THE SPATIAL AND TEMPORAL DISTRIBUTION OF ARTH ROPODS IN MICHIGAN CELERY AGROECOSYSTEMS By Jeremy J. Jubenville A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Entomology - Master of Science 2015 ABSTRACT THE SPATIAL AND TEMPORAL DISTRIBUTION OF ARTH ROPODS IN MICHIGAN CELERY AGROECOSYSTEMS By Jeremy J. Jubenville Understanding the sourc e of insects within an agricultural production system is essential to selecting the most suitable management practices. In this thesis, I examine the spatial patterns of arthropods in Michigan commercial celery systems and evaluate whether field margins are sources of insects therein. Insect abundance was measured at different distances within six different celery fields located in southwest Michigan during 2013 and 2014. Most herbivore s were evenly distributed throughout the field, although minor edge effects w ere detected when overall abundance was l ow. Herbivore abundance in margins, relative to the field, varied across groups; tarnished plant bug s and leafhoppers are key pests that may be using margins fo r habitat. Most predators were e venly distributed acros s the field. Syrphid flies were more abundant in margins, suggesting greater association with these areas than other groups. Parasitoid abundance was also consistently margin -centric across groups during both seasons. In 2014, I performed a mark -capture experiment in w hich three field margins were sprayed with a protein marker. Arthropods were subsequently captured and tested for the presence of the protein. Most groups had at least one individual testing positive for the presence of the marker, but the maj ority of specimens in every group tested negative. Variation in mark percentage, as well as the spatial distribution of marks, showed that some groups , such as tarnished plant bugs and syrphid flies, utilize the margins more than others. Any systematic cha nge in the way growers manage their margins would likely affect these groups to a greater degree than the other groups in the study. iii ACKNOWLEDGEMENTS I would like to thank Zsofia Szendrei for providing the opportunities and conditions that facilitated my academic and professional development. I could not have asked for a better advisor. I also would like to thank the other members of my graduate committ ee, Doug Landis and Phil Robertso n, for their wisdom and support; the most influential guidance always c ame in the form of a question. I greatly appreciate the southwest Michigan celery farmers who let me do this work in their fields: Ron Eding, Jake Willbrandt, Mark Cnossen, and Bruce & Eric Schreur. I am truly indebted to Jen Lau and Kay Gross for their ho spitality and support during my extended visit to Kellogg Biological Station. There is no way that I would have been able to complete this project without the help of Andy Galimberti, Alec Ashburn, Angie Lane, Jessica Kansman, and Gabriel King. I am thankf ul for the help of Mark Hammond and Cindy Mills, lab managers extraordinaire. My research and graduate degree were supported by Project GRE EEN, the Rhodes (Gene) Thompson Memorial Fellowship, a Michigan Vegetable Council scholarship, and two G.H Lauff scho larships from KBS. My summer undergraduate assistant, Gabriel King, was supported by the KBS URA program in conjunction the MSU College of Agriculture and Natural Resources. Finally, I would like to express my deepest love and appreciation for my wife, Mega n, for her support during the whole process. This would not have happened without her encouragement. I dedicate this thesis to my grandparents, John T. and W. Lou Bush, who showed me the value of hard work, persistence, and resourcefulnes s. I wish they could see me now. iv TABLE OF CONTENTS LIST OF TABLESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉÉÉÉÉ É...vii LIST OF FIGURESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .ÉÉÉÉÉÉÉÉÉÉÉÉÉ.viii CHAPTER 1: Spatial ecology of arthropods an d celery production in southwest Michigan ÉÉ..1 IntroductionÉÉ. ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.1 Weeds in celeryÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.. 3 Pathogens of celeryÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉ.4 Insect pests in celeryÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. É.ÉÉÉ.4 AphidsÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..5 Aster leafhopperÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉÉÉÉÉ..7 Tarnished plant bugÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉÉÉÉÉ8 Spatial ecology of arthropodsÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉÉÉ9 Disturbance and recolonization in celery fieldsÉ ÉÉÉÉÉÉÉÉÉÉÉ...ÉÉÉ..13 Thesis objectivesÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉ17 CHAPTER 2: The spatial and temporal distribution of pests and natural enemies in Michigan commercial celery systems ..ÉÉÉÉ.ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.19 IntroductionÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ........19 Materials and metho dsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É...É21 Study sites ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉÉ21 Sampling methods ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉ22 Statistical analysis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉ23 ResultsÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ....24 Overvi ewÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ24 Spatiotemporal distribution of pestsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .ÉÉ.25 AphidsÉÉÉÉÉ ÉÉÉÉÉ.ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. ..25 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ25 Weekly abundance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.25 Thrips ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.É26 Seasonal abun dance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ27 Weekly abund ance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.27 Potato leafhopperÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...27 Seasonal abun dance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ28 Weekly abund ance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.28 Aster leafhopper ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉ29 Seasonal abun dance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ29 Weekly abund ance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.30 Tarnished plant bugÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ30 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ31 Weekly abundance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.31 Spatiotemporal di stribution of natural enemiesÉÉÉÉÉÉÉÉÉÉÉÉÉ.32 v SpidersÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...ÉÉÉ .32 Seasonal abun dance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ32 Weekly abundance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.33 Coleomegilla maculata ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..É É33 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ34 SyrphidaeÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... ÉÉ34 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ35 ChalcidoideaÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉ35 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ36 Weekly abund ance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.36 MymaridaeÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.37 Seasonal abundanc e patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ37 Weekly abundance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.38 TrichogrammatidaeÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. É..38 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ39 BraconidaeÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉ.39 Seasonal abundance patternsÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ40 Weekly abund ance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ..É.40 DiscussionÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..49 AphidsÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ49 Thrips ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.50 Potato leafhopperÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... É50 Aster leafhopperÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... ÉÉÉÉ..51 Tarnished plant bugÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..É ..52 SpidersÉ. ÉÉÉÉÉÉÉÉÉ...ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ52 Coleomegilla maculata ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉÉ É.53 SyrphidaeÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉ ÉÉ..53 Parasit oidsÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉ ÉÉÉ54 SynthesisÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ55 CHAPTER 3: The movement of arthropods between marg in and field in Michigan celery agroecosystems ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉ 58 IntroductionÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ..ÉÉÉÉÉÉÉÉÉÉ..ÉÉ58 Materials and Metho dsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.É..61 Study s itesÉÉÉÉ É.ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.61 Mark -captureÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉ.62 ImmunoassaysÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..ÉÉ..64 Arthropod abundance patternsÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉÉÉÉ..65 Statistical analysisÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉ.66 ResultsÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ68 OverviewÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ....68 Aster leafho pperÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.69 Potato leafhop perÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉ.70 Tarnished plant bugÉÉÉÉÉÉÉ...ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.70 Syrphidae ÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.71 Coleomegillia maculata ÉÉÉÉÉÉÉ...ÉÉÉÉÉÉÉÉÉÉÉÉÉ.É.71 vi ChamaemyiidaeÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉ..71 BraconidaeÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉÉÉÉÉ ÉÉ..ÉÉÉ.72 Myma ridaeÉÉÉÉÉÉÉ...ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉ..72 ChalcidoideaÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉ73 ControlsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... ÉÉÉÉÉ.ÉÉ.ÉÉÉÉÉ73 Plant tissueÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉÉÉ73 Insect negative controlÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...ÉÉ74 DiscussionÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..ÉÉ78 False negatives and false positivesÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ82 Marking efficiencyÉÉ..É.ÉÉÉÉÉÉÉÉ........ ÉÉÉÉÉÉÉÉÉÉ.83 ContaminationÉÉÉÉÉ... ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ....84 Summary and App licationÉ ....................ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.85 CHAPTER 4: Conclusions and future research.ÉÉÉÉÉÉÉÉÉ ÉÉÉÉ..É ÉÉÉ.É.88 APPENDICES ÉÉÉ...ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉÉÉÉÉ...92 Appendix 1.1 Supplementary dataÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .....93 Appendix 1.2 RECORD OF DEPOSITION OF VOUCHER SPECIMENSÉÉÉÉ.. 109 LITERATURE CITEDÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ....110 vii LIST OF TABLES Table 1.1. Table of pesticide mixtures from a southwest Michigan celery farm in 2014. Insecticides, highlighted in red text, were added to every application of fungicide or herbicide ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.... É.18 Table 1.2. Pesticide schedule for two celery fields on one Michigan celery farm in 2014. Mix numbers correspond to those specified in Table 1.1. ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... É..18 Table 2.1. Significance of fixed effect (sample position) on weekly or monthly herbivore abundance in commercial celery fields in southwest Michigan during 2013 and 2014. ALH - aster leafhopper, PLH - potato leafhopper, TPB - tarnished plant bugÉÉÉÉÉÉÉÉÉÉ.. 41 Table 2.2. Significance of fixed effect (sample position) on weekly or monthly natural enemy abundance in commercial celery fields in southwest Michigan during 2013 and 2014. CMAC - Coleomegilla maculata .ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.É42 Table 3. 1. The number of arthro pods (n) recovered from yellow sticky traps deployed at four sampling positions across three celery farms in southwest Michigan. Percentage (%) indicates the proportion of taxa testing positive for the presence of chicken egg protein marker; hash sign (#) indicates the number of positive marks. ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...75 Table S1. The total number, frequency, and feeding functional group of arthropods captured on sticky traps in celery fields across southwest Michigan in 2013 and 2014. Frequency is percentag e (%) of traps with at least one individual. "(Aq)" indicates aquatic taxa. .............ÉÉ93 Table S2 . Number analyzed (N), mean (+/ - SEM) ELISA optical density value (OD), and the percentage of sample s (%) testing positive for presence of albumin protein marker. Leaf samples were acquired before and after protein application to margins. Sticky trap sample discs were acquired from the recovered traps. Post -application tests of celery foliage and sticky traps were used to detect contamination resulting from wind drift, precipitation, or condensation..ÉÉ.É97 Table S3 . Number analyzed (N), mean (+/ - SEM) ELISA optical density value (OD), and the percentage of individuals (%) testing positive for presence of albumin protein marker. All insects were collected from three different celery farms with sweep nets one week prior to protein marker application. ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ....ÉÉÉÉÉÉÉ.. .ÉÉ98 Table S4. Voucher specimens deposited at the Albert J. Cook Arthropod Research Collection (Michigan State University)ÉÉÉÉ .........................................................................................109 viii LIST OF FIGURES Figure 1.1. Spirea aphids feedi ng on a celery leaf. (Photo: Z. Szendrei) ÉÉÉÉÉ... É..ÉÉ...5 Figure 1.2 . Satellite images of celery fields from two different farms in (A) Allegan County and (B) Van Buren County in southwest Michigan. Arrows indicate some celery fields throughout the landscape in Van Buren County; each of these fields is approximately 50m wide. Note that many fields are long, narrow, and are bordered lengthwise by ditches ÉÉÉÉÉÉÉ.. É..É.14 Figure 1.3 . Photographs of marginal areas bordering celery fields from two different farms in (A) Van Buren County and (B) Allegan County in southwest Michigan. Margins and ditch banks are managed with mowing (A) or herbicides (B) ÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉÉÉÉÉÉ14 Figure 1.4 . Photographs of celery fields from two different farms in (A) Van Buren County and (B) Ottowa County in southwest Michigan. Note the sharp interface between crop, soil, and margin vegetati on.ÉÉÉÉÉ......................................................................................................15 Figure 2.1 . Generalized canopy sampling design. Transects for yellow sticky traps were established in six commercial celery fields located in three southwest Michigan counties in 2013 and 2014. Yellow dots denote sampling locations ÉÉÉÉÉÉÉÉÉÉ... É.. ÉÉÉ...ÉÉ .21 Figure 2.2 . Weekly or monthly mean abundance ( ±SEM) of herbivores captured in the celery canopy with yellow sticky traps at different dist ances into the field for 2013 and 2014. Asterisks indicate significant differences among sampling positions within a particular week ( ! = 0.05). ALH - aster leafhopper, PLH - potato leafhopper, TPB - tarnished plant bug.É... .................É.43 Figure 2 .3. Weekly or monthly mean abundance ( ±SEM) of predators captured in the celery canopy with yellow sticky traps at different distances into the field for 2013 and 2014. Asterisks indicate significant differences among sampling positions within a particular week ( ! = 0.05). CMAC - Coleomegilla maculata .ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..É44 Figure 2.4. Weekly mean abundance ( ±SEM) of parasitoids captured in the celery canopy with yellow sticky traps at different distances into the field for 2013 and 2014. Asterisks indicate significant differences among sampling positions within a particular week ( ! = 0.05)ÉÉ....É45 Figure 2.5 . Overall (seasonal) mean abundance ( ±SEM) of herbivores at different distances into the field for 2013 and 2014 seasons. Bars with the same letter within the same year are not significantly different (TukeyÕs HSD; ! = 0.05). ALH - asterleafhopper, PLH - potato leafhopper, TPB - tarnished plant bugÉÉÉ.. ÉÉ.ÉÉÉÉÉÉÉÉ. ÉÉÉÉÉÉÉÉ...46 Figure 2.6 . Overall (seasonal) mean abundance ( ±SEM) of predators at different distances into the field for 2013 and 2014 seasons. Bars with the same letter within the same year are not significantly different (TukeyÕs HSD; ! = 0.05). CMAC - Coleomegilla maculata .ÉÉÉ... .....47 ix Figure 2.7 . Overall (seas onal) mean abundance ( ±SEM) of parasitoids at different distances into the field for 2013 and 2014 seasons. Bars with the same letter within the same year are not significantly different (TukeyÕs HSD; ! = 0.05)ÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉ.É.ÉÉ.48 Figure 3.1. Mark -capture sampling design. Transects for yellow sticky traps were established in three commercial celery fields across three farms in southwest Michigan in July 2014. Yellow dots denote sampling locations. Orange rectangle denotes protein application areaÉÉÉ.ÉÉ62 Figure 3.2. Mean abundance of arthropods at four different sampling positions from data collected with yellow sticky traps in six fields across three different celery farms in southwest Michigan during the first week of July 2014. Data are stan dardized on a scale ranging from 0 -1 to enable comparisons across taxa. ÉÉÉÉ... ÉÉÉÉÉÉÉÉÉÉÉÉÉ... ÉÉÉÉ..É76 Figure 3.3. The relationship between the Margin -Field ratio (MF) and the proportion of individuals testing positive for the presence of the chicken egg protein marker. The MF ratio is the abundance of arthropods caught in the margin relative to those caught in the celery field (please see Methods section). Regression points and lines were plotted with different symbols and colors according to feeding fun ctional group. The black regression line is the result of all points in a singular analysis. É.ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... ÉÉ..77 Figure S1. Mean distribution of alate aphids across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. .ÉÉÉÉÉÉÉÉÉÉÉÉ..É99 Figure S2. Mean distribution of thrips across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...100 Figure S3. Mean distribution of potato leafhoppers across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale ÉÉÉÉÉÉÉÉÉÉÉ.. ....É101 Figure S4. Mean distribution of aster leafhopper s across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly differ ent. Note: each year is at a different scale. ...ÉÉ ÉÉÉÉ.ÉÉÉ... É..É102 Figure S5. Mean distribution of tarnished plant bugs across four different sampling locations over all studied celery fields in southwest Michigan in 2014. Abundance is proportional to c ircle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: eac h year is at a different scale. .ÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉÉ.. É.....103 x Figure S6. Mean distribution of spiders across four different samplin g locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a di fferent scale. .ÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ÉÉ.É..É104 Figure S7. Mean distribution of chalcidoid wasps across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. ÉÉÉÉÉÉÉÉÉÉÉÉ.É105 Figure S8. Mean distribution of mymarid wasps across four different sampling locations over all stud ied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. .ÉÉ.. ÉÉÉÉÉÉÉÉÉ..É106 Figure S9. Mean distribution of braconid wasps across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. ÉÉÉÉÉÉÉÉÉÉ.. É...É107 Figure S10. The number of individuals testing positive for the presence of a chicken egg protein marker applied to the margins of t hree celery fields in southwest Michigan. ÉÉÉ... ÉÉÉ..108 1 CHAPTER 1: Spatial ecology of arthropods and celery production in southwest Michigan Introduction Celery ( Apium graveolens ) is a biennial plant in the Apiaceae selected for its large succulent petioles. Although there is historical record of celery usage around the 4th century BCE as a medicinal herb, cultivation as an edible vegetable began in Europe around the 17th century (Levine 2013). It was first established as a commercial crop in the United States during the late 19th century in Kalamazoo, MI (Berger et al. 1988, Levine 2013). Michigan currently accounts for 5.5% of total U.S celery production with 98.5 million pounds h arvested from 1700 acres and a market value of just under $19 million (USDA NASS, 2015). Although most of the production in Michigan occurs in the southwest region of the state, there is notable acreage in the west central and eastern counties as well. Most, if not all, commercial celery production in Michigan occurs o n muck soils (histosols) (Zandstra et al. 1986, Wong et al. 2010). In temperate l atitudes, cultivated muck soils are usually remnant geographic water features that have either dried up due t o natural processes or have been intentionally drained for cultivation (Brady & Weil 2007, Lindbo & Kozlowski 2005). By their nature, muck soils are often found in low -lying areas where the water table has historically been near or above the surface. Field s in these systems are susceptible to flooding, which can damage a crop under prolonged circumstances. Growers manage excess water with a combination of drainage tiling, water pumps, ditches, and dikes (J. Jubenville pers. obs., Brady & Weil 2007, Lindbo & Kozlowski 2005). Consequently, celery farms tend to have long, narrow fields (e.g. 400 -800m x 50m) that are bordered on two or three sides by drainage ditches . 2 The organic matter c ontent of muck is high when compared to the more common mineral soils, res ulting in a fine -textured, nutrient -rich growing medium (Brady & Weil 2007, Lindbo & Kozlowski 2005). Muck systems are innately productive due to this high percentage of humus in the soil and a consistent water supply. When dry and bare, however, these soi ls are so light that they are highly susceptible to wind erosion. In the spring, some growers will cover freshly transplanted fields with a woven material to mitigate soil los s and protect the young plants from chafing by blowing soil (Zandstra et al. 1986, E. Schreur, pers. comm., J. Jubenville, pers. obs.). Retaining these soils is a persistent challenge; consistent cultivation results in annual losses from erosion and oxidation of organic matter (Lucas 1982, Brady & Weil 2007, Lindbo & Kozlowski 2005). Even with these drawbacks, the characteristics of muck soil make it an excellent growing medium for celery and many other vegetables (Zandstra et al. 1986, Brady & Weil 2007, Lindbo & Kozlowski 2005). There are two main types of cultivated celery, ÔPasca lÕ, which has green stalks (petioles), and ÔGolden HeartÕ (a.k.a. ÔKalamazooÕ) which Òself -blanchesÓ and has white stalks (Zandstra et al. 1986, Levine 2013). Three cultivars of Pascal celery that are currently popular with Michigan growers are ÔGreen BayÕ for favorable fresh market characteristics, ÔSabrosoÕ for processing, and ÔDutchessÕ, which is used as a dual -purpose cultivar. Michigan growers start celery from seed inside greenhouses in mid -February where they are nurtured for approximately eight week s. Seedlings are then transplanted into the field around mid -April at a density of ~40,000 plants/acre. Field preparation consists largely of amendments and cultivation, although drainage issues are occasionally addressed. Cover crops, if any, will have been turned under and a pre-emergent herbicide is sometimes applied (Hausbeck 2011, Zandstra 2012). Many growers will 3 use either a Brassica (oilseed radish) or Sorghum -sudan grass as a cover crop specifically for their anti -microbial properties (J. Eding pers. comm., Wong et al. 2010). Cold temperatures can cause premature bolting in celery plants, so early -Spring transplants are sometimes protected by row or field covers (Zandstra et al. 1986). Celery rem ains in the field 80-95 days depending on the culti var, growing conditions, and harvest schedule. Mechanical harvesting is a frequent option, although some growers employ workers to harvest by hand when the produce is designated for fresh market. Weeds in celery The characteristics of muck soils that make them desirable for growing vegetables are also favorable to weed growth . The list of weeds common among these systems is long, and several of them in our region have developed resistance to herbicide (R Eding pers. comm. M. Cnossen pers. comm., J. Willbrandt pers. comm., Zandstra 2012). Som e of the most common weeds are p igweed ( Amaranthus spp .), giant ragweed ( Ambrosia trifida ), lambÕs quarter (Chenopodium album ), common ragweed ( Ambrosia artemisiifolia ) velvetleaf ( Abutilon theophrasti ), smartw eed (Polygonum spp.), purslane ( Portulaca oleracea ), and yellow nutsedge (Cyperus esculentus ) (J. Jubenville pers. obs., Hausbeck, 2011). Growers are proficient at keeping their fields relatively free of weeds. Nevertheless, these species occur persistentl y along the numerous margins and ditches that occur in these systems. Plant species that are common and abundant within these marginal areas include, giant ragweed, velvetleaf, pigweed, wild carrot (Daucus carota ), poison hemlock ( Conium maculatum ), smartw eed, field bindweed (Convolvulus arvensis ), common nettle ( Urtica dioica ), common milkweed ( Asclepias syriaca ), yellow nutsedge, and assorted grasses. Margin weed management consists of cultivation, 4 mowing, or post -emergent herbicide (Hausbeck 2011, Zandstra 1986). Mowing may be the most common method employed and occurs in any given field approxi mately once per month. Pathogens of celery In the suite of plant pathogens that attack celery, some of the most important are foliar blights caused by the fu ngi Septoria apiicola and Cercospo ra apii , a bacterial blight caused by Pseudomonas syringae pv. apii , Fusarium yellows caused by the fungus Fusarium oxysporum , and the Aster yellows phytoplasma (Lacy et al. 1996, Hausbeck 2011). Colletotrichum acutatum is a fungus recently indentified from samples taken from commercial celery fields in Michigan (Rodr™guez -Salamanca et al. 2012). Although Colletotrichum spp. have been known to cause anthracnose in a variety of other fruit and vegetable commodities, this dis ease was previously unconfirmed in Michigan celery (Peres et al. 2005, Dillard 1992, Rodr™guez -Salamanca et al. 2012). Disease caused by the Septoria fungus occurs every year and is the most common pathogen problem in Michigan commercial celery fields (Hau sbeck 2011). Insect pests in celery Some of the major insect pests of celery are aphids ( Aphis spirae cola, Aphis helianthi, Myzus persicae , etc.), the tarnished plant bug (Lygus lineolaris ), the aster leafhopper (Macrosteles quadrilineatus ), and the variegated cutworm ( Peridroma saucia ). Minor pests include the two -spotted spider mite ( Tetranychus urticae ), the carrot weevil ( Listronotus oregonensis ), and thrips (Thysanopetera: Thripidae). The celery leaftier ( Udea rubigalis ), along with slugs, leafmi ners (Diptera: Agromyzidae), and loopers (Lepidoptera: Noctuidae) can cause great damage in large numbers, but are best described as sporadic pests in Michigan celery fields. 5 Aphids . Aphids are small soft -bodied insects that insert their piercing and suck ing mouthparts into plant phloem tissue to obtain sugars and nutrients. The general aph id life cycle in the northern United States is as follows (summarized from Dixon 1977, 1985): they overwinter as eggs on a primary host plant and hatch as wingless femal es in the spring. These females produce genetically i dentical offspring by viviparous parthenogenesis, with several generations continuing in this manner. As the weather warms, aphid populations will start to produce winged females that disperse to other p lant hosts and once again produce wingless offspring. As fall and winter approach, winged females fly back to their primary hosts to produce both male and female offspring for mating. Mated females will lay eggs on the primary plant host t o pass the winter and the process repeats itself the following spring. The spirea aphid, Aphis spiraecola (He miptera:Aphididae), has been identified (D. Lagos and D. Voegtlin, pers. comm.) as the species responsible for recent infestations in southwest Michigan cel ery fields. It is a small (1.2 -2.2mm) yellow -green aphid with black cauda and black siphunculi ( Fig. 1.1 ). The thorax in the alatae is dark brown or black, while that of the aptarae are Figure 1.1. Spirea aphids feeding on a celery leaf. (Photo: Z. Szendrei) 6 the same color as the abdomen. Aphis spiraecola originated in East Asia, was first recorded in the United States in 1907, and is now distributed throughout the world (Blackman and Eastop 2007). It is a highly polyphagous species with known hosts in twenty plant families, including the Apiaceae, and is one of the most important pests of citrus crops in the United States (Blackman and Eastop 2007). Aphis spiraecola can be difficult to identify in the field due to variable (Lowery et al. 2002) but generally similar morphology to many other aphid species (e.g. A. citricola, A. pomi , A. eugeniae) (Halbert and Voegtlin 1992, Footit et al. 2009, Blackman & Eastop 2007) . In celery, the spirea aphids can easily be confused with the cotton -melon aphid ( A. gossypii ) and, to a lesser extent, the green peach aphid (Myzus persicae ) and the sunflower aphid (A. helianthi). In warmer climates, A. spiraecola populations are usually perm anently parthenogenetic (Blackman and Eastop 2007). Popula tions in the northern U.S. and E ast Asia, however, are cyclically parthenogenetic and are known for using Spirea spp. as primary hosts for overwintering. Nevertheless, there are specific populations in Japan that use Citrus spp. as a primary host and research in Washington apple crops (Lowery et al. 2002) suggest that there are some populations that overwinter on app le trees ( Malus spp.). Thus, while Spirea spp. can be considered the most likely primary host for any given population, it is not guaranteed to be the only primary host. Aphid infestations can produce curling and distortion in leaves, fungal growth on excreted honeydew, and excessive amounts of exuvia, which together can render a vegetable plant unsuitable for fresh market. In celery, A. spiraecola tend to aggregate on young tissue near the top of the petioles; infestations can be obvious to the eye. Althoug h disease transmission is a concern with phloem -feeders , incidence seems to be lower with A. spiraecola until populations 7 grow large (Blackman and Eastop 2007). Far more concerning, however, is the development of pesticide resistance in Spirea aphids that have been reported in other crops (Hogmire et al. 1992, Smirle et al. 2010, Blackman & Eastop 2007) and is likely causing problems for Michigan celery growers (Z. Szendr ei, pers. comm.). Aster leafhopper . The aster leafhopper, Macrosteles quadrilineatus Forbes (Hempitera: Cicadellidae), is a small (~3mm) yellowish -green leafhopper with six black spots across the head and smoke -colored wings. It is arguably the most conc erning of insect pe sts to celery growers due to it s ability to vector the Aster Yellows phytoplasma (AYp), a cell wall -less bacteria -like organism that causes the Aster Yellows disease in vascular plants (Meade and Peterson 1964, Beanland et al. 2005, Chri stensen et al. 2005, Frost et al. 2011). Macrosteles quadrilineatus is distributed widely across the continental U .S. (both east and west of the Rocky Mountains) and is known to migrate thousands of miles every spring from southern locations to the northern states (Westdal et al. 1961, Meade and Peterson 1964, Hoy et al. 1992). The aster leafhopper can also overwinter as eggs on foliage as far north as Manitoba and Quebec (Westdal et al. 1961). Thus, locally observed populations in the Midwest are likely comprised of both migrant and overwintered individuals. The host range for the aster leafhopper is broad, having been r ecorded on over 300 plants and many commercial vegetable crops, including celery (Wallis 1962, Frost et al. 2011). Likewise, the AYp has been found in over 300 plant species and causes physiological distortions such as twisting, stunting, adventitious root s, and chlorosis in the infected host (Christensen et al. 2005, Frost et al. 2011). In commercial vegetable production, the quick spread of this disease can render an entire crop unmarketable (Hoy et al. 1992). 8 Tarnished plant b ug. The tarnished plant b ug, Lygus lineolaris (Hemiptera: Miridae), is an herbivore with a ge ographical range as broad as it s dietary niche. Its distribution encompasses the United States as well as southern Canada, and has been found as far north as the Yukon and Alaska (Kelton 1 975, Young 1986, Capinera 2001). The tarnished plant bug is a true generalist, having been observed to feed on over 300 plant species (Young 1986), although growth, development, and fecundity has been shown to vary with hosts. Previous research indicates a strong association with early secondary succession al communities and weedy margins (Snodgrass et al. 1984, Young 1986, Capinera 2001, Snodgrass et al. 2005, Outward et al. 2008), giving it a reputation as a Òborder speciesÓ. Tarnished plant bugs overwint er as adults and emerge early in the spring to feed on the local f lora. Oviposition occurs shortly thereafter with eggs ha tching in 1 -2 weeks depending on the temperature (Ridgeway and Gyrisco 1960). There are five instars that, with warm weather, can proceed to the adult stage in approximately twenty days (Ridgeway and Gyrisco 1960). Longevity in adults ranges from 40 -60 days for females and 30 -40 days for males (Capinera 2001). In southwest Michigan, there are probably three generations per year. Lygus bugs, including the tarnished plant bug, have an extensive suite of natural enemies. The most important parasitoids are reported to be the mymarid egg parisitoid Anaphes iole (Hymenoptera: Mymaridae) and several nymphal parasitoids from the Braconidae: Leio phron uniformes , Peristenus palipes , P. relictus , and P. pseudopallipes (Clancy and Pierce 1966, Scales 1973, Graham et al. 1986, Norton et al. 1992, Braun et al. 2001, Seymour et al. 2005, Pickett et al. 2009). Documented predators include Nabis spp., Orius spp., Geocoris spp., Zelus spp., the spined soldier bug ( Podisus maculiventris ), spiders, ants, and lady beetles (Leigh and Gonzalez 1976, Whalon and Parker 1978, Hagler 2011). 9 Damage to plants from tarnished plant bugs varies with the species. In g eneral, they seem to prefer new (meristematic) tissue and reproductive structures (Boersma and Luckman 1970, Khattat and Stewart 1975, Schafers 1980, Grafius and Morrow 1982, Fleischer and Gaylor 1987, Attanazov et al. 2002, Foshee et al. 2008). Because celery in Michigan is not grown for seed, lygus bugs will instead feed on the new tissue in the heart of the celery. Feeding damage to the young tissue causes discoloration and leaf distortion, often rendering the plant unmarketable (Stewart and Khoury 19 78, Boivin et al. 1991). Spatial ecology of arthropods One of the defining characteristics of agricultural systems is c onsistent periodic disturbance (Landis and Marino 1999 , Landis et al. 2000). Ecological disturbance can be defined as an event that causes a change in the spatial patterns of organisms in an ecological system (Wissinger 1997, Debinski and Holt 2000, Tscharntke et al. 2005b, Schowalter 2012) . These changes are often due to mortality and, in natural habitats, can be caused by events that ra nge in scale from small fires to regional droughts and natural disasters. Disturbance in an agricultural system often comes in the form of plant or pest management practices such as plowing, harvesting, pesticide application, and mowing. I n celery fields, most disturbance to the insect community comes from frequent pesticide applications. These disturbances are often intense and almost always result in a mass local extinction of whole groups of organisms, including pestiferous and beneficial insects. The purpose of this study is to explain the patterns of distribution and abundance of insect groups within celery fields. Explanation requires an analysis based on multiple spatial scales (Tscharntke et al. 2005b). To a certain extent, what we find in celery f ields and their border habitats is the result of filtering of regional species pools by the local environment and 10 biotic interactions (competition, predation, mutualism) (Leibold et al. 2004,Tscharntke et al. 2005a). Biodiversity at any particular location can also be influenced by the surrounding landscape, which functions at a scale in between the locality and the region (Steffan -Dewenter et al. 2002, Tscharntke and Brandl 2004, Tscharntke et al. 2005). Agric ultural landscapes are a mosaic of ecosystems that reflect human land -use patterns, where natural structural and ecosystem heterogeneity is replaced with a collection of simplified monocultural cropping systems (Tscharntke et al. 2002, 2005). Natural and s emi -natural fragments (hedge rows, field margins, etc.) are interspersed among crop fields, roads, and developed property to create heterogeneity and structural complexity that is much different than the previously existing natural areas (Tscharntke et al. 2002, 2005a, 2007). Spatial separation of habitats may require some species to exploit resources from a number of disconnected habitats (Steffan -Dewenter et al. 2002, Theis et al. 2003, Tscharntke and Brandl 2004). Therefore, understanding local populatio n and community dynamics in a landscape in which habitats of various size and shape are fragmented, poorly -connected, or completely isolated requires analysis at multiple spatial scales. There are a couple of influential models that have informed our unde rstanding of spatial ecology (Tscharntke and Brandl 2004, Rand and Louda 2006) and are often applied as a conceptual basis for predicting insect diversity in agricultural systems: island biogeography theory (IBT) (MacArthur and Wilson 1967) and metapopulat ion theory (Levins 1969). IBT predicts lower species richness with decreasing island size and increasing isolation. Although developed to predict biodiversity on ocean islands, it has been applied to predict diversity in terrestrial natural habitats embedd ed within an anthropogenically -altered matrix (i.e. the terrestrial ÒoceanÓ) with varying degrees of success (Fahrig and Merriam 1994, Kruess and 11 Tscharntke 1994, Rozenzweig 1995, Debinski and Holt 2000, Cook et al. 2002, Rousch et al. 2013). The greatest limitation of IBT in predicting biodiversity in a fragmented landscape is that natural habitats are not usually surrounded by a wholly hostile matrix (e.g. the ocean to an island) and may provide resources for habitat generalists (Janzen 1983, Marino and L andis 1996, Jonsen and Fahrig 1997, Landis et al. 2000, Cook et al. 2002, Tscharntke et al. 2002, 2005b, Tscharntke and Brandl 2004, Rand et al. 2006). Species spillover from several types of habitats can result in a greater alpha diversity for individual fragments and greater beta diversity for the landscape than IBT would predict (Tscharntke et al. 2002, Leibold et al. 2004, Tylianakis et al. 2005, 2006). Metapopulation theory is concerned with extinction and colonization dynamics across a collection of habitats within a given area. It is used to predict the likelihood of a particular species occupying habitat patches within a landscape and can be expressed as a ratio of the number of extinctions across patches to the number of colonizations of previously unoccupied patches over a given period of time. Like IBT, metapopulation theory has its own set of limitations that are, in this case, largely rooted in its general simplicity (Hanski 1998, Tscharntke et al. 2005b). Expansions of this theory have overcome some of the limitations by including spatially explicit components and accommodating community interactions and local population dynamics (Hanski 1998). The increase in realism and predictive accuracy, however, is accompanied by substantial complexity (Ha nski 1998), the discussion of which is beyond th e scope of this thesis. I used metapopulation theory here simply as general framework for understanding: 1) that there are source -sink dynamics and bi -directional movement patterns among insect populations ac ross habitat patches within agricultural landscapes (Shmida and Wilson 1985, Holt 1993, Pulliam 1998, Bianchi et al. 2006, Tscharntke et al. 2007), and 2) 12 without immigration, extinction is the only major force acting on local populations (Levins 1969, Han ski 1998). The combination of IBT and metapopulation theories lead to several predictions that are germane to the subject of insect pests and natural enemies in agricultural mosaics: the effects of dispersal ability on species persistence within a fragmen ted landscape, niche breadth, and the differential effects of habitat fragmentation across trophic levels. Greater dispersal ability allows for a greater chance to encounter and colonize isolated patches within the landscape (MacArthur and Wilson 1967, Han ski 1998, Fahrig 2003). Species with lower vagility experience a greater risk for regional extinction because emigrants from isolated patches may not be able to locate other suitable patches (Hanski 1998, van Nouhuys and Hanski 2002, Fahrig 2003). Fragment ation is not predicted to be as important as overall habitat loss to generalist species due to their ability to exploit resources from a variety of habitats (Jonsen and Fahrig 1997, Krauss et al. 2003). Even with lower dispersal ability, a generalist may o nly need to travel a short distance to find necessary resources. Specialists, however, are predicted to be at a disadvantage due to a combination of habitat loss, greater isolation of suitable habitats, and dependence on the persistence of another species (Holt et al. 1999, 2002, van Nouhuys and Hanski 2002). Should their obligate prey or host aggregate preferentially in agricultural fields, their necessary resource may disappear completely in the event of a pesticide application. Finally, because populati on densities at higher trophic levels tend to be lower and more variable, the risk for extinction of local populations increases with isolation and habitat reduction (Kruess and Tscharntke 1994, Holt 2002). This effect is predicted to be greater for parasi toids than predators due to their propensity for host specificity (van Nouhuys and Hanski 1999, Holt 2002). Consequently, local 13 extinction of natural enemy populations may release herbivores from top -down control and increase the likelihood of an outbreak (Kruess and Tscharntke 1994). Disturbance and recolonization in celery fields Field sizes and shapes for growing celery in southwest Michigan varies based on the total area of muck soil available at any particular location. Some gr owers engage in Òpocket muck farmingÓ (Bruce Klamer, pers. comm.) by exp loiting multiple small pockets of muck soil distributed over the landscape. These small fields are likely the remains of old ponds and are often subject to flooding after strong rains due to the fact that muc k occurs naturally in low -lying areas. Installing drainage technology in these small areas is probably not economical and so the growers simply tolerate the occasional flooding. Large areas of muck, on the other hand, benefit from the economy of scale and so growers will often construct and install a network of drainage ditches and tiling to lower the water table across the entire area of muck soil. A common result of this strategy is an aggregation of long narrow fields across the muck area, each bordered on two or three sides by drainage ditches. The actual w idth of these fields are based on grower equipment, but are often ~50m wide (J. Jubenville, pers. observation, Fig. 1.2A-B). The width of a drainage ditch varies, but is typically 2 -3m wide. Some growe rs cultivate the soil as close to the slope of the ditch as possible to maximize growing area and reduce weed p ressure, whereas some leave uncultivated space on both sides to create a fairly wide crop -free space between fields. This crop -free space can be 6-7m wide and is usually populated by a mixture of agricultural weeds and regional flora (Figs. 1.3A -B, 1.4A). Growers do not usually allow the plants in these areas to grow significantly and mow or spray herbicide to control them (Fig. 1.3A-B). Mowing occ urs, on average, once per month during the season; herbicide application usually happens just before transplanting and can be the only 14 Figure 1.2 . Satellite images of celery fields from two different farms in (A) Allegan County and (B) Van Buren County in southwest Michigan. Arrows indicate some celery fields throughout the landscape in Van Buren County; each of these fields is approximately 50m wide. Note that many fields are long, narrow, and are bordered lengthwise by ditches. margin management need ed for the life of the crop (R. Eding. pers. comm., Zandstra 1986). In addition to this management, growers will often leave cultivated, plant -free soil as a buffer zone between margins and the crop. This buffer zone results in a ÒhardÓ visual interface be tween the lush green crop, the dark muck soil, and the weedy margin -ditch area (1.4A -B). Thus, the Figure 1 .3. Photographs of marginal areas bordering celery fields from two different farms in (A) Van Buren County and (B) Allegan County in southwest Michigan. Margins and ditch banks are managed with mowing (A) or herbicides (B). 15 arrangement of the celery cr op, drainage ditches, and their margins re sults in smaller fields with a higher edge-to-area ratio than crops grown on upland mineral soils (e.g. corn/soy/wheat, squash, solanaceous fruit ). Figure 1.4 . Photographs of celery fields from two different farms in (A) Van Buren County and (B) Ottowa County in southwest Michigan. Note the sharp interface between crop, soil, and margin vegetation. Fungicide application for most growers occurs on a weekly calendar -day basis (E. Schr eur & R. Eding pers. comm ) and many will add a broad -spectrum insecticide to the spray tank as a matter of economy (R. Ed ing & M. Cnossen pers. comm., Table 1.1, Table 1.2 ). This weekly insecticide treatment eliminates virtually all a rthropods from celery fields, but insect s have often re-colonized these areas by the time of the next application (J. Jubenville, pers. obs.). In general, insect colonists could come from three places: natural areas within the landscape, 16 semi-natural or less -disturbed areas (e.g. hedgerows , margins, recreational areas) , and the aerial plankton species pool (i.e. migrant and wind -borne organisms) (Buntin et al. 1990). Understanding the sour ce of insects has imp ortant management implications. Margins may be especially important to insects on celery farms. The high number of margins in these systems could provide temporary refuge from pesticides and potentially serve as an immediate source of insect s to re -colonize the crop (Landis et al. 2000, Lee et al. 2001). Furthermore, field margins and other border habitats are the only potential sources within the complete control of the grower. If we think of a celery field immediately after insecticide ap plication as an uninhabited island, we can use a combination of IBT and metapopulation theory as well as our knowledge of grower management practice to predict both the sources of re -colonizing insects and their distribution patterns within the field. The length of time it takes for an insect colonist to reach these ÒislandsÓ depends upon its dispersal strength and the relative isolation of the field (MacArthur and Wilson 1967, Levins 1969, Hanski 1998, Schowalter 2012). Margins, the closest potential source of insects within these systems, experience high -levels of disturbance (mowing, herbicide), but not in the same form or frequency as the crop. The period between disturbances may be enough time for incipient insect populations to grow via recruitment as well as immigration. Thus, it is reasonable to consider the possibility of a source -sink relationship between the margins and the crop. In this scenario, the network of margins could serve as sources, refuges, and corridors that help facilitate the persist ence and dispersal of insects across the farm (Haddad 1999, 2003, Collinge 2000, Debinski and Holt 2000, Landis et al. 2000, Lee et al. 2001, Nicholls et al. 2001, Van Dyck and Baguette 2005, Conradt and Roper 2006, Holzschuh et al. 2009, Delattre et al. 2010). 17 Sources and distribution patterns of any particular gro up or species in this situation should be strongly correlated to its dispersal ability. We know from previous research that colonization of highly disturbed areas usually occurs at the edges fir st and proceeds inward (Schowalter 2012). Because these fields are so narrow, however, we can expect to find that high -dispersal species might be able to reach the middle of these fields (~25m) rather easily, showing uniform distribution patterns. Species that are less mobile may take longer to colonize the area, possibly reflected by a pattern of decreasing abundance with increasing distance from the field edge. Given the narrow field profile with a short distance to the middle, a consistent edge -oriented colonization pattern could signify that margins are a source for insects. If any of these insects also happen to be pests, then this knowledge can inform future margin management decisions. Thesis objective s The intent of this thesis i s to document and analyze the arthropod community in Michigan commercial celery fields. The first objective is to document the spatiotemporal abundance patterns of the arthropod community and determine the most important groups. The second objective is to evaluate the effic acy of current margin management practices and determine if margins are a major source of the insects that we find in the field. 18 Table 1.1. Table of pesticide mixtures from a southwest Michigan celery farm in 2014. Insecticides, highlighted in red text, were added to every application of fungicide or herbicide. Table 1.2. Pesticide schedule for two celery fields on one Michigan celery farm in 2014. Mix numbers correspond to those specified in Table 1.1. Grower MixProductUseInsecticide ClassMixture #1Permethrin 3.2insecticidepyrethroidChloronilfungicideNu Copfungicide-bactericideMixture #2AcephateinsecticideorganophosphateChloronilfungicideNu Copfungicide-bactericideMixture #3Permethrin 3.2insecticidepyrethroidDimethoate 4ECinsecticideorganophosphateChloronilfungicideNu Copfungicide-bactericideMixture #4Lannate LV insecticidecarbamateChloronilfungicideNu Copfungicide-bactericideMixture #5CaparolherbicideLoroxherbicideVolunteer herbicidePermethrin 3.2insecticidepyrethroidField 1Week DateMixtureInsecticide classReason for application131-MayMix #5pyrethroidWeeds/Aster leafhopper 15-JunMix #2organophosphateAster leafhopper317-JunMix #4carbamateAster leafhopper/Tarnished plant bug 425-JunMix #5pyrethroidWeeds/Aster leafhopper 530-JunMix #4carbamateAster leafhopper/Tarnished plant bug 68-JulMix #1pyrethroidAster leafhopper717-JulMix #2organophosphateAster leafhopper/Tarnished plant bug 823-JulMix #5pyrethroidWeeds/Aster leafhopper 931-JulMix #2organophosphateTarnished plant bug/Aphids 108-AugMix #3organophosphateTarnished plant bug/Caterpillars 1220-AugMix #1pyrethroidTarnished plant bug/Caterpillars Field 2Week DateMixtureInsecticide classReason for application17-JunMix #4carbamateAster leafhopper/Tarnished plant bug 425-JunMix #5pyrethroidWeeds/Aster leafhopper 530-JunMix #4carbamateAster leafhopper/Tarnished plant bug 52-JulMix #5pyrethroidWeeds/Aster leafhopper 68-JulMix #1pyrethroidAster leafhopper717-JulMix #2organophosphateAster leafhopper/Tarnished plant bug 931-JulMix #2organophosphateTarnished plant bug/Aphids 108-AugMix #3organophosphateTarnished plant bug/Caterpillars 1220-AugMix #1pyrethroidTarnished plant bug/Caterpillars 19 CHAPTER 2 : The spatial and temporal distribution of pests and natural enemies in Mic higan commercial celery systems Introduction Celery ( Apium graveolens L.) is a biennial plant in the Apiaceae, grown commercially as an annual, and selected for its large succulent petioles. It wa s first established as a commercial crop in the United States during the late nineteenth century in Kalamazoo, MI (Berger et al. 1988, Levine 2013). Michigan currently accounts for 5.5% of total U.S . celery production with 98.5 million pounds harvested fro m 1700 acres and a market value of just under $19 million (USDA NASS, 2015). Although most of the production in Michigan occurs in the southwest region of the state, there is notable acreage in the west central and eastern counties as well. Most, if not all, commercial celery production in Michigan occurs o n muck soils (Zandstra et al. 1986, Wong et al. 2010), which are remnant geographic water fea tures that have dried out due to natural processes, lowering of the local water table through ground water withdrawal, or have been intentionally drained for cultivation (Brady & Weil 2007, Lindbo and Kozlowski 2005). Muck soils have high organic matter content, high water retention capacity, and low bulk density, which are desirable attributes for growing transp lantable crops (Lindbo and Kozlowski 2005). Field sizes and shapes for celery in southwest Michigan vary based on the total area of muck soil available at any particular location and range from small pond -size ÒpocketsÓ of muck to remnant lake and river be ds covering hundreds of acres. While installing drainage technology in small pock ets is economically prohibitive, in large areas of muck , growers will often construct and install a network of drainage ditches and tiling to lower the water table acro ss the entire area (J. Jubenville pers. obs., Brady and Weil 2007, Lindbo and 20 Kozlowski 2005). While the widths of these large fields are based on individual grower equipment, they are often ~50m wide (J. Jubenville, per s. observation) . Consequently, celery farms in southwest Michigan tend to have long, narrow fields (e.g. 400 -800m x 50m) that are bordered on two or three sides by drainage ditches (Figs 1.2A-B). Ditches are bordered on both sides by weedy marginal areas that vary in width, but are usually about 2 meters wide and managed by mowing or herbicide application. Mowing occurs, on average, once per month during the season; herbicide application usually happens just before transplanting and can be the only margin management needed for the life of the crop (R. Eding. pers. comm., Zandstra 1986 , Fig 1.3A-B). In addition, growers will often leave cultivated, plant -free soil as a buffer zone between margins and the crop. This buffer zone results in a ÒhardÓ visual interface between the lush green crop, the dar k muck soil, and the weedy margin -ditch area (Fig. 1.4A-B). Major pests of celery include aphids (Aphis spirae cola, A. helianthi, Myzus persicae ), the tarnished plant bug ( Lygus lineolaris ), the aster leafhopper ( Macrosteles quadrilineatus ), and the variegated cutworm ( Peridroma saucia ). Minor pests include the two -spotted spider mite (Tetranychus urticae ), the carrot weevil ( Listronotus oregonensis ), and thrips (Thysanopetera: Thripidae). The celery leaftier ( Udea rubigalis ), along with slugs, leafmi ners (Diptera: Agromyzidae), and loopers (Lepidopt era: Noctuidae) can cause damage in large numbers, but are best described as sporadic pests in Michigan celery fields. Knowledge of the typical spatial and temporal abundance patterns exhibited by any particular pest is essential for designing efficient and conscientious pest management programs. Apart from occasional extension bulletin s, published descriptions of insects in Michigan celery are sparse. Reference material regarding community composition and relative abundance at the 21 field, local, and regional scale is lacking. Constructing and maintaining an inventory of arthropods across a sizable geographic area allows us to document and monitor taxa of interest and determine whether fluctuations in abundan ce are synchronous across the region. Moreover, determining field -scale spatiotemporal distribution patterns may allow growers to implement site -specific management practices in which insecticides are applied only to targeted locations within a field. Prac tices such as these could lower production costs and reduce the environmental impact of commercial agriculture in these potentially sensitive ecosystems. Providing a within -field refuge from pesticides may also help to conserve natural enemies and could re sult in more consistent pest control throughout the season (e.g. Morrison and Sz endrei 2013). I studied the celery arthropod community as ma naged by experienced growers on established farms with a history of continuous celery producti on. The purpose of th is study i s to document the relative abundance and spatiotemporal distribution of arthropods within the canopy of commercial cele ry crops in Michigan in order to determine important groups and detect aggregative tendencies. Special emphasis is placed on key pests and natural enemies. Materials and Methods Study s ites. Each growing season (2013 and 2014), I selected three commercial celery farms based on geographic separation and grower willingness to participate in cooper ative research projects. I n 2013, farms were located in Ottawa, Allegan, and Van Buren counties and were separated by distances t hat ranged from 23.4 km to 85.6 km (mean distance: 57.6 km). In 2014, farms were located in Allegan and Van Buren counties with a geographic al separation ranging from 35.2 km to 72.2 km (mean d istance: 57.0 km). Two of the farms participated both years, resulting in a total of four farms over the course of two se asons. Two newly transplanted fields 22 from each farm were selected for sampling, which resulted in a to tal of six fields per season. Fields on the same farm were selected to be as far apart as possible . Sampling side of the field was bordered by a drainage ditch whenever possible (~83% of study fields) . Sampling method s. I sampled canopy arthropods using yellow stic ky traps placed at canopy level. I established three transects that were arranged perpendicular to and spatially equidistant along the field edge and extended from the weedy margins into the field. Transect length was uniform and determined by t he distance between the edge and the middle of the most narrow field. Four fixed sampling positions were located along each transect at the following points : the margin, the crop edge, 10 meters into the crop, and 25 meters into the crop for a total of twe lve traps per field (Fig 2.1 ). I used non-baited 7.6 by 12.5 cm yellow sticky traps (Great Lakes IPM, Figure 2.1. Generalized canopy sampling design. Transects for yellow sticky traps were established in six commercial celery fields located in three southwest Michigan counties in 2013 and 2014. Yellow dots denote sampling locations. 23 Vestaburg, MI) placed at canopy level with wire loop hangers. During the first month, the canopy was low and I simply inserted the hangers into the soil . However, as the plants grew taller, I used 76.2 x 1.3 cm galvanized pipe to extend the length of the hanger and raise the traps to canopy level. Traps were replaced weekly between late May and early September in both 2013 and 2014. Arthropods of interest were counted and identified to the lowest possible taxonomic unit. Two specific groups were not counted as natural enemies during the 2013 season: Dolichopoidae (Diptera) and Scymnus spp. (Coleoptera: Coccinellidae). Thus, I only present data on these taxa for the 2014 season. Statistical a nalysis. I used R (R Core Team, 2014) and glmmADMB (Skaug, Fournier, Bolker, Magnussin, and Nielsen 2014) to perform a multilevel mixed -model P oisson regression of the relationship between distance from the edge and arthropod abundance. Data were subset and analyzed by week to accommodate distributional assumptions and improve the fit of the model. Sampling position was entered as a fixed effect. For random effects, I entered intercepts for farm, field, and transect, all of which were nested to account for spatial autocorrelation (transect within field within farm). Model selection and probability distributions (poisson, negative binomial, zero-inflated, etc.) were based on AIC criteria and PearsonÕs di spersion coeffic ient (Hilbe 2014 ). To test for significance of the fixed effect, I performed likelihood ratio tests of the full model against a model without the effect. Taxonomic groups were analyzed individually. Statistical separation of mean abundance between sampling positions was performed by post -hoc analysis using TukeyÕs HSD ( ! = 0.05). In cases where weekly counts of a taxonomic group were consistently too low to analyze, I pooled the data by month or by season. As a rule, I tried to 24 analyze the data in as many temporal subsets as possible in order to obtain the finest resolution of spatial distribut ion possible across the season. I analyzed overall seasonal abundance by pooling weekly data and performing a mixed -model linear regression of sampling position against arthropod abundance in R using lme4 (Bates, Maechler, Bolker, and Walker, 2014). Sampling position, week, and the intera ction of sampling position and week were entered as fixed effects. As before, farm, field, and transect were specified as nested random effects. Data were transformed (log e+1) to improve model fit and meet distributional assumptions of normality and homosc edasticity. To test for significance of the two fixed effects and their interaction, I iteratively dropped the interaction and effects from the model and performed likelihood ratio tests against the full model. P -values are reported in Table 2.1 . Taxonomic groups were analyzed individually. Statistical separation of mean abundance between sampling positions was performed by post -hoc analysis using TukeyÕs HSD (! = 0.05). Results Overview. I recorded a total of 73,366 and 118,185 arthropods in 2013 and 2014, respective ly, for a grand total of 191,551 captured arthropods (Table S1 ). The total herbivore counts were 62,987 in 2013 and 106,92 8 in 2014 and accounted for 85.9 % and 90.5% of yearly captures, respectively. The total number of predators was 3,060 in 2013 and 4,424 in 2014, accounting for 4.2% and 3.7 % of the yearly captures, respectively. Finally, I counted 5,622 parasitoids in 2013 and 6587 in 2014, which accounted for 7.7 % and 5.6% of yearly captures respectively. 25 Spatio -temporal distribution of pests. Aphids . Aphids were the most abundant pest group captured in both years, accounting for 45.8% and 57% of all counted herbivores in 2013 and 2014, respectively (Table S1 ). A total of 28,893 and 60,938 adult aphids were captured on yellow sticky traps in 2013 and 2014, respectively. Aphid abundance was significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=16.04, P<0.001). In both years, mean aphid numbers remained relatively low during the first nine weeks of the seaso n, but started to increase substantially around mid -August and continued to increase every week until the crop was harvested (Figs. 2.2A -B & S1). Seasonal abundance patterns: Overa ll in 2013, aphids displayed a margin -centric pattern with #48% more aphids captured in the margins and #40% more captured at the field edge than at either interior position (Fig. 2.5A). The overall influence of sampling position was statistically significant (Sample position (S): "2(3)=50.26, P<0.001; Fig. 2 .2A), as well a s the interaction with time (Sample position (S) x Time (T ): "2(36)=56.11, P=0.018; Figs. 2.2 A & S1 ). Overall in 2014 , the margin -centric pattern was not repeated . Instead, mean aphid capture was greatest at the 10 m position, with #35% more aphids captured than in the margins and #25% more than either of the oth er field positions (Fig. 2.5B) . Although the influence of sampling position was statistically significant for the season (S: "2(3)=34.95, P<0.001; Fig. 2 .2B), as well as the interaction with time ( SxT: "2(33)=73.11, P<0.001; Figs. 2.2B & S1 ), the differences between the overall means were not statistically separable with post hoc TukeyÕs HSD pair -wise comparison (Fig. 2 .5B). Weekly abundance patterns: Aphids were usually more abundant at the field edge than at the other positions during t he first 5 weeks of 2013 (Fig. S1 ); sample position significantly influenced aphid abund ance during this time (Table 2.1 ). Differences among positions were less 26 apparent during the next three weeks despite significant differe nces in weeks 7 and 8 (Table 2.1 ). Sampling position was not a significant predictor of aphid abundance during the final five weeks, although there appears to be a slight margin -oriented trend in we eks 12 and 13 (Fig. S1, Ta ble 2.1). Aphid abundance was significantly influenced by sampling position dur ing weeks 2 -8 in 2014 (Table 2.1 ). Mean margin captures were always among the most abundant, with numbers generally decreasing with distance into the field (Fig. S1 ). Relativel y speaking, these numbers are low compared with the last month of the 2014 season in which abundance incre ased sharply every week (Figs. 2.2B & S1 ). Similar to 2013, sampling position was not a significant predictor for late -season aphid abundance. The fin al week of 2014 was an exception, however, with aphid abundance significantly higher 10m into the field ( Table 2.1, Fig. S1 ). Thrips . Thrips were the second most abundant pest group captured in both years, accounting for 37.2% and 26.9% of all counted her bivores in 2013 and 2014, respectively (Table S1 ). A total of 23,483 and 28,736 thrips were captured on yellow sticky traps in 2013 and 2014, respectively. The majority of thrips were likely onion thrips, Thrips tabaci , but were too numerous to consistently identify in a timely manner. Thrips abundance was significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=59.60, P<0.001). In both years, thrips abundance was low to start the season, but increased dramatically during the first month (Figs. 2.1C-D). There were three noticeable peaks in abunda nce during the 2013 season: late June, mid -July, and mid -August. Thrips abundance during the 2014 season peaked in late June, decreased over the ne xt three weeks and remained relatively low for the rest of the season (Fig. 2.1D & Fig. S2). 27 Seasonal abundance patterns : Overall in 2013, thrips displayed an even distribution pattern across all sampling locations (Fig. 2.5C). The overall influence of sa mpling position was not statistically significant (S: "2(3)=0.57, P=0.905); neither was the overall interaction with time (SxT: "2(36)=29.51, P=0.769). Overall in 2014, thrips were evenly distributed across all sampling locations within the crop, but signi ficantly more abundant in the margins (Fig. 2.5D). In total, 56.4%, 40.2%, and 38.1% more thrips were captured in the margins than at the edge, 10m, and 25m positions, respectively. The overall influence of sampling position was statistically significant ( S: "2(3)=24.15, P<0.001), as well as the interaction with time (SxT: "2(33)=55.69, P=0.008, Figs. 2.5D & Fig. S2 ). Season mean thrips captures were 65.8%, 47.3%, and 45.1% greater in the margins than they were at the edge, 10m, and 25m positions, respectively (Figs. 2.1D & 2.5D). Weekly abundance patterns : Mean abundance within weeks was similar among all positions in 2013 (Fig. S4). With exception of the last week of June, sample position did not significantly influence thrips abundance (Table 2.1). By co ntrast, the edge position was almost always among the most abundant positions for weekly mean thrips captures in 2014. Within the crop, sample position significantly influenced thrips abundance in weeks 4, 6, and 9, although the differences were small (Fig . S2 , Table 2.1). Mean margin abundance was 73 -94% greater than the other positions during the first half of the season, but often statistically indistinguishable from the others during the second half (Figs. 2.2D & S2 , Table 2.1). By and large, thrips wer e evenly distributed across all positions during the second half of the season (Fig s. 2.1D & S2 ). Potato leafhopper . Potato leafhoppers (PLH) were the 3 rd most abundant pest group captured in both years, accounting for 5.7% and 6.1% of all counted herbivores in 2013 and 2014, 28 respectively (Table S1 ). A total of 3,582 and 6,524 adult PLH were captured on yellow sticky traps in 2013 and 2014, respectively. Potato leafhopper abundance was significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=117.83, P<0.001). In both ye ars, mean PLH abundance peaked in late June and early July; there was an additional peak at the end of the 2013 season (Fig. 2.2G -H & Fig. S3). Seasonal abundance patterns : Overall, in 2013, PLH were evenly distributed across all locations within the crop , but were significantly less abundant in the margins (Fig. 2.5G). In total, ~46% more PLH were captured on the field edge than at the other two crop locations and ~71% more PLH were captured at the edge than in the margins. The overall influence of sampli ng position was statistically significant (S: "2(3)=26.00, P<0.001), as well as the interaction with time (SxT: "2(36)=90.081, P<0.001, Figs. 2.5G & Fig. S3). Mean PLH captures were ~15% greater at the 10m and 25m positions and 65% greater at the edge posi tion than they were in the margins (Figs. 2.2G & 2.5G). In 2014, the capture totals for the margins and field edges were the greatest out of all sampling positions as well as being nearly identical. In total, ~32% and ~62% more PLH were captured at either the edge or the margin positions than at the 10m and 25m positions, respectively. The overall influence of sampling position was statistically significant (S: "2(3)=45.30, P<0.001), but the interaction with time was not (SxT: "2(33)=27.09, P=0.756, Figs. 2.5H & Fig. S3). Weekly abundance patterns : In all but the last three weeks of the season, PLH were evenly distributed across all three positions within the crop in 2013 (Fig. S3, Table 2.1). During the final two weeks, mean PLH abundance increased notabl y at the edge position, but remained stable at the other two interior positions. Weekly mean margin abundance was statistically 29 indistinguishable from the edge for seven out of the 12 weeks. In weeks 4,5,6, and 13, however, there were statistical differenc es in the number of PLH caught between margin and the field edge, ~1m apart. In total, sampling position significantly influenced PLH abundance in 8 out of the twelve weeks (Fig. 2.2G, Table 2.1). Potato leafhoppers were, for the most part, evenly distri buted for the first month of the 2014 growing season. The trend during the six weeks following peak abundance was edge -oriented, with abundance decreasing with distance into the field (Fig. S3). The trend was similar in the last two weeks, but the differen ces were not statistically significant (Table 2.1). As previously mentioned, mean and total abundance for PLH in the margin and edge positions were nearly identical in 2014 (Figs. 2.2H & S3). Aster leafhopper . Aster leafhoppers (ALH) were the 4 th most abu ndant pest group captured in both years, accounting for 3.5% and 3.9% of all counted herbivores in 2013 and 2014, respectively (Table S1 ). A total of 2,190 and 4,174 adult ALH were captured on yellow sticky traps in 2013 and 2014, respectively. Aster leafhopper abundance was significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=20.42, P<0.001). In both years, ALH numbers were highest during the first month of the crop, steadily declined during the second month, and rem ained relatively low during the final month ( Fig. 2.2 E-F & Fig. S4). Seasonal abundance patterns: Overall in 2013, ALH displayed an even distribution pattern across all sampling locations (Fig. 2.5 E). In total, 14.5% more ALH were captured on the edge of the crop than at any other sampling position. The overall influence of sampling position was not statistically significant (S : "2(3)=5.81, P=0.121, Fig. 2.2E ), and neither was the interaction with ti me (Sx T: "2(36)=48.91, P=0.071). Overall in 2014, ALH displayed an even distribution 30 across all locations within the crop, but were significantly less abundant in the margins (Fig. 2.5F). In total, ~22% more ALH were captured on the edge than at the other two field locations and ~102% mor e were captured at the edge than in the margins. The overall influence of sampling position was statistically significant (S: "2(3)=9.98, P=0.019 ), as well as th e interaction with time (SxT : "2(33)=81.88, P<0.001, Figs. 2.2F & S4). Mean ALH captures were ~60% greater at the 10m and 25m positions and ~90% greater at the edge position than t hey were in the margins (Figs. 2.5F ). Weekly abundance patterns: The edge position was always among the most abundant positions for weekly mean ALH captures in 2013. With in the crop, sample position significantly influenced ALH abundance in weeks 1, 6, 11, and 13, although the diff erences were often small (Fig. S2 , Table 2.1). Mean margin abundance was often similar to the edge position, but was significantly sma ller in we eks 3 & 4 (Fig. S4 , Table 2.1). In total, sampling position significantly influenced ALH abundance 6 out of 12 analyzable weeks in 2013 (Table 2.1). In all but two weeks of the season (weeks 9 & 11) , aster leafhoppers were evenly distributed across all th ree positions within the crop in 2014 (Fig. S4 ). In three of the first four weeks, when ALH were most abundant, mean abundance in the margins w as significant ly less than in the crop (Fig. 2.2F & S 4). During the second and third months, however, mean ALH abundance in the margins was usually not significantly different than any other sampling position (Fig. S4 , Table 2.1). Tarnish ed plant b ug. Tarnished plant bugs (TPB) were the 10 th most abundant pest group captured in 2013 and the 7 th most abundant pest group captured in 2014, accounting for 1% and 0.6% of all counted herbivores in 2013 and 2014, respectively (Table S1 ). They are included in 31 this analysis due to their historical importance as a polyphagous crop pest . A total of 210 and 742 adult TPB were captured on yellow sticky traps in 2013 and 2014, respectively. T arnished plant bug abundance was significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=142.4, P<0.001). In both years, TPB numbers were lowest during the first month of the crop, increased slightly during the second month, and were highest during the last month (Fig. 2.2I-J & Fig. S5 ). Cap ture numbers were low during 2013 and had to be pooled by mont h for analysis. Seasonal abundance patterns: Overall in 2013, TPB displayed an even distribution pattern across all sampling locations (Fig. 2.5 I). The overall influence of sampling position was not statistically significant (S: "2(3)=0.46, P=0.927); neither was the interaction with time (SxT : "2(36)=44.91, P=0.147). Overall in 2014, TPB were evenly distributed across all sampling locations within the crop, but significantly more abundant in the margins (Fig. 2.5 J). In total, 64.9%, 81.6%, and 133 % more TP B were captured in the margins than at the edge, 10m, and 25m positions, respectively. The overall influence of sampling position was statistically significant (S: "2(3)=31.30, P<0.001), but the interaction with time was not ( SxT: "2(33)=44.06, P=0.095, Figs. 2.5J & Fig. S5). Weekly abundance patterns : Due to low numbers in 2013 , weekly TPB captures were pooled across months for analysis. The influence of sampling position did not influence abundance during any month of 2013 (July: "2(3)=3.04, P=0.386; Au gust: "2(3)=4.07, P=0.254, Tab le 2.1). Twice as many individuals were captured at the interior field positions for the month of July (Fig 2.1I). In August, the pattern was reversed; more TPB we re captured in the margins and edge than in the interior field positions (Fig. 2.1I). 32 In all but one week of the season, tarnished plant bugs were evenly distributed across all three positions within the crop in 2014 (Figs. 2.1J, S5). The margin position was always among the most abundant positions for weekly TPB captures. During the first five weeks, there were no differences in mean abundance between the margin and all other positions. The following seven weeks, as overall numbers began to increase, TPB were usually more abundant in the margins than the other pos itions (Fig s. 2.1J, S5). Sampling position was a significant predictor of TPB abundance during weeks 6 -9 and had a marginally sig nificant effect during week 11 (Table 2.1, Fig. S5). Spatiotemporal distribution of natural enemies. Spiders . Spiders were the most abundant predator, accounting for 24.5% and 26.3% of all captured predators in 2013 and 2014 (Table S1 ). A total of 747 and 1,088 spiders were captured on yellow sticky traps in 2013 and 2014, respectively. Spider abundance was significantly influenc ed by year (multilevel mixed -model linear regression, year: "2(1)=28.37, P<0.001). In both years, mean spider numbers were relatively low during the first month of the season, but started to increase around the beginning of July (Fig 2 .3A-B & Fig. S6 ). Ab undance peaked in the third week of July in 2013 and began to steadily decrease and then stabilize at lower numbers around the second week of August (Fig 2.3) . In 2014, there was a relatively large amount of variation in abundance within and between weeks, especially after the first month. The trend was generally upwards starting in early July and reached a dynamic period of peak abundance between weeks 7 -11 and then sharply decreased in week 12 (Fig. 2 .3B & Fig. S6 ). Seasonal abundance patterns: Overall i n 2013, spiders displayed an even distribution pattern across all sampling locations (Fig. 2.6 A). In total, 52%, 74.5%, and 55.7% more spiders were 33 captured at the edge position than at the margin, 10m, and 25m positions, respectively. Season al mean spider captures were 46.9%, 72%, and 52.4% greater on the edges than they were at the margin, 10m, and 25m positions, respectively (Fig. 2 .6A & Fig. S6 ). However, the overall influence of sampling position was not statistically significant (S: "2(3)=4.79, P=0.18 8); neither was the overall interaction with time ( SxT: "2(36)=20.80, P=0.980). Likewise, in 2014, spiders were evenly distributed overall (Fig 2.6 B). In total, 44.6%, 48.9%, and 34.1% more spiders were captured at the 25m position in than at the margin, edge, and 10m positions, respectively. Although the influence of sampling position was statistically significant for the season (S: "2(3)=9.43, P=0.024 ), as well as the interaction with time ( SxT: "2(33)=52.51, P=0.017 ; Figs. 2 .3B & S6 ), the differences between the overall means were not statistically separable with post hoc TukeyÕs HSD pair -wise comparison ( Fig. 2.6 B). Weekly abundance patterns : In all but the week of peak abundance in 2013, spiders were evenly distributed across all sampling loca tions (Fig. S6 ). With the exception of the third week of July, sample position did not significantly influence spider abundance (Table 2.2). The differences during that week were relatively high, wit h mean edge abundance 87.5-173% greater than the other position s (Figs. 2.3A & S6). In contrast to 2013, the 25m position was among the most abundant positions for weekly mean spider captures in 2014 (Figs. 2.3B & S6 ). Sample position influenced spider abundance in weeks 5, 10, and 11; spiders were evenly distributed across positions the other nine weeks of 2014 (Table 2.2). Coleomegilla maculata . Of the coccinellids captured, Coleomegilla maculata (CMAC) was the most abundant, accounting for 82.8% and 52.1% of captured lady beetles in 2013 and 2014, respectively (Table S1 ). A total of 347 and 263 adult CMAC were captured on yellow sticky traps in 2013 and 2014, respectively. CMAC abundance was not significantly influenced by year 34 (multilevel mixed -model linear regression, year: "2(1)=1.25, P=0.269). In both years, CMAC numbers were highest during the first month of the crop and then decreased as time progressed (Fig. 2 .3C-D). Because capture numbers were low for both seasons, data were subset and analyzed by month. Seasonal abundance patterns: Overall in 2013, CMAC displayed an even distribution pattern across all sampling locations (Fig. 2.6 C). Seasonal mean CMAC captures were 34.4%, 22.9%, and 17.8% greater in the margins than they were at the edge, 10m, and 25m positions, respectively (Fig. 2C). However, the over all influence of sampling position was not statistically significant (S: "2(3)=3.31, P=0.346 ); neither was the overall interaction with time ( SxT: "2(36)=36.65, P=0.439 ). Overall in 2014, CMAC were evenly distributed ov er all sampling positions (Fig 2.6 D). Total capture was greatest at the edge and 25m positions; the totals were very similar (101 & 103, respectively) In total, #44% and #15% more CMAC were captured at the edge or 25m positions than at the margin and 10m positions in 2014. The overall infl uence of sampling position was not statistically significant (S: "2(3)=2.33, P=0.506); neither was the overall interaction with time ( SxT: "2(33)=20.53, P=0.956). Sampling position did not infl uence abundance for any month during 2014 (Fig 2.6C -D, Table 2. 2). Syrphidae . Hover flies were the 6 th most abundant predatory group captured in 2013 and the 9th most abundant in 2014 (Table S1 ). They are included here due to their historical and ecological importance as highly -mobile cosmopolitan generalist predators. A total of 288 and 127 adult syrphids were captured on yellow sticky traps in 2013 and 2014, respectively. Syrphid abundance was significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=34.37, P<0.001). In both years, syrphid numbers were low in early June, peaked in July 35 and slowly decline d over the month of August (Fig. 2 .6E-F). Because capture numbers were low for both sea sons, data were subset and analyzed by month. Seasonal abundance patterns: Overall in 2013, syrphids displayed an even distribution across all sampling locations ( Fig. 2.6 E). Season mean syrphid captures were 40.4%, 71.2%, and 78% greater in the margins th an they were at the edge, 10m, and 25m positions, respectively (Fig. 2.6E). The influen ce of sampling position was statistically significant (S: "2(3)=16.89, P<0.001 ) for the season ; but the interaction with time was not (SxT: "2(36)=25.60, P=0.901). Conversely , sampling position did not significantly influence abundance during any month in 2013 (Fig. 2.3E, Table 2.2). Overall in 2014, syrphids displayed an even distribution across all locations within the crop, but were significantly more abundant i n the margins than in the interior field positions (Fig. 2.6F). Seasonal mean syrphid captures were 75.6%, 153%, and 178% greater in the margins than they were at the edge, 10m, and 25m positions, respectively (Fig. 2 .6F). T he overall influence of sampling position was statistically significant (S: "2(3)=17.1, P<0.001), but the interaction with time was not (SxT: "2(33)=21.71, P=0.934, Fig . 2.3F). The differences between the overall means were statistically separable with post hoc Tuk eyÕs HSD pair -wise comparison ( !=0.05; Fig. 2.6 F). Furthermore, sampling position was found to influence abundance in July and August of 2014 (Fig. 2 .3F, Table 2.2). Chalcidoidea . Wasps in the superfamily Chalcidoidea were the most abundant group of parasitoids captured in both seasons, accounting for 46.9% and 54.3% of all captured parasitoids in 2013 and 2014, respectively (Table S1 ). A total of 1,958 and 3,581 chalcidoid wasps were captured on yellow sticky traps in 2013 and 2014, respectively. Chalcidoid abundance was 36 significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=130.32, P<0.001). In both years, mean chalcidoid numbers dipped considerably in the second week compared to the first week, then increased continuously until their first peak perio d in second week of July (Fig 2.4 A-B). Populations in both years dipped again and then began to increase at a relatively high rate around the third week of August and continued in that fashion until the crop was harvested. Seasonal abundance patterns: Overall in 2013, chalcidoid wasps displayed an even distribution across all locations within the crop, but were significantly more abundant in the margins (Fig. 2.7A). Total mean chalcidoid margin captures were greater than those at the edge, 10m, a nd 25m positions by factors of 2.17, 2.73, and 2.85, respectively (Fig 2.7 A). The overall influence of sampling position was statistically significant (S: "2(3)=86.1, P<0.001), as well as the interaction with time ( SxT: "2(36)=54.50, P=0.025 , Figs. 2.4A & S7). Overall in 2014, chalcidoid wasps were most abundant in the margins, and far less abundant at the field positions (Fig. 2.7 B). Total mean chalcidoid margin captures were greater than those at the edge, 10m, and 25m positions by factors of 2.93 5.51, and 7.47, respectively (Figs. 2.4B & 2.7 B). The overall influence of sampling position was statistically significant (S: "2(3)=352.8, P<0.001), as well as the interaction with time ( SxT: "2(33)=66.25, P<0.001, Figs. 2.7B & S7). Weekly abundance patterns : The margin position was always among the most abundant positions for weekly mean chalcidoid captures in 2013 (Fig. 15). In seven out of the 13 weeks, chalcidoids were distributed evenly across all three positions within the crop. The differences in abundance in the other six weeks, while statistically significant, were often small (Table 2.2, Fig. 37 S7). In total, sampling position significantly influenced chalcidoid abundance in all but one week (week 8) in 2013 (Table 2.2 ). Likewise, on a weekly basis, chalcidoids were always most abundant in the margins in 2014. During the first five weeks, chalcidoid abundance was even ly distributed across all three field locations. However, during wee ks 6-12, abundance at the field edge was usually significantly highe r than in the interior positions (10m & 25m). In total, sampling position significantly influenced chalcidoid abundance every week in 2014 (Table 2.2). Mymaridae . Mymarid wasps were the second most abundant group of parasitoids captured in both seasons, a ccounting for 22.6% and 24.1% of all captured parasitoids in 2013 and 2014, respectively (Table S1 ). A total of 946 and 1591 mymarids were captured on yellow sticky traps in 2013 and 2014, respectively. Mymarid abundance was significantly influenced by yea r (multilevel mixed -model linear regression, year: "2(1)=124.06, P<0.001). No mymarids were found the first three weeks of 2013. Starting in week four, there was considerable vari ation from week to week in 2013 (Fig 2 .4C). Although peak numbers were reached in week 10, they were also relatively high in weeks 5, 7, and 13. By comparison, weekly numbers were relatively stable and consistent in 2014. Although mean mymarid abundance peaked in week 9, those numbers were not considerably larger than many o ther weeks of the season (Fig. 2.5 D). Seasonal abundance patterns: Overall in 2013, mymarid wasps displayed an even distribution across all locations within the crop, but were significantly more abundant in the margins (Fig. 6C). Total mean mymarid margin captures were greater than those at the edge, 10m, and 25m positions by factors of 1.96, 2.3, and 2.82, respectively (Figs. 2.4C & 2.7 C). The overall 38 influence of sampling position was statistically significant (S: "2(3)=80.89, P<0.001), as well as the int eraction with time ( SxT: "2(36)=80.65, P<0.001, Figs . 2.4C & S8 ). Overall in 2014, mymarid wasps were once again most abundant in the margins, and evenly distributed across all three field positions (Fig. 2.7 D). Total mean mymarid margin captures were gr eater than those at the edge, 10m, and 25m positions by factors of 1.77, 2.34, and 2.32, respectively (Figs. 2.4D & 2.7 D). The overall influence of sampling position was statistically significant (S: "2(3)=86.85, P<0.001), but the interaction with time was not (SxT: "2(33)=42.06, P=0.1, Fig. 2.4D & Fig. S8 ). Weekly abundance patterns : The margin position was always among the most abundant positions for weekly mean m ymarid captures in 2013 (Fig. S8 ). Mymarids were distributed evenly across all three positions within the field in three out of ten weeks. The differences in abundance in the other seven weeks, while statistically significant, were often small (Table 2.2, Fig. S8 ). In total, sampling position significantly influenced mymarid abundance in all but two weeks (weeks 4 & 11) in 2013 (Table 2.2). As in the previous year, mymarids were always most abundant in the margins on a weekly basis in 2014. In six out of the 12 weeks, mymarids were distributed evenly across all three positions within the field . Abundance patterns during the other six weeks were usually edge-centric (Table 2.2, Fig. S8). In total, sampling position significantly influenced mymarid abundance eight out of 12 weeks in 2014 (Table 2.2). Trichogrammatidae . Trichogrammatid wasps were t he 4th most abundant group of parasitoids captured in both seasons, accounting for 4.4% and 6.8% of all captured parasitoids in 2013 and 2014, respectively (Table S1 ). A total of 185 and 448 trichogrammatid wasps were captured on yellow sticky traps in 2013 and 2014, respectively. Trichogrammatid abundance was 39 significantly influenced by year (multilevel mixed -model linear regression, year: "2(1)=95.72, P<0.001). In both years, wasp numbers were lowest in the first month and increased continuously over the following two months (Fig 2.4 E-F). Because capture numbers were low for both seasons, data were subset and analyzed by month. Seasonal abundance patterns: Overall in 2013, trichogrammatids had an edge -centric distribution pattern (Fig. 2.7 E) with mean captures being 9.25%, 29.2%, and 102 % greater in the margins than they were at the edge, 10m, and 25m positions, respectively (Figs. 2.4E & 2.7 E). The overall influen ce of sampling position was statistically significant (S: "2(3)=7.96, P=0.047) as well as interaction with time ( SxT: "2(36)=56.03, P=0.018 ). When subset by month , sampling position did not influence abundance for eith er month in 2013 (Fig 2.4 E, Table 2.2). Overall in 2014, trichogrammatid wasps were most abundant in the margins and decreased in abundance with increasing distance into the field (Fig. 2.7 F). Season mean trichogrammatid captu res were 22.5%, 90.4%, and 131 % greater in the margins than they were at the edge, 10m, and 25m positions, respectively (Figs. 2.4E & 2.7 E). The overall influence of sampling position was statistically significant (S: "2(3)=29.06, P<0.001 ), but the interaction with time was not ( SxT: "2(33)=28.20, P=0.705 , Figs. 3E). When su bset b y month , sampling position was found to influence abundance in July and August of 2014 (Fig. 2.4F , Table 2.2). Braconidae . Braconid wasps were the 5 th most abundant group of parasitoids captured in 2013 and the 3rd most abundant group captured in 2014, accounting for 4.4% and 7.9% of all captured parasitoids in their respective years (Table S1 ). A total of 184 and 520 braconids were captured on yellow sticky traps in 2013 and 2014, respectively. Braconid abundance was significantly influenced by ye ar (multilevel mixed -model linear regression, year: "2(1)=166.9, P<0.001). In 40 both years, wasp numbers were relatively stable across the whole season with peaks appearing in early-mid July and late August (Fig 2.4 G-H). Because capture numbers were low in 2 013, data were subset and analyzed by month for that year. Seasonal abundance patterns: Overall in 2013, braconid wasps displayed an even distribution across all locations within the crop, but were significantly more abundant in the margins (Fig. 2.7G). Season mean braconid margin captures were greater than those at the edge, 10m, and 25m positions by factors of 1.89, 3.66, and 3.26, respectively (Figs. 2.4G & 2.7 G). The overall influence of sampling position was statistically significant (S: "2(3)=37.71, P<0.001), as well as the interaction with time ( SxT: "2(36)=56.03, P=0.018). When subset by month , sampling position was found to influence abundance for all three months of 2013 (Fig. 2.4 G, Table 2.2). Overall in 2014, braconid wasps were mo st abundant in the margins and displayed a slight edge-centric pattern in the field (Fig. 2.7 H). Season mean braconid margin captures were greater than those at the edge, 10m, and 25m positions by factors of 2.37, 3.54, and 3.77, respectively (Figs. 2.4H & 2. 7H). The overall influence of sampling position was statistically significant (S: "2(3)=130.4, P<0.001), but the interaction with time was not ( SxT: "2(33)=37.78, P=0.260, Fig. 2.4H & Fig. S9 ). Weekly abundance patterns : The margin position was always am ong the most abundant positions for weekly mean br aconid captures in 2014 (Fig. S9 ). Braconid wasps were distributed evenly across all three positions within the crop in nine out of twelve weeks. Abundance patterns during the other three weeks were edge -centric (Table 2.2, Fig. S9 ). In total, sampling position significantly influenced braconid abundance in eight weeks in 2014 (Table 2.2). 41 Table 2.1. Significance of fixed effect (sample position) on weekly or monthly herbivore abundance in commercial celery fields in southwest Michigan during 2013 and 2014. ALH - aster leafhopper, PLH - potato leafhopper, TPB - tarnished plant bug. Taxon 20132014Taxon 20132014AphidweekdfChisqPweekdfChisqPThripsweekdfChisqPweekdfChisqP1340.84<0.001133.2120.36013196.6<0.001137.0620.0702335.71<0.0012312.840.005234.2850.2322320.62<0.0013326.89<0.0013325.19<0.001335.760.1243332.27<0.0014329.89<0.0014313.170.004439.340.0254315.520.0015325.2<0.0015345.94<0.0015364.37<0.0015327.49<0.001635.510.1386335.73<0.001630.8860.8296312.260.0077310.490.0157326.76<0.001731.9570.581734.8660.1828310.940.0128312.80.005830.6620.882832.1940.533936.610.085933.8380.279935.6340.131939.8130.0201034.9060.1791034.3520.2261035.5440.1361035.490.1391133.3220.3451137.0940.0691136.2520.11137.2390.0651234.9620.17512312.770.0051231.8080.6131236.1460.1051334.7460.1911333.320.345ALHweekdfChisqPweekdfChisqPPLHweekdfChisqPweekdfChisqP139.0090.0291323.93<0.0011NA130.9420.815236.5760.091238.6560.0342316.62<0.0012310.270.0163312.390.006332.2160.529334.8620.182332.2960.513438.0640.0454322.33<0.0014321.26<0.001436.390.094532.0460.5635310.920.0125350.25<0.0015316.040.0016312.250.007634.0760.2536334.29<0.0016319.66<0.001738.9440.030733.5060.320732.2480.5237312.140.007832.3490.503834.7330.192834.540.2098323<0.001933.7590.2899315.650.001933.9940.262939.3280.0251033.7860.2861036.710.08210313.140.00410310.530.0151139.2570.0261138.230.0421139.4620.0241137.250.06412NA12311.07 0.011 12318.643E-041236.1040.10713315.640.00113316.180.001TPBmonthdfChisqPweekdfChisqP1NA1NA233.0360.3862NA334.0660.254330.8730.832434.6790.197534.7190.1946319.51<0.0017316.51<0.0018326.17<0.0019320.36<0.0011034.0880.2521137.7640.0511231.5420.673 42 Table 2.2. Significance of fixed effect (sample position) on weekly or monthly natural enemy abundance in commercial celery fields in southwest Michigan during 2013 and 2014. CMAC - Coleomegilla maculata . Taxon 20132014Taxon 20132014CMACmonthdfChisqPmonthdfChisqPSyrphidaemonthdfChisqPmonthdfChisqP133.9580.266131.6740.643136.5160.0891353.120.150230.8660.834235.7740.123231.3260.7232310.330.016330.0980.992330.280.964332.6440.4503310.270.016AranaeweekdfChisqPweekdfChisqPChalcidoideaweekdfChisqPweekdfChisqP1NA133.9630.2651324.55<0.0011337.17<0.001231.3590.715236.8130.0782319.72<0.0012321.39<0.001334.5430.208337.8580.0493330.4<0.0013350.3<0.001434.5250.210431.470.6894311.15 0.011 4344.95<0.001533.2360.3575311.77 0.0085323.14<0.0015335.94<0.001634.7380.192635.6310.1316321.77<0.0016340.5<0.0017314.150.003734.3660.2257320.1<0.0017317.31<0.001833.7760.287831.1320.769834.980.1738323.2<0.001936.0140.111 932.6390.4519310.90.0129343.03<0.0011033.2190.35910318.340.0001039.4520.02410358.26<0.0011131.3130.7261138.8580.03111355.5<0.00111348.57<0.0011231.9650.5801236.4810.09012347.58<0.00112355.41<0.0011333.1980.362133111.2 <0.001MymaridaeweekdfChisqPweekdfChisqPBraconidaemonthdfChisqPweekdfChisqP1NA1312.020.0071321<0.0011346.08<0.0012NA2320.09<0.0012320.7<0.0012345.73<0.0013NA3339.07<0.001337.8580.049336.9240.074437.0620.070436.2920.0984343.59<0.0015330.53<0.0015361.29<0.0015329.8<0.0016318.56<0.0016316.250.0016321.64<0.0017330.75<0.001737.9260.048734.3470.226839.8140.020836.350.0968311.79 0.0089317.85<0.0019313.450.0049311.01 0.01210314.690.00210314.190.0031035.6320.1311133.6750.29911311.5 0.00911319.74<0.00112310.740.01312310.250.0171230.3520.95013318.56<0.001Trichogramm. monthdfChisqPmonthdfChisqP1NA133.7980.284232.920.404239.0120.029337.0940.0683319.720.000 43 Figure 2.2. Weekly or monthly m ean abundance (±SEM) of herbivores captured in the celery canopy with yellow sticky traps at different distances into the field for 2013 and 2014. Asterisks indicate significant differences among sampling positions within a particular week ( ! = 0.05). ALH - aster leafhopper, PLH - potato leafhopper, TPB - tarnished plant bug. 44 Figure 2.3. Weekly or monthly mean abundance ( ±SEM) of predators captured in the celery canopy with yellow sticky traps at different distances into the field for 2013 and 2014. Asterisks indicate significant differences among sampling positions within a particular week ( ! = 0.05). CMAC - Coleomegilla maculata . 45 Figure 2.4. Weekly mean abundance ( ±SEM) of parasitoids captured in the celery canopy with yellow sticky traps at different distances into the field for 2013 and 2014. Asterisks indicate significant differences among sampling positions within a particular week ( ! = 0.05). 46 Figure 2.5. Overall (seasonal) mean abundance ( ±SEM) of herbivores at different distances into the field for 2013 and 2014 seasons. Bars with the same letter within the same year are not significantly different (TukeyÕs HSD; ! = 0.05). ALH - asterleafhopper, PLH - potato leafhopper, TPB - tarnished plant bug. 47 Figure 2.6. Overall (seasonal) mean abundance ( ±SEM) of predators at different distances into the field for 2013 and 2014 seasons. Bars with the same letter within the same year are not significantly differ ent (TukeyÕs HSD; ! = 0.05). CMAC - Coleomegilla maculata . 48 Figure 2.7. Overall (seasonal) mean abundance ( ±SEM) of parasitoids at different distances into the field for 2013 and 2014 seasons. Bars with the same letter within the same year are not significantly different (TukeyÕs HSD; ! = 0.05). 49 Discussion In general, most captured arthropods were distributed evenly across the crop and were more abundant in the margins just outside the field. Edge effects were often detected when abundance was low, but usually disappear ed when abundance was near the seasonal peak. Furthermore, persistence of these patterns across the season vari ed according to group. Aphids. Aphid abundance wa s greater on the edges and margins of celery fields during the first half of the season in both years when relative abundance wa s low. Wh en abundance wa s higher, later in the season s, aphids were evenly distributed throughout the field. Orga nisms across a broad taxonomic range have been shown to be able to detect and respond to habitat edges (Haddad 1999, Haddad et al. 2003). Herbivorous pests have the tendency to aggregate initially on the edges of agric ultural fields (van Helden 2010). The suggested reason for this behavior is that the pest first locates the crop (habitat location) from the air, drops below the flight boundary layer and proceeds to locate a host plant (Taylor 1974 , van Helden 2010 ). The host location process usually results in the insect encountering the edge of the crop and then stopping. Michigan celery fields often have a sharp crop-soil interface (Fig. 1.4), which may influence host location and aggregation in certain pest groups. Aph ids have been found to aggregate at the edges of other crops after insecticide application, but may only be detectable at scales larger than 25m (Winder et al. 1999). For example, Winder et al. (1999) concluded that the edge effects for Sitobion avenae in winter wh eat extended from 30 to 60 meters into the field with constant density throughout. Because the margins are not direc tly sprayed with insecticides, these areas may function as small refu ge strips for arthropods. A phids (and many other groups) that I caught on sticky traps likely came from two places: margins or the aerial plankton pool. Aerial plankton 50 is the collective name for the pool of tiny organisms that fly or drift on atmospheric wind currents and include small light -bodied arthropods like spiders and aphids. I expected to catch more aphids at the field edge at the early stage s of re-colonization due to normal foraging/host -location activity of surviving herbivores from the margins. If the w ind -borne aphids are low in abundance, then foraging patterns from the margin may be detectable, even at such a small scale. However, when aphid abundance grows exponentially later in the season, the sheer number of wind -transported aphids will easily mask small edge effect s from residual aphid popul ations left in the margins after the most recent insecticide application. Thus, it is likely that the vast majority of aphids caught in the celery canopy have not come directly from the margins. Thrips. Thrips abundance was greatest from late June to mid -July in both 2013 and 2014. Thrips were found in relatively consistent numbers across all field positions in both years. In 2014, a minor negative edge effect was seen in the spatial distribution plot (Fig. S4) which shows more thrips were caught in the c enter of the field than at the edges (Figure S4). A possible explanation for this is that the center of the fields may experience reduced natural enemy activity due to population depletion from insecticide application. Overall, margin abundance was often h ighly similar to that found in the celery field itself. Potato leafhopper. Potato leafhoppers were more abundant during the first half of the season with peak numbers usually occurring in late June or early July. Comparatively, this occurs at approximate ly 1 -2 weeks later than the aster leafhopper. Although it is not cited as a disease vector, the stereotypical stunting and necrosis caused by PLH feeding (ÒhopperburnÓ) can cause 51 tremendous damage to a field of young transplants (Capinera 2001). For this r eason, chemical management of leafhoppers in newly -transplanted fields should extend into early July. The within -field distribution of potato leafhoppers was fairly consistent between the two years. In general, more PLH were found on the edges of the cele ry field than within the interior, although they were often evenly distributed within fields when numbers were relatively high (Fig. S3). This pattern is consistent with other studies in alfalfa (Emmen et al. 2004) and soybeans (Poston and Pedigo 1975), al though the fields of those crops are typically much larger than celery fields. Abundance in the margins, relative to the field locations, was not consistent between 2013 and 2014. PLH have been shown to develop successfully on pigweed ( Amaranthus spp.) and smartweed ( Polygonum spp.) (Capinera 2001), both of which grow readily in celery field margins. Consequently, a reasonable explanation for margin -centric abundance patterns may be directly linked to the presence and vegetative growth stage of preferred we ed hosts outside the celery field. Aster leafhopper. Aster leafhoppers we re evenly distributed across the celery crop in both 2013 and 2014. Similarly to aphids, a faint edge effect was detected when relative abundance wa s low, but this was not observed when ab undance wa s high. Un like the aphids, ALH abundance wa s often lower in the margins during periods of peak abundance. One explanation for this is that margins are often mowed or sprayed with herbicide around the time of planting , when seasonal ALH ab undance is at its peak . As a result, there are fewer resources to exploit in th ose areas. Oviposition site selection behavior may also factor into these patterns. Aster leafhoppers overwinter as eggs in grasses and weeds (Meade and Peterson, 1964), but s pring and summer generations m ay prefer to reproduce in the celery field (or other vegetable crop) where plants are 52 growing rapidly and the vascula r fluids are highly nutritious. Another potential explanation is that ALH may be disinclined to cross the bound ary of the field edge (van Helden 2010), especially if a sharp (bare soil) interface exists between the field and the margin. Tarnished plant b ug. Tarnished plant bug populations increase d with time and we re evenly distributed throughout celery fields dur ing both growing seasons. These insects are highly -mobile foragers that can easily disperse over 25m within seconds (J. Jubenville, pers. obs.). Margin abundance wa s usually similar to or greater than abundance within the field. T arnished plant bugs are true generalists and readily exploit weedy margins in other crops (Snodgrass et al. 1984, Young 1986, Capinera 2001, Snodgrass et al. 2005, Outward et al. 2008); margin management should be a priority in systems where TPB is a pest. Spiders. Spiders we re consistently the most abundant predator s found on canopy traps, although their distribution and t he abundance within the field was highly variable from week to week. Overall, they were evenly distributed across celery fields. The vast majority of the se spiders were immatures, and a lthough I was unable to indentify them to family, they were likely Linyphiid spiders, due to the great abundance of sheet webs consistently found throughout celery fields (J. Jubenville, pers. obs.). These spiders disperse in the winds by ballooning and are not thought to be capable of directed flight (Suter 1999) . As such, their distribution in the field should be strongly influenced by how the wind breaks on hedgerows and other tall objects (buildings, etc.) near the field. Due to their abundance, field penetration, persistence, and generalist diet, spiders are probably one of the most important predator s in these systems. Any possibility of reaping 53 natural pest control services from this group, however, is likely reduced by near-weekly insecticide applications to the field . Coleomegilla maculata . The most abundant coccinellid in celery systems is Coleomegilla maculata (CMAC). They can be found in similar abundance at any location in the field or margin. Because they experience the landscape on a scale much larger than 25m, an even distribution is to be expected. No edge effects were detected. Coleomegilla maculata is notable among coccinellids for a truly polyphagous diet. Gut content analysis has shown that they ingest pollen and fungal spores in addition to other arthropods (Webber and Lundgren 2009 ). One implication is that they can sustain themselves with alternative resources while arthropod prey are temporarily absent from agricultural fields. Conserving refuge habitat near the field could have a positive effect on CMAC abundance in the field (Landis et al. 2000). Syrphidae. Syrphid flies we re evenly distributed throughout the field. Like CMAC, syrphids are highly mobile generalist predators that likely experience the landscape at greater distances than the size of an average celery field in this study. Because of their great vagility, one might expect them to be equally abund ant across all locations. S yrphid abundance , however, was consistently greater in the margin s and decreased with distance into the field (Figure 2.6 ). One explan ation might be that margins are managed less frequently, harbor a greater diversity of flora, and likely serve as a refuge for alternative prey. Syrphid larvae are predatory , making the margin a preferable habitat for oviposition over celery fields that are often devoid of prey. Another potential explanation is that weeds in the margins and ditch b anks occasionally reach the 54 flowering stage. Because nectar is an important floral resource for adult syrphids, they likely prefer margins over flower -less celery fields (Hickman and Wratten 1996) . In celery margins, syrphid adults have frequently been seen feeding on the flowers of sma rtweed (J. Jubenville, pers. obs.). As with CMAC, it is likely that creating habitat with flowering resources near the field could have a positive effect on syphid abundance within these systems (Cowgill et al. 1993, Landis et al. 2000). Parasitoids. Edge effects, while variable across groups, we re apparent and persistent among parasitoids as a whole. Chalcidoid and braconid wasps were evenly distributed across the field in 2013, but clearly more abundant on the field edge in 2014. Mymarid wasps were evenly distributed in both years, but show ed a non-significant trend of decreasing abundance with increasing distance into the field. Trichogrammatid wasps displayed a margin -centric abundance pattern in both years, but this wa s more pronounced in 2014. In all cases, parasitoid abundance was notably greater in the margins than at any field location across the whole season. Dipteran species we re present in great abundance in these systems, especially near the ditches. It is lik ely that these species served as hosts to some of the chalcidoids that I captured. Because margins can be, to a certain extent, considered refuges from insecticides, populations residing in these areas are never depleted. This is could also be a reason for the sheer numerical dominance of Chalci doids , among parasitoids . If parasitoid populations we re maintained within the margins , then the edge effects I observed can be explained as a result of foraging behavior into the fields f rom these areas. In previous studies, parasitoids have been observed to remain in the general area of patches where they have had recent success in locating hosts (e.g. Waage 1978, 1979, Cronin 2009). More over, 55 the probability for the parasitoid returning to a patch seems to increase with exposure to a host chemical stimulus (Waage 1978). As margins are likely to be refuges for hosts, this might explain the persistent margin -centric abundance pattern. Longley et al. (1997) reported similar re-colonization patterns of parasi toids after insecticide treatment in winter wheat fields. The patterns they reported, however, were short -lived and disappeared after 5 days. Interestingly, they also reported that overall , primary parasitoid populations within the crop were below pre -spray levels 5 days after application with numbers near the field center still recovering after 12 days. Given that celery fields are sprayed with insecticides on a near -weekly basis, our results are quite similar to Longley et al. (1997) and could be explained as a product of the pest management regime. Because adult parasitoids derive their nourishment largely from nectar, enriching refuge areas with floral resources could have a positive effect on parasitoid longevity and abundance (Wackers 2001, 2004; Winkl er et al. 2009). Synthesis. Migration movement in insects has been defined as directed movement while temporar ily ignoring typical host location cues (Kennedy 1985). Even insect groups that are considered weak fliers, such as whiteflies, can travel several kilometers with directed flight upwind over a period of hours (Byrne 1999). In this context, the distance betwe en the edge and the center of many celery fields should not be difficult for most canopy arthropods to move across in a directed manner within a matter of seconds. As the results here show, however, different groups display d ifferent distribution patterns . Differences in distribution patterns may be indicative of differences in the source of these captured insects. A margin -centric pattern, one in which abundance is greatest in the margin and decreases with distance, suggests that the particular taxon is using the margin as habitat. Whereas a distribution pattern in which 56 abundance is uniform across all sampling locations (i.e. there is no central position of greatest abundance) carries no such suggestion. Most of the herbivore groups I sampled were evenly distributed across all four sampling location s. Small edge effects were detected in some groups when relative abundance was low, but the practical significance may be minimal . Leafhoppers were occasionally less abundant in the margins. This behavior may ha ve to do with host availability and habitat preference. Predators were also evenly distributed across all field locations. Of the predators, syrphid flies were the only group with significantly more individuals captured in the margins. Factors that likel y contribute to this behavior are the availability of alternative resources and refuge within the margins. On the whole, parasitoids were also more abundant in the margins. Mymarids were evenly distributed throughout the field, while braconid, chalcidoid, and trichogrammatid wasps displayed clear edge effects. As with the syrphids, alternative resources and refuge within the margins likely play a role in this pattern. Knowledge of spatial and temporal patterns of insect pests in agricultural fields can be useful for designing efficient scouting protocols, estimating pest densities, and developing management strategies that eliminate unnecessary pesticide application (Winder et al. 1999). My results suggest that sampling on the edge will generally capture th e same number of individuals as sampling from the center of most fields. However, there are many celery fields in southwest Michigan that are wi der than the fields we sampled and therefore may have different distribution patterns . Considering the small edge effects detected at 25m, I would expect that the effect would be more distinguishable in wider fields. The relationship between herbivores and the margins varies across groups. My results suggest that in some cases, the margins are used for resources and that these areas could serve as 57 a source for recolonizing the fields after insecticide application (e.g. tarnished plant bugs). Both leafhoppers (ALH, PLH) are known to feed and develop on weedy margin species (Wallis 1962, Capi nera 2001, Frost et al. 2011) commonly found in celery systems. With this in mind, it seems likely that leafhopper and tarnished plant bug populations would benefit from unmanaged margins and that current margin management practices limit their populations near the crop. Although current margin management practices dampen pest populations, they are also likely limiting the populations of important groups of natural enemies. Based on my results, it is likely that the Chalcidoidea, Mymaridae, Braconidae, Tri chogrammatidae, and Syrphidae would all benefit from conserving the vegetative structure within margins and ditch banks. These groups all have member species that are known to attack the suite of celery pests highlighted in this study and practices that pr omote stability and growth in their populations may benefit growers in the form of increased pest control (Landis et al. 2000). Nevertheless, while the idea of conserving and managing margins is intriguing, additional research is necessary to determine whe ther the net effect would be beneficial or simply a breeding ground for serious celery pests (Winkler et al. 2010). 58 CHAPTER 3: The movement of arthropods between margin and field in Michigan celery agroecosystems Introduction Land use within the modern agricultural landscape falls along a continuum of habitat stability, with areas of intense disturbance on one end and near -natural areas at the other. At the farm scale, crop fields are ephemeral habitats that experience frequent and intense disturbance while border areas are often managed less frequently and with varying levels of intensity. Pest management practices often eliminate arthropods within the field , but border areas are not usually managed in this way. The inequality in disturbance between adjacent areas may result in a source-sink dynamic with arthropods from border habitats re -colonizing the suddenly unoccupied field. Because local abundance can be influenced by factors at the landscape scale, it should not be assume d that margins are the primary source of insects within the field . Understanding the source of beneficial and pestiferous insects within a n agricultural production system is fundamental to selecting the most suitable management practices. Michigan celery is grown on muck soils, which occur naturally in lowland swampy areas. Growers will often construct and install a network of drainage ditches and tiling to lower the water table across the entire area of muck soil. A common result of this strategy is an aggregation of long narrow fields across the muck area, each bordered on two or three sides by drainage ditches. Ditches are bordered on both sides by weedy marginal areas that vary in width, but are usually about 2m wide and managed by mowing or herbicide a pplication. These marginal areas are reminiscent of Òbeetle banksÓ which provide a refuge for ground beetles and other arthropods from intense agricultural practices that might otherwise kill them (Landis et al. 2000). 59 Conserving beneficial arthropods alon g refuge strips may result in greater pest control and fewer pesticide applications (Landis et al. 2000). In many ways, the field -margin -ditch configuration in celery systems appears to be an ideal (albeit, unintentional), implementation of this concept. Many celery fields are narrow, with their centers approximately 25 meter s from the edge. Re -colonization of s uch narrow fields should be expected to happen quickly once the toxicity of pesticide residues begins to fade. In an earlier chapter I showed that arthropods fro m most taxonomic groups show little difference in abundance between the edge and the center of celery fields (CHAPTER 2). However , some of these groups are most abundant in the margins, indicating that border habitat in these systems could be serving as both a refuge and a source (CHAPTER 2). Ecological models predict herbivore abundance will be higher in the center than at the edge of large cereal fields due to predation pressure from the outside (Bianchi and van der Werf 2004). Empirical support of this prediction has been in shown in studies where natural enemy abundance is greater on the edge than the center, while herbivore abundance is greater in the center than the edge after insecticide applications (Duffield & Aebischer 19 94, Longley et al. 1997). This makes intuitive sense as predator and parasitoid populations generally track herbivore populations over time. Many celery fields, however, are considerably smaller than cereal fields (only 50m wide on average), thus natural e nemies should quickly track herbivores to the middle of the field. Furthermore, if margins in these systems are managed for the conservation of arthropods, a n increase in certain natural enemy populations could result in elevated natural pest control. In combination with accurate population estimation, economic thresholds , and regular scouting, this would be the foundation of an integrated pest management 60 strategy to reduce the ecological impact of intensive farming while maintaining economic viability. An undesirable outcome, however, is that herbivore populations will also increase as a result of conserving margin vegetation. The presence of alternative prey is generally viewed as a necessity for maintaining stability in local natural enemy populations whenever prey within the crop disappear as a result of insecticide application (Altieri 1999, Landis et al. 2000). All of the aforementioned pests (CHAPTER 2) are known to be polyphagous, some with a diet ary niche ranging across hundreds of plant species ( Wallis 1962, Young 1986, Capinera 2001, Blackman & Eastop 2007, Frost et al. 2011). In particular, the tarnished plant bug has a documented reputation as a Òborder speciesÓ that readily inhabits weedy margins (Snodgrass et al. 1984, Young 1986, Capinera 20 01, Snodgrass et al. 2005, Outward et al. 2008). Evaluating the extent to which arthropod species interact with the margins, then, is a necessary first step in assessing the potential viability of a margin conservation strategy in Michigan celery . A rel ated issue in this regard is whether natural enemies found within non -crop habitats move into the crop to forage for prey or hosts. The movement patterns of natural enemies between crop and non -crop habitat is often poorly understood and results in a gener al inability to predict and evaluate the efficacy of habitat management for improved pest control (Lavendero et al. 2004, Synder et al. 2005, but see Schellhorn et al. 2014). Therefore, it has been recommended, that future studies examine, among other thin gs, the movement patterns of insects between focal areas in addition to their overall abundance (Synder et al. 2005). Recently, an improved method for marking large areas in the environment has been shown to be a viable method for studying movement patterns in a broad range of insects. Immunomarking techniques using crude proteins (Jones et al. 2006, Hagler and Jones 2010, 61 Jones et al. 2011) are highly sensitive, relatively inexpensive, flexible, and appropriate for use in a wide range of applicatio ns. Insects can acquire the protein markers directly by being sprayed or indirectly by walking across a substrate with the protein residue. Insects can subsequently be captured by sticky traps and tested for the presence of the protein with an enzyme -linke d immunosorbent assay (ELISA). I used immunomarkers to ask whether insects in celery systems interact with the margins and whether they move into the field from the margins. Specifically, I (1) measure d the abundance and spatial distribution of arthropods in the fields and margins in Michigan commercial celery, (2) estimate d the degree to which the arthropod community interacts with the margin, (3) assess ed the movement of the arthropod community from the margin to the field , and (4) determine d whether mean relative abundance correlates with positive mark percentage. Materials and Methods Study sites. In the 2013 and 2014 growing seasons, I selected three Michigan commercial celery farms based on large production volume, geographic separation, and grower cooperation . Farms were located in Allegan and Van Buren counties with a geographic al separation ranging from 35.2 km t o 72.2 km (mean distance: 57.1 km). Three fields from each farm were selected for study. Two fields on each farm were used for investigating the abundance and spatial distribution of arthropods. To maintain the independence and comparability of observations, I selected fields that had been planted at the same time, but were as far apart as possible. A third field with overgrown margin vegetati on was selected for the mark -capture study. Over -grown margins were used because they come nearest to representing a refuge strip. In total, nine separate fields were used, six for abundance and distribution observations and three for the mark -capture study. 62 Mark -capture. In one field per farm, we established a sampling zone with four transects th at were arranged perpendicular to the field ed ge and extended from the margins into the field. The sampling zone was defined as the middle 150 meters of the entire field length with transects evenly spaced at 50m intervals (Fig 3.1). Four fixed sampling positions were located along each transect at th e following points: the margin, the field edge, 10m into the field , and 25m into the field for a total of sixteen sampling locations per field. A 76.2 x 1.3 cm galvanized pipe was hammered into the ground at each sampling position until the top edge was ju st above the canopy. The middle 250 m of each margin was sprayed until runoff with a solution of 10% albumin + 0.3 g/l of ethylenediamine tetra acetate (EDTA; ED4SS; Sigma -Aldrich) using a backpack sprayer with a three -nozzle boom at 30 -40 psi. The top and underside of all foliage was sprayed to a depth of approximately one meter, bringing the total area of each margin sprayed to ~250 m2. I placed yellow sticky traps ( 25.4 x 40.6 cm, Great Lakes IPM, Vestaburg, MI) at each of the sampling positions. Traps w ere removed from the field after 48 -72 hours and transported Figure 3.1. Mark -capture sampling design. Transects for yellow sticky traps were established in three commercial celery fields across three farms in southwest Michigan in July 2014. Yellow dots denote sampling locations. Orange rectangle denotes protein application area. 63 to the laboratory in a cooler. Individual insects were carefully removed from the sticky trap, each with a new wooden toothpick. Toothpicks were inserted into a 1.5 ml microcentrifuge tube and submerged in 1 ml solution of tris -buffered saline (TBS; 94158; Sigma -Aldrich) + 0.3 g/l EDTA. After 3 minutes, the insect and toothpick were removed and discarded. Samples were labeled and placed in a freezer ( -20¡C) to be tested in immunoassays for the presence of the protein at a later date. I conducted an array of tests to check for the possibility of false positives and contamination by spray drift. One week prior to spraying, I collected leaf samples from four dominan t plant species that were growing in the field margins on each farm to determine if the foliage possessed a protein that would trigger a false positive. Leaf samples were placed individually in 1 -gallon plastic storage bags and taken to the laboratory for analysis. A single leaf disc, 7mm in diameter, was removed from each sample using a hollow punch a nd placed in a 1.5 ml tube containing 1ml of the extraction buffer with forceps. After 3 minutes, the sample was removed and discarded. The hollow punch, the forceps, and the cutting mat upon which the samples were placed were all sanitized with an 80% ethanol solution and wiped with clean Kimwipes (Kimberly -Clark, Roswell, GA) after each sample. In addition, I tested celery foliage both for the possibility of a false positive and for contamination from spraying the marker. Four leaf samples were taken from four randomly selected celery plants in the first row nearest the margin and placed in individual storage bags the week before spraying. Upon removal of the sticky traps at the end of the field experiment, four additional leaf samples were taken from the row nearest the sprayed margin (within the bounds of the spray area) and placed into individual plastic storage bags. Leaf discs and assay samples were prepar ed immediately after returning to the laboratory and in the same manner as previously described for the margin foliage. Finally, I 64 used a sweep net to collect representative samples of the insect community in the celery fields and margins of all three farm s to determine if any species possessed a protein that would trigger a false positive. I selected a suitable range of taxa from this collection and acquired samples by submerging individuals in the extraction buffer as previously described. Samples in micr ocentrifuge tubes were labeled appropriately and stored in a freezer ( -20¡C) to be tested in immunoassays for the presence of the protein at a later date. Immunoassays. Indirect ELISAs (Crowther 2001) were performed on all samples as described by Jones et al. (2006, 2011). The primary antibodies utilized in all immunoassays were for chicken egg albumin produced in rabbit (rabbit anti -chicken) (C6534; Sigma -Aldrich), whereas secondary antibodies were donkey anti -rabbit with peroxidase conjugate (31458; Ther mo Scientific). The blocking buffer used in all assays was a solution of 0.3% Bovine Serum Albumin (BSA; B9433; Sigma -Aldrich) in phosphate -buffered saline (PBS; P4417; Sigma -Aldrich). As suggested by Jones et al. (2011), antibodies were mixed in a diluent buffer created from PBS and 1300 ppm Silwet L -77 for purposes of economy and assay optimization (Crowther 2001). Dilution ratios of 1:6000 for primary antibodies and 1:8000 for secondary antibodies were determined by the checkerboard titration process and used in all subsequent assays (Crowther 2001). All incubations were performed in an incubator oven (Lab -line Imperial III) set to 37 ¡C. All buffered solutions were prepared with filtered and deionized water (Barnstead Nanopure II). To test for the presen ce of the marker, 80 µl of each insect sample was placed in a separate well of a 96 -well microplate (Nunc -Immuno; M0661; Sigma -Aldrich). A solution of 10% antigen in sample extraction buffer (TBS + EDTA) was used for a positive control, with just the extra ction buffer used for negative control and blanks. Microplates were incubated for 65 120 minutes, after which, the contents were discarded. Wells were washed five times with 300 µl of PBS + 0.09% Triton X100 (PBST; X100; Sigma -Aldrich) followed by an addition of 300 µl of blocking buffer to every well. Microplates were then incubated for 60 minutes, after which the contents were discarded. The wells were washed two times with 300 µl of PBST and an 80 µl aliquot of properly diluted primary antibodies was placed into each well. The microplates were incubated for 30 minutes, the contents were subsequently discarded, and the plates washed five times with PBST. Secondary antibodies in 80 µl aliquots were added to each well and incubated for 120 minutes. The contents were discarded and the plates were washed three times with a solution of PBS + 2.3g/l sodium dodecyl sulfate (SDS; L4509; Sigma -Aldrich) followed by three washes with PBS. Once a chromogenic substrate (1-Step Ultra TMB; 34028; Thermo Scientific) in 80 µl aliquots was added to each well, plates were wrapped in aluminum foil and allowed to develop in the dark at 25 ¡C. After 20 minutes, the reaction was stopped with the addition of 80 µl of 2N H 2SO4 to each well. The optical density of each well was read with a plate reader (SpectraMax M5, Molecular Devices, California, USA) at a wavelength of 450 nm with 490 nm used as a reference standard . Wells with the extraction buff er (TBS + EDTA) were used as blanks to correct all readings. The values for optical density range from 0 -4, with higher numbers associated with darker colors caused by higher concentrations of the antigen. I used the conservative threshold of 4 standard deviations above the mean optical dens ity for negative control wells on a plate to determine whether a sample was positive for a mark (Jones et al. 2006, 2011). Arthropod abundance patterns. To assess the relationship between relative margin abundance and margin usage (expressed as percent po sitive for marker), I sampled the celery fields with 66 yellow sticky traps placed at canopy level to acquire spatial abundance data. In each of the remaining six fields (see Methods - Study sites), I established three transects that were arranged perpendicul ar to and spatially equidistant along the field edge and extended from the margins into the field. Transect length was uniform and determined by the distance between the edge and the middle of the most narrow field. Four fixed sampling positions were locat ed along each transect at the following points: the margin, the crop edge, 10m into the field , and 25m into the field for a total of twelve traps per field (Fig 2.3). I used non-baited 7.6 x 12.5 cm yellow sticky traps (Great Lakes IPM, Vestaburg, MI) plac ed at canopy level with wire loop hangers inserted into one end of 76.2 x 1.3 cm galvanized pipes. Traps were recovered after one week and arthropods of interest were counted and identified to the lowest possible taxonomic unit. Statistical analysis. To determine if there were significant differences in the proportion of insects testing positive for the marker, I performed Fisher's Exact Test (FET) in R (R Core Team, 2014) using the number of positive marks and the number of negative marks. This test r eturns the exact probability of observing the data on the null hypothesis that the proportions are equal. Data were pooled across taxa to generate the overall proportion of the community that tested positive at each sampling position. I performed pair -wise comparisons between all sampling positions using the Holm -Bonferroni correction for family -wise error rate. Due to low numbers of positive marks, data for individual taxa at the margin and edge positions were combined into one proportion and compared to t he combined data for the 10m and 25m positions. The resulting two categories represent insects that were collected in the immediate vicinity of the margins (within ~1m) and insects collected within the field interior (10 -25m away). This comparison 67 serves as an indicator of propensity to move from the margin into the field. To maintain consistency, the overall community proportions were also analyzed in this manner. I used R and lme4 (Bates, Maechler, Bolker, Walker, Christensen, S ingmann, Dai, and Grothend ieck 2015) to perform a mixed -model logistic regression of the relationship between distance from the margin and the proportion of insects that tested positive. Distance was entered as the fixed -effect predictor, field was a random effect, and a two -vector object containing the number of positive and the number of negative marks served as the binomial response variable. To determine if relative margin abundance can predict the degree of interaction with the margin (expressed as percent positive for marker) , I used data from two different sources. Data from the concurrent six-field spatial abundance study was used to derive a predictor statistic and data from the three -field mark -capture component (percent positive) was used as the response variable. For the predictor statistic, I calculated the mean number of individuals captured per 7.6 x 12.5 cm yellow s ticky trap at each sampling position across all si x fields for taxa of interest. I then created a ratio of mean margin abundance (M margin ) to the average mean abundance of the other three sampling positions (M edge , M10, M25), hereinafter referred to as the margin -field ratio (MF ratio) (3.1). I used R (R Core Team 2014) and a linear model ( lm) to perform a simple linear regression of the relationship between the margin -field ratio and the overall percent marked positive for arthopods of interest. The margin -field ratio was entered as the fixed effect and the percent positive was used as the re sponse variable. Regressions were performed on feeding functional groups (herbivores, predators, parasitoids) in addition to all of the groups combined. (3.1) 68 Results Overview . I removed a total of 1606 arthropods across 44 different taxonomic groups from 48 yellow sticky traps and tested them for the presence of the protein marker. Of these arthropods, 147 (9.2%) tested positive for albumin (Table 3.1). Across taxonomic groups, t he percentage of positive marks ranged from 0 -38.5% (Table 3.1); at least one of the groups was significantly different ( P=0.019, four -sample FisherÕs Exact Test, FET ). The proportion of positive marks was lowest at the field edge (6.5%, Table 3.1) with pr oportions 58%, 93%, and 19% greater at the margin, 10 -meter, and 25 -meter positions, respectively (Table 3.1). The difference in proportion between the edge and the 10m was the only pair -wise comparison with statistical significance (P=0.031, pair -wise FET , Holm -Bonferroni adjustment). The combined proportion of insects with a positive mark t hat were c aptured in the margin and edge was not significantly different from the combined proportion of protein -positive captures at the interior field locations (P=0. 255, FET). Among the 44 taxonomic groups I analyzed, 29 (65.9%) had at least one individual that tested positive for the marker (Table 3.1, Fig S10). Of these 29 groups, 19 different taxa (65.5%) had at least one individual testing positive at the 25m location and 23 taxa (79.3%) had at least one individual testing positive at either interior position (Table 3.1, Fig. S10). Despite significant differences for pair -wise comparisons, distance from the margin was not a significant predictor of the percenta ge of samples found positive for the marker (z=0.024, df=449, P=0.981, mixed -model logistic regression). To determine whether relative margin abundance can predict the degree of interaction with the margins (as measured by percent positive for marker), I used all taxonomic groups from the mark -capture study in which there were 10 or more individuals tested. I then calculated a 69 margin -field abundance (MF) ratio (as described in Methods section) for these groups and eliminated those for which there were no individuals captured in the field (i.e. ratio is undefined), resulting in a total of 27 taxonomic groups s uitable for evaluation . For reference, mean abundance (standardized for visual comparison purposes only ) at all sampling positions for the 27 taxa are displayed in Fig. 3.2. Overall, the MF ratio was not a significant predictor of the proportion of individuals that tested positive for the protein marker; the two parameters were not correlated (R2=0.006, F=0.161, df=1, 25, P=0.692, Fig 3.3). For herbivores, the MF ratio was not a significant predictor of the proportion of individuals that tested positive for the protein marker; the two parameters were not correlated (R 2=0.048, F=0.698, df=1, 14, P=0.417, Fig 3.3). For predato rs, the MF ratio was n ot a significant predictor of the proportion of individuals that tested positive for the protein marker; the two parameters were not correlated (R 2=0.198, F=1.23, df=1, 5, P=0.318, Fig 3.3). For parasitoids, the MF ratio was a significant predictor of the proportion of individuals that tested positive for the protein marker; the two parameters were correlated (R2=0.9304, F=26.73, df=1, 2, P=0.035, Fig 3.3). However, one regression point in the parasitoid analysis was highly influential. Once this point was removed from the analysis, the correlation was no longer significant (R 2=0.051, F=0.053, df=1, 2, P=0.856). Because proportions differed across sampling position and taxonomic groups, individual taxa of interest were analyzed separately: Aster leafhopper . I recovered and tested a total of 41 aster leafhoppers, of which 6 (14.6%) tested positive for the marker (Table 3.1). The proportion of positive marks was lowest at the edge position (7.1%, Table 3.1) with proportions at the margin, 10 -met er, and 25-met er positions greater by factors of 1.625, 1, and 2.5, respectively. When combined, the proportion of protein - 70 positive insects captured within 1m of the margins was not significantly different from the proportion captured at the interior field positions (P= 0.651, FET). Likewise, distance from the margin did not significantly influence the proportion of samples positive for the marker (z= 0.517, df=9, P=0.655, mixed -model logistic regression). Potato leafhopper. I recovered and tested a total of 156 potato leafhoppers, of which 8 (5.1%) tested positive for the marker (Table 3.1). The proportion of positive marks was lowest at the 25 -meter position (0%, Table 3.1). Proportions at the margin and 10 -meter positions were greater than the edge position by factors of 1.5 and 3.7, respectively (Table 3.1). When combined, the proportion of protein -positive insects captured within 1m of the margins was not significantly different from the proportion captured at the interior field positions (P=0.492, FET) . Distance f rom the margin was not a significant predictor of the proportion of samples testing positive fo r the marker (z= -0.697, df=9, P=0.268, mixed -model logistic regression). Tarnished plant b ug. I recovered and tested a total of 52 tarnished plant bugs, of wh ich 20 (38.5%) tested positive for the marker (Table 3.1); this was the highest percentage of any group. Proportions at the margin and edge positions were nearly the same (48.0 and 50.0%) and also #50.0% and #450.0% greater than the proportion positive at the 10 -meter and 25 -meter positions. When combined, the proportion of protein -positive insects captured within 1m of the margins was significantly different from the proportion captured at the interior field p ositions (P=0.038, FET). When modeled , distance was also found to significantly influence the proportion of samples positive for the marker (z= -2.181, df=9, P=0.029, mixed -model logistic regression). 71 Syrphidae. I recovered and tested a total of 25 syrphid flies, of which 7 (28.0%) tested positive for the marker (Table 3.1); this was the highest percentage of any natural enemy group. The proportion of positive marks was highest in the margins (57.1%); the other positions were lower (average 16.7%) but simil ar to each other (Table 3.1). When combined, the proportion of protein -positive insects captured within 1m of the margins was not significantly different from the proportion captured at the interior field positions (P=1.000, FET) . Distance from the margin was not a significant predictor of the proportion of positive marks (z= -0.738, df=9, P=0.461, mixed -model logistic regression). Coleomegilla maculata . I recovered and tested a total of 15 Coleomegilla maculata (CMAC) specimens, of which 4 (26.7%) tested p ositive for the marker (Table 3.1). This was the only lady beetle species to test positive for the marker and was also the natural enemy group with the 2 nd highest percentage. Positive marks were found only at the interior sampling locations, with the proportion at the 10 -meter location (2 positive/3 individuals) larger than that of the 25 -meter location (2/6) by a factor of 2 (Table 3.1). When combined, the proportion of protein -positive insects captured within 1m of the margins was not significantly diffe rent from the proportion captured at the interior field positions (P=0.103, FET). Distance from the margin was not a significant predictor of the proportion of positive marks (z=1.11, df=9, P=0.267). Chamaemyiidae. I recovered and tested a total of 50 cha maemyiid flies, of which 7 (14.0%) tested positive for the marker (Table 3.1); this was the third highest percentage among natural enemy taxa. The proportion of positive marks was lowest in the margins (0%, Table 3.1). Proportions at the other three sampli ng locations were similar to each other, differing largely by 72 the total number of tested specimen as opposed to the number of positive marks (2, 3, and 2 positive specimens at the edge, 10 -meter, and 25 -meter positions, respectively; Table 3.1). When combi ned, the proportion of protein -positive insects captured within 1m of the margins was not significantly different from the proportion captured at the interior field positions (P=0.225, FET) and distance from the margin did not significantly influence the p roportion of positive marks (z=1.152, df=9, P=0.337). Braconidae. I recovered and tested a total of 46 braconid wasps, of which 4 (8.7%) tested positive for the marker (Table 3.1); this was the highest percentage among parasitoid groups. All four of the p ositive marks were found on insects recovered from the margins (Table 3.1). When combined, the proportion of protein -positive insects captured within 1m of the margins was not significantly different from the proportion captured at the interior field posit ions (P=0.313, FET) . Distance was not a significant predictor of proportion positive (z= -0.002, df=9, P=0.998; simple logistic regression). Mymaridae. I recovered and tested a total of 76 mymarid wasps, of which 5 (6.6%) tested positive for the marker (Ta ble 3.1). The proportion of positive marks was highest at the 10 -meter position (18.2%, Table 3.1). However, the proportions at all four sampling locations were similar to each other, differing largely by the total number of tested specimen as opposed to t he actual number of positive marks (1, 1, 2, 1 positive marks at margin, edge, 10 -meter, 25 -meter positions, respectively; Table 3.1). When combined, the proportion of protein -positive insects captured within 1m of the margins was not significantly differe nt from the proportion captured at the interior field positions (P=0.378, FET) . Distance from the margin did not significantly 73 influence the proportion of positive marks (z=0.316, df=9, P=0.752, mixed -model logistic regression). Chalcidoidea. I recovered and tested a total of 114 chalcidoid wasps, of which 9 (7.9%) tested positive for the marker (Table 3.1). The proportion of positive marks was highest at the 10 -meter location (14.3%, Table 3.1). However, the proportions at all four sampling locations wer e similar to each other, differing largely by the total number of tested specimen as opposed to the actual number of positive marks (2, 2, 3, 2 positive marks at margin, edge, 10 -meter, 25 -meter positions, respectively; Table 3.1). When combined, the propo rtion of protein -positive insects captured within 1m of the margins was not significantly different from the proportion captured at the interior field positions (P=0.314, FET). Distance from the margin did not significantly influence the proportion of posi tive marks (z=0.565, df=9, P=0.572). Controls. Plant tissue . I sample d leaf tissue from a total of 19 different plant taxa, including the celery field , one week prior to the protein application; all samples tested negative for the presence of the protein marker ( Table S2 ). Eight leaf samples from the first row of celery in every mark -capture field were acquired at the end of the experiment to assess possible contamination. At least one celery sample from each farm tested positive for the presence of the pr otein marker, indicating contamination at all three study locations. The percentage of post -application celery samples testing positive for the presence of the marker ranged from 12.5 -25.0% percent across farms for a mean contamination percentage of 16.7% (Table S2). However, the proportion of contamination was not found to significantly influence the proportion of marked arthropods 74 across farms (t= -1.018, df=1, P=0.494). Samples from the sticky traps deployed in the mark -capture experiment were also tested to check for possible analogous protein cross -contamination as well as contamination from field activities; all samples tested negative for the protein ( Table S2). Leaf samples from post -application margin foliage were acquired to confirm the presence of the protein residue; all samples tested positive for the presence of the protein marker ( Table S 2). Insect negative control . Similar to the margin and celery foliage, insects from all three farms were collected with sweep nets one week prior to protein application to check for possible cross -contamination from analogous proteins. A total of 140 insects in 26 different taxonomic groups were collected; all samples tested negative for the presence of the protein ( Table S3 ). 75 Table 3.1. The number of arthropods (n ) recovered from yellow sticky traps deployed at four sampling positions across three celery farms in southwest Michigan. Percentage (%) indicates the proportion of taxa testing positive for the presence of chicken egg protein marker; hash sign (#) indicates the number of positive marks. MarginEdge10m25mTotal Taxon n%(#)n%(#)n%(#)n%(#)n%(#)HEMIPTERAMacrosteles quadrilineatus 1618.8(3)147.1(1)714.3(1)425.0(1)4114.6(6)Macrosteles spp. (other)20.0(0)20.0(0)00.0(0)10.0(0)50.0(0)Empoasca fabae454.4(2)342.9(1)3613.9(5)410.0(0)1565.1(8)Cicadellidae (other)128.3(1)1030.0(3)812.5(1)70.0(0)3713.5(5)Cercopidae10.0(0)00.0(0)00.0(0)1100.0(1)250.0(1)Aphididae320.0(0)464.3(2)3514.3(5)293.4(1)1425.6(8)Lygus lineolaris 2548.0(12)1050.0(5)633.3(2)119.1(1)5238.5(20)Lygus spp. (other)250.0(1)100.0(0)224.5(1)397.7(1)736.8(5)Miridae560.0(3)80.0(0)250.0(1)728.6(2)2227.3(6)Psyllidae130.0(0)40.0(0)520.0(1)250.0(1)248.3(2)Membracidae2040.0(8)450.0(2)333.3(1)30.0(0)3036.7(11) Cydnidae1100.0(1)616.7(1)812.5(1)250.0(1)1723.5(4)Aleyrodidae80.0(0)120.0(0)60.0(0)90.0(0)350.0(0)Anthocoridae621.6(1)5010.0(5)2015.0(3)185.6(1)1506.7(10)CHRYSOPIDAE 20.0(0)00.0(0)00.0(0)10.0(0)30.0(0)ARANAE277.4(2)385.3(2)267.7(2)140.0(0)1055.7(6)COCCINELLIDAECoccinella septempunctata00.0(0)00.0(0)10.0(0)10.0(0)20.0(0)Coleomegilla maculata10.0(0)50.0(0)366.7(2)633.3(2)1526.7(4)Harmonia axyridis40.0(0)30.0(0)20.0(0)10.0(0)100.0(0)Propylea quatuordecimpunctata 10.0(0)30.0(0)00.0(0)10.0(0)50.0(0)Symnus spp.00.0(0)20.0(0)10.0(0)20.0(0)50.0(0)CHRYSOMELIDAE Phyllotreta spp.520.0(1)30.0(0)30.0(0)00.0(0)119.1(1)Other220.0(0)273.7(1)80.0(0)90.0(0)661.5(1)COLEOPTERA (Other) Cantharidae00.0(0)20.0(0)20.0(0)10.0(0)50.0(0)Mycetophagidae10.0(0)00.0(0)00.0(0)00.0(0)10.0(0)Staphylinidae90.0(0)60.0(0)812.5(1)616.7(1)296.9(2)Curculionidae825.0(2)60.0(0)50.0(0)00.0(0)1910.5(2)ICHNEUMONOIDEABraconidae1723.5(4)160.0(0)80.0(0)50.0(0)468.7(4)Aphidiidnae10.0(0)00.0(0)00.0(0)10.0(0)20.0(0)Ichneumonidae10.0(0)333.3(1)10.0(0)00.0(0)520.0(1)Other10.0(0)00.0(0)00.0(0)1100.0(1)250.0(1)CHALCIDOIDEAMymaridae293.4(1)175.9(1)1118.2(2)195.3(1)766.6(5)Trichogrammatidae 40.0(0)30.0(0)10.0(0)40.0(0)120.0(0)Other336.1(2)365.6(2)2114.3(3)248.3(2)114 7.9(9)HYMENOPERA (Other) Vespidae 00.0(0)10.0(0)00.0(0)00.0(0)10.0(0)Formicidae00.0(0)20.0(0)00.0(0)00.0(0)20.0(0)DIPTERAChamaemyiidae120.0(0)1513.3(2)1323.1(3)1020.0(2)5014.0(7)Syrphidae757.1(4)911.1 (1)333.3(1)616.7(1)2528.0(7)THYSANOPTERAThripidae372.7(1)370.0(0)3910.3(4)345.9(2)1474.8(7)Aeolothripidae110.0(0)110.0(0)119.1(1)1010.0(1)434.7(2)PSCOPTERA616.7(1)40.0(0)10.0(0)30.0(0)147.1(1)ACRIDIDAE10.0(0)00.0(0)00.0(0)00.0(0)10.0(0)LEPIDOPTERA10.0(0)10.0(0)00.0(0)250.0(1)425.0(1)TOTAL 48510.3(50)4606.5(30)32612.6(41)3357.8(26)16069.2(147) 76 Figure 3.2. Mean abundance of arthropods at four different sampling positions fro m data collected with yellow sticky traps in six fields across three different celery farms in southwest Michigan during the first week of July 2014 . Data are standardized on a scale ranging from 0 -1 to enable comparisons across taxa. 77 Figure 3.3. The relationship between the Margin -Field ratio (MF) and the proportion of individuals testing positive for the presence of the chicken egg protein marker. The MF ratio is the abundance of arthropods caught in the margin relative to those caught in the cel ery field (please see Methods section). Regression points and lines were plotted with different symbols and colors according to feeding functional group. The black regression line is the result of all points in a singular analysis. 78 Discussion In my fie ld study, I investigated the movement and spatial distribution of canopy arthropods in Michigan celery fields. I measured mean abundance at various distances into the field , quantified movement patterns, and assessed the degree to which key groups interact with the margins. Over 65% of the groups analyzed for the presence of the marker had at least one individual test positive and almost 80% of those groups tested positive at either interior field position. This suggests that the majority of the canopy arthropod community in celery systems interact with the margins to some extent and that these groups move from the margins into the field. In terms of the whole community, distance did not significan tly influence the proportion of insects testing positive for the marker. Likewise, there was no significant difference between the proportion of protein -positive insects captured near the margins and those captured at the interior positions. These results indicate that the typical dispersal distance by any given arthropod is greater than 25 meters and that re -colonization of a field after a pesticide application occurs rapidly. It also suggests that, in general, the hard edges of the crop -soil -margin interf ace do not seem to hinder the canopy community from moving between margin and field , although there may be specific groups for which it does (e.g. TPB, syrphids). Most importantly, however, it shows that natural enemy groups would b e able to respond to prey aggregation at the innermost part of a celery field. The percentage of individuals within taxonomic groups that tested positive for the presence of the marker ranged from 0 -38.5% with an overall average of 9.2%. The distribution of proportions positive for the marker across the 27 taxa in the analysis was skewed low; only ten groups had proportions higher than the average. Because the percentages varied widely, I analyzed several key arthropod groups separately. While this method strengthened my ability to 79 assess differences in proportions across taxa, I acknowledge that the low number of positive marks imposes limitations on conclusions made about the movement patterns of specific athropod groups. Therefore, I focus primarily on overall percent ages, proximity to the margin, and the furthest distance travelled. The proportion of tarnished plant bugs (38.5%) and aster leafhoppers (14.5%) testing positive for the protein were the highest among key pests and well above the community average. Collectively, 8 5% of the protein -positive tarnished plant bugs and 66% of the protein -positive aster leafhoppers were captured within 1 meter of the margin, which suggests a high propensity to stay near the margin when moving about. For comparison , 90.9% of protein -marke d treehoppers, Acutalis spp. (Hemiptera: Membracidae), which are common in celery margins due to host plant associations with abundant ragweed species ( Ambrosia spp.), were captured within 1 meter of the sprayed area (Table 3.1; Fig. S10; J. Jubenville, pe rs. obs.). I speculate that the limited movement pattern in TPB and ALH could be indicative of habitat preference. Although they are both highly polypha gous species, if the margins serve as refuge and the plants in the margins are a consistent non -toxic fo od resource, then plant host habi tuation may explain this movement pattern (e.g. Bancroft 2005). Nevertheless, positive marks for both the TPB and ALH were found at all sampling positions within the celery field . This indicates that some proportion move in to the field interior and suggests that they can easily forage anywhere within the field from the margins. Aphids, thrips, and potato leafhoppers were well below the average percent positive, suggesting a relatively low degree of interaction with border areas. These groups are usually considered part of the aerial plankton pool and may only interact with the margins in this system on an incidental basis. Together, 25%, 14.3%, and 37.5% of protein -positive aphids, thrips, and 80 PLH, respectively, were captur ed at the margin and edge positions (within ~1 meter of the sprayed areas), suggesting that they are not as inclined to stay near the margins as TPB and ALH. Interestingly, potato leafhoppers had one of the lowest percentages (5.1%) of herbivores in the st udy despite being the most abundant cicadellid species during the two seasons (CHAPTER 2, Table S1 ). Potato leafhopper ecology is much like that of the aster leafhopper in that a small proportion of the population overwinters within Michigan , but the major ity migrate on the wind from southern states. Potato leafhoppers, like aster leafhoppers, are also agile flyers but clearly inte ract with the margins to a lesser extent ( 5.1 versus 14.5% positive). Among predators, only three taxonomic groups had protein -positive percentages higher than the study average: the pink spotted lady beetle, CMAC (26.7%), and two dipteran families, Syrphidae (28.0%) and Chamaemyiidae (14.0%). All of the protein positive CMAC specimens were captured at the interior field position s, suggesting little propensity to stay near the margins. Coleomegilla maculata is different than the other lady beetle species in this system. Not only is it far more abundant than others (CHAPTER 2 , Table S1 ), but it is also the only coccinellid that tested positive for the presence of the marker. In addition to insect prey, this lady beetle species is known to feed on floral resources and fungal spores (Webber an d Lundgren 2009), which are likely to be far more abundant in the margins than the field due to weekly fungicide use in the field . Flowers are also occasionally present within the margins and usually found on faster -developing species such as smartweed ( Polygnum spp.) and field bindweed ( Convolvulus arvensis ). Because of this, CMAC may frequently visit celery margins to forage for pre y, pollen, and spores, but the margins are likely one of many habitats it may visit on any given day. With the dipteran predators, 71% of the marked syrphids and 28.5% of the marked chamaemyiids were captured within 1 meter of the margin, indicating that syrphids have a higher 81 propensity to stay near the margins. Syrphid flies feed on nectar as adults and were consistently observed on smartweed flowers growing in field borders (J. Jubenville, pers. obs.). Floral resources are known to enhance local syrphid populations in agricultural landscapes (Cowgill et al. 1993, Blaauw & Isaacs 2012) and celery margins can, at times, provide a combination of flowers, refuge, and oviposition sites (i.e. nearby prey for offspring). A s such, it seems likely that syrphids use these margins as habitat when conditions are favorable. Chamaemyiids seem to be rangy fliers (J. Jubenville, pers. obs.) and likely use this vagility to forage within a variety of habitats. As with CMAC, chamaemyii ds are frequent visitors to margins, but their overall association is moderate compared to other groups. Both dipteran groups had positive marks at the furthest location from the sprayed areas, suggesting that they will forage throughout the field after in teracting with the margin. Among the parasitoids, Braconidae (8.7%), Mymaridae (6.6%), Trichogrammatidae (0.0%), and other Chalcidoidea (7.9%) scored near or below the community average of 9.2%. Despite these numbers, 100%, 40%, and 44% of the protein -positive braconid, mymarid, and chalcioid wasps, respectfully, were captured within 1 meter of the margin. This indicates that the degree to which the prominent parasitoid groups interact with the margins is highly variable and may also be influenced by other factors (e.g. host availability). Positive marks for mymarid and chalcidoid wasps were found at all sampling locations, indicating that they will move throughout the field after interacting with the margin. In the regression analysis, the points representing Mymaridae, Chalcidoid ea, and Braconidae were clustered closely together (Fig. 3.3), while the point representing the Trichogrammatidae (0% positive) was located well outside this cluster and near the origin. When analyzed, the regression shows a high ly significant positive relationship between the MF ratio 82 and the percent positive for parasitoids (R 2=0.9304, F=26.73, df=1, 2, P=0.035, Fig 3.3). However, when the Trichogrammatidae are removed from the analysis, thereby reducing it to a regression of th ree points, the correlation is no longer significant (R 2=0.051, F=0.053, df=1, 2, P=0.856). Caution should therefore be taken when forming conclusions based on this result . The majority of specimens in every group tested negative for the presence of the m arker. My interpretation of these results is that it is more likely that the majority of insects found within the celery fields come from sources other than the margins (but see False negatives and false positives and Marking efficiency). From an ecological standpoint, in order for margins to function as a source of arthropods, the rate of emigration would need to exceed the rate of immigration as well as the birth rate exceeding the death rate. Al though this is something that I did not directly me asure, the fact that margins are subject to intense ecological disturbance on a monthly basis (mowing, Fig 1.3) makes this an unlikely prospect for all but those species with the highest reproductive rates (e. g. aphids). Margins probably function as a refu ge most weeks of the season and so abundance would be expected to be higher in these areas. As I have shown, however, a margin -centric pattern does not seem to be a reliable predictor of overall habitat use (Fig 3.3). False negatives and false positives. The discovery of positive marks in this study is directly related to the threshold value for optical density. The OD of any particular sample must exceed the average OD of the negative control wells on the same plate by four standard deviations in order for it to be considered positive for the presence of the protein. This is a conservative threshold established by Jones et al. (2006) and is used in subsequent studies by Jones, Hagler, and others (e.g. Boina et al. 2009, Horton et al. 2009, Hagler and Jones 2010, Peck et al. 2014). 83 The cost of such a conservative approach is an inherent increase in false negatives. In their studies, Jones and Hagler were more interested in whether arthropods were moving from one location to another as well as the distance that a particular taxon moved from the sprayed location. They took care in making generalizations about how much any particular arthropod group utilized the sprayed areas. Likewise, they were also careful in how they compared the percent positive among taxa. The question of where to set the thresh old is comparable to the question of where we set our alpha value when performing statistical analyses. Using a small alpha increases the risk of not detecting a significant effect (false -negative) when one is actually present, whereas using a larger alpha increases the risk of declaring a significant effect (false -positive) when, in reality, one does not exist. This translates to my study in the following manner: if I increase the possibilit y of false negatives (higher O D threshold), then I risk under estimating the degree to which insects use the margins. If I increase the possibility of false positives (lower OD thresh old), then I not only risk over estimating how much arthropods use the marg ins, but I also risk over-estimating their movement patterns as well. This is of particular importance when considering the question of whether natural enemies will move out into the center of the field to prey upon pests . In this study, I prefer the conve ntional conservative approach. This means that I may be underestimating the movement patterns and the degree to which any particular taxon interacts with the margins. Marking efficiency. Another possible caveat is that different species may acquire positi ve marks with different efficiency. Hagler and Jones (2010) showed that it does not take long for a broad range of taxa to acquire positive marks from residual chicken egg protein on leaf tissue. In this 84 study they introduced unmarked arthropod assemblages acquired with sweep nets to field cages containing cotton plants previously sprayed with chicken egg whites. Subsequent sampling showed a high percentage of positive marks after one day and up to eleven thereafter. Overall percentages for the first four d ays after spraying ranged from 70% for Hymenoptera (mostly ants) to 96% for Coleoptera (many species). More relevant to my study is that 91% of Heteropterans tested positive after four days. These percentages indicate two important things: 1) most taxa wil l readily acquire residual egg protein i n a field study under two weeks in duration (assuming no rain), and 2) there is likely a difference in the rate in which taxa acquire the marker. Furthermore, tarnished plant bugs, the group with the highest percenta ge in my study, are part of a group (Heteroptera) that demonstrated high marker acquisition efficiency in the Hagler and Jones (2010) study. Therefore, it is likely that I am underestimating the degree to which taxa interact with the margins, but the diffe rence may be minimal. Contamination. During the protein application process, the spray boom was pointed in the opposite direction of the field at all times. Nevertheless, at least one leaf sample taken from the first row of celery on each farm at the end of the experiment tested positive for the protein maker (Table S2 ). This means that insects could acquire the protein mark from walking on some of the celery foliage. This result is not surprising given that the first celery row is ~1m from the margin foliage and highlights two important points: ELISA assays are highly sensitive and margin foliage coverage was probably good . When interpreting the results of this study, it should be noted that the contamination in the field could result in an overestimation of margin usage and movement patterns. That being said, th e amount of margin foliage covered in egg whites greatly 85 exceeded the amount that may have drifted into the field. Insects were far more likely to acquire marks from the margin than anywhere else. Summary and Application . Understanding the sources of insects, their movement patterns, and how they utilize key areas in agricultural systems can have a positive impact on crop production. This is particularly important in the case of plant disease vecto rs where simple management changes may have a profound effect on local disease ecology (e.g. Boina et al. 2009). Knowledge of aggregation tendencies can be used to successfully reduce the frequency and impact of pest management actions in some crops by imp lementing tactics such as targeted spraying (e.g. Morrison and Szendrei 2013) and local habitat manipulatio n (e.g. Swezey et al. 2007, 2013). The results of this study reveal differences in how various arthropod groups interact with the margins in Michiga n celery systems. Variation in mark percentage, as well as the spatial distribution of the marks, shows that some groups utilize the margins more than others and these groups are comprised of both key pests and natural enemies. A useful application of this information is to predict how these groups might respond to alternative margin management strategies. Any systematic change in the way growers manage their margins would likely affect these groups to a greater degree than the other groups in the study. Fo r instance, suppose we adopt a conservative approach, retain the current floral composition (mostly common agricultural weeds and grasses), and allow the margins to grow throughout the year. We could expect all of these groups to benefit from the habitat e nhancement, but perhaps most conspicuously, tarnished plant bugs. A situation such as this could result in a dramatic increase in pest pressure, especially if pest control within the margin is not equally enhanced by the increase in stability . 86 A more pro ductive approach may be to replace much of the current flora with a select mixture of flowering perennials. Syrphids, coccinellids, and parasitoids have all been found to exploit floral resources on the fringes of agricultural fields and subsequently dispe rse into the field . For example, Freeman Long et al. (1998) sprayed rubidium (Rb) on flowering strips and found Rb -labeled natural enemies as far as 75m into an orchard. Similarly, Nicholls et al. (2001) found that the presence of a flowering plant corrido r enhanced natural enemy colonization of an adjacent vineyard. Most pertinently, Nicholls et al. (2001) noted that the influence of the flowering corridor was limited by natural enemy dispersal distance. In celery systems, where fields are often narrow and the edge-to-area ratio is high, dispersal to the center of the field from the margins is not likely to be a limiting factor in colonization success. The prospect of strengthening pest populations with this strategy is a concern (e.g. Winkler et al. 2009, 2010). However, previous research in southwest Michigan has shown that a thoughtfully designed mixture of flowering perennials can result in greater abundance of key natural enemy groups without a significant increase in he rbivores. Blaauw and Isaacs (2012) found that cicadellid abundance declined dramatically and mirid abundance increased moderately with the addition of flowering perennials. Most notably, they documented an increase in pest control (aphids) on sentinel plants, but only when flowers were pr esent. This speaks to the influence that floral resources may have on pest regulation in managed environments. Conserving margins for beneficial insects in Michigan celery may be a relatively easy way to integrate an additional pest control strategy into these busy commercial systems. When combined with diligent scouting and economic thresholds, successful implementation of this strategy could provide growers with substantial economic benefits and reduce the environmental impact of intensive agricultural practices. Future research on this subject should consider how 87 habitat enhancement might affect key pest populations (e.g. Wackers 2001, Winkler et al. 2010) as well as the likelihood of pesticide drift into conservation areas. Additional work to establish economic thresholds for key pests would also be highly beneficial. 88 CHAPTER 4: Conclusions and future r esearch There were two primary objectives of this thesis: 1) quantify spatiotemporal abundance patterns of arthropods in and around celery fields and 2) evaluate the likelihood of field margins as a source of insects in the crop. I proceed to achieve these objectives with a combination of inferential observation and direct measurement. In chapter two, I analyze d the spatial and temporal abundance pa tterns of a broad range of arthropods across the season. Given the narrow field profile with a short distance to the middle, a consistent edge -centric colonization pattern could signify that margins are a source for any particular group. On the other hand, I would expect a high-dispersal group interacting with the margins on an incidental basis to be evenly distributed (on average) across the field. In chapter three, I implement a mark -capture experiment to assess margin usage across a broad range of taxa. Although this is a far more direct method of answering my question, it is not without its own set of implications. I inferred, generally, that if a species was associated with the margin then it was more likely to have a relatively high percentage of marke d individuals and that the majority of these marked individuals would be recovered in close proximity to the margin. When combined with the furthest distance a marked specimen is recovered, we can predict which species would be most likely to respond to ma rgin enhancements and how that might affect the crop. Land use within the modern agricultural landscape falls along a continuum of habitat stability, with areas of intense disturbance on one end and near -natural areas at the other. At the farm scale, crop fields are ephemeral habitats that bear frequent and intense disturbance while border areas often managed less frequently and with varying levels of intensity. Pest 89 management practices often cause arthropod extinction within the field , but border areas are not usually managed in this way. The inequality in disturbance between adjacent areas may result in a source-sink dynamic with arthropods from border habitats re -colonizing the suddenly unoccupied field. Whether this is considered a favorable circumstance depends on the degree to which pest species utilize these areas. Nevertheless, because local abundance can be influenced by factors at the landscape scale (Tscharntke 2005a), it should not be assumed that margins are the primary sour ce of any group of insects found within the field . Understanding the origin of beneficial and pestiferous insects within a production system is fundamental to selecting the most suitable management practice. Chapter two characterizes the abundance and dist ribution patterns of the canopy arthropods based on the results of yellow sticky trap sampling at different distances into the field. In general, groups were either: 1) evenly distributed across all sampling locations, 2) most or least abundant in the marg ins, but evenly distributed throughout the field, or 3) most abundant at the margins/field edge and decreasing in abundance with greater distance into the field. I hypothesized that groups fitting into the last two categories are those that might interact with the margins to a higher degree and may therefore benefit from habitat enhancement in these areas. Among these groups are historically important pests as well as natural enemies and so their relationship with the margin needed to be quantified in a mor e direct manner before margin enhancements could be considered. Celery systems are often comprised of groups of several long narrow fields separated lengthwise by drainage ditches (Fig 1.2). The fields are narrow enough that recolonization should occur ra pidly. Margins and ditch banks are the nearest potential source of arthropods and growers manage them with this in mind; practices such as mowing and herbicide application are common. The degree to which any arthropod group used the margins, however, was n ot 90 established and growers managed them under the assumption that pestiferous herbivores would use them as habitat. Border habitats in other systems have been manipulated to encourage natural enemy populations through the addition of perennial floral resou rces. These types of modifications are intended to attract predators and parasitoids currently present within the landscape with the expectation that, once there, they will disperse within the field to provide natural pest control. Research has shown this strategy to work in some situations (Bianchi et al. 2006) and may be an employable strategy in Michigan celery systems. Many celery fields are narrow, with their centers approximately 25 meters from the edge. Conceptually, such narrow fields should be quic kly re -colonized by mobile arthropods once the toxicity of pesticide residues begins to fade. In chapter three, I examined the degree to which the arthropod community interacts with the margins and how far they move out into the field. Margins were sprayed with a crude protein marker that would not normally be found in celery fields (i.e. chicken egg albumin). Arthropods acquired the marker directly by being sprayed or indirectly from walking on residue -covered foliage. We can use the results of this experi ment in several ways: we can quantify how far the insects move into the field after interacting with the margins, we can estimate the degree to which the insects interact with the margins, and we can ascertain which groups have a propensity to stay near th e margins. Over 90% of specimens tested negative for the protein, although the majority (65.9%) of groups had at least one positive mark. This suggests that most groups do interact with the margins, but only to a minor extent. Therefore, it does not appear that margins, as they are currently managed, serve as source of insects that are found in the celery field. I also assessed whether margin abundance (relative to field abundance, MF ratio) could predict margin usage (measured as overall percent positive) and found that there was no systematic relationship between the two. This was particularly important when 91 considering the spatiotemporal data for parasitoids, which provided clear margin -centric patterns (Fig. 2.7). From the mark -capture results, we can s ee that some groups utilize the margins more than others and these groups are comprised of both key pests and natural enemies. If margins were left to grow in their current state, I would expect tarnished plant bug and syrphid fly numbers to increase along with Chamaemyiidae and Coleomegilla maculata numbers to a lesser extent. Importantly, all of these groups forage throughout the field while interacting with the margins. Perennial flowering strips could differentially benefit natural enemy populations (ei ther through attraction or overall habitat stability) and I would anticipate that lady beetle, syrphids, anthocorids, and parasitoids would all benefit from margins that have been managed with this intent. Further research could explore the interaction of different margin treatments in pest and natural enemy abundance within celery fields. In summary, I conclude that the margins in Michigan celery systems are not currently used to a large extent by most arthropod groups. Two arthropod groups seem to use th em more than others: tarnished plant bugs and syrphid flies. Apart from these two groups, it does not seem likely that margins are a major source of insects that we find in the field . Furthermore, margin -centric patterns do not seem to be a reliable predic tor of overall margin utilization (e.g. parasitoids) or abundance in the celery field, so care must be taken when making inferences based only on spatial abundance patterns. 92 APPENDICES 93 Appendix 1.1 Supplementary data Table S1. The total number, frequency, and feeding functional group of arthropods captured on sticky traps in celery fields across southwest Michigan in 2013 and 2014. Frequency is percentage (%) of traps with at least one individual. "(Aq)" indicates aquatic taxa. 20132014OrderSuborder/ SuperfamilySuperfamily/ FamilySub-family/ group SpeciesTotal Freq. (%)Total Freq. (%)Feeding Functional Group Araneae74732.7108843.2PredatorOpiliones20.2PredatorColeopteraAnthicidaeNotoxus spp.343.770.7Detri/HerbivoreUnidentified16412.430.4OmnivoreBuprestidae20.240.5HerbivoreCantharidaeChauliognathus spp.421.3PredatorUnidentified162.0737.2Carabidae14811.6 606.1PredatorChrysomelidaeAcalymma vittatum70.970.8Pest/HerbivoreCerotoma trifurcata 40.470.8Diabrotica undecimpunctata 20.2Oulema melanopus70.7Unidentified45217.531620.5AlticiniPhyllotreta spp.392.8735.5Unidentified72026.1106423.1Cassidinae150.8Cleridae20.2CoccinellidaeCoccinella septempunctata141.570.8Coleomegilla maculata34726.636323.9Cycloneda munda20.210.1Harmonia axyridis272.3374.0Hippodamia convergens 40.510.1Hippodamia parenthesis 40.510.1Hyperaspis spp.10.1Propylea quatuordecimpunctata 91.0Psyllobora spp.50.6Scymnus spp.25315.8Unidentified121.5CurculionidaeDryophthorinae13712.411110.3HerbivoreScolytinae20.2HerbivoreElateridae10.1Histeridae00.030.4Lampyridae91.080.9Latridiidae67733.518712.0FungivoreMeloidae40.210.1Melyridae00.060.5Mordellidae13210.5172.0Herbivore/PredatorMycetophagidae906.2FungivoreScarabaeidae10.140.4Herbivore 94 Table S1 (contÕd) 20132014OrderSuborder/ SuperfamilySuperfamily/ FamilySub-family/ group SpeciesTotal Freq. (%)Total Freq. (%)Feeding Functional Group Silphidae10.1DetritivoreSilvanidae30.4FungivoreStaphylinidae45230.335026.9Tenebrionidae 957.2Detri/HerbivoreThroscidae40.5HerbivoreUnidentified58429.4221.9Various DipteraAsilidae20.230.4Chamaemyiidae252.250525.8Dolichopodidae00.022519.0Syrphidae28820.112711.5 Tachinidae 1249.2ParasitoidEphemeroptera10.1Algivore (Aq)HemipteraAuchenorrynchaCercopidae60.7807.4HerbivoreCicadellidaeEmpoasca fabae358275.5652487.3Pest/HerbivoreMacrosteles quadrilineatus 219050.6417451.6Macrosteles spp. (other)71018.2Agallia quadripunctata90.316611.3 Flexamia spp.401.2Norvellina spp.10.110.1Grape Leafhopper291.6Graphocephala coccinea151.3Colladonus clitellarius50.6Deltocephalus spp.30.4Draeculacephala zeae80.8Unidentified137550.692041.0Issidae10.1HerbivoreMembracidae1016.325413.0HerbivoreSternorynchaAphididae2889397.36093890.8Pest/HerbivoreAleyrodidae98323.362740.6PestPsyllidae23814.667630.0HerbivoreHeteropteraAnthocoridae31318.341625.4PredatorBerytidae10.1HerbivoreBlissidae10.1HerbivoreCydnidae30.4391.5HerbivoreLygaeidae Ischnodermus spp.20.2Unidentified20.110.1Lygaeoidae 10.110.1HerbivoreMesoveliidae10.1HerbivoreMiridaeLygus lineolaris 21010.474230.2Pest/Herbivore 95 Table S1 (contÕd) 20132014OrderSuborder/ SuperfamilySuperfamily/ FamilySub-family/ group SpeciesTotal Freq. (%)Total Freq. (%)Feeding Functional Group Lygus spp. (other)2027.426513.3Four-lined plant bug 120.8Trigonotylus caelestialium 70.3221.5Plagiognathus spp.111.2Coccobaphes spp.30.2Halticus bractatus50.1Unidentified18513.717712.2Nabidae110.950.5PredatorPentatomidae121.210.1HerbivorePiesmatidae272.6HerbivoreReduviidae20.310.1PredatorSaldidae10.1PredatorTingidae 60.7HerbivoreUnidentified657.530.4HerbivoreHymenopteraFormicidae120.9848.6PredatorApoideaAnthophila10.1111.2PollinatorColletidaeHylaeinae10.1Unidentified20.2Sphecidae202.0Vespoidea Tiphiidae 20.2ParasitoidPomilidae10.1Predator/parasitoidVespidae 40.5Unidentified141.010.1AculeataUnidentified10.170.7PredatorIchneumonoideaBraconidaeAphidiinae161.71369.5ParasitoidMicrogastrinae50.6Unidentified18416.452035.1Ichneumonidae464.8797.4Unidentified19617.8353.4ChalcidoideaMymaridae94647.3159167.6Trichogrammatidae 18515.344831.7Unidentified195857.0358175.6EvanioideaEvaniidae10.1DiaprioideaDiapriidae10.1CeraphronoideaCeraphronidae332.3Megaspilidae10.1292.8ProctotrupoideaProctotrupidae70.8ChrysidoideaBethylidae190.7192.2 96 Table S1 (contÕd) 20132014OrderSuborder/ SuperfamilySuperfamily/ FamilySub-family/ group SpeciesTotal Freq. (%)Total Freq. (%)Feeding Functional Group Dryinidae10.1CynipoideaUnidentified124439.2PlatigastroideaPlatygastridae40.2Scelionidae10.1ParasiticaUnidentified59329.060.7SymphytaArgidae Purslane sawfly130.5251.3HerbivoreUnidentified040.2Siricidae010.1Tenthrendinidae 10.160.7Unidentified515.520.2LepidopteraPieridaePieris rapae101.1HerbivorePhoebis sennae10.1Pyraloidea10.160.7Heterocera153.1Unidentified immature050.5Unidentified111.2737.5NeuropteraChrysopoidae131.590.9PredatorHemerobiidae40.520.2OdonataAnisopteraCoenagrionidae80.8PredatorLestidae20.2Unidentified60.7ZygopteraLibellulidae30.4Unidentified373.860.7OrthopteraAcrididae131.4HerbivoreTettigoniidae 30.4Unidentified161.3PlecopteraNemouridae10.120.2Detri/Algi/Herbivore (Aq)Unidentified20.2Pscocoptera16411.8 766.3HerbivoreThysanopteraAeolothripidae`52722.244626.8PredatorThripidae2348388.42873697.1HerbivoreTrichoptera Unidentified60.6161.6Various (Aq) 97 Table S2. Number analyzed (N), mean (+/ - SEM) ELISA optical density value (OD), and the percentage of sample s (%) testing positive for presence of albumin protein marker. Leaf samples were acquired before and after protein application to margins. Sticky trap sample discs were acquired from the recovered traps. Post -application tests of celery foliage and sticky traps were used to detect contamination resulting from wind drift, precipitation, or condensation. Before Spraying OD NegativeOD PositiveLeaf typeN(mean +/- SEM)N(mean +/- SEM)%Farm 1Chenopodium album40.095±0.0030--0.0Portulaca oleracea40.101±0.0010--0.0Polygonum pensylvanicum40.093±0.0010--0.0Amaranthus spp.40.103±0.0000--0.0Ambrosia trifida 40.106±0.0040--0.0Urtica dioica40.109±0.0030--0.0Grass (unknown)40.103±0.0020--0.0Celery (1st row)40.102±0.0060--0.0Farm 2Arctium spp.40.105±0.0010--0.0Typha spp.40.096±0.0010--0.0Urtica dioica40.093±0.0010--0.0Sassafras albidum40.106±0.0040--0.0Ascelpias syriaca40.089±0.0000--0.0Polygonum pensylvanicum40.090±0.0010--0.0Parthenocissus quinquefolia40.091±0.0020--0.0Celery (1st row)40.095±0.0000--0.0Farm 3Convolvulus arvensis40.101±0.0020--0.0Cyperus esculentus40.102±0.0010--0.0Silene Alba 40.105±0.0020--0.0Abutilon theophrasti40.102±0.0010--0.0Amaranthus spp.40.091±0.0010--0.0Cirsium vulgare 40.104±0.0020--0.0Ambrosia trifida 40.091±0.0010--0.0Grass (unknown)40.102±0.0060--0.0Celery (1st row)40.106±0.0030--0.0After Spraying OD NegativeOD PositiveType N(mean +/- SEM)N(mean +/- SEM)%Farm 1Celery (1st row)70.093±0.00110.113 -12.5Residue check0--40.750±0.031100.0Sticky trap check40.089±0.0000--0.0Farm 2Celery (1st row)60.094±0.00120.099±0.00025.0Residue check0--40.637±0.112 100.0Sticky trap check40.086±0.0020--0.0Farm 3Celery (1st row)60.092±0.00120.105±0.00825.0Residue check0--80.435±0.120100.0Sticky trap check40.089±0.0010--0.0 98 Table S3. Num ber analyzed (N), mean (+/ - SEM) ELISA optical density value (OD), and the percentage of individuals (%) testing positive for presence of albumin protein marker. All insects were collected from three different celery farms with sweep nets one week prior to protein marker application. ODTaxon N(mean +/- SEM)%HEMIPTERAAnthocoridae50.102±0.0020.0Aphididae70.091±0.0010.0Macrosteles quadrilineatus 40.089±0.0010.0Empoasca fabae80.092±0.0010.0Cicadellidae (other)70.099±0.0020.0Lygus lineolaris 160.099±0.0020.0Lygus spp. (other)20.102±0.0010.0Miridae10.0840.0Psyillidae40.104±0.0030.0Pentatomidae20.110 ±0.0040.0COCCINELLIDAEPropylea quatuordecimpunctata 20.091±0.0000.0Coccinella septempunctata10.110 0.0Coleomegilla maculata30.094±0.0030.0Harmonia axyridis10.1200.0Cycloneda munda10.1070.0COLEOPTERA (Other) Cantharidae40.092±0.0010.0Carabidae10.1040.0Chrysomelidae110.094±0.0020.0Curculionidae (weevil)100.096±0.0020.0ARANAE80.100±0.0030.0HYMENOPTERABraconidae120.100±0.0020.0Ichneumonidae20.104±0.0090.0Chalcidoidea80.100±0.0030.0DIPTERAChamaemyiidae10.0970.0Syrphidae140.100±0.0020.0LEPIDOPTERA (moth) 50.097±0.0050.0 99 Figure S1. Mean distribution of alate aphids across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 100 Figure S2 . Mean distribution of thrips across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 101 Figure S3. Mean distribution of potato leafhoppers across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 102 Figure S4 . Mean d istribution of aster leafhoppers across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within wee ks are not significantly different. Note: each year is at a different scale. 103 Figure S5. Mean distribution of tarnished plant bugs across four different sampling locations over all studied celery fields in southwest Michigan in 2014. Abundance is pro portional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 104 Figure S6. Mean distribution of spiders across four different sampling locati ons over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 105 Figure S7. Mean distribution of chalcidoid wasps across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 106 Figure S8. Mean distribution of mymarid wasps across four different sampling locations over all studied celery fields in southwest Michigan for 2013 a nd 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly different. Note: each year is at a different scale. 107 Figure S9. Mean distribution of braconid wasps across four different sampling locations over all studied celery fields in southwest Michigan for 2013 and 2014. Abundance is proportional to circle radius (mean aphids per sticky trap). Circles of the same color within weeks are not significantly d ifferent. Note: each year is at a different scale. 108 Figure S10. The number of individuals testing positive for the presence of a chicken egg protein marker applied to the margins of three celery fields in southwest Michigan. 109 Appendix 1.2 RECORD OF DEPOSITION OF VOUCHER SPECIMENS The specimens listed below have been deposited in the named museum as samples of those species or other taxa, which were used in this research. Voucher recognition labels bearing the voucher number have been attached or inc luded in fluid preserved specimens. Voucher Number: 2015-04 . Author and Title of thesis: Author: Jeremy Jubenville Title: The spatial and temporal distribution of arthropods in Michigan celery agroecosystems Museum(s) where deposited: Albert J. Cook Arthropod Research Collection, Michigan State University (MSU) Specimens: **If lowest taxonomic level is above family, lowest classification used for arthropod is indicated Table S4. Voucher specimens deposited at the Albert J. Cook Arthropod Research Collection (Michigan State University). 110 LITERATURE CITED 111 LITERATURE CITED Altieri, M. A. 1999. The ecological role of biodiversity in agroecosystems. Agriculture, Ecosystems and the Environment 74: 19 -31. Atanassov, A. P. W. Shearer, G. Hamilton, and D. Polk. 2002. Development and implementation of a reduced risk peach arthropod management program in New Jersey. Journal of Economic Entomology 95: 803 -812. Bancroft, J. S. 2005. Dispe rsal and abundance of Lygus hesperus in field crops. Environmental Entomology 34: 1517 -1523. Beanland, L., L. V. Madden, C. W. Hoy, S. A. Miller, and L. R. Nault. 2005. Temporal Distribution of aster leafhopper sex ratios and spatial pattern of aster yellows phytoplasma disease in lettuce. Annals of the Entomological Society of America 98: 756 -762. Bianchi, F. J. J. A. and W. van der Werf. 2004. Model evaluation of the function of prey innon crop habitats for biological control of ladybeetles in agricultural landscapes. Ecological Modelling 171: 177 -193. Bianchi, F. J. J. A., C. J. H. Booij, and T. Tscharntke. 2006. Sustainable pest regulation in agricultural landscapes: a review on landscape composition, biodiversity and natural pest control. Proc. Roy. Soc. Lond. B 273: 1715 Ð1727. Blaauw, B. and R. Isaacs. 2012. Larger wildflower plantings increase natural enemy density, diversity, and biological control of sentinel p rey, without increasing herbivore density. Ecological Entomology 37: 386 Ð394. Blackman, R.L. and V.F. Eastop. 2007. Taxonomic Issues. pp. 1 -29. In van Emden, H.F. and Harrington, R. (eds.) Aphids as Crop Pests. CABI, Wallingford, Oxon, UK. Boersma, D. B. , and W. H. Luckmann. 1970. Effects of tarnished plant bug feeding in soybean. J. Econ. Entomol. 63: 253 --256. Boivin, J. -P., R. Le Blanc, and J.A. Adams. 1991. Spatial dispersion and sequential sampling plan for the tarnished plant bug (Hemiptera: Mirida e) on celery. Journal of Economic Entomology 84: 158 -164. Boina, D. R., W. L. Meyer, E. O. Onagbola, and L. L. Stelinksi. 2009. Quantifying dispersal of Diaphorina citri (Hemiptera: Psyllidae) by immunomarking and potential impact of unmanaged groves on commercial citrus management. Environmental Entomology 38: 1250 -1258. Brady, N. C. and R. R. Weil. 2007. Histosols. pp. 99 -100. In The Nature and Properties of Soils (14th edition), Prentice Hall, Upper Saddle River, NJ. 112 Braun L., M. Erlandson, D. Baldwin, J. Soroka, P. Mason, R. Foottit, and D. Hegedus. 2001. Seasonal occurrence, species composition, and parasitism of Lygus spp. in alfalfa, canola, and mustard. The Canadian Entomologist 133: 565 -577. Buntin, G. D., L. P. Pedigo, and W. B. Showers. 1990. Temporal occurrence of the variegated cutworm (Lepidoptera: Noctuidae) adults in Iowa with evidence for migration. Environmental Entomology 19: 603 -608. Byrne, D. N. 1999. Migration and dispersal by the sweet potato whitefly, Bemisia tabaci . Agricultural a nd Forest Meteorology 97: 309 -316. Capinera, J. L. 2001. Handbook of Vegetable Pests. Academic Press (Elsevier), San Diego, CA. Christensen, N. M., K. B. Axelsen, M. Nicolaisen, and A. Schultz. 2005. Phytoplasmas and their interactions with hosts. Trends in Plant Science 10: 526 -535. CIE (Commonwealth Institute of Entomology) . 2001. Distribution Maps of Pests, No. 256. CAB International, Wallingford, Oxon, UK. Clancy, D. W. and H. D. Pierce. 1966. Natural enemies of some lygus bugs. Journal of Economic Entomology 59: 853 -858. Collinge, S.K. 2000. Effects of grassland fragmentation on insect species loss, colonization, and movement Patterns. Ecology 81: 2211 -2226. Conradt, L. and T. J. Roper. 2006. Nonrandom movements behaviou r at habitat boundaries in two butterfly species: implications for dispersal. Ecology 87: 125 Ð132. Cook, W. C., K. T. Lane, B. L. Foster, and R. D. Holt. 2002. Island theory, matrix effects and species richness patterns in habitat fragments. Ecology Lette rs 5:619Ð623. Cowgill, S. E., S. D. Wratten, and N. W. Sotherton. 1993. The effect of weeds on the numbers of hoverfly (Diptera: Syrphidae) adults and the distribution and composition of their eggs in winter wheat. Annals of Applied Biology 123: 499 -515. Cronin, J. T. 2009. Habitat edges, within -patch dispersion of ho sts, and parasitoid oviposition behavior. Ecology 90: 196 -207. Debinski, D. M., and R. D. Holt. 2000. A survey and overview of habitat fragmentation experiments. Conservation Biology 14: 34 2Ð355. Delattre, T., J -B. Pichancourt, F. Burel, P. Kindlmann. 2010. Grassy field margins as potential corridors for butterflies in agricultural landscapes: A simulation study. Ecological Modelling 221: 370 -377. 113 Dillard, H. R. 1992. Colletotrichum coccod es: the pathogen and its hosts. pp 225 -236. In Colletotrichum : Biology, Pathology and Control, J. A. Bailey and M. J. Jeger, eds. CAB International, Wallingford, Oxon, UK. Dixon, A. F. G. 1977. Aphid Ecology - Life -cycles, polymorphism, and population reg ulation. Annual Review of Ecology and Systematics 8: 329 -353. Dixon, A. F. G. 1985. Aphid ecology. Blackie & Son Ltd., Glasgow, Scotland, UK . Duffield, S. T. and N. J. Aebischer. 1994. The effect of spatial scale of treatment with dimethoate on invertebr ate population recovery in winter wheat. Journal of Applied Ecology 31: 263 -281. Emmen, D. A., S. J. Fleischer, and A. Hower. 2004. Temporal and spatial dynamics of Empoasca fabae (Harris) (Homoptera: Cicadellidae) in Alfafa. Environmental Entomology 33: 890-899. Fahrig, L. and G. Merriam. 1994. Conservation of fragmented populations. Conservation Biology 8: 50 Ð59. Fahrig, L. 2003. Effects of Habitat Fragmentation on Biodiversity. Annual Review of Ecology, Evolution, and Systematics 34: 487 -515. Fleischer, S. J. and M. J. Gaylor. 1987. Seasonal abundance of Lygus lineolaris (Heteroptera: Miridae) and selected predators in early season uncultivated hosts: implications for managing movement into cotton. Environmental Entomology 16: 379 -389. Foshee, W. G., R. T. Boozer, E.K. Blythe, D.L. Horton, J. Burkett. 2008. Management of plum curculio and catfacing insects on peaches in central Alabama: standard crop stage -based vs. integrated pest management -based approaches. International Journal of Fruit Sci ence 8: 188-199. Freeman Long, R., A. Corbett, C. Lamb, C. Reberg -Horton, J. Chandler, and M. Stimmann. 1998. Beneficial insects move from flowering plants to nearby crops. California Agriculture 52: 23 Ð26. Frost,K.E., D. K. Willis, and R. L. Groves. 20 11. Detection and Variability of Aster Yellows Phytoplasma Titer in Its Insect Vector, Macrosteles quadrilineatus (Hemiptera:Cicadellidae). Journal of Economic Entomology 104: 1800 -1815. Grafius, E. and E. A. Morrow. 1982. Damage by the tarnished plant bu g and alfalfa plant bug (Heteroptera: Miridae) to asparagus. Journal of Economic Entomology 75: 882 -884. Graham, H. M., C. G. Jackson, and J. W. Debolt. 1986. Lygus spp. (Hemiptera: Miridae) and their parasites in agricultural areas of southern Arizona. E nvironmental Entomology 15: 132 -142. 114 Haddad, N. M. 1999. Corridor use predicted from behaviors at habitat boundaries. The American Naturalist 153: 215 -227. Haddad, N. M., D. R. Bowne, A. Cunningham, B. J. Danielson, D. J. Levey, S. Sargent, and T. Spira. 2003. Corridor use by diverse taxa. Ecology 84: 609 -615. Hagler, J. R. and V. P. Jones. 2010. A protein -based approach to mark arthropods for mark -capture type research. Entomologia Experimentalis et Applicata 135: 177 -192. Hagler, J. R. 2011. An immuno logical approach to quantify consumption of protein -tagged Lygus hesperus by the entire cotton predator assemblage. Biological Control 58: 337 -345. Halbert, S.E. and D.J. Voegtlin. 1992. Morphological differentiation between Aphis spiraecola and Aphis pom i (Homoptera: Aphididae). The Great Lakes Entomologist 25: 1 -8. Hanski, I. 1998. Metapopulation dynamics. Nature 396: 41 -48. Hausbeck, M. 2011. Pest management in the future: a strategic plan for the Michigan celery industry. USDA Regional Pest Management Centers, Center Products/PMSPs. Online. Hickman, J. M. and S. D. Wratten. 1996. Use of Phacelia tanacetifolia strips to enhance biological con trol of aphids by overfly larvae in cereal fields. Journal of Economic Entomology 89: 832 -840. Hilbe, J. M. 2014. Modeling Count Data. Cambridge University Press, New York, NY. Hodek, I., G. Iperti, and M. Hodkova. 1993. Long distance flights in Coccine llidae (Coleoptera). European Journal of Entomology 90: 404 -414. Hogmire, H. W., H. W. Brown, J. J. Schmidt, and T. M. Winfield. 1992. Population development and insecticide susceptibility of apple aphid and spirea aphid (Homoptera: Aphididae) on apple. J ournal of Entomological Science 27: 113 -119. Holt, R. D., 1993. Ecology at the mesoscale: the influence of regional processes on local communities. In: Ricklefs, R.E., D. Schluter (eds.), Species diversity in ecological communities. University of Chicago Press, Chicago, IL, USA, pp. 77 Ð88. Holt, R. D. 2002. Food webs in space: On the interplay of dynamic instability and spatial processes. Ecological Research 17: 261 -273. Holt, R. D., J. H. Lawton, G. A. Polis, and N. D. Martinez. 1999. Trophic rank and the species Ðarea relationship. Ecology 80: 1495Ð1504. 115 Holzschuh, A., I. Steffan -Dewenter, and T. Tscharntke. 2009. Grass strip corridors in agricultural landscapes enhance nest-site colonization by solitary wasps. Ecological Applications 19: 123 -132. Horton, D. R., V. P. Jones, and T.R. Unruh. 2009. Use of a new immunomarking method to assess movement by generalist predators between a cover crop and tree canopy in a pear orchard. American Entomologist 55: 49 -56. Hoy, C. W., S. E. Heady, and T. A. Koch. 1992. Species composition, phenology, and possible origins of leafhoppers (Cicadellidae) in Ohio vegetable crops. Journal of Economic Entomology 85: 2336 -2343. Janzen, D. H. 1983. No park is an island: increase interference from outside as park size decreases. Oikos 41: 402 Ð410. Jones, V. P., J. R. Hagler, J. F. Brunner, C. C. Baker, T. D. Wilburn. 2006. An inexpensive immunomarking technique for studying movement patterns of naturally occurring insect populations. Environmental Entomology 35: 8 27-836. Jones, V. P., T. D. Melton, and C. C. Baker. 2011. Optimizing immunomarking systems and development of a new marking system based on wheat. Journal of Insect Science 11: 1 -16. Jonsen, I. D. and L. Fahrig. 1997. Response of generalist and specialist insect herbivores to landscape spatial structure. Landscape Ecology 12: 185 Ð197. Kelton, L. A. 1975. The Lygus bugs (Genus Lygus Hahn) of North America (Heteroptera: Miridae). Memoirs Entomological Society of Canada 95: 1 -101. Kennedy, J. S. 1985. Migration, behavioural and ecological. pp 5 -26. In Rankin, M.A. (ed.) Migration: Mechanisms and Adaptive Significance. Contributions in Marine Science Supplement Vol. 27. Khattat, A. R. and R. K. Stewart. 1975. Damage by t arnished plant bug to flowers and setting pods of green beans. Journal of Economic Entomology 68: 633 -635. Krauss, J., I. Steffan -Dewenter and T. Tscharntke. 2003. How does landscape context contribute to effects of habitat fragmentation on diversity and population density of butterflies? Journal of Biogeography, 30, 889 Ð900. Krishnamurthy, K. S. 2008. Celery, pp 401 -412. In Parthasarathy, V.A., Chempakam, B., and Zachariah, T. J., (eds.) Chemistry of Spices. CABI Publishing, Wallingford, Oxon, UK. Krues s, A. and T. Tscharntke. 1994. Habitat fragmentation, species loss, and biological control. Science 264: 1581 Ð1584. 116 Lacy, M. L., R.D. Berger, R.L. Gilberson, and E.L. Litle. 1996. Current challenges in controlling diseases of celery. Plant Disease 80:1084 -1091. Landis, D. & P. Marino. 1999. Landscape structure and extrafield processes: impact on management of pests and beneficials. pp 79 Ð104. In J. R. Ruberson (ed.) Handbook of pest management, Marcel Dekker Inc, New York, NY. Landis, D.A., S. D. Wratten , and G. M. Gurr. 2000. Habitat management to conserve natural enemies of arthropod pests in agriculture. Annual Review of Entomology 45: 175 Ð201. Lavandero, A., S. Wratten, J. Hagler, and M. Jervis. 2004. The need for effective marking and tracking techniques for monitoring the movements of insect predators and parasitoids. International Journal of Pest Management 50: 147 -151. Lee, J. C., F. D. Menalled, and D. A. Landis. 2001. Refuge habitats modify impact of insecticide disturbance on carabid beetle co mmunities. Journal of Applied Ecology 38: 472 -483. Leibold, M. A., M. Holyoak, N. Mouquet, P. Amarasekare, J. M. Chase, M. F. Hoopes, R. D. Holt, J. B. Shurin, R. Law, D. Tilman, M. Loreau, A. Gonzalez. 2004. The metacommunity concept: a framework for multi -scale community ecology. Ecology Letters 7: 601 Ð613. Leigh, T. F. and D. Gonzalez. 1976. Field cage evaluation of predators for control of Lygus hesperus Knight on cotton. Environmental Entomology 5: 948 -952. Levine, R. L. 2013. Celery, pp. 245 -248. In Cumo, Christopher (ed.) Encyclopedia of Cultivated Plants. ABC -CLIO, Santa Barbara, CA. Levins, R., 1969. Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entom ological Society of America 15: 237 Ð240. Lindbo, D. L. and D. A. Kozlowski. 2005. Histosols, pp. 830 -834. In Lal, Rattan (ed.) Encyclopedia of Soil Science (2 nd edition), CRC Press, Boca Raton, FL. Lowery, T., R. Footit, M. Smirle, and E. Peryea. 2002. Identification, biology, and insecticide resistance of green apple aphid and spirea aphid. pp 148 -156. In Entomology Research Review. Washington Tree Fruit Research Commission. Yakima, WA. Lucas, R. E. 1982. Organic soils (Histosols) formation, distributio n, physical, and chemical properties and management for crop production. In Farm Science Research Report 435, Michigan State University Agricultural Experiment Station, East Lansing, MI MacArthur, R. H., and E. O. Wilson. 1967. The theory of island biogeo graphy. Princeton University Press, Princeton, N.J. 117 Marino, P. C. and D. A. Landis. 1996. Effect of landscape structure on parasitoid diversity and parasitism in agroecosystems. Ecological Applications 6: 276 Ð284. Meade, A. B., and A. G. Peterson. 1964. Origin of populations of the six -spotted leafhopper, Macrosteles fascifrons , in Anoka County, Minnesota. J. Economic Entomology 57: 885 -8. Morrison III, W. R. and Z. Szendrei. 2013. Patterns of Spatial and Temporal Distribution of the Asparagus Miner (Dip tera: Agromyzidae): Implications for Management. Journal of Economic Entomology 106: 1218 -1225. Nicholls, C. I., M. Parrella, and M. A. Altieri. 2001. The effects of a vegetational corridor on the abundance and dispersal of insect biodiversity within a no rthern California organic vineyard. Landscape Ecology 16: 133 Ð146. Norton, A. P., S. C. Welter, J. L. Flexner, C. G. Jackson, J. W. Debolt, and C. Pickel. 1992. Parasitism of Lygus hesperus (Miridae) by Anaphes iole (Mymaridae) and Leiophron uniformis (Braconidae) in California Strawberry. Biological Control 2: 131 -137 Peck, G. W., H. J. Ferguson, V. P. Jones, S. D. OÕNeal, and D. B. Walsh. 2014. Use of a highly sensitive immunomarking system to characterize face fly (Diptera: Muscidae) dispersal from c ow pats. Environmental Entomology 43: 116 -122. Peres, N. A., L. W. Timmer, J. E. Adaskaveg, and J. C. Correll. 2005. Lifestyles of Colletotrichum acutatum . Plant Dis. 89:784 -796. Pickett, C. H., S. L. Swezey, D. J. Nieto, J. A. Bryer, M. Erlandson, H. Goulet, and M. D. Schwartz. 2009. Colonization and establishment of Peristenus relictus (Hymenoptera: Braconidae) for control of Lygus spp. (Hemiptera: Miridae) in strawberries on the California Central Coast. Biological Control 49: 27 -37. Poston, F. L. and L. P. Pedigo. 1975. Migration of plant bugs and the potato leafhopper in a soybean-alfalfa complex. Environmental Entomology 4: 8 -10. Pulliam, H. R., 1988. Sources, sinks, and population regulation. American Naturalist 132: 652 Ð661. Rand, T. A. and S. M. Louda. 2006 Spillover of Agriculturally Subsidized Predators as a Potential Threat to Native Insect Herbivores in Fragmented Landscapes. Conservation Biology 20: 1720 Ð1729. Rand, T., J. M. Tylianakis, and T. Tscharntke. 2006. Spillover edge effects: t he dispersal of agriculturally-subsidized insect natural enemies into adjacent natural habitats. Ecology Letters 9: 603 Ð614. 118 Ridgeway, R. L. and G. G. Gyrisco. 1960. Effect of temperature on the rate of development of Lygus lineolaris (Hemiptera: Miridae) . Annals of the Entomological Society of America 53: 691-694. Rodr™guez -Salamanca, L. M., T. B. Enzenbacher, J. M. Byrne, C. Feng, J. C. Correll, and M. K. Hausbeck. 2012. First report of Colletotrichum acutatum causing leaf curling and petiole anthracnos e on celery (Apium graveolens ) in Michigan. Plant Disease 96:1383. Rosch, V., T. Tscharntke, C. Scherber, and P. Batary. 2013. Landscape composition, connectivity and fragment size drive effects of grassland fragmentation on insect communities. Journal of Applied Ecology 50: 387 -394. Rosenzweig, M. L. 1995. Species Diversity in Space and Time. Cambridge University Press, Cambridge, UK. Scales, A. L. 1973. Parasites of the Tarnished Plant Bug in the Mississippi Delta. Environmental Entomology 2: 304 -306. Schaeffers, G. A. 1980. Yield effects of tarnished plant bug feeding on June -bearing strawberry varieties in New York State. Environmental Entomology 73: 721 Ð725. Schellhorn, N. A., F. J. J. A. Bianchi, and C.L. Hsu. 2014. Movement of entomophagous arthropods in agricultural landscapes: links to pest suppression. Annual Review of Entomology 59: 559 -81. Schowalter, T. D. 2012. Insect responses to major landscape -level disturbance. Annual Review of Entomology 57: 1 Ð20. Seymour, L. M., T. M. Mowry, W. H. Day, and J. D. Barbour. 2005. Parasitism of Lygus spp. nymphs by the parasitoid wasp, Peristenus howardi , in the alfalfa seed -growing region of the Pacific Northwest. Journal of Insect Science 5: 1 -8. Shmida, A. and M. V. Wilson. 1985. Biological de terminants of species diversity. Journal of Biogeography 12: 1 -20. Smirle, M. J. C. L. Zurowski, D. T. Lowery, and R. G. Footit. 2010. Relationship of Insecticide Tolerance to Esterase Enzyme Activity in Aphis pomi and Aphis spiraecola (Hemiptera: Aphidid ae). Journal of Economic Entomology 103: 374 -378. Snodgras, G. L., W. P. Scott, and J. W. Smith. 1984. Host Plants and Seasonal Distribution of the Tarnished Plant Bug (Hemiptera: Miridae) in the Delta of Arkansas, Louisiana, and Mississippi. Environment al Entomology 13: 110 -116. Snodgrass, G. L., W. P. Scott, C. A. Abel, J. T. Robbins, J. Gore, and D. D. Hardee. 2005. Tarnished Plant Bug (Heteroptera: Miridae) Populations near fields after early season herbicide treatment. Environmental Entomology 34: 7 05-711. 119 Steffan -Dewenter I, U. Munzenberg, C. Burger, C. Thies, and T. Tscharntke. 2002. Scale dependent effects of landscape structure on three pollinator guilds. Ecology 83: 1421 Ð32. Stewart, R. K. and H. Khoury. 1976. The biology of Lygus lineolaris (Palisot de Beauvois) (Hemiptera: Miridae) in Quebec. Annals of the Entomological Society of Quebec 21: 52 -63. Suter, R. B. 1999. An aerial lottery: the physics of ballooning in a chaotic atmosphere. Journal of Arachnology 27: 281 -293. Swezey, S. L., D. J . Nieto, and J. A. Bryer. 2007. Control of western tarnished plant bug Lygus hesperus Knight (Hemiptera: Miridae) in California organic strawberries using alfalfa trap crops and tractor-mounted vacuums. Environmental Entomology 36: 1457 -1465. Swezey, S. L ., D. J. Nieto, J. R. Hagler, C. H. Pickett, J. A. Bryer , and S. A. Machtley. 2013. Dispersion, Distribution, and Movement of Lygus spp. (Hemiptera: Miridae) in Trap -Cropped Organic Strawberries. Environmental Entomology 42: 770 -778. Taylor, L. R. 1974. Insect migration, flight periodicity, and the boundary layer. Journal of Animal Ecology 43: 225 -238. Thies, C., I. Steffan -Dewenter, T. Tscharntke. 2003. Effects of landscape context on herbivory and parasitism at different spatial scales. Oikos 101: 18 Ð25. Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology 75: 2 Ð16. Tscharntke T, I. Steffan -Dewenter, A. Kruess, and C. Thies. 2002 . Contribution of small habitat fragments to conservation of insect communities of grassland -cropland landscapes. Ecological Applications 12:354 Ð63. Tscharntke, T. and R. Brandl. 2004. Plant -insect interactions in fragmented landscapes. Annual Review of Entomology 49: 405 -430. Tscharntke, T., A. M. Klein, A. Kruess, I. Steffan -Dewenter , and C. Thies. 2005a. Landscape perspectives on agricultural intensification and biodiversity Ðecosystem service management. Ecology Letters 8: 857 Ð874. Tscharntke, T., T. A. Rand, and F. J. J. Bianchi. 2005b. The landscape context of trophic interaction s: insect spillover across the crop Ðnoncrop interface. Ann. Zool. Fennici 42: 421 Ð432. Tscharntke T., R. Bommarco, Y. Clough, T. O. Crist, D. Kleijn, T. A. Rand, J. M. Tylianakis, S. van Nouhuys, and S. Vidal. 2007. Conservation biological control and enemy diversity on a landscape scale. Biological Control 43: 294 -309. 120 Tylianakis, J. M., A. M. Klein, and T. Tscharntke. 2005. Spatiotemporal variation in the effects of a tropical habitat gradient on Hymenoptera diversity. Ecology 86: 3296 Ð3302. Tylianakis , J. M., A. M. Klein, T. Lozada, and T. Tscharntke, 2006. Spatial scale of observation affects !, ", and #, diversity of cavity -nesting bees and wasps across a tropical land -use gradient. Journal of Biogeography 33: 1295 Ð1304. Van Dyck, H. and M. Baguette . 2005. Dispersal behaviour in fragmented landscapes: routine or special movements? Basic and Applied Ecology 6: 535 Ð545. van Helden, M. 2010. Spatial and Temporal Dynamics of Arthropods in Arable Fields. pp. 55 -64. In Oerke, E -C., Gerhards, R., Gunter, M ., and Sikora, R.A. (eds.) Precision crop protection - the challenge and use of heterogeneity. Springer, Dordrecht, Heidelberg, London, New York. van Nouhuys, S. and I. Hanski. 2002. Colonization rates and distances of a host butterfly and two specific pa rasitoids in a fragmented landscape. Journal of Animal Ecology 71: 639 Ð650. Wagge, J. K. 1978. Arrestment responses of the parasitoid, Nemeritis canescens , to a contact chemical produced by its host, Plodia interpunctella . Physiological Entomology 3: 135 -146. Wagge, J. K. 1979. Foraging for patchily -distributed hosts by the parasitoid, Nemeritis canescens. Journal of Animal Ecology 48: 353 -371. Wackers, F. L. 2001. A comparison of nectar and honeydew sugars with respect to their utilization by the hymeno pteran parasitoid Cotesia glomerata . Journal of Insect Physiology 47: 1077 Ð1084. Wackers, F. L. 2004. Assessing the suitability of flowering herbs as parasitoid food sources: flower attractiveness and nectar accessibility. Biological Control 29: 307 Ð314. Wallis, E. L. 1962. Host plant preference of the six -spotted leafhopper. Journal of Economic Entomology 55: 998 -999. Webber, D. C. and J. G. Lundgren. 2009. Assessing the trophic ecology of the Coccinellidae: Their roles as predators and as prey. Biological Control 51:199 -214. Westdal P. H., C. F. Barrett, H. P. Richardson. 1961. The six-spotted leafhopper, Macrosteles fascifrons (ST†L.) and Aster Yellows in Manitoba. Canadian Journal of Plant Science 41: 320-331. Whalon, M. E. and B. L. Parker. 1978. Immunological identification of tarnished plant bug predators. Annals of the Entomological Society of America 71: 453 -456. 121 Winder, L., J. N. Perry, and J. M. Holland. 1999. The spatial and temporal distribution of the grain aphid Sitobion avenae in winter wheat. Entomologia Experimetalis et Applicata 93: 277-290. Winkler, K., F. Wackers, and D. M. Pinto. 2009. Nectar -providing plants enhance the energetic state of herbivores as well as their parasitoids under field conditions. Ecological Entomology 34: 221Ð227. Winkler, K., F. L. Wackers, A. J. Termorshuizin, J. C. van Lenteren. 2010. Assessing risk and benefits of floral supplements in conservation biological control. Biocontrol 55: 719 -727. Wissinger, S. A. 1997. Cyclic colonization in predictab ly ephemeral habitats: a template for biological control in annual crop systems. Biological Control 10: 4 -15. Wong G., M. Ngouajio, and K S. Charles. 2010. Brassica biofumigants improve onion and celery production systems. Journal of Sustainable Agricultu re 34: 2-14. Young, O. P. 1986. Host Plants of the Tarnished Plant Bug, Lygus lineolaris (Heteroptera: Miridae). Annals of the Entomological Society of America 79: 747 -762. Zandstra, B. H., T. A. Dudek, E. J. Grafius, M. L. Lacy, and D. D. Warncke. 1986. Commercial recommendations: celery. Michigan State University Extension Bulletin 1308. Zandstra, B. H. 2012. Celery Weed Control Update. Great Lakes Fruit, Vegetable, & Farm Market Expo., Celery education session. Grand Rapids, MI.