Entomology Publications Entomology

4-2014 Quality Over Quantity: Buffer Strips can be Improved with Select Native Species Kelly Ann Gill Iowa State University, [email protected]

R. Cox Centro Internacional del Mejoramiento de Maíz y Trigo

Matthew E. O'Neal Iowa State University, [email protected]

Follow this and additional works at: http://lib.dr.iastate.edu/ent_pubs Part of the Biodiversity Commons, Entomology Commons, and the Terrestrial and Aquatic Ecology Commons The ompc lete bibliographic information for this item can be found at http://lib.dr.iastate.edu/ ent_pubs/48. For information on how to cite this item, please visit http://lib.dr.iastate.edu/ howtocite.html.

This Article is brought to you for free and open access by the Entomology at Iowa State University Digital Repository. It has been accepted for inclusion in Entomology Publications by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Quality Over Quantity: Buffer Strips can be Improved with Select Native Plant Species

Abstract Native attractive to beneficial insects may improve the value of buffer strips by increasing biodiversity and enhancing the delivery of insect-derived ecosystem services. In a 2-yr field experiment, we measured the response of insect communities across nine buffers that varied in plant diversity. We constructed buffers with plants commonly found in buffers of USDA-certified organic farms in Iowa (typically a single species), recommended for prairie reconstruction, or recommended for attracting beneficial insects. We hypothesized that the diversity and abundance of beneficial insects will be 1) greatest in buffers composed of diverse plant communities with continuous availability of floral resources, 2) intermediate in buffers with reduced species richness and availability of floral resources, and 3) lowest in buffers composed of a single species. We observed a significant positive relationship between the diversity and abundance of beneficial insects with plant community diversity and the number of . More beneficial insects were collected in buffers composed of species selected for their attractiveness to beneficial insects than a community recommended for prairie restoration. These differences suggest 1) plant communities that dominate existing buffers are not optimal for attracting beneficial insects, 2) adding flowering perennial species could improve buffers as habitat for beneficial insects, 3) buffers can be optimized by intentionally combining the most attractive native species even at modest levels of plant diversity, and 4) plant communities recommended for prairie reconstruction may not contain the optimal species or density of the most attractive species necessary to support beneficial insects from multiple guilds.

Keywords Habitat management, Floral provisioning, Biodiversity-ecosystem function, Pollination, Biological control

Disciplines Biodiversity | Entomology | Terrestrial and Aquatic Ecology

Comments This article is from Environmental Entomology 43 (2014): 298–311, doi:10.1603/EN13027. Posted with permission.

Rights This article is the copyright property of the Entomological Society of America and may not be used for any commercial or other private purpose without specific written permission of the Entomological Society of America.

This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/ent_pubs/48 Quality Over Quantity: Buffer Strips can be Improved with Select Native Plant Species Author(s): K. A. Gill , R. Cox , and M. E. O'Neal Source: Environmental Entomology, 43(2):298-311. 2014. Published By: Entomological Society of America URL: http://www.bioone.org/doi/full/10.1603/EN13027

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. COMMUNITY AND ECOSYSTEM ECOLOGY Quality Over Quantity: Buffer Strips Can be Improved With Select Native Plant Species

1 2 1,3 K. A. GILL, R. COX, AND M. E. O’NEAL

Environ. Entomol. 43(2): 298Ð311 (2014); DOI: http://dx.doi.org/10.1603/EN13027 ABSTRACT Native plants attractive to beneÞcial insects may improve the value of buffer strips by increasing biodiversity and enhancing the delivery of insect-derived ecosystem services. In a 2-yr Þeld experiment, we measured the response of insect communities across nine buffers that varied in plant diversity. We constructed buffers with plants commonly found in buffers of USDA-certiÞed organic farms in Iowa (typically a single species), recommended for prairie reconstruction, or recommended for attracting beneÞcial insects. We hypothesized that the diversity and abundance of beneÞcial insects will be 1) greatest in buffers composed of diverse plant communities with continuous availability of ßoral resources, 2) intermediate in buffers with reduced species richness and availability of ßoral resources, and 3) lowest in buffers composed of a single species. We observed a signiÞcant positive relationship between the diversity and abundance of beneÞcial insects with plant community diversity and the number of ßowers. More beneÞcial insects were collected in buffers composed of species selected for their attractiveness to beneÞcial insects than a community recommended for prairie restoration. These differences suggest 1) plant communities that dominate existing buffers are not optimal for attracting beneÞcial insects, 2) adding ßowering perennial species could improve buffers as habitat for beneÞcial insects, 3) buffers can be optimized by intentionally combining the most attractive native species even at modest levels of plant diversity, and 4) plant communities recommended for prairie reconstruction may not contain the optimal species or density of the most attractive species necessary to support beneÞcial insects from multiple guilds.

KEY WORDS habitat management, ßoral provisioning, biodiversityÐecosystem function, pollina- tion, biological control

The diversity and abundance of beneÞcial insects are A loss of native plant diversity is evident across the positively inßuenced by plant-derived resources such midwestern United States. Historically, the state of as nectar, pollen, nesting substrates, and overwinter- Iowa was dominated (Ϸ79.5%) by tallgrass prairie ing sites surrounding cultivated land (Westrich 1996, ecosystems, but during the past 150 yr most of IowaÕs Elliott et al. 2002, Steffan-Dewenter et al. 2002, Landis native vegetation has been replaced by agricultural et al. 2005, Klein et al. 2007, Zhang et al. 2007, Kwaiser systems, and now Ͻ0.1% of native prairie remains in and Hendrix 2008, Tscharntke et al. 2008, Wackers et the state (Samson and Knopf 1994, Smith 1998). In al. 2008, Le Fe´on et al. 2011). Patches of noncrop general, such a loss of plant diversity in an agricultural vegetation within agricultural landscapes can provide landscape reduces the resources required by beneÞ- these resources, allowing beneÞcial insects to persist cial insects to survive and deliver ecosystem services near agricultural Þelds before, during, and after peri- to surrounding crops (Landis et al. 2005). Reconstruc- ods when insect-derived ecosystem services are pro- tion of prairie plant communities in agricultural land- vided to annual crops (for reviews, see Landis et al. scapes can conserve beneÞcial insects by increasing 2000, Bianchi et al. 2006, Isaacs et al. 2009). Plant the amount of perennial habitat surrounding annual resources are exploited at varying times and levels cropping systems. Many of the perennial plant species across different guilds of beneÞcial insects (pollina- found in prairies are attractive to beneÞcial insects tors, predators, and parasitoids), making the season- (Fiedler and Landis 2007a, Frank et al. 2008, Tuell et long availability of noncrop vegetation an important al. 2008), although variation exists in the attractiveness component of agricultural landscapes. to insects among individual plant species. Species composition of the plant community likely inßuences the extent to which reconstructed prairie can con- 1 Department of Entomology, Iowa State University, Ames, IA tribute to the conservation of beneÞcial insects and 50011. delivery of insect-derived ecosystem services (Fiedler 2 Centro Internacional del Mejoramiento de Maõ´z y Trigo (CIMMYT), and Landis 2007b, Tuell et al. 2008, Isaacs et al. 2009). Km. 45, Carretera, Mexico-Veracruz, El Bata´n, Texcoco, Edo. de Me´xico, CP 56130 Me´xico. Buffer strips are typically recognized for their role 3 Corresponding author, e-mail: [email protected]. in soil and water conservation practices (e.g., grass

0046-225X/14/0298Ð0311$04.00/0 ᭧ 2014 Entomological Society of America April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 299

Þlter strips, riparian buffers; Clark and Reeder 2007). randomly assigned to plots using a randomized com- Recently, buffer practices have been incorporated plete block design. into requirements for organic production systems. Re- Buffer Treatments. We constructed nine different quirement ¤ 205.202 ([USDAÐNOP] United States buffers to test our hypotheses regarding the impact of Department of AgricultureÐNational Organic Pro- plant diversity on the diversity and abundance of ben- gram 2009) states that buffers are required of organic eÞcial insects. The nine buffer treatments were con- producers seeking certiÞcation from the USDA when structed from plants used in buffers of certiÞed or- organically managed land is adjacent to land not under ganic farms in Iowa (simple buffers), recommended organic management. The stated purpose of this man- for prairie reconstruction (diverse buffers), or rec- datory buffer zone is to prevent the unintended ap- ommended for attracting beneÞcial insects (diverse plication of a prohibited substance (e.g., pesticide and and forb-only buffers). pollen from genetically modiÞed crops) to the crop Simple Buffers. Input from organic farmers was or contact with a prohibited substance applied to used to develop a subset of four buffer treatments. In adjoining land that is not under organic manage- 2008, we surveyed organic farmers who were certiÞed ment (USDAÐNOP 2009). There are no speciÞca- by the top three certifying agencies in Iowa (Midwest tions in this requirement regarding the composition Organic Services Association, Inc., Organic Crop Im- of vegetative buffer zones. Reconstructing native provement Association, and the Iowa Department of perennial plant communities in buffer zones may in- Agriculture and Land Stewardship). From this survey crease biodiversity and ecosystem services, improving (data not shown, R. Cox), we determined most or- the value of buffers. ganic farms (72%) had perennial grasses or a crop Our goal is to develop best management practices species in their buffer strips. These data informed for designing and establishing perennial multi-species which plant species were used in the four buffer treat- buffers that are compatible with agricultural land- ments composed of a single plant species (hereafter scapes and attractive to beneÞcial insects. In a simple referred to collectively as “simple buffers”). Individual garden-variety experiment conducted over 2 yr, we plant species or cultivars were selected using the fol- compared the response of insect communities to buf- lowing considerations: 1) crop and noncrop species fers composed of individual plant species commonly used commonly by organic producers who completed found in buffers on USDA-certiÞed organic farms in our survey, 2) species compatible with local agricul- Iowa, and buffers composed of multiple perennial tural Þeld conditions (e.g., full-sun, noninvasive), and plants. We included a buffer composed of plants rec- 3) species that have ecological and economic beneÞts ommended for prairie reconstruction, and an equally in addition to potentially conserving beneÞcial insects diverse buffer composed of prairie plants selected for (e.g., erosion control, crops harvested and sold as their attractiveness to beneÞcial insects. Altogether, conventionally produced, and species that may be we constructed nine buffer treatments with increasing used or sold as forage). The four simple buffer treat- plant diversity and availability of ßoral resources. We ments are monocultures of switchgrass (Panicum vir- hypothesized that the diversity and abundance of ben- gatum L.), alfalfa (Medicago sativa L.), willow (Salix eÞcial insects in these buffers would be 1) greatest in matsudana Koidzumi), and corn (Zea mays L.). buffers with diverse plant communities with contin- Diverse Buffers. We established two buffer treat- uous availability of ßoral resources, 2) intermediate in ments as diverse plant communities (hereafter re- buffers with reduced plant species richness and avail- ferred to collectively as “diverse buffers”) to test the ability of ßoral resources, and 3) lowest in buffers hypothesis that buffers composed of multiple plant composed of a single species. species would attract a more diverse and abundant community of beneÞcial insects than simple buffers. Diverse plant communities were composed of a mix- ture of grasses and forbs that met the following cri- Materials and Methods teria: 1) perennial species native to the north-central Site Description and Experimental Design. The region of the United States; 2) species that, in com- study site was established at Iowa State UniversityÕs bination, produce ßowers throughout the growing Field Extension Education Laboratory (FEEL) lo- season; 3) species with low to moderate aggressive cated in Boone County, IA (42Њ 00.318Ј N, 93Њ 47.272Ј growth; and 4) species commercially available in local W). The site is a 17-ha demonstration farm divided genotypes (ecotypes). into multiple plots devoted to crop-related research. We established one diverse buffer treatment Adjacent Þelds and the surrounding landscape were based on recommendations from the Iowa Natural composed of corn and soybean crops. On 23 June 2009, Resources Conservation Service ([USDAÐNRCS] we constructed 36 garden-style plots measuring 2 by United States Department of AgricultureÐNatural Re- 2 m bordered by 5- by 15-cm pressure-treated lumber. source Conservation Service 2010]; Table 1). This These plots were distributed along a 55- by 24-m bare- treatment is referred to as the “CP-IA buffer,” as the soil Þeld in a grid formation of four blocks (oriented species in this buffer were identiÞed for conservation west to east) with nine plots per block. Nine buffer targeting restoration of rare and declining habitats treatments were designed with plant communities (e.g., IowaÕs native tallgrass prairie). Goals of this that vary in diversity and complexity (described be- practice include increasing plant diversity and pro- low). Each treatment was replicated four times and viding habitat and food for wildlife. The description 300 ENVIRONMENTAL ENTOMOLOGY Vol. 43, no. 2

Table 1. Species selected for the CP-IA buffer and their associated characteristics

Common namea ScientiÞc name Bloom timeb Growth habitc Spotted geranium Geranium maculatum L. MayÐJune FB Pale purple coneßower Echinacea pallida (Nutt.) Nutt. JuneÐJuly FB Blackeyed susan Rudbeckia hirta L. JuneÐAug. FB Smooth oxeye Heliopsis helianthoides (L.) Sweet JuneÐAug. FB CulverÕs root Veronicastrum virginicum (L.) Farw. JuneÐAug. FB Purple prairie clover Dalea purpurea Vent. JuneÐAug. LG-FB Canada wildrye Elymus canadensis L. JulyÐSept. CS-GR Sideoats grama Bouteloua curtipendula (Michx.) Torr. JulyÐSept. WS-GR Big bluestem Andropogon gerardi Vitman Aug.ÐSept. WS-GR Indiangrass Sorghastrum nutans (L.) Nash Aug.ÐSept. WS-GR Rough dropseed Sporobolus clandestinus (Biehler) Hitchc. Aug.ÐSept. WS-GR Prairie sage Artemisia ludoviciana Nutt. Aug.ÐOct. SS ShortÕs aster Symphyotrichum shortii (Lindl.) G.L. Nesom Aug.ÐOct. FB Stiff goldenrod Oligoneuron rigidum (L.) Small Aug.ÐOct. FB

a The CP-IA buffer includes 14 native perennial species: Þve grasses, eight forbs, and a shrublet, which are a subset of species selected from conservation practice CP25, that in combination, bloom throughout the season. b Species are ordered by bloom periods (earliestÐlatest for Iowa) when conspicuous ßowers or inßorescences are present. The duration of ßowering can be from 3 wk to 3 mo depending on the species and environmental conditions. c Growth habit codes indicate functional groups: FB, forb/herb; LG-FB, leguminous forb; CS-GR, cool-season graminoid; WS-GR, warm- season graminoid; SS, sub shrub. also indicates healthy prairie habitats can be a source 2012]; details available at nativeplants.msu.edu; Table of ßowers for pollinating insects (USDAÐNRCS 2010), 2). but it is unclear how attractive this buffer would be for Forb-Only Buffers. Three buffer treatments were beneÞcial insects from multiple guilds. This buffer was established with only forbs (Table 3) to assess the not designed with the primary goal of increasing the response of beneÞcial insects to plant communities diversity and abundance of and natural enemies, with a reduction in plant species richness and resource and its ability to do so has not yet been tested. availability. We hypothesized that beneÞcial insect A second, diverse buffer treatment was established diversity and abundance will be intermediate in these to test the hypothesis that diverse plant communities forb-only buffers. The selection criteria for these spe- can be optimized to attract beneÞcial insects. This cies were consistent with (1Ð4) of the diverse buffers, buffer was designed to increase the diversity and using only the most attractive forbs from the MSU Best abundance of bees and natural enemies. We selected Bet buffer. The most species rich of these treatments, a combination of plant species identiÞed by Fiedler referred to as “MSU5,” contained Þve species of forbs, and Landis (2007a) and Tuell et al. (2008), who iden- which provided ßowering resources from two or more tiÞed species that were highly attractive to natural species blooming throughout the growing season. enemies and bees and exhibited relatively low attrac- Treatments referred to as “MSU3” and “MSU2” were tiveness to arthropod pests. Twelve species were se- systematic reductions of the MSU5 treatment, de- lected for the “MSU Best Bet” buffer (honoring the signed by reducing the phenological overlap of species work conducted at Michigan State University [MSU in bloom.

Table 2. Species selected for the MSU Best Bet buffer and their associated characteristics

Common namea ScientiÞc name Bloom timeb Growth habitc Canadian anemone Anemone canadensis L. MayÐJune FB Meadow ziziad Zizia aptera (A. Gray) Fernald MayÐJune FB Pinnate prairie coneßower Ratibida pinnata (Vent.) Barnhart JuneÐAug. FB Swamp milkweed Asclepias incarnata L. JuneÐAug. FB Switchgrass Panicum virgatum L. JulyÐAug. WS-GR Canada wildrye Elymus canadensis L. JulyÐSept. CS-GR Common boneset Eupatorium perfoliatum L. JulyÐSept. FB Cup plant Silphium perfoliatum L. JulyÐSept. FB Prairie ironweedd fasciculata Michx. JulyÐSept. FB Little bluestem Schizachyrium scoparium (Michx.) Nash Aug.ÐOct. WS-GR New England aster Symphyotrichum novae-angliae (L.) G.L. Nesom Aug.ÐOct. FB Smooth blue aster Symphyotrichum laeve (L.) A´ .Lo¨ve&D.Lo¨ve Aug.ÐOct. FB

a The MSU Best Bet buffer includes 12 native perennial species: three grasses and nine forbs, selected based on those individually rated “Best” for relative attractiveness to either (or both) natural enemies and bees in evaluations by Fiedler and Landis (2007) and Tuell et al. (2008). Selection was further restricted to Iowa ecotypes that, in combination, bloom throughout the season. b Species are ordered by bloom periods (earliestÐlatest for Iowa) when conspicuous ßowers or inßorescences are present. The duration of ßowering can be from 3 wk to 3 mo depending on the species and environmental conditions. c Growth habit codes indicate functional groups: FB, forb/herb; CS-GR, cool-season graminoid; WS-GR, warm-season graminoid. d Vernonia missurica and Zizia aurea used by Fiedler and Landis (2007) were not available as plugs through our local provider and were replaced with V. fasciculata and Z. aptera; similar species in the same genus. April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 301

Table 3. Species included in forb-only buffers Field and Plot Maintenance. A 4-m distance was maintained between each plot in all directions to allow a ScientiÞc Buffer treatments Common name b for mowing between plots. All plots were mulched name MSU5 MSU3 MSU2 once in late October 2009, with clippings of clean oat Meadow zizia Zizia aptera XXXstraw to control weeds and protect establishing seed- Swamp milkweed Asclepias incarnata XX lings (plugs) from frost and animal damage. The straw Pinnate prairie Ratibida pinnata X was removed during early April 2010 before the es- coneßower Cup plant Silphium perfoliatum XXXtablishment of simple buffer treatments. Annual rye- New England aster Symphyotrichum X grass (Lolium multiflorum Lamarck) was sown as novae-angliae ground cover between plots on 24 May 2010 and

a mowed throughout the sampling period. Plots were Forbs from the MSU Best Bet buffer were selected to create three not mulched with straw after the 2010 growing season, additional treatments. The planting density of each forb-only buffer remained the same as the diverse buffers, but species richness was as a thatch layer from Þrst-year plant material was left reduced. Bloom periods and growth habits are as in Table 2. An “X” in plots. In both years, weeds were removed between indicates species present in each forb-only buffer. b and within plots to maintain species composition with Authors for species as in Table 2. special attention to weed removal immediately before insect sampling. Plant Measurements. Plant diversity, percent Plant Establishment. Switchgrass plugs (1-yr-old ground cover, canopy height, and ßower abundance plants, Ion Exchange Inc., Harpers Ferry, IA) were were measured in each buffer treatment to determine transplanted by hand on 21 April 2010. Plugs were whether plant characteristics account for variation in planted at a density of one plug per 929 sq cm resulting beneÞcial insect diversity and abundance. The num- in 25 plants per replicate plot. Plugs were positioned ber of plants and plant species per plot were counted 15 cm from the plot borders on all sides, and 30 cm and SimpsonÕs diversity index (1/D) was calculated spacing was maintained between plants. for each plot (Simpson 1949). Analyses were based on Alfalfa seed was purchased locally (BrekkeÕs Town Þnal measurements taken in August 2010 and 2011 to and Country Store, Ames, IA) and sown on 9 April represent the maximum end-of-season plant diversity 2010. Seed was hand broadcast using the standard rate and to account for the annual establishment of corn of 8Ð9 kg per ha, resulting in 0.009 kg of seed per plants. For each year, resulting diversity values were replicate plot. Owing to the small amount of seed summed among replicates and mean plant diversity being used, seeds for each plot were weighed, por- was calculated per buffer treatment. SimpsonÕs diver- tioned, and combined with coarse sand to add bulk to sity indices were calculated using the “vegan” package the material to ensure an even distribution when version 2.0Ð1 in R version 2.14.1 (Oksanen et al. 2011, broadcasting. R Development Core Team 2011). Willow cuttings were taken from established willow Buffer treatments were designed to achieve varia- stands (Small Potatoes Farm, Minburn, IA) in Febru- tion in the amount and timing of ßoral resources. To ary 2010. Once the root mass was adequately devel- determine whether we achieved this variation, the oped, shrubs were obtained from the farm and trans- number of ßowers was counted two times per month, planted on 21 April 2010. Willow shrubs ranging from coinciding with arthropod collection. abun- 61 to 91 cm in height were planted at a density of three dance was measured for plots containing conspicuous shrubs per replicate in a triangle formation with 122 ßowers; therefore, corn and switchgrass were not cm spacing between each shrub. measured. The bloom period of willows preceded the Corn seed (DEKALB DKC 61Ð72 Roundup Ready annual establishment of corn plants, and therefore, Corn; Monsanto Co., St. Louis, MO) was sown by hand our sampling period, so willows also were excluded on 7 May 2010 and 11 May 2011 with three rows per from ßoral measurements. replicate plot and 15-cm plant spacing and 76-cm row For buffer treatments with conspicuous ßowers, spacing, resulting in a density of 35 plants per replicate ßower abundance was measured by counting the plot. In 2010, corn plants that did not emerge were number of individual open ßowers on each plant. replanted on 1 June 2010. Individual ßowers were deÞned as ßower heads for Plugs (1-yr-old plants, Ion Exchange Inc., Harpers spp. and Geraniaceae spp., umbels per Ferry, IA) were used to establish diverse and forb-only cluster for Asclepiadaceae spp., solitary ßowers for plant communities. All plugs were transplanted by Ranunculaceae spp., and spikes and racemes for Scro- hand on 16 September 2009. These Þve buffer treat- phulariaceae and Fabaceae spp. For each year, ßower ments (MSU2, MSU3, MSU5, CP-IA, and MSU Best data were summed across the six sample dates among Bet) were planted earlier than other treatments to replicates, and mean ßower abundance was calculated allow time required for perennial species to establish. based on the total number of observations (n ϭ 24) per Plugs were planted at a density of 25 per plot, and buffer treatment. individual species placement within plots was kept the The height of each plant was measured to the tallest same across replications to reduce within-treatment point. Mean canopy height was calculated as the sum variation among replicates. Plugs were positioned 15 of all plant heights per plot over the total plant heights cm from the plot borders on all sides, and 31-cm of each buffer treatment. Five random subsamples per spacing was maintained between plants. plot were taken to measure percent ground cover by 302 ENVIRONMENTAL ENTOMOLOGY Vol. 43, no. 2 tossing a 31- by 31-cm quadrat into plots and visually that provide either biological control (a combination estimating the proportion of ground covered by veg- of predators and parasitoids) or pollination (bees). etation within each quadrat. Percentages for each toss To describe the diversity of beneÞcial insects, we were estimated by two different individuals and av- calculated species richness as the number of taxo- eraged over the total number of estimates recorded in nomic units in each vacuum sample. For each year, the each buffer treatment. Analyses were based on Þnal resulting values were summed across the six sampling measurements taken in August 2010 and 2011, to rep- dates among replicates, and mean species richness was resent the maximum, end-of-season height and ground calculated based on the total number of observations cover and to account for the annual establishment of (n ϭ 24) per buffer treatment. Species richness esti- corn plants. For each year, resulting values were mates for bees and natural enemies were restricted to summed among replicates, and means for canopy the number of taxonomic units identiÞed to species or height and ground cover were calculated per buffer classiÞed as morphospecies. and natural enemy treatment (n ϭ 4). taxa with undetermined identiÞcations and spiders Arthropod Collection, Identification, and Guild As- were omitted from measures of species richness owing signment. Arthropod (insect and spider) communities to unresolved species-level identiÞcations. However, were sampled in each plot throughout the 2010 and we did include spiders in all estimates of the abun- 2011 growing season (June, July, and August). We dance of natural enemies. Diversity indices of species used vacuum sampling methods adapted from Fiedler richness were calculated using the “vegan” package and Landis (2007a). A Þne mesh white paint strainer version 2.0Ð1 in R version 2.14.1 (Oksanen et al. 2011, was placed over the air intake on a gas-powered leaf R Development Core Team 2011). blower (Troy-Bilt, model no. TB320BV), and vegeta- To describe the abundance of beneÞcial insects, we tion in each plot was vacuumed for 30 s while moving calculated the number of individuals in each vacuum continuously around each plot to contact the foliage sample. For each year, the resulting values for each and ßowers on all sides. The mesh strainer with the sample were summed across the six sampling dates sample was then removed and placed into a clear among replicates, and mean abundance was calculated plastic resealable bag. An unused clean mesh strainer based on the total number of observations (n ϭ 24) per was used for sampling subsequent plots. Vacuum sam- buffer treatment. Mean abundance was calculated pling occurred during the Þrst and third week of each separately for herbivores, predators, parasitoids, pol- month during the sampling period with no Ͻ12 d linators, detritivores, and fungivores for each treat- between sampling events. To ensure high insect ac- ment. All guilds were included to describe the pro- tivity and consistency among samples, vacuum sam- portion of the insect community comprised by each pling was restricted to mid-day during favorable group. Analyses focused on bees and natural enemies weather conditions (warm sunny days with cloud of insect pests. cover Ͻ30% and wind gusts below 5 mph). After each Statistical Analyses. A paired t-test was used to test sampling event, insects were transported to the lab for differences between expected and observed plant and frozen until processed. Voucher specimens were diversity (SimpsonÕs diversity index, 1/D) of each deposited in the Department of Entomology, Insec- buffer treatment. For expected diversity, 1/D was tary at Iowa State University, Ames, IA calculated as if all species in each plot established as When possible, insects were identiÞed to species. planned. This was compared with observed diversity, Spiders were identiÞed to order (Araneae). When 1/D calculated for each plot based on species that species identiÞcation could not be resolved, individ- actually established (PROC TTEST, SAS 9.2, SAS In- uals were identiÞed to the lowest taxonomic unit pos- stitute 2008; 1/D calculated using the “vegan” package sible or organized into morphospecies, and given a version 2.0Ð1 in R version 2.14.1, Oksanen et al. 2011, unique identiÞer for reference and classiÞcation of R Development Core Team 2011). We used ANOVA duplicates. Following identiÞcation, individuals were to test for variation among means for observed plant grouped into guilds: herbivores, predators, parasitoids, diversity, canopy height, percent ground cover, and pollinators, detritivores, fungivores, and “other” based ßower abundance among buffer treatments. This on species accounts described in identiÞcation keys model included treatment (nine buffer treatments) and reviewed literature. The group referred to as and block (four replicate plots) as Þxed effects and the “other” includes species with nonfeeding adults, blood interaction of treatment and block (PROC GLM, SAS feeders, and unresolved feeding habits. Insects occu- 9.2, SAS Institute 2008). When signiÞcant differences pying different guilds in different stages of their life in plant measurement data were detected, a post hoc cycles (i.e., herbivores, predators, and parasitoids) mean comparisons test was performed using least sig- were classiÞed based on feeding behaviors of their niÞcant differences (LSD), StudentÐNewmanÐKuels immature stages. For this study, the pollinator guild (SNK) procedure (␣ ϭ 0.05; PROC GLM, 9.2, SAS was deÞned as managed Apis mellifera L. and wild, Institute 2008). Ground cover data were arcsine non-Apis bee species. square root transformed before analysis, and trans- Initially, we described the species composition of formed data were used to determined signiÞcant dif- the entire insect community among buffer treatments ferences, and untransformed means are presented in to determine differences in overall diversity compared results. Plant measurement data were analyzed sepa- with diversity within guilds. Further analyses focused rately for 2010 and 2011 to account for variation be- on the diversity and abundance of beneÞcial guilds tween years. Results of analyses pertaining to canopy April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 303

height, percent ground cover, and ßower abundance, (AICc; Burnham and Anderson 2002). For each year, were reported separately by year, but values for plant explanatory variables included were plant diversity diversity across treatments did not vary between years (SimpsonÕs diversity index, 1/D), canopy height, per- and are only reported once (using 2010 data) to rep- cent ground cover, and ßower abundance. Percent resent both 2010 and 2011. ground cover data were arcsine square root trans- We tested multiple hypotheses related to the rela- formed before analysis, and transformed data were tionship between buffer treatments and insect diver- used to generate models. Response variables included sity and abundance. All buffer treatments were in- the diversity and abundance of bees and natural en- cluded to test the null hypothesis that the diversity and emies. We report the “best-Þt model” (i.e., the model abundance of insect communities did not vary among with the minimum AICc value) and “competing mod- the nine different buffer treatments. Additional hy- els” (i.e., any model for the same response variable potheses pertained to a subset of the nine treatments. having an AICc value with a difference less than two In all procedures described below, data were analyzed is considered strongly supported; Burnham and An- separately for 2010 and 2011 to account for variation derson 2002). Models with differences in AICc values between years. greater than two (compared with the best-Þt model) We compared the species composition of insect were considered too weak to support these data and communities (included species of all guilds) among are not shown. Model selection was performed using buffer treatments using nonmetric multidimensional the “AICcmodavg” package version 1.24 in R version scaling (nMDS). Treatments were ordinated using the 2.14.1 (R Development Core Team 2011, Mazerolle BrayÐCurtis dissimilarity matrix and plotted in two 2012). dimensions. We used multi-response permutation procedures (MRPP) to test the null hypothesis of no difference among treatments. In addition, we Þtted Results the abundance of insects in each guild as regressed vector arrows. Arrows point in the direction of in- Plant Diversity. No signiÞcant differences were creasing abundance for each guild, and arrow length found between expected and observed mean plant indicates amount of proportional correlation with the diversity (SimpsonsÕs diversity index, 1/D) within our ordination. The “vegan” package version 2.0Ð1 in R established buffer treatments (P Ͼ 0.05). As expected, version 2.14.1 (Oksanen et al. 2011, R Development we observed a signiÞcant difference in plant diversity Core Team 2011) was used to conduct nMDS, MRPP, among the nine buffer treatments (F ϭ 40.16; df ϭ 8, and vector Þtting procedures. 35; P Ͻ 0.0001; Table 4). These differences were con- We used ANOVA to test the null hypothesis of no sistent with the desired treatments assigned to each difference in the diversity of bees and natural enemies plot. among buffer treatments (PROC GLM, SAS 9.2, SAS Plant Measurements. The number of ßowers among Institute 2008). This model included treatment (nine buffer treatments with conspicuous ßowers increased buffer treatments) and block (four replicate plots) as from 2010 to 2011. The abundance of ßoral resources Þxed effects and the interaction of treatment and available per vacuum sampling event varied signiÞ- block as random effects. When signiÞcant differences cantly across buffer treatments in 2010 (F ϭ 13.61; df ϭ in diversity data were detected, a post hoc mean com- 5, 215; P Ͻ 0.0001) and 2011(F ϭ 12.86; df ϭ 5, 215; P Ͻ parison test was performed with the LSD SNK pro- 0.0001; Table 4). The MSU Best Bet, MSU5, MSU3, and cedure (␣ ϭ 0.05) to identify differences (PROC MSU2 buffer treatments had signiÞcantly more ßow- GLM, SAS 9.2, SAS Institute 2008). ers in bloom per sampling event compared with the We used ANOVA to test the null hypothesis of no CP-IA buffer and alfalfa (Table 4). There were sig- difference in the abundance of bees and natural en- niÞcant differences in canopy height and percent emies among buffer treatments. This model included ground cover among buffer treatments in 2010 (can- treatment (nine buffer treatments), block (four rep- opy height: F ϭ 10.23; df ϭ 8, 35; P Ͻ 0.0001; ground licate plots), and time (six sampling events) as Þxed cover: F ϭ 13.23; df ϭ 8, 35; P Ͻ 0.0001) and 2011 effects and the interaction of treatment and block as (canopy height: F ϭ 9.19; df ϭ 8, 35; P Ͻ 0.0001; ground random effects (PROC GLM, SAS 9.2, SAS Institute cover: F ϭ 17.47; df ϭ 8, 35; P Ͻ 0.0001). 2008). The abundance of herbivores was also analyzed Insect Community Composition. Vacuum sampling to determine whether buffer treatments vary in their yielded 14,632 insects in 2010 and 22,261 insects in attractiveness to herbivores, particularly pest species. 2011. Samples collected in 2010 were primarily com- When signiÞcant differences in abundance data were posed of herbivores (59%), followed by beneÞcial detected, a post hoc mean comparison test was per- insects composed of predators, parasitoids, and bees formed using the LSD SNK procedure (␣ ϭ 0.05) to (28% pooled). Detritivores, fungivores, and “other” identify differences (PROC GLM, SAS 9.2, SAS Insti- accounted for the remaining 13% of the total insect tute 2008). community. Herbivores remained the dominant guild We determined which plant characteristics ex- in 2011, accounting for 73% of collected insects and the plained the most variation in the diversity and abun- proportion of beneÞcial groups decreased to (17%) dance of bees and natural enemies using multiple relative to the total. Detritivores, fungivores, and linear regression analysis and AkaikeÕs Information “other” accounted for the remaining 10% of the total Criterion for model selection, adjusted for sample size insect community. 304 ENVIRONMENTAL ENTOMOLOGY Vol. 43, no. 2

Table 4. Mean ؎ SEM for plant characteristics measured among buffer treatments during 2010 and 2011

2010 2011 a Buffer Diversity Canopy ht Ground Flower Canopy ht Ground Flower (cm)b cover (%)c no.d (cm)b cover (%)c no.d Corn 1 Ϯ 0e 159 Ϯ 13b 20 Ϯ 4c Ñ 156 Ϯ 13b 18 Ϯ 4d Ñ Willow 1 Ϯ 0e 135 Ϯ 46a 5 Ϯ 0.2c Ñ 297 Ϯ 49a 4 Ϯ 1d Ñ Switchgrass 1 Ϯ 0e 115 Ϯ 10b 74 Ϯ 6a Ñ 119 Ϯ 10b 74 Ϯ 6ab Ñ Alfalfa 1 Ϯ 0e 55 Ϯ 4b 85 Ϯ 3a 8 Ϯ 2b 60 Ϯ 4b 81 Ϯ 3a 9 Ϯ 3b MSU2 2 Ϯ 0d 104 Ϯ 11b 37 Ϯ 11b 64 Ϯ 12a 115 Ϯ 21b 37 Ϯ 10c 70 Ϯ 13a MSU3 3 Ϯ 0.1c 89 Ϯ 29b 45 Ϯ 4ab 52 Ϯ 13a 91 Ϯ 26b 47 Ϯ 4bc 58 Ϯ 15a MSU5 5 Ϯ 0.1b 79 Ϯ 11b 68 Ϯ 6ab 30 Ϯ 7a 99 Ϯ 14b 71 Ϯ 6ab 31 Ϯ 7a MSU Best Bet 12 Ϯ 0.3a 117 Ϯ 32b 53 Ϯ 5ab 67 Ϯ 14a 109 Ϯ 6b 54 Ϯ 6bc 68 Ϯ 14a CP-IA 12 Ϯ 0.4a 77 Ϯ 15b 60 Ϯ 9a 23 Ϯ 4b 83 Ϯ 6b 61 Ϯ 9bc 24 Ϯ 4b

Means within columns followed by common letters are not signiÞcantly different at P Ͻ 0.05. a Plant diversity is the mean value of SimpsonÕs diversity index (1/D), values increase as diversity increases. Diversity values were the same for both years and are reported once to represent both 2010 and 2011. b Canopy represent end-of-season canopy height measured in August data during 2010 and 2011. c Percent ground cover represents end-of-ground cover measured in August data in 2010 and 2011. SigniÞcant differences are based on arcsine square root transformed data; untransformed means are presented here. d Flower number is the mean abundance of ßowers per buffer type per sampling event.

In both years, alfalfa treatments experienced out- vacuum samples varied signiÞcantly by year for nat- breaks of the potato leafhopper, Empoasca fabae Har- ural enemies (F ϭ 14.03; df ϭ 1, 431; P Ͻ 0.0001), bees ris (Hemiptera: Cicadellidae). This species was the (F ϭ 20.02; df ϭ 1, 431; P Ͻ 0.0001), and herbivores most common herbivore in both 2010 (22% of all (F ϭ 13.52; df ϭ 1, 431; P ϭ 0.0001). Owing to the herbivores and 13% of the total community) and 2011 variation in insect abundance by year, data were an- (47% of all herbivores and 35% of the total commu- alyzed separately for 2010 and 2011. nity). Leafhopper abundance in our experiment may We used nMDS ordination to show the conÞgura- not represent how outbreaks of E. fabae typically oc- tion of treatments based on the species composition of cur in an on-farm setting. Alfalfa would have been cut insects (Fig. 1). Ordinations were plotted in two di- and harvested when E. fabae infestations occur (Lefko mensions, and stress for the Þnal solutions was 0.02 and et al. 1999). We did not cut the alfalfa during this 0.067 for 2010 and 2011, respectively; these values are experiment, allowing populations of E. fabae to persist. considered ideal for species abundance data (Clarke Other than E. fabae, no other economically important 1993, McCune and Grace 2002). Based on the MRPP pests were observed. When E. fabae is omitted, the tests, the insect communities collected among buffer recalculated ratios for each guild relative to the total treatments were signiÞcantly different in 2010 (A ϭ are more similar between years (2010: 53% herbivores 0.04; P Ͻ 0.001) and 2011 (A ϭ 0.07; P ϭ 0.009). Vector and 33% beneÞcial groups; 2011: 60% herbivores and arrows show the abundance of beneÞcial insects of 24% beneÞcial groups). The abundance of insects in multiple guilds is increasing in the directions of the

Fig. 1. Nonmetric multidimensional scaling ordinations (nMDS plots) of the species composition based on the BrayÐ Curtis dissimilarity indices for (a) 2010 and (b) 2011 samples depicting the conÞguration of treatments in relation to the community dissimilarity; treatments in proximity have a more similar species composition than treatments separated by greater distances. The plotted arrows indicate the insect guilds correlated with treatments, and the arrow points in the direction of the most rapid change in increasing abundances. Correlations were signiÞcant at P Յ 0.05. April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 305

Table 5. Total and mean ؎ SEM taxonomic units per buffer treatment for bees and natural enemies in 2010 and 2011

Corn Willow Switch-grass Alfalfa MSU2 MSU3 MSU5 MSU Best Bet CP-IA Bee taxa Totala 1 2 3 6 13 14 13 18 14 2010 Mean Ϯ SEM 0d 1 Ϯ 0.6d 1 Ϯ 0.7d 2 Ϯ 0.9cd 5 Ϯ 1.4bc 5 Ϯ 1.6bc 7 Ϯ 1.5b 10 Ϯ 1.2a 6 Ϯ 1.8b 2011 Mean Ϯ SEM 0.3 Ϯ 0.2b 0b 0b 1 Ϯ 0.2b 4 Ϯ 0.6a 4 Ϯ 0.6a 2 Ϯ 1.2ab 3 Ϯ 1.2ab 1 Ϯ 0.4ab Natural enemy taxa Totala 37 32 50 58 50 51 59 57 57 2010 Mean Ϯ SEM 17 Ϯ 2b 10 Ϯ 1b 36 Ϯ 1a 43 Ϯ 6a 36 Ϯ 4a 38 Ϯ 6a 40 Ϯ 3a 48 Ϯ 5a 45 Ϯ 3a 2011 Mean Ϯ SEM 16 Ϯ 4d 15 Ϯ 3d 25 Ϯ 4cd 46 Ϯ 4a 22 Ϯ 0.6cd 28 Ϯ 2bcd 36 Ϯ 2abc 40 Ϯ 5a 27 Ϯ 2bcd

Means within columns followed by common letters are not signiÞcantly different at P Ͻ 0.05. a Total refers to the total number of species/unique morphospecies within a guild summed across all samples collected in 2010 and 2011. Species captured in both years were only counted once for totals.

MSU Best Bet, MSU5, and MSU3 buffers. The insect were most diverse in the MSU5 buffer, followed by communities in simple buffers, especially corn and alfalfa, the diverse buffers, and the other forb-only willow, have the least number of shared species in buffers, varying by nine or fewer species (Table 5). On comparison with all other buffer treatments. a per plot basis, we found two to four times the mean Pollinator Diversity. We observed a bee commu- number of taxa observed in plots with multiple plant nity composed of 24 taxonomic units representing Þve species than corn or willow. families, with MSU Best Bet buffer collecting the most The mean number of species per plot varied signif- (18 taxonomic units). We found two to three times icantly across buffer treatments in 2010 (F ϭ 10.22; more bee species in diverse and forb-only buffers than df ϭ 8, 215; P Ͻ 0.0001) and 2011 (F ϭ 8.35; df ϭ 8, 215; in simple buffers (Table 5), although this relationship P Ͻ 0.0001). In 2010, we observed signiÞcantly fewer varied by year. During each year, we observed signif- natural enemy taxa within corn and willow treatments, icant differences in bee diversity among buffer treat- but no signiÞcant differences were observed among ments. The mean number of bee species collected the remaining treatments. In 2011, we observed sig- varied signiÞcantly across buffer treatments in 2010 niÞcantly more natural enemy taxa within alfalfa and ϭ ϭ ϭ ϭ (F 6.25; df 8, 215; P 0.0002) and 2011 (F 5.73; the MSU Best Bet buffers compared with the other ϭ ϭ df 8,215; P 0.0004). In 2010, we observed the most treatments, excluding the MSU5 treatment, which had bee species within the MSU Best Bet buffer; in 2011 we as many natural enemies as all treatments but corn and did not observe signiÞcant differences among any of willow (Table 5). the treatments with multiple plant species (Table 5). Most natural enemies collected among buffer treat- Bee diversity was lowest in simple buffers composed ments are considered widely distributed and common of only one species. In 2010, no bees were captured in across our region. During 2010, parasitoids accounted corn treatments, and in 2011, there were no bees for a greater proportion of natural enemy taxa com- captured in either willow or switchgrass treatments. pared with predators (60% and 40%, respectively), and The majority of bees collected among buffer treat- in 2011 parasitoids were slightly less dominant than ments were species native to North America. Excep- predators (49% and 51%, respectively). Most of the tions include a few introduced species such as the honey bee (A. mellifera) and the alfalfa leafcutting natural enemy species captured were generalists such bee, Megachile rotundata F., both found only in 2011. as Orius insidiosus (Say) (Hemiptera: Anthocoridae), Megachile rotundata was observed only in the MSU Nabis spp. (Hemiptera: Nabidae), Harmonia axyridis Best Bet buffers while A. mellifera was observed in (Pallas) (Coleoptera: Coccinellidae), tachinid ßies MSU5, MSU Best Bet, and CP-IA buffers. The majority (Diptera: Tachinidae), and pteromalid wasps (Hyme- of taxa (79%) we captured were ground-nesting bees, noptera: Pteromalidae). Some of the natural enemies but at least one cavity-nesting species was represented we observed are omnivorous (e.g., Coleomegilla macu- in all families except Andrenidae. Ground-nesting spe- lata (DeGeer), syrphids ßies, free-living adult parasi- cies exhibit different levels of sociality ranging from toids) and supplement their diet with plant-derived annual eusocial (e.g., Bombus spp.), communal (Aga- foods, such as pollen and nectar (Triplehorn and John- postemon spp.), solitary (Melissodes spp.), and varia- son 2005). These taxa can be biological control agents tions thereof ( spp.). In contrast, cavity- of agronomic insect pests including the soybean aphid nesting species are all solitary nesters (Packer et al. (Aphis glycines Matsumura; Rutledge et al. 2004, 2007). Most species collected among buffer treat- Costamagna et al. 2008), potato leafhopper (E. fabae; ments are considered common or locally abundant in O¨ stman and Ives 2003, Weiser Erlandson and Obrycki our region (Michener 2000, Packer et al. 2007; see also 2010), and European corn borer (Ostrinia nubilalis Ascher and Pickering 2012 for geographic distribution (Say) (Musser and Shelton 2003). In several studies, maps). these natural enemies were positively associated with Natural Enemy Diversity. We observed a natural plant community diversity and ßowering plants used enemy community composed of 87 taxonomic units in insectary plantings or maintained in Þeld margins representing 41 families. Overall, natural enemies (Colley and Luna 2000, Harmon et al. 2000, Fiedler 306 ENVIRONMENTAL ENTOMOLOGY Vol. 43, no. 2

Fig. 2. Mean Ϯ SEM abundance of bees across buffer treatments collected in (a) 2010 and (b) 2011 per plot. Means with common letters are not signiÞcantly different. and Landis 2007b, Lundgren 2009, Lundgren et al. est number of Lasioglossum specimens (24) was ob- 2009, Al-Dobai et al. 2012). served in the MSU Best Bet and MSU3 buffers, ac- Pollinator Abundance. We observed a pollinator counting for 44% and 50% of the total bee abundance community composed of 325 individual bees. During in these buffer treatments, respectively. In contrast, each year, we observed signiÞcant differences in the Bombus griseocollis (DeGeer) (Hymenoptera: Api- abundance of bees among buffer treatments. The dae) were the most abundant bees collected in 2011. mean number of bees per plot varied signiÞcantly B. griseocollis was only present in samples from the across buffer treatments in 2010 (F ϭ 6.47; df ϭ 8,215; MSU Best Bet and forb-only buffers, with the greatest P Ͻ 0.0001) and 2011 (F ϭ 4.33; df ϭ 8,215; P Ͻ 0.0001). number of individuals (14) observed in the MSU2 During 2010 we did not capture a single bee in corn; buffer, accounting for 46% of the total bee abundance in 2011 we did not capture a single bee in either willow in that treatment. or switchgrass. Bees were more abundant in diverse Natural Enemy Abundance. We collected 7,520 nat- and forb-only buffer treatments, from which we cap- ural enemies, of which predators accounted for a tured three to four times the mean number of indi- greater proportion than parasitoids (57% and 43%, viduals per plot than in simple buffer treatments (Fig. respectively). Natural enemies were more abundant 2). Although these differences were not signiÞcant, in diverse and forb-only buffer treatments containing noticeably more bees were found within forb-only and multiple plant species and alfalfa, from which we col- MSU Best Bet buffers compared with the CP-IA buffer lected 2Ð10 times the number of natural enemies per (Fig. 2). Overall, the most bees (75) were collected plot than the remaining simple buffer treatments (Fig. from the MSU Best Bet buffer. 3). The greatest number of natural enemies (1,602) Lasioglossum spp. (Hymenoptera: Halictidae) were was observed in the MSU Best Bet buffers. The mean the most abundant bees collected in 2010. This group number of natural enemies per plot varied signiÞ- was present in all treatments except alfalfa. The great- cantly across buffer treatments in 2010 (F ϭ 9.15; df ϭ

Fig. 3. Mean Ϯ SEM abundance of natural enemies across buffer treatments collected in (a) 2010 and (b) 2011 per plot. Means with common letters are not signiÞcantly different. April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 307

Table 6. Summary of model selection statistics used for evaluating the influence of plant community characteristics on the species richness and abundance of bees and natural enemies vacuum collected in 2010 and 2011

a ⌬ ␻ 2 b Response variable Ki AICc AICc i Adj. R Variables in model 2010 Bee Richness 4 18.20 0.00 0.99 0.54 Plant diversity, ßower no. Abundance1 4 35.97 0.00 0.58 0.35 Flower no., plant diversity Abundance2 5 36.65 0.68 0.41 0.29 Flower no. 2011 Bee Richness 3 Ϫ77.00 0.00 0.99 0.60 Flower no. Abundance 3 Ϫ23.77 0.00 0.99 0.56 Flower no. 2010 Natural enemy Richness1 6 44.42 0.00 0.69 0.80 Plant diversity, ßower no., ground cover, (canopy ht) Richness2 5 46.09 1.67 0.30 0.78 Plant diversity, ground cover Abundance 5 204.14 0.00 0.88 0.57 Plant diversity, ßower no., ground cover 2011 Natural enemy Richness1 5 63.46 0.00 0.44 0.37 Plant diversity, ßower no., ground cover Richness2 4 63.83 0.37 0.36 0.33 Plant diversity, ground cover, (canopy ht) Abundance1 5 183.20 0.00 0.56 0.41 Plant diversity, ßower no., ground cover Abundance2 4 184.21 1.01 0.34 0.36 Plant diversity, ground cover

ϭ ϭ For each best-Þt and competing model we present the response variable, Ki the no. of variables in each model, AICc AkaikeÕs Information ⌬ ϭ ␻ ϭ Criterion adjusted for sample size, AICc difference in AICc score between best-Þt and competing models, i Akaike weights as an estimate of the relative likelihood of a given model against all other models, and Adj. R2 ϭ R-square adjusted for the no. of terms in the model. a Where response variables are listed twice within guilds and years, the Þrst model is the best-Þt model based on the min. AICc and greater Adj. R2 values. Competing models are listed second and only competing models with ⌬AICc Ͻ 2 are shown. b Plant diversity is SimpsonÕs 1/D, ßower number is mean ßower abundance per sampling event. Variables in parentheses indicate a negative relationship. (For details regarding plant characteristics see Methods and Materials: plant measurements; see Table 4 for plant characteristics comparisons across treatments).

8, 215; P ϭ 0.0008) and 2011(F ϭ 8.79; df ϭ 8, 215; P Ͻ emy abundance and species richness. In addition to 0.0001). In 2010, we observed signiÞcantly more nat- the positive relationships, we observed a signiÞcant ural enemies per plot within the MSU Best Bet buffers negative relationship between natural enemy species compared with all other buffers; in 2011 we did not richness and canopy height for the best-Þt model in observe signiÞcant differences among the MSU Best 2010 and competing model in 2011. Bet, MSU5, and alfalfa buffers (Fig. 3). In both years, O. insidiosus was the most abundant Discussion predator and was present in samples from all buffer treatments, but the most (523) were observed in the We successfully established nine different plant MSU Best Bet buffer, accounting for 36% of the total communities with sufÞcient aboveground growth in natural enemy community in this buffer treatment. both years to observe consistent trends in the diversity Pteromalid wasps were the most abundant parasitoid and abundance of beneÞcial insects. Diversity and family and were present in samples from all buffer abundance of bees and natural enemies was 1) great- treatments. The MSU5 buffer had the most pteroma- est in buffers with diverse plant communities with lids (384), comprising 34% of the total natural enemy continuous availability of ßoral resources, 2) interme- community in this treatment. diary in buffers reduced in species richness and avail- Model Comparisons. BeneÞcial insect diversity and ability of ßoral resources, and 3) lowest in buffers abundance exhibited positive relationships with sev- composed of a single species. Based on these obser- eral of the plant characteristics measured among buf- vations, we propose that buffer strips could be opti- fer treatments. All best-Þt and competing models were mized with native plants to attract multiple guilds and signiÞcant (P Ͻ 0.05). During 2010, we observed a species of beneÞcial insects. Overall, our results sug- signiÞcant positive relationship between bee species gest that 1) plant communities used in current buffer richness and plant diversity and the number of ßowers strips on organic farms (at least within Iowa) are not in bloom (Table 6). In 2010, a competing model based optimal for conserving beneÞcial insects, 2) the ad- solely on ßower abundance was also signiÞcant. In dition of ßowering perennial species can improve buf- 2011, the variables in the best-Þt models for both bee fer strips as habitats for beneÞcial insects, 3) combi- species richness and abundance were reduced to a nations of native perennial plants can attract signiÞcant positive relationship with the number of beneÞcial insects even at modest levels of plant di- ßowers and no competing models. During both years versity, and 4) plant communities recommended for we observed a signiÞcant positive relationship be- prairie reconstruction may not contain the most at- tween species richness and abundance of natural en- tractive native species at densities necessary to attract emies and plant diversity, the number of ßowers in multiple guilds of beneÞcial insects. bloom, and ground cover in the best-Þt models. In We hypothesized that diversity and abundance of 2010, there was evidence for a competing model for beneÞcial insects would be limited in single-species natural enemy species richness and in 2011 there was plant communities (simple buffers) compared with evidence for a competing model for both natural en- moderately diverse (forb-only buffers) and mixtures 308 ENVIRONMENTAL ENTOMOLOGY Vol. 43, no. 2 of forbs and grasses (diverse buffers). Our results ture of switchgrass. We suggest that the value of switch- agree with our predictions; however, we also observed grass as a buffer for organic farms will depend on how it signiÞcant variation in beneÞcial insect communities is established and managed. among the simple buffers. These results indicate that In contrast to the other simple buffer treatments, some simple buffers may be more suitable habitats alfalfa had several characteristics (e.g., percent than others. Among the simple buffers, bee and nat- ground cover, ßoral resources) that, as indicated by ural enemy communities were more diverse and abun- our analyses, have a positive relationship with the dant on perennial plants than the annual plantings of diversity and abundance of beneÞcial insect commu- corn. Perennial buffer strips may be more hospitable nities. Natural enemy communities were signiÞcantly refuges for beneÞcial insects than ephemeral plant more diverse and abundant in alfalfa compared with communities. However, we did not observe a signif- corn and willow in both years, and additionally to icant difference in beneÞcial insect abundance be- switchgrass in 2011. However, the same was true for tween corn and willow, an introduced perennial. The herbivores in alfalfa in both years of our study. Alfalfa low abundance of insects on willow may be a product can provide multiple resources for beneÞcial insects; of our sampling methodology. Willow ßowers in the however, the management used in Iowa to manage spring, which may provide insects with food resources. pests (e.g., E. fabae) below economic thresholds in- However, we sampled later in the season when those cludes insecticide applications and early alfalfa har- resources were not present on willow and were more vest (Lefko et al. 1999). Insecticides are not compat- abundant on other plants. Our data suggest that for a ible with organic production, and early harvest can buffer to be attractive to beneÞcial insects, it should remove habitat and prey, and therefore, natural ene- consist of plants, which combined, provide season- mies. We did not manage our plots in this manner and long ßoral resources and vegetative ground cover, in uncut alfalfa plots became infested with E. fabae in addition to being perennial. both years. Switchgrass is a native perennial grass commonly Unmanaged pest populations in our alfalfa plots may used in conservation programs (USDAÐNRCS 2011) be partially responsible for recruiting larger popula- and is being explored for bioenergy (Prochnow et al. tions of natural enemies. Therefore, our results may 2009). The results of our study indicate that switch- overestimate the ability of alfalfa to attract natural grass monocultures were not effective for increasing enemies. In addition, bee diversity and abundance in beneÞcial insect abundance or diversity. In both years, alfalfa was not signiÞcantly different from corn in 2010 natural enemy communities in switchgrass did not and willow and switchgrass in 2011, where bees were signiÞcantly differ from corn and willow. Bee com- not observed. Despite these results, alfalfa can be an munities did not signiÞcantly differ among any of the attractive option for a buffer when it doubles as a simple buffers, and bees were absent in switchgrass in harvestable forage crop. This may apply to a subset of 2011. In contrast to corn and willow, switchgrass organic farmers involved in livestock production. In shared some characteristics (e.g., a greater percentage this situation, beneÞts may increase when alfalfa is of ground cover) that, as indicated by our analyses, harvested in strips (Weiser et al. 2003), such that not have a positive relationship with the diversity and all habitat for beneÞcial insects is removed at once. abundance of natural enemies. However, switchgrass, Among the nine treatments, in both years, the MSU like corn, lacks components (e.g., ßoral resources) Best Bet buffer was consistently one of the most at- found in plant communities that had a more diverse tractive for both bees and natural enemies. Regarding and abundant beneÞcial insect communities, making it bees, halictids were particularly diverse and abundant a sub-optimal candidate for buffer strips. in the MSU Best Bet buffer. Several species of halictids Gardiner et al. (2010) observed that the beneÞcial are responsible for pollinating crops including Þeld- insect communities varied signiÞcantly across Þelds grown tomato (Greenleaf and Kremen 2006), water- (Ͼ2 ha) of corn, switchgrass, and mixed prairie. Spe- melon (Kremen et al. 2004), and canola (Morandin ciÞcally, bee abundance and species richness of lady and Winston 2005). Therefore, an abundance of hal- beetles was greater in switchgrass monocultures and ictids may lead to pollination of crops across multiple mixed prairie polycultures compared with corn (Gar- bloom periods. In 2011, the most abundant bee species diner et al. 2010). We also observed the greatest insect in MSU Best Bet and forb-only buffers was the bum- abundance and diversity in plant communities com- blebee, B. griseocollis. Bombus species are also known posed of prairie plant species; however, beneÞcial to pollinate the crops listed above, and are especially insect communities did not differ signiÞcantly be- effective pollinators of crops that require sonication or tween corn and switchgrass. As noted by Gardiner et buzz pollination. Halictid and Bombus spp. were con- al. (2010), the switchgrass Þelds that they used were sistently observed in the MSU Best Bet even when planted as part of the Conservation Reserve Program populations of these insects ßuctuated between years and contained a mean of 27 plant species. Gardiner et among the other buffer treatments. Several species of al. (2010) suggested that if switchgrass is planted as a parasitoids (braconids and pteromalids), and the dedicated biofuel crop, it would be managed to pro- predatory, O. insidiosus, were also abundant in the mote single-species stands, which would likely have MSU Best Bet buffer. Like pollinators, natural enemies few beneÞcial insects. The community of beneÞcial have a well-established role in agroecosystems. These insects we observed in our switchgrass plots was likely natural enemies can attack a wide range of herbivo- the product of management that produced a monocul- rous insect pests. To summarize, the MSU Best Bet April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 309 buffer supported a diverse and abundant beneÞcial who consider incentives that promote management insect community that can provide a suite of ecosys- practices to acquire multiple ecosystem services from tem services that complement organic and conven- buffer strips. tional cropping systems. In summary, the results from our Þeld experiment Despite having the same species richness of grasses indicate plant communities that dominate existing and forbs, we observed differences in the beneÞcial buffer strips and lands designated for conservation insect community between the CP-IA and MSU Best may not be optimal for beneÞcial insects. Adding ßow- Bet buffers. Typically, reconstructed prairies have a ering perennial species can improve buffer strips as greater proportion of grass compared with forbs; how- habitats for beneÞcial insects, especially bee pollina- ever, we manipulated these ratios to contain a greater tors. Moreover, buffer strips can be further optimized proportion of forbs in both the MSU Best Bet (76% by intentionally combining the most attractive native forbs and 24% grass) and CP-IA (68% forbs and 32% species even at modest levels of plant diversity, such grass). This forb-rich ratio was used to optimize the that ßowering resources are available throughout the plant communities, so each of these diverse treatments growing season. In conclusion, conservation of bene- provided ßoral resources to accommodate a range of Þcial insects appears to be a product a high density of insect species. The forbs in the MSU Best Bet buffer the attractive native species (i.e., quality), and not produced a greater number of ßowers than the CP-IA, necessarily a product of a habitat made up of many contributing to a difference (that was not always sig- native plant species (i.e., quantity). niÞcant) in the diversity and abundance of bees and natural enemies. Differences among these forb-rich mixtures reinforce the importance of the decision- Acknowledgments making process in targeted conservation efforts. To We thank Brent Pringnitz for assistance with Þeld main- conserve beneÞcial insects with buffers by carefully tenance and research demonstrations, Nicholas Schmidt for reintroducing native ßowering plants will require con- assistance with plot construction, Ion Exchange Inc. for as- sidering their density at the farm and landscape scale. sistance with seedlings and planting advice, and Rick and For example, in a landscape-scale study conducted in Stacey Hartmann for propagating and providing willow cut- the same eco-region as our research site, attractive tings. We acknowledge the effort of many individuals who forb species similar to those in the MSU Best Bet buffer helped with planting, weeding, data collection, and assisted treatment were present in a large (Ͼ1,619 ha) recon- in all aspects of this study. We thank Drs. Lisa Schulte Moore structed prairie embedded in cropland composed of a and Mary Harris for reviewing an earlier version of this cornÐsoybean rotation (Schmidt et al. 2011). Despite manuscript, as well as the anonymous reviewers whose sug- gestions greatly improved this manuscript. This research was the proximity of prairies to cultivated Þelds, no in- funded by The Leopold Center for Sustainable Agriculture. crease in natural enemy abundance or diversity was observed in adjacent crops. Schmidt et al. (2011) sug- gested that densities of these plant species in tradi- References Cited tional prairie restorations may not be optimal for en- hancing both biological diversity and functional Al-Dobai, S., S. Reitz, and J. Sivinski. 2012. Tachinidae (Dip- diversity at the landscape scale. At smaller scales, dif- tera) associated with ßowering plants: estimating ßoral ferences in plant diversity may result in an observable attractiveness. Biol. Control 61: 230Ð239. Ascher, J. S., and J. Pickering. 2012. Discover Life bee spe- effect on the insect community, as noted by Rebek et cies guide and world checklist (Hymenoptera: Apoidea: al. (2005). Anthophila). (http://www.discoverlife.org) (accessed The importance of conserving beneÞcial insects to 10 October 12) maintain ecological processes is being increasingly Bianchi, F.J.J.A., C.J.H. Booij, and T. Tscharntke. 2006. Sus- recognized, and previous work has focused on the tainable pest regulation in agricultural landscapes: a re- relationship between successful crop production and view on landscape composition, biodiversity and natural insect-derived services, like biological control. In a pest control. Proc. R. Soc. Lond. Biol. 273: 1715Ð1727. review of studies that investigate how habitat man- Burnham, K. P., and D. R. Anderson. 2002. Model selection agement can promote conservation biological control, and multimodel inference: a practical information-theo- retic approach, 2nd ed. Springer, New York, NY. Fiedler et al. (2008) found that most studies described Clark, W. R., and K. F. Reeder. 2007. Agricultural buffers the relationship of natural enemies to speciÞc plants. and wildlife conservation: a summary about linear prac- Among the 34 studies reviewed, the majority of the 165 tices, pp. 45Ð55. In J. B. Haußer (ed.), Fish and wildlife plant species documented were exotic. Gaps in re- response to farm bill conservation practices, technical search regarding the advantages of using native plants review 07-1. (USDA) United States Department of Ag- over exotic plants for enhancing ecosystem services riculture (NRCS) Natural Resource Conservation ser- mediated by beneÞcial insects are increasingly being vice, and (FSA) Farm Service Agency in partnership with addressed (Fiedler and Landis 2007a, Frank et al. 2008, The Wildlife Society, Bethesda, MD. Tuell et al. 2008). These studies, along with the results Clarke, K. R. 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18: 117Ð presented here, suggest that buffer strips have the 143. potential to provide beneÞts beyond simply meeting Colley, M. R., and J. M. Luna. 2000. Relative attractiveness requirements for organic certiÞcation. Realizing these of potential beneÞcial insectary plants to aphidophagous beneÞts will require outreach to farmers and policy hoverßies (Diptera: Syrphidae). Environ. Entomol. 29: writers and representatives of conservation agencies 1054Ð1059. 310 ENVIRONMENTAL ENTOMOLOGY Vol. 43, no. 2

Costamagna, A. C., D. A. Landis, and M. J. Brewer. 2008. Lundgren, J. C., K.A.G. Wyckhuys, and N. Desneux. 2009. The role of natural enemy guilds in Aphis glycines sup- Population responses by Orius insidiosus to vegetational pression. Biol. Control 45: 368Ð379. diversity. BioControl 54: 135Ð142. Elliott, N. C., R. W. Kieckhefer, G. J. Michels, and K. L. Giles. Mazerolle, M. J. 2012. AICcmodavg: model and multimodel 2002. Predator abundance in alfalfa Þelds in relation to inference based on (Q)AIC(c). R package version 1.25. aphids, within-Þeld vegetation, and landscape matrix. En- (http://CRAN.R-project.org/packageϭ AICcmodavg). viron. Entomol. 31: 253Ð260. McCune, B., and J. Grace. 2002. Analysis of ecological com- Fiedler, A. K., and D. A. Landis. 2007a. Attractiveness of munities. MjM Software Design, Gleneden Beach, OR. Michigan native plants to arthropod natural enemies and Michener, C. D. 2000. The bees of the world. John Hopkins herbivores. Environ. Entomol. 36: 751Ð765. University Press, Baltimore, MD. Fiedler, A. K., and D. A. Landis. 2007b. Plant characteristics Morandin, L. A., and M. L. Winston. 2005. Wild bee abun- associated with natural enemy abundance at Michigan dance and seed production in conventional, organic, and native plants. Environ. Entomol. 36: 878Ð886. genetically modiÞed canola. Ecol. Appl. 15: 871Ð881. Fiedler, A. K., D. A. Landis, and S. D. Wratten. 2008. Max- [MSU] Michigan State University. 2012. Native plants and imizing ecosystem services from conservation biological ecosystem services. (http://nativeplants.msu.edu) (ac- control: the role of habitat management. Biol. Control 45: cessed 17 June 2012) 254Ð271. Musser, F. R., and A. M. Shelton. 2003. Predation of Ostrinia Frank, S. D., P. M. Shrewsbury, and O. Esiekpe. 2008. Spa- nubilalis (: Crambidae) eggs in sweet corn by tial and temporal variation in natural enemy assemblages generalist predators and the impact of alternative foods. on Maryland native plant species. Environ. Entomol. 37: Environ. Entomol. 32: 1131Ð1138. 478Ð486. Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Gardiner, M. A., J. K. Tuell, R. Isaacs, J. Gibbs, J. S. Ascher, Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, H. H. D. and A. Landis. 2010. Implications of three biofuel Stevens, and H. Wagner. 2011. Vegan: community ecol- crops for beneÞcial arthropods in agricultural landscapes. ogy package. R package version 2.0-1. (http://CRAN.R- ϭ Bioenergy Res. 3: 6Ð19. project.org/package vegan) ¨ ¨ Greenleaf, S. S., and C. Kremen. 2006. Wild bee species Ostman, O., and A. R. Ives. 2003. Scale-dependent indirect increase tomato production and respond differently to interactions between two prey species through a shared surrounding land use in northern California. Biol. Con- predator. Oikos 102:505Ð514. serv. 133: 81Ð87. Packer, L., J. A. Genaro, and, C. S. Sheffield. 2007. The bee genera of eastern Canada. Canadian Journal of Arthropod Harmon, J. P., A. R. Ives, J. E. Losey, A. C. Olson, and K. S. IdentiÞcation No. 3, 25 September 2007. (http://www. Rauwald. 2000. Coleomegilla maculata (Coleoptera: biology.ualberta.ca/bsc/ejournal/pgs03/pgs_03.html) Coccinellidae) predation on pea aphids promoted by (accessed 26 June 2012) (doi: http://dx.doi.org/10.3752/ proximity to dandelions. Oecologia 125: 543Ð548. cjai.2007.03) Isaacs, R., J. Tuell, A. K. Fiedler, M. M. Gardiner, and, D. A. Prochnow, A., M. Heiermann, M. Plo¨chl, T. Amon, and P. J. Landis. 2009. Maximizing arthropod-mediated ecosys- Hobbs. 2009. Bioenergy from permanent grassland Ð A tem services in agricultural landscapes: the role of native review: 2. Combustion. Bioresour. Technol. 21: 4945Ð plants. Front Ecol. Environ. 7: 196Ð203. 4954. Klein, A. M., B. Vaissiere, J. H. Cane, I. Steffan-Dewenter, ` R Development Core Team. 2011. R: a language and envi- S. A. Cunningham, C. Kremen, and T. Tscharntke. 2007. ronment for statistical computing. R Foundation for Importance of crop pollinators in changing landscapes for Statistical Computing, Vienna, Austria. (http://www.R- world crops. Proc. R. Soc. Lond. Biol. 274: 303Ð313. project.org) Kremen, C., N. M. Williams, R. L. Bugg, J. P. Fay, and R. W. Rebek, E. J., C. S. Sadof, and L. M. Hanks. 2005. Manipu- Thorp. 2004. The area requirements of an ecosystem lating the abundance of natural enemies in ornamental service: crop pollination by native bee communities in landscapes with ßoral resource plants. Biol. Control 33: California. Ecol. Lett. 7: 1109Ð1119. 203Ð216. Kwaiser, K. S., and S. D. Hendrix. 2008. Diversity and abun- Rutledge, C. E., R. J. O’Neil, T. B. Fox, and D. A. Landis. dance of bees (Hymenoptera: Apiformes) in native and 2004. Soybean aphid predators and their use in inte- ruderal grasslands of agriculturally dominated land- grated pest management. Ann. Entomol. Soc. Am. 97: scapes. Agric. Ecosyst. Environ. 124: 200Ð204. 240Ð248. Landis, D. A., S. D. Wratten, and G. M. Gurr. 2000. Habitat Samson, F., and F. Knopf. 1994. Prairie conservation in management to conserve natural enemies of arthropod North America. BioScience 44: 418Ð421. pests in agriculture. Annu. Rev. Entomol. 45: 175Ð201. SAS Institute. 2008. SAS/STAT, version 9.2. SAS Institute, Landis, D. A., F. D. Menalled, A. C. Costamagna, and T. K. Cary, NC. Wilkinson. 2005. Manipulating plant resources to en- Schmidt, N. P., M. E. O’Neal, and L. A. Schulte Moore. 2011. hance beneÞcial arthropods in agricultural landscapes. Effects of grassland habitat and plant nutrients on soy- Weed Sci. 53: 902Ð908. bean aphid and natural enemy populations. Environ. En- Le Fe´on, V., F. Burel, R. Chifflet, M. Henry, A. Ricroch, B. E. tomol. 40: 260Ð272. Vaissie`re, and J. Baudry. 2013. Solitary bee abundance Simpson, E. H. 1949. Measure of diversity. Nature 163: 688Ð and species richness in dynamic agricultural landscapes. 688. Agric. Ecosyst. Environ. 166: 94Ð101. Smith, D. D. 1998. Iowa prairie: original extent and loss, Lefko, S. A., M. E. Rice, and L. P. Pedigo. 1999. Producer preservation and recovery attempts. J. Iowa Acad. Sci. perception and pest management practices in Iowa al- 105: 94Ð108. falfa. J. Prod. Agric. 12: 257Ð263. Steffan-Dewenter, I., U. Munzenberg, C. Burger, C. Thies, Lundgren, J. G. 2009. Nutritional aspects of non-prey foods and T. Tscharntke. 2002. Scale-dependent effects of in the life histories of predaceous Coccinellidae. Biol. landscape context on three pollinator guilds. Ecology 83: Control 51: 294Ð305. 1421: 1432. April 2014 GILL ET AL.: BUFFER STRIPS IMPROVED WITH NATIVE PLANT SPECIES 311

Triplehorn, C. A., and N. F. Johnson. 2005. Borror and De- [USDA–NRCS]. United States Department of Agriculture. LongÕs Introduction to the study of insects, 7th ed. Thom- 2011. United States Department of Agriculture. The son Brooks/Cole, Belmont, CA. plants database natural resource conservation service, Tscharntke, T., R. Bommarco, Y. Clough, T. O. Crist, D. national plant data team, Greensboro, NC. (http://plants. Kleijn, T. A. Rand, J. M. Tylianakis, S. van Nouhuys, and usda.gov) (accessed 27 June 2012) S. Vidal. 2008. Conservation biological control and en- Wackers, F. L., P.C.J. van Rijn, and G. E. Heimpel. 2008. emy diversity on a landscape scale. Biol. Control 45: Honeydew as a food source for natural enemies: making 238Ð253. the best of a bad meal. Biol. Control 45: 176Ð184. Tuell, J. K., A. K. Fiedler, D. A. Landis, and R. Isaacs. 2008. Weiser, L. A., J. J. Obrycki, and K. L. Giles. 2003. Within- Visitation by wild and managed bees (Hymenoptera: Þeld manipulation of potato leafhopper (Homoptera: Ci- Apoidea) to eastern U.S. native plants for use in conser- cadellidae) and insect predator populations using an un- cut alfalfa strip. J. Econ. Entomol. 96: 1184Ð1192. vation programs. Environ. Entomol. 37: 707Ð718. Weiser Erlandson, L. A., and J. J. Obrycki. 2010. Predation of [USDA–NOP] United States Department of Agriculture– immature and adult Empoasca fabae (Harris) (Hemiptera: National Organic Program. 2009. National Organic Pro- Cicadellidae) by three predatory insect species. J. Kans. gram, Production and Handling Preamble. (http:// Entomol. Soc. 83: 1Ð6. www.ams.usda.gov/AMSv1.0/ams.fetchTemplateData.do? Westrich, P. 1996. Habitat requirements of central Euro- ϭ ϭ template TemplateN&navID NOSBlinkNOSBMeetings pean bees and the problems of partial habitats, pp. 1Ð16. ϭ ϭ &rightNav1 NOSBlinkNOSBMeetings&topNav &left In A. Matheson, S. L. Buchmann, C. OÕToole, P. Westrich, ϭ ϭ ϭ ϭ Nav &page NOPOrganicStandards&resultType &acct and I. H. Williams (eds.), The conservation of bees. Ac- nopgeninfo) (accessed 27 June 2012) ademic Inc., USA. [USDA–NRCS] United States Department of Agriculture– Zhang, W., T. H. Ricketts, C. Kremen, K. Carney, and S. M. Natural Resource Conservation Service. 2010. Restora- Swinton. 2007. Ecosystem services and dis-services to tion of rare and declining habitat program fact sheet, CRP agriculture. Ecol. Econ. 64: 253Ð260. Practice CP25. Natural Resource Conservation Service, Washington DC. Received 28 January 2013; accepted 9 January 2014.