ADVANCING OUR UNDERSTANDING OF ASTER YELLOWS EPIDEMIOLOGY TOWARD IMPROVED DISEASE MANAGEMENT
By
Kenneth E. Frost
A dissertation submitted in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
(Plant Pathology)
at the
UNIVERSITY OF WISCONSIN – MADISON
2012
Date of final oral examination: 11-02-2012
The dissertation is approved by the following members of the Final Oral Committee: Russell Groves, Associate Professor, Entomology Amy O. Charkowski, Associate Professor, Plant Pathology Jed B. Colquhoun, Professor, Horticulture Paul D. Esker, Assistant Professor, Plant Pathology David Kyle Willis, Associate Professor, Plant Pathology
i Acknowledgments
I thank my adviser, Russ Groves, for his excellent criticism and support over the course of this research. I also thank my committee members, Amy Charkowski, D. Kyle
Willis, Paul Esker and Jed Colquhoun for their time, guidance, and advice. Along the way, conversations with Tom German, Jeff Wyman, Brian Flood, and Paul Mitchell have been enjoyable and helpful. I want to give a shout out to all the current and former Groves Lab members – Carol Groves, Emily Mueller, Scott Chapman, Shahideh Nouri, Anders Huseth, JP
Soto-Arias, Chen Zhang, David Lowenstein, Natalie Hernandez, and Sarah Schramm have all been invaluable sources of information and constructive criticism over the last five years. I thank them for their insightful discussion in and around Russell Labs and at the terrace. I should also thank the workers at the Hancock Wisconsin Agricultural Experiment Station for their willingness to lend a hand and the carrot growers of Wisconsin for their cooperation and support. Finally, I would like to thank my family for their support throughout this work.
ii Abstract
Rational: Each year, Wisconsin vegetable producers grow carrots on an average of 4,500 acres grossing over $6 million dollars in revenues. Unfortunately, this crop is threatened annually by the occurrence of aster yellows (AY), a disease caused by the aster yellows phytoplasma (AYp). The AYp is transmitted primarily by the aster leafhopper (ALH) in a persistent-propagative manner and current disease control focuses on controlling the insect vector. The decision to intercede and implement an insecticide spray is based upon the Aster
Yellows Index (AYI), a calculation of the allowable numbers of ALH for a crop at a particular infectivity rate. Since its development, the AYI has been refined to include treatment thresholds for crops with differing susceptibility to AY. We have also advanced our tools for detecting AYp in plants and insects and have come to understand that the components of the index, abundance and infectivity, vary spatially and temporally in Wisconsin. However, these advances in technology and understanding have not been used to their highest potential for managing AY.
Project Goal: To advance our understanding of the epidemiology of aster yellows in
Wisconsin towards the development and implementation of a comprehensive management plan
Objectives: I) To determine the probability that an AYp-infected leafhopper will transmit the pathogen to a susceptible plant. II-A) To evaluate the relative importance of assigned causes of variability associated with aster leafhopper abundance and infectivity. B) To determine seasonal trends in factors associated with AY risk. III) Refine and characterize the current integrated pest management program in processing carrot production to include reduced-risk
iii technologies.
To address these objectives we developed molecular diagnostic tools to ensure molecular AYp detection was accurate, precise, and reflects biology, used historical pest survey sampling to identify seasonal trends in factors associated with AY risk, created an ALH development model to better understand conditions favorable for the sporadic occurrence of high AY risk periods, and combined this information together with econometric assessments for improved decision making in pest management and risk avoidance.
Impact and Outcomes: By ensuring molecular AYp detection is accurate, by identifying seasonal trends in association with AY risk, and by advancing predictive tools to address the sporadic occurrence of high AY risk periods, we can minimize costs associated with unwarranted pesticide applications and reduce yield losses due to advanced preparation for
ALH infestations. This project aims to incorporate biologically relevant information about the AY disease system from multiple scales using available and emerging technologies to improve on-farm AY management decisions. These improvements to diagnostic, detection and vector monitoring systems are anticipated to benefit agricultural producers, crop consultants and other stakeholders and will be essential for properly timing vector control measures to times when plant protection is most needed.
iv Table of Contents Acknowledgments ...... i Thesis Abstract ...... ii Chapter 1: Literature Review ...... 1 Problem defined ...... 2 The aster yellows phytoplasma ...... 2 The aster leafhopper ...... 4 Mode of transmission ...... 5 Aster Leafhopper migration ...... 5 Aster Leafhopper influence on aster yellows in Wisconsin ...... 6 Aster Yellows management ...... 7 Research rational and significance ...... 8 References ...... 12 Chapter 2: Detection and variability of aster yellows phytoplasma (Candidatus Phytoplasma asteris) titer in its insect vector, Macrosteles quadrilineatus (Hemiptera: Cicadellidae) ...... 16 Abstract ...... 17 Introduction ...... 18 Materials and Methods ...... 21 Results ...... 33 Discussion ...... 39 References ...... 50 Tables ...... 57 Figures ...... 63 Chapter 3: Factors influencing aster leafhopper (Macrosteles quadrilineatus) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields ...... 67 Abstract ...... 68 Introduction ...... 69
v Materials and Methods ...... 73 Results ...... 83 Discussion ...... 86 References ...... 96 Tables ...... 102 Figures ...... 107 Appendix A ...... 111 Appendix B ...... 116 Chapter 4: Seasonal pattern of aster leafhopper (Macrosteles quadrilineatus) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields ...... 118 Abstract ...... 119 Introduction ...... 120 Materials and Methods ...... 123 Results ...... 130 Discussion ...... 133 References ...... 143 Tables ...... 148 Figures ...... 153 Chapter 5: Feasibility of alternative management strategies for the control of aster yellows in Wisconsin carrot ...... 158 Abstract ...... 159 Introduction ...... 160 Materials and Methods ...... 163 Results ...... 168 Discussion ...... 171 References ...... 176 Tables ...... 179 Figures ...... 188 Chapter 6: Concluding Remarks and Future Directions ...... 190
vi Summary ...... 191 Future directions ...... 194 Tables and Figures ...... 198
1
Chapter 1: Literature Review
2
Problem defined. Aster yellows (AY) is a disease caused by the aster yellows phytoplasma
(AYp), which is a small prokaryote that is taxonomically placed in the provisional genus,
Candidatus. The AYp is obligately associated with its plant and insect hosts and has not been successfully cultured in the laboratory, which has slowed research progress due to the inability to obtain a pure culture or construct and examine mutants. The AYp has an extensive and diverse host range infecting over 350 plant species including many common vegetable crops, ornamental plant species, field crops, as well as several non-crop wild host species
(Peterson 1973, Lee and Gunderson 2000, Hollingsworth et al. 2008, ). A large proportion of these plant species have been determined to be hosts of the pathogen through environmental surveys and greenhouse bioassays, but little is known about disease symptoms expressed by non-crop species and only a portion of the non-crop species may serve as epidemiologically important inoculum sources for spread to valued crops. The symptoms caused by AYp are variable, but the most common disease phenotypes are vein clearing, chlorosis, stunting, twisting and proliferation of stems and the development of adventitious roots (Kunkle 1926,
Lee et al. 2000). These symptoms lead to direct yield and quality losses and processing problems that result from malformed roots and an inability to obtain clean raw product.
The aster yellows phytoplasma. The AYp is a small prokaryote that is taxonomically placed in the provisional genus Candidatus (i.e. 'Ca. Phytoplasma asteris') which corresponds to the
16SrI group according to a proposed phytoplasma phylogeny based on 16S rRNA gene sequences (Lee et al. 2000, IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group 2004). Similar to other mollicutes, the AYp is thought to have undergone reductive evolution because of its long coevolutionary history with both is plant and insect 3 hosts resulting in a reduced genome (~ 660 to 1300 kb) (Bove 1997, Bai et al. 2006).
Nevertheless, the 'streamlined' genomes of phytoplasmas, and other mollicutes, are likely the reason these organisms have either not been cultured or require complex culture mediums.
The AYp is obligately associated with its plant and insect hosts and recent studies examining disease phenotypes together with molecular methods show that AYp organisms exist as a diverse population of strains in the agricultural environment (Lee et al. 2003, Zang et al. 2004) which often group by host association. Six strains previously characterized in
Ohio have recently been detected in Wisconsin carrot, but 30-50% of the pathogen isolates collected could not be characterized using previously developed primer-typing methods
(Zhang et al. 2004, Frost et al. 2008). Unfortunately, we have an incomplete understanding of how these diverse populations of AYp genotypes influence disease epidemics or the role of the insect or plant host in maintaining this pathogen diversity in the agroecosystem. It is known that some AYp strains are transmitted by the leafhopper at different rates suggesting that pathogen variability can influence both the acquisition efficiency and the latent period of AYp in the insect (Murral et al. 1996). Additionally, AYp strains have a complex pathogenic relationship with a diverse host range including members of both monocots and dicots.
Analyses of the genetic diversity of AYp have begun to elucidate differences between many of the AYp strains (Lee et al. 2003, Zang et al. 2004). While not explicitly studied in the Midwest agroecosystem, a conclusion emerging from studies of similar pathosystems is that the AYp strains will cluster within groups based upon host association (Pooler et al. 1995, Hendson et al. 2001).
Specific to AYp, Lee et al. (2003) reported that strains cluster according to plant hosts 4 grown in a common garden in Texas. There are many non-crop plant species that have been implicated as AYp hosts (Schultz 1979) but, to our knowledge, there have been no systematic or comprehensive surveys of wild plants that support populations of AYp that could serve as primary inoculum sources. Moreover, the genetic relationships associated with the ability of these strains, or sets of strains, to survive in reservoir hosts or cause disease on an economic host have not been investigated. In severe years, local AYp sources are thought to greatly contribute to epidemic development in Wisconsin crops (Schultz 1979).
The Aster Leafhopper. Although twenty-four leafhopper species are known to carry and transmit AYp organisms, the aster leafhopper (ALH), Macrosteles quadrilineatus Forbes, is usually considered the primary vector of the AYp due to its absolute and relative abundance
(to other potential vectoring species) in susceptible crops. The ALH has a widespread distribution in the continental (lower 48) United States and, more recently, a Macrosteles species, closely related to M. quadrilineatus, has been introduced to Hawaii (Le Roux and
Rubinoff 2009). In Wisconsin, local leafhopper populations will often overwinter in small grains and perennial weeds as eggs which hatch and pass through five nymphal instars prior to adulthood. The ALH is polyphagous and has been reported to feed on more than 300 plant species, including many species that are susceptible to the AYp. Reports of ALH host preferences in the field vary by plant community composition (Lee and Robinson 1958,
Wallis 1962, Schultz 1979), plant physiological state (Peterson 1973), season and geographic location of the study (Lee and Robinson 1958, Wallis 1962, Peterson 1973). Not all plant species are likely infected at the same rates and are also not distributed in the landscape in the same way. Additionally, the composition of the crop hosts and the spatial arrangement of plant 5 host species can influence epidemic severity since leafhoppers disperse more frequently and farther when present in less preferred hosts (Zhou et al. 2003).
Mode of transmission. The AYp is thought to be circulative and propagative in the ALH and vector competence involves acquisition, pathogen replication and circulation in the insect, and transmission to a susceptible host (Matthews 1991). The ALH acquires AYp by feeding on vascular tissues of infected plants for extended periods of time (hours to days). The phytoplasma then moves through the alimentary canal to the lumen of the midgut. The phytoplasma then migrates from the midgut lumen, through the epithelium, to spaces between the basal plasmalemma and the basal lamina and into the hemocoel (Fletcher et al. 1998,
Kwon et al. 1999) during a two to three week latency period; or the period after acquisition during which the insect cannot transmit the pathogen. AYp organisms then circulate in the insect body and replicate in the hemolymph and other insect tissues including muscle cells
(Maramorosch 1952, Sinha and Chiykowski 1967, Fletcher et al. 1998). The pathogen then moves into salivary ducts, and is introduced into plant hosts with insect saliva during the egestion phase of feeding (Fletcher et al. 1998, Kwon et al. 1999). Once infectious, the leafhopper may transmit AYp to healthy plants during relatively short inoculation feeding periods lasting several minutes to a few hours (Maramorosch 1953). A leafhopper remains infected and able to transmit AYp for the remainder of its adult life. Although transovarial transmission of AYp is not thought to occur within the ALH, there is mounting evidence that suggests transovarial transmission of phytoplasmas does occur for some insect-pathogen combinations (Alma et al. 1997, Kawakita et al. 2000).
Aster leafhopper migration. A defining feature of the ALH’s biology is an early season 6 migration of the insect from the Gulf-states to the Upper Midwest. The migratory behavior of the ALH is documented to greatly influence the potential for aster yellows epidemics in the upper Midwest regions of the United States (Chiykowski and Chapman 1965, Drake and
Chapman 1965). Whether the ALH migration affects Wisconsin is highly dependent on weather and wind patterns since the ALH travels over large distances in surface and upper air wind currents, usually in a south-to-north direction. ALHs are thought to move into upper-
Midwestern fields presumably from the Gulf States in early spring (Chiykowski and Chapman
1965, Drake and Chapman 1965) and from the central and northern Great Plains later in the growing season (Hoy et al. 1992). The migratory behavior together with the mode of pathogen transmission (persistent and propagative) by the ALH enables the insect to carry and transmit the pathogen over great distances. Presumably, the first AYp to enter Wisconsin’s carrot, onion and potato fields is vectored by adult female leafhoppers migrating from grain crops in the southern U.S. and particularly the Gulf States in late April to mid-May. These migrations coincide with the early germination and establishment of susceptible carrot and small grain cover crops.
Aster leafhopper influence on AY in Wisconsin. ALH movement patterns can be classified into two groups that are important for spreading the AYp based on where the insects acquire the pathogen. The first group is comprised of ALHs that overwinter in Wisconsin (or offspring of early season migrants) and acquire the AYp locally spreading it into our susceptible crops.
This movement pattern likely occurs every season and is responsible for AYp spread later in the growing season. Insect phenology (the local, temperature-driven, seasonal population dynamics of the insect) is a probable driver of this ALH movement pattern and, coupled with 7 the prevalence of local AYp sources, could influence AY disease pressure (Canto et al. 2009,
Granadinos 2009, Schultz 1979). This movement pattern likely exerts a more constant influence on AY disease pressure which represents the baseline disease pressure. The second movement pattern is comprised of ALHs that arrive in Wisconsin from distant locations with some individuals infected with AYp, spreading the pathogen to our susceptible crops.
However, these influxes tend to occur sporadically and their frequency of occurrence is difficult to quantify. These movement patterns are likely driven by surface weather patterns leading to their stochastic nature (Hurd 1920, Huff 1963, Chiykowski and Chapman 1965,
Hoy et al. 1992). It is not known if the dispersal of migrant leafhoppers upon, or after, arrival or the dispersal of locally overwintering leafhoppers, are more important epidemiologcal drivers of AYp epidemics in a given year. However, an increased understanding of the frequency at which these different ALH movement patterns occur and influence AY pressure can lead to better disease management due to advanced preparation for (or anticipation of) high AY risk periods.
Aster yellows management. A critical factor for a successful disease control program relies on a detailed understanding of the components which directly affect disease development.
Knowledge of the organism being targeted, environmental influence on disease development, and host plant susceptibility are vital to timing and scheduling chemical applications for control of aster yellows. To control aster yellows, the decision to intercede and implement a pest control practice is usually based upon calculation of the Aster Yellows Index (AYI)
(Chapman 1971). This index is a calculation of the allowable numbers of insects for a crop, based on the percentage of insects that are inoculative. The AYI has improved timing of 8 pesticide applications by targeting periods when plant protection is most needed. The AYI has since been refined to include treatment thresholds for crops with differing susceptibility to AY.
The AYI was originally calculated using data obtained from infectivity bioassays which directly measured the ability of an individual ALH to infect a susceptible host
(Chapman 1971). A major constraint associated with implementation of this bioassay to farm- scale management practices is the time needed for symptom development of infected plants.
Crop scouting and molecular diagnostic tools incorporating polymerase chain reaction (PCR) have been adopted to determine the percentage of leafhoppers that are infected with mollicutes (Bloomquist and Kirkpatrick 2002, Munyaneza et al. 2010). These molecular diagnostic tools have decreased the lag between finding infected leafhoppers and prescribed sprays. However, infected leafhoppers do not always transmit pathogens to susceptible crops
(i.e. latent period), so there is not a direct relationship between the proportion of infected leafhoppers and the proportion of infective leafhoppers and limited information exists to determine the proportion of PCR-positive individuals that are inoculative. Additionally, it is nowt known that the critical components underlying AYI estimates, including ALH abundance and infectivity, vary spatially and temporally in Wisconsin (Rice-Mahr et al. 1993). While important, the advances in AYp detection technology and the knowledge of the spatial and temporal variability of AY risk have not been used to their highest potential for managing AY in Wisconsin.
Research Rational and Significance: Each year, Wisconsin vegetable producers grow carrots on an average of 3,200 acres, onions on 2,200 acres, and potatoes on approximately
1,800 acres, grossing over $14 million dollars in revenues (USDA-NASS, 2011). 9
Unfortunately, these crops are threatened annually by the occurrence of AYp which is obligately transmitted by the ALH. Currently, the decision to intercede and implement a pest control practice (e.g. insecticide spray) is based upon the calculation of the Aster Yellows
Index (AYI; described above). The insecticide sprays, when implemented, primarily utilize
Group 3 synthetic pyrethroids (IRAC, Mode-of Action Classification http://www.irac- online.org/) that target the ALH. Although effective in controlling these insects, the IRAC
Group 3 compounds have impacts on other potentially beneficial insects present in the crop.
Additionally, these compounds are broadly characterized as having a wide spectrum of activity often with acute oral neurotoxicity to mammals, notable chronic effects as endocrine disruptors, and are classified as both mutagenic and carcinogenic.
The synthetic pyrethroids are a class of chemicals that have been introduced over the past three decades for a variety of insecticidal uses including both agricultural and urban/domestic applications. These materials currently comprise the backbone of low-cost registrations which are relied upon for use against the ALH. The synthetic pyrethroids were conditionally re-registered beginning in 1984 for use on selected crops and currently, EPA is assessing risks to non-target organisms for ten synthetic pyrethroids: bifenthrin, cyfluthrin, cypermethrin, deltamethrin, fenpropathrin, fenvalerate, cyhalothrin, tefluthrin, tralomethrin, and permethrin. Several of these synthetic pyrethroids remain conditionally registered for use on vegetables grown in muck soils, however, each of these chemicals are highly lipophilic and, in aquatic environments, tend to strongly adsorb to sediments. Based on the affinity for synthetic pyrethroids to partition to sediments, EPA identified the need to re-evaluate the environmental fate and potential toxicity of synthetic pyrethroids in aquatic sediments, 10 especially in ecologically sensitive areas including low-land muck soils.
Recently, some states have placed certain pesticide products containing pyrethroids into re-evaluation. The re-evaluation is again based on recent monitoring surveys and toxicity studies revealing the widespread presence of synthetic pyrethroid residues in the sediment of both agricultural and urban dominated waterways (Werner et al. 2002, Weston et al. 2004,
Gan et al. 2005). Under section 4 of the Federal Insecticide, Fungicide, and Rodenticide Act
(FIFRA), US EPA continues to re-evaluate existing pesticides to ensure that they meet current scientific and regulatory standards. US EPA has recently completed several Reregistration
Eligibility Decisions for the synthetic pyrethroid group of insecticides and will likely continue to do so in the future under section 4(g)(2)(A) of FIFRA. With the advent of novel, reduced risk, and less broad spectrum registrations for many homopterous, sucking insect pests, the continued reregistration eligibility of this class of insecticides could be in jeopardy.
Growers currently achieve adequate to good control of aster yellows in their crops and
AYp has typically been controlled in Wisconsin through repetitive applications of compounds in the synthetic pyrethroid group. In a given year, it is common for as many as 5-7 applications of insecticide to occur on a 7-10 day calendar schedule on the same crop. In part, the rationale behind these repetitive spray applications results from the fact that a single insecticide application is inexpensive and may cost only $1.5 to $3.0 per acre, not including cost of application. However, cheap and adequate disease control has slowed the refinement of our current IPM strategies and inhibited the development of new control tactics specific to this disease system.
Considering the lack of new epidemiological knowledge generated for this disease 11 system over the last two decades, combined with the current use pattern of broad spectrum pyrethroid insecticides on Wisconsin vegetable crops and the E.P.A. re-evaluation of those synthetic pyrethroid compounds, we feel it is important to generate new, research-based and sustainable forms of pest management approaches to control this vector-borne disease now and into the future.
12
References Cited
Alma A, Bosco, D., Danielli, A., Bertaccini, A., Vibio, M. and A. Arzone, 1997. Identification of phytoplasmas in eggs, nymphs and adults of Scaphiodeus titanus Ball reared on healthy plants. Insect Mol. Biol. 6:115-121.
Bia, X., Zhang, J., Ewing, A., Miller, S.A., Radek, A. J., Shevchenko, D.V., Tsukerman, K., Walunas, T., Lapidus, A., Campbell, J.W., and S.A. Hogenhout. 2006. Living with genome instability: the adaptation of phytoplasmas to diverse environments of their insect and plant hosts. J. Bacteriol. 188(10):3682-3696.
Bloomquist, C.L. and B.C. Kirkpatrick. 2002. Frequency and seasonal distribution of pear psylla infected with the pear decline phytoplasma in California pear orchards. Phytopathology 92:1218-1226.
Bove, J.M. 1997. Spiroplasmas: infectious agents of plants, arthropods and vertebrates. Wein Klin. Wochenschr. 109:604-612.
Canto, T., Aranda, M.A., and A. Fereres. 2009. Climate change effects on physiology and population processes of hosts and vectors that influence the spread of hemipteran- borne plant viruses. Glob. Change Biol. 15:1884-1894.
Chapman, R.K. 1971. Prediction of aster leafhopper migrations and the resultant aster yellows problem in Wisconsin. Proc. North Cent. Branch Entomol. Soc. Am. 26:96- 98.
Chiykowski L.N. and R.K.Chapman. 1965. Migration of the six-spotted leafhopper in central North America. Wisc. Coll. ALS Res. B. 261: 21-45.
Drake D.C. and R.K. Chapman. 1965. Evidence for long distance migration of the six- spotted leafhopper in Wisconsin. Wisc. Coll. ALS Res. B. 261: 1-20.
Fletcher, J., Wayadande, A. Melcher, U. and F. Ye. 1998. The phytopathogenic mollicute- insect vector interface: a closer look. Phytopathology 88:1351-1358.
Frost, K. E., Groves, Carol L. and R. L. Groves. 2008. Detection of multiple strains of the aster yellows phytoplasma in Wisconsin carrot fields [abstract]. Phytopathology 98:S55.
Gan, J. Lee, S.J., Liu, W. P., Haver, D. L. and J.N. Kabashima. 2005. Distribution and persistence of pyrethroids in runoff sediments. J. Eviron. Qual. 34:836-841.
Granadinos. 2004. Managing aster leafhoppers with integrated approaches in Wisconsin 13
carrot production. M.S. Thesis, University of Wisconsin, Madison.
Hendson, M., Purcell, A.H., Chen, D., Smart, C., Guilhabert, M., and B. Kirkpatrick. 2001. Genetic diversity of Pierce’s disease strains and other pathotypes of Xylella fastidiosa. Appl. Environ. Microbiol. 67:895-903.
Hollingsworth, C.R., Atkinson, L.M., Samac, D.A., Larsen, J.E., Motteberg, C.D., Abrahamson, M.D., Glogoza, P., and I.V. MacRae. 2008. Region and field level distributions of aster yellows phytoplasma in small grain crops. Plant Dis. 92:623-630.
Hoy, C.W., Heady, S.E., and T.A. Koch. 1992. Species composistion, phenology, and possible origins of leafhoppers (Cicadellidae) in Ohio vegetable crops. J. Econ. Entomol. 85:2336-2343.
Huff, F.A. 1963. Relation between leafhopper influxes and synoptic weather conditions. J. of Appl. Meteorol. 2(1):39-43.
Hurd, W.E. 1920. Influence of the wind on the movements of insects. Mon. Weather Rev. 48(2):94-98.
IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group. 2004. 'Candidatus Phytoplasma', a taxon for the wall-less, non-helical prokaryotes that colonize plant phloem and insects. J. Syst. Evol. Microbiol. 54:1243-1255.
Kawakita, H., Siki, T., Wei, W., Mitsuhashi, W., Watanabe, K., and M. Sato. 2000. Identification of mulberry dwarf phytoplasmas in the genital organs and eggs of leafhopper Hishimonoides sellatiformis. Phytopathology 90:909-914.
Kunkle, L.O. 1926. Studies on aster yellows. Am. J. Bot. 13(10):646-705.
Kwon, M.-O., Wayadande, A.C., and J. Fletcher. 1999. Spiroplasma citri movement into the intestines and salivary glands of its leafhopper vector, Circulifer tenellus. Phytopathology 82(12):1144-1151
Lee, I.-M., Davis, R.E. and D. E. Gunderson. 2000. Phytoplasma: Phytopathogenic mollicutes. Annu. Rev. Microbiol. 54:221-255.
Lee, I.-M., Martini, M., Bottner, K. D., Dane, R.A., Black, M.C. and N. Troxclair. 2003. Ecological implications from a molecular analysis of phytoplasmas involved in an aster yellows epidemic in various crops in Texas. Phytopathology 93:1368-1377.
Lee, P.E. and A. G. Robinson. 1958. Studies on the six-spotted leafhopper Macrosteles fascifrons (Stal) and aster yellows in Manitoba. Can. J. Plant Sci. 38:320-327. 14
Le Roux, J.J. and D. Rubinoff. 2009. Molecular data reveals California as the potential source of an invasive leafhopper species, Macrosteles sp. nr. severini, transmitting the aster yellows phytoplasma in Hawaii. Ann. Appl. Biol. 154(3):419-427.
Maramorosch K. 1952. Direct evidence for the multiplication of aster-yellows virus in its insect vector. Phytopathology 42(2):59-64.
Maramorosch K. 1953. Incubation period of aster-yellows virus. Am. J. Bot. 40:797-809.
Matthews R.E.F. 1991. Plant Virology 3rd edition. Academic Press, Inc. San Diego, CA 92101.
Munyaneza, J.E., Crosslin, J.M., Upton, J.E., and J. L. Buchman. 2010. Incidence of the beet leafhopper-transmitted virescence agent phytoplasma in local populations of the beet leafhopper, Circulifer tenellus, in Washington State. J. Insect Sci. 10: 18.
Murral D.J., Nault L.R., Hoy C. W., Madden L.V. and S. A. Miller. 1996. Effects of temperature and vector age on transmission of two Ohio strains of aster yellows phytoplasma by the aster leafhopper (Homoptera: Cicadellidae). J. Econ. Entomol. 89(5): 1223-1232.
Peterson, A.G. 1973. Host plant and aster leafhopper relationships. Proc. North Cent. Branch Entomol. Soc. Am. 28:66-70.
Pooler M.R., Myung I. S., Bentz J., Sherald J., and J. S. Hartung. 1997. Detection of Xylella fastidiosa in potential insect vectors by immunomagnetic separation and nested polymerase chain reaction. Lett. Appl. Microbiol. 25:123-6.
Rice-Mahr, S.E., Wyman, J.A., and R. K. Chapman. 1993. Variability in aster yellows infectivity of local populations of the aster leafhopper (Homoptera: Cicadellidae) in Wisconsin. J. Econ. Entomol. 86:1522-1526.
Schultz, G. A. 1979. Epidemiology of the aster leafhopper and aster yellows in relation to disease control. Ph.D. Thesis, University of Wisconsin, Madison.
Sinha R.C. and L. N. Chiykowski. 1967. Initial and subsequent sites of aster yellows virus infection in a leafhopper vector. Virology 33:702-708.
Wallis, R. L. 1962. Host plant preference of the six-spotted leafhopper. J. Econ. Entomol. 55:998-999.
Werner, I., L.A. Deanovic, D.E. Hinton, J.D. Henderson, G.H. de Oliveira, B.W. Wilson, 15
W. Krueger, W.W. Wallender, M.N. Oliver, and F.G. Zalom. 2002. Toxicity of stormwater runoff after dormant spray application of diazinon and esfenvalerate (Asana (R)) in a French prune orchard, Glenn County, California, USA. Bull. Environ. Contam. Toxicol. 68:29–36.
Weston, D.P., You, J., and M.J. Lydy. 2004. Distribution and toxicity of sediment-associated pesticides in agriculture-dominated water bodies in California's Central Valley. Environ. Sci. Technol. 38:2752–2759.
Zhang J., Hogenhout S.A., Nault L.R., Hoy C.W. and S.A. Miller. 2004. Molecular and symptom analyses of phytoplasma strains from lettuce reveal a diverse population. Phytopathology 94(8): 842-849.
Zhou X, Hoy C.W., Miller S.A. and L.R. Nault. 2003. Marking methods and field experiments to estimate aster leafhopper (Macrosteles quadrilineatus) dispersal rates. . Environ. Entomol.32(5): 1177-1186.
16
Chapter 2: Detection and variability of aster yellows phytoplasma (Candidatus
Phytoplasma asteris) titer in its insect vector, Macrosteles quadrilineatus (Hemiptera:
Cicadellidae)
17
Abstract. The aster yellows phytoplasma (AYp) is transmitted by the aster leafhopper (ALH),
Macrosteles quadrilineatus Forbes, in a persistent and propagative manner. To study AYp replication and examine the variability of AYp titer in individual ALHs, we developed a quantitative real-time PCR (qPCR) assay to measure AYp concentration in insect DNA extracts. Absolute quantification of AYp DNA was achieved by comparing the amplification of unknown amounts of an AYp target gene sequence, elongation factor TU (tuf), from whole insect DNA extractions, to the amplification of a dilution series containing known quantities of the tuf gene sequence cloned into a plasmid. The capabilities and limitations of this method were assessed by conducting time course experiments that varied the incubation time of AYp in the ALH from 0 to 9 days following a 48 hour acquisition access period (AAP) on an AYp- infected plant. Average AYp titer was measured in 107 ALHs and, expressed as Log10
(copies/insect), ranged from 3.53 (±0.07) to 6.26 (± 0.11) occurring at 1 and 7 days after the
AAP. AYp titers per insect and relative to an ALH chromosomal reference gene, cp6 wingless
(cp6), increased approximately 100-fold in insects that acquired the AYp. High quantification cycle values obtained for ALHs not exposed to an AYp-infected plant were interpreted as background and used to define a limit of detection for the qPCR assay. This method will improve our ability to study biological factors governing AYp replication in the ALH and determine if AYp titer is associated with frequency of transmission.
Keywords: Aster leafhopper, aster yellows, aster yellows phytoplasma, Macrosteles quadrilineatus, qPCR 18
Aster yellows (AY) is a widespread disease of plants caused by the aster yellows phytoplasma (AYp), a small, wall-less prokaryotic organism that is currently placed in the provisional genus Candidatus (Lee et al. 2000, IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group 2004). The AYp has an extensive and diverse host range infecting over 350 plant species including many common vegetable, ornamental, and agronomically important field crops, and several non-crop plant species (Kunkel 1926,
Chiykowski 1965, Chiykowski and Chapman 1965, Chiykowski 1967, Westdal and
Richardson 1969, Peterson 1973, Lee et al. 1998, Lee et al. 2000, Lee et al. 2003,
Hollingsworth et al. 2008). The most common disease phenotypes are vein clearing, chlorosis, stunting, twisting and proliferation of plant stems, and the development of adventitious roots
(Kunkle 1926, Bloomquist 2002). In vegetable crops, these symptoms can lead to yield and quality losses. For root vegetables, processing problems can result from an inability to obtain clean raw product due to adventitious root growth and associated field soil contamination.
Although, more than twenty-four leafhopper species are known to acquire and transmit AYp organisms (Mahr 1989, Christensen et al. 2005), the aster leafhopper (ALH),
Macrosteles quadrilineatus Forbes, is considered to be the primary vector of the AYp due to its prevalence in Midwestern susceptible crops (Drake and Chapman 1965, Hoy et al. 1992).
The ALH is a polyphagous insect species which utilizes over 300 different plant species for food, oviposition, and shelter and many of these are susceptible to AYp infection, (Wallis
1962, Peterson 1973). ALH host plant species can be classified into two primary groups based on utilization patterns to include: 1) feeding hosts or 2) feeding and reproductive hosts. Other factors such as plant community composition (Lee and Robinson 1958, Wallis 1962, Schultz 19
1979), plant physiological state (Peterson 1973) and seasonal or geographic location (Lee and
Robinson 1958, Wallis 1962, Peterson 1973) can also affect host preferences of ALH in the field. In Wisconsin, cultivated grains are hosts for overwintering eggs and also serve as early feeding and reproductive hosts for the ALH (Drake and Chapman 1965). In addition to grain crops, the ALH feeds on and is moderately abundant in mixed broadleaf weeds and grasses that border crop fields (Shultz 1979).
AYp has been reported to be circulative and propagative in the ALH (Maramorosch
1952, Sinha and Chiykowski 1967, Lee et al. 2000). Vector competence involves acquisition, pathogen replication and circulation to result in successful transmission to a susceptible host
(Matthews 1991). Using indirect methods (see below) and histological studies of other fastidious prokaryotic organisms, a model has been developed for acquisition and transmission of AYp. Briefly, the ALH acquires AYp by feeding on vascular tissues of infected plants for extended periods of time (hours to days) (Murral et al. 1996). The phytoplasma moves from the mouthparts through the alimentary canal to the lumen of the midgut (Sinha and Chiykowski 1967, Fletcher et al. 1998). Once in the midgut lumen, the organism moves through the epithelium (Fletcher et al. 1998, Kwon et al. 1999), circulates in the insect body and replicates in the hemolymph and other insect tissues (Maramorosch 1952a, Sinha and
Chiykowski 1967, Fletcher et al. 1998). Ultimately, AYp infects the accessory salivary glands where it replicates, moves into the salivary ducts and is introduced into plant hosts with insect saliva during the egestion phase of feeding (Fletcher et al. 1998, Kwon et al. 1999). These events occur during a two to three week latency period or the period after acquisition during which the insect cannot transmit the pathogen. Once infectious, the leafhopper may transmit 20
AYp to healthy plants during relatively short inoculation feeding periods and remains infectious for the remainder of its adult life (Maramorosch 1953b). Although transovarial transmission of AYp is not thought to occur, there is mounting evidence that transovarial transmission of phytoplasmas does occur for some insect-pathogen combinations (Alma et al.
1997, Kawakita et al. 2000).
Since AYp organisms have an obligatory association with their plant and insect hosts and have not been successfully cultured in the laboratory, much of what is known about phytoplasma replication in the insect has been derived using indirect methods of measurement. For example, replication of AYp organisms in their insect host was studied using dilution series experiments where insect extracts were microinjected into populations of uninfected insects (Black 1941, Maramorosch 1952a, Maramorosch 1952b, Maramorosch
1955). The infectivity of the resulting microinjected populations was measured and the limiting dilution (the dilution at which leafhoppers lost the ability to transmit) was determined after serial transfers of the pathogen. These early experiments measured 100-fold, or higher, increases of AYp in its insect vector. More recently, DNA hybridization (Rahardja et al. 1992,
Bloomquist and Kirkpatrick 2002), competitive PCR (Liu et al. 1994) and qPCR (Marzachi and Bosco 2005, Bosco et al. 2007a, 2007b) have been used in an attempt to directly quantify the titer of phytoplasma organisms in their insect host.
Recent research in pathogen-vector interactions suggests that the titer of circulative, propagative pathogens in their insect vectors may influence the likelihood of successful transmission events to a susceptible host plant. For example, Rotenberg et al. (2009) demonstrated that single thrips (Frankliniella occidentalis) containing higher titers of Tomato 21 spotted wilt virus (TSWV) transmitted the virus more frequently to susceptible plants. Only a few studies have reported phytoplasma titer of an insect and its relationship to transmission success (Bosco et al. 2007a, Galetto et al. 2009). Additionally, the relationship between AYp titer and frequency of transmission has not been examined for individual ALH. To better understand AYp replication in the ALH, our objectives were (i) to develop a quantitative assay to measure AYp titer in individual leafhoppers, (ii) to examine the variability of AYp titer in
ALHs, and (iii) to use the assay to characterize the temporal dynamics of AYp titer among a population of AYp-infected insects.
Materials and Methods
Aster leafhopper colony. An aster leafhopper colony was established from adult ALH populations collected in wheat (Triticum spp.) fields in central Missouri, (April 2009) and southern Wisconsin (May-June 2009). Adult leafhoppers were initially maintained on oat
(Avena sativa L.) seedlings in a controlled environment with a 16: 8 (L:D) photoperiod (24° C light; 19° C dark). To ensure a phytoplasma-free colony of leafhoppers, the field caught adult female leafhoppers were initially placed on oat seedlings and allowed a 36 hour oviposition period after which time all adult leafhoppers were removed. First instar nymphs resulting from those eggs were subsequently moved onto new oat seedlings not previously visited by
ALHs. Additionally, leafhoppers were periodically checked for phytoplasma or AYp infection by PCR, using primers P1 and 16sSr (Lee et al. 2006) or F4 and R1 (Davis and Lee, 1993; reaction conditions described below). All oat plants were grown in a glasshouse and plants were established by sowing oat seed into 10 cm square pots containing Metro Mix 300 (Sun
Gro Horticulture, Canada CM Ltd.) with each pot receiving approximately 1 gram of 22
Osmocote time-release fertilizer (The Scotts Company LLC, USA).
Phytoplasma isolate collection. AYp isolates were initially obtained by placing 3 groups of
20-30 field-caught adult ALHs per group onto three 96-well flats containing carrot seedlings for a 96 hour acquisition access period (AAP). Carrot plants that subsequently expressed symptoms typical of aster yellows were individually transplanted into 15 cm pots and assayed for the presence of phytoplasma by PCR using the universal phytoplasma nested primer set P1 and 16sSr followed by a second round of amplification with R16F2n and R16R2 (Lee et al.
2006; reaction conditions described below). AYp infection was confirmed using AYp- specific16s rDNA primers F4 and R1. Plants testing PCR-positive for the presence of the AYp were placed into insect-proof cages and 10 fourth- or fifth-instar ALH nymphs were isolated on each infected plant for a 48 hour AAP. Nymphs were then aspirated and transferred to rye
(Secale cereal L.) seedlings where they were maintained for 14 days to complete adult eclosion. As adults, they were again transferred (~5 leafhoppers for each plant species) and isolated onto either 6-week old aster (Callistephus chinensis) ‘Aster-Tiger Paws’ or periwinkle (Catharanthus roseus) plants. Adult ALH placed onto periwinkle were allowed an indefinite inoculation access period (IAP) to recover the AYp in planta for long term storage.
Prior to use in isolation, Chinese aster (Callistephus chinensis) seed was sown into 96 well seedling trays containing Metro Mix 300 and amended with Osmocote time-release fertilizer at a rate 2.5 kg per M3 of soil. When the aster plants had reached the 2-3 true leaf stage, they were transplanted to 10 cm square pots. Aster plants were in the 5-7 leaf stage when phytoplasma isolation was initiated. Plants were maintained in insect-proof cages in the greenhouse at approximately 25-29° C under natural light supplemented with a 16:8 L:D 23 photoperiod.
DNA extraction from leafhoppers. A cetyltrimethylammonium bromide (CTAB) method
(modified from Doyle and Dickson, 1987) was used to extract DNA from individual insects.
Briefly, insects were placed in 1.5 ml microfuge tubes and washed with 400 μl of CTAB buffer (2% CTAB, 1.2 M NaCl, 100 mM TRIS-HCl, 20 mM EDTA and 0.2% β-mercapto ethanol) which was later discarded. Twenty μl of CTAB buffer was added back to each tube and individual insects homogenized with sterile blue Kontes pestles (Kimble Chase Life
Science Research, Vineland NJ). The buffer volume in each tube was brought to 600 μl with
CTAB buffer and tubes were incubated for 30 minutes in a 60° C water bath. Six hundred μl of chloroform:isosamyl alcohol (24:1) was added to each tube and tubes were inverted 20 times. After centrifugation at 12,000g for 15 minutes, the aqueous phase was transferred to a clean 1.5 ml microfuge tube and 600 μl of cold isopropanol was added to each tube. Samples were incubated overnight at -20° C and then centrifuged at 16,100g for 15 minutes. The supernatants were discarded and each pellet was washed with 1000 μl of 70% ethanol.
Samples were again centrifuged at 16,100g for 15 minutes, and again supernatants were discarded and pellets dried in a SpeedVac model SS1 (Savant Instruments, Inc. Farmingdale,
NY). DNA extracts were re-suspended in 50 μl of sterile distilled water and the quantity and quality of the extracted DNA was assessed by scanning 1.5 μl of each sample in a NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc. Waltham, MA). Extracted DNAs were stored at -80° C until used in polymerase chain reaction (PCR). Prior to quantitative (qPCR), all leafhopper DNA extracts were diluted to15 ng/μl.
Design of qPCR primers and calibration curves. Two AYp gene sequences, elongation 24 factor TU (tuf: GenBank accession AJ271323) and lysyl-tRNA synthetase (lysS: GenBank accession AJ271323), were used as AYp targets for amplification. Both gene targets were selected based on their performance in an evaluation of 10 candidate AYp gene targets (data not shown). Primers for these two genes consistently generated similar results and only primed the specified AYp targets of interest. The remaining 8 candidate primers were eliminated due to non-specific amplification of unknown targets in uninfected insects (i.e. possibly uncharacterized endosymbionts) which produced inconsistent results among primer sets. One aster leafhopper gene sequence, Wingless (cp6: GenBank accession FJ001411), was selected as a target for amplification of the leafhopper chromosomal DNA and served as a reference gene for relative quantification among samples with unknown quantities of AYp target sequence and also confirmed that the DNA extraction method being used produced consistent results among samples. All primers (Table 1) were designed using Beacon
Designer (PREMIER Biosoft International) and target sequence locations with significant structure were avoided.
To develop calibration curves of our AYp and leafhopper gene targets, PCR products of each target were generated with each of our designed primer pairs. Two μl of DNA extracts from AYp infected periwinkle or ALHs in varying concentrations were used as template for the reactions. Amplified PCR products were separated by gel electrophoresis, purified using activated silica beads (modified from Vogelstein and Gillespie 1979), and sequenced at the
University of Wisconsin-Madison Biotechnology Center DNA sequencing facility
(http://www.biotech.wisc.edu/). Using BLAST, PCR products were compared to the NCBI database and the original gene sequence used to design the primers. Products with the 25 appropriate sequences were cloned into a pGEM-T Easy plasmid (Promega, Madison, WI) according to the manufacturer’s instructions. Escherichia coli DH5α were transformed with the plasmid constructs containing target sequences. Transformants were selected using blue- white selection and assayed for the presence of the cloned target using PCR. Plasmids were purified from overnight cultures of E. coli transformants using a QIAprep Spin Miniprep Kit
(Qiagen) and standard plasmids were termed pCP6-5, pTUF-2, and pLysS-4. Purified plasmid preparations were quantified using a NanoDrop spectrophotometer and DNA concentrations
(expressed as plasmid copy number/μl) were calculated as follows assuming the average weight of a nucleotide base pair was 660 Daltons:
DNA (copies/μl) = DNA (ng/μl) / ((DNA (bp) * 1x109 (ng/g) * 660 (Da/bp)) / 6.022x1023
(copies/mol))
Two independent 10-fold dilution series for each gene target based on plasmid copy number/μl were prepared in concentrations ranging from 108 to 10 copies/μl. All standards were diluted in 0.1 X TE buffer (pH = 8.0). Initially, the analytical sensitivity of our primers was evaluated using the full dilution series and primer efficiency (E), based on standards ranging from 106 to 10 copies/μl, was calculated as:
E = 10-1/slope of dilution curve
To evaluate the variability associated with calibration curve preparation, four total reactions of each standard (i.e. two technical replicates per concentration for each independently prepared calibration curve) ranging from 106 to10 copies/ μl were run on a single plate (data analysis described below).
All experimental 96-well plates with samples containing unknown amounts of AYp 26 target included a 10-fold dilution series of standards that ranged from 106 to 10 copies/μl for quantification purposes. Primer efficiencies were calculated as described above and plate-to- plate variability of primer efficiency was characterized (see data analysis).
PCR and qPCR conditions. The presence of phytoplasma in plant and insect tissue extracts was detected using PCR or nested PCR. Nested PCR reactions were performed as described by Lee et al. (2006) with minor modifications. The first-round of amplification utilized universal phytoplasma primers P1 and 16sSr. Reactions were carried out in 25μl of 1X GoTaq
Green Master mix (Promega, Madison, WI, USA) containing 1 μM of each forward and reverse primer, two μl of DNA extract in varying concentrations as template and water. For a nested amplification, one μl of the diluted (1:30) PCR product from the first round of amplification was used as template and the universal phytoplasma primers R16F2n and
R16R2 were used in the PCR mixture. Briefly, the reactions were denatured at 94°C for 10 minutes, followed by 38 cycles of 94°C for 1 minute, annealing at 55°C for 2 minutes, and extension at 72°C for 3 minutes. The last cycle was followed by a final extension of 10 minutes at 72°C, and held at 4°C. Reaction conditions for the AYp specific primer set F4 and
R1 (Davis and Lee 1993), were the same as above with slight changes to the thermo cycler program. Reactions were denatured at 94°C for 2 minutes followed by 30 cycles of 94°C for
30 seconds, annealing at 55°C for 30 seconds, and extension at 72°C for 1 minute. The last cycle was followed by a final extension of 10 minutes at 72°C and held at 4°C. Positive and negative controls were run with each set of reactions and all reactions were conducted in a
MyiQ Cycler thermo cycler (Bio-Rad Laboratories, Inc. Hercules, CA). Ten μl of each PCR product was subjected to electrophoresis in a 1.2% agarose gel, stained with ethidium bromide 27 and visualized under UV light. Positive detections resulted in the production of a 1600 bp amplicon (P1 and 16sSR), a 1200 bp amplicon for the nested reaction (R16F2n and R16R2) or a 660 bp amplicon for primers F4 and R1.
Real-time qPCR reactions were performed in iQ 96-well PCR plates sealed with
Micro-Seal 'B' film (Bio-Rad Laboratories, Inc. Hercules, CA). Each reaction contained 1X iQ SsoFastTM EvaGreen® Supermix (Bio-Rad Laboratories, Inc. Hercules, CA), 0.2 μM of each forward and reverse primer (Table 1), 8 μl of each leafhopper extract (at 15 ng DNA/μl) or 8 μl plasmid standard (at varying copies/ μl), and sterile distilled water for a final reaction volumes of 20 μl. DNA extracts containing unknown quantities of AYp or leafhopper DNA were assayed in triplicate (3 technical replicates) and plasmid DNA standards at concentrations ranging from 106 to 10 copies/ μl (or 8.0 x 106 to 80 copies per reaction) were assayed in a total of four reactions (4 technical replicates, two from each duplicate dilution series). The reactions were denatured at 94°C for 10 minutes followed by 40 cycles of 94°C for 30 seconds and annealing at 55°C for 30 seconds. The last cycle was followed by a final extension of 10 minutes at 72°C, and a melt curve was created by increasing the temperature from 65°C to 95°C by 0.5 degree increments per 10 seconds. Real-time PCR reactions were conducted in a MyiQ instrument fitted with a One-Color Real-Time PCR Detection System
(Bio-Rad Laboratories, Hercules, CA).
Aster Leafhopper Weights. The average leafhopper weight was estimated to better understand if differences between AYp or ALH target concentrations of the ALH sexes was due to DNA yield per male and female insect or, in the case of AYp, differential growth of the phytoplasma in male versus female ALH. A random selection of 50 adult male and 50 adult 28 female leafhoppers was collected from the leafhopper colony. Carbon dioxide was used to immobilize the leafhoppers and insects were weighed in batches of 5 individuals on an analytical balance (Mettler-Toledo International Inc., Switzerland). The average weight per insect for each batch was calculated and the average weight for ALHs was reported as the mean of the ten batches for both male and female leafhopper groups.
Time course experiment. To assess the potential utility of our qPCR method, we produced a population of leafhoppers with varying AYp titer. The leafhopper population was created by providing a group of 200 ALH a 48-hour AAP on an AYp-infected aster after which time the insects were removed from the acquisition host and placed on rye seedlings to allow for the propagation of the phytoplasma within the leafhoppers. The rye plants, on which incubation occurred, were grown in 10 cm circular pots and covered with a mesh-top, cylindrical plastic tube and held in an environmental cabinet under similar light and temperature as the ALH culture (see above). Following the AAP, the first group of 15 ALHs was sampled immediately
(0 days post AAP) and subsequent sampling occurred on 1, 2, 3, 5, 7 and 9 days post-AAP. An additional group of 15 leafhoppers not exposed to an AYp-infected aster served as a control group. At sampling, individual insects were placed into 1.5 ml microfuge tubes, sex was determined, and leafhopper specimens were stored at -80° C until DNA extraction (described above). The presence of phytoplasma in ALHs was confirmed using a single round of amplification with the primer sets P1/16sSr, followed by a second round of amplification with
R16F2n/R16R2.
Data analysis. Data export and organization. Quantification cycles (Cq) were calculated automatically by the MyiQ Optical System Software (Version 1.0, BioRad Laboratories, 29
Hercules, CA) and reports with relevant data (i.e. Cq values, starting quantities, starting quantities in unknown samples and etc.) were exported as text (*.txt) files. Exported data were concatenated in a single spreadsheet, combined with the biological data recorded for each sample, and imported into R version 2.11.1 (R Development Core Team 2009) for statistical analysis. Unless specified, all functions used in the analysis can be found in the base distribution of R and are italicized in the text.
Standard curve repeatability. A linear regression approach (i.e. ANCOVA) was used to examine the variability associated with standard curve preparation and test the hypothesis that slopes and intercepts of different standard curve preparations were equivalent (Burns et al.
2005). The intent of this analysis was to determine the repeatability of standard curve preparation. Linear models were fit in R using the lm function and test statistics were extracted using the anova function.
A linear mixed effects model was used to examine the plate-to-plate variability of the standard curves run on the plates containing samples for quantification. This model
th corresponded to the simple linear regression of the estimated Cq value (Yij) in the j reaction
th within the i plate on the known starting concentration of the reaction (xij) and can be formulated as:
Yij = (B0 + b0i) + (B1 + b1i) xij + εij
2 2 2 b0i ~ N (0, σi ); b1i ~ N (0, σs ); εij ~ N (0, σr ).
In this model, B0 and B1 are the fixed effects parameters corresponding to the intercept and slope. b0i and b1i are the random effects vectors assumed to be independent and normally distributed for different 96-well plates and εij are the within-plate errors assumed to be 30 independent of the random effects. The intent of this analysis was to examine the overall, or average, slope and intercept values of the population of five independent 96-well plates (for each primer target) and to explicitly examine plate-to-plate variability associated with our standard curves. The homogeneity of variance of our standard curves throughout the target concentration range was evaluated by examining the residual plots of the mixed models (not shown). We also examined the within-plate Cq value variances averaged for the five plates and the standard deviation of the average variation (among plates) for each reaction containing different known starting quantities of standard. Linear mixed models were fit in R using the lme function (Package nlme: Pinheriro and Bates 2000) and parameter estimates and test statistics were extracted using the summary and anova functions.
Insect samples. Calibration curves were run on each 96 well plate and the same standard curve preparation (for each primer set) was used for all experimental (time course) insect samples. Thus, estimates of starting quantities of the AYp targets in the experimental samples were adjusted for plate-to-plate variability and estimated starting quantities could be compared among all 96-well reaction plates. The estimated copy number for each of the three technical replicates were aggregated for each leafhopper and average copy number for each insect was used for further calculations and statistical analysis. An AYp positive detection was defined as having a copy number 3s or greater than the mean copy number of the control group of ALHs for each qPCR primer set.
A simple logistic regression model was used to evaluate the effect of AYp concentration (Xi) on the ability to detect AYp using conventional and nested PCR (Kutner et al. 2004). The outcomes of one hundred conventional and nested PCR reactions of ALH DNA 31 extracts with varying starting copy numbers of AYp were used for this analysis. The results of
PCR and nested PCR were coded in binary fashion (Y = 1 if the PCR resulted in a visually detectable band under UV light in an agarose gel and Y = 0 if there was no visible band) and initial AYp copy number was estimated using our qPCR results. The standard logistic regression model has the form:
Yi = exp(Zi) / (1 + exp(Zi))
where Zi = B0 + B1 Xi
This equation can be rearranged and the slopes (B1) and intercepts (B0) of the regression models are more easily interpreted in the context of the following regression equation:
Ln (Yi/1-Yi) = B0 + B1 Xi where the odds of a positive detection is Yi/1-Yi and Xi is the starting copy number (Log10) in the PCR reaction. For each regression model, the number of copies necessary to have the probability of a positive detection (Y) be 0.5 (similar to ED50) is estimated by substituting 0.5 for Y and solving for X, the number of AYp copies (log-transformed). The logistic regression models were fit in R using the glm function with a logit link function (family binomial)
(package MASS: Venables and Ripley 2002). Parameter estimates, (chi-squared) test statistics, model predictions and ED50 estimates (± SE) were extracted using the summary, anova, predict, and dose.p functions.
The titer of AYp present in individual ALH extracts was estimated assuming one AYp organism possessed a single copy of the target DNA and time was considered a categorical covariate. A linear model was then used to test the null hypothesis that there were no differences of AYp titer between sexes or among ALH that underwent different incubation 32 times. Linear models were fit in R using the lm function and differences among the mean length of incubation time were determined using Tukey's 'Honest Significant Difference' method (function TukeyHSD). Parameter estimates and test statistics were extracted using the summary and anova functions. When data were unbalanced, the drop1 function was used to extract the marginal sum of squares for model factors or the effect of a factor conditional on all other terms entering the model first.
To better define the population dynamics of AYp in ALH, iteratively reweighted nonlinear regression analysis was used to fit a 3 parameter logistic growth model to absolute
AYp titer expressed as Log10-transformed copies per insect (N) and relative AYp titer expressed as Log10-transformed AYp gene copies per cp6 gene copies (R) averaged by incubation time. The logistic growth equation was:
Yi = Φ/ ( 1 + exp((((4*μ)/Φ)*(λ – Xi)) + 2) where Yi is AYp titer relative to AYp titer at time 0 (i.e. Ni – N0 or Ri – R0), Xi was incubation time and i indexes incubation time. Parameters Φ, μ, and λ represent the asymptote, the maximum growth rate, and the lag time associated with bacterial population growth
(Zwietering et al. 1990). Nonlinear models were fit using the nls function with start values estimated from visual examination of the plotted data points. Parameter estimates, standard errors, and test statistics were extracted using the summary function and pseudo-R2 values
(pseudo-R2 = 1 – (variance of residuals / total variance))) were used as a measures of goodness of fit (Schabenberger and Pierce 2002).
Finally, Pearson’s correlation coefficient and simple linear regression was used to compare estimates of AYp copy number per µl (Log10) of tuf and lysS in each ALH DNA 33 extract. The lm, cor.test, anova and summary functions were used to complete this analysis.
Results
Calibration curve preparation for absolute quantification of AYp. Two independent standard curves were prepared for each of the three primer sets shown in Table 1 using serial dilutions of pCP6-5, pTUF-2, and pLysS-4 plasmid. For each standard curve, regressions of the Cq value on the standard starting quantity showed that a significant proportion of the variability in quantification cycle could be predicted by the starting quantity (Figure 1). All regression slopes were significantly different than zero with coefficients of determination for all regressions greater than 0.99 (Table 2; Figure 1). Additionally, there was no significant effect of dilution series preparation on the slope (Table 2: B1 x Standard Preparation) or intercepts (Table 2: Standard Preparation) of the best fit lines for each primer set. All plasmid standards examined together, the standard deviation of the mean Cq value increased with increasing dilution of the plasmid targets (t = -3.8; df = 16; P < 0.005, R2 = 0.48). However, when 80 or more copies of plasmid template was used per reaction, the precision of the technical replicates remained high with the standard deviation among Cq values for technical replicates averaging 0.40 (± 0.17), 0.25 (± 0.10), and 0.42 (± 0.17) for CP6, lysS and tuf gene targets, respectively.
Intra- and inter-assay variability. A linear mixed effects model was used to examine the plate-to-plate variability of the standard curves run on each of the experimental plates containing AYp DNA extracted from leafhopper samples. The typical, or average, slope (B1) and intercept (B0) values for each primer target run on 5 independent 96-well plates did not vary (i.e. standard errors of the fixed effects were all less than 0.05) and average efficiencies 34 for our primer sets were 1.99, 1.86 and 1.92 for cp6, lysS and tuf genes, respectively (Table
3). Plate-to-plate variability of the slopes (σs) was low when compared to the magnitude of the slope estimates at 1%, 0% and 3% of the magnitude of the typical slope (B1) of the cp6, lysS and tuf gene primer sets, respectively. Similarly, the plate-to-plate variability of the intercepts
(σi) was 2.7%, 2.5% and 4.6% of the magnitude of the typical intercept estimates (B0). The residual variability (σr) of 0.22, 0.27 and 0.32 for pCP6-5, pLysS-4, and pTUF-2 represents the within plate error and other experimental error. These values can be interpreted as a measure of variation due to technical replication and can be used as an additional measure of calibration curve repeatability similar to the average variance of Cq values of pCP6-5, pLysS-
4 and pTUF-2 standards (Table 3). In general, the precision of the technical replicates remained high with an average variance of the Cq values of 0.18 (± 0.05) when 80 copies of plasmid template was used per reaction (all primer sets together). Although the variance of the mean Cq value increased with increasing dilution of the plasmid targets (t = 4.53; df = 88; P <
0.001).
Aster leafhopper samples. Of the 200 aster leafhoppers used in this experiment, 15 were used as a control group and 185 were allowed access to an AYp-infected plant. Due to mortality, only 93 live ALHs that had access to the AYp source plant were recovered resulting in a total of 108 ALH DNA extracts, of which, one was not quantified because of poor DNA quality. In general, female ALH weighed more than male ALH and weights averaged 1.32 mg and 0.90 mg, respectively. Similarly, more total DNA was extracted from individual female insects (9278 ± 74 ng) than male insects (2956 ± 138 ng). The CTAB DNA extraction method yielded high quality DNA with 260/280 ratios averaging 2.11 (± 0.004) and 260/230 ratios 35 averaging 1.30 (± 0.03) for all ALH extracts. The average yields for our CTAB DNA extractions from ALH were higher than Chen et al. (2010), who reported yields of 2200 ng/mg tissue from CTAB extractions of western corn rootworm beetles (Diabrotica virgifera vergifera).
Of the 92 leafhoppers exposed to an AYp-infected plant, AYp was detected in 37 and
95 percent of the insects using the primer set P1/16sSr and R16F2n/R16R2 (nested in
P1/16sSr ), respectively (Table 4). AYp was detected in 73 and 82 percent of the ALH samples using the lysS, and tuf gene primer sets, respectively. For lysS and tuf gene sequence targets, the limits of detection calculated were 58 and 22 copies per reaction for lysS and tuf genes, respectively, and below the minimum range of our standard curve (80 copies/reaction).
For the average sized female (9280 ng) these detection limits correspond to 4485 and 1700 copies per insect for the lysS and tuf gene targets, respectively. For the average male (2960 ng) ALH these detection limits are reduced to 1430 and 543 copies per insect for the lysS and tuf gene targets, respectively.
Logistic regression was used to examine the relationship between the starting copy number in PCR reactions and the outcome of the conventional PCR assay (Table 5). Using a cutoff of 0.5, these models predicted the observed conventional PCR outcome 90% and 88% of the time when copy number was calculated using AYp concentration estimated from lysS and tuf gene primer sets, respectively. The copy number necessary to have a 0.5 probability of detecting AYp by conventional PCR was estimated to be (Log10) 3.86 (± 0.16) and 3.72 (±
0.16), respectively for both the lysS and tuf gene primer sets. For nested PCR, the logistic regression models predicted the observed outcomes 87% and 88% using copy number 36 estimates obtained from lysS and tuf gene primer sets, respectively. For nested PCR, the copy number necessary to have a 0.5 probability of detecting AYp was estimated to be (Log10) 1.78
(± 0.17) and 1.50 (± 0.16) corresponding to 60 and 32 copies per reaction.
AYp titer as a function of sex and time. No AYp was detected in the control group of ALHs
(those not having access to an AYp-infected plant) and the highest AYp titers were among individual insects that underwent longer incubation times (Table 6). Averaging over sex, AYp titer measured using the lysS gene primer set and expressed as copies per insect (Log10), ranged between 3.94 (± 0.18) to 6.34 (± 0.13) occurring at 1 and 7 days post acquisition.
However, the variability of AYp titer in individual leafhoppers was greater ranging from 3.68 to 7.02 total copies. Similar results were obtained using the tuf gene primer set with average titer ranging between 3.50 (± 0.07) to 6.27 (± 0.12) copies at 1 and 7 days after acquisition and variation among individual insects ranging from 3.19 to 6.93 copies.
In the subset of AYp-infected ALHs, when AYp concentration was expressed as copies per ng DNA (Log10) there was no significant interaction of sex and incubation time (Time x
Sex effect: tuf: F = 1.00; df = 5, 62; P = 0.42; lysS: F = 1.06; df = 4, 55 ; P = 0.39). Linear models were refit without the interaction term and AYp concentration differences were detected among incubation times (Time effect; tuf: F = 45.6; df = 6, 67; P < 0.001 ; lysS: F =
28.1; df = 6, 59; P < 0.001) and between male and female insects (Sex effect; tuf: F = 17.66; df = 1, 67; P < 0.001 ; lysS: F = 9.14; df = 1, 59 ; P < 0.005 ). AYp titer was approximately 3- fold higher in male ALH than in female ALH.
When AYp titer was expressed as copies per insect, which accounts for approximate insect size, there was no significant interaction of sex and incubation time (Time x Sex effect: 37 tuf: F = 1.17; df = 5, 62; P = 0.33; lysS: F = 0.86; df = 4, 55; P = 0.50). Again, linear models were refit without the interaction term. However, no differences of AYp concentration were detected between male and female insects (Sex effect; tuf: F = 0.56; df = 1, 67; P = 0.46; lysS:
F = 0.05; df = 1, 59; P = 0.83) and titer differed among incubation times (Time effect; tuf: F =
47.2; df = 6, 67; P < 0.0001; lysS: F = 29.5; df = 6, 59; P < 0.0001). Using 'Tukey's Honest
Significant Difference' method, infected insects could be grouped into approximately three categories based on their AYp titer per insect (Table 6).
To better define the population dynamics of AYp in potentially infective ALHs, iteratively reweighted nonlinear regression analysis was used to fit a 3-parameter logistic growth model to AYp titer (absolute copy number per insect and relative to a ALH chromosomal reference gene) as a function of time (Figure 2A). The non-linear models described the average AYp titer in aster leafhoppers over time with overall model fits having pseudo-R2 values of 0.97 and 0.98, for lysS and tuf genes, respectively. Estimated AYp lag
-1 times were 2.06 and 1.51 days and maximum growth rates were (Log10) 1.22 and 0.81 day for lysS and tuf genes corresponding to doubling times of approximately 355 and 535 minutes.
AYp titer increased approximately 100-fold and became asymptotic six days after completion of the AAP. Similar results were obtained when AYp titer was expressed relative to the number of ALH cp6 gene copies present in the sample (Figure 2B).
Comparison of lysS primer set to tuf primer set. The tuf and lysS gene primers sets were both designed to target gene sequences of AYp. As expected, the two primer sets produced similar estimates of target sequence concentration which, when compared, were significantly correlated (Pearson’s: R = 0.97, t = 37.5, df = 94, P < 0.001). Regression analyses showed that 38 a significant proportion of the variability in titer as measured by the tuf gene primer set could be predicted by estimates of AYp titer obtained using the lysS gene primer set (Figure 3:
Slope: B1 = 1.02; t = 37.5; df = 94; P < 0.0001; Intercept: B0 = 0.22; t = -0.22; df = 94; P <
0.0001). The slope of the regression line was not significantly different from one (t = 0.68; df
= 94; P = 0.25) and the y-intercept was significantly different from zero suggesting a constant bias may be present among AYp titers measured by the different primer sets.
Concentrations of cp6 (ALH) in insect DNA extracts. To ensure the accuracy of our DNA extraction methodology and ability to measure and dilute insect samples for AYp detection, the ALH cp6 gene target was amplified to be used as a reference chromosomal marker. When
CP6 concentration was expressed as copies per ng DNA (Log10), the set of values was normally distributed (Shapiro-Wilk: W = 0.99, P = 0.36, n = 107) with a mean of 2.70 (±
0.02). Differences were detected among insects undergoing different incubation times (Time effect; F = 2.69; df = 7, 92; P = 0.01) and between leafhopper sex (Sex effect; F = 8.75; df =
1, 92 ; P < 0.005 ) with DNA extracts from male insects having about 2-fold more copies of
CP6 than females insects per ng DNA (M: 2.99 ± 0.03; F: 2.61 ± 0.02). There was no interaction of sex and incubation time on cp6 gene sequence concentration (copies/ng DNA)
(Time x Sex effect: F = 0.93; df = 6, 92; P = 0.48).
With cp6 gene sequence concentration expressed as copies per insect, the number of cp6 copies in male ALHs was not different from female ALHs (M: 3,115,136 ± 270,081; F:
4,110,462 ± 171,458; Sex effect; F = 0.67, df = 1, 92, P = 0.41) but the number of cp6gene copies did differ among incubation times (Table 6: Time effect; F = 2.15, df = 7, 92, P < 0.05).
Again, there was no interaction of sex and incubation time on cp6 gene concentration 39
(copies/ng DNA) (Time x Sex effect: F = 0.58; df = 6, 92; P = 0.75).
Discussion
Several studies have reported the development of qPCR assays for the identification and quantification of mollicutes in their insect and plant hosts (Marzachi and Bosco 2005, Lee et al. 2006, Inoue et al. 2009, Lopes et al. 2009, Wenbin et al. 2008, Wenbin et al. 2009a,
Wenbin et al. 2009b, Yvon et al. 2009). To our knowledge, however, the current study is the first to report a method for directly quantifying AYp titer in the ALH (M. quadrilineatus
Forbes) and contributes to the growing body of research of phytoplasma replication in their insect host by describing the AYp growth pattern and titer variation among individual ALH.
One of the primary contributions of this study was to demonstrate qPCR as a reliable and accurate method for measuring AYp titer in ALHs and detecting differences in AYp titers among insect individuals. Because absolute quantification of AYp DNA was achieved by comparing the amplification of unknown amounts of an AYp target gene sequence to the amplification of a dilution series containing known starting quantities of the targets sequences cloned into a plasmid, the factors leading to variation of the calibration curves within and among experiments was evaluated to examine the capabilities and limitations of the method.
A critical factor that relates directly to the reproducibility of any qPCR assay (for absolute quantification) is the accurate measurement of the initial plasmid standard DNA concentrations, which we estimated using a spectrophotometer, and preparation of calibration curves (Rutledge and Cote 2003, Burns et al. 2005, Bustin et al. 2009, Montes-Borrego et al.
2011). Since we could not reliably use spectrophotometry to estimate the low DNA concentrations that exist in our calibration curves (they were below our spectrophotometer’s 40 limit of detection) we examined the variation that may occur among calibration curve preparations made using two independently prepared and measured plasmid standards.
Several studies have reported differences among slopes (Atallah et al. 2009) and slopes and intercepts (McNeil et al. 2004, Montes-Borrego et al. 2011) of different standard curve preparations and/or runs. We used an ANCOVA approach to compare the slopes and intercepts of two independently prepared standard curves, which demonstrated that the variation due to standard curve preparation was less than the variation due to the technical replication (or within plate variation) of the standard curves (Burns et al. 2005). Our result is similar to Montes-Borrego et al. (2011) in that the reproducibility of our standard curve was not affected by the origin of the plasmid standard or our ability to measure the plasmid standard and prepare a dilution series.
Intra- and inter-assay variation of standard curves has been examined using multiple methodologies. Specifically, Rutledge and Cote (2003) and Burns et. al. (2005) have discussed the use of calculating standard deviations among (and within) standard curve replicates and ANCOVA for the characterization of calibration curve variability, respectively.
In our study, we used a linear mixed model to simultaneously examine both intra- (within plate) and inter-assay (plate-to-plate) variation. Because this model associates common random effects to observations sharing the same level of a classification factor, it most accurately represents the covariance structure induced by the inherent 96-well plate groupings
(Pinheiro and Bates 2000). This approach allowed us to examine the typical slope (B1) and intercept (B0) values for standard curves run on a “population” of five independent 96-well plates (for each primer target). It also allowed us to characterize plate-to-plate variability 41 associated with the slope (σs) and intercept (σi) of our standard curve as well as the residual within-plate variability (σr). From these values (Table 3) we can approximate 95% confidence intervals for the typical slope and intercepts. For example, plate-to-plate variation in the slope of cp6 gene calibration curve corresponds to a standard deviation of 0.04 Cq-values per log10
(Copy Number), and as a result, slope values as low as (-3.35 – 2*0.04) = -3.43 or as high as -
3.27 Cq-values per log10 (Copy Number) among calibration curves run on different 96-well plates would be expected. For the cp6 gene primer set, this corresponds to the calculated efficiency of 98.9% ± 1.7% which is consistent with the range slopes that would be estimated by fitting individual linear regressions for each run (Murtaugh 2007) and consistent with the findings of Rutledge and Cote (2003). Few studies have reported the intra- and inter-assay variation of slope (i.e. efficiency) and intercept in this way and no guidelines exist to determine acceptable levels of slope variation for a qPCR assay. Formal tests to determine differences among slopes of calibration curve slopes are available, but are often not reported.
Currently, the amount of sample DNA added to each of our qPCRs is measured using spectrophotometry and standardized to 120 ng. We originally thought this standardization could be avoided if a chromosomal ALH gene target whose concentration (copies/ng DNA) among insect DNA extracts remained stable could be used as a reference to standardize our assay for all insect individuals (Marzachi and Bosco 2005, Wong and Medrano 2005, Bustin et al. 2009). However, female ALHs had fewer copies of the cp6 gene per ng of DNA (Log10) than male ALHs. This difference might be explained by the presence of RNA, since our DNA extraction did not include a ribonuclease step, or extracrhomosomal DNA in our insect extracts. For example, female ALHs are larger than male ALHs, having approximately twice 42 the volume and weighing 150% more than males. Therefore, female ALH’s cells and bodies likely contain more non-genomic DNA and RNA (i.e. mitochondrial DNA, endosymbiont
DNA, and etc.) than male ALH and, to our knowledge, the relative contributions of the different DNA and RNA sources to the total nucleotide pool have not been measured for male and female ALHs. We assumed that these sources of variability, including the presence of
RNA in our samples, occurred for every insect in our experiment and could be considered a consistent source of random variation. Thus our deduced copy number might be biased, but the relative differences among experimental treatments would not be expected to vary interactively with the occurrence of these errors because they were present in all samples at some average level.
Calculations of the absolute AYp titer on a per insect basis with and without an ALH reference chromosomal marker were equivalent; mathematically the units cancel to give the same result. To further evaluate the utility a ALH reference chromosomal marker, we directly compared the relative AYp titer, calculated with and without the use of the cp6 gene target
(Figure 2), and found that using the cp6 gene as a reference did not change the results of our statistical analysis. Parameter estimates for the nonlinear regressions were not different (test statistics not reported) although, visually, the use of the cp6 gene did improve the consistency among parameter estimates for AYp titer measured using the lysS and tuf gene targets. This result implies that the additional qPCR step to quantify an ALH reference chromosomal marker may not be needed if DNA extraction methods are consistent among samples and yield high quality DNA. In the future, the quality and concentration of DNA extracts should always be measured and reported to ensure the accuracy (or validity) of the qPCR assay 43 results. We will continue to use the ALH cp6 gene target a reference to study relative AYp growth in the ALH, and to identify experimental errors associated with DNA extraction or samples in which PCR inhibitors may be present. Relative titer maybe useful for studying
AYp growth in ALH or other AYp hosts because it is a ratio and can be scaled to a reference value or treatment (i.e. in our experiment – Time 0). However, absolute copy number is often needed to fully understand and interpret the biological relevance of data and for a majority of our analyses AYp titer was expressed in absolute terms as AYp copies per insect or per nanogram of DNA.
Since the AYp has not been successfully cultured in the laboratory and DNA extracts typically contain a background of host DNA, we were initially concerned that prokaryotic endosymbionts naturally colonizing the ALH might contain DNA sequences sufficiently conserved to cause false positives with a specific primer set. Therefore, multiple AYp gene targets were amplified to provide additional evidence for the analytical specificity of the assay. If estimates of target copy number obtained using primers designed to amplify two different single-copy AYp gene sequence targets were similar, then the assay was likely targeting the same organism. Estimates of AYp target in ALH DNA extracts obtained by amplifying two different AYp sequence targets in each DNA extract were significantly correlated and the slope of the regression line between the estimates was not significantly different from one. In addition to providing evidence for the specificity of the qPCR assay, this finding further supported the conclusion that plasmid standards could be accurately quantified and the repeatability of calibration curve preparation is high. The costs of future 44 experimentation could be significantly reduced by quantification of the AYp using a single primer set.
In the control ALH group, we did not detect phytoplasma in any insect extracts using nested PCR. The average Cq values for the control group were 36.40 and 35.87 for the lysS and tuf gene targets, respectively, which were higher than the Cq values obtained for the most dilute standard of the lysS and tuf gene calibration curves at 35.37 and 33.42, respectively. We interpreted the qPCR signal associated with the control group of insects as background noise that may occur from nonspecific amplification products in the SsoFast EvaGreen reactions.
Additional evidence supporting this interpretation came from an examination of the melt curves associated with the control group qPCR products. Melt peaks for this subset of reactions were poorly defined and of low amplitude. Thus, the background noise was used to define a diagnostic limit of detection (LOD) of three standard deviations above the mean starting copy number of the control group of ALHs. Individual insects were considered to be infected with AYp if the estimated copy number was greater than the LOD, which represented the clinical sensitivity of the assay. For insects undergoing shorter incubation times (0, 1 and
2 days), the percent positive detections using qPCR was higher than the percent positive detections when using conventional PCR but lower than the percent positive detections when using nested PCR. However, in insects undergoing longer incubation times (2 or more days), the percent positive detections were consistent between qPCR and nested PCR.
Many studies have compared the outcomes of conventional and nested PCR methods to qPCR methods for the detection of specific pathogen and the comparison of methodologies has largely been qualitative in nature (Crosslin et al. 2006, Wen et al. 2009, Wenbin et al. 45
2009, Zhang et al. 2010, Montes-Borrego et al. 2011). Because the outcome of (conventional) endpoint PCR depends on the starting copy number of the targeted sequence and a large number of unknown experimental variables (Freeman et al. 1999), a positive (or negative) detection can be thought of as having a probability distribution conditional on the initial copy number in the reaction. In this study, we used logistic regression to relate the initial copy number of AYp target present in each PCR reaction tube (estimated using qPCR) to the binary outcome of our conventional PCR and nested PCR assay outcomes (Kutner et al. 2004). We found that the starting copy number did relate significantly to the outcome of the conventional assay and approximately 6,000 and 125,000 copies, respectively, were necessary to have a 0.5 and 0.95 probability of detecting AYp from an environmental sample when a single round of amplification was used (38 cycles). These values dropped to approximately 40 and 850 when a second nested round of amplification was used for detection. Because, we did not control for numerous possible errors in our conventional PCR assays (i.e. conventional PCR was not replicated, reactions contained varying concentrations of DNA template, primer efficiencies not quantified and etc.), these estimates represent an approximate calculation of the copy numbers of target in a reaction necessary to give a positive detection with conventional PCR and should be refined with further research. The intent of this analysis was to point out that the comparison of molecular detection methods could be made more quantitative; it was not to evaluate the primers being used or assess our conventional PCR assays.
We also examined the variability of AYp titer in ALHs and characterized growth dynamics of AYp titer within a population of AYp-infected insects. Using indirect methods,
Sinha and Chiykowski (1967) first reported the phenology of AYp recovery from M. 46 quadrilineatus (formerly M. fascifrons) tissues after a three day AAP and found phytoplasmas in the ailementry canal, hemolymph, and salivary gland at, three, six and 12 days post-AAP, respectively. Previous studies have measured Chrysanthemum yellows phytoplasma titer in the insect vectors, Euscelis incisus Kirschbaum, Euscelidius variegatus Kirschbaum and
Macrosteles quadripunctulatus Kirschbaum after a 7 or 10 day AAP (Bosco et al. 2007a,
Bosco et al. 2007b). In those studies, the relatively long AAP and sampling intervals may have masked some of the finer scale phytoplasma population dynamics occurring in the insect. In our study, we found that AYp titer in ALH increased over a period of approximately five days post AAP and became asymptotic. We used a logistic growth curve model to estimate parameters that are commonly used to describe bacterial growth in culture
(Zwietering et al. 1990). We found that AYp had long doubling times, greater than 500 minutes, which is not uncommon for some pathogenic bacteria such as Mycobacterium tuberculosis (James et al. 2000). We also found that AYp titer increased approximately 100- fold in the insect, which is consistent with some of the original findings of Black (1941), but growth slowed after six days suggesting there is some upper limit to the AYp population size in the ALH. In the future, a consistent modeling approach of mollicute growth in their host would allow for easier comparison among multiple studies.
Under the assumption that an individual AYp organism possessed a single copy of target sequence, the average AYp titer per insect increased approximately 100-fold to 106.3 over 7 days. The number of AYp organisms present in an individual ALH, measured using tuf and lysS gene primer sets ranged from approximately 103.2 (i.e. the LOD for a single insect) to
107.0 over the course of the experiment and varied as much as 100-fold within an incubation 47 time group. These estimates are consistent with previous attempts to measure AYp titers in their insect vectors (Bloomquist and Kirkpatrick 2002). When expressed as copies per ng of
DNA, the highest AYp titer in an individual insect was 2.4 x 103 copies per ng at 7 days post
– AAP which is lower than previous reports of 3.1 x 104 copies per ng at 33 days post – AAP in the chrysanthemum yellows phytoplasma-leafhopper system (Marzachi and Bosco 2005).
For our experiment, this suggests that Ayp titers might have continued to increase if the ALH were given longer incubation periods.
Male ALHs had higher AYp titers than female ALHs when phytoplasma concentration was expressed as copies per ng of DNA. This difference might be explained by the fact that female insects are larger and have more tissues that are not susceptible to AYp infection, such as the fat bodies and mycetomes (Sinha and Chiykowski 1967), but still contribute to the total
DNA yield. However, when AYp titer was expressed on a per insect basis, females and males harbored approximately the same number of AYp organisms. Similar to our findings,
Rotenberg et al. (2009) reported that male thrips harbor higher TSWV concentrations per unit
RNA but female thrips harbor more molecules of TSWV on a per insect basis. In that study and in our study, the quantitative differences of pathogen load between sexes was likely due to the relative size differences of pathogen susceptible tissues between male and female insects. Rotenberg et al. (2009) also observed that male thrips transmit TSWV more frequently than female thrips even though males harbor fewer virus molecules than females.
The authors hypothesized that the higher transmission efficiency of male thrips was likely due to feeding behavior and not virus titer. Similarly, Beanland et al. (1999) has reported differences in the ability of male and female ALHs to transmit the AYp, with females more 48 likely to transmit than male insects. However, it is not known if the differences in transmission ability between male and female ALH were due to AYp titer, ALH feeding behavior or AYp distribution within the insect body, which is known to vary in concentrations among tissues of some insect vectors (Galetto et al 2009).
The biotic and abiotic factors that influence the variation of AYp titer in ALH have not been well characterized. For example, temperature is known to influence the latent period of
AYp-infected ALH (Maramorosch 1953, Murral et al. 1996), but the underlying mechanism for the temperature effect is not known and may simply be that AYp organisms grow more slowly in the insect at lower temperatures. It is the genetic composition of the ALH and AYp that determine vector competency and there may be a genetic basis for variation of AYp titer in ALH. Multiple AYp strains exist in the environment (Lee et al. 2003, Zhang et al. 2004) which differentially affect ALH fitness and have the potential to alter ALH population dynamics in the field (Beanland et al. 2000). Again, the mechanism for the fitness effect of
AYp on the ALH is not known but may be related to phytoplasma growth in the ALH. The use of qPCR as a tool may be applied to address the biological relevance of AYp variability and growth within insect individual and within and among ALH populations. To date, AYp titer variation has not been described within or among field caught ALH populations and the relationship between AYp titer variation and ALH infectivity has not been established.
However, the existence of AYp titer variability among ALHs and having the tools to manipulate and measure that variability is necessary for completing experiments to relate
AYp titer to a leafhopper’s ability to transmit.
Acknowlegements. We thank Thomas German at the University of Wisconsin, for the use of 49 his laboratory space and equipment. We thank Emily Mueller and Thomas German for their constructive criticism and comments on earlier versions of the manuscript. Funding support was provided by the USDA Specialty Crops Research Initiative through the Wisconsin
Specialty Crop Block Grant Program (MSN 129013). 50
References cited
Alma A., Bosco, D., Danielli, A., Bertaccini, A., Vibio, M., and A. Arzone. 1997. Identification of phytoplasmas in eggs, nymphs and adults of Scaphiodeus titanus Ball reared on healthy plants. Insect Mol. Biol. 6:115-121.
Atallah, Z.K., Bae, J., Jansky, S.H., Rouse, D.I., and W.R. Stevenson. 2007. Multiplex real-time quantitative PCR to detect and quantify Verticillium dahliae colonization in potato lines that differ in response to verticillium wilt. Phytopathology 97:865-872.
Beanland, L., Hoy, C.W., Miller, S.A., and L.R. Nault. 1999. Leafhopper (Homoptera: Cicadellidae) transmission of aster yellows phytoplasma: does gender matter? Environ. Entomol. 28(6):1101-1106.
Beanland, L., Hoy, C.W., Miller, S.A., and L.R. Nault. 2000. Influence of aster yellows phytoplasma on the fitness of aster leafhopper (Homoptera: Cicadellidae). Ann. Entomol. Soc. Am. 93(2):271-276.
Black, L.M. 1941. Further evidence for multiplication of the aster-yellows virus in the aster leafhopper. Phytopathology 31:120-135.
Bloomquist, C.L. 2002. Aster yellows and beet leafhopper transmitted virescence agent yellows. Pages 58-59 in: Compendium of Umbelliferous crop diseases.
Bloomquist, C.L. and B.C. Kirkpatrick. 2002. Frequency and seasonal distribution of pear psylla infected with the pear decline phytoplasma in California pear orchards. Phytopathology 92:1218-1226.
Bosco, D., Galletto, L., Leoncini, P., Saracco, P., Raccah, B., and C. Marzachi. 2007a. Interrelationships between “Candidatus phytoplasma ateris” and its leafhopper vectors (Homoptera: Cicadellidae). J. Econ. Entomol. 100(5): 1504-1511.
Bosco, D., Galletto, L., Leoncini, P., Saracco, P., Raccah, B., and C. Marzachi. 2007b. Patterns of chrysanthemum yellows phytoplasma multiplication in three leafhopper vector species (Cicadellidae Deltocephalinae). B. Insectol. 60(2): 227-228.
Burns, M.J., Nixon, G.J., Foy, C.A. and N. Harris. 2005. Standardisation of data from real- time quantitative PCR method – evaluation of outliers and comparison of calibration curves. BMC Biotechnol. 5:31.
Bustin S.A., Benes, V., Garson, J.A., Hellemans, J., Huggett, J., Kubista, M., Mueller, R., Nolan, T., Pfaffl, M.W., Shipley, G.L., Vandesompele, J., and C.T. Wittwer. 2009. The MIQE guidelines: minimum information for publication of quantitative real-time 51
PCR experiments. Clin. Chem. 55(4):611-622.
Chen, H., Rangasamy, M., Tan, S.Y., Wang, H. and B.E. Siegfried. 2010. Evaluation of five methods for total DNA extraction from western corn rootworm beetles. PLoS ONE 5(8):e11963.
Chiykowski L.N. and R.K. Chapman. 1965. Migration of the six-spotted leafhopper in central North America. Wisconsin Agricultural Experiment Station Research Bulletin 261: 21-45.
Chiykowski L.N. 1965. The reaction of barley varieties to aster yellows virus. Can. J. Bot. 43:373-378.
Chiykowski L.N. 1967. Reaction of some wheat varieties to aster yellows virus. Can. J. Plant Sci. 47:149-151.
Christensen, N.M., Axelsen, K.B., Nicolaisen, M. and A. Schultz. 2005. Phytoplasmas and their interactions with hosts. Trends Plant Sci. 10:526-535.
Crosslin, J.M., Vandemark, G.J., and J.E. Munyaneza. 2006. Development of a real-time, quantitative PCR for detection of the Columbia Basin potato purple top phytoplasma in plants and beet leafhoppers. Plant Dis. 90:663-667.
Davis, R.E. and I.-M. Lee. 1993. Cluster-specific polymerase chain reaction amplification of 16S rDNA sequences for detection and identification of mycoplasmalike organisms. Phytopathology 83(9):1008-1011.
Doyle, J.J. and E.E. Dickson. 1987. Preservation of plant samples for DNA restriction endonuclease analysis. Taxon 36(4): 715-722.
Drake D.C. and R.K. Chapman. 1965. Evidence for long distance migration of the six- spotted leafhopper in Wisconsin. Wisconsin Agricultural Experiment Station Research Bulletin 261: 1-20.
Fletcher, J., Wayadande, A. Melcher, U., and F. Ye. 1998. The phytopathogenic mollicute- insect vector interface: a closer look. Phytopathology 88:1351-1358.
Freeman, W.M., Walker, S.J., and K.E. Vrana. 1999. Quantitative RT-PCR: pitfalls and potential. Biotechniques 26:112-125.
Galetto, L., Nardi, M., Saracco, P., Bressan, A., Marzachi, C., and D. Bosco. 2009. Variation in vector competency depends on chrysanthemum yellows phytoplasma distribution within Euscelidius variegatus. Entomol. Exp. Appl. 131: 200-207. 52
Hollingsworth, C.R., Atkinson, L.M., Samac, D.A., Larsen, J.E., Motteberg, C.D., Abrahamson, M.D., Glogoza, P., and I.V. MacRae. 2008. Region and field level distributions of aster yellows phytoplasma in small grain crops. Plant Dis. 92:623-630.
Hoy, C.W., Heady, S.E., and T.A. Koch. 1992. Species composition, phenology, and possible origins of leafhoppers (Cicadellidae) in Ohio vegetable crops. J. Econ. Entomol. 85:2336-2343.
Inoue, H., Ohnishi, J., Ito, T., Tomimura, K., Miyata, S., Iwanami, T., and W. Ashihara. 2009. Enhanced proliferation and efficient transmission of Candidatus Liberibacter asiaticus by adult Diaphorina citri after acquisition feeding in the nymphal stage. Ann. Appl. Biol. 155:29-36.
IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group. 2004. 'Candidatus Phytoplasma', a taxon for the wall-less, non-helical prokaryotes that colonize plant phloem and insects. J. Syst. Evol. Microbiol. 54:1243-1255.
James, B.W., Williams, A., and P.D. Marsh. 2000. The physiology and pathogenicity of Mycobacterium tuberculosis under controlled conditions in a defined medium. J. Appl. Microbiol. 88:669-677.
Kawakita, H., Siki, T., Wei, W., Mitsuhashi, W., Watanabe, K., and M. Sato. 2000. Identification of mulberry dwarf phytoplasmas in the genital organs and eggs of leafhopper Hishimonoides sellatiformis. Phytopathology 90:909-914.
Kunkle, L.O. 1926. Studies on aster yellows. Am. J. Bot. 13(10):646-705.
Kutner, M.H., Nachtsheim, C.J. and J. Neter. 2004. Applied linear regression models (4th edition). The McGraw-Hill Companies, Inc. New York, New York 10020.
Kwon, M.-O., Wayadande, A.C., and J. Fletcher. 1999. Spiroplasma citri movement into the intestines and salivary glands of its leafhopper vector, Circulifer tenellus. Phytopathology 82(12):1144-1151
Lee, I.-M., Bottner, K.D., Munyaneza, J.E., Davis, R.E., Crosslin, J.M., du Toit, L.J., and T. Crosby. 2006. Carrot purple leaf: a new spiroplasmal disease associated with carrots in Washington State. Plant Dis. 90:989-993.
Lee, I.-M., Davis, R.E., and D. E. Gunderson. 2000. Phytoplasma: Phytopathogenic mollicutes. Annu. Rev. Microbiol. 54:221-255.
Lee, I.-M., Gundersen-Rindal, D.E. and A. Bertaccini. 1998. Phytoplasma: ecology and 53
genomic diversity. Phytopathology 88(12):1359-1366.
Lee, I.-M., Martini, M., Bottner, K. D., Dane, R.A., Black, M.C., and N. Troxclair. 2003. Ecological implications from a molecular analysis of phytoplasmas involved in an aster yellows epidemic in various crops in Texas. Phytopathology 93:1368-1377.
Lee, P.E. and A.G Robinson. 1958. Studies on the six-spotted leafhopper Macrosteles fascifrons (Stal) and aster yellows in Manitoba. Can. J. Plant Sci. 38:320-327.
Lopez, S.A., Bertolini, E., Frare, G.F, Martins, E.C., Wulff, N.A., Teixeira, D.C., Fernandes, N.G. and M. Cambra. 2009. Graft transmission efficiencies and multiplication of ‘Candidatus Liberibacter americanus’ and ‘Ca. Liberibacter asiaticus’ in citrus plants. Phytopathology 99:301-306.
Liu, H.W., Goodwin, P.H., and C.R. Kuske. 1994. Quantification of DNA from the aster yellows mycoplasmalike organism in aster leafhoppers (Macrosteles fascifrons Stal) by a competitive PCR reaction. Syst. Appl. Microbiol. 17:274-280.
Mahr, S.E.R. 1989. Development and aster yellows-infectivity of the aster leafhopper. Ph.D. Thesis, University of Wisconsin, Madison.
Maramorosch, K. 1952a. Direct evidence for the multiplication of aster-yellows virus in its insect vector. Phytopathology 42(2):59-64.
Maramorosch, K. 1952b. Multiplication of aster-yellows virus in its vector. Nature 168:194- 195.
Maramorosch, K. 1953. Incubation period of aster-yellows virus. Am. J. Bot. 40:797-809.
Maramorosch, K. 1955. Relative contents of aster-yellows virus in inocula of vectors during four serial transfers. Am. J. Bot. 42:676-678.
Matthews, R.E.F. 1991. Plant Virology 3rd edition. Academic Press, Inc. San Diego, CA 92101.
Marzachi, C. and D. Bosco. 2005. Relative quantification of chrysanthemum yellows (16Sr I) phytoplasma in its plant and insect host using real-time polymerase chain reaction. Mol. Biotechnol. 30: 117-127.
McNeil, M., Roberts, A.M.I., Cockerell, V. and V. Mulholland. 2004. Real-time PCR assay for quantification of Tilletia caries contamination of UK wheat seed. Plant Pathol. 53:741-750.
54
Montes-Borrego, M., Munoz-Ledesma, F.J., Jimenez-Diaz, R.M., and B.B. Landa. 2011. Real-time PCR quantification of Peronospora arborescens, the opium poppy downy mildew pathogen, in seed stocks and symptomless infected plants. Plant Dis. 95:143- 152.
Murral, D.J., Nault, L.R., Hoy, C.W., Madden, L.V., and S.A. Miller. 1996. Effects of temperature and vector age on transmission of two Ohio strains of aster yellows phytoplasma by the aster leafhopper (Homoptera: Cicadellidae). J. Econ. Entomol. 89(5): 1223-1232.
Murtaugh, P.A. 2007. Simplicity and complexity in ecological data analysis. Ecology 88(1):56-62.
Peterson, A.G. 1973. Host plant and aster leafhopper relationships. Proc. North Cent. Branch Entomol. Soc. Am. 28:66-70.
Pinheiro, J.C. and D.M. Bates. 2000. Mixed-effects models in S and S-plus. Springer-Verlag New York, Inc. New York, New York 10010.
Rahardja, U., Whalon, M.E., Garcia-Salazar, C. and Y.T. Yan. 1992. Field detection of X- disease mycoplasma-like organism in Paraphlepsius Irroratus (Say) (Homoptera: Cicadellidae) using a DNA probe. Environ. Entomol. 21:81-88.
R Development Core Team (Version 2.11.1). 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/.
Rotenberg, D., Kumar, N.K.K., Ullman, D.E., Montero-Astua, M., Willis, D.K., German, T.L. and A.E. Whitfield. 2009. Variation in tomato spotted wilt virus titer in Frankliniella occidentalis and its association with frequency of transmission. Phytopathology 99:404-410.
Rutledge, R.G. and C. Cote. 2003. Mathematics of quantitative kinetic PCR and the application of standard curves. Nucleic Acids Res. 31(16):e93
Schabenberger, O. and F. J. Pierce. 2002. Contemporary Statistical Models for the Plant and Soil Sciences. CRC Press, Boca Raton, FL.
Schultz, G.A. 1979. Epidemiology of the aster leafhopper and aster yellows in relation to disease control. Ph.D. Thesis, University of Wisconsin, Madison.
Sinha, R.C. and L.N. Chiykowski. 1967. Initial and subsequent sites of aster yellows virus infection in a leafhopper vector. Virology 33:702-708. 55
Venables, W.N. and B. D. Ripley. 2002. Modern applied statistics with S (4th edition). Springer-Verlag New York, Inc. New York, New York 10010.
Vogelstein, B. and D. Gillespie. 1979. Preparative and analytical purification of DNA from agarose. Proc. Natl. Acad. Sci. U.S.A. 76(2):615-619.
Wallis, R.L. 1962. Host plant preference of the six-spotted leafhopper. J. Econ. Entomol. 55:998-999.
Wen, A., Mallik, I., Alvarado, V.Y., Pasche, J.S., Wang, X., Li, W., Levy, L., Scholthof, H.B., Mirkov, T.E., Rush, C.M., and N.C. Gudmestad. 2009. Detection, distribution, and genetic variability of ‘Candidatus Liberibacter’ species associated with zebra chip complex disease of potato in North America. Plant Dis. 93:1102-1115.
Wenbin, L., Abad, J.A., French-Monar, R.D., Rascoe, J., Wen, A., Gudmestad, N.C., Secor, G.A., Lee, I.-M., Duan, Y., and L. Levy. 2009. Multiplex real-time PCR for detection, identification and quantification of ‘Candidatus Liberibacter solanacearum’ in potato plants with zebra chip. J. Microbiol. Meth. 78:59-65.
Wenbin, L., Levy, L., and J.S. Hartung. 2009. Quantitative distribution of ‘Candidatus Liberibacter asiaticus’ in citrus plants with citrus Huanglongbing. Phytopathology 99:139-144.
Wenbin, L., Li, D., Twieg, E., Hartung, J.S., and L. Levy. 2008. Optimized quantification of unculturable Candidatus Liberibacter spp. in host plants using real-time PCR. Plant Dis. 92:845-861.
Westdal, P.H. and H.P. Richardson. 1969. The susceptibility of cereals and wild oats to an isolate of the aster yellows pathogen. Can. J. Botany 47:755-760.
Wong, M.L. and J.F. Medrano. 2005. Real-time PCR of mRNA quantitation. Biotechniques 39(1):75-85.
Yvon, M., Thebaud, G., Alary, R. and G. Labonne. 2009. Specific detection and quantification of the phytopathogenic agent ‘Candidatus Phytoplasma prunorum’. Mol. Cell. Probe. 23:227-234.
Zhang, J., Hogenhout, S.A., Nault, L.R., Hoy, C.W., and S.A. Miller. 2004. Molecular and symptom analyses of phytoplasma strains from lettuce reveal a diverse population. Phytopathology 94(8): 842-849.
Zhang, X., Zhou, G., and X. Wang. 2010. Detection of wheat dwarf virus (WDV) in wheat 56
and vector leafhopper (Psammotettix alienus Dahlb.) by real-time PCR. J. Virol. Methods 169:416-419.
Zwietering, M.H., Jongenburger, I., Rombouts, F.M. and K. van’t Riet 1990. Modeling of the bacterial growth curve. App. Environ. Microb. 56(6):1875-1881. 57
Tables and Figures.
Table 1. Designed qPCR primer set names, sequences, sources, efficiencies and product sizes. Primer Target Efficiency Product Name 5' Sequence 3' organism 1 Size cp6F GGGCAAGAAGGGCAAGTA Aster 2 2.01 91 cp6R AGGCTCCAGATACACTAGGTC Leafhopper
CCAGGTTCTGTTAAGCCTCATT Aster tufF C Yellows 1.97 198 AACTACTAATTCAGCGTTGTCA Phytoplasma tufR CC 3 lysSF CTTGAGAATTGCCACCGAATTG Aster Yellows GCATATCAGCATAAGCCAAGTA 1.97 158 lysSR Phytoplasma AG 3 1 Primer efficiencies were calculated as E = 10(-1/slope) and slopes were established from dilution curves of target DNA fragments. 2 Primers were designed from GenBank accession (FJ001411). 3 Primers were designed from GenBank accession (AJ271323).
58
Table 2. ANOVA tables for regression analysis evaluating the effect of independent preparation of dilution series standards. Ho: Slopes do not vary between Ho: Intercepts do not vary between calibration curve preparations calibration curve preparations Standard Parameter F-value P-valuea Parameter F-value P-valueb
pCP6-5 B1 2537 < 0.001 B1 2640 < 0.001 Standard Standard 0.06 0.81 0.06 0.80 Preparation Preparation B x Standard 1 0.18 0.68 Preparation
pTuf-2 B1 3493 < 0.001 B1 3549 < 0.001 Standard Standard 0.54 0.47 0.55 0.47 Preparation Preparation B x Standard 1 0.66 0.42 Preparation pLysS-4 B1 5247 < 0.001 B1 5336 < 0.001 Standard Standard 0.37 0.47 0.37 0.54 Preparation Preparation B x Standard 1 0.65 0.42 Preparation a P-values calculated using df = 1, 20 b P-values calculated using df = 1, 21
59
Table 3. Regression analysis of standard curves and standard deviations of Cq values for different concentrations of target DNA sequences on circular plasmids calculated from five 96-well plates per target Plate-to-plate Target Conc. Average Cq Target Parameters ± SEa variabilityb (cps/reaction)c Varianced (Std. Dev.) e pCP6-5 B1 = -3.35 (0.02)* σs = 0.04 8,000,000 0.04 (0.03)
B0 = 36.8 (0.45)* σi = 1.00 800,000 0.05 (0.03) σr = 0.22 80,000 0.02 (0.01) 8,000 0.03 (0.02) 800 0.03 (0.03) 80 0.15 (0.20) pLysS-4 B1 = -3.72 (0.01)* σs = 0.00 8,000,000 0.01 (0.01)
B0 = 39.2 (0.44)* σi = 0.98 800,000 0.03 (0.01) σr = 0.27 80,000 0.02 (0.01) 8,000 0.07 (0.03) 800 0.08 (0.05) 80 0.24 (0.19) pTUF-2 B1 = -3.52 (0.05)* σs = 0.11 8,000,000 0.01 (0.01)
B0 = 37.3 (0.77)* σi = 1.73 800,000 0.06 (0.04) σr = 0.32 80,000 0.03 (0.02) 8,000 0.04 (0.02) 800 0.09 (0.06) 80 0.16 (0.26) a Slopes (B1) and intercepts (B0), obtained from the linear mixed model can be interpreted as the result of a “typical” standard curve observed in our experiments. Fixed effects followed by * are significant at p < 0.0001 by an F-test df = 1, 112. b Standard deviation of the random effects (i.e. on the same scale as the fixed effects) and can be interpreted as a measure of the variability of the slope (σs) and intercept (σi) among plates and the residual, within-plate, variability (σr). c Target concentrations were calculated as described above and copy numbers were based on 8 ul of standard per reaction. d Average Cq variance can be interpreted as a measure of within plate variability of 4 PCR reactions on run on five 96-well plates while its SD represents a measure of the among plate variability. e Average primer efficiency can be calculated as E = 10(-1/B1) . 60
Table 4. Number of positive AYp detections (%) for aster leafhoppers undergoing different incubation times following a 48 hour acquisition access period in experiment 2. Primer sets used for AYp detection Incubation # of (Days) ALHs P1/16sSr R16F2n/R16R2na lysSF/lysSRb tufF/tufRb C 15 0 ( 0 ) 0 (0) 0 (0) 0 (0) 0 15 0 ( 0 ) 14 (93) 6 (40) 7 (60) 1 15 1 (7) 14 (93) 8 (53) 11 (73) 2 15 1 (7) 14 (93) 9 (60) 13 (87) 3 15 8 (53) 13 (87) 14 (93) 14 (93) 5 14 9 (64) 14 (100) 13 (93) 13 (93) 7 15 14 (93) 15 (100) 14 (93) 14 (93) 9 3 1 (33) 3 (100) 3 (100) 3 (100) a Second round of amplification nested in primer set P1/16sSr. b A positive detection had a value of 3s or greater than the mean of the control.
61
Table 5. Simple logistic regression analysisa of 100 detection outcomes of PCR and nested PCR reactions used to detect varying amounts of AYp in ALH DNA extracts Odds Predictor Estimate SE χ2 df Pr ( >| χ2|) Ratio Intercept - B0 -8.87 1.66 LysS - B 2.30 0.43 59.9 1 < 0.0001 9.97 PCR 1 Intercept - B0 -7.94 1.49 Tuf - B1 2.14 0.40 56.9 1 < 0.0001 8.50 Intercept- B0 -4.09 1.49 Nested LysS - B1 2.29 0.69 64.1 1 < 0.0001 9.87 PCR Intercept- B0 -3.66 1.27 Tuf - B1 2.44 0.68 55.3 1 < 0.0001 11.47 a “Logistic” regression model fit in R with glm function and arguments were set to binomial family with a logit link function.
62 62
Table 6. Mean copy number (± SE) of AYp and insect genome sequence targets, expressed as Log10 copies per insect (cp/in), for aster leafhoppers undergoing different incubation times following a 48 hour acquisition access period. Gene Targeta Incubation # of lysS # of tuf # of cp6 (Days) ALHb log(cp/in)c ALHb log(cp/in)c ALHs log(cp/in) C 15 NDd 15 ND 15 6.65 (0.05)A,B 0 6 3.94 (0.18)A 9 3.63 (0.16)A 15 6.53 (0.04)A,B 1 8 3.92 (0.07)A 10 3.50 (0.07)A 15 6.58 (0.05)A 2 9 4.16 (0.12)A 13 4.19 (0.12)B 15 6.58 (0.06)A,B 3 14 5.10 (0.17)B,C 14 4.79 (0.16)B 15 6.54 (0.05)A,B 5 13 5.73 (0.21)B,C 13 5.64 (0.19)C 14 6.51 (0.05)A,B 7 14 6.34 (0.13)C 14 6.27 (0.12)C 15 6.42 (0.07)B 9 3 5.96 (0.26)C 3 5.73 (0.32)C 3 6.58 (0.18)A,B a Titers for a specific gene target followed by the same letter are not different (Tukey’s HSD; P = 0.05). b Number of aster leafhoppers testing PCR positive and used in the mean calculation. A positive detection had a value of 3s or greater than the mean of the control (C). c Only ALH extracts testing positive for AYp were used in the mean comparisons of the titer estimates obtained from LysS and Tuf targets of experimental treatments (i.e. they defined as different from the control). d ND = Not detected.
63
A cp6 gene
Y = −3.35 x + 36.77 15 20 25 30 35
1 2 3 4 5 6
B lysS gene
Y = −3.72 x + 39.17 15 20 25 30 35 Quantification Cycle (Cq)
1 2 3 4 5 6 Figure 1. Calibration curves used to quantify the copy number of ALH sequence target (cp6 – Panel A) or AYp C sequence target (lysS – Panel B and tuf – tuf gene Panel C) in individual ALH DNA extracts. Points for each starting quantity represent four total reactions of each standard (i.e. four technical replicates per starting concentration) ranging from 106 to10 copies per μl run on five different 96-well plates. The four technical replicates resulted in a total of replicates for each Y = −3.52 x + 37.28 starting quantity grouped by 96-well plate as1 (□), 2 (Δ), 3 (●), 4(◊) and 5(▲). 15 20 25 30 35 Unbroken lines represent the typical slope and intercept estimated using a linear mixed 1 2 3 4 5 6 effects model and dashed lines describe a Standard Starting Quantity 95% prediction interval for each gene (Copies/ul) sequence calibration curve. 64 2.5 A ) 0 - N
i lysS gene tuf gene Φ = 2.09 ± 0.18 Φ = 2.30 ± 0.38
1.0 1.5 2.0 t = 11.4; p < 0.001 t = 5.9; p = 0.004 λ = 2.06 ± 0.21 λ = 1.51 ± 0.64 t = 9.9; p < 0.001 t = 2.3; p = 0.08 AYp Titer 0.5 Absolute (N μ = 1.22 ± 0.35 μ = 0.81 ± 0.49 t = 3.5; p = 0.03 t = 1.7; p = 0.17 2 2 R = 0.97 R = 0.98
−0.50 0.0 2 4 6 8 B ) 0 - R i lysS gene tuf gene Φ = 2.28 ± 0.21 Φ = 2.25 ± 0.31 t = 11.0; p < 0.001 t = 7.3; p < 0.005 λ = 1.86 ± 0.21 λ = 1.69 ± 0.47
AYp Titer t = 8.9; p < 0.001 t = 3.6; p = 0.02 Relative (R μ = 1.03 ± 0.22 μ = 0.91 ± 0.45 t = 4.6; p < 0.01 t = 2.1; p = 0.12 R2= 0.97 R2= 0.97
Alimentary Hemolymph Salivary
−0.5 0.0 0.5Canal 1.0 1.5 2.0 2.5 Glands
0 2 4 6 8 Incubation Time (Days) Figure 2. A) AYp titer is expressed as Log10-transformed copies per insect (N) averaged by incubation time relative to AYp concentration at Time 0 versus incubation time. B) AYp titer is expressed as Log10-transformed AYp (lysS or tuf) gene copies per cp6 gene copies (R) averaged by incubation relative to AYp concentration at Time 0 versus incubation time. In A and B, the solid and dashed lines represent the best fit 3-parameter logistic growth curve (iteratively reweighted least squares) describing AYp titer in the ALH as a function of time.
65 Figure 2 cont’d. The logistic growth curve had the form: Yi = Φ/ ( 1 + exp((((4*μ)/Φ)*(λ – Xi)) + 2) where Yi is AYp titer relative to AYp titer at time 0, Xi was incubation time and i indexes incubation time. Parameter estimates (± SE), test statistics and p-values for the asymptote (Φ), the maximum growth rate (μ), and the lag time (λ) associated with bacterial population growth are reported for each of the regression fits and are included in the respective panel. Time is days after a 48 hour AAP. Four degrees of freedom were used for all t-tests and a pseudo-R2 value (pseudo-R2 = 1 – (SSR/ SST)) was used as a measure of goodness of fit (Schabenberger and Pierce 2002). The bar in the lower portion of panel B is a temporal portrayal of AYp recovery from the ALH alimentary canal, hemolymphand salivary glands as reported by Sinha and Chiykowski (1967). 66
Y = 1.02 x − 0.22 P < 0.001 R2 = 0.94 (Copies/µl) 10 Target Target lysS gene Limit of Detection - 0.83 tuf gene Concentration Log
tuf gene Limit of Detection - 0.47 0 1 2 3 4 5
0 1 2 3 4 5 lysS gene Target Concentration Log10 (Copies/µl) Figure 3. AYp concentrations quantified using the tuf gene target sequence versus AYp concentrations quantified using the lysS gene target sequence. AYp titer concentrations measured using both primer sets were significantly correlated (R = 0.95; p < 0.001). 67
Chapter 3: Factors influencing aster leafhopper (Macrosteles quadrilineatus) abundance
and aster yellows phytoplasma infectivity in Wisconsin carrot fields
68
Abstract
In Wisconsin, vegetable crops are threatened annually by infection of the aster yellows phytoplasma (AYp), the causal agent of aster yellows (AY) disease, vectored by the aster leafhopper. Aster leafhopper abundance and infectivity are influenced by processes operating across different temporal and spatial scales. We applied a multilevel modeling approach to partition variance in multi-field, multi-year, pest scouting data sets containing temporal and spatial covariates associated with aster leafhopper abundance and infectivity. Our intent was to evaluate the relative importance of temporal and spatial covariates to infer the relevant scale at which ecological processes are driving AY epidemics and identify periods of elevated risk for AYp spread. The relative amount of aster leafhopper variability among and within years (39%) exceeded estimates of variation among farm locations and fields (7%). Similarly, time covariates explained the largest amount of variation of aster leafhopper infectivity
(50%). Leafhopper abundance has been decreasing since 2001 and reached its minimum in
2010. The average seasonal pattern indicated that periods of above average abundance occurred between 11 June and 1 August. Annual infectivity appears to oscillate around an average value of 2% and seasonal periods of above average infectivity occur between 19 May and 15 July. The coincidence of the expected periods of high leafhopper abundance and infectivity increases our knowledge of when the insect moves into susceptible crop fields and when it spreads the pathogen to susceptible crops, representing a seasonal interval during which management of the insect can be focused.
Keywords: Macrosteles quadrilineatus, aster yellows phytoplasma, aster yellows, insect migration, variance component analysis 69
Aster yellows (AY) is a widespread disease of plants caused by the aster yellows phytoplasma
(AYp), a small, wall-less prokaryotic organism that is currently placed in the provisional genus Candidatus (Lee et al. 2000, IRPCM Phytoplasma/Spiroplasma working team –
Phytoplasma taxonomy group 2004). The AYp has an extensive and diverse host range infecting over 350 plant species including many common vegetable, ornamental, and agronomically important field crops, and several non-crop plant species (Kunkel 1926,
Chiykowski 1965, Chiykowski and Chapman 1965, Chiykowski 1967, Westdal and
Richardson 1969, Peterson 1973, Lee et al. 1998, Lee et al. 2000, Lee et al. 2003,
Hollingsworth et al. 2008). Plant-to-plant spread of AYp in the field generally occurs as a result of transmission by more than twenty-four leafhopper species (Mahr 1989, Christensen et al. 2005). However, the aster leafhopper, Macrosteles quadrilineatus Forbes, is considered to be the most important vector of the AYp due to its prevalence in susceptible, Midwestern crops (Drake and Chapman 1965, Hoy et al. 1992).
The AYp is persistently transmitted by the aster leafhopper and both nymphs and adults can acquire the pathogen. Once infected, an individual aster leafhopper can remain infective for the remainder of its life. The aster leafhopper is a polyphagous insect species that utilizes over 300 different plant species for food, oviposition, and shelter (Wallis 1962,
Peterson 1973), many of which are susceptible to AYp infection. Aster leafhopper host plant species can be classified into two primary groups based on utilization patterns to include: 1) feeding hosts or 2) feeding and reproductive hosts. Other factors such as plant community composition (Lee and Robinson 1958, Wallis 1962, Schultz 1979), plant physiological state
(Peterson 1973), season and geographic location (Lee and Robinson 1958, Wallis 1962, 70
Peterson 1973) can also affect host preferences of aster leafhopper in the field. In Wisconsin, cultivated grains are hosts for overwintering eggs and also serve as early season feeding and reproductive hosts for the aster leafhopper (Drake and Chapman 1965). In addition to grain crops, the aster leafhopper feeds upon and is moderately abundant in mixed broadleaf weeds and grasses that border crop fields (Shultz 1979).
Each spring, the aster leafhopper migrates from the Gulf Coast states to the Upper
Midwest (Chiykowski and Chapman 1965). The aster leafhopper generally migrates in a
South to North direction but, in flight, leafhopper movement is greatly influenced by synoptic weather systems making it difficult to predict when and where the aster leafhopper will arrive.
The migratory behavior together with the mode of pathogen transmission by the aster leafhopper enables the insect to acquire and transmit the pathogen over great distances. Large numbers of migrating aster leafhoppers have been reported to influence the potential for AY epidemics in vegetable crops grown in Wisconsin and in other Midwestern States
(Chiykowski 1965, Drake and Chapman 1965, Chiykowski and Chapman 1965, Chapman
1973, Hoy et al. 1992). The severity of AY outbreaks is thought to be directly related to the infectivity and the abundance of aster leafhoppers immigrating into a susceptible crop
(Chapman 1971).
In Wisconsin, AY management has focused primarily on controlling the insect vector, the aster leafhopper. The aster yellows index (AYI), was developed as a risk assessment tool to describe periods in the growing season when protection of a susceptible crop was most needed (Chapman 1971, Chapman 1973). Simply, the AYI metric is the product of aster leafhopper (relative) abundance and infectivity. Insecticide sprays are then recommended if 71 the AYI exceeds an allowable threshold that is based on the relative susceptibility of the crop to infection by AYp. Originally, the AYI was calculated using an infectivity estimate determined from a series of early season (migratory) leafhopper collections and bioassays on susceptible Chinese aster (Callistephus chinensis). This infectivity estimate was used for the entire growing season while aster leafhopper abundance was determined weekly, or more frequently, for each field throughout the summer (Mahr 1993).
Following observations that aster leafhopper abundance and infectivity in and around carrot fields was dynamic in time and space (Mahr et al. 1993), efforts were made to estimate infectivity for each field throughout the summer to obtain a more site and time-specific AYI.
In many pathogen- disease systems, including the aster yellows patho-system in Wisconsin, contemporary tools for pathogen detection (i.e. nucleic acid based detection methods) have been adopted to estimate the infection frequencies (Bloomquist and Kirkpatrick 2002,
Munyaneza et al. 2010). However, even with the availability of contemporary tools, significant annual and site-specific variation of pathogen detection in the insect vector frequently occurs. In most cases, the relationship between pathogen presence in the vector and the vector’s ability to successfully transmit the pathogen is not known. In turn, many producers avoid risk of pathogen spread by using inexpensive, prophylactic applications of pyrethroid insecticides, a management practice that circumvents the utility of the AYI. An improved understanding the factors that influence variation in aster leafhopper abundance and infectivity will further improve the implementation of the AYI.
Environmental processes that drive plant disease epidemics occur at multiple temporal and spatial scales. For example, large (temporal and spatial) scale climate patterns may 72 influence the risk for fusarium head blight (FHB) development in the Midwestern U.S. (Kriss et al. 2012). However, smaller scale weather fluctuations such as a short-term dry period around wheat at anthesis can counteract the overall impact of a generally wet year by reducing the number of primary fungal infections leading to reduced FHB severity (De Wolf et al.
2003, Kriss et al. 2012). Investigating the patterns of aster leafhopper abundance or infectivity variation across different temporal and spatial scales will provide insight into the processes that drive the variation in annual AY epidemics. For example, if aster leafhopper migration was important for producing variation in the aster leafhopper infectivity, then it might be expected that inter-annual variation of aster leafhopper infectivity would be high relative to intra-annual variation. Additionally, an improved understanding of variation across different temporal and spatial scales can also inform future sampling strategies (Wheatley and Johnson
2009). For instance, if inter-annual variation of aster leafhopper infectivity was comparatively large relative to intra-annual variation, then repeated sampling within a season would explain very little about aster leafhopper infectivity. Unfortunately, experiments that manipulate environmental processes across multiple spatial and temporal scales (simultaneously) are difficult to perform. Yet, compiled data from observational studies that include spatial and temporal information at multiple scales offer an opportunity to obtain information about the scale at which ecological factors, contributing to abundance and infectivity variation, occur
(Magnuson 1990, Sagarin and Pauchard 2010).
Here we present an approach for parsing sources of variation in Poisson-distributed count data, similar to the method described by Duffy et al. (2010) that examined binomial data. Specifically, we applied this approach to analyze a multi-year, multi-location data set of 73 aster leafhopper abundance (2001-2011) obtained from pest scouting records in Wisconsin carrot fields. We also used a similar approach to examine sources of variation associated with aster leafhopper infectivity (1994 to 2008) collected from similar fields, locations and years.
For each data set, we quantified the variance components associated with annual, seasonal and geographic variability. The primary goal of this study was to identify the scale of the processes that drive AY epidemics and identify periods of time in the growing season when crop protection is most needed. Our specific objectives in this study were to: I) evaluate the relative importance of time (i.e. year and calendar date) and space (i.e. farm and field) in explaining the variability observed in aster leafhopper abundance and AYp-infectivity, II) identify periods of time in the growing season where aster leafhopper abundance and infectivity was above and below average, corresponding to periods of elevated or low risk, respectively.
Materials and Methods
Aster leafhopper abundance. Field sampling was conducted using sweep nets in commercial carrot fields to monitor the relative abundance of aster leafhopper in specific areas of
Wisconsin from 2001 through 2011. A total of 237 fields were sampled over the 11 year span of this survey resulting in an average of 31 fields per year with multiple fields re-sampled in successive years due to crop rotation practices. The fields were clustered geographically into
6 distinct growing regions in the Central Plain, the Western Upland and Eastern Ridge eco- regions of Wisconsin. The approximate distance among fields ranged from 0.1 to 15 km within a farm and 15 to 200 km among farms. In Wisconsin, carrots are direct seeded in mid-
April through early June and a cover crop (e.g. oats, wheat, or rye) is concomitantly 74 established to prevent wind damage to the developing carrot crop during seedling emergence.
Carrot seedlings typically emerge in late-May and early June and the crop is usually harvested from late August through mid-November, depending on the growing season.
In all years, aster leafhopper monitoring began prior to carrot emergence, usually in mid-May, and terminated one to three weeks prior to carrot harvest, and no later than
September 20 for all field sites in all years. Early sample dates, those prior to May 25, occurred primarily in rye, wheat, or oat since the carrot crop does not typically emerge until after that date. At each location, the relative abundance of leafhopper adults associated with the carrot canopy was determined by standard sweep net sampling along 2 to 18 transects extending into the carrot crop toward the middle of the field. Twenty-five to 100 pendulum sweeps per transect were conducted using a standard sweep net (38 cm diameter) and all aster leafhopper stadia were counted. Counts were enumerated as adult aster leafhoppers per 25 sweeps. Decimals, occurring when more than 25 sweeps were conducted, were rounded to the nearest integer. Fields were sampled weekly unless weather or grower management did not allow for sampling.
Aster leafhopper collections and AYp infectivity. Aster leafhopper infectivity was monitored using a transmission bioassay and, for the commercial carrot production area of
Wisconsin, records of infectivity were available from 1994 through 2008. In total, infectivity was estimated from among 378 aster leafhopper populations; approximately 25 populations of which were from multiple geographic locations and several dates throughout each growing season.
When possible, more than 200 leafhoppers were collected in sweep nets and placed 75 onto oat seedlings for transport back to the laboratory for transmission bioassays. Typically,
204 leafhoppers were placed in pairs onto 102 Chinese aster (Callistephus chinensis) plants and insects were allowed to feed for a 48 hour inoculation access period (IAP). Disease symptoms were assessed after a two-week incubation period and percent infectivity was calculated as:
Infectivity = # of diseased plants / Total # of leafhoppers
The total number of leafhoppers was used as the denominator because infectivity levels are often low and a diseased plant was more likely due to a single infective leafhopper rather than the presence of two infective leafhoppers on the same plant.
Statistical analysis. Factors contributing to variation of aster leafhopper abundance. A generalized linear mixed modeling (GLMM) approach based on Poisson regression (log-link) with random intercepts was used to examine the relative importance of year, week, farm and field on the abundance of the aster leafhopper (Pinhero and Bates 2000, Madden et al. 2002,
Nita et al. 2008, Bolker et al. 2009). The multilevel model (Gelman and Hill 2007) had the following form:
Yi(abcd) ~ Poisson(µi(abcd)) (Model 1)
g(µi(abcd)) = loge(µi(abcd)) = X β + log(effort) + εa + εb + εc + εd
2 εa ~ N(0, σ a)
2 εb ~ N(0, σ b)
2 εc ~ N(0, σ c)
2 εd ~ N(0, σ d), where Yi(abcd) was the total aster leafhopper count for a field and the total aster leafhopper 76 count for a field was offset by the number of transects walked (or sampling effort) in each field. The regression coefficient for the offset term log(effort), by definition, was constrained to one. The fixed effects term, β, represented the model intercept and was interpreted as the statewide seasonal average aster leafhopper abundance in carrot fields. εa, εb, εc, and εd, were the random effects (or intercepts) for year, day, farm, and field, respectively. They represented the variance components associated with the temporal and spatial “blocks” of this model. The variance components of aster leafhopper abundance were quantified on the aster leafhopper count given by g(µi(abcd)) and variance components were assessed in terms of variances (or standard deviations) on the latent, or loge scale of the model.
Mixed-effects models, in general, are used because they associate random effects to observations sharing the same level of a classification factor. Thus, mixed-effects models are useful because they can accurately represent the covariance structure that exists among samples when repeated measurements are taken at the same location or time (Pinheiro and
Bates 2000); essentially we are assuming all observations from a given source (or subject) are correlated. Often, when research emphasis is placed on estimating fixed regression coefficients, random effects are included in a model to account for the covariance among sample groupings prior to estimating the regression coefficients. However, in our case, the variance-covariance structure itself was of interest and all factors in our analyses were considered random since the primary goal was to examine the nature of different spatial and temporal levels from which the data were thought to have originated (Pinhero and Bates 2000,
Bayeen et al. 2008, Nita et al. 2008).
Model 1 represents the case in which the variance can be divided into separate 77 components for year (εa), week (εb), farm (εc), and field within farm (εc), and where the
2 2 2 2 magnitude of the variance components (i.e. σ a, σ b, σ c, or σ d) could be interpreted as a measure of the relative importance of the different spatial and temporal factors associated with aster leafhopper count. There are many ways that a GLMM can be defined to examine the interactions among different combinations of covariate groupings and to quantify their associated variances (Duffy et al. 2010). In our case we were not interested in the contributions of specific years as much as we were interested in the variation due to year.
Additionally, we were more interested in the day-to-day variability of aster leafhopper abundance than the variability of aster leafhopper abundance on a specific day. The flexibility of the GLMM allowed us to specify random variables for year (εa) and day (εb) and model
2 2 that variability as εa + εb (i.e. σ a + σ b). We could also examine the variability of a specific
2 2 day by including a day-year interaction term, εab (or σ ab), in the model. Thus a σ ab > 0 implies that annual aster leafhopper abundance varies interactively with calendar day and the
2 variability among day-year pairs would be more adequately modeled as εa + εb + εab (i.e. σ a +
2 2 σ b + σ ab). Our biological interpretation of this approach is analogous to that of regression analyses where both intercept and slope are allowed to vary among treatments. Estimates of the variability of each grouping (i.e. year, day, and year-day) are obtained and the relative amount of variability described by each level of grouping is reflective of the importance of each factor. Important terms are those that describe larger amounts of variability. Applied specifically to insect abundance data, εa is an estimate of the annual variation of aster leafhopper abundance, εb is an estimate of variation in seasonal aster leafhopper abundance
(or phenology), and εab is an estimate of how aster leafhopper phenology varies interactively 78 among years. Thus, the inclusion of different “interaction terms” as random effects in the
GLMM leads to numerous ways to partition the variances associated with aster leafhopper count, providing insight about the underlying biology and spatial or temporal scales at which processes important for aster leafhopper population dynamics are occurring (Duffy et al.
2010).
In our data set, the temporal and spatial grouping of covariates occurs at different scales. For example, year and day represent a different spatial grain size. Studying the patterns of aster leafhopper abundance variation at different temporal scales can provide insight about the scale of the underlying ecological processes driving aster leafhopper prevalence (Levin
1992, Wheatley and Johnson 2009). Large variation of aster leafhopper abundance among years, relative to other sources, might suggest that aster leafhopper numbers are influenced by climatic or biological factors (i.e. El Niño and La Niña cycles, winter mortality or early generation survivorship at southerly latitudes). In contrast, large variation within years might be better explained by processes such as differences in grower management or synoptic weather events. Similarly, insights can be gained by examining variation occurring among and within geographic location. For example, large variation of aster leafhopper abundance among geographic locations might imply that the local habitat (i.e. non-crop reproductive host plants) surrounding crop a field is important. Alternatively, small variations in aster leafhopper abundance among geographic locations might suggest larger scale processes, occurring across all locations, drive insect abundance (i.e. mean annual temperatures).
In general, a full model that included all the random effects of interest was constructed and Akaike information criterion (AIC) and likelihood ratio tests (LRT) were used to evaluate 79 if the inclusion of random effects parameters were justified in the model. Parameter estimates for a selected submodel are reported in the text (Table A1 contains parameter estimates for the full model and various submodels). All models were fit using the glmer (lme4: version
0.999375-39; Bates et al. 2011) function in the lme4 package of R (lme4: version 0.999375-
39; Bates et al. 2011, R version 2.15.0; R Development Core Team 2012) which allows for the analysis of crossed classified data as crossed random effects (Pinhero and Bates 2000,
Baayen et al. 2008).
Factors contributing to variation of aster leafhopper infectivity. A linear mixed modeling
(LMM) approach, similar to the GLMM approach previously used for aster leafhopper abundance, was used to examine the relative importance of year, farm and calendar day, on aster leafhopper infectivity. In this data set, calendar day, corresponded to weekly estimates of leafhopper infectivity and sample dates were represented as Julian date at the mid-point of the sample week. Again, the multilevel model representing a simple case for describing infectivity had the form:
Yi(abc) = μ + εa + εb + εc + ε i(abc) (Model 2)
2 εa ~ N(0, σ a)
2 εb ~ N(0, σ b)
2 εc ~ N(0, σ c)
2 ε i(abc) ~ N(0, σ r), where Yi(abc) was the estimated aster leafhopper infectivity – the square root-transformed proportion of leafhoppers able to transmit the AYp. As described above, this LMM was also extended to examine the interactions among the random effects terms, partitioning the 80 variance of aster leafhopper infectivity among known spatial and temporal “blocks”, providing insight about the scales at which processes important for influencing variation of aster leafhopper infectivity are operating. Models were again fit using the lmer function
(lme4: version 0.999375-39; Bates et al. 2011) and AIC and LRT were used to evaluate if the inclusion of random effects parameters were justified in the model (See Table A2 for parameter estimate of the full model).
Model diagnostics. The variance assumptions of regression analysis are often made for statistical purposes. For example, if there is not constant variance, standard errors may be biased leading to unreliable statistical tests. Trends occurring in the model residuals would violate the assumption of independent response variables and often are a result of erroneous model structure. However, identifying trends in the residuals may reveal useful biological patterns and may imply the pattern of a trend which can be directly fit in subsequent modeling efforts. The results we present here emphasize the utility of visually examining the possible patterns of variation in data (i.e. residuals). A more direct modeling of these data trends was subsequently performed (Frost et al. 2012), but was not the focus of this paper.
A series of residual plots were used to assess the assumptions of the random effects model and determine if the errors in the model predictions behave in the same way within each level of grouping in the data. For mixed effects models, there are several different types of residuals that can be obtained due to the different group levels of the model and each type of residual is useful for evaluating model assumptions (Pinheiro and Bates 2000, Nobre and Singer 2007).
Here, we focus on plots of the conditional modes of the random effects versus temporal
(group) indices because we were interested in the temporal aspects relating to pathogen 81 transmission. Thus, we present plots examining the conditional modes of the random effects for the population expected values for year, ordered by year, and calendar day, ordered by day to demonstrate the temporal trends of insect abundance and infectivity data, among and within year. Conditional modes of the random effects for the various levels of the models were extracted using the ranef function and plotted using the qplot function of the ggplot2 package
(Wickham 2009). Trend lines were generated using a generalized additive model.
Correlations among years, farms, and fields. The inclusion of random effects into a regression model has an effect on the structure of the model’s variance-covariance matrix
(Zuur et al. 2009). If the mixed effects regression models are specified appropriately, the
GLMM and LMM framework can be used to examine these induced correlations among farms within year or year-week groups (Zuur et al. 2009; See example Appendix B). For example, a GLMM used to examine correlations among farms within years can be formulated as:
µp(y) = exp(β1x1 + β2x2 + β3x3 + εp(y))
εp(y) ~ N(0, Σp)
Here β1 through β3 are fixed effects for the mean leafhopper abundance of farm1 through farm
3 and the design matrix uses dummy variables (i.e. 0 or 1) to represent the farm category. The random variable, εp(y), is independent, normally distributed, and its (symmetric) covariance matrix is:
2 Σp = σ1 ρ1σ1σ2 ρ2σ1σ3 2 ρ1σ1σ2 σ2 ρ3σ2σ3 2 ρ2σ1σ3 ρ3σ2σ3 σ3
82 which accounts for correlations (i.e. ρ1 – ρ3) of farms within each year. In a similar way, the
GLMM can be formulated to examine correlations of farms within years, weeks or year-week combinations. In turn, these correlations allow us to examine if similarities exist among locations that may be important for insect abundance. As in Duffy et al. (2010), we assessed the distribution of the ρ estimate using parametric bootstrapping (100 bootstrap estimates).
However, we did not try to imply the significance of the correlation based on bootstrapping because ρ was explicitly defined by fitting data to the GLMM (or LMM). The correlations obtained from the variance-covariance matrices were used to calculate (Euclidean) distance matrices. The hclust function was used to conduct hierarchical clustering of the distance matrices and produce dendrograms for visualization of correlative relationships among variables.
Outside of the GLMM context, the correlation of aster leafhopper abundance among field combinations within a farm was also examined for all farms in all years. For example, we were interested to know if aster leafhopper counts from two fields, on the same farm and sampled at the same time, would be similar. Due to the large number of correlation coefficients produced in this analysis, we chose to graphically visualize the distributions of the coefficients using density plots. The density plots were produced using the stat_density function in the ggplot2 package (Wickham 2009), which uses kernel density estimation and densities were scaled to one within each farm. Following visual examination, the lm function was used to conduct an analysis of variance (ANOVA) to examine the effect of farm and year on within farm correlations among fields. The marginal sum of squares was used to evaluate the importance of factors in the ANOVA model. 83
Results
Quantitative variability in aster leafhopper abundance. Aster leafhopper count data for an individual field on a sample date was variable, ranging from 0 to 60 aster leafhoppers per 25 sweeps over the 11 year interval, and included a large number of occasions when no aster leafhopper were caught. Data such as these have traditionally been log-transformed and analyzed using ordinary linear models with normal errors. However, plots of the field variance versus the field mean suggested that the assumption of constant variance was not valid and revealed that the variance, while not equal to the mean, appeared to be proportional to the mean. This property of the data was more consistent with Poisson-distributed data and we used a GLMM, Poisson family (log-link), to examine the aster leafhopper count data.
The average aster leafhopper abundance for all fields in all years was estimated by model 1 to be 0.44 (95% CI: 0.23 – 0.85) aster leafhoppers per 25 sweeps. The temporal
“blocks” of the GLMM described a larger proportion of aster leafhopper count variability
(Table 1). For example, year, day and the day x year interaction terms accounted for approximately 39% of the aster leafhopper count variability, whereas, farm and field accounted for 7% of the total aster leafhopper count variability. When temporal blocks were allowed to interact with spatial blocks, the largest proportion of aster leafhopper count variation was described at shorter temporal scales. For example, the day x year x farm term described 26% of the total variation of aster leafhopper counts and the day x year x field term, which corresponded to the observation level (or residual) variability, was estimated to be
0.924 (reported as σ), approximately 28% of the total variance. The remainder of the random effects terms that we examined did not account for a large proportion of aster leafhopper 84 count variance and were excluded from the final model.
Aster leafhopper abundance model diagnostics. Plots of the conditional modes of year random effects, ordered by year, indicated that aster leafhopper abundance has been decreasing since 2001and reached its minimum in 2010 (Fig. 1A). Additionally, the seasonal
(or within-year) pattern that resulted from plotting the day conditional modes, ordered by calendar day, indicated that periods of above average aster leafhopper abundance occurred between 11 June and 1 August, representing a seasonal “window” during which higher aster leafhopper abundance occurs (Fig. 1B). These plots indicated there was a pattern among years and within years that could be more directly modeled and this was the focus of our second paper (Frost et al. 2012).
Quantitative variability of aster leafhopper infectivity. Over 15 years of measurement, aster leafhopper infectivity ranged between zero and 14%. These data were bounded (i.e. by 0 and 100%) and a histogram of infectivity indicated that the data were not normally distributed. Therefore, the infectivity data were square root-transformed prior to regression analysis. The average infectivity estimated by model 2 was 1.9% (95% CI: 1.2% - 2.9) although we would predict the average annual infectivity to fall between 0.2% and 5.6% (i.e., from supplemental Table A1: 0.139 ± 2*√(0.0142 + 0.0472)). Similar to aster leafhopper abundance, farm (or location) did not explain a large amount of the variability in aster leafhopper infectivity (Table 1). However, year, week and year x week groups described approximately 50% aster leafhopper infectivity variability while the remaining 50% of the variability could not be attributed to a known factor in our data set (i.e. residual variance).
Aster leafhopper infectivity model diagnostics. The largest proportion of variance was 85 explained by year and, therefore, we plotted the conditional modes of the random effects for the population expected values of year, ordered by year (Fig. 2A). This plot suggested that annual aster leafhopper infectivity oscillated 2% among years, although more data would be necessary to estimate the periodicity of infectivity. Plots of the week conditional modes of the random effects, ordered by week, indicated that periods of above average aster leafhopper infectivity occurred between 19 May and 15 July (Fig. 2B). Again, these plots indicated that among year and within year seasonal patterns of aster leafhopper infectivity that could be more directly modeled (Frost et al. 2012).
Correlations of aster leafhopper abundance. Correlations of aster leafhopper abundance among years ranged from -0.73 to 0.88 with no distinct grouping that emerged within years
(dendrogram not shown). Additionally, correlations were plotted versus the lag between years with no apparent association occurring (not shown). All farms were positively correlated within year groupings (Table 2) and correlation coefficients ranged from 0.59 to 0.95 suggesting that the effect of year on aster leafhopper abundance was consistent among farms.
Within year-week correlation coefficients ranged from -0.17 to 0.40 suggesting the aster leafhopper abundance estimates among farms, at this shorter time scale, were less correlated
(Table 3). The similarity of farms at these different scales can be visualized using correlation as a distance measure to produce dendrograms. Based on hierarchical clustering of the correlation coefficients, farms generally formed two branches and appeared to group by
(similar) geographic location within year and year-week scales (Fig. 3). This clustering of farms may also be partially explained by similarities in habitat characteristics in the landscape surrounding the farms. 86
To determine if aster leafhopper counts from two fields at the same farm would be similar, we initially used density plots to examine the distribution of correlation coefficients of aster leafhopper abundance among field (within farm) combinations. On average, the distributions of correlation coefficients among field combinations were approximately normal for all farms in all years (not shown). The correlation coefficients of field combinations for a farm varied interactively with year (Year x Farm effect: F = 4.8; df = 27, 1495; P < 0.001) and a general interpretation for the main effects of year (F = 17.0; df = 10, 1495; P < 0.001) and farm (F = 3.7; df = 3, 1495; P = 0.011) was not possible.
Correlations of aster leafhopper infectivity. Aster leafhopper infectivity among all farms was positively correlated within year (Table 4) and remained positively correlated within year-week combinations (Table 5). Unlike aster leafhopper abundance, farms did not cluster by geographic location based on correlations within a year. However within year-week groupings, farms formed two clusters possibly based on geographic location. Specifically, farms in Southern Wisconsin grouped together and away from farms in Central Wisconsin
(Fig. 4).
Discussion
A successful disease control program relies on a detailed understanding of the critical factors that directly influence epidemic development. For plant pathogens spread by arthropods, insect vector abundance and transmission capability, or infectivity, have been reported as important factors that influence plant disease severity in a given growing season
(Chapman 1971, Chapman 1973, Madden et al. 2000, Jeger et al. 2004). In this paper, we present an approach for examining factors affecting variation in observed insect abundance 87 and infectivity and apply this approach to a long-term observational data set. In our approach, the importance of factors contributing to aster leafhopper abundance and infectivity variation was determined by a factor’s relative contribution to the explanation of total variation.
We found that geographic location, farm or field alone, was not a factor that contributed (significantly) largely to the observed variation of insect abundance, relative to other sources of variation. However, aster leafhopper abundance varied greatly among years.
Immigration of the aster leafhopper, presumably from the Gulf States in early spring
(Chiykowski and Chapman 1965) and later from the central and northern Great Plains (Hoy
1992), has long been considered the principle source for infectious aster leafhoppers in susceptible carrot. The trajectory of air movement and position of cold fronts could affect the geographic extent of adult insect arrival (e.g. depositions zones) (Hurd 1920, Huff 1963,
Westbrook and Isard 1999, Zhu 2006) and it would be expected that major synoptic weather events, which occur over larger geographic extents, could lead to low variability of insect abundance at larger spatial scales (i.e. scales larger than the extent of our observational study).
Leafhopper abundance also varied greatly at the smaller temporal and spatial scales, within years and farms. For example, more than 50% of aster leafhopper count variability was described by the interaction terms of day x year x farm and day x year x field. Synoptic weather and wind patterns also occur at these shorter time-scales and are known to correlate with leafhopper influxes (Chiykowski 1965, Hoy 1992, Huff 1963), which could have influenced weekly aster leafhopper abundances. However, this result may be more indicative of unique crop production practices implemented on different carrot fields at these shorter time scales. 88
Similar to aster leafhopper abundance, temporal factors accounted for the largest proportion of the variability of aster leafhopper infectivity, which was dominated by the among year variance component (31%). Farm (or location) was not a factor that contributed largely to infectivity. Taken together, the results of our variance component analysis of aster leafhopper abundance and infectivity are consistent with the hypothesis that the aster leafhopper immigration contributes, in part, to the annual risk of AY epidemics in Wisconsin.
Nearly 50% of the aster leafhopper infectivity variation could not be partitioned to a known factor in our data set. This large residual variability could be due to multiple causes, such as AYp strain variability (Lee et al. 2003, Zhang et al. 2004, Frost et al. 2008) or misidentification of the AYp as the cause of the disease symptoms observed in the bioassay plants. In this study, infectivity was determined using infectivity bioassays and examines only those insects in the field that are already able to transmit. Currently, it is common to determine the percentage of insects that are carrying a pathogen using polymerase chain reaction (PCR) assays and the seasonal pattern of pathogen detections in their insect vectors has been documented for numerous pathosystems (Beanland et al. 1999, Bloomquist and
Kirkpatrick 2002, Munyaneza et al. 2010, Bressan et al. 2011). Although PCR can specifically detect the presence of a pathogen, the relationship between a PCR detection and transmission capability of an individual insect is rarely known. Further documentation of the seasonal pattern of infectious insect vectors and research examining the relationship between
PCR detection and transmission capability of the insect is necessary to provide accurate pest management recommendations. A comparison of PCR detections and percent capable vectors from the same insect population may provide information about the importance of local 89 inoculum in the environment since insects acquiring AYp locally would be less likely to pass through the necessary latent period prior to being assayed (or controlled). We hypothesize that one possible explanation for the difference between the percentage of infectious individuals and PCR-positive detections may result from the inoculum contribution of the local environment.
Although the analysis did not directly describe the temporal patterns of aster leafhopper abundance, plots examining the conditional modes of the year random effects indicated that average annual aster leafhopper abundance decreased through the interval 2001 to 2010. We cannot explain why aster leafhopper abundance steadily decreased over this period of time, although periodic or quasi-periodic climate patterns, such as the diurnal temperature cycles associated with El Niño and La Niña, or synoptic weather patterns could potentially impact overwintering survival and seasonal insect phenology in Southern latitudes
(Westbrook et al. 1997, Diffenbaugh et al. 2008, Morey et al. 2012), seasonal migratory cues
(Carlson et al. 1992, Isard and Irwin 1993), and weather patterns conducive to leafhopper transport and dispersion (Huff 1963, Carlson et al. 1992, Westbrook and Isard 1999).
Additionally, the magnitude of migrating aster leafhopper population and its trajectory from the aster leafhopper source regions is likely to be affected by the among year variation in the abundance of small grains acreage planted in the source regions and the location of those acres with respect to seasonal wind and weather patterns.
We observed that, on average, there is a period of elevated aster leafhopper abundance between 11 June and 1 August in Wisconsin. This observation is consistent with previous reports of aster leafhopper phenology in Wisconsin and in the Midwest (Hoy et al. 1992, 90
Mahr et al. 1993). In Wisconsin, aster leafhopper overwinter as eggs (Drake and Chapman
1965) and eclosion and subsequent development of aster leafhopper to the adult stage is linearly related to temperature (Jensen 1981, Mahr 1989). It is typical to accumulate enough thermal units by June 11 for aster leafhopper to have developed into winged adults. Thus, the above average aster leafhopper captures observed around June 11 may, in part, be due to the emergence of the local leafhopper population. This seasonal pattern of leafhopper abundance, elevated in mid-to-late June through mid-July followed by a decline in late-July to August, has also been observed in other temperate regions of the United States and Europe for different leafhopper species. For example, Circulifer tenellus (beet leafhopper), Psammotettix alienus (European grass-feeding leafhopper), and Graphocephala atropunctata (blue-green sharpshooter) abundances all peak in June followed by a decline in late July and August
(Lindblad and Areno, 2002, Munyaneza et al., 2008, Gruber and Daugherty 2012). It may be that these leafhoppers all overwinter, or diapause, in the same life stage (i.e. eggs) and have to develop through a similar number of instar stadia (i.e. 4-5 instars) leading to a reasonably synchronous emergence as adults among species.
To our knowledge, this is the first large systematic study that reports over a decade of insect infectivity estimates. Because these data were collected from among and within 14 growing seasons, we were able to examine the inter- and intra-annual variability of aster leafhopper infectivity. Although highly variable among years, the average natural infectivity was estimated as 2% and we would predict infectivity to range between 0.2% and 5.6% for any given year. Additionally, there may be long term trends of annual infectivity which, if modeled, could help to anticipate high infectivity years. However, more years of data are 91 necessary to determine if the periodicity of natural leafhopper infectivity occurs and to quantify such oscillations (i.e. wavelengths, amplitudes and etc.). Gruber and Daugherty
(2012) reported seasonal data on the proportion of G. atropunctata that transmitted Xylella fastidiosa from two historical data sets and concluded that infectivity was either constant or increasing exponentially over the season. In contrast, we found that natural infectivity of aster leafhopper increased early in the season, from mid-May through late-June, and then decreased in mid-July. Thus, periods of above average natural infectivity typically occurred between 19
May and 15 July. Taken together, the coincidence of the expected periods of high leafhopper abundance and infectivity represent a potential ‘treatment window’ in which management of the insect could be focused.
The landscape surrounding each farm (or farm location) can influence aster leafhopper abundance or infectivity because each location supports a unique composition of predominant plant species. For example, the aster leafhopper uses over 300 different plant species for food, oviposition, and shelter (Wallis 1962, Peterson 1973) and many of these are species are susceptible to AYp infection (Kunkel 1926, Chiykowski 1965, Chiykowski and Chapman
1965, Chiykowski 1967, Westdal and Richardson 1969, Peterson 1973, Lee et al. 1998, Lee et al. 2000, Lee et al. 2003, Hollingsworth et al. 2008). Thus, the species composition surrounding each farm likely influences the reproductive capability of the aster leafhopper and/or the prevalence of AYp in the local environment. The habitats surrounding farms may also interact with seasonal weather to further influence leafhopper development and infectivity. It is known that the development of the aster leafhopper, under the same temperature conditions, occurs more slowly on C. chinensis than it does on either oats or 92 barley (Mahr 1989). Therefore, the observed farm-to-farm variability of leafhopper phenology may be the result of local weather operating similarly at all farms and times (years and weeks), but operating interactively with the local landscape (different hosts). It is therefore interesting that carrot farms appeared to group differently at the year and year-week scales based on the observed correlations in leafhopper abundance. In contrast, farms tended to group similarly at the year and year-week scales based on the correlation in leafhopper infectivity.
Studying the pattern of variation and correlations across different temporal and spatial scales is informative for developing future sampling strategies (Wheatley and Johnson 2009,
Zuur et al. 2009, Sagarin and Pauchard 2010). For aster leafhopper infectivity, the largest amount of variation was observed in among year samples. In turn, successive, in-season sampling for aster leafhopper infectivity may provide little additional information to explain risk for disease development. This is, perhaps, why the earlier AYI proposed and implemented by R. K. Chapman was a successful AY management tool (Chapman 1973).
Additionally, farm location, although not a large contributor to variation of infectivity, may be important if we wish to quantify farm-to-farm variation of infectivity. For example, based on our correlation analysis in Wisconsin, it would be best to determine infectivity for farms located in the Southern and Central parts of the state to maximize the among location variability. Aster leafhopper abundance varied at the smaller scales, within year and among fields. Additionally, the correlation of aster leafhopper abundance among fields and within farms varied interactively with year. The practical application of this outcome suggests that scouting to determine aster leafhopper abundance for a specific field and date combination 93 will remain necessary for accurate, site-specific insect management recommendations, even if two fields occur at the same farm location.
Compiled data from observational studies are useful to obtain information about the scale at which ecological factors contributing to aster leafhopper abundance and infectivity occur, because manipulating environmental processes across multiple spatial and temporal scales is difficult. Unfortunately, data collected over long periods of time are at risk of problems associated with accuracy and precision (Baumgartner et al. 1998); long-term data sets are often collected by multiple individuals and for multiple purposes. In our case, the consistency with how the data were collected helped to reduce some of the inherent variability often associated with long-term datasets. For research purposes, it is often necessary to obtain representative samples for all possible conditions that may occur in the biological system of interest. The value of long-term data sets is that they provide a relevant range of observations from the systems we wish to describe (Magnuson 1990). For example, large data sets usually include data points from observations during aberrant or extreme environmental conditions.
In the future, the availability and integrity of large pest scouting data sets will likely increase as agricultural data collection becomes less demanding and data storage becomes less expensive using commercialized software (Sagarin and Pauchard 2010). Several efforts are currently being made to streamline data collection efforts (i.e. Ag Connections, Inc. Murray,
KY, U.S.A, scoutpro.org), but similar to other areas of biology, methods will still need to be developed to thoroughly examine these data (Sagarin and Pauchard 2010). The methods presented in this paper may be applicable for analyzing other multi-year, multi-location observational pest scouting data sets to reveal patterns of variability driving plant disease or 94 pest epidemics. Applied specifically to the aster yellows disease system, this methodology improves our understanding of the spatial and temporal patterns of variation of aster leafhopper abundance, informs efforts to directly model the seasonal patterns of leafhopper abundance and infectivity to deduce AY risk, and further increases our knowledge of when the insect moves into susceptible crop fields and when it spreads the pathogen to susceptible crops. 95
Acknowledgements: We thank Dr. Emily Mueller for her constructive criticism and comments on earlier versions of the manuscript. Funding support was provided by the USDA
Specialty Crops Research Initiative through the Wisconsin Specialty Crops Block Grant
Program (MSN 129013) and from the Wisconsin Muck Farmers Association. 96
References
Baayen, R. H., Davidson, D. J., and D. M. Bates. 2008. Mixed-effects modeling with crossed random effects for subjects and items. J. Mem.Lang. 59:390-412.
Bates, D., Maechler, M. and B. Bolker. 2011. lme4: Linear mixed-effects models using S4 classes. R package version 0.999375-42. { http:// CRAN.R- project.org/package=lme4}
Manly, B.F.J. 1998. Sampling and modelling of insect populations. In: Population and community ecology for insect management and conservation. Baumgartner, J., Brandmayr, P., and B.F.J. Manly, ed. A.A. Balkema Publishers, Brookfield, VT, USA. pp 3-20.
Beanland, L., Hoy, C.W., Miller, S.A., and L.R. Nault. 1999. Leafhopper (Homoptera: Cicadellidae) transmission of aster yellows phytoplasma: does gender matter? Environ. Entomol. 28(6):1101-1106.
Bloomquist, C.L. and B.C. Kirkpatrick. 2002. Frequency and seasonal distribution of pear psylla infected with the pear decline phytoplasma in California pear orchards. Phytopathology 92:1218-1226.
Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Stevens, M. H. H., and J.S. White. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends Ecol.Evol. 24(3):127-137.
Bressan. A., Garcia, F.J.M, and E. Boudon-Padieu. 2011. The prevalence of ‘Candidatus Arsenophonus phytopathogenicus’ infecting the plant hopper Pentastiridius leporinus (Hemiptera: Cixiidae) increase nonlinearly with the population abundance in sugar beet fields. Environ. Entomol. 40(6):1345-1352.
Canto, T., Aranda, M.A., and A. Fereres. 2009. Climate change effects on physiology and population processes of hosts and vectors that influence the spread of hemipteran- borne plant viruses. Glob. Change Biol. 15:1884-1894.
Chapman, R.K. 1971. Prediction of aster leafhopper migrations and the resultant aster yellows problem in Wisconsin. Proc. North Cent. Branch Entomol. Soc. Am. 26:96- 98.
Chapman, R.K. 1973. Integrated control of aster yellows. Proc. North Cent. Branch Entomol. Soc. Am. 28:71-92.
Chiykowski L.N. 1965. The reaction of barley varieties to aster yellows virus. Can. J. Bot. 97
43:373-378.
Chiykowski L.N. 1967. Reaction of some wheat varieties to aster yellows virus. Can. J. Plant Sci. 47:149-151.
Chiykowski L.N. and R.K. Chapman. 1965. Migration of the six-spotted leafhopper in central North America. Wisconsin Agricultural Experiment Station Research Bulletin 261: 21-45.
Christensen, N.M., Axelsen, K.B., Nicolaisen, M. and A. Schultz. 2005. Phytoplasmas and their interactions with hosts. Trends Plant Sci. 10:526-535.
Carlson, J.D., Whalon, M.E., Landis, D.A., and S.H. Gage. 1992. Springtime weather patterns coincident with long-distance migration of potato leafhopper into Michigan. Agr. Forest Meteorol. 59:183-206.
De Wolf, E.D., Madden, L.V. and P.E. Lipps. 2003. Risk assessment models for wheat fusarium head blight epidemics based on within-season weather data. Phytopathology 93(4):428-435.
Diffenbaugh, N.S., Krupke, C.H., White, M.A., and C.E. Alexander. 2008. Global warming presents new challenges for maize pest management. Environ. Res. Lett. 3: 044007 (9 pp.)
Drake D.C. and R.K. Chapman. 1965. Evidence for long distance migration of the six- spotted leafhopper in Wisconsin. Wisconsin Agricultural Experiment Station Research Bulletin 261: 1-20.
Duffy, M.A., Caceres, C.E., Hall, S.R., Tessier, A.J., and A.R. Ives. 2010. Temporal, spatial, and between-host comparisons of patterns of parasitism in lake zooplankton. Ecology 91(11):3322-3331.
Frost, K. E., Esker, P.D., van Haren, R., Kotolski, L., and Russell L. Groves. 2012. Seasonal pattern of aster leafhopper (Hemiptera: Cicadellidae) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields. Environ. Entomol. XX:XX-XX.
Frost, K. E., Groves, Carol L. and Russell L. Groves. 2008. Detection of multiple strains of the aster yellows phytoplasma in Wisconsin carrot fields [abstract]. Phytopathology 98:S55.
Gelman, A. and J. Hill. 2007. Data analysis using regression and multilevel/hierarchical models. Cambridge University Press, Cambridge, UK. 98
Gruber, B.R. and M. P. Daugherty. 2012. Understanding the effects of multiple sources of seasonality on the risk of pathogen spread to vineyards: vector pressure, natural infectivity, and host recovery. Plant Pathol. [Online] Doi: 10.1111/j.1365- 3059.2012.02611.x. pp. 11
Hollingsworth, C.R., Atkinson, L.M., Samac, D.A., Larsen, J.E., Motteberg, C.D., Abrahamson, M.D., Glogoza, P., and I.V. MacRae. 2008. Region and field level distributions of aster yellows phytoplasma in small grain crops. Plant Dis. 92:623-630.
Hoy, C.W., Heady, S.E., and T.A. Koch. 1992. Species composition, phenology, and possible origins of leafhoppers (Cicadellidae) in Ohio vegetable crops. J. Econ. Entomol. 85:2336-2343.
Huff, F.A. 1963. Relation between leafhopper influxes and synoptic weather conditions. J. Appl. Meteorol. 2(1):39-43.
Hurd, W.E. 1920. Influence of the wind on the movements of insects. Mon. Weather Rev. 48(2):94-98.
IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group. 2004. 'Candidatus Phytoplasma', a taxon for the wall-less, non-helical prokaryotes that colonize plant phloem and insects. J. Syst. Evol. Microbiol. 54:1243-1255.
Isard, S.A. and M.E. Irwin. 1993. A strategy for studying the long-distance aerial movement of insects. J. Agric. Entomol. 10:283-297.
Jeger, M.J., Holt, J., van den Bosch, F., and L.V. Madden. 2004. Epidemiology of insect- transmitted plant viruses: modeling disease dynamics and control interventions. Physiol. Entomol. 29:291-304.
Jensen, J.O. 1982. Management of the aster leafhopper and aster yellows in Wisconsin. Ph.D. Thesis, University of Wisconsin, Madison
Kriss, A.B., Paul, P.A. and L.V. Madden. 2012. Variability in fusarium head blight epidemics in relation to global climate fluctuations as represented by the El Niño- Southern Oscillation and other atmospheric patterns. Phytopathology 102(1): 55-64.
Kunkle, L.O. 1926. Studies on aster yellows. Am. J. Bot. 13(10):646-705.
Lee, I.-M., Davis, R.E., and D. E. Gunderson. 2000. Phytoplasma: Phytopathogenic mollicutes. Annu. Rev. Microbiol. 54:221-255.
99
Lee, I.-M., Gundersen-Rindal, D.E. and A. Bertaccini. 1998. Phytoplasma: ecology and genomic diversity. Phytopathology 88(12):1359-1366.
Lee, I.-M., Martini, M., Bottner, K. D., Dane, R.A., Black, M.C., and N. Troxclair. 2003. Ecological implications from a molecular analysis of phytoplasmas involved in an aster yellows epidemic in various crops in Texas. Phytopathology 93:1368-1377.
Lee, P.E. and A.G Robinson. 1958. Studies on the six-spotted leafhopper Macrosteles fascifrons (Stal) and aster yellows in Manitoba. Can. J. Plant Sci. 38:320-327.
Levin, S.A. 1992. The problem of pattern and scale in ecology. Ecology 73(6):1943-1967.
Lindblad, M. and P. Areno. 2002. Temporal and spatial population dynamics of Psammotettix alienus, a vector of wheat dwarf virus. Int. J. Pest Manage. 48:233-238.
Maden, L.V., Jeger, M.J., and F. van den Bosch. 2000. A theoretical assessment of the effects of vector-virus transmission mechanism on plant disease epidemics. Phytopathology 90(6): 576-594.
Madden , L.V., Turechek, W.W., and M. Nita. 2002. Evaluation of generalized linear mixed models for analyzing disease incidence data obtained by designed experiments. Plant Dis. 86:316-325.
Magnuson, J.T. 1990. Long-term ecological research and the invisible present. Bioscience 40(7):495-501.
Mahr, S.E.R. 1989. Development and aster yellows-infectivity of the aster leafhopper. Ph.D. Thesis, University of Wisconsin, Madison.
Mahr S. E. R., Wyman J. A., and R.K. Chapman. 1993. Variability in aster yellows infectivity of local populations of the aster leafhopper (Homoptera: Cicadellidae) in Wisconsin. J. Econ. Entomol. 86(5): 1522-1526.
Morey, A.C., Hutchison, W.D., Venette, R.C. and E.C Burkness. 2012. Cold hardiness of Helicoverpa zea (Lepidoptera: Noctuidae) pupae. Environ. Entomol. 41(1):172-179.
Munyaneza, J.E., Crosslin, J.M., Upton, J.E., and J.L. Buchman. 2010. Incidence of the beet leafhopper-transmitted virescence agent phytoplasma in local populations of the beet leafhopper, Circulifer tenellus, in Washington State. J. Insect Sci. 10:18.
Nita M., Ellis, M.A. and L.V. Madden. 2008. Variation in disease incidence of phomopsis cane and leaf spot of grape in commercial vineyards in Ohio. Plant Dis. 92:1053-1061.
100
Nobre, J.S. and J.M. Singer. 2007. Residual analysis for linear mixed models. Biometrical J. 49(6):863-875.
Peterson, A.G. 1973. Host plant and aster leafhopper relationships. Proc. North Cent. Branch Entomol. Soc. Am. 28:66-70.
Pinheiro, J.C. and D.M. Bates. 2000. Mixed-effects models in S and S-plus. Springer-Verlag New York, Inc. New York, New York 10010.
R Development Core Team (Version 2.15.0). 2012. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/.
Sagarin, R. and A. Pauchard. 2010. Observational approaches in ecology open new ground in a changing world. Front. Ecol. Environ. 8(7):379-386.
Schultz, G.A. 1979. Epidemiology of the aster leafhopper and aster yellows in relation to disease control. Ph.D. Thesis, University of Wisconsin, Madison.
Suzuki R. and H. Shimodaira. 2011. pvclust: Hierarchical clustering with p-values via multiscale bootstrap resampling. R package version 1.2-2. URL http://CRAN.R-project.org/package =pvclust.
Wallis, R.L. 1962. Host plant preference of the six-spotted leafhopper. J. Econ. Entomol. 55:998-999.
Westbrook, J.K. and S.A. Isard. 1999. Atmospheric scales of biotic dispersal. Agr. Forest Meteorol.97:263-274.
Westbrook, J.K., Wolf, W.W., Lingren, P.D., Raulston, J.R., Lopez Jr., J.D., Matis, J.H, Eyster, R.S, Esquivel, J.F., and P.G. Schleider. 1997. Early-season migratory flights of corn earworm (Lepidoptera: Noctuidae). Environ. Entomol. 26(1): 12-20.
Westdal, P.H. and H.P. Richardson. 1969. The susceptibility of cereals and wild oats to an isolate of the aster yellows pathogen. Can. J. Botany 47:755-760.
Wheatley, M. and C. Johnson. 2009. Factors limiting our understanding of ecological scale. Ecol. Complex. 6:150-159.
Wickham, H. 2009. ggplot2: elegant graphics for data analysis. Springer New York. ISBN 978-0-387-98140-6, URL http://had.co.nz/ggplot2/book.
Zhang, J., Hogenhout, S.A., Nault, L.R., Hoy, C.W., and S.A. Miller. 2004. Molecular and 101
symptom analyses of phytoplasma strains from lettuce reveal a diverse population. Phytopathology 94(8): 842-849.
Zhu, M., Radcliffe, E.B., Ragsdale, D.W., MacRae, I.V. and M.W. Seeley. 2006. Low- level jet streams associated with spring aphid migration and current season spread of potato viruses in the U.S. northern Great Plains. Agr. Forest Meteorol. 138:192-202.
102
Tables and Figures.
Table 1. Variance estimates for aster leafhopper abundance and infectivity from best fitting GLMM or LMM, respectively. Variance Variance % of total Component Estimate1 Variance Abundance Year x Week x Field2 0.924 28.0 Year x Week x Farm 0.892 26.2 Year 0.860 24.3 Day 0.478 7.5 Year x Day 0.452 6.7 Farm 0.431 6.1 Field 0.193 1.2
Infectivity Year 0.047 31.0 Year x Week 0.033 15.4 Week 0.017 4.0 Residual 0.054 49.7 1 Reported as a standard deviation; Percent of total variance calculated using variances (i.e. σ2). 2 Corresponds to observation level, or residual, variability.
103
Table 2. Correlation of aster leafhopper abundance among farms within year groups. Location Farm 1 Farm 2 Farm 3 Farm 4 Farm 5 Farm 2 0.59 Farm 3 0.89 0.89 Farm 4 0.89 0.72 0.91 Farm 5 0.80 0.80 0.92 0.85 Farm 6 0.95 0.65 0.89 0.87 0.70
104
Table 3. Correlation of aster leafhopper abundance among farms within year-week groups. Year Farm 1 Farm 2 Farm 3 Farm 4 Farm 5 Farm 2 -0.17 Farm 3 0.18 0.22 Farm 4 0.05 0.01 0.03 Farm 5 0.03 0.40 0.23 0.00 Farm 6 0.40 -0.09 0.16 -0.04 0.05 105
Table 4. Correlation of aster leafhopper infectivity among farms within year groups. Location Farm 1 Farm 2 Farm 3 Farm 4 Farm 5 Farm 6 Farm 7 Farm 2 0.64 Farm 3 0.74 0.63 Farm 4 0.74 0.64 0.73 Farm 5 0.67 0.58 0.66 0.67 Farm 6 0.63 0.53 0.62 0.63 0.71 Farm 7 0.52 0.44 0.51 0.52 0.57 0.60 Farm 8 0.08 0.03 0.08 0.08 -0.00 -0.03 0.01
106
Table 5. Correlation of aster leafhopper infectivity among farms within year-week groups. Location Farm 1 Farm 2 Farm 3 Farm 4 Farm 5 Farm 6 Farm 7 Farm 2 0.33 Farm 3 0.39 0.30 Farm 4 0.33 0.39 0.40 Farm 5 0.34 0.17 0.30 0.20 Farm 6 0.25 0.19 0.25 0.23 0.43 Farm 7 0.23 0.19 0.18 0.20 0.26 0.40 Farm 8 0.06 0.06 0.07 0.15 0.01 0.01 0.02
107 1.5 A ● 1.0 ● ●
0.5 ● ● ● ● 0.0 ● ●
−0.5
−1.0
Conditional Modes (Year) ● ● −1.5
2002 2004 2006 2008 2010 2001 2003 2005 2007 2009 2011 Year ● B ●
● ● ● ● 0.5 ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ●● ● ● ●●● ● ● ● ● ● ●● ● ● ● ● ● 0.0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ●● ● ● Conditional Modes (Day) ● ● ● −0.5 ● ● ● ● ● ● ●
140 160 180 200 220 240 260 May June July August Sept. Day of Year Figure 1. Conditional modes, or predictions at the population level given the random effects, for A) annual aster leafhopper abundance, ordered by year and B) seasonal aster leafhopper abundance, ordered by day. Generalized additive models were used to add smoothed curves to each plot to examine annual and seasonal trends of aster leafhopper abundance. 108
● A ●
0.05 ● ● ● ●
0.00 ● ● ● ● ● ●
● −0.05 ● Conditional Means (Year)
●
1994 1996 1998 2000 2002 2004 2006 2008 Year
● B ● 0.01 ● ● ● ● ● ● 0.00
● ● ● −0.01 ● ● Conditional Means (Week)
−0.02
●
140 160 180 200 220 May June July Aug. Day of Year Figure 2. Conditional modes, or predictions at the population level given the random effects, for A) annual aster leafhopper infectivity, ordered by year and B) seasonal aster leafhopper infectivity, ordered by day. Generalized additive models were used to add smoothed curves to each plot to examine annual and seasonal trends of aster leafhopper infectivity. 109
Farm 5 (M) A Farm 4 (S) B Farm 3 (M) Farm 2 (M)
Farm 5 (M) Farm 3 (M)
Farm 6 (S) Farm 4 (S)
Farm 1 (S) Farm 6 (S)
Farm 2 (M) Farm 1 (S)
0.25 0.20 0.15 0.10 0.05 0.400.450.500.550.600.650.700.75 Abundance Correlation Distance Abundance Correlation Distance (Farms within Year groups) (Farms within Year-Day groups)
C Figure 3. Dendrograms for visualizing the correlative relationships of aster Farm 4 Farm 6 leafhopper abundance among farms within Farm 1 A) year and B) year-day groups. Farms Farm 2 with predominantly sandy soils and with a Farm 3 soil organic fraction exceeding 65% are Farm 5 denoted as (S) and (M), respectively. C) Approximate geographic locations of farms sampled for aster leafhopper abundance and groupings implied by dendrograms (Similar farms shaded similarly). Farm Groupings Based on Abundance (Year-Day groups) 110
Farm 4 (M) Farm 6 (M) A B Farm 3 (M) Farm 5 (M) Farm 1 (S) Farm 7 (M) Farm 2 (M) Farm 4 (M) Farm 6 (M) Farm 1 (S) Farm 5 (M) Farm 3 (M)
Farm 2 (M) Farm 7 (M)
Farm 8 (S) Farm 8 (S)
0.20.40.60.81.0 0.400.450.500.550.600.650.70 Infectivity Correlation Distance Infectivity Correlation Distance (Farms within Year) (Farms within Year-Week)
C Figure 4. Dendrograms for visualizing the correlative relationships of aster leafhopper Farm 8 infectivity among farms sampled within A) Farm 1 year and B) year-week groups. Farms with Farm 2 Farm 3 predominantly sandy soils and with a soil Farm 4 organic fraction exceeding 65% are denoted Farm 5 as (S) and (M), respectively. C) Approximate geographic locations of farms and groupings Farm 7 implied by dendrograms (Similar farms shaded similarly). Farm 6
Farm Groupings Based on Infectivity (Year - Week groups) 111
Appendix A.
Example of R code used to examine the relative importance of factors that accounting for variation of aster leafhopper abundance and infectivity. Tables A1 and A2 contain the output from these models. library(lme4) ### Load lme4 package
### Create all grouping levels alh$YEAR<-factor(alh$YEAR) alh$JULIAN<-factor(alh$JULIAN)
###### Set up interactions ######## alh$YW<-interaction(alh$YEAR, alh$JULIAN); alh$YW<-factor(alh$YW) alh$FY<-interaction(alh$YEAR, alh$FARM); alh$FY<-factor(alh$FY) alh$FW<-interaction(alh$JULIAN, alh$FARM); alh$FW<-factor(alh$FW) alh$FYW<-interaction(alh$YEAR, alh$JULIAN, alh$FARM); alh$FYW<- factor(alh$FYW) alh$FdY<-interaction(alh$YEAR, alh$FIELD); alh$FdY<-factor(alh$FdY) alh$FdW<-interaction(alh$JULIAN, alh$FIELD); alh$FdW<-factor(alh$FdW) alh$FdYW<-interaction(alh$FIELD, alh$YEAR, alh$JULIAN); alh$FdYW<- factor(alh$FdYW)
#### Fit full model fglme1<-glmer(Tot_Cnt ~ 1 + (1 | YEAR) + (1 | JULIAN) + (1 | FARM) + (1 | FIELD) + (1|YW) + (1|FY) + (1|FW) + (1|FYW) + (1|FdY) + (1|FdW) + (1|FdYW), data = alh, offset = log(effort), family = "poisson", verbose = TRUE)
#### Fit submodel fglme2<-glmer(Tot_Cnt ~ 1 + (1 | YEAR) + (1 | JULIAN) + (1 | FARM) + (1 | FIELD) + (1|YW) + (1|FY) + (1|FYW) + 112
(1|FdY) + (1|FdW) + (1|FdYW), data = alh, offset = log(effort), family = "poisson", verbose = TRUE)
#### Use LRT to compare models anova(fglme1, fglme2)
#### Continue modeling process to produce parameter estimates in table A1 113
Table 1. GLMM coefficients describing aster leafhopper abundance. Model Model Model Model 1 Model 3 Model 5 Parameter 2 4 6 Fixed µ -0.82 -0.82 -0.82 -0.83 -0.82 -0.84 (0.33) (0.33) (0.33) (0.32) (0.32) (0.32) Random σa (Year) 0.851 0.851 0.851 0.861 0.860 (24.3%) 0.864 σb (Day) 0.488 0.488 0.488 0.476 0.478 (7.5%) 0.506 σc (Farm) 0.440 0.440 0.439 0.430 0.431 (6.1%) 0.426 σd (Field) 0.151 0.151 0.190 0.193 0.193 (1.2%) 0.192 σa x σb 0.472 0.472 0.473 0.451 0.452 (6.7%) - σa x σc 0.194 0.194 0.197 - - - σa x σd 0.134 0.134 - - - - σb x σc 0.000 - - - - - σb x σd 0.370 0.370 0.342 0.339 - - σa x σb x σc 0.871 0.871 0.870 0.895 0.892 (26.2%) 0.989 σa x σb x σd 0.843 0.843 0.858 0.859 0.924 (28.0%) 0.924 Evaluation df 12 11 10 9 8 7 AIC 13674 13672 13671 13671 13670 13676 χ2 deviance 0 1.3 2.0 0.91 7.9 P deviance 1 0.26 0.16 0.34 <0.005
114 library(lme4) ### Load lme4 package
### Create all grouping levels
alhi.sb$Year2<-factor(alhi.sb$Year) alhi.sb$YW<-interaction(factor(alhi.sb$Year), factor(alhi.sb$Week3)); alhi.sb$YW<- factor(alhi.sb$YW) alhi.sb$YF<-interaction(factor(alhi.sb$Year), alhi.sb$FL); alhi.sb$YF<-factor(alhi.sb$YF) alhi.sb$WF<-interaction(factor(alhi.sb$Week3), alhi.sb$FL); alhi.sb$WF<- factor(alhi.sb$WF)
##### Start modeling - Start with all RE terms drop those estimated at zero (or as per
#### Fit full model flme1<-lmer(sqrt(Prop) ~ 1 + (1 | Year) + (1 | Week3) + (1 | Farm) + (1 | YW) + (1 | YF) + (1 | WF), alhi.sb, REML = FALSE)
#### Fit submodel flme2<-lmer(sqrt(Prop) ~ 1 + (1 | Year) + (1 | Week3) + (1 | Farm) + (1 | YW) + (1 | WF), alhi.sb, REML = FALSE)
#### Use LRT to compare models anova(flme1, flme2)
#### Continue modeling process to produce parameter estimates in table A2 115
Table 2. Quantitative variability of aster leafhopper infectivity. Estimates Parameters Model 1 Model 2 Model 3 Model 4 Model 5 Fixed 0.139 0.139 0.139 0.139 0.141 µ (0.014) (0.014) (0.014) (0.014) (0.013) Random 0.047 σ 0.047 0.047 0.047 0.044 a (Year) (31.0%) σb (Week) 0.017 0.017 0.017 0.017 (3.9%) - σc (Farm) 0.000 0.000 0.000 - - 0.033 σ x σ 0.033 0.033 0.033 0.047 a b (15.4%) σa x σc 0.000 - - - - σb x σc 0.000 0.000 - - - 0.059 σ 0.059 0.059 0.059 0.059 r (resid) (49.7%) Evaluation df 8 7 6 5 4 AIC -925 -927 -929 -931 -929 χ2 deviance 0 0 0 4.0 P deviance 1 1 1 <0.05
116
Appendix B
Example of the R code used to examine the correlations among farms and covariance matrix induced by inclusion of random effects. Here we express the model in matrix notation as:
G(ui) = Xi x β + Zi x bi + εi where bj ~ N(0, D), εi ~ N(0, Σi), and i indexes year.
For a 3 farm example the covariance matrix for the random effects is:
2 d 11 d12 d13 2 D = d21 d 22 d23 2 d31 d32 d 33
This model can also be expressed:
G(ui) ~ N(Xi x β, Vi) where Vi = Zi x D x Zi′ + Σi
Here, 1 0 0 Zi = Zi′ = 0 1 0 0 0 1 and 1 0 0 2 Σi = σ x 0 1 0 0 0 1
Resulting in 2 2 σ + d 11 d12 d13 2 2 Vi = d21 σ + d 22 d23 2 2 d31 d32 σ + d 33
The notation is different than what is presented in the body of the paper. Here the covariance of any two farms within year is d and the variance of a farm within year is σ2 + d2. The
2 2 correlation between any two farms can be determined using division (i.e. d21 / sqrt( σ + d 11)
2 2 x sqrt( σ + d 22).
117
## R code to examine correlation of farms within year groupings fglmeCor4<-glmer(Tot_Cnt ~ 0 + FARM + (0 + FARM | YEAR), data = alh, offset = log(alh$effort), family = "poisson") summary(fglmeCor4)
118
Chapter 4: Seasonal pattern of aster leafhopper (Macrosteles quadrilineatus) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields 119
Abstract
In Wisconsin, vegetable crops are threatened annually by the aster yellows phytoplasma (AYp), which is obligately transmitted by the aster leafhopper. Using a multi- year, multi-location data set, seasonal patterns of leafhopper abundance and infectivity were modeled. A seasonal aster yellows index (AYI) was deduced from the model abundance and infectivity predictions to represent the expected seasonal risk of pathogen transmission by infectious aster leafhoppers. The primary goal of this study was to identify periods of time during the growing season when crop protection practices could be targeted to reduce the risk of AYp spread. Based on abundance and infectivity, the annual exposure of the carrot crop to infectious leafhoppers varied by 16-fold and 70-fold, respectively. Together, this corresponded to an estimated 1000-fold difference in exposure to infectious leafhoppers.
Within a season, exposure of the crop to infectious aster leafhoppers varied 3-fold due to abundance and 9-fold due to infectivity. Periods of above average aster leafhopper abundance occurred between 11 June and 2 August and above average infectivity occurred between 27
May and 13 July. A more comprehensive description of the temporal trends of ALH abundance and infectivity provides new information defining when the ALH moves into susceptible crop fields and when they transmit the pathogen to susceptible crops.
Keywords: Macrosteles quadrilineatus, aster yellows phytoplasma, migration, dispersal, arthropods in relation to plant disease 120
Aster yellows (AY) is a widespread disease of plants caused by the aster yellows phytoplasma (AYp), a small, wall-less prokaryotic organism that is currently placed in the provisional genus Candidatus (Lee et al. 2000, IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group 2004). The AYp has an extensive and diverse host range infecting over 350 plant species including many common vegetable, ornamental, and agronomically important field crops, and several non-crop plant species (Kunkel 1926,
Chiykowski 1965, Chiykowski and Chapman 1965, Chiykowski 1967, Westdal and
Richardson 1969, Peterson 1973, Lee et al. 1998, Lee et al. 2000, Lee et al. 2003,
Hollingsworth et al. 2008). The most common disease phenotypes are vein clearing, chlorosis, stunting, twisting and proliferation of plant stems, and the development of adventitious roots
(Kunkle 1926, Bloomquist 2002). In vegetable crops, these symptoms can lead to direct yield and quality losses and, for root vegetables specifically, processing problems can result from an inability to obtain clean raw product due to adventitious root growth and associated field soil contamination.
AYp has been reported to be circulative and propagative in the ALH (Maramorosch
1952, Sinha and Chiykowski 1967, Lee et al. 2000), and vector competence involves acquisition, pathogen replication and circulation to result in successful transmission to a susceptible host (Matthews 1991). Plant-to-plant spread of AYp in the field can occur as a result of transmission by more than twenty-four leafhopper species (Mahr 1989, Christensen et al. 2005). However, the aster leafhopper, Macrosteles quadrilineatus Forbes, is considered to be the primary vector of the AYp due to its prevalence in Midwestern susceptible crops
(Drake and Chapman 1965, Hoy et al. 1992). 121
Similar to most plant pathogens spread by arthropods, risk for spread of AY to a susceptible crop is a function of aster leafhopper abundance and transmission capability, or infectivity (Madden et al. 2000, Jeger et al. 2004). In Wisconsin, aster yellows management has focused primarily on controlling the insect vector, the aster leafhopper, and an AY risk index, known as the aster yellows index (AYI), was developed to describe the maximum allowable numbers of infectious leafhoppers and define periods of time when plant protection is most needed (Chapman et al. 1971). The AYI metric is the product of aster leafhopper infectivity, or percent of infectious aster leafhoppers, and the magnitude of the aster leafhopper population, or the number of aster leafhoppers in 100 (pendulum) sweeps using a standard 38 cm sweep net. Originally, the AYI was used to make insecticide spray recommendations based on a series of early season leafhopper collections (Chapman 1971,
Chapman 1973). However, following the observations that aster leafhopper abundance and infectivity, in and around carrot fields, varied spatially and temporally (Mahr et al. 1993), efforts were made to refine AYI estimates for a specific date and field. However, even with the availability of contemporary tools, significant annual and site-specific variation in pathogen detection in the insect vector frequently occurs. In most cases, the relationship between pathogen presence in the vector and the vector’s ability to successfully transmit the pathogen is not known. In turn, many producers avoid risk of pathogen spread by using inexpensive, prophylactic insecticide applications, a management practice that circumvents the utility of the AYI.
In Wisconsin, aster yellows has typically been controlled using repetitive applications of insecticidal compounds in the synthetic pyrethroid group. Although successful from the 122 perspective of managing insect pests in a cost-effective manner, this approach presents considerable risk, since these insecticides are older, broad-spectrum compounds with documented mammalian toxicity (Wolansky and Harrill 2008). The chemicals in this group are also harmful to aquatic organisms, are lipophilic, and in aquatic environments, tend to adsorb to organic sediments (Gan et al. 2008). Monitoring surveys in the U.S. have detected the presence of synthetic pyrethroid residues in the sediment of both agricultural and urban dominated waterways (Werner 2002, Weston 2004). These findings have prompted concerns about pyrethroid exposure to non-target areas, especially ecologically sensitive areas such as wetlands, which include the low-land muck soils where the majority of Wisconsin carrot is grown. Thus, it has been our goal to reduce the nearly exclusive reliance on synthetic pyrethroid insecticides for the management of aster yellows in carrot. We hypothesize that an improved understanding of the expected AY risk based on historical data, combined with an understanding of the “window” of time during which these events occur, will improve aster yellows management. Specifically, grower adoption of reduced-risk (RR) insecticides may be improved by targeting insecticide applications to periods during which AY risk is high thereby reducing the number of applications of the more expensive RR insecticides necessary to control aster yellows improving the cost-efficiency of these newer tools.
We have previously examined factors that contribute to the variability of aster leafhopper abundance and infectivity (Frost et al. 2012). However, we did not directly model the patterns of variation associated with those factors. An outcome of our previous analysis noted significant residual trends in the seasonal patterns of aster leafhopper abundance and infectivity which could be directly modeled. Nonparametric regression and additive models 123 are extensions of linear models that allow for non-linear relationships between the response variable and multiple predictor variables. In these models, the goal is primarily to describe the data in a way in which complex functions serve as best predictors making the models flexible and preserving model interpretability (Hastie et al. 2009). Recently their application to examine trends in long term data-sets has become useful. For example, these methodologies have been used to examine environmental drivers affecting long-term trends in water quality
(Ferguson et al. 2008), sperm whale habitat preference (Pirotta et al. 2011), shifts in insect phenology due to climatic variation (Hodgson et al. 2011), and seasonal trends of aphid dispersal into agricultural fields (Nault et al. 2009). Our primary goal in the current study is to identify periods of time in the growing season when crop protection is most needed. In this paper, generalized additive mixed models are used to: I) describe the pattern that represents the expected seasonal aster leafhopper abundance and infectivity in the carrot crop, II) predict the expected seasonal aster leafhopper abundance and infectivity to deduce a seasonal AYI that best represents the periods of risk for exposure to infectious leafhopper, and III) retrospectively quantify the frequency and magnitude of aster leafhopper infestation events.
For IPM practitioners, the identification of temporal trends of abundance and infectivity can improve our knowledge of where leafhoppers acquire the pathogen, when they move into susceptible fields, and when they spread the pathogen to susceptible crops.
Materials and Methods
Data description. A detailed description of the data set, sample sites and sampling methodologies, has previously been reported (Frost et al. 2012). Briefly, field sampling was conducted using sweep nets in commercial carrot fields to monitor the relative abundance of 124 aster leafhopper in the common carrot production areas of Wisconsin from 2001 through
2011. A total of 237 fields were sampled over the 11-year span of this survey. Abundance was estimated for an average of 31 fields per year with several fields re-sampled in multiple years due to crop rotation practices. Abundance was determined by sweep net sampling along 2 to
18 transects that were walked into the carrot crop toward the middle of the field. Twenty-five to 100 (pendulum) sweeps per transect were conducted using a standard sweep net (38 cm diameter) and aster leafhoppers were counted and enumerated as aster leafhoppers per 25 sweeps. Decimal values were rounded to the nearest integer. Aster leafhopper infectivity was monitored using an infectivity bioassay in the commercial carrot production areas and records of infectivity were available from 1994 through 2007.
The rate of infectious aster leafhoppers was estimated for a total of 378 insect populations from 1994 through 2008. Approximately 25 populations were collected and assayed per year with multiple geographic locations and several dates represented throughout each growing season. At each sample location and date, aster leafhoppers were collected in sweep nets and placed onto oat seedlings for transport back to the laboratory. When possible,
204 leafhoppers were placed, in pairs, onto 102 Chinese aster (Callistephus chinensis) plants and allowed to feed for the duration of the experiment. Disease symptoms were assessed after a two week incubation period and percent infectivity was calculated as:
Infectivity = # of diseased plants / Total # of leafhoppers
The total number of leafhoppers was used as the denominator because infectivity levels are low and a diseased plant was more likely due to a single infective leafhopper rather than the presence of two infective leafhoppers on the same plant. 125
Statistical analysis. We used generalized additive mixed models (GAMM) to describe the annual and seasonal trends of aster leafhopper abundance and infectivity. A GAMM is an extension of a generalized additive model that relaxes the underlying assumption that the data are independent by allowing observations to be correlated (Zuur et al. 2009). One advantage of using a GAMM to describe seasonal leafhopper abundance is that we can estimate the underlying trend from the data without assuming the trend has any specific functional form
(Wood 2006). In these models, the goal is primarily to describe the data in a way in which complex functions serve as best predictors – without needing to understand the complex mathematical representation of the model; trends are represented by mathematical functions, but those mathematical representations are not particularly intuitive when written down
(Wood 2006, Agresti 2007). Therefore, to evaluate the form of the function, or trend in the data, we produced model predictions given a new set of data containing all model covariates.
From our models, the seasonal predictions of leafhopper abundance and infectivity were used to deduce the expected AYI for the average growing season to define the interval in which elevated risk for crop exposure to infectious leafhoppers was greatest. Additionally, the strategy we used to confront the presence of annual variability associated with leafhopper abundance and infectivity was to include seasonal estimates for the years with the highest and lowest observed annual abundance and infectivity. This approach is in contrast to our previous study where we chose to characterize the distributions for each level of grouping in our data.
Aster leafhopper abundance. A GAMM was used to describe the seasonal pattern of aster leafhopper abundance (Yi) as a function of calendar day (xi) with each field-year combination 126 and observation represented as a random effect in the model. This GAMM could be represented as:
Yij ~ Poisson(g(µij(abc))) (Model 1)
g(µij(ab)) = loge(µij(abc)) = offset(effort) + βi + f(dj) + ε(a) + ε(b) + ε(c)
2 ε(a) ~ N(0, σ farm),
2 ε(b) ~ N(0, σ field),
2 ε(obs) ~ N(0, σ obs), where βi corresponded to the average abundance estimate for each year (i) and f(dj) was a smoothing function (penalized cubic regression spline) of the calendar day covariate (di). The model contains an offset corresponding to the log-transformed number of transects (or effort) used to estimate leafhopper abundance, and ε(a), ε(b), and ε(c) are the random effects for each farm, field, and observation, respectively. The model was specified using the gamm4 function in the gamm4 package (Wood 2006) of R (version 2.15.0; R Development Core Team 2012) and generalized cross validation was used to estimate the value of the smoothing parameter for the unknown scale parameter (Wood 2004, Wood 2006, Zuur et al. 2009).
The component smooth function f(di), centered on the scale of the linear predictor, was plotted versus calendar date (di) together with the partial residual of the model to represent the seasonal trend of aster leafhopper abundance. Thus, the period of elevated leafhopper abundance was estimated by visual examination of the plotted data (i.e. f1(di) vs. di). Predictions (PA) of leafhopper abundance, and associated standard errors (SE) of the predictions, were obtained from the fitted GAMM for each calendar day in low, typical, and high abundance years. Confidence intervals for each day were estimated as PA ± 2*SE. 127
Confidence intervals were back-transformed to the response scale and multiplied by four, to obtain leafhoppers per 100 sweeps, before being used in the calculation of the seasonal AYI.
ALH infectivity. A GAMM was also used to examine seasonal trends of leafhopper infectivity. This model was used to fit aster leafhopper infectivity (Yi; sqrt-transformed) to calendar day (xi) with year represented as a random effect in the model. This GAMM could be represented as follows:
2 Yi ~ N(g(µi), σ ) (Model 2)
g(µij) = µij = βi + f (di) + εa + εij
2 εi ~ N(0, σ i)
2 εij ~ N(0, σ r), where βi corresponded to the mean infectivity estimate for each year (i) and f(di) was a smoothing function (penalized cubic regression spline) of calendar day (di). In this model, εa corresponded to the random effects term for farm and εij represented the residual error. The method used to fit this function was the same as the one previously described.
The component smooth function, f(di), of the model fit was plotted versus calendar date (di) representing the seasonal trend of aster leafhopper infectivity. Again, the period of elevated infectivity was estimated by visual examination of the plotted data (i.e. f1(di) vs. di).
Predictions (PI) of leafhopper infectivity, and associated standard errors (SE) of the predictions, were obtained from the fitted GAMM for each day in low, typical, and high infectivity years. Similar to abundance, infectivity confidence intervals were estimated for infectivity as PI ± 2*SE. Predictions and confidence intervals were squared and expressed as percentages before being used in the calculation of the seasonal AYI. 128
Aster Yellows Index. The number of aster leafhoppers and the associated rate of infectious individuals in the leafhopper population affect the exposure of a crop to infection by a pathogen and subsequent disease development. Therefore, the seasonal AY risk was deduced using the daily model predictions of aster leafhopper abundance (PAi; Model 1 above) and infectivity (PIi; Model 2) into an AYI metric as follows:
AYIi = PAi * PIi (Model 3) where i indexes calendar day. The seasonal AYI was calculated for low, typical, and high abundance and infectivity years to represent the range of the observed data. Since the calendar day range of the infectivity data set was shorter than the abundance data set, predictions of infectivity after calendar day 228 were estimated as the prediction on day 228. The AYI is essentially an assessment of the potential annual and seasonal “risk” of crop exposure to aster leafhoppers capable of transmitting AYp. In the results section, we define the relative exposure “risk” as the exposure of the carrot crop to infectious leafhoppers relative to the exposure of some reference group, usually the low-exposure AYI. For example, holding infectivity constant, a field with an abundance of two leafhoppers (per 100 sweeps) has a 2- fold greater exposure to infectious leafhoppers than a field with an abundance of one leafhopper (per 100 sweeps). Similarly, a field in which the rate of infectious leafhoppers is
3% has a 3-fold greater exposure to infectious leafhoppers than a field in which only 1% of the leafhoppers are capable of transmitting AYp.
Correlation of Abundance and Infectivity. The correlation between annual estimates of leafhopper abundance and infectivity was calculated for the years in which the data overlapped (2001- 2008). To examine if the average seasonal abundance and infectivity were 129 correlated, a preliminary examination of the cross-correlation between leafhopper abundance and infectivity model predictions was conducted. Since infectivity was only measured weekly, model predictions were extracted for the weeks in which infectivity estimates were available.
The cross-correlations were calculated and plotted versus the weekly time lags and the peaks in the cross-correlations represent the phase shift between leafhopper abundance and infectivity within the growing season.
Quantifying the frequency of leafhopper influxes. We defined an influx event as an occurrence when the observed number of aster leafhoppers at a field was greater than the allowable number of leafhoppers calculated using predetermined AYI values of 25, 50, 75, and 100, corresponding to high, medium, medium-low, and low susceptibility, respectively.
These AYI values represented the nominal thresholds that have been determined based on professional experience and experiments examining host plant resistance to aster yellows
(Pedigo1989, Foster and Flood 2005). For this calculation, infectivity was allowed to vary across the growing season as predicted by model 2 for the typical growing season. Allowable leafhoppers were calculated as follows:
Allowable leafhoppersi = AYI / PIi where i indexes calendar day. Indicator variables were used to code the occurrence of an event or non-event (allowable leafhoppersi < observed leafhoppersi = 1; allowable leafhoppersi > observed leafhoppersi = 0). The total number and proportion of events in each magnitude category was calculated for each year, week, farm and field. A GAMM (family = binomial) was used to examine the seasonal probability of detecting leafhopper abundances 130 above an AYI threshold value given an average seasonal infectivity. These models could be represented as follows:
Yi(abc) ~ Binomial(µi(abc), ni(abc)) (Model 4)
g(µi(abc)) = logit(µi(abc)) = f(di) + ε i(a) + ε i(b) + ε i(c)
2 ε i(a) ~ N(0, σ a),
2 ε i(b) ~ N(0, σ b),
2 ε i(c) ~ N(0, σ c), where f(di) was a smoothing function (penalized cubic regression spline) of calendar day (di).
In this model, εi(a), εi(b), and εi(c) corresponded to the random effects term year, farm, and field, respectively. As described above, this model was specified using the gamm4 function in the gamm4 package (Wood 2006). The component smooth function, f1(di), on the scale of the response, was plotted versus calendar date (di) to represent the expected seasonal probability of observing aster leafhopper abundance that would prompt a control practice.
Results
Abundance trends. Annual. Here we chose to present year modeled as a fixed effect because we were primarily interested in characterizing the within season trends and the annual trends can also be inferred from the graphical presentation of the annual means (Fig. 1). Overall, the average annual aster leafhopper abundance decreased from 2001 to 2011, but also varied greatly among years with average abundance ranging from 0.08 to 1.27 insects per 25 pendulum sweeps (Table 1; Fig. 1A). This implies that, for the year with the highest observed aster leafhopper abundance, the exposure of the crop to infectious individuals would be approximately 16-fold greater than the year with the lowest leafhopper abundance (i.e. 131 holding infectivity constant).
Seasonal. Within year, a plot of the component smooth function for abundance versus calendar date indicated that periods of above average aster leafhopper abundance occurred between 11 June and 2 August (Fig. 1B). The component smooth function for aster leafhopper count ranged from -0.6 to 1.0 (loge-scale), which would represent a 5-fold change in the exposure of the crop to infectious leafhoppers over the course of the growing season, holding infectivity constant. However, for the period in which a pest management practice would be implemented, 1 June through 31 August, aster leafhopper abundance was only 3- fold higher than the lowest estimated leafhopper abundance for the entire growing season.
Residual plots indicated that some seasonal (within year) trends remained after fitting model
1. These trends could be removed by fitting separate smoothing functions for each year (not shown), however, the overall pattern observed by year was described by the simpler model.
Additionally, we presented the simpler model here because it was more consistent with the objectives of our study to determine if, on average, there was a critical time interval when crop protection is most needed.
Infectivity trends. Annual. Similar to abundance, aster leafhopper infectivity varied among years and the average annual infectivity ranged between 0.09 and 6.25% (Table 2). Thus, the exposure of the crop to infectious leafhoppers in the year with the highest rate of infectivity was 75-fold higher than the year with the lowest rate of infectivity (i.e. if leafhopper abundance was held constant). Visually, there was no overt trend in the average annual leafhopper infectivity from 1994 to 2008 (Fig. 2A).
Seasonal. Within year, plots of the component smooth function of infectivity versus calendar 132 date indicated that periods of above average infectivity occurred between 27 May and 13 July
(Fig. 2B). Similar to leafhopper abundance, residual plots indicated that some seasonal
(within year) trends remained after fitting model 2. These trends were removed by fitting separate smoothing functions for each year (not shown). Our rational for presenting the simpler model remains as previously stated.
Seasonal aster yellows risk. From May through August, the typical (or expected) seasonal aster leafhopper abundance ranged from 1 to 3 leafhoppers per 100 sweeps (Fig. 3A).
However, for the year with the highest observed leafhopper population sizes, the expected leafhopper abundance varied between 3 and 9 leafhoppers per 100 sweeps. Similarly, the typical infectivity between May and mid-August ranged from 0.3% to 2.6%, representing a 9- fold increase in exposure of carrot to infectious individuals due to the seasonality (Fig. 3B).
The deduced AYI for the state of Wisconsin incorporates the seasonal variability of leafhopper abundance and infectivity estimates (Fig. 3C). In a typical year, the seasonal AYI ranged from approximately 1 to 8. In years when both leafhopper abundance and infectivity are high, the expected AYI peaks at 60 although an AYI of 130 is not unexpected given the error associated with the seasonal trend. Taken together, estimates of a critical ‘risk window’ or timing interval in which aster leafhopper management could be focused, were similar to previous estimates (Table 3; Frost et al. 2012).
Correlation of abundance and infectivity. There was no correlation between annual estimates of leafhopper abundance and infectivity for the time interval 2001 through 2008 (R
= 0.05, P = 0.90). Within a season, the correlation between the aster leafhopper abundance and infectivity was examined using a cross-correlogram (Fig. 4). On average, over the 11 year 133 term of this data set, within season leafhopper abundance and infectivity estimates were negatively correlated at time lags that ranged between -3 and -5 weeks. In contrast, leafhopper abundance and infectivity were positively correlated at time lags of 0 to 3 weeks.
Frequency of occurrence of aster leafhopper influxes. Over the 11 years of this data set, there were 583 occasions when the observed aster leafhopper abundance (an “influx” event) was greater than the allowable abundance (established AYI thresholds) given an average expected seasonal infectivity (Table 4). This corresponds to approximately 1.8 events per field across all years. However, the highest number of influx events (146) occurred in 2002, corresponding to an average of 4.4 events per field. In contrast, only 1 influx event occurred in 2009 for all 25 fields sampled corresponding to an estimated 0.04 events per field. In general, the number of influx events has decreased since 2001 which is consistent with the trend of the annual leafhopper abundance over the same time, and the magnitude of the aster leafhopper events has also decreased. Among farms, influx events, determined as a proportion of total observations, were not evenly distributed and ranged from 3.5% to 29.5% (Table 5).
Moreover, predicted influx events were not evenly distributed throughout the growing season (Fig. 5). The seasonal dynamics of the probability of observing an influx event was similar to the pattern observed for the expected seasonal abundance with the peak probability
(14.8%) of observing any event occurring on July 2 (184). However, the timing of the peak probability of observing an influx event was earlier for larger influx events. For example, the peak probability of observing an event with an associated AYI of 25, 50, 75 and 100, occurred at 11 July (193), 5 July (187), 1 July (183), and 26 June (178), respectively.
Discussion 134
For aster yellows management in Wisconsin, the aster yellows index combines insect vector abundance and transmission capability to describe the maximum allowable numbers of infectious leafhoppers that can be tolerated on a susceptible crop providing a an indication of when crop protection is most needed (Chapman 1971, Chapman 1973). In this paper, we used a multi-year data set, and multi-location modeling approach to generate reliable estimates for the expected seasonal patterns of aster leafhopper abundance and transmission capability, or infectivity. The predicted seasonal leafhopper abundance and infectivity were then used to deduce a seasonal aster yellows index to represent the expected seasonal exposure to infectious aster leafhoppers that may spread AYp to a susceptible crop. Additionally, we used the expected seasonal infectivity to retrospectively determine the frequency at which the observed insect abundance would have exceeded the established AYI thresholds resulting in an insecticide application.
Annual trends of abundance and infectivity. The exposure of a carrot crop to infectious aster leafhoppers was approximately 16-fold greater in the year with the highest observed aster leafhopper abundance than the year with the lowest leafhopper abundance.
Similarly, the exposure of the crop to infectious leafhoppers was 70-fold higher in the year with the highest rate of infectivity when compared to the year with the lowest rate of infectivity. Taken together, these observations suggest that years in which high aster leafhopper abundance occurs co-incidentally with high infectivity can result in 1000-fold greater exposure of the carrot crop to infectious leafhoppers when compared to years in which low leafhopper abundance is co-incident with low infectivity. These observations are consistent with the large annual variability of aster yellows pressure previously reported in 135
Wisconsin (Chapman 1971, Chapman 1973, Mahr et al. 1993) and consistent with our previous findings of large annual variability of leafhopper abundance and infectivity (Frost et al. 2012).
We found a low correlation among annual estimates of aster leafhopper abundance and infectivity, which would support the hypothesis that aster leafhopper abundance and infectivity are independent quantities varying among years. It has been determined that fitness is increased in aster leafhopper individuals infected by AYp (Beanland et al. 2000, Sugio et al. 2011), but it is not known how the increase in fitness translates to the in-field population dynamics of infectious insects. Our correlations were calculated using the eight years of data for which we had estimates of both abundance and infectivity. However, it is likely that a fitness effect may appear in the field as a lagged correlation (i.e. high abundance lags high infectivity by a year or two) and this would require many more paired annual estimates of leafhopper abundance and infectivity than are currently available.
A year when the aster leafhopper is abundant and a high proportion of the population is capable of transmitting AYp may occur with a relatively low frequency and many years may pass before the two conditions coincide (Magnuson 1990). Assuming independent events and using previous estimates of among year variability (i.e. loge(abundance) ~ N( -0.82,
0.8602) and sqrt(infectivity) ~ N(0.14, 0.0472); (Frost et al. 2012)), we simulated values for the average annual leafhopper abundance and infectivity. These abundance and infectivity values were then back-transformed and used to calculate AYI values to represent a 1000 year time frame. We found that an average annual AYI that was 2-, 3-, 5- and 10-fold greater than the median annual AYI would occur, on average, every 4, 6, 13 and 53 years, respectively. 136
These time estimates may represent a null model from which the assumption of independence among years may be tested. Nevertheless, the high among year variability of abundance and infectivity is consistent with the hypothesis that the spring migration of the aster leafhopper contributes to the annual risk of AY epidemics in Wisconsin. Additionally, the large annual variability in the rate of infectious leafhoppers suggests that, in addition to weather events affecting insect trajectories and deposition (Hurd 1920, Huff 1963, Westbrook and Isard 1999,
Zhu 2006), the prevalence and heterogeneity of AYp-infected feeding hosts of the leafhopper in the landscape along the migratory route, or acquisition trajectory, may also be important factor contributing to annual infectivity (Carter 1961, Lee et al. 2003).
Seasonality of abundance and infectivity. Aster leafhopper abundance varied by approximately 3-fold during the period of the growing season in which pest management would normally be implemented. Additionally, the exposure of the carrot crop to infectious leafhoppers within a year varied by as much as 9-fold. Taken together, carrot growers can expect the relative exposure to infectious aster leafhoppers to vary by as much as 30-fold throughout the growing season. However, the periods during which aster leafhopper abundance and infectivity tend to be above average often overlap within the growing season.
Thus, without information about insect abundance and infectivity for a specific field, the coincidence of these expected periods of high population sizes and increased infectivity represent a timing interval in which management of the insect could be focused to limit pathogen spread.
Within a season, however, it is unlikely that aster leafhopper abundance and infectivity are independent. In theory, the mixing of aster leafhopper populations with 137 differing proportions of infectious aster leafhoppers can result in variable proportions of infectious leafhoppers in the population. Recently, Bressan et al. (2011) described a pathosystem in which the abundance of the plant hopper, Pentastiridius leporinus, was directly related to the proportion of individuals carrying the pathogen, ‘Candidatus
Arsenophonus phytopathogenicus’, in the population. A key difference between the system that Bressan et al. (2011) described and our system is that transovarial, or vertical, transmission is not known to occur in our system. Bressan et al. (2011) also measured the proportion of insects carrying the pathogen which may be different than the proportion of infectious planthoppers.
In Wisconsin, a dilution of aster leafhopper infectivity may occur as aster leafhoppers that overwinter in Wisconsin as eggs (local population) begin to emerge from their habitats in early June, although we have no direct evidence for this phenomenon (Drake and Chapman
1965). A preliminary examination of the cross-correlation of the average within season dynamics of aster leafhopper abundance and infectivity showed that infectivity was negatively correlated with abundance with a -3 to -5 week lag. This is counterintuitive given that leafhopper fitness is increased by AYp infection (Beanland et al. 2000, Sugio et al. 2011).
However, the negative correlation could be due to grower management resulting from plant protection practices implemented when infectivity is known to be high. Leafhopper abundance was positively correlated with infectivity at a 0 to 3 week lag suggesting that higher vector abundance results in AYp spread and higher rates of infectious leafhoppers in subsequent generations, an observation consistent with the findings of Sisterson (2009). In general, more research is needed to examine the within season relationship between 138 leafhopper abundance and infectivity in the field to determine the reasons for the correlations at time lags of 3 to 4 weeks, which is approximately the generation time of the aster leafhopper (Mahr 1989). Thus, the emergence of local populations, or offspring from early migratory populations, variability in birth and death rates between inoculative and non- inoculative individuals, and insect management practices could all contribute to the observed phase shifts between abundance and infectivity within the growing season.
Frequencies and magnitudes of aster leafhopper influxes. The arrival of large numbers of insects into a field (or an influx) has the potential to affect epidemic progression and the spring migration of the aster leafhopper has long been considered the principle source of early season infectious aster leafhoppers (Chiykowski and Chapman 1965, Hoy 1992). We have defined a leafhopper influx as an aster leafhopper abundance that exceeds a predetermined AYI threshold given the expected seasonal aster leafhopper infectivity. The selected AYI thresholds are used by growers to make spray decisions and are nominal thresholds determined by practitioner experience and known cultivar resistance to aster yellows (Pedigo 1989, Foster and Flood 2005). We found that the occurrence of insect influxes that would elicit an insecticide application coincided with the period of time when higher numbers of leafhoppers are expected. Our approach did not distinguish between an increase of insect numbers occurring because of population growth, local insect movements, or long-distance insect movements. However, the probability of influx events occurring in
May is lower than the probability of influxes occurring in mid to late-June. A similar phenology has been observed for Circulifer tenellus (beet leafhopper), Psammotettix alienus
(European grass-feeding leafhopper), and Graphocephala atropunctata (blue-green 139 sharpshooter), all of which achieve peak abundance in June followed by population declines in late July and August (Lindblad and Areno, 2002, Munyaneza et al., 2008, Gruber and
Daugherty 2012). It may be that these leafhoppers all overwinter, or diapause, in the same life stage (i.e. eggs) and have to develop through a similar number of instar stadia (i.e. 4-5 instars) leading to a reasonably synchronous emergence as adults among species.
Alternatively, the occurrence of the observed leafhopper influxes beginning in early June may be due to the emergence and dispersal of the local leafhopper population from their overwintering host (i.e. winter wheat) to more succulent irrigated vegetable crops present in the landscape at that time (Carter 1961).
The number of occurrences when a spray would be recommended has decreased since
2001 which is consistent with the overall decrease in the annual aster leafhopper abundance occurring over this time period. Additionally, the number of occurrences when a spray would be recommended varied among farm (or farm locations). This could be due to the differential influence of the landscape surrounding each farm and its effect upon the reproductive capability of the aster leafhopper in these local environments. For example, the aster leafhopper uses over 300 different plant species for food, oviposition, and shelter (Lee and
Robinson 1958, Wallis 1962, Peterson 1973), and the distribution and abundance of these species surrounding each farm likely differs. It could also be due, in part, to differences in grower management of their crop where some growers tolerate higher insect abundances prior to applying an insecticide; some growers tolerate a modest, but ephemeral leafhopper population prior an insecticide application, and finally other growers may prefer to apply more regular, prophylactic insecticide treatments which functionally maintain low leafhopper 140 populations.
Management Implications: Control of pathogens transmitted by insects in a persistent manner tends to be less difficult than pathogens transmitted non-persistently (Chapman 1973,
Madden et al. 2000). The decision to intercede and implement a pest control action requires an understanding of the level of insect infestation a crop can tolerate without incurring economic loss (Pedigo 1989). In Wisconsin, growers currently achieve adequate control of aster yellows in their crops using repetitive applications of insecticidal compounds in the synthetic pyrethroid group. In a given year, it is common for as many as 5-7 applications of an insecticide to be applied on a 7-10 day calendar schedule on the same crop. In part, the rationale behind these repetitive spray applications results from the fact that a single insecticide application is inexpensive when compared to newer, reduced-risk and less broad spectrum insecticides that target fewer homopterous pests.
In this study, we identified a timing interval in which management of the aster leafhopper could be focused. This interval, or ‘risk window’ results from the coincidence of above average leafhopper abundance and higher observed infectivity. Additionally, because this timing interval occurs early in the season, there exists the opportunity to deploy newer, reduced-risk, systemic insecticides with flexible application methods (i.e. seed treatments, in- furrow, or layby incorporation) as potential pest management alternatives for long term control of the aster leafhopper. When compared to current management practices (and if successful), the use of these new insecticides and associated delivery systems have the potential to increase the sustainability and profitability of carrot production, enhance natural enemy populations and biological control, and reduce adverse effects on farm workers and 141 applicators, as well as the local environment. 142
Acknowledgments. We thank Dr. Emily Mueller for her constructive criticism and comments on earlier versions of the manuscript. Funding support was provided by the USDA Specialty
Crops Research Initiative through the Wisconsin Specialty Crops Block Grant Program (MSN
129013). 143
References
Agresti. A. 2007. An Introduction to Categorical Data Analysis, Second Edition. Wiley- Interscience, Hoboken, NJ.
Beanland, L., Hoy, C.W., Miller, S.A., and L.R. Nault. 2000. Influence of aster yellows phytoplasma on the fitness of aster leafhopper (Homoptera: Cicadellidae). Ann. Entomol. Soc. Am. 93(2):271-276.
Bloomquist, C.L. 2002. Aster yellows and beet leafhopper transmitted virescence agent yellows. Pages 58-59 in: Compendium of Umbelliferous crop diseases.
Bressan. A., Garcia, F.J.M, and E. Boudon-Padieu. 2011. The prevalence of ‘Candidatus Arsenophonus phytopathogenicus’ infecting the plant hopper Pentastiridius leporinus (Hemiptera: Cixiidae) increase nonlinearly with the population abundance in sugar beet fields. Environ. Entomol. 40(6):1345-1352.
Carter, W. 1961. Ecological aspects of plant virus transmissions. Annu. Rev. Entomol. 6:347- 370.
Chapman, R.K. 1971. Prediction of aster leafhopper migrations and the resultant aster yellows problem in Wisconsin. Proc. North Cent. Branch Entomol. Soc. Am. 26:96- 98.
Chapman, R.K. 1973. Integrated control of aster yellows. Proc. North Cent. Branch Entomol. Soc. Am. 28:71-92.
Chiykowski L.N. 1965. The reaction of barley varieties to aster yellows virus. Can. J. Bot. 43:373-378.
Chiykowski L.N. 1967. Reaction of some wheat varieties to aster yellows virus. Can. J. Plant Sci. 47:149-151.
Chiykowski L.N. and R.K. Chapman. 1965. Migration of the six-spotted leafhopper in central North America. Wisconsin Agricultural Experiment Station Research Bulletin 261: 21-45.
Christensen, N.M., Axelsen, K.B., Nicolaisen, M. and A. Schultz. 2005. Phytoplasmas and their interactions with hosts. Trends Plant Sci. 10:526-535.
Drake D.C. and R.K. Chapman. 1965. Evidence for long distance migration of the six- spotted leafhopper in Wisconsin. Wisconsin Agricultural Experiment Station Research Bulletin 261: 1-20.
144
Ferguson, C.A., Carvalho, L., Scott, E.M., Bowman, A.W., and A. Kirika. 2008. Assessing ecological responses to environmental change using statistical models. J Appl. Ecol. 45:193-203.
Foster, R. and B.R. Flood. 2005. Vegetable insect management. Purdue Research Foundation, West Lafayette, IN.
Frost, K.E., Esker, P.D., van Haren, R., Kotolski, L, and R.L. Groves. 2012. Factors influencing Macrosteles quadrilineatus (Hemiptera: Cicadellidae) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields. Environ. Entomol. XXXXX.
Gan, J., Spurlock, F., Hendley, P., and D.P. Weston. 2008. Synthetic Pyrethroids: occurrence and behavior in aquatic environments. American Chemical Society, Washington, D.C.
Gruber, B.R. and M. P. Daugherty. 2012. Understanding the effects of multiple sources of seasonality on the risk of pathogen spread to vineyards: vector pressure, natural infectivity, and host recovery. Plant Pathol. [Online] Doi: 10.1111/j.1365- 3059.2012.02611.x. pp. 11
Hastie T., Tibshirani, R., and J. Friedman. 2009. The Elements of Statistical Learning – Data Mining, Inference and Prediction (2nd edition). Springer-Verlag New York, LLC, New York, New York.
Hodgson, J.A., Thomas, C.D., Oliver, T.H., Anderson, B.J., Brereton, T.M., and E.E. Crones. 2011. Predicting insect phenology across space and time. Global Change Biol. 17:1289-1300.
Hollingsworth, C.R., Atkinson, L.M., Samac, D.A., Larsen, J.E., Motteberg, C.D., Abrahamson, M.D., Glogoza, P., and I.V. MacRae. 2008. Region and field level distributions of aster yellows phytoplasma in small grain crops. Plant Dis. 92:623-630.
Hoy, C.W., Heady, S.E., and T.A. Koch. 1992. Species composition, phenology, and possible origins of leafhoppers (Cicadellidae) in Ohio vegetable crops. J. Econ. Entomol. 85:2336-2343.
Huff, F.A. 1963. Relation between leafhopper influxes and synoptic weather conditions. J. Appl. Meteorol. Clim. 2(1):39-43.
Hurd, W.E. 1920. Influence of the wind on the movements of insects. Mon. Weather Rev. 48(2):94-98. 145
IRPCM Phytoplasma/Spiroplasma working team – Phytoplasma taxonomy group. 2004. 'Candidatus Phytoplasma', a taxon for the wall-less, non-helical prokaryotes that colonize plant phloem and insects. J. Syst. Evol. Microbiol. 54:1243-1255.
Jeger, M.J., Holt, J., van den Bosch, F., and L.V. Madden. 2004. Epidemiology of insect- transmitted plant viruses: modeling disease dynamics and control interventions. Physiol. Entomol. 29:291-304.
Kunkle, L.O. 1926. Studies on aster yellows. Am. J. Bot. 13(10):646-705.
Lee, I.-M., Davis, R.E., and D. E. Gunderson. 2000. Phytoplasma: Phytopathogenic mollicutes. Annu. Rev. Microbiol. 54:221-255.
Lee, I.-M., Gundersen-Rindal, D.E. and A. Bertaccini. 1998. Phytoplasma: ecology and genomic diversity. Phytopathology 88(12):1359-1366.
Lee, I.-M., Martini, M., Bottner, K. D., Dane, R.A., Black, M.C., and N. Troxclair. 2003. Ecological implications from a molecular analysis of phytoplasmas involved in an aster yellows epidemic in various crops in Texas. Phytopathology 93:1368-1377.
Lee, P.E. and A.G Robinson. 1958. Studies on the six-spotted leafhopper Macrosteles fascifrons (Stal) and aster yellows in Manitoba. Can. J. Plant Sci. 38:320-327.
Lindblad, M. and P. Areno. 2002. Temporal and spatial population dynamics of Psammotettix alienus, a vector of wheat dwarf virus. Int. J. Pest Manage. 48:233-238.
Maden, L.V., Jeger, M.J., and F. van den Bosch. 2000. A theoretical assessment of the effects of vector-virus transmission mechanism on plant disease epidemics. Phytopathology 90(6): 576-594.
Magnuson, J.T. 1990. Long-term ecological research and the invisible present. Bioscience 40(7):495-501.
Mahr, S.E.R. 1989. Development and aster yellows-infectivity of the aster leafhopper. Ph.D. Thesis, University of Wisconsin, Madison.
Mahr S.E.R., Wyman J.A., and R.K. Chapman. 1993. Variability in aster yellows infectivity of local populations of the aster leafhopper (Homoptera: Cicadellidae) in Wisconsin. J. Econ. Entomol. 86(5): 1522-1526.
Maramorosch, K. 1952. Direct evidence for the multiplication of aster-yellows virus in its insect vector. Phytopathology 42(2):59-64. 146
Matthews R.E.F. 1991. Plant Virology 3rd edition. Academic Press, Inc. San Diego, CA 92101.
Munyaneza, J.E., Crosslin, J.M., Upton, J.E., and Buchman, J.L. 2010. Incidence of the beet leafhopper-transmitted virescence agent phytoplasma in local populations of the beet leafhopper, Circulifer tenellus, in Washington State. Journal of Insect Science 10: 18.
Nault, B.A., Shah, D.A., Straigh, K.E., Bachmann, A.C., Sackett, W.M., Dillard, H.R., Fleischer, S.J., and F.E.Gildow. 2009. Modeling temporal trends in aphid vector dispersal and cucumber mosaic virus epidemics in snap bean. Environ. Entomol. 38(5):1347-1359.
Pedigo, L.P. 1989. Entomology and Pest Management. Macmillan Pub. Co., New York, NY. 646 pp.
Peterson, A.G. 1973. Host plant and aster leafhopper relationships. Proc. North Cent. Branch Entomol. Soc. Am. 28:66-70.
Pirotta, E., Matthiopoulos, J., MacKenzie, M., Scott-Hayward, L., and L. Rendell. 2011. Modelling sperm whale habitat preference: a novel approach combining transect and follow data. Mar. Ecol. Prog. Ser. 436:257-272.
R Development Core Team (Version 2.15.1). 2012. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/.
Sinha, R.C. and L.N. Chiykowski. 1967. Initial and subsequent sites of aster yellows virus infection in a leafhopper vector. Virology 33:702-708.
Sisterson, M.S. 2009. Transmission of insect-vectored pathogens: effects of vector fitness as a function of infectivity status. Environ. Entomol. 38(2):345-355.
Sugio, A., Kingdom, H.N., MacLean, A.M., Grieve, V.M. and S. A. Hogenhout. 2011. Phytoplasma protein effector SAP11 enhances insect vector reproduction by manipulating plant development and defense hormone biosynthesis. P. Natl. Acad. Sci. USA 108(48):E1254 – 1263.
Wallis, R.L. 1962. Host plant preference of the six-spotted leafhopper. J. Econ. Entomol. 55:998-999.
Werner, I., L.A. Deanovic, D.E. Hinton, J.D. Henderson, G.H. de Oliveira, B.W. Wilson, 147
W. Krueger, W.W. Wallender, M.N. Oliver, and F.G. Zalom. 2002. Toxicity of stormwater runoff after dormant spray application of diazinon and esfenvalerate (Asana (R)) in a French prune orchard, Glenn County, California, USA. Bull. Environ. Contam. Toxicol. 68:29–36.
Westbrook, J.K. and S.A. Isard. 1999. Atmospheric scales of biotic dispersal. Agr. Forest Meteorol. 97:263-274.
Westdal, P.H. and H.P. Richardson. 1969. The susceptibility of cereals and wild oats to an isolate of the aster yellows pathogen. Can. J. Botany 47:755-760.
Weston, D.P., J. You, and M.J. Lydy. 2004. Distribution and toxicity of sediment-associated pesticides in agriculture-dominated water bodies in California's Central Valley. Environ. Sci. Technol. 38:2752–2759.
Wheatley, M. and C. Johnson. 2009. Factors limiting our understanding of ecological scale. Ecol. Complex. 6:150-159.
Wolansky, M.J. and J.A. Harrill. 2008. Neurobehavioral toxicology of pyrethroid insecticides in adult animals: a critical review. Neurotoxicology and Teratology 30:55- 78.
Wood, S.N. 2004. Stable and efficient multiple smoothing parameter estimation for generalized additive models. J. Am. Stat. Assoc.. 99:673-686
Wood S.N. 2006. Generalized Additive Models: An Introduction with R. Chapman and Hall/CRC Press. Boca Raton, FL
Zhu, M., Radcliffe, E.B., Ragsdale, D.W., MacRae, I.V. and M.W. Seeley. 2006. Low- level jet streams associated with spring aphid migration and current season spread of potato viruses in the U.S. northern Great Plains. Agr. Forest Meteorol. 138:192-202
Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev A.A., and G.M. Smith. 2009. Mixed effects models and extensions in ecology with R. Springer Science + Business Media, LLC, New York, New York, 10013.
148
Tables and Figures.
Table 1. Model coefficients (Model 1) estimated by fitting x, y and z to aster leafhopper abundance as a smooth function of calendar day and year as a smooth function of calendar day and year. Parameters Estimate ± SE |z-value| P-value Fixed β2001 0.17 (0.20) 0.85 0.39 β2002 0.24 (0.20) 1.17 0.24 β2003 -0.75 (0.21) 3.49 < 0.001 β2004 0.14 (0.20) 0.72 0.47 β2005 -0.58 (0.22) 2.66 < 0.008 β2006 -0.69 (0.20) 3.35 < 0.001 β2007 -0.52 (0.21) 2.48 0.013 β2008 -1.04 (0.21) 4.97 < 0.001 β2009 -2.28 (0.22) 9.91 < 0.001 β2010 -2.55 (0.21) 11.36 < 0.001 β2011 -1.03 (0.21) 2.32 0.02 Smooth d.f. Chi-sq P-value s(Calendar Day) 6.57 109 < 0.001 Random σfarm 0.38
σfield 0.19
σfyw 0.99
σobs. 0.93 P-values are approximate. R-sq. (adj.) = 0.15 149
Table 2. Model coefficients (Model 2) estimated by fitting x, y and z square root-transformed ALH infectivity as a smooth function of calendar day and year. Parameters Estimate ± SE |t-value| P-value Fixed β1994 0.12 (0.01) 11.7 < 0.001 β1995 0.14 (0.01) 14.4 < 0.001 β1996 0.19 (0.01) 17.8 < 0.001 β1997 0.18 (0.02) 11.1 < 0.001 β1998 0.18 (0.01) 13.2 < 0.001 β1999 0.10 (0.01) 8.9 < 0.001 β2000 0.22 (0.01) 19.5 < 0.001 β2001 0.13 (0.01) 10.1 < 0.001 β2002 0.13 (0.01) 10.1 < 0.001 β2003 0.08 (0.02) 5.5 < 0.001 β2004 0.20 (0.01) 13.7 < 0.001 β2005 0.03 (0.01) 2.8 0.005 β2006 0.09 (0.02) 4.2 < 0.001 β2007 0.10 (0.02) 4.8 < 0.001 β2008 0.25 (0.03) 8.4 < 0.001 Smooth d.f. F P-value s(Calendar Day) 5.41 6.6 <0.001 P-values are approximate. R-sq. (adj.) = 0.36. 150
Table 3. Windows of risk for Aster Yellows phytoplasma spread. Calendar Day Window Start End Length 163 214 Frost et al. Aster 51 11 June 1 Aug (XXXX) Leafhopper 163 214 Abundance 51 This Paper 11 June 1 Aug. 140 197 Frost et al. Aster 57 19 May 15 July (XXXX) Leafhopper 145 196 Infectivity 51 This Paper 24 May 14 July 155 210 AYI 55 This Paper 3 June 28 July
151
Table 4. Number of occasions, in each year, when the observed ALH abundance was greater than the allowable abundance as calculated using an AYI of 25, 50, 75 and 100, given the average seasonal infectivity. AYI AYI AYI AYI # # # Year 25 50 75 100 Events Fields Obs. 2001 78 28 15 23 144 34 438 2002 69 31 16 30 146 33 478 2003 25 6 2 2 35 21 309 2004 52 13 7 14 86 34 548 2005 33 1 2 1 37 26 386 2006 28 15 8 4 55 35 477 2007 28 6 2 2 38 31 415 2008 15 9 2 5 31 34 420 2009 1 0 0 0 1 25 315 2010 3 0 0 0 3 28 403 2011 12 3 1 1 17 35 442 Total 344 112 55 82 593 336 4631
152
Table 5. Number of occasions, on each farm, when the observed ALH abundance was greater than the allowable abundance as calculated using an AYI of 25, 50, 75 and 100, given the average seasonal infectivity. AYI AYI AYI AYI # # Location 25 50 75 100 Events Obs. Farm 1 142 44 14 11 211 2060 Farm 2 18 6 8 11 43 146 Farm 3 120 47 23 53 243 1029 Farm 4 2 1 0 0 3 17 Farm 5 32 11 10 6 59 409 Farm 6 30 3 0 1 34 970 Totals 334 112 55 82 583 4631
153
● ● ● 0 A
●
) ± 95% CI ●
e ● ●
−1 ● ●
−2
●
● Abundance Est. (Log
20012002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Year 2 B
1
0 s(Calendar Day, d.f. 6.57) −1
140 160 180 200 220 240 260 Calendar Day
Figure 1. A) Model coefficients (± 2 x SE) representing the average annual leafhopper abundance (log-scale) for 2001 through 2011. B) Component smooth function representing the trend of aster leafhopper abundance as a smooth function of calendar day plotted with partial residuals from model 1. Partial residuals for a well-fitting model should be scattered evenly around the curve to which they relate. 154
0.30
A ● 0.25
●
0.20 ●
● ● ●
0.15 ●
● ●
●
● 0.10 ●
● ●
0.05 ● Infectivity Est. (Sqrt) ± 95% CI
1994 1996 1998 2000 2002 2004 2006 2008 1995 1997 1999 2001 2003 2005 2007 Year 0.2 B 0.1
0.0 Day, 5.41) Day, ar
–0.1 s(Caland
–0.2
140 160 180 200 220 Calendar Day Figure 2. A) Model coefficients (± 2 x SE) representing the average annual leafhopper infectivity (square root-transformed) for 1994 through 2008. B) Component smooth function representing the trend of aster leafhopper infectivity as a smooth function of calendar day plotted with partial residuals. Expected AYI Expected Infectivity (%) Expected Abundance (100 swps) 100 125 10 10 12 75 50 0 2 4 6 8 25 2 4 6 8 0 0 A C B a June May June May a June May 4 6 8 0 220 200 180 160 140 4 6 8 0 2 240 220 200 180 160 140 4 6 8 0 2 240 220 200 180 160 140 Calendar Day Typical Typical Typical High High High Low Low Low July July July August August August Extrapolated Infectivity infectivity estimates. computed fromabundance and typical, andlowestimated AYI values values inexampleseasonswithhigh, C) plotted withpartialresiduals. a smoothfunctionofcalendarday trend ofasterleafhopperinfectivityas smooth functionrepresentingthe for 1994through2008.B)Component sweeps. y(squareroot-transformed) of abundanceper100pendulum multiplied by4toobtainanestimate are onthedatascaleandhavebeen leafhopper abundance.Predictions representing theaverageseasonal Figure 3. A) Expected Aster Yellows Index(A Model1prediction 155 YI) 156 0.5 ve rages) rre lation 0.0 Cross Co Cross (Seasonal A (Seasonal −0.5
−6 −4 −2 0 2 4 6 Lag (Weeks) Figure 4. Cross-correlation of the weekly abundance and rate of infectious leafhoppers for the average season. 157
0.15
AYI ALL AYI 25 AYI 50 AYI 75 AYI 100 0.10
0.05 Predicted Probability of observing an Event
0.00
120 140 160 180 200 220 240 May June July August Calendar Day Figure 5. The seasonal probability of detecting leafhopper abundances above an AYI of 25, 50 75, or 100 given an average seasonal infectivity. 158
Chapter 5: Feasibility of alternative management strategies for the control of aster
yellows in Wisconsin carrot
159
Abstract: Insect management programs to control aster yellows (AY) in processing and fresh market carrot crops in Wisconsin rely on frequent foliar applications of synthetic pyrethroid insecticides to control the aster leafhopper, the vector of the aster yellows phytoplasma
(AYp). Although successful for managing AY in a cost-effective manner, this approach presents considerable risk to worker safety and the environment because synthetic pyrethroids are broad-spectrum insecticides with documented toxicity to aquatic organisms and mammals.
This research documents the strengths and weaknesses of a proposed refinement to the current
IPM program for control of AY that uses reduced-risk (RR) insecticides. Specifically, aster leafhopper abundance, AY abundance, and crop yield and quality were evaluated for the current and proposed AY management programs. Historical pest scouting data was then used to examine the potential cost structures associated with the two IPM programs. Aster leafhopper abundance and AY symptoms were significantly lower in the proposed RR IPM program when compared to the currently used foliar insecticide program. Yields among the proposed programs were not significantly different. Based on historical data and current
(2012) insecticide prices and associated application costs, the proposed RR program will cost more than the currently used foliar program in 2 of 11 years if a conservative threshold is used to inform the foliar insecticide applications in the current IPM program. The RR systemic insecticide program for ALH management was technologically and operationally feasible and may be economically feasible in high ALH years. The ability to choose the most cost- effective program in any given year will minimize plant protection costs in the long term, but relies on accurate predictions of the annual AY risk to inform grower management decisions.
Keywords: aster yellows phytoplasma, IPM, reduced-risk insecticides 160
In Wisconsin, effective, economical, and efficient long term management of the key pests and pathogens in carrot continues to be a challenge. A pest that that is of particular importance is Macrosteles quadrilineatus, or aster leafhopper (ALH). The ALH is important primarily due to its ability to vector the aster yellows phytoplasma (AYp), which is the causal agent of aster yellows (AY) disease (Chikowski and Chapman 1965, Hoy et al. 1992, Lee et al. 1998, Lee et al. 2000). In the field, plant-to-plant spread of the AYp by the ALH is thought to occur in a persistent and propagative manner (Maramorosch 1952a, 1952b, Sinha and
Chiykowski 1967, Chapman 1973, Lee et al. 2000). Symptoms of AY are highly variable and can lead to direct yield and quality losses (Kunkle 1926, Bloomquist 2000). For root vegetables, processing problems can result from an inability to obtain clean raw product due to adventitious root growth and associated field soil contamination.
In Wisconsin, AY management has focused primarily on control of the insect vector, the aster leafhopper (Chapman 1973, Schultz 1979, Jensen 1982, Granadino 2004). An AY risk index, known as the aster yellows index (AYI), was developed to describe the maximum allowable numbers of infectious leafhoppers and define periods of time when plant protection was most needed (Chapman 1971, Chapman 1973 ). The AYI metric is the product of aster leafhopper infectivity, or percent of infectious aster leafhoppers, and the magnitude of the aster leafhopper population, or the number of aster leafhoppers in 100 (pendulum) sweeps using a standard 38 cm sweep net (Chapman 1971). Currently, efforts are made to refine AYI estimates for a specific date and field (Mahr et al. 1993, Frost et al. 2012b). However, even with the availability of contemporary tools, significant annual and site-specific variation in pathogen detection in the insect vector frequently occurs. In most cases, the relationship 161 between pathogen presence in the vector and the vector’s ability to successfully transmit the pathogen is not known.
The most recent IPM recommendation and cost effective program to control ALH (or
AY) in carrot and many other vegetable crops utilizes the AYI and typically results in frequent foliar applications of synthetic pyrethroid insecticides. Although successful from the perspective of managing insect pests in a cost-effective manner, this approach presents considerable risk, since these insecticides are older, broad-spectrum compounds with documented mammalian toxicity (Wolansky and Harrill 2008). The chemicals in this group are also harmful to aquatic organisms, are lipophilic, and, in aquatic environments, tend to adsorb to organic sediments (Gan et al. 2008). Monitoring surveys in the U.S. have detected the presence of synthetic pyrethroid residues in the sediment of both agricultural and urban dominated waterways (Werner 2002, Weston 2004). These findings have prompted concerns about pyrethroid exposure to non-target areas, especially ecologically sensitive areas such as wetlands, which include the low-land muck soils where the majority of Wisconsin carrot is grown. Thus, it has been our goal to reduce the nearly exclusive reliance on synthetic pyrethroid insecticides for the management of aster yellows in carrot.
In our previous work, we modeled the historic pattern of aster leafhopper abundance and infectivity, which are two important risk factors contributing to the spread of AYp (Frost et al. 2012a, Frost et al. 2012b). We also hypothesized that an improved understanding of the expected AY risk based on historical data, combined with an understanding of the “window” of time during which these events occur would improve aster yellows management. Thus, grower adoption of reduced-risk (RR) insecticides may be improved by targeting insecticide 162 applications to periods during which AY risk is high and reducing the number of applications of the more expensive RR insecticides necessary to control aster yellows, thereby improving the cost-efficiency of the newer insecticides. Our primary goal in the current study is to refine and implement a pest management program based on RR insecticides and an application technology that: 1) minimizes farm worker exposure to high-risk pesticides and newer RR insecticides, 2) reduces environmental risks by utilizing insecticides with a more friendly environmental profile to reduce or eliminate drift and run-off into water resources, and 3) creates incentives for adoption by the grower community by documenting the potential for enhanced profitability.
In general, feasibility studies, such as this, aim to objectively and rationally characterize the advantages and disadvantages of the existing and proposed pest management programs/strategies (Kinston 2004). Associated with this study, we are ultimately interested in the prospects for grower adoption of newer RR insecticides, but also realize that costs associated with pest control often drive management decisions. We have previously discussed the applications of long-term data sets for the examination of variability and long term trends
(Frost et al. 2012a, Frost et al. 2012b). However, long-term observational data sets can also play a key role in assessing the future uncertainty associated with disease outbreaks. For example, constructing tables containing the frequency of past disease outbreak events can help growers better understand the future risks of disease outbreaks and determine the potential costs associated with different pest management strategies.
In this study, we examine the operational, technological, and economic feasibility of a proposed systemic insecticide program for the control of AY in carrot and compare it to the 163 foliar insecticide program currently used in Wisconsin. Specifically, in this study we 1) compare the response of the ALH, AY disease symptoms, crop yield and quality of the carrot crop associated with the proposed ALH management program compared to the conventional foliar insecticide spray programs used in Wisconsin carrot production and 2) use historical pest scouting data to compare the cost structures associated with the different management programs, including the costs associated with the new insecticide products and application technologies.
Materials and Methods
Technological and operational feasibility. Experimental design. Experiments were conducted, on farm, at three commercial field locations in central Wisconsin on sandy loam soils near Hancock, WI. The field sites were selected based on a field history of known pressure from both aster leafhopper and nematode pests, northern root knot nematode
(NRKN; Meloidogyne hapla) and root lesion nematode (RLN; Pratylenchus penetrans).
Experiments were arranged as a randomized stripped block design with 10 experimental replications of the insecticide seed treatments which included: 1) untreated control –fungicide only seed treatment, 2) Cruiser seed treatment (0.1 mg a.i. / seed), and 3) Avicta® seed treatment (0.116 mg a.i. / seed) (Table 1). Total seed amounts for each variety were split into
3 equal parts and seed treatments were applied. At each experimental field site, replicates varied in plot length based on the total length of the field. In 2011, for the slicing carrot variety (cv. Enterprise), the experimental plot was approximately 730 m in length with 10 replicates resulting in plots that were 73 m in length resulting in a plot size of 0.4 hectares.
The variety ‘Enterprise’ was seeded at a rate of 1.5 million seeds per hectare and each plot 164 consisted of 3 banded carrot rows spaced 0.5 m apart on a single 1.8 m bed. For the dicing carrot variety (cv. ‘Canada’), the plot length was 2,390 m with 10 experimental replicates resulting in plots that were 239 m in length for a plot area of approximately 0.013 ha. For the variety ‘Canada’, each plot consisted of 3 paired carrot rows spaced 0.5 m apart on a single
1.8 m bed and was seeded at a rate of 562,500 seeds per hectare.
At the third location, a systemic insecticide was applied in-furrow at planting and experimental treatments included: 1) untreated control –conventionally managed carrot crop
(5 applications of esfenvalerate at 5.8 oz. / acre), 2) thiamothoxam in-furrow with water
(Platinum 75 SG applied at 4.01 oz. / acre), and 2) thiamothoxam in-furrow with liquid fertilizer (Platinum 75 SG applied at 4.01 oz. a.i. / acre).
Plant emergence. To investigate the potential for phytotoxicity and direct stand loss due to phytophagous insect pests (e.g. seed corn maggot, wireworms, etc), stand counts were obtained at approximately three weeks after planting. Emergence was estimated as the total number of seedlings in 6.1m of row and recorded for each plot for two weeks after the initial emergence estimate.
Aster leafhopper abundance. Adult and immature aster leafhopper abundance was estimated weekly from mid-June through mid-August by conducting 50 pendulum sweeps in each plot and counting all life stages collected in each sweep net sample.
Nematode abundance. Twice, in early-July and mid-August, the abundance of NRKN and
RLN were assessed for each plot. Briefly, ten 2.5cm diameter cores of soil were extracted from the top 10 cm of each plot and bulked. Soils samples were stored at 4C until nematodes were extracted and identified. Nematodes were extracted from the soil samples using the 165
Baermann funnel technique (Jenkins 1964). A 48 hour incubation period was used to separate root fragments and a centrifugation sugar floatation process to separate the soil fraction.
Disease and yield assessment. The percentage of symptomatic plants (i.e. expressing leaf yellowing or reddening, witches’ brooming, leaf twisting and etc.) in a 6.1m bed section of each experimental plot was recorded to measure the extent of AY disease present in each plot.
During crop harvest, total yield and crop quality (e.g. percentage of forked, stunted, and hairy- roots) was recorded from each plot and crop defects were expressed as a percentage of the total yield.
Data analysis. All data were analyzed using a one-way analysis of variance (ANOVA) and
Fisher’s protected (P < 0.05) LSD was used for post-hoc comparisons of treatment means.
Analyses were carried out in R using the lm function and the LSD.test function (Package: agricolae).
Economic feasibility. In Wisconsin, the carrot crop is primarily grown for total yield and there is no market value associated with the quality of the carrot roots. Given an equivalent yield among the currently used foliar IPM program and the proposed RR IPM program, we would then expect that a grower’s management decision to be influenced primarily by the costs associated with the different IPM programs. Therefore, we conducted an economic feasibility analysis to compare the costs associated with the currently used foliar program and the proposed RR program. Of the several RR program options, we selected the program that consisted of one application of thiamethoxam applied in-furrow during planting as an example to illustrate our methodology. We then used historical pest scouting data to examine the frequency of pest influx events and to estimate the number of foliar sprays that would be 166 prescribed by the AYI to determine the potential costs associated with different pest management strategies.
Price information. Insecticide prices were obtained in August 2012 by personal communication with several distributers of agricultural chemicals in the vegetable growing region of Wisconsin (Personal communication). The price used to calculate the costs associated with the foliar pyrethroid program and RR systemic insecticide program was $0.43 and $4.50 per ounce, respectively. For the examples in this manuscript, application rates were
5.8 and 4.0 ounces per acre resulting in a total insecticide cost of $2.50 and $18.00 per acre for the pyrethroid and neonicotinoid insecticides, respectively. Application costs included the cost of fuel, labor, and capital depreciation and were estimated to be $4.00 and $7.00 per acre for aerial and ground application, respectively.
Estimating frequency of prescribed sprays. Similar to our previous work, we defined an influx event as an occurrence when the observed number of aster leafhoppers at a field was greater than the allowable number of leafhoppers calculated using predetermined AYI values of 25,
50, 75, and 100 (Frost et al. 2012b). Influx events were again coded as 1 for the occurrence of an influx event or 0 for no influx event. Again, the total number of events in each magnitude category were aggregated (summed) to the level of field which we considered to be the level at which management occurs. Thus, the number of events corresponded directly to the number of sprays that would have been prescribed by the AYI and the aggregated data contains the distribution of fields with X prescribed insecticide sprays for the 11 year duration of the data set. The data were also aggregated by field and year to examine the change in the shape of the distribution of sprays among fields. 167
Fitting empirical probability density functions describing prescribed sprays. The resulting distribution of fields having X spray events consisted of zeros or positive integer values and was discrete. Therefore we choose the Poisson distribution as the mathematical function to represent these data and used the empirical data to estimate model parameters for the Poisson distribution. The Poisson distribution has a single parameter (λ) and has a density function described by:
p(x) = λ^x exp(-λ)/x! for x = 0, 1, 2, … and the mean and variance are λ = E(X) = Var(X). The fitting of the
Poisson distribution was carried out in R using the fitdistr function (Package: MASS).
Estimating P(foliar cost > RR systemic cost). Given the price information, the cost of the RR systemic program was scaled to the number of foliar sprays that could be applied for the cost of the systemic program. For example, given the above prices, a single ground application of a systemic insecticide would cost $25.00 corresponding to the dollar value of approximately
2.6 ground applications of the synthetic pyrethroid (i.e. ($18 + $7.0) / ($2.5 + $7.0) = 2.6).
From the estimated probability density for the fitted Poisson distributions, the probability that the foliar program would cost more than the RR systemic program was calculated.
Conceptually, this is the sum of the discrete probabilities, as described by the Poisson distribution with parameter λ, that a randomly chosen field will have more than 2.6 foliar applications of the pyrethroid insecticide. Probabilities were calculated in R using the ppois function with the estimated model parameters (λ) for the Poisson distribution.
Estimating value of accurate prediction and management selection. The “value” of selecting the most cost effective IPM program was estimated for the parameters (λ) of the Poisson 168 distribution estimated for each year and AYI combination. Value was considered to be the cost difference between the probability density weighted cost for the current foliar IPM program and the truncated density weighted cost given the systemic option. This calculation assumes that there is perfect knowledge of the future disease pressure and, given that knowledge, growers select the most cost effective IPM program for their fields.
Results
Technological and operational feasibility. Plant emergence. There was no difference in plant emergence among the seed treatments for either variety (Table 2). Again, plant emergence was not different on for the dicing variety ‘Canada’, but stand counts for the slicing variety
‘Enterprise’ were significantly higher for the systemic treatment (thiamethoxam, abamectin) when compared to the untreated control.
ALH abundance. The cumulative abundance of the aster leafhopper was significantly lower in the systemic insecticide program when compared to the foliar insecticide program (Table 3).
Additionally, yields were not significantly different among the two programs. The operational/technological feasibility of the systemic insecticide program was not different from the standard foliar program was regarded as a legal application.
Nematode abundance. The NRKN was not observed in soils sampled from any of the three locations. However, RLN was observed in significant numbers that ranged from 0 to 430 among all locations sampled in July 2010 (Table 3). The RLN counts increased on the August
2012 sample date and ranged between 0 and 3300 among all locations sampled.
Disease symptoms. Above ground AY symptoms, including leaf reddening or yellowing and witches’ brooming, were low among all treatments with an average of 2 symptomatic plants 169 in 6.1m of row among all locations. Within locations, differences of symptom abundance occurred among insecticide treatments (Table 4). At two locations (A, C) at least one systemic treatment performed significantly better than the current foliar program (on the slicing variety ‘Enterprise’). At the third location (B) and with the dicing variety ‘Canada’, the systemic seed treatments performed worse (i.e. had more disease) than the current foliar programs.
Below ground symptoms typically attributed to nematode feeding, including root stunting, forking, or the presence of galls, varied among locations. Within locations, differences in percent of roots affected by below ground symptoms occurred among insecticide treatments
(Table 4). At one location (A) (and on the slicing variety ‘Enterprise’), the systemic treatment that included abamectin, with insecticidal and antihelmintic activity, performed better than the current foliar program. Results were opposite at a second location (and on the dicing variety
‘Canada’) where both systemic insecticide seed treatments had a higher percentage of symptomatic roots. There were no differences among treatments at the third location (C).
Yield and percent defects. There was no difference in yield or percent root defects at either of the locations where systemic insecticides were delivered as seed treatments when compared to the foliar pyrethroid program (Table 5). Treatments that had a systemic insecticide delivered in-furrow numerically yielded more than the field average.
Economic feasibility. Fitted parameter estimates Poisson distribution. Parameter estimates
(λ) describing the Poisson distribution that best fit the empirical frequency of the number of prescribed sprays for a random field for 2001-2010 combined at spray thresholds of 25, 50, 75 and 100 ranged from 0.24 to 1.76 (Table 6). However, parameter estimates varied greatly 170 among years and were affected by the abundance of aster leafhoppers observed in the field and the AYI thresholds used to prescribe foliar insecticide applications. For example, λ ranged from 0.04 in 2009 to 4.42 in 2002 which results in a differently shaped distribution of spray events for a randomly chosen field (Fig. 1). However, when an AYI 100 threshold was used, λ only ranged from 0.00 in 2009 to 0.91 in 2002 and the shift in the shape of the distribution describing the number of spray events for a randomly chosen field was less dramatic.
P ($ Foliar > $ Systemic). The number of foliar sprays relates directly to the cost associated with the foliar insecticide program and price information allows for a comparison among the foliar and systemic IPM programs. Using price information, the cost of the proposed systemic
IPM program was mapped onto the distribution of AYI prescribed sprays if given the recommendations for the current foliar IPM program (Fig. 1). In general, we found that the cost of a systemic management program was often greater than the cost of the current foliar program (Table 7). However, similar to λ, the probability that the foliar insecticide program would cost more than the systemic insecticide program varied among years and depended on aster leafhopper abundance and AYI thresholds used to prescribe insecticide applications. For example, in some years (i.e. 2001 and 2002), implementing a foliar program would have cost more than a systemic program on average if the risk averse AYI 25 threshold were used to prescribe sprays. However, in 2009 or 2010, there would have been almost no cost-advantage to using a systemic insecticide on any field.
Cost savings with perfect information and decision making. The theoretical cost of using the current foliar IPM program ranged from 0.38 to 42.03 dollars per acre for an AYI 25 (Table 171
8). For an AYI 100, the theoretical cost was much lower and ranged from 0.00 to 8.64 dollars per acre. Using the fitted parameters for the Poisson distributions, the cost savings, expressed in dollars per acre, was calculated assuming perfect prediction of disease pressure and perfect selection of the most cost-effective management option for all fields. For some years and AYI thresholds, the selection of the most cost-effective management option has the potential to save as much as 18.85 dollars per acre (Table 9). Selecting the most cost-effective program can save as much as 45 percent of insecticide and associated applications costs when compared to the total cost of a using a strictly foliar program.
Discussion
For AY management in Wisconsin, the aster yellows index (AYI) informs growers when it is necessary to implement a pest control practice (e.g. insecticide spray). Currently, growers rely almost exclusively on synthetic pyrethroid insecticides that target not only the aster leafhopper, but impact other beneficial insects that may be present in the crop. One of the primary contributions of this study was to demonstrate the utility of a proposed reduced risk (RR), AY management strategy that is legally, technologically and operationally feasibility. The economic feasibility was also assessed using historical pest scouting data and, when compared to current AY management, the proposed RR IPM program is cost-effective in years when aster leafhopper abundances are high. Additionally, the economic feasibility of this management approach will increase as the price of RR insecticides decreases. The proposed RR IPM program represents a new AY control option to complement the currently used management strategy. 172
Many studies have evaluated the efficacy of different insecticide products, on farm
(Huber et al. 1979, Burkness et al. 2001, Nault et al. 2004, Koch et al. 2005). However, few studies have separated and discussed the concepts of technological, operational, and legal feasibility and often ignore aspects involving the economic feasibility (Kingston 2004).
Technological feasibility refers to the ability of our carrot producers to implement the proposed management program. One advantage to conducting on-farm research, in cooperation with producers, is that we can learn if the equipment, insecticide tools, and/or expertise necessary for the implementation of a programmatic change, exist. Often, the successful implementation of on-farm research provides a qualitative measure of a management program’s technological feasibility=. Operational feasibility refers to how well a proposed management program works by both controlling pests and disease outbreaks and taking advantage of the control opportunities identified in previous work. In this study, we have shown that the proposed RR IPM program controls the aster leafhopper and AY in the carrot crop. Additionally, the carrot yields were not adversely affected by any of the proposed refinements to the current foliar IPM program. In fact, an increase in carrot yield was observed at one location with an in-furrow application of thiamathoxam. In this study, we have largely ignored the question of legal feasibility because the insecticides used all possessed a national 3c registration for use on carrot in the state of Wisconsin at the rates used in this experiment (with the exception of abamectin).
In Wisconsin, carrot is primarily grown to achieve a maximum tonnage and negotiated prices for the crop are not a function of root quality. For this reason, pest management decisions are primarily influenced by the costs associated with plant protection. In this study 173 we also examined the economic feasibility of a programmatic refinement to AY management because we were concerned that the greater costs associated with the RR IPM program which may inhibit widespread grower adoption of the RR programs. By examining the frequency of past disease outbreak events, we derived the potential costs associated with two different pest management options. We found that in some years, producers might reduce costs by selecting the RR systemic program. In the long run, given perfect information and management decisions, the selection of the most cost-effective management option for any given field has the potential to save as much as 18.50 dollars per acre or 45 percent savings when compared to the total cost of a using a strictly foliar program. It is also necessary to point out that our analysis strictly followed the AYI thresholds when examining the past frequency of observed aster leafhopper abundances that would have warranted an insecticide application. Crop producers tend to be risk averse and apply insecticide more frequently than the AYI would prescribe.
The ability to choose the most cost-effective pest management program in a given year can minimize plant protection costs in the long term. However, this value cannot be tapped if tools to predict the annual AY risk are not developed such that growers can accurately plan their management program. Thus, the potential cost savings associated with management options could also be thought of as the “value” of a good predictive model for annual leafhopper pressure. Having an understanding of the long-term value associated with the ability to predict pest and disease pressure provides a way to determine how much capital investment should be allocated to the development of predictive tools.
In general, feasibility studies also aim to objectively and rationally characterize the 174 advantages and disadvantages of the existing and proposed pest management programs/strategies. The use of the RR systemic insecticide program has many beneficial attributes that growers like. For example, for the growers it is risk averse and avoids among and within year risk (Frost et al. 2012b). The neonicotiniods are broad spectrum and are efficatious against both leafhoppers and aphids, the two largest carrot pests in Wisconsin, while affecting fewer non-target insect species by allowing more potential for biological control. These insecticides have flexible delivery – in-furrow, seed treatment, layby – and have long residuals that are rate dependent. Probably the largest disadvantage associated with the use of a RR systemic is that the current cost of a single application of the RR compound is much more expensive than the typical pyrethroid. Additionally, the decision to apply must be made early in the season and making the most cost-effective decision for any given year would hinge upon some knowledge of the pest pressure. Finally, there are emerging ecological and environmental issues associated with widespread usage of the systemic neonicotinoids.
The primary thrust of this work has been to refine the management of AY disease.
However, additional opportunities exist for the management of NRKN and RLN which are major pathogens of vegetables throughout the United States. Rotating the carrot crop with a non-host crop such as RKN-resistant soybean, sorghum sudangrass, or several other grain crops, if economically possible, can be effective in reducing nematode damage (Reynolds,
L.B. et al. 2000, Atkins et al. 2003, Kratochvil et al. 2004). However, the crop rotations used on many commercial farms are of limited value because most crops grown, such as potatoes, snap beans, and onion, are susceptible to the nematode allowing populations to survive or 175 increase in size. As a result, nematode control is most often achieved with the use of pre-plant nonfumigant-types of nematicides, which is the primary pesticide tool registered for use in
Wisconsin. Oxamyl is a carbamate used to control insects, mites, and nematodes often considered the industry standard for use, which we would like to eliminate and reduce exposures to humans and the environment. We evaluated the potential for a new, reduced risk, and less broad spectrum seed treatment registration (i.e. abamectin) targeting the NRKN and
RLN, which increasingly provide pest management alternatives for long term control of nematode pests and disease symptoms resulting from feeding and serve as safer and more environmentally compatible options for the replacement of carbamates.
Our primary goal in this study was to refine and implement a pest management program based on RR insecticides and an application technology to minimize farm worker exposure to insecticides, to reduce environmental risks and to create incentives for adoption by the grower community through the documentation of enhanced profitability. This approach also allows for more comprehensive, long-term management strategies that can be tailored to meet the diverse needs of processing carrot producers.
176
References
Atkins, S.D., Hidalgo-Diaz, L., Kalisz, H., Mauchline, T.H., Hirsch, P.R. and B.R. Kerry. 2003. Development of a new management strategy for the control of root-knot nematodes (Meliodogyne spp) in organic vegetable production. Pest Manag.Sci. 59(2):183-189.
Bloomquist, C.L. 2002. Aster yellows and beet leafhopper transmitted virescence agent yellows. Pages 58-59 in: Compendium of Umbelliferous Crop Diseases.
Burkness, E.C., Hutchison, W.D., Bolin, P.C., Bartels, D.W., Warnock, D.F., and D.W. Davis. 2001. Field efficacy of sweet corn hybrids expressing a Bacillus thruingiensis toxin for management of Ostrinia nubilalis (Lepidoptera: Crambidae) and Helicoverpa zea (Lepidoptera: Noctuidae). J. Econ. Entomol. 94(1):197-203.
Chapman, R.K. 1971. Prediction of aster leafhopper migrations and the resultant aster yellows problem in Wisconsin. Proc. North Cent. Branch Entomol. Soc. Am. 26:96- 98.
Chapman, R.K. 1973. Integrated control of aster yellows. Proc. North Cent. Branch Entomol. Soc. Am. 28:71-92.
Chiykowski L.N. and R.K. Chapman. 1965. Migration of the six-spotted leafhopper in central North America. Wisc. Coll. ALS Res. B. 261: 21-45.
Drake D.C. and R.K. Chapman. 1965. Evidence for long distance migration of the six- spotted leafhopper in Wisconsin. Wisc. Coll. ALS Res. B. 261: 1-20.
Felipe de Mendiburu. 2012. agricolae: Statistical Procedures for Agricultural Research. R package version 1.1-1. http://CRAN.R-project.org/package=agricolae.
Frost, K.E., Esker, P.D., van Haren, R., Kotolski, L, and R.L. Groves. 2012a. Factors influencing Macrosteles quadrilineatus (Hemiptera: Cicadellidae) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields. Environ. Entomol. In review.
Frost, K. E., Esker, P.D., van Haren, R., Kotolski, L., and Russell L. Groves. 2012b. Seasonal pattern of aster leafhopper (Hemiptera: Cicadellidae) abundance and aster yellows phytoplasma infectivity in Wisconsin carrot fields. Environ. Entomol. In review.
Gan, J., Spurlock, F., Hendley, P., and D.P. Weston. 2008. Synthetic Pyrethroids: occurrence and behavior in aquatic environments. American Chemical Society, 177
Washington, D.C.
Granadino L. A. 2004. Managing aster leafhoppers with integrated approaches in Wisconsin carrot production. M.S. Thesis, University of Wisconsin, Madison.
Hoy, C.W., Heady, S.E., and T.A. Koch. 1992. Species composition, phenology, and possible origins of leafhoppers (Cicadellidae) in Ohio vegetable crops. J. Econ. Entomol. 85:2336-2343.
Huber, R.T, Moore, L., and M.P. Hoffmann. 1979. Feasibility study of area-wide pheromone trapping of male pink bollworm moths in a cotton insect pest management program. J. Econ. Entomol. 72:222-227.
Jensen, J.O. 1982. Management of the aster leafhopper and aster yellows in Wisconsin. Ph.D. Thesis, University of Wisconsin, Madison
Kingston, J. 2004. Conducting feasibility studies for knowledge based systems. Knowl- Based Sys. 17: 157-164.
Koch, R.L., Burkness, E.C., Hutchison, W.D., and T.L. Rabaey. 2005. Efficacy of systemic insecticide seed treatments for protection of early-growth-stage snap beans from bean leaf beetle (Coleoptera: Chrysomelidae) foliar feeding. Crop Prot. 24:734-742.
Kratochvil, R.J., Sardanelli, S., Everts, K., and E. Gallagher. 2004. Evaluation of crop rotation and other cultural practices for management of root-knot and lesion nematodes. Agron. J. 96:1419-1428.
Kunkle, L.O. 1926. Studies on aster yellows. Am. J. Bot. 13(10):646-705.
Lee, I.-M., Davis, R.E., and D. E. Gunderson. 2000. Phytoplasma: Phytopathogenic mollicutes. Annu. Rev. Microbiol. 54:221-255.
Mahr S.E.R., Wyman J.A., and R.K. Chapman. 1993. Variability in aster yellows infectivity of local populations of the aster leafhopper (Homoptera: Cicadellidae) in Wisconsin. J. Econ. Entomol. 86(5): 1522-1526.
Maramorosch, K. 1952a. Direct evidence for the multiplication of aster-yellows virus in its insect vector. Phytopathology 42(2):59-64.
Maramorosch, K. 1952b. Multiplication of aster-yellows virus in its vector. Nature 168:194- 195.
Nault, B.A., Taylor, A.G., Urwiler, M., Rabaey, T., and W.D. Hutchison. 2004. 178
Neonicotinoid seed treatments for managing potato leafhopper infestations in snap bean. Crop Prot. 23:147-154.
R Development Core Team (Version 2.15.1). 2012. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/.
Reynolds, L.B., Potter, J.W., and B.R. Ball-Coelho. 2000. Crop rotations with Tagetes sp. is an alternative to chemical fumigation for control of root-lesion nematodes. Agron. J. 92:957-966.
Schultz, G.A. 1979. Epidemiology of the aster leafhopper and aster yellows in relation to disease control. Ph.D. Thesis, University of Wisconsin, Madison.
Sinha, R.C. and L.N. Chiykowski. 1967. Initial and subsequent sites of aster yellows virus infection in a leafhopper vector. Virology 33:702-708.
Venables, W.N. and B. D. Ripley. 2002. Modern applied statistics with S (4th edition). Springer-Verlag New York, Inc. New York, New York 10010.
Werner, I., L.A. Deanovic, D.E. Hinton, J.D. Henderson, G.H. de Oliveira, B.W. Wilson, W. Krueger, W.W. Wallender, M.N. Oliver, and F.G. Zalom. 2002. Toxicity of stormwater runoff after dormant spray application of diazinon and esfenvalerate (Asana (R)) in a French prune orchard, Glenn County, California, USA. Bull. Environ. Contam. Toxicol. 68:29–36.
Weston, D.P., J. You, and M.J. Lydy. 2004. Distribution and toxicity of sediment-associated pesticides in agriculture-dominated water bodies in California's Central Valley. Environ. Sci. Technol. 38:2752–2759.
Wolansky, M.J. and J.A. Harrill. 2008. Neurobehavioral toxicology of pyrethroid insecticides in adult animals: a critical review. Neurotoxicol Teratol 30:55-78. 179
Tables and Figures.
Table 1. List of the experimental treatments including the field location, insecticides, application rates and technology used. Active Target Location Product Ingredient Delivery Rate Organism Cultivar A UTC - - - - ‘Enterprise’ A Cruiser 5FS Thiamethoxam Seed Trt 0.1 mg a.i. / seed ALH ‘Enterprise’ A Avicta Thiamethoxam, Seed Trt 0.1 mg a.i. / seed ALH ‘Enterprise’ Abamectin 0.016 mg a.i. / seed Nematode B UTC - - - - ‘Canada’ B Cruiser 5FS Thiamethoxam Seed Trt 0.1 mg a.i. / seed ALH ‘Canada’ B Avicta Thiamethoxam, Seed Trt 0.1 mg a.i. / seed ALH ‘Canada’ Abamectin 0.016 mg a.i. / seed Nematode C Asana XL (Std) Esfenvalerate Foliar 5 appl. @ 8.0 fl oz / acre ALH ‘Canada’ C Platinum 75SG Thiamethoxam In-furrow 4.01 fl oz / acre ALH ‘Canada’ C Platinum 75SG, Thiamethoxam In-furrow 4.01 fl oz / acre ALH ‘Canada’ liquid fertilizer
180
Table 2. Seedling emergence (±SD), expressed as number of plants in 6.1m of row, for experimental treatments at two locations. Active Location Ingredient Stand Count 1 Stand Count 2 A Esfenvalerate 59.0 (7.7) 38.6 (5.5) a A Thiamethoxam 50.7 (7.3) 44.3 (5.3) ab
Thiamethoxam b A 53.8 (10.1) 46.8 (6.4) Abamectin B Esfenvalerate 37.9 (7.5) 34.3 (6.9) B Thiamethoxam 32.9 (8.5) 32.0 (6.1)
Thiamethoxam B 30.8 (7.5) 31.0 (7.8) Abamectin * Treatment comparisons were made within a location ** Overall F-test was not significant (P < 0.05) for comparisons not followed by letters 181
Table 3. Cumulative aster leafhopper counts (±SD) and root lesion nematode population sizes (±SD) for experimental treatments at three locations in 2011. Active Cumulative RLN RLN Location Ingredient ALH July August A Esfenvalerate 9.9 (4.0) a 105.6 (90.2) 661.1 (510.0) a A Thiamethoxam 8.0 (3.4) ab 106.2 (88.3) 1379.4 (734.5) ab
Thiamethoxam b b A 4.9 (2.4) 90.6 (66.0) 1020.8 (508.5) Abamectin B Esfenvalerate 3.5 (2.1) 13.8 (20.4) a 122.1 (113.5) B Thiamethoxam 4.3 (1.5) 46.2 (43.5) a 379.5 (422.2)
Thiamethoxam a B 3.6 (2.0) 13.2 (18.1) 480.7 (494.3) Abamectin C Esfenvalerate - - - Thiamethoxam C 2.5 (1.3) 55.8 (115.6) - + H2O Thiamethoxam C 2.8 (2.4) 10.4 (16.1) - + Liq * Treatment comparisons were made within a location ** Overall F-test was not significant (P < 0.05) for comparisons not followed by letters – Indicates data were not collected 182
Table 4. Above ground aster yellows symptoms (±SD), expressed as number of symptomatic plants in 6.1m of row, and below ground carrot defects (±SD), such as stunting, forking, or galling, expressed as % of total roots examined, for experimental treatments at all locations. Active # Aster Yellows % RLN Location Ingredient Symptoms Symptoms A Esfenvalerate 2.9 (0.3) a 20.3 (5.1) ab A Thiamethoxam 1.7 (0.4) ab 31.1 (4.0) a
Thiamethoxam b b A 1.0 (0.3) 12.2 (5.5) Abamectin
B Esfenvalerate 0.1 (0.1) a 7.5 (2.4) a B Thiamethoxam 1.1 (0.3) b 23.8 (3.8) b
Thiamethoxam ab b B 1.0 (0.3) 31.4 (5.8) Abamectin C Esfenvalerate 1.4 (0.4) ab -
Thiamethoxam a C 1.6 (0.3) 13.7 (6.0) + H2O
Thiamethoxam b C 0.4 (0.2) 1.7 (1.1) + Liq * Treatment comparisons were made within a location ** Overall F-test was not significant (P < 0.05) for comparisons not followed by letters – Indicates data were not collected 183
Table 5. Total yield (±SD) and percent root defects (±SD), expressed as the percent of the total yield, for experimental treatments at all locations. Active Yield Percent Location Ingredient (Tons/Acre) Defect A Esfenvalerate 25.6 (3.1) 18.6 (8.1) A Thiamethoxam 26.6 (4.0) 15.8 (6.3) Thiamethoxam A 26.1 (46.0) 14.6 (6.2) Abamectin B Esfenvalerate 19.5 (4.7) 47.4 (13.3) B Thiamethoxam 19.4 (4.8) 34.8 (10.1) Thiamethoxam B 21.1 (4.6) 43.9 (12.7) Abamectin C Esfenvalerate 31.1 - Thiamethoxam C 32.8 (3.3) * 6.2 (3.9) + H2O Thiamethoxam C 34.1 (3.3) ** 3.7 (4.0) + Liq * Treatment comparisons were made within a location ** Overall F-test was not significant (P < 0.05) for comparisons not followed by letters – Indicates data were not collected γ Overall field yield Significantly different than overall field yield (Esfenvalerate) at ϕ (P < 0.05) and ϕϕ (P < 0.01) by one sample t-test 0.01).
184
Table 6. Parameter estimates of the poisson distributions used to describe the number of prescribed sprays for a random field if AYI spray thresholds of 25, 50, 75 and 100 were used. Rate of events for a randomly chosen field in a year (λa) Year AYI 25 AYI 50 AYI 75 AYI 100 2001 4.24 (0.35) 1.94 (0.24) 1.12 (0.18) 0.68 (0.14) 2002 4.42 (0.37) 2.33 (0.27) 1.39 (0.21) 0.91 (0.17) 2003 1.67 (0.28) 0.48 (0.15) 0.19 (0.10) 0.10 (0.07) 2004 2.53 (0.27) 1.00 (0.17) 0.62 (0.13) 0.04 (0.11) 2005 1.42 (0.23) 0.15 (0.08) 0.12 (0.07) 0.04 (0.04) 2006 1.57 (0.21) 0.77 (0.15) 0.34 (0.10) 0.11 (0.06) 2007 1.23 (0.20) 0.32 (0.10) 0.13 (0.06) 0.06 (0.05) 2008 0.91 (0.16) 0.47 (0.12) 0.21 (0.08) 0.15 (0.07) 2009 0.04 (0.04) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 2010 0.11 (0.06) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 2011 0.49 (0.12) 0.14 (0.06) 0.06 (0.04) 0.03 (0.03) Total 1.76 (0.07) 0.74 (0.05) 0.41 (0.03) 0.24 (0.03) a λ represents the fitted rate (± SE) 185
Table 7. The probability that the current foliar insecticide management program would have cost more than a systemic insecticide management program (2001 – 2011). P(Foliar > Systemic) Year AYI 25 AYI 50 AYI 75 AYI 100 2001 0.79 0.31 0.10 0.03 2002 0.82 0.41 0.17 0.06 2003 0.23 0.01 0.00 0.00 2004 0.46 0.08 0.02 0.01 2005 0.17 0.00 0.00 0.00 2006 0.21 0.04 0.01 0.00 2007 0.13 0.00 0.00 0.00 2008 0.06 0.01 0.00 0.00 2009 0.00 0.00 0.00 0.00 2010 0.00 0.00 0.00 0.00 2011 0.01 0.00 0.00 0.00 Total 0.260 0.039 0.008 0.002
186
Table 8. Theoretical cost of using current foliar IPM program (dollars per acre) given strict adherence to the AYI prescription. AYI 25 AYI 50 AYI 75 AYI 100 2001 40.23 18.44 10.62 6.43 2002 42.03 22.17 13.24 8.64 2003 15.83 4.52 1.81 0.90 2004 24.03 9.50 5.86 3.91 2005 13.52 1.46 1.10 0.37 2006 14.93 7.33 3.26 1.09 2007 11.65 3.06 1.23 0.61 2008 8.66 4.47 1.96 1.39 2009 0.38 0.00 0.00 0.00 2010 1.02 0.00 0.00 0.00 2011 4.61 1.35 0.54 0.27
187
Table 9. Theoretical cost savings, expressed in dollars per acre and as a percent of current foliar program costs, if prediction were perfect and selection of management options were perfect between current and proposed IPM programs. AYI 25 AYI 50 AYI 75 AYI 100 2001 17.32 (43) 2.97 (16) 0.69 (6) 0.17 (3) 2002 18.85 (45) 4.68 (21) 1.25 (9) 0.38 (4) 2003 2.01 (13) 0.06 (1) 0.00 (0) 0.00 (0) 2004 5.68 (24) 0.50 (5) 0.13 (2) 0.04 (1) 2005 1.32 (10) 0.00 (0) 0.00 (0) 0.00 (0) 2006 1.72 (12) 0.02 (3) 0.02 (1) 0.00 (0) 2007 0.88 (8) 0.02 (1) 0.00 (0) 0.00 (0) 2008 0.38 (4) 0.06 (1) 0.00 (0) 0.00 (0) 2009 0.00 (0) 0.00 (0) 0.00 (0) 0.00 (0) 2010 0.00 (0) 0.00 (0) 0.00 (0) 0.00 (0) 2011 0.06 (1) 0.00 (0) 0.00 (0) 0.00 (0) Total 48.22 8.31 2.09 1.18
188
1.0
Lambda 0.04 0.8 0.91 2.6 Sprays 4.42
0.6
Probability P ($ Foliar > $ Systemic) 0.4
0.2
0.0
0 5 10 15 Number of Sprays Figure 1. Distribution of fields with X prescribed sprays as described by the Poisson distribution with lambda of 0.04, 0.91 and 4.42 to represent the range of parameter estimates obtained over the 11 year data set. The price of the proposed RR systemic insecticide program was scaled to reflect the cost-equivalent number of foliar applications of a pyrethroid insecticide. The probability that the foliar program would cost more than the systemic program is equal to the area under the (probability density) to the right of 2.6. 189
1.0
A 0.8
D 0.6
C 0.4 Probability(systemic > foliar)
0.2 B
0.0 0 1 2 3 4 5 6 Lambda Figure 2. The probability that the systemic program will cost more than the foliar program for varying lambda values, to represent differing insect pressure, and different RR systemic program options or prices. Program options include: A) 1 aplication of systemic insecticide at 2003 prices. B) 1 application of systemic insecticide at 2012 prices. C) 1 application of systemic insecticide at 2012 prices + 1 application of foliar pyrethroid at 2012 prices. D) 1 application of systemic insecticide at 2012 prices + 2 application of foliar pyrethroid at 2012 prices. 190
Chapter 6: Concluding remarks and future directions
191
In Wisconsin, aster yellows management has focused primarily on controlling the insect vector, the aster leafhopper (ALH), and an AY risk index, known as the aster yellows index (AYI), was developed to describe the maximum allowable numbers of infectious leafhoppers during a discrete time period when plant protection is most needed. The AYI metric is the product of aster leafhopper infectivity, or percent of infectious aster leafhoppers, and the magnitude of the aster leafhopper population. Originally, the AYI was used to make insecticide spray recommendations based on a series of early season leafhopper collections, but following the observations that aster leafhopper abundance and infectivity in and around carrot fields varies, efforts were made to estimate the AYI for specific fields and dates.
Contemporary tools (i.e. PCR) are currently used to detect the pathogen in the ALH.
However, the relationship between pathogen presence in the vector and the vector’s ability to successfully transmit the pathogen is not known. In turn, many producers avoid risk of pathogen spread by using inexpensive, prophylactic insecticide applications, a management practice that circumvents the utility of the AYI. Although successful from the perspective of managing insect pests in a cost-effective manner, this approach presents considerable risk, since these insecticides are broad-spectrum compounds with documented mammalian toxicity. The chemicals in this group are also harmful to aquatic organisms, are lipophilic, and in aquatic environments, tend to adsorb to organic sediments. These circumstances have prompted concerns about pyrethroid exposure to non-target areas, especially ecologically sensitive areas such as wetlands, which include the low-land, organic muck soils where the majority of Wisconsin carrot is grown. Thus, it has been our goal to reduce the nearly exclusive reliance on synthetic pyrethroid insecticides for the management of aster yellows in 192 carrot. Specifically, we sought to improve grower adoption of reduced-risk (RR) insecticides by targeting insecticide applications to periods during which AY risk is high thereby reducing the number of applications of the more expensive RR insecticides necessary to control aster yellows improving the cost-efficiency of these newer tools. However, reliable information about ALH abundance and infectivity for specific fields and the coincidence of these expected periods of high insect population sizes and high rates of infectivity were not known.
The primary goal for completing the outlined objectives articulated in this thesis was to advance our understanding of the epidemiology of aster yellows in Wisconsin towards the development and implementation of a comprehensive disease and pest management plan for carrot. Our approach essentially tried to address the two problems described above by (Prong
I) developing the molecular tools to accurately determine the rate of infectious individuals in a field population of ALH and (Prong II) analyzing historical ALH scouting data to identify seasonal trends in factors associated with periods of elevated AY risk.
Molecular Prong: To examine the relationship between pathogen presence and titer in the vector and the vector’s ability to successfully transmit the pathogen, I developed a qPCR assay to quantify the titer of the aster yellows phytoplasma (AYp) in its insect vector,
Macrosteles quadrilineatus (Chapter 2, Frost et al 2011). I also characterized AYp’s growth pattern and titer within the leafhopper for approximately 9 days after acquisition. One of the primary contributions of this study was to demonstrate qPCR as a reliable and accurate method for measuring AYp titer in the aster leafhopper and detecting differences of AYp titer among groups of insects. Additionally, chapter 2 offers a detailed analysis of the capabilities 193 and limitations of the developed qPCR assay and discussion topics to include: 1) Examining the variation of qPCR calibration curves, 2) defining limits of detection for a qPCR assay, and
3) modeling growth of AYp in its insect vector.
Empirical Prong: To identify patterns of variability and describe the seasonal trends in leafhopper abundance and infectivity associated with a higher risk for AY spread in carrot,
I first needed to apply contemporary analytical methods to examine factors affecting insect abundance and infectivity. Specifically, chapters three and four describe a statistical analysis approach that is applied to a multi-year, multi-location observational pest scouting data set to obtain information about the scale at which ecological factors contribute to the variability of leafhopper abundance and infectivity. From these data and analysis, I was able to deduce seasonal “windows” of elevated risk for spread of the aster yellows phytoplasma to susceptible crops. One of the primary contributions of this study was to demonstrate the utility of long-term data sets for improving our understanding of the spatial and temporal patterns of variation of insect abundance and infectivity. However, these chapters also highlight and discuss the use of some contemporary statistical methodologies including generalized linear mixed models and generalized additive mixed models for the analysis of long-term ecological data. More practically speaking, I also identified a timing interval, or ‘risk window’, that results from the coincidence of above average leafhopper abundance and elevated levels of mean infectivity. Taken together, the coincidence of the expected periods of high leafhopper abundance and infectivity represent a potential ‘treatment window’ in which management of the insect could be focused if no current information about a specific field is available.
Future directions 194
Molecular prong (Chpt. 2): Needless to say, there are dangling threads associated with this research and experiments are needed before this method could be fully implemented to estimate the rate of infectious individuals in a population. The existence of AYp titer variability among ALHs and having the tools to manipulate and measure that variability are necessary factors for completion of a set of replicated, dose-response experiments needed to relate AYp titer to the leafhopper’s ability to successfully transmit. Figure 1 outlines two future studies that could be completed to relate transmission success to phytoplasma titer in individual insects. The flexibility associated with these methodologies will also allow the determination of ALH transmission efficiency as a function of environmental variation, AYp- genetic variation, or leafhopper genetic variation and would further allow for the description of AYp titer variation in field caught ALH populations. The improvements to AYp detection will be directly applicable to the AYp management by providing more accurate estimates of the “true” rate of infectious ALHs and, subsequently, more accurate calculations of the
Aster Yellows Index. Thus, the AYI could be more accurately and quickly implemented on a field by field basis and used as originally intended with far greater accuracy.
Empirical prong (Chapt 3 & 4): In large part, chapter 5 was an initial attempt to validate or refute whether or not the timing interval, identified in chapters 3 & 4, could be targeted to control the aster leafhopper and spread of AYp. Since the timing interval identified in chapters three and four occurs early in the season, there existed the opportunity to deploy at plant newer, reduced-risk, systemic insecticides with flexible application methods, such as seed treatments and in-furrow or layby incorporation, as potential pest alternatives for control of the aster leafhopper. In 2011, the primary outcome of our field work was to implement a 195 systemic insecticide program for the control of the ALH and aster yellows which was (legally) technologically and operationally feasible. The abundance of the ALH and aster yellows symptoms was significantly lower in these systemic insecticide programs when compared to the foliar insecticide program based primarily on the use of synthetic pyrethroids
(esfenvalerate: Asana XL). Additionally, although yields were not significantly different among the two programs, we did observe substantial improvements in total yield (e.g. mean =
11% increase) documented both the operational/technological feasibility of the systemic insecticide program
Actuarial tools: Long-term observational data sets combined with mathematical and statistical analysis tools can play a key role in assessing the future uncertainty associated with disease outbreaks. For example, constructing tables containing the frequency of past disease outbreak events can help growers better understand the future risks of disease outbreaks and determine potential costs associated with different pest management strategies (Table 1). This information will allow pest managers to examine the price structures associated with the management of future pest risks, including the costs associated with new insecticide products and application technologies, and plan for those risks and associated costs. The information can also be used to examine the value of real management options. For example, having the ability to choose the most cost-effective program in a given year will minimize plant protection costs in the long term and further increase the overall sustainability of the IPM program. Of course, tools to predict the annual AY risk will need further development so that growers are accurately planning their management programs based upon the “value” of a good predictive model for annual leafhopper and disease pressure. Associated values could be 196 estimated as the cost savings associated with the option to implement a management plan which is best suited for the specific year. Prediction of pest and disease outbreaks, or disease potential, can easily be regarded as a ‘holy grail’, and is among the most difficult goals to achieve for nearly all cropping systems – resulting in many contemporary vector-borne disease forecasting tools that possess moderate to low predictive value.
Predictive tools: By advancing predictive tools to address the sporadic occurrence of ‘risk intervals’, we can minimize costs associated with unwarranted pesticide applications and reduce yield losses due to advanced preparation for ALH infestations. An existing ALH development model can begin to identify geographically defined regions in the Midwest U.S., where winged ALH may be present. Predicting ALH dispersal and deposition is also possible by simulating air parcel trajectories using the HYSPLIT (HYbrid Single-Particle Lagrangian
Integrated Trajectory) model. Specifically, a fixed number of initial “insects” can be transported in the model simulation by the average air movement (i.e. wind; observed or forecasted) and its turbulent component. Thus, the simulated transport and deposition of the
ALH “particles” by the HYSPLIT model can provide an estimate of the likelihood of an insect influx event occurring in Wisconsin susceptible crops in a specific year. For example, we provide an example of a day-degree development model for the month of April (Fig. 2) and a candidate air parcel trajectory for an associated time interval (Fig. 3).
Project impact and outcomes: The purpose of this work has been to increase the sustainability of carrot crop production in Wisconsin by improving production practices.
Sustainability goals included reduction of foliar-applied pesticides to decrease cost, environmental impact, and exposure to farm workers. It also sought to enhance the economic 197 return experienced by growers by lowering input costs while maintaining the yield and quality of the carrot crop.
198
Tables and Figures.
Table 1. Number of influx events, in each year, when the observed ALH abundance was greater than the allowable abundance as calculated using an AYI of 25 provided an average seasonal infectivity estimate. The number of events was divided by 0.6 to estimate the actual number of sprays applied by growers. Current (2012) insecticide prices are used for the systemic or contact products that are available. Application costs were estimated to be seven and four dollars per acre for ground (G) and aerial (A) application, respectively.
Contact Systemic Programs Events 1 App + 1 App + Year per Field Ground Air 1 App 1 Clean 2 Clean 2001 4.24 67.08 45.88 2002 4.42 69.93 47.83 2003 1.67 26.42 18.07 2004 2.53 40.03 27.83 2005 1.42 22.46 15.36 34.49 G 43.98 G 2006 1.57 24.84 16.99 25.00 G 2007 1.23 19.46 13.31 31.49 A 37.98 A 2008 0.91 14.40 9.85 2009 0.04 0.63 0.43 2010 0.11 1.74 1.19 2011 0.49 7.75 5.30 34.49 G 43.98 G Ave 1.76 26.79 18.33 25.00 31.49 A 37.98 A 379.41 G 483.83 G Total 294.73 201.58 275.00 346.41 A 417.83 A Total cash outlay (11 yr) if you have 206.31 G 226.71 G 245.70 G the ability to select the “best” program 148.83 A 157.70 A 164.19 A ($ per Acre per 11 yrs) Long term (11 yr) “value” of accurately predicting plant protection needs and 88.42 G 68.02 G 49.03 G selecting best program ($ per Acre per 52.75 A 43.88 A 37.39 A 11 yrs)
199
● ● ● ●● ● ● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ●●● ●●● ● ● ●●●●●●●● ●● ● Success 1.0 ●●● ● ● ●●●●● ● ●●● ● ● sh old Thre ical Crit Transmission
● ● ●● ● ● 0.0 ● ● ●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ●●●●●●● ●●●●●●●●●●● ● Fail ● ●●●● ●●●● ●●●●●●● ● ●● ●
−2 0 2 4 6 8 AYp Titer per Copy ALH cp6 Figure 1. A depiction of the date needed to determine the dose-response relationship between insect phytoplasma titer and insect infectivity a population of leafhoppers with varying phytoplasma titer will be created. These data could be collected either experimentally or observationally. Experimentally, it would be necessary to create a population of ALHs that varied in their AYp titer and their ability to successfully transmit couple be assessed using a bioassay. Observationally, ALHs could be collected from the field and a bioassay used to determine their ability to successfully transmit. In both proposed studies, the AYp titer in the ALHs would need to be determined following the bioassay so that tranmission success could be related to AYp titer. The disease incidence data obtained from these studies could be expressed as: Incidence = P(disease) = Σ P( N = n) * P(I | N), The Σ P( N = n) * P(I | N) term describes the dose-response relationship between the AYp-titer and the ability of a leafhopper to infect a plant. Here, P(N = n) describes the distribution of AYp-titer in the infected insects and P(I | N) describes the dose response relationship between titer and the ability of an insect to infect a plant which is summed over all values of N=n. In this analysis, the distribution of AYp-titer in infected individuals will be assumed to be lognormally distributed and described using 2 parameters, the mean (μ) and the variance (σ2). The dose-response relationship will also be described using 2 parameter probit model (λ, τ2). 2 2 Pdisease (μ, σ, λ, τ) = Φ(μ – λ / √(σ + τ )) = Σ P( N = n) * P(I | N) where PII is the probability that an infected leafhopper is infective. Nonlinear regression will then be used to fit copy number to disease, yielding estimates (λ, τ2) of the dose- response relationship in the greenhouse. 200
100 100 April 5, 2012 100 100
100
100
100
100
100 100
Egg 100 200 1st instar 2nd instar
300 3rd instar 4th instar 100 Latitude
400 5th instar Adult
500
700
900 700
800 800 900 1100
800 25 30 35 40 45 50 −110 −105 −100 −95 −90 −85 −80 Longitude Figure 2. Temperature data coupled with a thermal time development model, will be used to predict the predominant life stage of ALH present in the Midwestern United States. The basic approach is to I) collect temperature data from the National Oceanic and Atmospheric Administrations (NOAA) website, II) calculate day-degree (dd) accumulations for the ALH, III) model distance-based relationships of dd accumulations and IV) use simple kriging to predict ALH developmental stage for gridded points in the Midwest. This work was initiated for the purposes of studying the biogeography of the AY disease system and, more generally, large-scale (>250 km) insect movement in the agricultural landscape. 201
50
▲ ● ▲ 45 ● ■ ■● -75 ■ ■ -80 -105 -85 -100 -95 ▲ ▲ ■ -90 ▲▲ ▲ ● ●●▲ ▲ ● ■■ ●● ●■ ● ■▲■▲■ Latitude ▲● ▲ ■ 40 ▲■ ▲ ●■●▲■▲●▲ ●● ●■▲▲■●▲ ●▲■■ ▲■● ● ■■ ▲■ ● ●▲ ▲ ■ ▲ ■●▲▲ ■● ■● ▲ ●▲■●▲ ● ■●▲■● ■●▲■●▲■ ▲■●▲■●▲ 35
Longitude Figure 3. An example of a fixed number of initial “insects” transported by the HYSPLIT model simulation by bulk air flow from regions in the Midwest U.S. where winged ALH may be present. A location identified by the day-degree model (Fig. 2) was used as an ALH source region and dispersion of the insects was simulated from that area in Northern Arkansas. For this date, the simulated transport and deposition of ALH “particles” (i.e. the colored lines) by the HYSPLIT model suggest there is the potential for an insect influx event to occur in Wisconsin.