows. However, as this estimator is originally derived from a trapping period of 14 days caution should be exercised in applying the results to trap capture obtained from trap periods significantly shorter or longer than 14 days.
A range of values for C and V were used, encompassing expectations of these values based on published pesticide prices, grower-supplied estimates and estimates from enterprise budgets for Florida peach orchards (Morgan and Olmstead 2013b,
Olmstead et al. 2013). These values included costs of control ranging from $0 to $100 per acre and per acre crop values of $5000, $7500, $10,000 and $12,500. Because the value for K is difficult to estimate, economic Injury levels were generated at K values of
0.25, 0.50. and 0.75. Our assumptions on what values for K are most appropriate for management of stinkbugs in peaches are presented in the discussion section.
Economic Threshold Calculations
To determine the magnitude of increase/decrease in mean E. servus trap capture from ti-1 to ti for each trapping period a generalized linear model was employed, with E. servus trap capture at ti set as the response value predicted by E. servus trap capture at ti-1 , with an interaction term for each trapping period. The resulting coefficients provide an expectation of the factor of change for each trapping period, such that the current mean trap value at any period, when multiplied by the regression coefficient for the specified trap period, will produce a best-estimate of the trap capture for the following period. The inverse of the regression coefficient of each period (referred to here as the
“change factor”) represents the percentage of the EIL at which the ET should be set for each period. For a conservative ET the change factor was produced using the
46
regression coefficient for each period plus one standard error. The resulting ET from this conservative approach represents the population density at which approximately
84.1% of predicted future population densities are expected to fall below the EIL.
Model validation was conducted using a modified form of leave-one-out (LOO) validation, where regression results were produced from a data set with a single site- year omitted (the training set). The model generated from the LOO training set was then used to predict values for the omitted site, and the observed error of each trapping period was calculated. Validation using this method allows us to evaluate how the model preforms when new data is presented, providing an expectation of model performance in future Florida peach seasons.
Results
Relationship between trap capture and fruit injury at harvest
Populations of Euschistus servus, as measured by trap capture, followed a trend that was consistent across sites and between years (Figure 3-1, Figure 3-2). This trend can be described as an initially slow rate of increase during the bloom phase of peach development, with rapid increases occurring during the early stages of fruit set and sizing. Populations generally hit their season long maximum approximately 10 weeks post-bloom, after which they trended downwards or maintained a stable level. The period of stabilization and decrease occurred shortly before harvest.
At total of 8,333 pieces of fruit were examined for injury at harvest. Percent fruit injury at harvest ranged from 2.3% to 11.9%. Cat-facing injury was more prevalent at harvest than sap-type injury (paired t-test, t = 6.4781, df = 7, p-value = 0.00034). The range of injury values fell below 15%, limiting our ability to investigate the relationship between trap capture and fruit injury at higher population densities. Within the range of
47
values observed during this study a significant correlation between AUC and percent fruit injury was observed when modeled via both linear and logistic regression (Figure 3-
3). At the lower values of this relationship linear regression provided improved model fit when compared to the binomial-logistic model (Figure 3-3); however, at very large AUC values the linear relationship predicts fruit injury will exceed the 100% limit. Modeling injury as a probability via logistic regression bounds the response variable between 0 and 1 and thus may provide greater accuracy at population densities above those observed during this study.
Stink bug capture was converted to the under the curve (AUC), with mean trap capture at each time period divided by the number of trapping days per period, producing an estimate of stink bug-day equivalents for each site x year combination.
When probability of fruit injury per unit AUC was modeled as a logistic relationship the logit probability of injury increased by 0.684 ± 0.052 for each unit increase in AUC
(logistic regression, F(1,6):55.64, P<.001) (Figure 3-3). The results can be interpreted as follows:
푒−3.977819+.683952푥 (3-2) 푃 = 1 + 푒−3.977819+.683952푥
Where 푃 is the probability of fruit injury and 푥 is the area under the curve. An approximation of 푥 can be made as follows:
푥 ≈ 푏푧/푑2 (3-3)
Where 푏 is the number of E. servus captured during a period of length 푑. The value 푧 is the interval of interest, equivalent to the length of time the observed pest density will be present in the orchard. The resulting estimate of 푃 is the probability of an individual fruit
48
incurring either cat facing or sap-type injury over a length of time 푧. This estimate is adaptable to varying lengths of trapping periods and provides an approximation of injury over varying time intervals. For season long predictions, the estimated time of exposure to injury is 70 days, approximating the length of time between shuck split and harvest, where fruit are present and vulnerable to injury. At a constant population density
(measured as mean insects/trap/day), the AUC is equivalent to the population density multiplied by 70, with each day being equivalent to 1/70th of the AUC.
linear regression indicated a strong linear relationship between AUC and fruit injury (R2 =0.91, P < 0.0001). The resulting equation for predicting percent injury can be expressed as follows:
푃 = 4.0522푥 + 0.3955푐 (3-4) where x is the AUC and can be approximated for shorter intervals following equation 3-
3. The y-intercept value (0.3955) from the linear regression equation represents the percent injury occurring when the AUC is equal to zero over a 70 day period (equivalent to zero capture of E. servus over an entire season). The value 푐 represents a correction for the y-intercept, necessary when applying equation 3-4 for intervals other than the 70 day-season long period used in this model. The correct value can be determined from equation 3-5.
푐 = 1 − 푧/70 (3-5)
Where 푧 is the interval of interest, i.e. the number of days over which the percent injury incurred is being calculated.
49
EIL Calculation
The relationship between mean trap capture during the trapping period and probability of injury incurred serve as the value I in equation 3-1, relating the population density to injury incurred, and allowing the development of economic injury levels for varying levels of 퐶, 푉 and 퐾.
EIL values are presented in figure 3-4 based on the linear relationship, and in figure 3-5 for the logistic relationship. Values presented in figures 3-4 and 3-5 represent values that solve equation 3-1. The proportional reduction in pest population, K, is treated as the percent reduction in trap approach rate obtained over a 14-day trapping period. Values for K of 0.25,0.50, 0.75 and 1 are included in figures 3-4 and 3-5 as differently colored lines. Solutions are presented as separate graphs for per acre crop values of $5,000, $7,500, $10,000 and $12,500.
Predicted Rate of Increase – Model Validation
The predicted rate of increase for each trapping period was described by the coefficient from linear regression model specified by the model:
E. servust ~ E. servust-1 : period. The resulting regression equation was significant
(F(7,57)=50.57,P<0.001, R2=0.8443). When capture data for a site x year combination was omitted from the training data as part of the cross-validation procedure, the model’s accuracy for the omitted site did not differ significantly from a model with the complete training set included (t-test of square prediction errors, t=1.4454, df=97.79, p = 0.1516).
The mean squared prediction error for the full model was 4.80, while the mean squared prediction error of the leave-one-out model was 7.79. Figure 3-6 and 3-7 present the predicted trap capture generated by the leave-one-out and full model, as well as the observed data, for 2017 and 2018 respectively.
50
Economic Thresholds
Thresholds were set for each trapping period at E. servus densities below the specified EIL, such that the expected increase at each period would produce a population density equal to the EIL. The product of the EIL and the inverse of the period-specified coefficients provides the best estimate of an ET value for each period
(Table 3-1). The resulting value can be interpreted as the population density which would be expected to reach the EIL after 14 days post sampling. Figure 3-8 presents ET values generated from an EIL of 5.53 E. servus per trap, as determined using the linear relationship of AUC to injury, a K values of 0.25, a crop value of $10,000/acre and a control cost of $40/acre. When the EIL of 5.53 was set, thresholds for periods 1,2,3,4,5 and 6 were 3.98, 2.10, 4.24, 4.59, 10.62 and 5.53 E. servus per trap, respectively.
When applied to the 2017 and 2018 trap capture data observed in this study, these ET values produced the correct treatment decision in 14 of 16 instances where the EIL had not previously been exceed in 2017, and 14 of 16 such instances in 2018 (Figure 3-8).
Discussion
Economic thresholds have been widely used for indirect pests, particularly those whose population densities are capable of rapid increase via reproduction within the crop (Onstad 1987). The application of economic injury levels and economic thresholds to stink bugs requires careful consideration of the specific factors driving population dynamics, as well as the mode by which these pests cause yield loss.
Once the relationship between pest density (as represented by trap capture) and injury is identified, the next step in establishing and EIL is identifying the value of the crop, the cost of the control action, and the K values. The value of an acre of Florida peaches varies by grower and by year as an interaction in obtained yield and fruit
51
prices. Budgets produced by Morgan & Olmstead (2013) suggests a mature, 4th year orchard will produce 8700 lbs. of fruit per acre, and command a price of $1.75 per lb.
Grower reported yields run significantly lower, in the 3000 to 5000lb per acre range
(Penca 2019, unpublished data). Additionally, the costs of materials being applied as well as the cost for sprayer equipment and labor vary between operations. While dynamic EILs have been encouraged (Poston et al. 1983, Pedigo et al. 1986), the method for setting the ET used in this study is in itself dynamic, though highly dependent on a reliable population trend during the Florida peach growing season.
Refinement of this approach is possible, and estimations would greatly benefit from additional years of data and a wider range of study sites. A more robust dataset can also enable the utilization of weather variables to predict E. servus population dynamics
(Ni et al. 2017).
The value K, representing the proportional reduction in pest density, is difficult to estimate. In our work the value for K must represent the reduction in pest density of a single application of a cost C. Additionally, the proportional reduction in trap capture from time tn to tn+1 must occur independently of natural fluctuations in population. In all applied settings K will fall below 1; factors that reduce the realized K value include pesticide lethality at the dose applied and the degree of achieved spray coverage. In the case of less-mobile pests, particularly those which reproduce rapidly within the cropping area, achieved K can be high when applications are made early, as significant mortality and persistent residual effects will prevent the population from reaching exponential levels of increase. In the case of stink bugs in peaches, we suspect the realized K value is low, likely below 0.5. This expectation results from the observation that 1) stink bugs
52
are highly mobile, polyphagous pests, and a significant portion of the population which may contribute to injury in the orchard during the interval between sampling may be found outside the orchard at the time of spraying, and thus will only be exposed to residual pesticides (McPherson and McPherson 2000), 2) Reproduction within the orchard plays a minor role in population density during very early season peach production, and 3) stink bugs, especially E. servus, exhibit some resiliency to pesticides and require adequate coverage exposure for mortality to occur (Hoffman et al. 1987,
Snodgrass et al. 2005). A review of pesticide spray trials targeting stink bugs in peaches suggests our expectations of a low K value may be justified. When compared to untreated control trees, the average reduction in injury resulting from multiple applications was 45.68% (Table 3-2). This reduction represents the cumulative effect of a series of sprays; however, in our application of equation 3-1 K should reflect the proportional reduction in injury from a single application, with the per-spray K value being lower than the total reduction in injury obtained. In the review of studies in table 3-
2, assuming each application had an equal and additive effect on injury reduction, the average K value observed was 0.1130 (Table 3-2). Assuming a generous K value of
0.25, a per acre value of $10,000 and cost to spray of $40 per acre, the resulting EIL for
Florida peaches is approximately 5.53 E. servus per trap over a 14-day trapping period.
The majority of sampling days fell below this EIL. However, should the K value be treated as 0.50 the EIL would be halved to 2.76, a value which was surpassed in the majority of sampling dates. An accurate perception of pesticide efficacy is essential in determining when it is rational to make a pesticide application; future research on
53
accurate K values for single sprays, multiple sprays, and the effect of application timing, is needed to improve EIL values for southeastern peaches.
The economic thresholds used in this study take advantage of a population trend that was observed to operate in Florida independent of site and year. The rapid increase in E. servus population density was timed with the period around shuck split and early fruit sizing. Peaches produced at the southeastern sites (St. Lucie county) were approximately 10 to 20 days ahead of peaches produced in both Polk and Lake
Counties. Based on these results, the period of greatest increase falls between mid-
February to mid-March, with populations generally reaching their season long maximums in late March and early April. In traditional peach growing areas of the southeastern USA E. servus populations tend to increase rapidly in mid to late May
(Horton et al. 2016); it is conceivable that the population peaks observed in Florida during March-April represent a similar phenomenon, advanced by 30-60 days on account of Florida’s warmer climate.
A second useful trend observed in our stink bug monitoring is the tendency for populations to reach their season-long maximums approximately 42 to 56 days after shuck split. Sites whose populations have not approached the EIL by 56 days are generally unlikely to exceed this level by harvest and management activities can respond accordingly with attention paid to the pre-harvest intervals of the available pesticides as harvest approaches. The timing of the Florida peach harvest corresponds with the end of the winter dry-season and the onset of warmer temperatures. The combination of warm & dry weather may be responsible for rapid increase in might populations in late April and early May (Penca, personal observations). The ability to
54
forego applications of pesticides may conserve natural enemies of mites, mitigating the risk of mite-induced defoliation during the harvest period (Hoyt and Caltagirone 2012).
While pheromone baited traps aim to reduce the effort required for sampling of pests, several important considerations must be made when combing traps with thresholds. The relationship between trap capture and injury at harvest, the EILs informed by this relationship, as well as the general trends in population dynamics that direct the ET values, are all generated using the specific trap design, deployment and pheromone lure described in this study. Variations to any of these can impact trap capture, resulting in a population index with a different relationship to the actual population in the environment than the relationship been used in the development of the
EIL and ET values. For example, changes in pheromone blends and dosage can result in drastically different trap capture (Khrimian et al. 2008, Cottrell and Horton 2011).
Adjustments to the design trap collection jar and cone, or inclusion of insecticidal ear tags, can alter trap capture retention efficiency (Cottrell 2001, Hogmire and Leskey
2006). Changes to trapping procedures that improve capture efficiency will lead to higher trap captures but will not relate to higher risk of fruit injury. Crucially, values generated in this study will not be applicable when a trapping procedure with differing trapping efficiency is used. This fact highlights the limits of a complex trap design, particularly one where multiple variables (i.e., trap color, size, materials, capture jar size, capture jar cone size, trap placement, pheromone lure, length of pheromone usage) are at play.
Lastly, the observations developed in this study represent a cat-facing pest complex dominated by Euschistus servus. This reflects the current situation in Florida;
55
however, other pests, including Leptoglossus phyllopus can be abundant in peach orchards and will contribute to cat facing and sap-type injury during the season (Mizell
2005, Foshee et al. 2008, Akotsen-Mensah et al. 2017). It is possible E. servus trap capture may serve as a proxy for L. phyllopus pest pressure; however, such a correlation would need to be developed through further research. While E. servus appears to be the primary catfacing pest of early season peaches in Florida, the recent discovery of a reproductive population of Halyomorpha halys in a Lake County peach orchard (Penca and Hodges 2018) raises concerns that H. halys may become a significant contributor to cat facing injury. If H. halys expands its range in Florida, and increases in abundance, it will become necessary to reevaluate the EILs and ETs developed in this work.
56
Figure 3-1. Population trends for adult E. servus during the 2017 season. Vertical bars represent the trapping date nearest to the commencement of fruit harvest.
57
Figure 3-2. Population trends for adult E. servus during the 2018 season. Vertical bars represent the trapping date nearest to the commencement of fruit harvest.
58
Figure 3-3. Description of the linear relationship between stink bug trap capture and percent injury over the entire injury interval (top) and restricted to the range of observed values (bottom). Trap capture has been converted to stink bug – day equivalents by calculating the area under the curve (AUC) for E. servus trap capture from shuck split to harvest.
59
V=$5,000 V=$7,500
V=$10,000 V=$12,500
Figure 3-4. EIL estimates based on a logistic relationship between E. servus population density and probability of fruit injury. Estimates are for control measures with K values equal to 0.25 (blue line), 0.5 (red line), 0.75 (orange line) and 1.0 (black line). EIL values are for single applications under 14-day treatment windows.
60
V=$5,000 V=$7,500
V=$10,000 V=$12,500
Figure 3-5. EIL estimates based on a linear relationship between E. servus population density and probability of fruit injury. Estimates are for control measures with K values equal to 0.25 (blue line), 0.5 (red line), 0.75 (orange line) and 1.0 (black line). EIL values are for single applications under 14-day treatment windows.
61
Figure 3-6. Predicted and observed E. servus trap capture, by trapping period for the 2017 season. The “full” predictions (red lines) represent model estimates from the complete data set. The “LOO” predictions (green lines) represent model estimates from the validation procedure, where data for each site was omitted from the training data used to make the site prediction.
62
Figure 3-7. Predicted and observed E. servus trap capture, by trapping period for the 2018 season. The “full” predictions (red lines) represent model estimates from the complete data set. The “LOO” predictions (green lines) represent model estimates from the validation procedure, where data for each site was omitted from the training data used to make the site prediction.
63
Table 3-1. Estimates of the rate of change, by period, as determined by linear regression with the previous period trap capture as the predictor variable. 1 2 Period Phenological Stage Month B0 ± SE ET adj. factor 1 Bloom – Shuck split January – February 1.05 ± 0.95 0.72 2 Shuck split – Sizing 1 February 2.37 ± 0.28 0.38
3 Sizing 1 – Sizing 2 February – March 1.13 ± 0.12 0.80 4 Sizing 2 – Ripening March 1.10 ± 0.10 0.83 5 Ripening – Harvest 1 March – April 0.43 ± 0.09 1.92
6 Harvest 1 April 0.84 ± 0.16 1.00 1All coefficients significant at P=0.001 2The ET adj. factor is the inverse of the model coefficients + 1 SE
Figure 3-8. Evaluation of an economic thresholds based on 2017 (red) and 2018 (blue) trap capture at an EIL of 5.53 bugs per trap, derived from the linear injury relationship, a K value of 0.25, a cost of treatment of $30 per acre, and a crop value of $7,500 per acre. Economic thresholds for each period are denoted by the blue lines. The orange shade area represents the range between ET and EIL where treatment should be made to avoid reaching the EIL.
64
Table 3-2. Review of treatment efficacy studies for reduction of catfacing injury in peaches.
# of % Catfacing % Catfacing % Catfacing % Reduction Treatment Source Applications Treatment Control Reduction / Application Altacor 35WG 4 30.00 81.50 -63.19 -15.80 Frank & Biggs 2014
Imidan 70WP + Lannate LV 4 31.00 81.50 -61.96 -15.49 Frank & Biggs 2014 azinphosmethyl 50WSB + phosmet (2) + Danitol 5 33.6 65.6 -48.78 -9.76 Nielsen et al. 2007 2.4EC (2) azinphosmethyl 50WSB + phosmet (2) +Baythroid 5 38.4 65.6 -41.46 -8.29 Nielsen et al. 2007 2E (2) azinphosmethyl 50WSB + phosmet (2) + Assail 5 70.4 65.6 7.32 1.46 Nielsen et al. 2007 30SG 2.4EC (2) azinphosmethyl 50WSB + phosmet (2) +Imidan 5 50.4 65.6 -23.17 -4.63 Nielsen et al. 2007 70WP azinphosmethyl 50WSB + phosmet (2) + Lannate 5 54.4 65.6 -17.07 -3.41 Nielsen et al. 2007 92SP 2.4EC (2) azinphosmethyl 50WSB + phosmet (2) + Danitol 5 40.8 65.6 -37.80 -7.56 Nielsen et al. 2007 Warrior E (2) Asana XL 6 54.00 68.00 -20.59 -3.43 Nielsen & Rucker 2013
Danitol 2.4 EC 6 45.00 68.00 -33.82 -5.64 Nielsen & Rucker 2013
Ammo 2.5 EC 6 4.50 24.00 -81.25 -13.54 Pfeiffer et al. 1985
FMC 54800 2EC 6 2.10 24.00 -91.25 -15.21 Pfeiffer et al. 1985
Carzol 92SP 6 9.80 24.00 -59.17 -9.86 Pfeiffer et al. 1985
Baythroid XL 6 15.30 44.70 -65.77 -10.96 Rucker et al. 2009
Danitol SC 6 18.00 44.70 -59.73 -9.96 Rucker et al. 2009
Warrior 6 19.30 44.70 -56.82 -9.47 Rucker et al. 2009
Imidan 70WP 6 24.70 44.70 -44.74 -7.46 Rucker et al. 2009
Phaser 50WP 1 33.33 36.67 -9.11 -9.11 Waldstein et al. 2004
65
Table 3-2 (Continued).
Phaser 50WP 1 41.67 36.67 13.64 13.64 Waldstein et al. 2004
Phaser 50WP 2 26.67 36.67 -27.27 -13.64 Waldstein et al. 2004
Danitol 2.4 EC 1 20.00 36.67 -45.46 -45.46 Waldstein et al. 2004
Pounce 3.2 EC 1 26.67 36.67 -27.27 -27.27 Waldstein et al. 2004
Pounce 3.2 EC 1 33.33 36.67 -9.11 -9.11 Waldstein et al. 2004
Pounce 3.2 EC 2 18.33 36.67 -50.01 -25.01 Waldstein et al. 2004
Assail 70WP 7 3.10 10.60 -70.75 -10.11 Waldstein et al. 2005
Esteem 35WP 7 6.80 10.60 -35.85 -5.12 Waldstein et al. 2005
Pounce 3.2EC 7 2.50 10.60 -76.42 -10.92 Waldstein et al. 2005
Provado 1.6F 7 5.60 10.60 -47.17 -6.74 Waldstein et al. 2005
Damoil 7 5.00 10.60 -52.83 -7.55 Waldstein et al. 2005
Endigo 2.71ZC + PermUp 3.2EC 4 0.00 2.00 -100.00 -25.00 Walgenbach & Schoof 2014
Actara 25WG + PermUp 3.2EC 4 0.50 2.00 -75.00 -18.75 Walgenbach & Schoof 2014
Imidan 70WP + PermUp 3.2EC 4 1.00 2.00 -50.00 -12.50 Walgenbach & Schoof 2014
Average 4.63 23.94 39.345 -45.68 -11.30
66
CHAPTER 4 DEVELOPMENT AND THRESHOLD-INDEPENDANT EVALUATION OF SEQUENTIAL SAMPLING PLANS FOR STINK BUG INJURY IN FLORIDA PEACHES
Stink bugs are among the most important direct pests of commercial peach,
Prunus persica (L.), production in the southeastern United States (Foshee et al. 2008,
Horton et al. 2016). In areas of the southeastern United States where large-scale commercial peach production occurs the major stink bug pest species include the brown stink bug, Euschistus servus (Say), the green stink bug, Chinavia hilaris (Say), and the southern green stinkbug, Nezara viridula (L.). The exotic-invasive brown marmorated stink bug, Halyomorpha halys (Stål), has been reported in the major peach producing states of South Carolina, Georgia and Alabama, but widespread injury to peach associated with H. halys has not been reported in the Southeast. Non-pentatomid
Heteroptera, including the tarnished plant bug, Lygus lineolaris (Palisot de Beauvois)
(Miridae), and the leaffooted bug, Leptoglossus phyllopus (L.) (Coreidae), can also contribute to fruit injury (Atanasso et al. 2002, Akotsen-Mensah et al. 2017). Together these pests form the catfacing pest complex of peaches in the southeastern United
States, with the term “catfacing” describing the malformed and pinched appearance of fruit injury caused by these pests (Chandler 1955, Rings 1957). Even under extensive management, feeding by members of the catfacing pest complex can result in losses in the 3 to 10 percent range, though fruit injury can be as high as 80% or more in mid-
Atlantic orchards exposed to outbreak populations of H. halys (Woodside 1950, Killian and Meyer 1984, Leskey, Short, et al. 2012).
Sampling to estimate pest density is a central tenet of integrated pest management (IPM) (Stern et al. 1959); however, an efficient sampling regime for
67
estimating yield loss and pest pressure has not been developed for catfacing pests in southeastern peach production. Sampling using damage as a proxy for stink bug presence has been successfully used in peach (Joseph et al. 2014), as well as other systems, including cotton (Reay-Jones et al. 2010). Sampling using fruit injury has several advantages over direct counts of insects, specifically, this approach is not affected by changes in insect presence due to environmental conditions at time of sampling. Injury-based sampling has the added advantage of being simple, as specialized equipment such as beat-sheets, traps and lures are not required. Simplified monitoring plans reduce the likelihood of user error that may produce inaccurate estimates, a feature that is particularly important when sample counts are compared to a threshold derived from a standardized sampling method.
In many cases a large sample size is required to produce accurate estimates of mean population density or mean percent fruit injury (Binns and Nyrop 1992). Often the sample size required to achieve an accurate estimate is impractically large, necessitating either the acceptance of a high degree of uncertainty (low precision) at a manageable sample size or the adoption of classification-based sampling approach.
Sequential sampling is one such approach. Sequential sampling is a method that allows for a variable sample size to be taken, with sampling proceeding until the cumulative sample count falls below a minimum stop point or above a maximum stop point (Wald
1945, Iwao 1975, Fowler and Lynch 1987). Sequential sampling can significantly reduce the sample size needed when compared to a random sample, with the reduction in sample size being achieved by terminating sampling early in the sample process when
68
the population mean/density being sampled is much smaller or much larger than the critical mean/density being used for classification.
Peach production is relatively new to the southern half of peninsular Florida, with approximately 1,400 acres in production (Morgan and Olmstead 2013a). The impact of the catfacing pest complex on the Florida peach industry is not known. Thus, an efficient injury-based sampling plan for stink bug injury in Florida peaches is needed, both as a decision-making tool for growers and as a research tool for development of
IPM practices. Two key pieces of information are needed for the development and evaluation of a sequential sampling plan for stink bug injury: (1) the spatial distribution of catfacing injury in peach orchards must be quantified according to an appropriate model, such as the relationship between mean and variance described by Taylor’s power law (TPL), and; (2) an economic threshold, or critical density, must be set as this value serves as a reference point for classification by the sequential sampling plan
(Binns et al. 2000). In this work we resolve the spatial distribution of catfacing injury through a multi-year survey of stink bug injury in Florida peach orchards. As an economic threshold for catfacing injury has not been established for southeastern peaches we implement three hypothetical thresholds and evaluating the rate of erroneous classification of pest density and average sample number via a threshold- independent simulation approach.
Methods
Data Collection
Sampling occurred at three peach orchards in Lake, Polk and St. Lucie County,
Florida. Sampling occurred at four time points for both the 2017 and 2018 season. Each
69
sample was taken 13-14 days apart, beginning after thinning (ca. February) and terminating at the commencement of harvest (ca. March/April). End-of season sampling occurred the week prior to the commencement of harvest at 3 additional sites in 2017 and 2 additional sites in 2018, producing a total of 14 and 13 unique site x date combinations during 2017 and 2018, respectively.
A sample unit consisted of 30 fruit from a single tree, visually inspected for injury, with each fruit categorized as either 1) uninjured, 2) catfacing injury or 3) sap injury. Sap injury took the form of transparent sap exuding from injured sites on the peach, whereas catfacing took the form of dimpling or a creased deformation of the fruit. Both catfacing and sap injury were combined to represent total injury associated with catfacing pests.
Individual trees were sampled by row starting at a corner of the block and sampling every third tree until the end of the block or the 25th tree (maximum 7 trees), whichever came first. The 1st row and every third row after row 1 was sampled up to row 19
(maximum 7 rows) or until the end of the orchard block, whichever came first. Sample sites, dates, and number of sample trees per site are summarized in Table 4-1.
Estimates of Taylor’s Power Law Parameters
The spatial structure of stink bug injury was described via Taylor’s power law
(equation 4-1) which models the relationship between the variance 푠2 and the mean 푥
(Taylor 1961).
푠2 = 푎푥푏 (4-1)
The variables 푎 and 푏 correspond to the resulting slope and y-intercept of the regression equation produced from linear regression of the log transformed mean and variance of each unique sample. 70
Development of Sequential Sampling Plans
Stop lines for sequential sampling plans were generated using equation 4-2, from
Binns et al. (2000) based off the method of sequential sampling developed by Iwao
(1975).
푎푢 푏 푎푢 푏 푈 = 푛 (푢 + 푍 √ 0 ) 퐿 = 푛 (푢 − 푍 √ 0 ) 푛 0 푎/2 푛 푛 0 푎/2 푛 (4-2)
Where 푈푛 and 퐿푛 are the upper and lower stop points at sample unit n. The critical density 푢0 is equivalent to the economic threshold. The value 푍푎/2 refers to the z score for half the alpha level. In our case the confidence level used was 90%, thus the half alpha value is .05, and the value of 푍푎/2 is 1.645. The values a and b are the parameters of Taylor’s power law solved for previously. As thresholds have not been developed for catfacing pests of Florida peaches, hypothetical economic thresholds of
0.6, 1.2 and 1.8 injured fruit per sample were used, corresponding with 2, 4, and 6 percent injury, respectively. Separate sequential plans were developed for each economic threshold.
Fixed Precision Sampling for Damage Estimates
The optimum sample size needed to achieve an estimate of fruit injury within a fixed level of precision was calculated using the general distribution formula of Wilson and Room (1983) (equation 4-3) where 푛 is the optimal sample size, 푥 is the injury density, a and b are the parameters of Taylor’s power law, and 퐷 is the fixed level of precision, expressed as a portion of the mean. Fixed precision sampling estimates were calculated for values of 퐷 = 0.10 and 퐷 = 0.20.
71
2 푍푎/2 푛 = ( ) 푎푥푏−2 퐷 (4-3)
Simulation-Based Evaluation of Sampling Procedures
To compare the sequential sampling plans against fixed-size simple random sample (SRS) of sample size 10 a total of 5,000 bootstrap simulations were performed using sampling with replacement drawn from all 26 empirical samples. Using bootstrapping to evaluate sampling plans is attractive as it allows the population to be described by the empirical observations. Alternative approaches, such as Monte-Carlo sampling from a descriptive distribution such as a negative binomial with a k value calculated from observational data, has been used, though difficulty in finding a common k value limits the reliability of this approach (Sylvester and Cox 1961, Wilson and Room 1983, Carleton et al. 2013).
At each sample unit the cumulative injury count was compared to the stop values for the sequential sampling plan developed for each threshold. The final decision for the simulation of sequential sampling was classified as either “treat” or “do not treat” and was made depending on whether the cumulative injury count passed above the upper sequential stop line or below the lower sequential stop line. The sequential sampling plan was constrained by a lower limit of 3 samples and an upper limit of 40 samples. For sequential simulations that reached the maximum sample size of 40, the cumulative injury count was divided by the sample number and the resulting average injury was compared to the threshold and a classification decision was made. The classification decision for the SRS method was generated via sampling with replacement of each empirical dataset and comparing the sample mean with the threshold value. When the sample mean exceeds the threshold value the decision was 72
“treat”, otherwise the decision was “do not treat”. The correct decision was determined by taking the mean injury density of each empirical data set and comparing it to the threshold. Each instance of simulation could thus be classified as either 1) correct-no treatment, 2) correct-treatment, 3) type I error (a false positive; treatment made when true density was below threshold) or 4) type II error (a false negative, no treatment made when true density was above threshold).
The likelihood of type I and type II error for the simulations of the sequential and fixed random sampling plans was modeled via binomial-logistic regression. The probability of error at a distance from the threshold 푢0 can be estimated via the coefficient of the binomial regression following equation 4-4, where 푝(푥) is the probability of error when the absolute difference between the true density of the sample set and the threshold density (푢0) is equal to 푥, e is Euler’s number and β is the coefficient of the binomial regression of the simulation results where the response variable was classified as either a correct (0) or incorrect (1) and the predictor was the distance of the empirical injury density to the threshold (푥).
1 푝(푥) = (4-4) 1 + 푒−훽푥
The definite integral of the right-hand side of equation 4 across the interval zero to ∞ provides the cumulative error probability of the interval. The standard logistic function is symmetrical around the inflection point of 0.5, thus the definite integral across the interval 0 to ∞ is equivalent to half the cumulative error probability of the interval to −∞ 푡표 ∞. Estimating error using this method produces equivalent values for type I and type II error probabilities. To account for differences in type I and type II error
73
rates, the method was repeated separately for error values occurring when the empirical injury density was below or above the critical density, allowing for separate estimates for cumulative probability of type I and type II error. Models were fit without an intercept, forcing the probability of error 푝(푥) at 푥 = 0 to equal 0.5.
The average sample number for the sequential sampling plan was determined using the simulation approach described above, where the number of samples needed to reach a decision was recorded for each of the 5,000 simulations of the 27 empirical data sets. The resulting estimates of sampling size thus covered the range of injury densities observed during the study. Sample sizes for the sequential sampling plans were bracketed by the upper (40) and lower (3) limit. To determine when a sequential sampling plan was more likely to result in a reduction in sample size when compared to a fixed random sample of size 10, we plotted the mean sample number for each sample site as a function of the difference between the sample site injury density and the threshold injury density.
Results
The coefficients of Taylor’s power law, as determined by the linear regression of the mean and variance for observed catfacing and sap/gummosis fruit injury were a =
1.183 ± 0.023 and b = 0.997 ± 0.039 (F = 643.1; df=1,24; P <.0001; R2 = 0.96). The slope of the log-log regression of mean and variance (b) was not significantly different from 1 (T-test, P = 0.4715. df=23), suggesting stink bug injury occurred in a random spatial arrangement in the peach orchard at the scale measured.
Graphical representations of sequential sampling plans, with stop lines, are presented in Fig. 1. Sequential plans were generated using hypothetical thresholds
74
ranging from 0.6, 1.2 and 1.8 injured fruit per samples, corresponding with 2, 4, and 6% injury total, respectively.
Optimal sample sizes needed for estimates of injury density at a fixed level of precision are presented in Fig. 2. At the overall mean injury level observed in this study
(1.2 injured fruit per 30 fruit sample, equivalent to 4.0% injured fruit) the optimal sample size to achieve an estimate within 10% and 20% of the mean was 267.4 and 66.9 samples, respectively. The optimum sample size rapidly increases as mean injury decreases, requiring 1071.1 and 267.8 samples to produce an estimate within 10% and
20% of the mean when the true mean injury level is 1%.
Binomial regression of the error rates of both sampling plans indicated a significant interaction between the probability of an erroneous classification and the absolute distance between sample site injury density and the hypothetical economic threshold. As the injury density approached the critical density of the hypothetical threshold the rate of error approached the maximum error rate of 50% for both sampling plans. The rate at which the likelihood of an erroneous classification fell from 0.50 to 0 was modeled via binomial regression. Separate models were produced to estimate probability of type I and type II errors, along with a symmetrical model which does not differentiate between error type (Fig. 3). The cumulative probability of error for each model, along with the regression coefficient and difference in error probability between sampling method is given in Table 2.
The mean sample size of the sequential sampling plans at each threshold are presented in Fig. 4. The average sample number was greatest when the empirical injury density was near the threshold density, and was close to the minimum of 3 sample units
75
when the empirical density was ± 1 percent point of the threshold density. Results of loess regression (Fig. 5) approximated the mean sample size of the sequential sampling plans to be less than 10 at distances from the threshold greater than 0.58 (95% CI =
0.53-0.63).
Discussion
A random arrangement of fruit injury within the peach orchards occurred at the scale measured. Random distributions are believed to be less prevalent in nature than aggregated distributions (Taylor et al. 1978); however, research on the spatial distribution of stink bugs suggests near-random distributions may be common. For instance, Pilkay et al. (2015) studied the spatial arrangement of stink bugs in wheat, corn, soybean, cotton and peanut in the southeastern United States, and found evidence of random distributions of adult stink bugs (predominately E. servus and N. viridula) in 34 of 56 (60.7%) sites sampled. When stink bug injury was the object being sampled results have varied. For instance Reay-Jones et al. (2010) sampled stink bug injury to cotton bolls (predominately caused by C. hilaris, E. servus and N. viridula) and found the TPL coefficient b to be 0.933 ± 0.144, suggesting the data was near Poisson- distributed and indicating a prevalence of random spatial distributions. Research on the spatial distribution of stink bug injury to peaches has been conducted in areas where the invasive brown marmorated stink bug, H. halys is the predominant catfacing pest.
Using Moran’s I as a measure of spatial autocorrelation, H. halys injury in peaches was predominately aggregated in New Jersey peach orchards, with only 3 of 15 site x date combinations showing evidence of random distribution (Hahn et al. 2017). The increased aggregation or clustering of H. halys has been attributed to its “border-driven”
76
behavior, in which H. halys invades an agricultural area from an adjacent wooded area, with high densities accumulating near the interface between habitats (Rice et al. 2014,
Blaauw et al. 2015a). If observations of H. halys in Florida peach orchards indicate establishment, and if H. halys becomes a major contributor to peach injury in Florida, it is possible that the spatial distribution of stink bug injury will shift towards a more aggregated distribution, thus necessitating a revision to the sampling plans developed for pre-H. halys invasion injury distributions (Penca and Hodges 2018).
It took between 1 and 2 hours to visually inspect 30 fruit per tree from 42 trees and was a labor-intensive process. As such, the expectation that a sample number greater than 42 will be obtained by growers is unlikely. For low densities of fruit injury,
42 sample units was below the optimal sampling size across most of the observed injury densities at both precision levels evaluated. At a precision level of 0.10, 42 samples would be an optimum sample size when 25% of the fruit in the orchard were injured, far exceeding what was observed in Florida orchards during this study. At a precision level of 0.20, 42 samples would be an optimum sample size when 6.3% of the fruit in the orchard were injured, a value closer to what was observed during the study, but likely much higher than a suitable economic threshold for fruit injury. The large optimal sample number at injury densities within the likely range of an economic threshold provides a strong justification for developing classification-based sampling methods which reduces sampling size and classification error.
Sequential sampling plans developed here provided a modest reduction in rates of type I and type II error when compared to the simple random sample. Interestingly, the estimated cumulative probability of Type I and Type II errors (0.115 and 0.087) for
77
the sequential sampling method, when combined, were very near 0.10, the predetermined error rate used in the development of the sequential sampling plans.
Similarly, the cumulative error probability of the symmetric model of the sequential sampling error was 0.104. As relative differences in classification error rate were highly dependent on the distance of the empirical injury density from the threshold, it was necessary to include distance from threshold in any comparison between sampling methods. To achieve this, we standardized the predictor value to be the absolute distance between the empirical injury density and the hypothetical threshold. This approach provided three main advantages: 1) it allows for regression with a more representative range of predictor values, and; 2) it allows for a comparison of the fixed random sampling plan and the sequential sampling plans independent of the hypothetical threshold. This method also forces the y-intercept to be equivalent to 0.50, indicating equivalent probability of type I and type II errors and a cumulative error probability of 1 when the population density is equal to the critical density. This is expected, as a binary classification of x̄ < 푢표 or x̄ > 푢표 will always produce an incorrect classification when x̄ = 푢표.
By adjusting injury density to represent the absolute distance from threshold, and simulating sampling using hypothetical thresholds that fit within the observed range found in the study sites, we can provide a threshold-independent estimate of the error- rate of a sequential sampling plan. Ultimately the realized error rate and mean sample size of the sequential sampling plan will depend on the combination of the threshold used and the distribution of injury in the orchards being evaluated. The sample sites used in the evaluation of these plans do not cover the entire range of possible mean
78
injury densities and distributions. We expect these results to hold true for orchards whose mean-variance relationship is well described by the TPL parameters generated by our sample sites and where the injury density falls near the 1 to 11% range observed in the sample sites during this study. The performance of these plans at critical densities outside the observed range may be limited as the TPL estimations of a and b did not include samples at mean injury densities higher than 12%.
Further modifications to sampling methods, with the goal of reducing incorrect classifications of pest populations and lowering the sample size and work required, will assist in the uptake of pest monitoring by growers for stink bug management. One area where improvements can be made is sample unit selection. In the simulated random sampling scheme used in this study, the real-world location of each random sample was not manipulated. Considerations of the relative location of each sample, such that a fixed sample size is taken over a representative area of the orchard (i.e. a belt transect), may improve the efficiency of sampling and be easily transferred to growers (Carleton et al. 2013). The selection of sample units within the orchard can be further optimized in response to any behavioral trends in stink bug dispersal within the orchard, particularly trends that may be detectable at different spatial scales than what was used in this study.
While simple to use, sequential sampling plans are more complex than a fixed- size simple random sample and may not be adopted by growers as readily. Additionally, a new sequential plan must be developed for specific economic thresholds, which should, in theory, change in response to fluctuations in commodity prices and control costs (Pedigo et al. 1986, Onstad 1987). Revenue for a 4th year Florida peach orchard
79
has been estimated at $15,225 per acre (Morgan and Olmstead 2013b), thus every 1- unit reduction in percent stink bug injury equates to a $152.25 per acre increase in revenue. While the efficacy of spray programs in reducing catfacing injury must be evaluated and compared with costs before a full economic threshold is developed, the significant savings generated from a modest decrease in fruit injury suggests a low economic threshold may be justified. In the absence of an established economic threshold for stink bugs in Florida peaches our development and evaluation of sequential sampling plans required the use of hypothetical, but probable, thresholds.
However, our approach allowed for evaluating the performance of these plans independent of the hypothetical threshold, thus providing a useful benchmark for future efforts to enhance the performance of sampling plans for stink bug injury in Florida peaches.
Table 4-1. Sample site location, orchard block size, sampling units and sample dates. A sampling unit consisted of 30 fruit per tree, visually inspected for injury. County Block Size Trees 2017 Dates 2018 Dates (Hectares) Sampled Lake 1.17 29 3/5, 3/19, 4/1, 4/14 2/28, 3/15, 3/27, 4/12 3/19, 4/2, 4/14, 4/29 3/1, 3/14, 3/27, 4/12 Polk 1.32 42
Polk 1.42 49 4/2 -
St. Lucie 1.63 49 4/2 3/29
St. Lucie 1.44 42 2/19, 3/5, 3/19, 4/1 2/15, 3/1, 3/15, 3/29
80
Figure 4-1. Sequential sampling plans for assessing density of stink bug injury to Florida peaches. Threshold of 2%, 4%, and 6% correspond with 0.6,1.2 and 1.8 injured fruit per sample, respectively.
Figure 4-2. Optimum sample size for estimation of fruit injury density at a fixed level of precision.
81
Figure 4-3. Approximation of the probability of erroneous classification using either a sequential or simple random sample (SRS) method. Lines represent the predicted probabilities from the binomial-logistic regression. Points represent estimated probabilities from bootstrap simulations of sampling from empirical data sets.
82
Table 4-2. Coefficients of binomial-logistic regression of the error rates resulting from simulated sampling. The cumulative error probability (CEP) is the overall likelihood of an erroneous classification of pest density across the range of distances-from-threshold. For the symmetric error model, the CEP value represents an equal probability of type I and type II error.
Reduction Error Method β se p CEP in CEP Type I Fixed -5.297 2.095 0.0040 0.1309 Random 0.0160 Sequential -6.032 1.733 0.0022 0.1149 (12.23%)
Type II Fixed -4.754 2.091 0.0339 0.1458 Random 0.0592 Sequential -8.004 3.774 0.0230 0.0866 (40.60%)
Symmetric Fixed -5.093 1.327 0.0001 0.1361 Random 0.0317 Sequential -6.636 1.893 0.0004 0.1044 (23.29%)
Figure 4-4. Average sample number by distance from threshold based on simulations of 27 empirical data sets. Sampling began with three samples and the maximum sample number was constrained to 40 samples.
83
Figure 4-5. Loess regression approximation of the average sample number as a function of distance to threshold. The intersection of line segments represents the point at which the average sample number fell below 10, the sample number used for the simple random sampling simulations. When the absolute distance from the threshold is greater than the horizontal line (x=0.577) sequential sampling will have an average sample number lower than 10, the sample number used for the simple random sampling.
84
CHAPTER 5
PYRIPROXYFEN TREATMENT TERMINATES HALYOMORPHA HALYS REPRODUCTIVE DIAPAUSE, WITH AN INDIRECT EFFECT ON ITS EGG PARASITOID TRISSOLCUS JAPONICUS*1
Native to eastern Asia, Halyomorpha halys (Stål) (Hemiptera: Pentatomidae) has become an invasive pest of global concern following its introduction and rapid spread through North America and Europe (Hoebeke and Carter 2003, Wermelinger et al.
2008, Zhu et al. 2012, Haye et al. 2015). Also known as the brown marmorated stink bug (BMSB), H. halys is a highly polyphagous pest, feeding on over 200 species of plants. In its native and invasive range BMSB occurs as a pest of tree fruit, causing an estimated 36 million USD in losses to tree fruit producers in the mid-Atlantic region of the United States in 2010 (Leskey, Hamilton, et al. 2012).
Current control of BMSB in agricultural settings relies heavily on the use of broad spectrum pesticides (Nielsen et al. 2008). The resumption of a calendar spray program for managing BMSB has disrupted established integrated pest management (IPM) programs, leading to an increase in the amount of pesticides being used by growers in the mid-Atlantic states (Blaauw et al. 2015). As an alternative to chemical control, classical biological control has been proposed as a means of managing BMSB populations, with the hymenopteran egg parasitoid Trissolcus japonicus (Ashmead)
(Hymenoptera: Scelionidae) currently being evaluated as a candidate for release (Yang et al. 2009, Ogburn et al. 2016). A sentinel egg mass study in Maryland detected the presence of T. japonicus in the landscape in 2014 (Talamas et al. 2015) with a
* Reprinted with permission from the Journal of Pest Science 85
subsequent detection occurring in 2015 in Washington State (Milnes et al. 2016) and
2016 in Oregon (Hedstrom et al. 2017).
In addition to its impact as an agricultural pest, H. halys also acts as a nuisance pest, with large numbers of shelter-seeking bugs entering homes during cooler months and overwintering in a state of reproductive diapause (Inkley 2012). For most temperate
Pentatomidae the environmental cue that initiates diapause is a decline in day length below a critical period (Saulich and Musolin 2012). This decline in photoperiod leads to a cessation of juvenile hormone (JH) production by the corpora allata, which causes regression of the reproductive organs into a non-reproducing state. Conversely, an increase in day length triggers the resumption of JH production, leading to oogenesis and resumption of reproductive activity (Kotaki and Yagi 1989, Denlinger 2002).
The insect growth regulator (IGR) class of pesticides includes a group of compounds that act as JH analogs. Commonly used members of this pesticide group include methoprene, hydroprene and pyriproxyfen. Of these three, pyriproxyfen has shown greater JH-like activity, likely due to a higher binding affinity for the JH receptor methoprene-tolerant (Met) (Hatakoshi et al. 1991, Charles et al. 2011). As a potent JH- agonist, pyriproxyfen has the potential to terminate the adult reproductive diapause of
H. halys (Numata and Hidaka 1984, Amiri et al. 2012). Our study aims to identify the effective doses of pyriproxyfen required for diapause termination in H. halys. In addition, pyriproxyfen treatment may lead to an increase in H. halys egg production with implications for hymenopteran egg parasitoids (Duan et al. 1995). As such, we investigate the ability of both H. halys and T. japonicus to complete development on pyriproxyfen-induced eggs.
86
Materials and Methods
Specimen Sources and Rearing
Specimens of H. halys were collected from Washington County, Maryland in
October of 2016 and held in mesh enclosures under a diapause maintaining photoperiod (10L:14D), and at 20-25 °C, 60% RH for a 15-day period to allow acclimation prior to treatment. Immediately following chemical treatment, specimens were transferred to a clear plastic container (25 cm diameter, 8.9 cm height) with mesh screen ventilation for the remainder of the bioassay. Rearing containers were lined with cotton paper and specimens were provided shelled raw peanuts, and water-soaked cotton, ad libitum. A stock colony of H. halys from the same collection site was reared separately under identical conditions with the exception of a long day photoperiod (16L:
8D), so as to terminate diapause. The long day stock colony provided eggs for use as controls when establishing baseline H. halys hatchability and T. japonicus viability.
Trissolcus japonicus specimens used in this study were sourced from a colony maintained at the Florida Department of Agriculture, Division of Plant Industry in
Gainesville, Florida. Trissolcus japonicus were reared from H. halys eggs in a 10-dram plastic vial provisioned with honey.
Pyriproxyfen Bioassay
Treatments consisted of technical grade pyriproxyfen (Chem Service Inc., West
Chester, PA) in acetone solvent, prepared via serial dilution at concentrations of 2%,
1%, 0.1%, 0.01%, 0.001%, along with an acetone control. A micro-applicator was used to apply 5 μl of treatment solution to the ventral abdomen of CO2 anesthetized adult H. halys. Applied doses were 0.05, 0.50, 5.00, 50.00 and 100.00 μg per insect. Each
87
treatment consisted of four replicates of 20 female and 10 male specimens and lasted
35 days from the date of pyriproxyfen application.
Oviposition and Oocyte Development Dose Response
Cages were checked daily and the total number of eggs was recorded for each day. To account for the effect of female mortality the number of eggs produced by each treatment at each observation was divided by the number of females surviving at the time of observation. The number of eggs per female for each day of the experiment was then summed to produce the standardized egg total for each treatment replicate, referred to here as “cumulative eggs per female”. Mean oviposition was analyzed using a quasi-Poisson generalized linear model selected due to the frequency of zero counts and heteroscedasticity. Mean hatchability was compared between treatments using least-squares means Tukey’s HSD.
Following the completion of the experiment, female H. halys were culled and dissected to determine the stage of oocyte development. A binary rating system was developed, based on the ovary grading system of Hodek (1971) in which the first and second stage of Hodek’s schema was considered representative of the diapause condition. Ovaries in which ovarioles presented any sign of oocyte differentiation, ranging in appearance from a swollen vitellarium up to fully mature chorionated eggs, were rated as showing signs of oogenesis, and thus not in a state of diapause. The early stages of oocyte development, appearing as a visually distinct swelling and pigmentation of the ovarioles below the nutritive tip, were encountered rarely.
Specimens that did not show signs of ovariole development were typified by having the widest point of each ovariole occurring at the nutritive tip (Saunders 1983). Binomial
88
response data was fitted to a two-parameter log-logistic function, where the inflection point represents the ED50. Calculation of ED10, ED50, ED90 and respective 95% confidence intervals was performed in R (R Core team 2015) using the ‘ED’ function of the drc package (Ritz and Streibig 2005).
H. halys Hatchability and T. japonicus Viability
To determine the effect of varying doses of pyriproxyfen on H. halys hatchability, a portion of each clutch was set aside in a 10-dram tube and monitored daily. Control eggs from H. halys colony specimens held under a long day (16L: 8D) photoperiod were reared under identical conditions to determine a baseline rate of hatchability for the experimental conditions. The number of emerged 1st instars was counted to determine the portion of hatch for each clutch tested. Results were fitted to a generalized linear model following a binomial (logit) regression. As each clutch served as a replicate composed of a varying number of eggs, the dependent variable was composed of the total number of successes and failures for each clutch tested, allowing clutch size to serve as a weight for each replicate in the regression model. Mean hatchability was compared between treatments using least-squares means Tukey’s HSD.
To determine the ability of the parasitoid T. japonicus to complete development on eggs produced by pyriproxyfen-treated adult H. halys, egg clutches from the bioassay study were exposed to a primigravid female T. japonicus in a 10 dram tube and held in an environmental chamber (26°C, 60%RH, 16L: 8D) for a minimum of 20 days, after which the number of emerged T. japonicus were counted and divided by the number of eggs provided to determine the percent viability. Signs of attempted parasitism, including observed oviposition behavior, darkening of egg clutches, and
89
parasitoid emergence/presence of dead parasitoid pupae, were observed in all clutches tested. As T. japonicus viability from treatment clutches was zero, only the mean control viability, standard error, and number of clutches per treatment was reported.
Results
Oviposition Response
There was a consistent positive relationship between pyriproxyfen dose and the cumulative number of eggs per female (Figure 5-1). Treatments of 5, 50 and 100 μg pyriproxyfen initiated oviposition between 20 and 22 days following exposure. The lowest dosage that produced an oviposition response was 0.5 μg, which produced a single clutch of 25 eggs 34 days after treatment. A total of 1549 eggs were produced during the experimental period, with the highest dose treated, 100 μg, producing 691 eggs.
Oocyte Development Dose Response
There was a strong dose dependent relationship between pyriproxyfen treatment and incidence of diapause termination in female H. halys (Figure 5-2). There was a clear visual distinction between ovaries in the diapause stage and ovaries showing signs of oocyte development, with diapausing ovaries maintaining a white/clear coloration and having the germarium/nutritive tip as their widest point. Oocyte development was indicated by a swelling of the vitellarium in intermediate cases, and with fully formed, chorionated oocytes visible in more extreme cases (Figure 5-3). The calculated effective doses and 95% confidence intervals which induced oocyte development and diapause termination in 10, 50 and 90 percent of the study population
90
were, respectively; ED10 : 0.15 (0.040-0.26), ED50 : 1.14 (0.61-1.67), ED90: 8.75 (2.42-
15.07) μg pyriproxyfen per female H. halys.
H. halys Hatchability and T. japonicus Viability
Significant differences were observed in hatchability of treatment eggs and eggs produced from untreated H. Halys colony specimens, with a majority of clutches tested having zero percent hatchability (Table 5-1). Viability of T. japonicus in all eggs induced via pyriproxyfen application was zero, differing significantly from a parasitoid viability of
71.8% in eggs from insects reared under long day length conditions (Table 5-1). All egg clutches presented to parasitoids showed a combination of the three signs of attempted parasitism (oviposition behavior, darkened eggs, emerged adult parasitoids/unemerged parasitoid pupae); however, the possibility that parasitoids failed to oviposit on some of the individual eggs within a clutch could not be ruled out. Following dissection of eggs parasitized by T. japonicus we observed that parasitoid development had frequently proceeded up to the pupal stage (Figure 5-4).
Discussion
Our results show that the juvenile hormone analog pyriproxyfen has the ability to induce oocyte development in diapausing adult H. halys, leading to oviposition of eggs with significantly reduced hatchability. The disruption of JH regulation can alter the normal timing of Pentatomidae reproduction when applied during periods of diapause transition. In the temperate United States, H. Halys enters diapause when the photoperiod drops below a critical period of around 14 hours, coinciding with the onset of cooler weather (Watanabe 1979, Nielsen and Hamilton 2009). Pyriproxyfen applications made during this time period may prolong the reproductive activity of H.
91
halys, causing oviposition of eggs when resources are low and environmental conditions are not conducive for nymphal survival. An increase in winter mortality from the expenditure of stored energy reserves during an extended reproductive period is also likely to occur (Hahn and Denlinger 2011). In the spring, diapausing H. halys exit their hibernaria as temperatures warm, with a lag period between spring emergence and the resumption of reproductive activity (Nielsen and Hamilton 2009). During this lag period, adults begin the reconstitution of their reproductive organs, with feeding required to meet their energetic demands. Exposure to pyriproxyfen during the early spring can accelerate the reconstitution of the ovaries and resumption of oogenesis and oviposition. Our study suggests that the eggs induced by pyriproxyfen exposure will have lower hatch rates and are unlikely to contribute to a population increase.
In most Pentatomidae studied, including H. halys, diapause is imaginal and facultative, with photoperiod providing the cue for modulation of JH production by the corpora allata (Saulich and Musolin 2012). The ability of JH to relay photoperiod information and regulate the genes involved in reproduction has been studied in the linden bug, Pyrrhocoris apterus (L.) (Heteroptera: Pyrrhocoridae). Juvenile hormone induced reproductive activity in adult P. apterus by binding to the JH receptor Met, which, in concert with the circadian proteins Clock and Cycle, upregulated expression of
Par domain protein 1 isoform 1 (Pdp1iso1) and downstream reproduction genes while simultaneously suppressing Cry2 and downstream diapause genes. Absence of JH led to dominant expression of Cry 2 over Pdp1iso1 , and conversely diapause associated genes were expressed (Bajgar et al. 2013b). Upregulation of Pdp1 iso1 and termination of diapause was also observed following exogenous application of the JH analog
92
methoprene (Bajgar et al. 2013a). Pyriproxyfen has high binding affinity to the JH receptor Met (Charles et al. 2011), and likely interacts with the JH-diapause regulating pathway in a similar manner as both JH and methoprene.
While more direct effects of pesticides on parasitoids, either by direct pesticide application to the parasitoid or to its host, have been well studied (Varma and Singh
1987, Rosenheim and Hoy 1988, Rahat et al. 2005, Saber et al. 2005, Stanley and
Preetha 2016), the present study is, to our knowledge, the first to show significant indirect effects to an egg parasitoid when the host species is treated with pesticides prior to both oogenesis and oviposition. The ultimate cause of the observed low H. halys hatchability and complete T. japonicus mortality remains unclear. One possibility is the vertical transmission of pyriproxyfen from the treated parent to the egg. Application of the pyriproxyfen to adult Sarcophaga ruficornis (F.) (Diptera: Sarcophagidae) resulted in significant larval mortality, suggesting vertical transmission of pyriproxyfen from adult to offspring is possible (Singh and Kumar 2015). Development of T. japonicus frequently progressed up to the pupal stage, as evidenced by dissections of unhatched eggs, mirroring observations in other Hymenoptera that have shown JH analogs interfering with the pupal-adult transformation (Hsiao and Hsiao 1969). The potential for metabolism of pyriproxyfen during the 20 to 34-day period prior to oviposition, and the lack of a dose dependent response in H. halys hatchability, reduces, but does not rule- out, the likelihood that vertically- transmitted pyriproxyfen acted directly on the developing parasitoids. Future studies designed to detect the presence of pyriproxyfen and its associated metabolites from within the yolk of pyriproxyfen-induced eggs will provide more conclusive evidence regarding this potential mode of exposure.
93
An alternative explanation for the observed T. japonicus mortality is that pyriproxyfen exposure lead to improper oocyte formation or insufficient provisioning of nutrients into the egg by the host, leading to partial, but incomplete development of both
H. halys and T. japonicus. In Heteroptera, JH is the primary hormone modulating vitellogenesis (Davey 1997). Pyriproxyfen treatment may have inhibited normal endocrine regulation of vitellogenesis via its high JH-receptor binding affinity, or by initiating vitellogenesis without adequate nutrition or ecdysteroid titer (Bownes 1989,
Davey 1997). Evidence suggests that other Trissolcus species are sensitive to host egg resource availability, with an increase in body size, fecundity and longevity associated with larger host eggs (Arakawa et al. 2004). In addition to needing sufficient yolk protein for food, there is evidence that low molecular-weight host compounds are a necessary component in parasitoid development (Nettles Jr. 1990, Cônsoli et al. 2001). When reared on artificial diet lacking the necessary host factors, Trissolcus basalis (Wollaston) failed to develop beyond pupation, resembling what was observed in the present study
(Volkoff et al. 1992).
The failure of T. japonicus to complete development on pyriproxyfen-induced eggs is particularly concerning. IGR compounds such as pyriproxyfen are touted as being “safe on beneficials” and are frequently recommended in IPM programs (Ishaaya and Horowitz 1992, Medina et al. 2003). While direct acute mortality to beneficial insects is lower in IGR compounds when compared to insect neurotoxins (Delbeke et al.
1997), pyriproxyfen-induced eggs can serve as a dead-end for parasitoid populations, preventing the production of the next generation. If T. japonicus is to succeed as a biological control agent of H. halys, pest management practices must be developed that
94
support their establishment (El-Wakeil et al. 2013, Stanley and Preetha 2016) and take into consideration the indirect and sublethal effects of pesticides on parasitoid populations (Desneux et al. 2007). To this end, further investigations into the indirect effects of host exposure to IGR compounds on parasitoid development, including the ecological impacts to parasitoid populations in the field, are needed.
Figure 5-1. Mean ± SE of cumulative eggs produced per female. “cd” = control dark, reared under 10L:14D. “cl” = control light, reared under 16L:8D. Bars not sharing the same letter are significantly different (least-squares means Tukey’s HSD, P≤0.01).
95
Figure 5-2. Percent diapause termination as a response to pyriproxyfen treatment in female H. halys. Dose response curve is fitted to a two-parameter log-logistic function.
96
Figure 5-3. Dissected H. halys ovaries representing various stages of oocyte formation. (1) Diapause condition with germarium wider than vitellarium. (2) Intermediate stage considered non-diapausing. The developing follicle apparent in the vitellarium and is wider than the germarium. (3) Mature oocytes with chorion. Mature eggs have descended to the lateral oviduct. Ovaries (2) and (3) are considered post-diapause. Photos courtesy of the author.
Table 5-1. H. halys hatchability and T. japonicus viability on pyriproxyfen-induced eggs (Mean ± SE). For H. halys % hatchability, means followed by different letters denote significance at p≤0.01 (least-squares means Tukey’s HSD). Nno hatch describes the number of clutches that had zero hatchability H. halys hatchability T. japonicus viability Treatment Nclutches Nno hatch Mean % Hatchability Nclutches Mean % Viability (μg) 0.5 9 4 19.15 ± 7.94a 9 0 50 24 18 9.60 ± 4.59a 17 0 100 27 22 6.19 ± 2.93a 19 0 Control 17 0 77.71 ± 3.71b 21 71.88 ± 6.91
97
Figure 5-4. Trissolcus japonicus pupa. When reared on pyriproxyfen-induced H. halys eggs, T. japonicus ceased development at the pupal stage, with no pupal to adult emergence observed. Photo courtesy of the author.
98
CHAPTER 6 USE OF PYRIPROXIFEN INTO INCREASE PARASITOID YIELD IN A BIOLOGICAL CONTROL REARING PROGRAM
Hymenopteran egg parasitoids are frequently employed in classical biological control programs against a variety of native and adventive pest species (Waage and
Hassell 1982, Corrêa-Ferreira and Moscardi 1996, Smith 1996, Mills 2005, 2009).
Recently egg parasitoids have been the subject of investigation for the control of the invasive Heteropterans Halyomorpha halys (Stål) (Pentatomidae), Bagrada hilaris
(Burmeister) (Pentatomidae) and Megacopta cribraria (Fabricius) (Plataspidae) (Yang et al. 2009, Leskey, Hamilton, et al. 2012, Gardner et al. 2013, Ruberson et al. 2013,
Sforza et al. 2017). A year-round supply of host eggs used for parasitoid rearing is essential for research, while large quantities of host eggs are essential for augmentative releases. The use of field collected Heteropterans as a source of host eggs is not feasible on a year-round basis due to reproductive diapause, as egg production is terminated during this period (Davey 1997, Saulich and Musolin 2012). Moving specimens into rearing containers with an artificial light source set to a long photoperiod is the most common means for breaking diapause in a laboratory setting; however, this approach has several drawbacks, including the fact that not all species or individuals within a species will respond to photoperiod alone, as well as the lag time required for the artificial photoperiod to achieve its desired effect and induce oviposition. In both research and applied settings reproductive diapause of parasitoid hosts poses a challenge for biological control practitioners.
The invasive M. cribraria is a pest of soybean on the southeastern United States.
Following its initial discovery in the vicinity of Atlanta, Georgia, M. cribaria has spread
99
rapidly, and can be found from Maryland south to Florida, and west to Arkansas and
Louisana (Eger Jr. et al. 2010, Suiter et al. 2010, Liang et al. 2018). Megacopta cribraria has a strong preference for the kudzu vine, Pueraria montana Lour (Merr.) variety lobata (Willd.) (Fabaceae), itself an invasive weed found widely in the southeastern
United States; however, it has been reported to feed on other legumes, including economically important species such as soybean (Zhang et al. 2012, Blount et al.
2015). Crop damage to alternative host legumes is greatest in areas where large patches of kudzu vine are in close proximity to agricultural areas, with high levels of adjacent infestations leading to dramatic losses in soybean yield (Seiter et al. 2013).
The rapid expansion of M. cribaria in its native range, and reports of yield loss to soybean, spurred investigations into potential biological control agents for M. cribaria.
The hymenopteran egg parasitoid Paratelenomous saccharalis (Dodd) (Scelionidae) was selected as a potential candidate due to its known association with the M. cribaria in its native range (Hoshino et al. 2017). Paratelenomous saccharalis was imported into the US for evaluation under quarantine condition; however, in 2013, before it could be approved for release, P.saccharalis was reared from egg masses of M. cribaria collected from the wild in Georgia and in Alabama (Gardner et al. 2013). The unintentional arrival of an adventive biological control agent has been described as
“fortuitous biological control”, in contrast to “classical biological control” which relies on the careful and purposeful introduction of foreign natural enemies (DeBach 1971).
Paratelenomous saccharalis has since been found in Florida and other southeastern states where M. cribaria is present (Medal et al. 2015).
100
Mass rearing of Paratelenomous saccharalis requires concurrent mass rearing of its host M. cribaria such that a consistent and abundant supply of eggs is available for parasitoid rearing and for maintenance of the host colony. Field-collected specimens are often used to augment host colonies, and in the case of M. cribraria the high density and ease of collection makes use of field specimens an efficient means of increasing host colony size, quality, and subsequent egg production. One difficulty in relying on field-augmentation is the decline in M. cribaria egg production exhibited by specimens collected during the fall season when the overwintering adult generation is transitioning into diapause. Reproductive diapause is a crucial part of M. cribraria overwintering success (Golec and Hu 2015). A decline in photoperiod, indicative of the change in seasons, provides the environmental cue for the commencement of diapause, which, at a physiological level, involves the insect corpora alata ceasing production of juvenile hormone (JH), JH being the hormone which promotes vitellogenesis (Davey 1997,
Saulich and Musolin 2012). The termination of egg production by overwintering adults has obvious benefits, as energetic resources can be reserved for winter survival, whereas attempts at reproduction during winter would be futile as nymphs are unable to survive winter conditions.
Pyriproxyfen is a juvenile hormone analog with a reported strong affinity for the binding site of the JH-receptor (Charles et al. 2011). Exogeneous application of pyriproxyfen has been shown to terminate diapause in other species of Heteroptera, resulting in the production of eggs from specimens previously in diapause (Numata and
Hidaka 1984, Amiri et al. 2012, Penca and Hodges 2017). In this work we aim to determine if pyriproxyfen will terminate diapause in M. cribraria, resulting in an increase
101
in egg production. We then aim to see if pyriproxyfen-induced eggs are suitable for rearing of P. saccharalis. Last, the net change in parasitoid productivity will be evaluated to determine if exogenous pyriproxyfen application, alone or in combination with a long day photoperiod, can be used to increase the efficiency of a parasitoid rearing program.
Materials and Methods
Specimen Collection
Specimens of M. cribaria were collected from a kudzu patch in Alachua County,
Florida (29.806183, -82.529999). A surplus of M. cribaria were collected with a sweep net and held in a large plexiglass container for 24 hours before treatment. Two experimental replications occurred, one for each of two field collection dates: collections for experiment 1 were made in mid-August, whereas collections for experiment 2 were made in late-September. The daylength at time of collection was approximately 13.25 hours for the mid-August collection (experiment 1) and approximately 12 hours for the late-September collection (experiment 2). A subsample of 20 females from each collection date were dissected to provide insight into the reproductive status of the population at the time of collection; none of the dissected females were gravid; all presented regressed ovarioles without mature oocytes indicating the population was likely in a state of reproductive diapause.
Treatments and Rearing Conditions
Chemical treatments consisted of technical grade pyriproxyfen (Chem Service,
Westchester, PA.) in acetone solvent applied at two concentrations; a high dose of
1.0% and a low dose of 0.1% pyriproxyfen, along with an acetone control. A 2 ul drop
102
was applied to the ventral abdomen of both male and female M. cribraria and allowed to evaporate before specimens were returned to their rearing container. Each of the four replicates consisted of 20 female and 10 male M. cribaria. Treated insects were held in a 1L clear plastic cup with a vented lid. A piece of filter paper was placed at the bottom of the cup, with a paper towel hung on the edge to provide an ovipositional surface.
Food consisted of 4 cuttings of pigeon pea, Cajanus cajan (L.) (Fabaceae), placed in moist rockwool set inside a shallow plastic dish. Pigeon pea cuttings were changed twice weekly.
The pyriproxyfen treatments and acetone controls were divided into two separate photoperiod regimes. Four replicates from each treatment (1% and 0.1% pyriproxyfen and acetone) were placed into either a long day (LD) (16L:8D) and short day (SD)
(10L:14D) growth chamber held at 25C and 65% RH. Inclusion of long day and short- day treatments allowed for the investigation of the influence of photoperiod on diapause termination, as photoperiod manipulation is frequently used to break/prevent diapause in laboratory reared insects.
The resulting experimental design included 4 replicates, each containing 30 M. cribaria (20F:10M), for each of the 6 treatments (1% pyriproxyfen + 16L:8D, 1% pyriproxyfen + 10L:14D, 0.1% pyriproxyfen + 16L:8D, 0.1% pyriproxyfen + 10L:14D, acetone + 16L:8D, acetone + 10L:14D). The experiment was repeated separately for both collection dates.
Mortality and Egg Data Collection
Cages were checked daily for mortality and egg production. Dead specimens were removed, and their quantity and sex recorded. The number of egg clutches, and
103
number of eggs per clutch, was recorded from each cage daily. Eggs were removed from the cages and set aside for rearing of either M. cribaria or P. saccharalis. For viability studies with M. cribaria eggs were placed in a clear 10-dram plastic vial with a snap-top lid and placed in the long day (16L:10D) photoperiod growth chamber. Total number of emerged 1st instar M. cribaria was recorded over a 28-day period for each clutch.
Parasitoid Exposure
Clutches of 10-30 fresh (less than 24 hrs old) eggs of M. cribaria were fixed to a
1 cm by 4 cm length of card-stock paper using a small piece of double-sided tape. The source container, date and number of eggs was recorded on the index card label inside the tube. An approximately 1 sq. cm piece of tissue paper was soaked with honey such that the fibers contained honey, but no large globules of honey exuded from the tissue paper. The honey-soaked tissue paper was added to each vial as a parasitoid food source. Paratelenomous saccharalis were sourced from a lab colony held at the Florida
Department of Agriculture and Consumer Services, Division of Plant Industry (FDACS-
DPI) in Gainesville, Florida. The FDACS-DPI P. saccharalis colony originated from specimens collected at the same kudzu patch where the M. cribaria used in this study were sourced from. Female P. saccharalis were held in a container with male P. saccharalis for at least 48 hours to allow opportunity for mating. Approximately 2-4 presumably mated female P. saccharalis were added to the vial containing eggs of M. cribaria. Immediately following exposure to eggs the parasitoids were observed and oviposition behavior was recorded. Parasitoids were removed from the exposure tubes after 5-7 days, allowing sufficient time for oviposition.
104
Parasitoid-exposed eggs were held in a growth chamber and monitored for parasitoid emergence for at least 28 days. The time to emergence for P. saccharalis reared in the FDACS-DPI source colony ranged from approximately 11 to 14 days. After at least 28 days the sex and number of emerged parasitoids was recorded. Additionally, the color of unhatched eggs was recorded as parasitized eggs generally turn from their usual salmon-pink to a gray color (Gardner et al. 2013). Eggs with a darkened appearance were dissected under a microscope and the presence/absence of non- emerged parasitoids was recorded. All eggs used in the parasitoid viability studies were sourced from the experiment 2 treatments.
Data Analysis
Experiment 1 and experiment 2 were separated in all analysis due to significant differences in control mortality between both experiments. Only eggs from experiment 2 were used in host and parasitoid viability studies. The effect of treatment on female M. cribaria mortality was analyzed via Kaplan Meier survivorship curves. This approach allows us to examine treatment differences over the course of the experiment, as opposed to comparing mortality at the end of study alone. For long running studies of mortality, the survival rate of all treatments will converge at zero, thus treatment effects may go unnoticed if only data from the end of the study are evaluated. Egg production between treatments was compared via a generalized linear model with daylength and chemical treatment as explanatory variables. The negative binomial distributions provided the best model fit for egg production analysis. Viability of M. cribaria from treatment and control eggs was modeled via logistic regression with a binary outcome
(viable/non-viable) assigned to each egg, with a weight term assigned based on clutch
105
size. Viability of P. saccharalis, recorded as percent emergence from exposed eggs, was modeled using the same approach as used for viability of M. cribaria, with percent viability recorded for each clutch, and a weight term assigned based on clutch size.
Combining the estimates for egg production and parasitoid viability for each treatment gives us a sensible estimate of total parasitoid production. However, this simple estimate does not provide us with a resulting probability distribution for parasitoid production for each treatment; making it difficult to conduct contrasts-type tests and determine the overall effective of each treatment on variable of interest, parasitoid productivity. Our solution was to conduct Monte-Carlo simulations for each treatment based on the probability distribution of the models for both egg production and parasitoid viability. Each simulation produced a value for number of eggs produced and probability of parasitoid emergence, with the produced values drawn from the modeled distribution of each treatment. We then took the product of these two values, which is equivalent to the number of parasitoids produced from that simulation (i.e. yield).
Simulations were repeated 2000 times for each treatment. The yield estimates for the
2000 simulations provide an accurate estimate of the distribution of parasitoid yield expected for each treatment and can be compared via a generalized linear model to determine treatment effects on yield. Post-hoc tests for all models were conducted using least squares with Tukey’s adjustment for multiple comparisons.
Results
Mortality
Mortality of M. cribaria occurred more rapidly in experiment 1 (median time to death: 17 days; 95% C.I.: 14-19 days) than in experiment 2 (median time to death: 39
106
days; 95% C.I.: 39-40 days) (Figure 6-1), suggesting a different age structure of the source population for each experiment. Kaplan Meier survival analysis provided a representation of treatment differences between experiments. In experiment 1 only the high-dose pyriproxyfen treatment experienced mortality which differed from that of the control treatment (Figure 6-2). In experiment 2 both photoperiod and pyriproxyfen dose had a significant impact on mortality when compared to the short daylength, acetone control (Figure 6-3). Pairwise comparisons of mortality from Kaplan Meier survival analysis is provided for experiment 1 and experiment 2 in tables 6-1 and 6-2.
Egg Production
In experiment 1, under long day photoperiod (16L:8D) egg production was significantly higher in pyriproxyfen treatments than in the short day (10L:14D) acetone control. In both experiments the high and low doses of pyriproxyfen induced egg production. Daylength was shown to be able to terminate diapause, with a photoperiod of 16L:8D leading to increased egg production in all treatments in both experiments, including the acetone treated controls. Under short day conditions (10L:14D) the acetone-treated controls did not break diapause and no egg production was observed.
Significantly higher egg production was observed in experiment 2 than experiment 1, primary as a result of the higher mortality seen in experiment 1. Treatment differences, as estimated from a negative binomial generalized linear model, along with contrasts determined through least squares means, are presented in figures 6-4 (experiment 1) and 6-5 (experiment 2).
107
Host and Parasitoid Viability
Viability of M. cribaria eggs was extremely low (<3%) in all treatments, except for the 16L:8D acetone control (Table 6-3), suggesting pyriproxyfen treatment resulted in eggs with reduced viability. Overall emergence of P. saccharalis was lower than what has been observed in the FDACS-DPI stock colony (N. Goltz, personal communication)
(Table 6-4). The long day control and the long day, 0.1% pyriproxyfen treatment had the highest rate of parasitoid viability. A lower photoperiod and higher dose of pyriproxyfen both were implicated in decreased parasitoid viability.
Parasitoid Yield
Both photoperiod and chemical treatment influenced parasitoid yield, with a significant interaction occurring between factors (negative binomial GLM, p<0.0001). All treatments were significantly different at an alpha level of 0.05 (least squares mean with
Tukey’s method for multiple comparisons) (Table 6-5). Under long day (16L:8D) conditions, both the low dose and high dose treatments produced significantly more parasitoids than the long day acetone control. The lower dose treatment of 0.1% pyriproxyfen producing the most parasitoids, primarily due to the significantly higher parasitoid mortality observed in the high dose treatment (Figure 6-6).
Discussion
While juvenile hormone analogs have primarily been used to control pests through disruption of the insect lifecycle, their utility in manipulating insect behavior for other purposes has not gone unnoticed. The juvenile hormone analog methoprene has been shown to increase silk production in Bombyx mori by up to 24% (Murakoshi et al.
1972, Miranda et al. 2002). In the commercial production of baculovirus the exogenous
108
application of a juvenile hormone analog resulted in a 2.7 to 2.9-fold increase in baculovirus yield, largely due to the supernumerary molt of the lepidopteran host (Lasa et al. 2007). In previous studies with Halyomorpha halys, pyriproxyfen was used to induce egg production, resulting in eggs with extremely low viability. When pyriproxyfen- induced eggs were exposed to the parasitoid T. japonicus, the egg parasitoid readily oviposited in the host eggs; however, no viable parasitoids were reared from parasitized eggs, limiting the utility of this method for T. japonicus rearing (Penca and Hodges
2017).
While the significance of this work is largely in its application for parasitoid rearing, there is potential for utilization in the field. The diapause terminating effect of pyriproxyfen can produce a significant reduction in overwintering survival when a diapausing population is exposed to a sufficient dose. The induction of oviposition in a field situation is of minor concern, as our results indicate viability of these eggs to be extremely low, and winter survival of nymphal stages of M. cribaria (and other
Heteroptera) is severely limited. Another possible technique is in “prepping” a field for the release of an egg parasitoid. Applications of pyriproxyfen can induce mass oviposition in the treatment area. Augmentative releases in this area, if timed to correspond with the treatment effect, may benefit the released parasitoids by increasing the likelihood they will find suitable hosts.
Further research is needed to identify the optimum dose of pyriproxyfen to maximize egg production and parasitoid viability. A potential split-application approach may be warranted, as this may allow continuous activation of the JH receptor during the study period. While pyriproxyfen is noted as having a strong affinity for the JH binding
109
site, the metabolization of pyriproxyfen over the course of the study may have reduced the strength of the signal promoting vitellogenesis (Charles et al. 2011). We speculate that maintaining a consistent level of JH-receptor activation by the JH analog over the study period may improve the quality of the resulting eggs.
The potential of this approach for enhancing production of egg parasitoids is substantial; however, certain conditions must be met. First, preliminary, species-specific work must be done to optimize the dose and treatment schedule of the host, and to determine if the parasitoid is able to successfully develop in the pyriproxyfen-induced eggs. Secondly, the method described here is most useful when specimens are readily field collected and have already undergone reproductive diapause. Exogenous IGR applications may increase egg production in non-diapausing specimens, though the benefits should be weighed against any effects on host mortality and egg quality; the described technique may not be worthwhile for host colonies that are already producing eggs.
The results of this study suggest pyriproxyfen in combination with an increased photoperiod can be used to support mass rearing of P. saccharalis. A long day photoperiod, coupled with 0.1% pyriproxyfen resulted in a 1.64-fold increase in parasitoid yield, whereas 1% pyriproxyfen resulted in a slightly lower 1.45-fold increase in parasitoid yield. If parasitoid viability in the 1% pyriproxyfen + 16L:8D treatment was increased such that it was equivalent to the parasitoid viability seen in long day control treatment, an estimated 3.73-fold increase in parasitoid yield would be obtained.
Increasing the number of parasitoids produced per host unit can significantly lower the
110
costs of a parasitoid rearing program, particularly when the goal is mass rearing parasitoids for augmentative or inoculative biological control programs
Figure 6-1. Survival probability for experiment 1 (red) and experiment 2 (blue) with all treatments combined.
111
Figure 6-2. Survivorship curves of M. cribraria from experiment 1, separated by photoperiod, for the three chemical treatments (Control = Acetone, High = 1.0% Pyriproxyfen, Low=0.1% Pyriproxyfen). Specimens were collected in mid-August when the natural photoperiod was approximately 13.25L: 10.75D.
112
v
Figure 6-3. Survivorship curves of M. cribraria from experiment 2, separated by photoperiod, for the three chemical treatments (Control = Acetone, High = 1.0% Pyriproxyfen , Low=0.1% Pyriproxyfen). Specimens were collected in late September, when the natural photoperiod was approximately 12L:12D.
Table 6-1. Pairwise comparisons of mortality using the Log-Rank test. Experiment 1.
Control Control 1% Pyr. 1% Pyr. 0.1% Pyr.
14L:10D 16L:8D 14L:10D 16L:8D 14L:10D Control 0.3013 16L:8D 1% Pyr. 0.0117* .0013* 14L:10D 1% Pyr. 0.0282* .0013* 0.0760 16L:8D 0.1% Pyr. 0.8930 0.2858 0.0326* 0.0578 14L:10D 0.1% Pyr. 0.4665 0.7650 0.0013* 0.0032* 0.3964 16L:8D Asterisks (*) indicate significance at p<0.05
113
Table 6-2. Pairwise comparisons of mortality using the Log-Rank test. Experiment 2.
Control Control 1% Pyr. 1% Pyr. 0.1% Pyr. 14L:10D 16L:8D 14L:10D 16L:8D 14L:10D Control 16L:8D <0.0001* 1% Pyr. 14L:10D <0.0001* <0.0001* 1% Pyr. 16L:8D <0.0001* <0.0001* 0.38 0.1% Pyr. 14L:10D <0.0001* <0.0001* <0.0001* <0.0001* 0.1% Pyr. 16L:8D <0.0001* 0.38 <0.0001* <0.0001* <0.0001* Asterisks (*) indicate significance at p<0.05
Figure 6-4. Egg production for experiment 1, separated by photoperiod and pyriproxyfen treatment. Bars with the same letter are not significant different (Least Squares, Tukey’s method of multiple comparisons, p<0.05).
114
Figure 6-5. Egg production for experiment 2, separated by photoperiod and pyriproxyfen treatment. Bars with the same letter are not significant different (Least Squares, Tukey’s method of multiple comparisons, p<0.05).
Table 6-3. Estimated M. cribaria hatch probability, as determined by binomial regression. Results are from eggs produced from experiment 2. Rows with the same letters are not significantly different at p<0.05 (Least Squares Means). Dose Photoperiod Hatch Prob. ± SE
Control 10L:14D No Eggs Produced
Control 16L:8D 0.280±0.019a
0.1% pyriproxyfen 10L:14D 0.005±0.005b
0.1% pyriproxyfen 16L:8D 0.022±0.006b
1% pyriproxyfen 10L:14D 0.000±0.000b
1% pyriproxyfen 16L:8D 0.015±0.004b
115
Table 6-4. Estimated P. saccharalis hatch probability, as determined by negative binomial regression. Results are from eggs produced from experiment 2. Rows with the same letters are not significantly different at p<0.05 (Least Squares Means).
Dose Photoperiod Parasitoid Hatch Prob. ± SE
Control 10L:14D NA
Control 16L:8D 0.347±0.034a
0.1% pyriproxyfen 10L:14D 0.090±0.014b
0.1% pyriproxyfen 16L:8D 0.288±0.021a
1% pyriproxyfen 10L:14D 0.109±0.015bc
1% pyriproxyfen 16L:8D 0.174±0.017c
Table 6-5. Estimated parasitoid yield, as determined by generalized linear regression. Rows without common letters are significantly different at p<0.05 (Least Squares Means).
Dose Photoperiod Estimated Parasitoid Yield ± SE
Control 10L:14D NA
Control 16L:8D 101.24±0.82a
0.1% pyriproxyfen 10L:14D 21.92±0.19b
0.1% pyriproxyfen 16L:8D 169.28±1.36c
1% pyriproxyfen 10L:14D 40.18±0.33d
1% pyriproxyfen 16L:8D 146.77±1.18e Estimated parasitoid yield is given per replicate (20 Females : 10 Males).
116
Figure 6-6. Display of density estimates produced from simulated (n=2000) estimates of parasitoid yield. The area under all curves sum to 1.
117
CHAPTER 7 CONCLUSIONS
This work provides the first insights into the catfacing pest complex in subtropical
Florida peaches. Based on trap and visual surveys we can conclude, with a high degree of certainty, that Euschistus servus is the primary catfacing pest in Florida peaches. We have also found evidence that Leptoglossus phyllopus may be a key member of the catfacing pest complex, but that yellow pyramid traps are not an effective method for monitoring L. phyllopus populations. The second most abundant species in the trap survey was Thyanta perditor, a stink bug species not found north of Florida. This species was not observed feeding on peaches during the visual survey, and thus its importance in the catfacing pest complex may not align with its observed abundance.
Lastly, our survey has provided the first evidence that Halyomorpha halys is occurring in
Florida peach orchards, with observations occurring over multiple years in Polk and
Lake County sites, and with reproduction observed in Lake County. This has resulted in
H. halys being declared locally established in Lake County, Florida by the Florida
Department of Agriculture and Consumer Services, Division of Plant Industry.
Comparisons of beta diversity suggested Lake and St. Lucie county have catfacing pest complexes with significantly different species compositions, and that the species composition in Polk was intermediate between the other two regions. While there is evidence that moderate differences in species composition are present between the study regions, management tactics and local-level landscape differences may be responsible for most of the regional differences in pentatomid species composition. The expansion of this work to include additional study sites across the study regions is
118
needed to determine if the observed differences in species composition are a result of regional characteristics or site-specific variation.
Damage surveys suggest catfacing injury in Florida can vary over a considerable range, from as low as 1% to over 12%. There was a statistically significant change in E. servus abundance between years that was consistent between sites and regions. A significant relationship between injury-at-harvest and total E. servus trap capture during the growing season was observed. From this relationship we were able to produce an estimate of injury based on trap capture that could be used in the development of economic injury levels. The resulting economic injury levels varied based on control costs, crop prices, and efficacy of control (K in the EIL equation). When input values were set to reflect reported values from growers and from University-produced enterprise budgets, the likely value for an economic injury level was centered around
2.5 to 7.5 stink bugs per trap when traps are deployed over a 14-day period. Our work suggests that for stink bugs, values of K are generally low, and thus a higher EIL may be justified.
A trend in population increase was detected that held consistent between sites and years and was associated with the phenological stage of fruit development. This trend allowed for the generation of economic thresholds based on the established EIL value which accurately anticipate the expected change in population density during the following 14-day period. Thresholds were lowest at shuck split and the first half of fruit sizing, as these were the stages where E. servus populations were expected to increase most rapidly. Thresholds increased for trapping periods closest to harvest,
119
reflecting a trend in which E. servus populations stabilized, or in some cases declined, in the period just prior to harvest.
Spatial analysis of stink bug injury indicated a random distribution throughout the orchard. Development of sequential sampling plans based on hypothetical thresholds were able to improve precision over a simple random sample of 10 units; however, this effect was negligible when true mean injury differed from the threshold by more than 1- unit percent. The average sample number for the sequential sampling plans had an inverse relationship with the distance from threshold, with sequential sampling plans requiring fewer than 10 samples when the distance from threshold was great than 1-unit percent. As such we can conclude that, independent of the threshold used, the sequential sampling plans provide a gain in accuracy when the distance from threshold is less than 1-unit percent and reduce the required sample size when the distance from threshold is greater than 1-unit percent. Beyond the direct applications our developed sequential sampling plans, the methods used provide a novel approach for the evaluation of sequential sampling plans independent of the critical threshold.
Our result on pyriproxyfen suggest this potent juvenile hormone analog terminates diapause, leading to a dramatic increase in egg production in both species of invasive Heteroptera tested (Halyomorpha halys and Megacopta cribaria). Termination of diapause was correlated with a significant reduction in survival during the experimental period. These results point to the potential use of pyriproxyfen for control through diapause-interference. Such control would be particularly useful for systems where the overwintering pest population is a significant contributor to crop injury.
120
We were also able to demonstrate the use of pyriproxyfen to increase egg parasitoid yield in a biological control rearing program. While Trissolcus japonicus was not able develop in pyriproxyfen-induced eggs, Paratelenomous saccharalis was able to develop, with only a modest reduction in parasitoid viability. The resulting 1.64-fold increase in parasitoid production is promising, and future work to enhance this technique may result in ever greater gains in rearing efficiency. Our results with T. japonicus suggest use of pyriproxyfen for parasitoid rearing should be evaluated on a species by species basis, as T. japonicus was not successfully reared from pyriproxyfen-induced eggs.
121
LIST OF REFERENCES
Akotsen-Mensah, C., R. R. Balusu, J. Anikwe, and H. Y. Fadamiro. 2017. Evaluating potential trap crops for managing leaffooted (Hemiptera: Coreidae) and phytophagous stink bug (Hemiptera: Pentatomidae) species in peaches. Agric. For. Entomol. 19: 332-340.
Aldrich, J. R., M. P. Hoffmann, J. P. Kochansky, W. R. Lusby, J. E. Eger, and J. A. Payne. 1991. Identification and attractiveness of a major pheromone component for nearctic Euschistus spp. stink bugs (Heteroptera: Pentatomidae). Environ. Entomol. 20: 477–483.
Amiri, A., A. R. Bandani, and A. Darvishzadeh. 2012. Effects of the insect growth regulators methoxyfenozide and pyriproxyfen on adult diapause in sunn pest Eurygaster integriceps (Hemiptera: Scutelleridae). J. Agr. Sci. Tech. 14: 1205– 1218.
Arakawa, R., M. Miura, and M. Fujita. 2004. Effects of host species on the body size, fecundity, and longevity of Trissolcus mitsukurii (Hymenoptera: Scelionidae), a solitary egg parasitoid of stink bugs. Appl. Entomol. Zool. 39: 177–181.
Atanasso, A., P. W. Shearer, G. Hamilton, and D. Polk. 2002. Development and implementation of a reduced risk peach arthropod management program in New Jersey. J Econ Entomol. 95: 804-812.
Bajgar, A., D. Dolezel, and M. Hodkova. 2013. Endocrine regulation of non-circadian behavior of circadian genes in insect gut. J. Insect Physiol. 59: 881–886.
Bajgar, A., M. Jindra, and D. Dolezel. 2013. Autonomous regulation of the insect gut by circadian genes acting downstream of juvenile hormone signaling. Proc. Natl. Acad. Sci. U. S. A. 110: 4416–4421.
Barzman, M., P. Bàrberi, A. N. E. Birch, P. Boonekamp, S. Dachbrodt-Saaydeh, B. Graf, B. Hommel, J. E. Jensen, J. Kiss, P. Kudsk, J. R. Lamichhane, A. Messéan, A.-C. Moonen, A. Ratnadass, P. Ricci, J.-L. Sarah, and M. Sattin. 2015. Eight principles of integrated pest management. Agron. Sustain. Dev. 35: 1199–1215.
Binns, M. R., and J. P. Nyrop. 1992. Sampling Insect Populations for the Purpose of IPM Decision Making. Annu. Rev. Entomol. 37: 427–453.
Binns, M. R., J. P. Nyrop, and W. van der Werf. 2000. Sampling and monitoring in crop protection: the theoretical basis for developing practical decision guides. CABI, Wallingford, Oxon, UK.
122
Blaauw, B. R., P. M. Brennan, D. Lockwood, and D. Rictchie. 2018. Southeastern peach, nectarine, and plum pest management and culture guide. University of Georgia Cooperative Extension, Athens, GA.
Blaauw, B. R., D. Polk, and A. L. Nielsen. 2015. IPM-CPR for peaches: incorporating behaviorally-based methods to manage Halyomorpha halys and key pests in peach. Pest Manag. Sci. 71: 1513-1522.
Blount, J. L., G. D. Buntin, and A. N. Sparks. 2015. Host preference of Megacopta cribraria (Hemiptera: Plataspidae) on selected edible beans and soybean. J. Econ. Entomol. 108: 1094–1105.
Bownes, M. 1989. The roles of juvenile hormone, ecdysone and the ovary in the control of Drosophila vitellogenesis. J. Insect Physiol. 35: 409–413.
Bundy, C. S., and R. M. McPherson. 2000. Dynamics and seasonal abundance of stink bugs (Heteroptera: Pentatomidae) in a cotton–soybean ecosystem. J. Econ. Entomol. 93: 697–706.
Carleton, R. D., S. B. Heard, and P. J. Silk. 2013. A simulation approach to assessing sampling strategies for insect pests: an example with the balsam gall midge. PLoS One. 8: e82618.
Chandler, S. C. 1955. Biological studies of peach catfacing insects in Illinois. J. Econ. Entomol. 48: 473–475.
Chao, A. 1984. Nonparametric estimation the number of classes in a population. Scand. J. Stat. 11: 265-270.
Chao, A., C.H. Chiu, and L. Jost. 2014. Unifying species diversity, phylogenetic diversity, functional diversity, and related similarity and differentiation measures through hill numbers. Annu. Rev. Ecol. Evol. Syst. 45: 297-324.
Charles, J.-P., T. Iwema, V. C. Epa, K. Takaki, J. Rynes, and M. Jindra. 2011. Ligand-binding properties of a juvenile hormone receptor, methoprene-tolerant. Proc. Natl. Acad. Sci. U. S. A. 108: 21128–21133.
Cônsoli, F. ., E. Conti, L. . Dangott, and S. . Vinson. 2001. In vitro culture of the teratocytes of Trissolcus basalis (Hymenoptera, Scelionidae) and their requirements for host-derived components. Biol. Control. 22: 176–184.
Corrêa-Ferreira, B. S., and F. Moscardi. 1996. Biological control of soybean stink bugs by inoculative releases of Trissolcus basalis. Entomol. Exp. Appl. 79: 1–7.
Cottrell, T. E. 2001. Improved trap capture of Euschistus servus and Euschistus tristigmus (Hemiptera: Pentatomidae) in pecan orchards. Florida Entomol. 84: 731- 732.
123
Cottrell, T. E., and D. L. Horton. 2011. Trap capture of brown and dusky stink bugs (Hemiptera: Pentatomidae) as affected by pheromone dosage in dispensers and dispenser source. J. Entomol. Sci. 46: 135–147.
Cullen, E. M., and F. G. Zalom. 2000. Phenology-based field monitoring for consperse stink bug (Hemiptera: Pentatomidae) in processing tomatoes. Environ. Entomol. 29: 560–567.
Davey, K. G. 1997. Hormonal controls on reproduction in female heteroptera. Arch. Insect Biochem. Physiol. 35: 443–453.
DeBach, P. 1971. Fortuitous biological control from ecesis of natural enemies, pp. 293- 307. In Entomological essays to commemorate the retirement of professor K. Yamsumatsu. Hokuryucan Publ. Co, Tokyo, Japan.
Delbeke, F., P. Vercruysse, L. Tirry, P. Clercq, and D. Degheele. 1997. Toxicity of diflubenzuron, pyriproxyfen, imidacloprid and diafenthiuron to the predatory bug Orius laevigatus (Heteroptera: Anthocoridae). Entomophaga. 42: 349–358.
Denlinger, D. L. 2002. Regulation of diapause. Annu. Rev. Entomol. 47: 93–122.
Desneux, N., A. Decourtye, and J.-M. Delpuech. 2007. The sublethal effects of pesticides on beneficial arthropods. Annu. Rev. Entomol. 52: 81–106.
Duan, J. J., R. J. Prokopy, C. M. Yin, C. Bergweiler, and H. Oouchi. 1995. Effects of pyriproxyfen on ovarian development and fecundity of Rhagoletis pomonella flies. Entomol. Exp. Appl. 77: 17–21.
Eger Jr., J. E., L. M. Ames, D. R. Suiter, T. M. Jenkins, D. A. Rider, and S. E. Halbert. 2010. Occurrence of the Old World bug Megacopta cribraria (Fabricius) (Heteroptera: Plataspidae) in Georgia: a serious home invader and potential legume pest. Insecta mundi. 0121: 1–11.
Ehler, L. E. 2006. Perspective Integrated pest management (IPM): definition, historical development and implementation, and the other IPM. Pest Manag. Sci. 62: 787– 789.
El-Wakeil, N., N. Gaafar, A. Sallam, and C. Volkmar. 2013. Side effects of insecticides on natural enemies and possibility of their integration in plant protection strategies, pp. 3–56. In Trdan, S. (ed.), Insecticides: development if safer and more effective tchnologies. InTech, Rjeka, Croatia.
Evans, H. E. 1957. Ethological studies on digger wasps of the genus Astata (Hymenoptera, Sphecidae). J. New York Entomol. Soc. 65: 159–185.
FAO. 2006. ISPM (International standards for phytosanitary measures) No 5. Glossary of phytosanitary terms. IPPC, Rome, Italy.
124
Ferguson, J. F., J. X. Chaparro, D. M. O’Malley, and L. Harrison. 2006. Options for subtropical peach production in Florida. Proc. Fla. State Hort. Soc. 119: 29–31.
Foshee, W. G., R. T. Boozer, E. K. Blythe, D. L. Horton, and J. Burkett. 2008. Management of plum curculio and catfacing insects on peaches in central Alabama: standard crop stage–based vs. integrated pest management–based approaches. Int. J. Fruit Sci. 8: 188–199.
Fowler, G. W., and A. M. Lynch. 1987. Sampling plans in insect pest management based on wald’s sequential probability ratio test. Environ. Entomol. 16: 345–354.
Frank, D. L., and A. R. Biggs. 2014. Control of oriental fruit moth and other insect pests in peach, 2013. Arthropod Manag. Tests. 39.
Frank, J. H., and E. D. McCoy. 1995. Precinctive insect species in florida. Florida Entomol. 78: 21-35.
Gardner, W. A., J. L. Blount, J. R. Golec, W. A. Jones, X. P. Hu, E. J. Talamas, R. M. Evans, X. Dong, C. H. Ray, G. D. Buntin, N. M. Gerardo, and J. Couret. 2013. Discovery of Paratelenomus saccharalis (Dodd) (Hymenoptera: Platygastridae), an egg parasitoid of Megacopta cribraria F. (Hemiptera: Plataspidae) in its expanded North American range. J. Entomol. Sci. 48: 355–359.
Gent, D. H., E. De Wolf, and S. J. Pethybridge. 2011. Perceptions of risk, risk aversion, and barriers to adoption of decision support systems and ipm perceptions of risk, risk aversion, and barriers to adoption of decision support systems and integrated pest management: an introduction. Phytopathology. 101: 640–643.
Golec, J. R., and X. P. Hu. 2015. Preoverwintering copulation and female ratio bias: life history characteristics contributing to the invasiveness and rapid spread of megacopta cribraria (heteroptera: plataspidae). Environ. Entomol. 44: 411–417.
Hahn, D. A., and D. L. Denlinger. 2011. Energetics of insect diapause. Annu. Rev. Entomol. 56: 103–121.
Hahn, N. G., C. Rodriguez-Saona, and G. C. Hamilton. 2017. Characterizing the spatial distribution of brown marmorated stink bug, Halyomorpha halys Stål (Hemiptera: Pentatomidae), populations in peach orchards. PLoS One. 12.
Hatakoshi, M., Y. Shono, H. Yamamato, and M. Hirano. 1991. Effects of the juvenile hormone analog pyriproxyfen, on Myzus persicae and Unaspis yanonensis. Appl. Entomol. Zool. 26: 412–414.
Haye, T., T. Gariepy, K. Hoelmer, J.-P. Rossi, J.-C. Streito, X. Tassus, and N. Desneux. 2015. Range expansion of the invasive brown marmorated stinkbug, Halyomorpha halys: an increasing threat to field, fruit and vegetable crops worldwide. J. Pest Sci. 88: 665–673. 125
Hedstrom, C., D. Lowenstein, H. Andrews, B. Bai, and N. Wiman. 2017. Pentatomid host suitability and the discovery of introduced populations of Trissolcus japonicus in Oregon. J. Pest Sci. 90: 1169-1179.
Hillebrand, H. 2004. On the generality of the latitudinal diversity gradient. Am. Nat. 163: 192-211.
Hodek, I. 1971. Termination of adult diapause in Pyrrhocoris apterus (Heteroptera: Pyrrhocoridae) in the field. Entomol. Exp. Appl. 14: 212–222.
Hoebeke, E. R., and M. E. Carter. 2003. Halyomorpha halys (Stål) (Heteroptera: Pentatomidae): a polyphagous plant pest from Asia newly detected in North America. Proc. Entomol. Soc. Washingt. 105: 225–237.
Ver Hoef, J. M., and P. L. Boveng. 2007. Quasi-poisson vs. negative binomial regression: How should we model overdispersed count data? Ecology. 88: 2766- 2772.
Hoffman, M. L., L. T. Wilson, and F. G. Zalom. 1987. Control of stink bugs in tomatoes. Calif. Agric. 41: 4–6.
Hogmire, H. W., and T. C. Leskey. 2006. An improved trap for monitoring stink bugs (Heteroptera : Pentatomidae) in apple and peach orchards. J. Entomol. Sci. 41: 9– 21.
Horton, D., P. Brannen, B. Bellinger, D. Lockwood, and D. Ritchie. 2016. Southeastern peach, nectarine and plum pest management and culture guide. Univ. Georg. Coop. Ext. Bull. 1171.
Hoshino, K., T. Adati, D. M. Olson, and K. Takasu. 2017. Seasonal occurrence and interspecific interactions of egg parasitoids of megacopta cribraria (heteroptera: Plataspidae) in Japan. Environ. Entomol. 46: 487-493.
Hoyt, S. C., and L. E. Caltagirone. 2012. The developing programs of integrated control of pests of apples in washington and peaches in california. P. 395-421.In Biological Control. Plenum Press, New York. 511 pp.
Hsiao, C., and T. H. Hsiao. 1969. Insect hormones: their effects on diapause and development of hymenoptera. Life Sci. 8: 767–774.
Hsieh, T. C., K. H. Ma, and A. Chao. 2016. iNEXT: an R package for rarefaction and extrapolation of species diversity (Hill numbers). Methods Ecol. Evol. 7: 1451-1456.
Inkley, D. B. 2012. Characteristics of home invasion by the brown marmorated stink bug (Hemiptera: Pentatomidae). J. Entomol. Sci. 47: 125–130.
126
Ishaaya, I., and A. R. Horowitz. 1992. Novel phenoxy juvenile hormone analog (pyriproxyfen) suppresses embryogenesis and adult emergence of sweetpotato whitefly (Homoptera: Aleyrodidae). J. Econ. Entomol. 85: 2113–2117.
Iwao, S. 1975. A new method of sequential sampling to classify populations relative to a critical density. Res. Popul. Ecol. (Kyoto). 16: 281–288.
Johnson, D. T., B. A. Lewis, B. D. McCraw, B. Carroll, B. Jervis, K. Striegler, B. Boozer, P. Mulder, W. G. Foshee, J. McVay, and R. F. Mizell. 2002. Development and implementation of a peach integrated pest management program in the southern USA. In Acta Hortic. 592: 1-8.
Joseph, S. V, J. W. Stallings, T. C. Leskey, G. Krawczyk, D. Polk, B. Butler, and J. C. Bergh. 2014. Spatial distribution of brown marmorated stink bug (Hemiptera: Pentatomidae) injury at harvest in mid-Atlantic apple orchards. J. Econ. Entomol. 107: 1839–48.
Jost, L. 2006. Entropy and diversity. Oikos. 113: 363-375.
Khrimian, A., P. W. Shearer, A. Zhang, G. C. Hamilton, and J. R. Aldrich. 2008. Field trapping of the invasive brown marmorated stink bug, Halyomorpha halys, with geometric isomers of methyl 2,4,6-decatrienoate. J. Agric. Food Chem. 56: 197–203.
Killian, J. C., and J. R. Meyer. 1984. Effect of orchard weed management on catfacing damage to peaches in North Carolina. J. Econ. Entomol. 77: 1596–1600.
Kotaki, T., and S. Yagi. 1989. Hormonal control of adult diapause in the brown-winged green bug, Plautia stali Scott (Heteroptera : Pentatomidae). Appl. Entomol. Zool. 24: 42–51.
Krispyn, J. W., and J. W. Todd. 1982. The red imported fire ant as a predator of the southern green stink bug on soybean in Georgia. J. Georg. Entomol. Soc. 17: 19– 26.
Lasa, R., P. Caballero, and T. Williams. 2007. Juvenile hormone analogs greatly increase the production of a nucleopolyhedrovirus. Biol. Control. 41: 389–396.
Laumann, R. A., M. C. B. Moraes, A. Khrimian, and M. Borges. 2011. Field capture of Thyanta perditor with pheromone-baited traps. Pesqui. Agropecuária Bras. 46: 113–119.
127
Leskey, T. C., G. C. Hamilton, A. L. Nielsen, D. F. Polk, C. Rodriguez-Saona, J. Christopher Bergh, D. Ames Herbert, T. P. Kuhar, D. Pfeiffer, G. P. Dively, C. R. R. Hooks, M. J. Raupp, P. M. Shrewsbury, G. Krawczyk, P. W. Shearer, J. Whalen, C. Koplinka-Loehr, E. Myers, D. Inkley, K. A. Hoelmer, D. H. Lee, and S. E. Wright. 2012. Pest status of the brown marmorated stink bug, Halyomorpha halys in the USA. Outlooks Pest Manag. 23: 218–226.
Leskey, T. C., and H. W. Hogmire. 2005. Monitoring stink bugs (Hemiptera: Pentatomidae) in mid-Atlantic apple and peach orchards. J. Econ. Entomol. 98: 143–153.
Leskey, T. C., B. D. Short, B. R. Butler, and S. E. Wright. 2012. Impact of the invasive brown marmorated stink bug, Halyomorpha halys (Stål), in mid-Atlantic tree fruit orchards in the United States: case studies of commercial management. Psyche A J. Entomol. 2012: 1–14.
Liang, W., L. Tran, R. Washington-Allen, G. Wiggins, S. Stewart, J. Vogt, and J. Grant. 2018. Predicting the potential invasion of kudzu bug, Megacopta cribraria (Heteroptera: Plataspidae), in North and South America and determining its climatic preference. Biol. Invasions.
McPherson, J. E., and R. M. McPherson. 2000. Stink bugs of economic importance in America north of Mexico. CRC Press, Boca Raton.
McPherson, R. M., J. R. Pitis, L. D. Newsom, J. B. Chapin, and D. C. Herzog. 1982. Incidence of tachinid parasitism of several stink bug (Heteroptera: Pentatomidae) species associated with soybean. J. Econ. Entomol. 75: 783–786.
Medal, J., A. S. Cruz, K. Williams, S. Fraser, D. Wolaver, T. Smith, and B. J. Davis. 2015. First record of Paratelenomus saccharalis (Hymenoptera: Platygastridae) on kudzu bug Megacopta cribraria (Heteroptera: Plataspidae) in Florida. Florida Entomol. 98: 1250–1251.
Medal, J., T. Smith, and A. S. Cruz. 2013. Biology of the brown marmorated stink bug Halyomorpha halys (Heteroptera: Pentatomidae) in the laboratory. Florida Entomol. 96: 1209–1212.
Medina, P., G. Smagghe, F. Budia, L. Tirry, and E. Viñuela. 2003. Toxicity and absorption of azadirachtin, diflubenzuron, pyriproxyfen, and tebufenozide after topical application in predatory larvae of Chrysoperla carnea (Neuroptera: Chrysopidae). Environ. Entomol. 32: 196–203.
Michaud, J. P., and A. K. Grant. 2003. IPM-compatibility of foliar insecticides for citrus: Indices derived from toxicity to beneficial insects from four orders. J. Insect Sci. 3.
Mills, N. 2005. Selecting effective parasitoids for biological control introductions: Codling moth as a case study. Biol. Control. 34: 274–282. 128
Mills, N. 2009. Egg parasitoids in biological control and integrated pest management, pp. 389–411. In Egg Parasitoids Agroecosystems with Emphas. Trichogramma. Springer Netherlands, Dordrecht.
Milnes, J. M., N. G. Wiman, E. J. Talamas, J. F. Brunner, K. A. Hoelmer, M. L. Buffington, and E. H. Beers. 2016. Discovery of an exotic egg parasitoid of the brown marmorated stink bug, Halyomorpha halys (Stål) in the Pacific Northwest. Proc. Entomol. Soc. Washingt. 118: 466–470.
Miranda, J. E., S. A. de Bortoli, and R. Takahashi. 2002. Development and silk production by silkworm larvae after topical application of methoprene. Sci. Agric. 59: 585–588.
Mizell, R. 2005. Stink bugs and leaffooted bugs are important fruit, nut, seed and vegetable pests. ENY-718/IN534. University of Florida Cooperative Extension. Gainesville, FL.
Mizell, R. F. I., T. C. Riddle, and A. S. Blount. 2008. Trap cropping system to suppress stink bugs in the southern coastal plain. Proc. Florida State Hortic. Soc. 121: 377–382.
Mizell, R. F. I., and W. L. Tedders. 1995. A new monitoring method for detection of the stinkbug complex in pecan orchards. Proc. Southeast. Pecan Grow. Assoc. 88: 36– 40.
Morgan, K. L., and M. A. Olmstead. 2013a. A diversification strategy for perennial horticulture in Florida. Horttechnology. 23: 482–489.
Morgan, K., and M. Olmstead. 2013b. Peach orchard establishment and production planning budgets for florida. Proc. Florida State Hortic. Soc. 126.
Morrison, W. R., D.-H. Lee, B. D. Short, A. Khrimian, and T. C. Leskey. 2016. Establishing the behavioral basis for an attract-and-kill strategy to manage the invasive Halyomorpha halys in apple orchards. J. Pest Sci. (2004). 89: 81–96.
Morrison, W. R., C. R. Mathews, and T. C. Leskey. 2016. Frequency, efficiency, and physical characteristics of predation by generalist predators of brown marmorated stink bug (Hemiptera: Pentatomidae) eggs. Biol. Control. 97: 120–130.
Mumford, J. D., and G. A. Norton. 1984. Economics of decision making in pest management. Annu. Rev. Entomol. 29: 147–74.
Murakoshi, S., C.F. Chang, and S. Tamura. 1972. Increase in silk production by the silkworm, bombyx mori L., due to oral administration of a juvenile hormone analog. Agric. Biol. Chem. 36: 695–696.
129
National Agricultural Statistics Service. 2016. Noncitrus fruits and nuts. United States Department of Agriculture. 107pp.
Nettles Jr., W. C. 1990. In vitro rearing of parasitoids: Role of host factors in nutrition. Arch. Insect Biochem. Physiol. 13: 167–175.
Ni, X., T. E. Cottrell, G. D. Buntin, X. Li, W. Wang, and H. Zhuang. 2017. Monitoring of brown stink bug (Hemiptera: Pentatomidae) population dynamics in corn to predict its abundance using weather data. Insect Sci. 26: 536-544.
Nielsen, A. L., S. Fleischer, G. C. Hamilton, T. Hancock, G. Krawczyk, J. C. Lee, E. Ogburn, J. M. Pote, A. Raudenbush, A. Rucker, M. Saunders, V. P. Skillman, J. Sullivan, J. Timer, J. Walgenbach, N. G. Wiman, and T. C. Leskey. 2017. Phenology of brown marmorated stink bug described using female reproductive development. Ecol. Evol. 7: 6680–6690.
Nielsen, A. L., and G. C. Hamilton. 2009. Life history of the invasive species Halyomorpha halys (Hemiptera: Pentatomidae) in northeastern United States. Ann. Entomol. Soc. Am. 102: 608–616.
Nielsen, A. L., and A. Rucker. 2013. Brown marmorated stink bug control comparing danitol and asana on peach, 2012. Arthropod Manag. Tests. 38.
Nielsen, A. L., P. W. Shearer, and G. C. Hamilton. 2008. Toxicity of insecticides to Halyomorpha halys (Hemiptera: Pentatomidae) using glass-vial bioassays. J. Econ. Entomol. 101: 1439–42.
Nielsen, A. L., P. W. Shearer, G. C. Hamilton, and A. Rucker. 2007. Late-season catfacing insect control on peach, 2006. Arthropod Manag. Tests. 32.
Numata, H., and T. Hidaka. 1984. Termination of adult diapause by a juvenile hormone analogue in the bean bug, Riptortus clavatus. Zoolog. Sci. 1: 751–754.
Ogburn, E. C., R. Bessin, C. Dieckhoff, R. Dobson, M. Grieshop, K. A. Hoelmer, C. Mathews, J. Moore, A. L. Nielsen, K. Poley, J. M. Pote, M. Rogers, C. Welty, and J. F. Walgenbach. 2016. Natural enemy impact on eggs of the invasive brown marmorated stink bug, Halyomorpha halys (Stal) (Hemiptera: Pentatomidae), in organic agroecosystems: a regional assessment. Biol. Control. 101: 39–51.
Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, and H. Wagner. 2013. Package ‘vegan.’ R Packag. ver. 2.5–2.
Olmstead, M., J. X. Chaparro, J. G. Williamson, R. E. Rouse, R. F. I. Mizell, P. Harmon, and J. F. Ferguson. 2013. Florida subtropical peaches: production practices. Univ. Florida Coop. Ext. Gainesville, Florida.
130
Onstad, D. W. 1987. Calculation of economic-injury levels and economic thresholds for pest management. J. Econ. Entomol. 80: 297–303.
Orr, D. 2009. Biological control and integrated pest management, pp. 207–239. In R. Peshin and A.K. Dhawan (eds.), Integrated pest management: innovation- development process. Springer, Dordrecht, Netherlands.
Palmer, M. W. 1990. The estimation of species richness by extrapolation. Ecology. 71: 1195–1198.
Panizzi, A. R., and D. C. Herzog. 1984. Biology of Thyanta perditor (Hemiptera: Pentatomidae). Ann. Entomol. Soc. Am.
Pedigo, L. P., and D. G. Buntin. 1993. Handbook of sampling methods for arthropods in agriculture. CRC Press, Boca Raton, Florida.
Pedigo, L. P., S. H. Hutchins, and L. G. Higley. 1986. Economic injury levels in theory and practice. Annu. Rev. Entomol. 31: 341–368.
Penca, C., and A. Hodges. 2018. First report of brown marmorated stink bug (Hemiptera: Pentatomidae) reproduction and localized establishment in Florida. Florida Entomol. 101: 708–711.
Penca, C., and A. C. Hodges. 2017. Pyriproxyfen treatment terminates Halyomorpha halys reproductive diapause, with an indirect mortality effect on its egg parasitoid Trissolcus japonicus. J. Pest Sci. (2004). 90.
Pfeiffer, D. G., L. F. Ponton, and M. Varn. 1985. Peach, catfacing and oriental fruit moth trial, 1984. Insectic. Acaric. Tests. 10: 41–41.
Pilkay, G. L., F. P. F. Reay-Jones, M. D. Toews, J. K. Greene, and W. C. Bridges. 2015. Spatial and temporal dynamics of stink bugs in southeastern farmscapes. J. Insect Sci. 15.
Poston, F. L., L. P. Pedigo, and S. M. Welch. 1983. Economic injury levels: reality and practicality. Bull. Entomol. Soc. Am. 29: 49–53.
R Core Team. 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
Rahat, S., G. M. Gurr, S. D. Wratten, J. Mo, and R. Neeson. 2005. Effect of plant nectars on adult longevity of the stinkbug parasitoid, Trissolcus basalis. Int. J. Pest Manag. 51: 321–324.
131
Reay-Jones, F. P. F., J. K. Greene, M. D. Toews, and R. B. Reeves. 2009. Sampling stink bugs (Hemiptera: Pentatomidae) for population estimation and pest management in Southeastern cotton production. J. Econ. Entomol. 102: 2360– 2370.
Reay-Jones, F. P. F., M. D. Toews, J. K. Greene, and R. B. Reeves. 2010. Development of sampling plans for cotton bolls injured by stink bugs (Hemiptera: Pentatomidae). J. Econ. Entomol. 103: 525–532.
Rice, K. B., C. J. Bergh, E. J. Bergmann, D. J. Biddinger, C. Dieckhoff, G. Dively, H. Fraser, T. Gariepy, G. Hamilton, T. Haye, A. Herbert, K. Hoelmer, C. R. Hooks, A. Jones, G. Krawczyk, T. Kuhar, H. Martinson, W. Mitchell, A. L. Nielsen, D. G. Pfeiffer, M. J. Raupp, C. Rodriguez-Saona, P. Shearer, P. Shrewsbury, P. D. Venugopal, J. Whalen, N. G. Wiman, T. C. Leskey, and J. F. Tooker. 2014. Biology, ecology, and management of brown marmorated stink bug (hemiptera: pentatomidae). J. Integr. Pest Manag. 5: 1–13.
Rider, D. A., and J. B. Chapin. 1992. Revision of the genus Thyanta Stal, 1862 (Heteroptera, Pentatomidae). 2. North America, Central America, and the West Indies. J. New York Entomol. Soc. 100: 42–98.
Rings, R. W. 1957. Types and seasonal incidence of stink bug injury to peaches. J. Econ. Entomol. 50: 599–604.
Ritz, C., and J. C. Streibig. 2005. Bioassay analysis using R. J. Stat. Softw. 12: 1–22.
Rosenheim, J. A., and M. A. Hoy. 1988. Sublethal effects of pesticides on the parasitoid Aphytis melinus (Hymenoptera: Aphelinidae). J. Econ. Entomol. 81: 476– 483.
Ruberson, J. R., K. Takasu, G. David Buntin, J. E. Eger, W. A. Gardner, J. K. Greene, T. M. Jenkins, W. A. Jones, D. M. Olson, P. M. Roberts, D. R. Suiter, and M. D. Toews. 2013. From Asian curiosity to eruptive American pest: Megacopta cribraria (Hemiptera: Plataspidae) and prospects for its biological control. Appl. Entomol. Zool. 48: 3–13.
Rucker, A., P. Shearer, and K. Tollerup. 2009. Late-season catfacing insect control on peach, 2008. Arthropod Manag. Tests. 34.
Saber, M., M. J. Hejazi, K. Kamali, and S. Moharramipour. 2005. Lethal and sublethal effects of fenitrothion and deltamethrin residues on the egg parasitoid Trissolcus grandis (Hymenoptera: Scelionidae). J. Econ. Entomol. 98: 35–40.
Saulich, A. K., and D. L. Musolin. 2012. Diapause in the seasonal cycle of stink bugs (Heteroptera, Pentatomidae) from the Temperate Zone. Entomol. Rev. 92: 1–26.
132
Saunders, D. S. 1983. A diapause induction-termination asymmetry in the photoperiodic responses of the Linden bug, Pyrrhocoris apterus and an effect of near-critical photoperiods on development. J. Insect Physiol. 29: 399–405.
Seiter, N. J., J. K. Greene, and F. P. F. Reay-Jones. 2013. Reduction of soybean yield components by Megacopta cribraria (Hemiptera: Plataspidae). J. Econ. Entomol. 106: 1676–1683.
Sforza, R. F. H., M. C. Bon, G. Martel, M. Augé, M. Roche, R. Mahmood, and L. Smith. 2017. Initial evaluation of two native egg parasitoids for the control of Bagrada hilaris, and invasive stink bug in Western USA, p. 221. In Symp. Biol. Control Arthropods.
Sharpe, R. H. 1969. Sub-tropical peaches and necatarines. Proc. Florida State Hortic. Soc. 82: 302-306.
Sherman, W. B., P. M. Lyrene, and R. H. Sharpe. 1996. Low-chill peach and nectarine breeding at the University of Florida. Proc. Fla. State Hort. Soc. 109: 222–223.
Singh, S., and K. Kumar. 2015. Effects of juvenoid pyriproxyfen on reproduction and F1 progeny in myiasis causing flesh fly Sarcophaga ruficornis L. (Sarcophagidae: Diptera). Parasitol. Res. 114: 2325–2331.
Smith, E. P., and G. van Belle. 1984. Nonparametric estimation of species richness. Biometrics. 40: 119-129.
Smith, S. M. 1996. Biological control with Trichogramma : advances, successes, and potential of their use. Annu. Rev. Entomol. 41: 375–406.
Snodgrass, G. L., J. J. Adamczyk, and J. Gore. 2005. Toxicity of insecticides in a glass-vial bioassay to adult brown, green, and southern green stink bugs (Heteroptera: Pentatomidae). J. Econ. Entomol. 98: 177-181.
Stam, P. A., L. D. Newsom, and E. N. Lambremont. 1987. Predation and food as factors affecting survival of Nezara viridula (L.) (Hemiptera: Pentatomidae) in a soybean ecosystem. Environ. Entomol. 16: 1211–1216.
Stanley, J., and G. Preetha. 2016. Pesticide toxicity to parasitoids: exposure, toxicity and risk assessment methodologies, pp. 99–152. In Pestic. Toxic. to Non-Target Org. Springer Netherlands, Dordrecht.
Stern, V., R. Smith, R. Van Den Bosch, and K. Hagen. 1959. The integrated control concept. Hilgardia. 29: 81-101.
133
Suiter, D. R., J. E. Eger, W. A. Gardner, R. C. Kemerait, J. N. All, P. M. Roberts, J. K. Greene, L. M. Ames, G. D. Buntin, T. M. Jenkins, and G. K. Douce. 2010. Discovery and distribution of Megacopta cribraria (Hemiptera: Heteroptera: Plataspidae) in northeast Georgia. J. Int. Pest. Mgmt. 1: 1-4.
Sylvester, E. S., and E. L. Cox. 1961. Sequential plans for sampling aphids on sugar beets in Kern County, California. J. Econ. Entomol. 54: 1080–1085.
Talamas, E. J., M. V. Herlihy, C. Dieckhoff, K. A. Hoelmer, M. Buffington, M.-C. Bon, and D. C. Weber. 2015. Trissolcus japonicus (Ashmead) (Hymenoptera, Scelionidae) emerges in North America. J. Hymenopt. Res. 43: 119–128.
Taylor, L. R. 1961. Aggregation, variance and the mean. Nature. 189: 732–735.
Taylor, L. R., I. P. Woiwod, and J. N. Perry. 1978. The density-dependence of spatial behaviour and the rarity of randomness. J. Anim. Ecol. 47: 383.
Varma, G. C., and P. P. Singh. 1987. Effect of insecticides on the emergence of Trichogramma Brasiliensis (Hymenoptera: Trichogrammatidae) from parasitized host eggs. Entomophaga. 32: 443–448.
Volkoff, N., S. B. Vinson, Z. X. Wu, and W. C. Nettles. 1992. In vitro rearing of Trissolcus basalis (Hym., Scelionidae an egg parasitoid of Nezara viridula (Hem., Pentatomidae). Entomophaga. 37: 141–148.
Waage, J. K., and M. P. Hassell. 1982. Parasitoids as biological control agents – a fundamental approach. Parasitology. 84: 241-268.
Wald, A. 1945. Sequential method of sampling for deciding between two courses of action. J. Am. Stat. Assoc. 40: 277–306.
Waldstein, D. E., T. A. Bradshaw, and J. A. Waldstein. 2005. Control of insects on peach, 2004. Arthropod Manag. Tests. 30.
Waldstein, D. E., M. K. Kering, and J. A. Bradshaw. 2004. Early season control of san jose scale and catfacing insects on peach, 2003. Arthropod Manag. Tests. 29.
Walgenbach, J. F., and S. C. Schoof. 2014. Peach insecticide trial, 2013. Arthropod Manag. Tests. 39.
Watanabe, M. 1979. Ecology and extermination of Halyomorpha halys 4: the relationship between day length and ovarian development. Annu. Rep. Toyama Inst. Heal. 3: 33–37.
Weber, D. C., W. R. Morrison, A. Khrimian, K. B. Rice, T. C. Leskey, C. Rodriguez- Saona, A. L. Nielsen, and B. R. Blaauw. 2017. Chemical ecology of Halyomorpha halys: discoveries and applications. J. Pest Sci. (2004).
134
Wermelinger, B., D. Wyniger, and B. Forster. 2008. First records of an invasive bug in Europe: Halyomorpha halys Stål (Heteroptera: Pentatomidae), a new pest on woody ornamentals and fruit trees? Bull. la Société Entomol. Suisse. 81: 1–8.
Wilson, L. T., and P. M. Room. 1983. Clumping patterns of fruit and arthropods in cotton, with implications for binomial sampling. Environ. Entomol. 12: 50–54.
Woodside, A. M. 1947. Weed hosts of bugs which cause cat-facing of peaches in Virginia. J. Econ. Entomol. 40: 231–233.
Woodside, A. M. 1950. Cat-facing and dimpling of peaches. Bull. Virginia Agric. Exp. Stn. 435.
Yang, Z. Q., Y. X. Yao, L. F. Qiu, and Z. X. Li. 2009. A new species of Trissolcus (Hymenoptera: Scelionidae) parasitizing eggs of Halyomorpha halys (Heteroptera: Pentatomidae) in China with comments on its biology. Ann. Entomol. Soc. Am. 102: 39–47.
Yonce, C. E., and R. F. I. Mizell. 1997. Stink bug trapping with a pheromone. Southeast. Pecan Grow. Assoc. 90: 54–56.
Zhang, Y., J. L. Hanula, and S. Horn. 2012. The biology and preliminary host range of Megacopta cribraria (Heteroptera: Plataspidae) and its impact on kudzu growth. Environ. Entomol. 41: 40–50.
Zhu, G., W. Bu, Y. Gao, and G. Liu. 2012. Potential geographic distribution of brown marmorated stink bug invasion (Halyomorpha halys). PLoS One. 7.
135
BIOGRAPHICAL SKETCH
Cory Penca was born and raised in Coral Springs, Florida, where he spent his younger days catching bugs and lizards with his brother, with plenty of encouragement from his parents. He studied botany and biology as an undergraduate at the University of Florida. After completion of his bachelor’s degree, Cory was introduced to entomology as a practical discipline while working as a laboratory technician for the
Florida Department of Agriculture and Consumer Services, Division of Plant Industry.
While at the Division of Plant Industry, Cory completed a master’s degree in natural resource policy and administration at the University of Florida, advised by Damian
Adams. In the Summer of 2015 Cory enrolled in the Doctor of Plant Medicine program at the University of Florida, and in the Fall of 2015 was admitted as a PhD student under the advisement of Dr. Amanda Hodges. His PhD research focused on studying the catfacing pest complex in Florida peaches and researching the use of juvenile hormone analogs for diapause interruption and parasitoid rearing.
During his time in Gainesville, Cory played rugby with the Gainesville Hogs
Rugby Football Club, winning multiple state championships. He enjoyed exploring and fishing the waters of Florida in his canoe with his fiancé, Sarah. Cory hopes to apply his education towards a fulfilling career in the service of those who grow our food.
136