Does floral farmscaping really improve insect biological control in vegetable systems of the Coastal Plain?

Final Report for LS09-220

Project Type: Research and Education
Funds awarded in 2009: $165,000.00
Projected End Date: 12/31/2012
Region: Southern
State: Georgia
Principal Investigator:
Peter Hartell
University of Georgia
Co-Investigators:
John Ruberson
University of Georgia
Expand All

Project Information

Abstract:

Floral farmscaping is the planting of flowering plants in proximity of target crops in order to attract and enhance the populations, fitness, and biological control efficacy of natural enemies. Flowering plants provide food resources such as pollen, floral, and extrafloral nectar for natural enemies. These food resources can be critical for survival and reproduction of natural enemies, and have therefore provided a means of manipulating natural enemies to enhance their biological control efficacy for pest management, in cropping systems. Flowering plants differ in their capacity to supply these food resources; therefore, it is important in designing a farmscaping system to screen potential flower plants to identify and work with those that attract and support desired natural enemies, while excluding those that might compromise the intended goal of pest suppression.

Predatory arthropods play important roles in natural pest control in agroecosystems. Simplification of modern agroecosystems through monoculture cropping practices has led to decreased abundance, diversity, and impact of these predatory insects. However, due to side effects of chemical insecticides on the environment and human health, natural pest control has grown in emphasis. Abundance, fitness, and biological control efficacy of predatory arthropods can be enhanced by providing them access to non-prey food sources. In the present study, we assessed the suitability of three flower treatments (buckwheat, Fagopyrum esculentum (Moench); a combination of fennel, Foeniculum vulgare (Mill.), and dill, Anethum graveolens (L.); and a combination of sunflower, Helianthus annuus (L.), and yarrow, Achillea millefolium (L.)) for enhancing predator abundance and predation of sentinel eggs of the beet armyworm, Spodoptera exigua (Hübner), in organic broccoli and cucumber systems in Athens and Tifton, Georgia, from 2010 to 2012. There was no evidence for effects of the flower treatments on predator abundance or efficacy. There were few and inconsistent significant differences among treatments in number of sucking predators/plant in cucumber (only Athens in 2011) and total numbers of predators/plant in cucumber (only Athens in 2010). Predation of beet armyworm eggs did not differ among treatments within locations and years. The few significant differences likely reflected random events rather than the effect of treatments, since they were unusual and inconsistent across locations and years.

We assessed the suitability of three flower treatments (buckwheat, Fagopyrum esculentum (Moench); a combination of fennel, Foeniculum vulgare (Mill.) and dill, Anethum graveolens (L.); and a combination of sunflower, Helianthus annuus (L.) and yarrow, Achillea millefolium (L.)) for enhancing biological control of lepidopteran pests in an organic broccoli production system over three years at two locations. Lepidopteran pest composition varied across years and locations with Plutella xylostella (L.), being the dominant pest in Athens in 2010 and Tifton in 2010 and 2011, while Pieris rapae (L.) was dominant in Athens in 2011 and Tifton in 2012. Diadegma insulare (Cresson) was the dominant parasitoid of P. xylostella in both locations and across the years of the study, while parasitism of P. rapae in Athens in 2011 and in Tifton in 2011 and 2012 was dominated by Tachinids, and by Pteromalus puparum (L.) in Tifton in 2010. There were inconsistent significant differences among treatments, such as in % parasitism of all lepidopteran pupae/plant in Athens in 2011; P. rapae larval density in Tifton in 2011; % parasitism of P. rapae pupae/plant in Athens in 2011; and P. xylostella larva density in Athens in 2011. These significant differences likely reflected random events rather than the effect of treatments, since they were unusual and inconsistent across locations and years.

We investigated the effects of two flowering plants (buckwheat, Fagopyrum esculentum (Moench) and Indian blanket, Gaillardia pulchella Foug.) on adults of the southern green stink bug, Nezara viridula (L.) (Heteroptera: Pentatomidae) parasitoid, Aridelus rufotestaceus (Tobias) (Hymenoptera: Braconidae). We also assessed the suitability of three flower treatments (buckwheat; a combination of fennel, Foeniculum vulgare (Mill.) and dill, Anethum graveolens (L.); and a combination of sunflower, Helianthus annuus (L.) and yarrow, Achillea millefolium (L.)) for enhancing parasitism of lepidopteran pests in an organic broccoli production system, and predation of sentinel eggs of the beet armyworm, Spodoptera exigua (Hübner) (Lepidoptera: Noctuidae), in organic broccoli and cucumber systems. Aridelus rufotestaceus lived longer on flowers and 5% honey solution than on water alone. Feeding on Indian blanket and 5% honey solution increased production of mature ova. Apart from few inconsistent significant differences among treatments in the response variables, the flower treatments did not enhance parasitism of lepidopteran pests, as well as predation of S. exigua eggs. The results imply that F. esculentum and G. pulchella can benefit A. rufotestaceus for managing N.viridula. Our results on parasitism of lepidopteran pests and predation of S. exigua eggs might have been confounded by the size of the plots, interactions among predators and available prey, and history of the land.

The lab experiment with Aridelus rufotestaceus demonstrated the value of flowering plant resources for parasitoids; however, the lab data failed to translate into significant differences for either predation or parasitism in the field. The field studies were dominated by insect pests and natural enemies that have significant mobility, and the small plots used (12×12 meters) may have been too small to allow pests and natural enemies to function independent of the larger landscape. If this is the case, then it would indicate that there is a spatial threshold below which floral farmscaping may be ineffective, and this may have important ramifications for smallholders and limited-land producers, especially if their production system is embedded in a diversified landscape on a larger scale. Limited progress was made on Objective 2 (characterizing the species of natural enemies and pests coming to the flowers and building a database on the species recorded with the intent to develop a field guide). A variety of species were recorded, and these are listed in the text of the report. However, the systematic analyses were not completed as planned.

Project Objectives:

OBJECTIVE 1: Quantify the impact of floral farmscaping on vegetable pest management, specifically on biological control of arthropod pests by the full suite of naturally occurring enemies in broccoli and cucumbers. We will test several hypotheses to address Objective 1, based on the generally accepted, but unproven, ideas regarding farmscaping:
(1) Beneficial arthropod abundance and diversity will be greater in plots with floral farmscaping than plots without floral farmscaping, and furthermore, buckwheat will enhance abundance and diversity more than the other farmscape treatments because of its abundant nectar;
(2) Abundance and diversity of natural enemies will decrease with increasing distance from the central floral resource, and
(3) Biological control of arthropod pests will be enhanced when flower farmscaping is present in plots, and efficacy will decrease with increasing distance from the center floral resource.

OBJECTIVE 2: Develop a database of arthropod natural enemies attracted to selected flower candidates for farmscaping in the southeastern United States. We will evaluate a selected suite of flowering plants for the pests and natural enemies they attract for future screenings in farmscape systems. These surveys will form the basis for development of an online field guide to arthropod natural enemies and pests attracted to various flowering plants. The field guide will be initiated with this project, but will require several additional years to complete.

We will characterize the density, species diversity and richness of pest and beneficial species in the agronomic crop and the farmscape, the rates of predation and parasitism of pests in the agronomic crops, and correlate the yield and quality of the produce with the overall pest abundance and damage. These data will permit us to test the outlined hypotheses, and assess mechanisms underlying the outcomes.

Unfortunately, limited progress was made on Objective 2 (characterizing the species of natural enemies and pests coming to the flowers and building a database on the species recorded with the intent to develop a field guide). A variety of species were recorded, and these are listed in the text of the report. However, the systematic analyses were not completed as planned.

Introduction:

Beneficial organisms provide critical ecosystem services, among which natural pest suppression is of considerable importance to mankind (Campbell et al., 2012). Natural pest control is important for global food security and has been valued in global cropping systems at $14 billion/year (Costanza et al., 1997), $4.5 billion/year in the United States alone (Losey and Vaughan, 2006), and up to $100/ hectare/ year in Canterbury, New Zealand (Sandhu et al., 2008). It is even more critical in food production in developing countries, where many farmers depend almost entirely on natural pest control for pest management (Wyckhuys et al., 2013).

Natural pest control is provided by natural enemies such as predators and parasitoids, which can be introduced into native ecosystems from exotic origins to provide permanent pest control of typically exotic pests (classical biological control), periodically released to establish control (augmentative biological control), or whose environment can be manipulated to enhance their populations or efficacy for pest control (conservation biological control) (DeBach, 1964; Hajek, 2004; Perdikis et al., 2011). The last approach (conservation biological control) is the most sustainable because it focuses on the resident natural enemies, integrates with production practices, and reduces the problems that beset other pest management approaches, such as environmental degradation, pest resistance to chemical pesticides, and ecosystem risks associated with introduction of exotic species. One way to implement conservation biological control is through farmscaping.

Farmscaping is a pest management strategy that involves the use of insectary plants, hedgerows, cover crops, beetle banks, and water reservoirs to provide valuable resources that attract and enhance populations of beneficial organisms, such as insects, birds, and bats, and enhance their populations for pest control (Dufour, 2000). It is an ecological and sustainable approach to pest management that must be able to integrate economically, environmentally, and socially with production practices.

When flowering plants are used for farmscaping, it is termed floral farmscaping. Therefore, floral farmscaping is the planting of flowering plants in proximity to target crops in order to attract and enhance the populations, fitness, and biological control efficacy of natural enemies of pests, for enhanced natural pest control. Floral farmscaping is a form of conservation biological control (Ehler, 1998; Landis et al., 2000a). It also can be viewed as a form of ecological engineering, which involves habitat manipulation with cultural techniques to enhance biological pest control (Gurr et al., 2004).

Historical background of floral farmscaping
One of the earliest mentions of natural enemies visiting flowers was by Froggatt (1902), who reported Scolia formosa (Guérin-Méneville), a parasitoid of the grey cane beetle, Lepidoderma albohirtum (Waterhouse), visiting a flower. Subsequent records include Allen (1929); King and Holloway (1930); Nishida (1958).

However, not until the early 1930’s was the connection between flower visit and biological control first made by Clausen et al. (1933), who reported that adult food (honeydew from aphids and nectar from flowers of umbelliferous and polygonaceous plants) was a major factor for Tiphia matura (Allen and Jaynes) (Hymenoptera: Tiphiidae), a parasitoid of Popillia cupricollis (Hope) (Coleoptera: Scarabaeidae), limiting distribution and biological control effectiveness. Later, Wolcott (1942) reported that Larra americana (Saussure), a parasitoid of mole cricket, Scapteriscus vicinus (Scudder), was successfully introduced from Brazil into Puerto Rico because two local weeds: Borreria verticillata (L.) and Hyptis atrorubens (Poit.) provided nectar for the wasp.

One of the earliest records of targeted planting of insectary plants to provide food source and shelter for parasitoids was in New Zealand apple orchards in the 1960’s, in which the Australian shrubs Acacia spp., Eucalyptus spp., Grevillea rosmarinifolia (A.Cunn.), Hakea laurina (R.Br.), and Citrus sp. were planted to support populations of the parasitoids Trichogramma spp. and Apanteles spp. for control of the light brown apple moth, Epiphyas postvittana (Walker) (Collyer and van Geldermalsen, 1975). Subsequently, Yan and Duan (1988) reported that planting white sweet clover, Melilotus albus (Desr.) between rows of apple trees had positive effects on predator community in the trees. And from 1990, the literature on the use of insectary plants as food sources to enhance population of natural enemies for pest control has grown considerably, for example, Haley and Hogue (1990); Bugg et al. (1991); Wyss (1995); Stephens et al. (1998); Bostanian et al. (2004); Blaauw and Isaacs (2012); Díaz et al. (2012a); Gontijo et al. (2013).

Benefits of floral farmscaping
Floral farmscaping can confer the following benefits when adequately implemented in the field. It can lead to reductions in the amount of pesticides that a farmer may need to use as part of an integrated pest management (IPM) program. This can reduce operation costs for the farmer, health risks associated with chemical residues from the pesticides application to humans, and deleterious effects on natural enemies, thereby conserving them. For example, (Yan et al., 1997) reported that a section of apple orchard managed with alfalfa, Medicago sativa (L.)/Lagopsis supina (Steph.)), cover crop experienced mite (Tetranychus ulmi (Koch) and T. vevennensis (Zacher)) infestations below economic threshold and did not require any pesticide application unlike the section without the cover crop. It also can provide an opportunity to conserve native flora, when native plants are integrated into the system. Further, it can be an additional source of income to farmers, especially when high-value flowering plants are used (Dufour, 2000; Landis et al., 2000a).

Plant food resources for natural enemies
Many natural enemies (parasitoids and predators) are omnivores and, therefore, utilize non-host food (plant-based diets) in addition to host or prey (animal-based diets) in order to optimize their life histories (Coll and Guershon, 2002; Eubanks and Styrsky, 2005). Although natural enemies feed on a combination of these diets, the extent and timing at which they utilize these food resources vary. Many of them feed on a combination of these two diets throughout their feeding life stages (predators such as coccinellids and heteropteran predators), and therefore are referred to as lifelong omnivores (Eubanks and Styrsky, 2005). Others feed on one of these diets at only certain life stages, (e.g., syrphid flies, which feed solely on prey in larval stages and only on a plant-based diet as adults), and are referred to as life-history omnivores (Wäckers and van Rijn, 2005). Yet others feed on a combination of these diets only in the adult stage (e.g., parasitoids that feed on hosts during their larval stages and on a combination of host and plant-based diets in the adult stage).

Omnivory by these natural enemies is particularly well-suited for biological control of pests in ephemeral agroecosystems because feeding on multiple trophic levels allows these natural enemies to survive and remain in the area when hosts or prey are scarce, reducing their risk of starvation or emigration from the taget crop area (Eubanks and Denno, 1999; Beckman and Hurd, 2003; Welch et al., 2012). And as a result, they may continue to feed on pests at low densities, driving them to local extinction, thereby benefiting pest management. This is in contrast to what might happen to obligate, and especially specialist predators at low prey densities (Eubanks and Styrsky, 2005).

Plant-based diets include nectar (floral and extrafloral), phloem sap, and pollen (Olson et al., 2005; Wäckers, 2005a; Wäckers and van Rijn, 2005; Lundgren, 2009a). Floral nectar is among the rewards that plants use to recruit pollinators and is derived from both phloem and xylem sap or phloem sap alone, and is secreted by nectaries. It is composed primarily of carbohydrates, but may contain some amino acids, vitamins, lipids and secondary plant metabolites (Fahn, 1988; Wäckers, 2005a).

Extrafloral nectar is produced from glands located in various plant parts external to the flower, such as stems, leaves, fruits, and bracts. Extrafloral nectar is usually not involved in plant pollination, but is used mainly to recruit parasitoids and predators for plant defense (Koptur, 1992). Similar to floral nectar, extrafloral nectar is composed primarily of carbohydrates, with some amino acids, lipids, and vitamins (Fahn, 1988; Wäckers, 2005a).

Pollen is the means of transferring male genetic information of plants from anthers to the stigma, and so is important for plant reproduction. It is also an important reward offered by insect-pollinated flowers to pollinators. Pollen is composed primarily of free amino acids and proteins with variable amounts of carbohydrates, lipids, and sterol (Roulston and Cane, 2000; Wäckers, 2005a).

These plant food resources have been reported to improve natural enemy development rates, survival, fecundity, dispersal, and distribution (Addison et al., 2000; Eubanks and Styrsky, 2005; Witting-Bissinger et al., 2008b; Díaz et al., 2012a; Géneau et al., 2012; Portillo et al., 2012). Improvement in these life history traits can be important in biological control. For example, decreased development time can provide opportunity for natural enemies to produce more generations in a season. Increased longevity can increase the length of time natural enemies may have access to prey and hosts (pests), while increased fecundity increases number of offspring natural enemies may have, and thereby enhances the numerical response of the enemies to pest populations, and increasing natural enemy populations with a concomitant increase in pest consumption.

Because natural enemies are associated with these food resources and because these food resources can enhance their life history traits, floral farmscaping provides an opportunity for biological control practitioners to deploy these resources to enhance biological control in cropping systems (Lundgren, 2009b). Planting of insectary plants in association with target crops for the purpose of pest management in cropping systems is based on this association.

Floral plants as sources plant-derived food
Pollen, floral nectar, and extrafloral nectar are produced by flowering plants. These flowering plants can be annual, in which they complete their lifecycle in one growing season (e.g., dill, Anethum graveolens (L.)), or perennial, in which they complete their lifecycle in more than two years (e.g., deergrass, Muhlenbergia rigens (Benth.) Hitchc). Perennial flowering plants tend to be better suited for insectary plants because they provide a more persistent resource for natural enemies, unlike the annuals that die at the end of the season, and may have to be replanted (Long et al., 1998; Landis et al., 2000a; Sokhangoy et al., 2012).

Flowering plants vary in their capacity to supply these food resources, in quantity, quality, accessibility, and length of supply. With respect to nutrient contents, some nectar is ‘sucrose-dominant’, for example buckwheat, Fagopyrum esculentum (Moench) nectar, while some are ‘hexose-dominant,’ for example coriander, Coriandrum sativum (L.). Sugar consumption may increase osmotic pressure in insects, with physiological consequences such as destabilization of water balance. This increase is more rapid with consumption of nectars dominated by monosaccharides such as glucose and fructose than disaccharide-dominant nectars (Baker and Baker, 1983; Vattala et al., 2006). The quantity of nectar that a flower produces can determine the number of visitors it receives as insects are able to discriminate flowers based on nectar volume (Goulson, 1999). The size and shape of nectaries can significantly affect the community of natural enemies that benefits from a particular plant. Nectaries of some plants are highly accessible, typically characterized by shallow and wide corollae apertures, for example buckwheat (0.54 mm deep and 6.59 mm wide;(Baggen et al., 1999a)), while others limit access to certain natural enemy species by being too deep or narrow for many species to utilize. Some plants, such as Indian blanket, Gaillardia pulchella (Foug.) (observed to flower for 7-8 months in the field in southern Georgia), flower for prolonged periods of time, while some flower for a shorter period, for example buckwheat (observed to flower for 3-4 weeks in the field in southern Georgia), limiting their utility for longer growing seasons without staggered plantings.

Because of these variations, flowering plants differ in the types of natural enemies that they attract and their value to those natural enemies, and it is important to evaluate them in order to identify those that maximize these food resources for natural enemy utilization before they can be deployed as insectary plants.

Deployment of insectary plants
Insectary plants can be deployed in the field as ‘flower strips’, alternated with target crops, or they can planted at borders surrounding the target crops. They can also be planted at the center field to create a “halo effect”, so that when natural enemies are attracted they can move into the crop section to control the pests.

Natural enemies vary in the distances that they can travel; for example, Hippodamia convergens (Guérin-Méneville) can travel more than 1 km (Sivakoff et al., 2012), while Tachinids are known to have long flight ranges of up to 200 m (Romina et al., 2011; Pfannenstiel et al., 2012). Therefore, it is important to locate insectary plants appropriately in space within or around crop fields, in order to maximize natural enemy area of influence.

Because flowers differ in when they initiate flowering and how long they subsequently flower, it is important in choosing flowering species mixes to select those that can complement each other temporally and extend the availability of appropriate floral resources for natural enemies.

Does floral farmscaping enhance biological control?
Since the importance of floral resources for fitness of natural enemies became evident, numerous studies have been carried out to determine how they can be used to enhance biological control and manage agricultural pests. A look at those studies shows that results have been equivocal. Some of those studies show positive effects of planting insectary plants in association with target crops with respect to pest control, some show negative effect of insectary plants, with increased risk of exacerbating pest problems, while others show negligible or neutral effects of insectary plants on pest control (see attached Table 1).

Possible reasons for the mixed results in previous studies
Potential reasons for the conflicting results obtained with floral farmscaping include:
1) Natural enemy feeding habit: many predators and parasitoids in agroecosystems are polyphagous, and feed on multiple taxa of prey (Coll and Guershon, 2002; Welch et al., 2012). These polyphagous natural enemies may switch prey based on preference [preferred (high-quality) vs. alternative (low-quality) prey] or nutritional demands and, therefore, the presence of preferred prey may influence how these polyphagous natural enemies respond to alternative prey. This presents problems in biological control, especially when the target prey is not the preferred prey for natural enemies, with the predators not responding as desired to pests of interest (Welch et al., 2012). Natural enemies can also feed on plant food and become satiated and not respond adequately to pests.

2) Competition for resources: natural enemies may compete among themselves for floral resources, when they are limited, thereby resulting in competitive exclusion of some. In some cases some beneficial insects may interfere with the ability of others to utilize floral resources. For example, Campbell et al. (2012) showed that parasitoid visitation to a flower of short corolla length was reduced by 50% when the flowers were mixed with other flowers of long corollae length, possibly due to competitive interference from bumble bees, which prefer flowers with long corolla length.

3) Wrong insectary plants: flowers differ in their capacity to supply plant food resources and, therefore, using wrong plants may end up not attracting the desired natural enemies. Floral resources may be more nutritious than prey and may draw the natural enemies away from the prey. On the contrary, floral resources may be nutritionally poor may be inadequate to sustain natural enemies in time of prey scarcity. Insectary plants may differentially favor recruitment of pests relative natural enemies.

4) Relative size of insectary plant land: the area of land devoted to insectary plants relative to target crop land can be too small to produce effects. Pfiffner and Wyss (2004) recommended that at least 10% of intensively cultivated area be set aside as wildflower strips for natural enemy conservation. In the lettuce-alyssum system in California, ~4% of cultivated land is devoted to natural enemy conservation, and this system has been successful in managing the currant-lettuce aphid (Nasonovia ribisnigri Mosley) with attracted hoverflies (Gillespie et al., 2011). The work by Gillespie et al. (2011) on the lettuce-alyssum system suggests that the area set aside for natural enemies can be reduced to 2% of the crop area without any effect on hoverfly abundance in the crop section. There are likely system-specific variations in the extent of area that must be devoted to insectary plants relative to target crop to achieve positive biological control results.

5) Natural enemy mobility: natural enemies vary in the distance and speed at which they travel in the field and therefore using insectary plants that attract mostly slow- moving natural enemies may not be as effective as those that attract considerable number of fast-moving natural enemies. Fast-moving natural enemies with long dispersal range, depending on their foraging strategies, may likely encounter more pests and consequently improve biological control.

Present study
In view of these mixed results and potential system differences in how insectary plants can be used to enhance biological control, it is apparent that more work is needed to answer the question: does floral farmscaping enhance biological control? Additional studies are necessary to develop location- and system-specific farmscape systems for pest management.

Hypotheses Tested
  1. The availability of flowers will lower pest density and enhance their parasitism in an organic broccoli cropping system.

    Buckwheat will lower pest density and enhance parasitism more than other flowers tested in an organic broccoli system because of its copious nectar production.

    The presence of flowers will enhance predator abundance and predation of Spodoptera exigua (Hübner) eggs in organic broccoli and cucumber systems.

    Buckwheat, in particular, will enhance predator abundance and predation of S. exigua eggs more than other flowers tested in organic broccoli and cucumber cropping systems because of copious nectar production and ease of its access in buckwheat.

    Flowering plants will enhance longevity and fecundity of an important model parasitoid, Aridelus rufotestaceus (Tobias) (Hymenoptera: Braconidae), more than water alone.

    Buckwheat will enhance parasitoid longevity and fecundity (of A. rufotestaceus) more than other flower treatments because of buckwheat’s copious nectar production and ready accessibility.

Cooperators

Click linked name(s) to expand
  • Juan Carlos Diaz-Perez
  • Dawn Olson
  • Rick Reed
  • David Riley
  • Relinda Walker

Research

Materials and methods:
Parasitism and Farmscaping in Broccoli

Floral plants
The treatments used were: 1) broccoli (control), 2) buckwheat, Fagopyrum esculentum (Moench) (Caryophyllales: Polygonaceae), 3) a combination of fennel, Foeniculum vulgare (Mill.) (Apiales: Apiaceae) and dill, Anethum graveolens (L.) (Apiales: Apiaceae), and 4) a combination of sunflower, Helianthus annuus (L.) (Asterales: Asteraceae) and yarrow, Achillea millefolium (L.) (Asterales: Asteraceae). All of these flowering plants were selected because they are either native to North America (in the case of sunflower) or cultivated for food and industrial uses, as well as for attractants for natural enemies in cropping systems.

Sunflower is grown mainly for its seeds, which are a source of oil for food, feed for animals, and other industrial raw materials (Lu and Hoeft, 2009; Fernández-Martínez et al., 2010), and is known to attract a diverse array of insects, including natural enemies (Jones and Gillett, 2005; Adedipe and Park, 2010). It is an annual plant that produces capitulae of 2 to 30 cm in diameter, depending on the variety (Cronn et al., 1997; Fambrini et al., 2007). The capitulum bears two kinds of flowers: ray and disc florets. Ray florets sometimes have nectaries that are usually smaller than those of disc florets and, therefore, produce less nectar than disc florets. Average disc floret corolla length ranges from 7.23 to 10.22 mm, while the mean nectar production per floret per day ranges from 0.24 to 0.38 µL (Atlagi? et al., 2003; Wist and Davis, 2006; Hadisoesilo and Furgala, 1986).

Yarrow is native to Western Asia and Europe, but is grown in most temperate regions, including the United States. It is mainly cultivated for medicinal uses and it is becoming popular as an attractant for natural enemies (Applequist and Moerman, 2011; Dib et al., 2012). It is perennial and as a member of Asteraceae produces a flower head about 2 to 4 mm wide comprising about five 5 to six 6 ray florets and 10 to 30 disk florets, with corollae 2.2 to 3 mm long (Zhang et al., 1996; Sulborska and Weryszko-Chmielewska, 2006 ; Warwick and Black, 1982).

Dill is an annual flowering plant that is native to the Mediterranean region and is used mainly as a spice and medicine (Sokhangoy et al., 2012; Carrubba et al., 2008; Tian et al., 2012). It is also being used as a companion plant in cropping systems (Winkler et al., 2010). It produces flowers with no corolla depth, with average aperture size of 2.63 mm and easily accessible nectar (Winkler et al., 2009).

Fennel is a perennial flowering herb that is native to the Mediterranean region and cultivated for spice and medicinal use (Gross et al., 2008). It produces flowers with no corolla depth and easily accessible nectar (Winkler et al., 2009).

Buckwheat is native to Asia and is widely cultivated in many regions of the world, including the United States, for food and habitat management (Li and Zhang, 2001; Ohnishi, 1990; Wijngaard and Arendt, 2006; Lee and Heimpel, 2008). It has shallow corollae with wide apertures, which make its nectar easily accessible to many insects (Sim and Choi, 1999; Vattala et al., 2006).

Experimental design
The experiment was conducted in two locations: Athens (The University of Georgia’s Athens Horticulture Research Farm) and Tifton (The University of Georgia’s Tifton Horticulture Research Farm), Georgia, from 2010 to 2012. The study was conducted in spring 2010 and 2011 in Athens and spring 2010, 2011, and 2012 in Tifton. The land used in Athens was in transition to organic certification, while that used in Tifton was organically certified and had been in organic production for two years before the start of the experiment. The broccoli (var. Windsor F1), dill (var. Bouquet), fennel (var. Bronze), and buckwheat seeds used in the study were organic and were purchased from Johnny’s Selected Seeds (http://www.johnnyseeds.com/default.aspx), while sunflower (var. Sunbright F1) was not organic and was obtained from the same vendor. Yarrow (White) was not organic and was obtained from Peaceful Valley Farm & and Garden Supply (http://www.groworganic.com)
The year prior to spring planting, the land in Athens and Tifton was planted with cover crops [Athens: Austrian winter peas, Pisum sativum (L.) (Fabales: Fabaceae) and oats, Avena sativa (L.) (Poales: Poaceae), planted in late fall; Tifton: sunn hemp, Crotalaria juncea (L.) (Fabales: Fabaceae), planted in early summer]. The flower plants were started in the greenhouse between January and February of each planting season to make sure that the flowers were at the flowering stage at the time of transplanting. Flowers were also replanted as needed every 3-4 weeks to ensure that flowers were present in the field during sampling. Broccoli for both locations was started in the greenhouse between January and February by planting in Fafard germination mix in multi-celled trays with circular cells of 7.6 cm diameter. In Athens, broccoli (15-20 cm tall) and flower seedlings were transplanted to the field between 1 and 30 April in 2010 and between 4 and 20 April in 2011. In Tifton, broccoli (15-20 cm tall) and flower seedlings were transplanted on 22 and 23 March in 2010, 14 and 15 March in 2011, and 14 and 15 March in 2012.

Field Layout
Broccoli at each location and in each year was transplanted into 16 blocks (4 replicates of 4 treatments) measuring 12×12 meters (144 m2) and separated from one another by 3-m borders of bare soil on all sides of the plot. Each block contained 6 twin-row beds, each measuring 12 m long x 1.2 m wide. Broccoli was planted in Athens and Tifton at a spacing of 0.46 m between rows and 0.46 m between plants within rows. A treatment plot measuring 2×2 meters was established in the center of each block in which the respective flower or control treatments were placed, with the flower plants planted on the adjacent halves of the third and fourth beds. The treatments with two flowers – dill/fennel and yarrow/sunflower – were planted in such that one flower species was planted on one half of one bed and the other flower on the other half of the other bed, and alternated across the blocks to avoid bias. Forty flowering plants were planted in each treatment plot, 20 plants on each half of the bed, in Athens. Forty Forty-eight flowering plants were planted in each treatment plot, 24 plants on each half of the bed, in Tifton. The flowers were close to flowering when transplanted and were transplanted immediately after broccoli. Ten broccoli or cucumber (depending on the trial) plants were planted in the control plots, five plants on each half of the bed, in both locations.

The treatment plots were laid out in a randomized complete block design in Athens (because of land constraints) and Latin square design in Tifton. The plots in Athens were fertilized with feathermeal (11-14% N) at the rate of 3 g/planting hole, and in Tifton pelletized poultry litter was applied at a rate of 14673480 kg/ha (3% N for 104 kg/ha of N) to the field in mid-February of each year. Water was supplied with drip irrigation as needed at both locations. Weeds were controlled was done by tillage and hand-pulling in 2010, and with by tillage and black plastic mulch (0.25 ml) covering the beds in 2011 and 2012 at both locations.

Data collection and analyses
After transplanting the plants (broccoli and flowers), they were allowed 1-2 weeks to establish before sampling began. In each location, broccoli plants were sampled for lepidopteran larvae and pupae at the center of the block next to the flowering plants, and 4 m away from the center in the four cardinal directions, resulting in five sampling positions (east, west, north, south, and center) per block; i.e., five plants per block and 80 plants in each location. The plants were sampled once a week between 8:00 am and 12:00 pm in Athens 2010 (May 22 to June 8), Athens 2011 (April 26 to June 23), Tifton 2010 (April 20 to May 25), Tifton 2011 (April 13 to June 7), and Tifton 2012 (March 29 to May 24). Each plant was sampled by examining the leaves, stem, and broccoli head and collecting all the lepidopteran larvae and pupae on itobserved. Larvae and pupae collected from each plant were individually held in diet cups (35 cm2) and feeding stages were provided fresh broccoli leaves until pupation. All specimens were held in the laboratory at 26 ± 2°C and RH of 60 ± 10% until either the lepidopteran adult or parasitoids emerged. Dominant Most dominant parasitoids were identified to species level and the remainder to family.

Yield data for broccoli were obtained by harvesting broccoli heads and measuring their diameter. A maximum of 24 broccoli heads (depending on availability) were harvested per block (four heads/row on three rows of broccoli adjacent to both sides of flower treatment plot) and a maximum of 384 heads/location (depending on availability). Weekly harvests were conducted for 2-4 weeks, depending on condition of crop.

Data on hosts/plant, parasitized hosts/plant, and % parasitism of hosts/plant were collected for total lepidopteran hosts (larvae and pupae), Pieris rapae (larvae and pupae), and Plutella xylostella (larvae and pupae) for each location and each year, and analyzed using generalized linear mixed models (repeated measures two-way ANOVA), with block as a random effect (SAS, 2010). Percentage parasitism was transformed using arcsin?% before analysis with generalized linear mixed models (repeated measures two-way ANOVA). Data on broccoli head diameter for each harvest, location and year, were also analyzed using generalized linear mixed models (repeated measures two-way ANOVA) with block as a random effect.

Predation and Farmscaping in Broccoli and Cucumbers

Floral plants
The treatments used were: 1) broccoli or cucumber (control), 2) buckwheat, Fagopyrum esculentum (Moench) (Caryophyllales: Polygonaceae), 3) a combination of fennel, Foeniculum vulgare (Mill.) (Apiales: Apiaceae) and dill, Anethum graveolens (L.) (Apiales: Apiaceae), and 4) a combination of sunflower, Helianthus annuus (L.) (Asterales: Asteraceae) and yarrow, Achillea millefolium (L.) (Asterales: Asteraceae). As noted in the parasitism section above, we selected these flowers because they are either native to North America (in case of sunflower) or grown as commercial crops or as attractants for natural enemies in agricultural systems.

Experimental design
The experiment was carried out at experimental plots in two locations, Athens and Tifton, Georgia, from 2010 to 2012. The study was conducted in spring and summer 2010 and 2011 in Athens, and spring 2010, spring and summer 2011, and spring 2012, in Tifton. Broccoli was grown in spring, while cucumber was grown in summer in both locations. The land used in Athens was in transition to organic certification, while that used in Tifton was organically certified and had been in organic production for two years before the start of the experiment. The broccoli (var. ‘Windsor F1’), cucumber ( var. ‘Marketmore 76’) dill (var. ‘Bouquet’), fennel (var. ‘Bronze’), and buckwheat seeds used in the study were organic and were purchased from Johnny’s Selected Seeds (http://www.johnnyseeds.com/default.aspx), while sunflower (var. ‘Sunbright F1’) was not organic and was obtained from the same vendor. Yarrow (White) was not organic and was obtained from Peaceful Valley Farm & Garden Supply (http://www.groworganic.com)

The year prior to spring and summer planting, the land in Athens [(Austrian winter peas, Pisum sativum (L.) (Fabales: Fabaceae), 22.4 kg/ha and oat, Avena sativa (L.) (Poales: Poaceae), 11.2 kg/ha,planted in late fall)] and Tifton [(sunn hemp, Crotalaria juncea (L.) (Fabales: Fabaceae), 33.6 kg/ha planted in early summer)] was planted with cover crops. The flower treatment plants were started in the greenhouse in January and February of each year to make sure that the plants were at the flowering stage at the time of transplanting. Flowers were also replanted in the plots every 3-4 weeks as needed to ensure that flowers were present in the field throughout the sampling period. Broccoli for both locations was started in the greenhouse in January and February, while cucumber in Athens was started in May. All transplant starters were planted in Fafard germination mix in 7.6-cm diameter wells of 24-well trays, under 26 ± 2°C and 14:10 (L:D) photoperiod.

Broccoli (15-20 cm tall) and flowering plants for the broccoli trials in Athens were transplanted to the field from 1 to 30 April in 2010 and from 4 to 20 April in 2011, while cucumber (15-20 cm tall) and flowering plants for cucumber trials were transplanted from 23 June to 7 July in 2010 and from June 15 to 21 in 2011. Broccoli (15-20 cm tall) and flowering plants in Tifton were transplanted on 22 and 23 March in 2010, 14 and 15 March in 2011, and 14 and 15 March in 2012, while cucumbers and the accompanying flower plants were transplanted on 5-6 May 2011.

Field Layout
Broccoli and cucumber were transplanted into 16 blocks (4 replicates of 4 treatments), each measuring 12×12 meters, i.e. 144 m2, and separated from one another by 3 m borders of bare soil, with each block containing 6 twin-row beds, each measuring 12 m long x 1.2 m wide, in each location and each year. Broccoli and cucumbers were planted in Athens and broccoli in Tifton at a spacing of 0.46 m between rows and 0.46 m between plants within rows. A treatment plot measuring 2×2 meters was established in the center of each block in which the respective flower or control treatments were placed. The treatment plots were laid out in a randomized block design in Athens (because of land limitations) for both broccoli and cucumber and in a Latin square design for broccoli and a randomized block design for cucumber in Tifton.

The flowering plants were planted on the adjacent halves of the centers of the third and fourth beds of each plot in the 2×2 m subplot. The treatments with two flower species (dill/fennel and yarrow/sunflower) were planted such that one flower species was planted on one half of one bed and the other flower on the other half of the other bed, and sides were alternated across the blocks to avoid bias. In Athens, 40 flowering plants were planted in each treatment plot, 20 plants on each half of the bed, and 20 plants for each of the flower species in treatments with two flower species. In Tifton, 48 flowering plants were planted in each treatment plot, 24 plants on each half of the bed, and 24 plants for each of the flower species in treatments with two flower species. The flowers were close to flowering when transplanted and were transplanted immediately after broccoli transplanting. Ten broccoli or cucumber (depending on the trial) plants were planted in the control plots, five plants on each half of the bed, in both locations.

The plots in Athens were fertilized with feathermeal (11-14% N) at the rate of 3 g/planting hole (ca. 17 kg/ha of N) and in Tifton pelletized poultry litter was applied at a rate of 1467 kg/ha (3% N for 44 kg/ha of N) to the field in mid-February of each year. Water was supplied with drip irrigation as needed. Weed control was done by tillage and hand-pulling in 2010 and with tillage and black plastic mulch (0.25 ml) covering the beds in 2011 and 2012 at both locations.

Beet armyworm eggs
Beet armyworm egg masses used in the experiment were obtained from a beet armyworm culture maintained at the Entomology Department of the University of Georgia, Tifton Campus. The beet armyworm colony was maintained on artificial diet prepared with pinto bean meal, wheat germ, agar, yeast, ascorbic acid, sorbic acid, and vitamin mixture (Burton, 1969). The beet armyworm colony was held at 25 + 1°C and 14:10 L:D.

Data collection and analyses
After transplanting (broccoli, cucumber, and flowers), plants were allowed about 1-2 weeks to establish before sampling. Individual broccoli and cucumber plants were sampled for predators at the center of the block next to the flowering plants (1 plant), and 1 plant each 4 m away from the center in the four cardinal directions, resulting in five sampling positions (east, west, north, south, and center) per block; i.e., five plants per plot and 80 plants in each location; except in broccoli in Tifton in 2011, where we had three sampling positions (north, central, and south) per block; i.e., three plants per plot and a total of 48 plants. The plants were sampled once a week between 8:00 am and 12:00 pm in Athens 2010 (May 22 to June 8 for broccoli and July 20 to September 5 for cucumber), Athens 2011 (April 26 to June 23 for broccoli and July 6 to July 26 for cucumber), Tifton 2010 (April 20 to May 25), Tifton 2011 (April 13 to June 7 for broccoli and June 2 to July 8 for cucumber), and Tifton 2012 (March 29 to May 24).

Beet armyworm moths of the lab colony were provisioned with a white paper towel oviposition substrate (28×18.5 cm). Sheets with eggs were collected within 12 hours of initiation of oviposition to allow use of young eggs in the trials. Egg masses of 20 to 60 eggs (and with few moth scales on them) on the paper towel were counted under the microscope and cut out to be used as sentinel prey. One egg mass was stapled on the underside of one leaf of the sampled plants (i.e., five egg masses per plot and 80 egg masses per location, except in broccoli in Tifton in 2011, where we used three egg masses per plot and a total of 48 egg masses) and photographed 24, 48, and 72 h [using a Canon dSLR camera (Canon EOS Digital Rebel XT) at distance of about 5 cm from egg masses] after placing the eggs. The photographs were examined and predated eggs at each photo interval were determined by subtracting the eggs remaining at each time from the total eggs that were put out initially, and carefully evaluating the remaining eggs for signs of predation. Sucked eggs were recognized by their collapsed pyramidal shape. Chewed eggs were characterized by extensive damage to the chorions, with yolk sometimes spilled on the paper containing the eggs, and sometimes a part of the paper also was chewed. Some eggs hatched prior to the 72-hour period late in the season when field temperatures were high (hatched eggs were readily recognized by transparent, dry, and empty chorions left after the larvae emerged) and they were excluded from the analyses. Missing eggs were considered incidental loss and were excluded from the analyses. This omission undoubtedly led to an underestimation of predation, but incidental dislodgement of eggs occurs (from rain, heavy dews, wind vibration on leaves, etc.), and in the absence of clear data on the proportion lost due to abiotic factors, we were unable to adequately assign lost eggs to a definitive fate. Further, the number of missing eggs was unaffected by treatment (see below), so outcomes were not affected by their omission.

Data on broccoli yield were obtained by harvesting broccoli heads and measuring their diameter. A maximum of 24 broccoli heads (depending on availability) were harvested per date per block (four heads/row on three rows of broccoli adjacent to each side of flower treatment plot) and a maximum of 384 heads/location (depending on availability). Data on cucumber yield in Athens were obtained by harvesting all fruits on the plants sampled for predators and measuring their diameter. Tifton’s cucumber yield data were obtained by enclosing the plants sampled for predators with a 1×1 m quadrat (with the plants at the center of the quadrat) and harvesting all the fruits within the quadrat. Tifton’s cucumber yield data were converted to per plant before analysis to match those of Athens. Weekly harvests were conducted for 2-4 weeks, depending on crop condition.

Data on predator density (number of chewing predators/plant, number of sucking predators/plant, and total number of predators/plant), percent egg masses accessed by predators by 72 h, percent chewed beet armyworm eggs/plant by 72 h, percent sucked beet armyworm eggs/plant by 72 h and total percent predation of beet armyworm eggs/plant by 72 h for each location and each year were collected and analyzed using generalized linear mixed models (repeated measures two-way ANOVA), with block as a random effect (SAS Institute, 2010). Percentage predation was transformed using arcsin?% before analysis with generalized linear mixed models (repeated measures two-way ANOVA). Data on broccoli head and cucumber diameter for each harvest, location and year, were also analyzed using generalized linear mixed models (repeated measures two-way ANOVA) with block as a random effect.

Effects of Nectar on Parasitoid Fitness

Floral plants
Buckwheat and Indian blanket were assessed for their suitability as non-host food sources for A. rufotestaceus. Buckwheat was chosen because of its abundant nectar production and its flower architecture that favors nectar accessibility, as well as its history of use in similar studies and in commercial production. Buckwheat has shallow corollae with wide apertures, which make its nectar easily accessible to many insects (Sim and Choi, 1999; Vattala et al., 2006). Although native to Asia, it is widely cultivated in many regions of the world, including the United States (Ohnishi, 1990). It starts flowering about one month after sowing and continues flowering for about 6 weeks (Li and Zhang, 2001; Quinet et al., 2004). Its seeds are inexpensive and can be purchased readily from most flower seed companies.

Indian blanket was selected because it is native to the US and has an extended flowering period. It is an annual flowering plant, although some can persist beyond one growing season, and it has the capacity to bloom all year round, depending on the climate (Hammond et al., 2007). It produces flowers with narrow and elongated corollae that can interfere with nectar access for foragers with short mouthparts (Mani and Saravanan, 1999). As a member of Family Compositae, its nectar production is relatively limited but it can be sustained for a long time (Mani and Saravanan, 1999; Hammond et al., 2007).
Organic buckwheat seed was procured from Johnny’s Selected Seeds (http://www.johnnyseeds.com) with product ID: 966G.36. Organic Indian blanket seed (variety SWF230) was obtained from Peaceful Valley Farm Supply (http://www.groworganic.com). Organic seed were chosen to minimize risks of any residual pesticides that might be associated with conventional seed.

The two plant species were planted in organic germination mix, Fafard 20 (obtained from GROSouth; http://www.grosouth.com/), in a greenhouse located at the University of Georgia Entomology Department, Tifton Campus, under these conditions: 26 ± 2°C, 14:10 (light:dark, L:D) photoperiod, and relative humidity (RH) of 60 ± 10%. Buckwheat was sown every three weeks for the duration of the study (from the end of August until December 2011 ) to maintain a constant supply of flowers. Indian blanket was planted once and produced flowers throughout the experimental period. All plants were watered as needed, starting with once every two days and changing to once a day as the plants grew larger.

Parasitoids
Aridelus rufotestaceus were obtained from a culture maintained at the Entomology Department of the University of Georgia, Tifton Campus. The parasitoids had been in colony for two years (ca. 20 generations) and were reared on N. viridula nymphs maintained on shelled sunflower seeds and snap beans at 25 + 1°C and L:D 14:10 after exposure to parasitoids. Newly emerged parasitoids were sexed and only females were used for the experiment (males were rare, less than 5% of emerging parasitoids).

Longevity
Survivorship of adult female A. rufotestaceus was determined with the following food treatments: (1) flowering buckwheat plant, (2) flowering Indian blanket plant, (3) 5% honey solution, and (4) water (control). The number of wasps used for the treatments were 15, 12, 16, and 16 for buckwheat, Indian blanket, 5% honey solution, and water respectively. Newly emerged female wasps were individually placed in transparent plastic cages (15.5 x 10.5 x 5.5 cm) with a hole cut in one side and sealed with a cloth screen to ensure ventilation and permit water and 5% honey solution replacement. Circular holes were also cut in the bottom of each cage to permit introduction of the flowering plants, with the gap around the stems of the flowering plants plugged with cotton batting to prevent the wasps from escaping. One flower head of Indian blanket was used per cage and a cluster of flowers (approximately 20 flowers) of buckwheat was used per cage to ensure an abundant nectar supply. Water and a 5% honey solution were offered in microcentrifuge tubes with holes punctured in the lids and a cotton wick was introduced through the hole to ensure a constant supply of the fluids through capillary action. Water was offered in all the treatments in addition to the four treatments (buckwheat, Indian blanket, 5% honey solution, and water). The cages were held at 25 + 1°C and L:D 14:10, and wasps were observed twice daily until they died.

Fecundity
Fecundity of female A. rufotestaceus was assessed at emergence and five days after receiving the aforementioned four treatments by counting the mature ova in dissected females. A total of 10 newly emerged females (< 24 h post-emergence) and 12, 15, 16, and 9 wasps fed on buckwheat, Indian blanket, 5% honey solution, and water treatments, respectively, for five days, under the same conditions as in the longevity experiment, were collected and immobilized on ice. The females were dissected in PBS (Phosphate Buffered Saline) solution to extract the ovaries. The extracted ovaries were slide mounted and all mature eggs in the ovaries counted at 40x magnification.

Before each female was dissected for the fecundity study, head width, right hind tibia length, and right forewing were measured with an ocular micrometer. These metrics were used as covariates to ensure that any observed differences were attributable to treatments rather than possible size differences in the parasitoids across treatments.

Sugar Analyses
To analyze the sugar contents of the wasps and treatments, the wasps were allowed to feed on the treatments (buckwheat, Indian blanket, and 5% honey solution) for 24 h after emergence, whereas nectar from the flower treatments was obtained by rupturing the nectary gland and soaking up the nectar with a small section (4 cm2) of Kimwipe® tissue. Water and 5% honey solution samples were also obtained using Kimwipe® tissues. The fed wasps and Kimwipe® tissue containing the treatments were held in microcentrifuge tubes and placed in a freezer at -80?C until sample preparation.

Individual wasps (4 for each treatment) were dissected by cutting them just behind the prothorax so that only the abdomen with propodeum remained. The abdomen with propodeum was placed in 100µl of HPLC-grade water and ground with a plastic pestle. All fluid was removed with a pipette and placed in a vial.

The Kimwipe® tissues used to obtain the nectar from the flower, honey solution and the water as control were separately placed in 200µl of HPLC-grade water and left for 15 min. to dissolve the sugars. Subsequently, 100µl of the extracted fluid was removed with a pipette and placed into a vial.

Extracts from insects and food treatments were transferred to 2-mL glass autosampler vials and taken to dryness under a stream of N2 gas. After addition of 40 µL of anhydrous pyridine and 200 µL of N,O-bis-(trimethylsilyl)trifluoroacetamide (BSTFA) with 1 % trimethylsilyl chloride (TMCS), vials were sealed with screw-caps fitted with Teflon faced septa, and heated at 70oC for 2 h. After cooling to room temperature, 300 µL of n-hexane and 2 µL of a 0.5 ug uL-1 solution of phenanthrene-d10 (P-d10) in hexane were added to each vial. The phenanthrene-d10 was used as an internal standard. GC-MS analyses were performed on a ThermoQuest-Finnigan DSQII system (ThermoFisher Scientific, San Jose, CA). The GC column was a 30 m DB5MS® (Agilent, Santa Clara, CA, USA) with inner diameter, 0.25 mm, and film thickness, 0.25 µm. Helium carrier gas flow was fixed at 1.5 mL min-1. Injections were in the splitless mode at 220oC with pressure surged to 250 kPa for 1 min after injection. Column over temperature at injection, 60oC, was held for 1 minute and then increased to 250oC at 10oC min-1 and held for 10 minutes. Data acquisitions were in the selected ion monitoring mode. Ions monitored were m/z = 147, 204, 217, 437 (fructose); 147, 91, 204, 217, (glucose); 217, 361, 437 (sucrose); 191, 204, 217, 361 (maltose), and 188 (phenanthrene-d10). Ions in bold italics were used for quantitation. Confirmation criteria included retention time within ±0.05 min, detection of all target ions, and the relative response ratio between the quantitation ion and the next most abundant ion within ± 20% of analytical standards prepared in the same way as samples (Becker et al., 2013). All chemicals and standards were obtained from Sigma-Aldrich (St. Louis, MO, USA).

Statistical Analyses
Fecundity and longevity of female A. rufotestaceus fed on buckwheat, Indian blanket, 5% honey solution, and water (control) were analyzed with generalized linear models (one-way ANOVA) (SAS, 2010). Fecundity and longevity data were square root-transformed to normalize the distribution and eliminate significance of replication.
Fecundity data were regressed against head width, right hind tibia, and right forewing, and were analyzed for significant differences between treatments using generalized linear models (one-way ANOVA).
Sugars were not normally distributed and thus were analyzed using Kruskal–Wallis non parametric one-way analysis of variance (SAS, 2010).

Research results and discussion:
Parasitism and Farmscaping in Broccoli

Results
Lepidopteran species (larvae and pupae) obtained in our samples included Plutella xylostella, Pieris rapae, Trichoplusia ni (Hübner), Spodoptera exigua (Hübner), and Spodoptera eridania (Cramer). Table 1 presents all observed lepidopterans (larvae and pupae), while Tables 2 and 3 focus on P. rapae and P. xylostella, respectively, because they were the most abundant hosts obtained. In Tifton, P. rapae constituted approximately 3, 12, and 84% of all the lepidopteran hosts obtained in 2010, 2011 and 2012, respectively, while P. xylostella constituted 93, 80, and 5% in 2010, 2011, and 2012, respectively (Fig. 1). In Athens, P. rapae constituted approximately 31 and 53% of all lepidopteran hosts in 2010 and 2011, respectively, while P. xylostella constituted 62 and 33% in 2010 and 2011, respectively (Fig. 1). Therefore, the remainder of this work will focus on P. xylostella and P. rapae.

The parasitoids obtained from sampled hosts included the ichneumonid Diadegma insulare (Cresson) and the braconid Microplitis plutellae (Muesbeck), both of which are parasitoids of P. xylostella; and the pteromalid Pteromalus puparum (L.) and flies of the family Tachinidae, which were collected only from P. rapae. Additionally, some specimens of the family Chalcididae were reared from both lepidopteran species, with apparently different parasitoid species attacking the two lepidopterans. In Athens, D. insulare, M. plutellae, and Chalcididae contributed 71, 22, and 7% respectively to parasitism of P. xylostella in 2010, while in 2011 they contributed 60, 20 and 20% respectively (Fig 2). In Tifton, D. insulare and M. plutellae accounted for 70 and 30%, respectively, of parasitism of P. xylostella in 2010, and in 2011, D. insulare, M. plutellae, and Chalcididae accounted for 92, 5, and 3%, respectively, of parasitism of P. xylostella. In contrast, in 2012, D. insulare accounted for 100% of P. xylostella parasitism (Fig. 2).

No parasitoids were recovered from P. rapae in Athens in 2010, while in 2011, Tachinidae, Chalcididae, and P. puparum accounted for 68, 13, and 19% of the parasitism (Fig. 3). In Tifton, P. puparum and Tachinidae were respectively responsible for 100% of P. rapae parasitism in 2010 and 2011, while in 2012, Tachinidae, Chalcididae, and P. puparum accounted for 77, 12, and 10%, respectively, of P. rapae parasitism (Fig. 3). There were no significant differences among the treatments in any of the variables for total lepidopteran hosts (pupae and larvae) (Table A1), except in % parasitism of pupae in Athens in 2011 (F = 4.46, df = 3, 9, P = 0.04), in which the dill/fennel combination treatment yielded the highest % parasitism of pupae (52.10 ± 16.50 %, n = 8) and differed significantly from the other treatments: sunflower and yarrow combination (17.50 ± 8.10%, n = 14), buckwheat (15.60 ± 11.40%, n = 9), and broccoli (2.50 ± 2.50%, n = 10) (Table 1).

For Pieris rapae (Table A2) there were significant differences among treatments in larvae/plant in Tifton 2011 (F = 4.15, df = 3, 9, P = 0.04) and % parasitism of pupae in Athens 2011 (F = 4.43, df = 3, 9, P = 0.04). Broccoli treatment yielded significantly more larvae/plant (0.40 ± 0.10, n = 12) than buckwheat (0.30 ± 0.00, n = 12), dill/fennel (0.20 ± 0.00, n = 10), and yarrow/sunflower combinations (0.30 ± 0.00, n = 9) in Tifton in 2011 (Table 2). The dill/fennel treatment yielded significantly higher % parasitism of pupae (52.10 ± 16.50, n = 8) than broccoli (2.50 ± 2.50, n = 10), buckwheat (15.60 ± 11.40, n = 9), and sunflower/yarrow combination (18.90 ± 8.60, n = 13) in Athens in 2011 (Table 2).

For Plutella xylostella (Table A3), there were significant differences among the treatments only in larval density in Athens in 2011 (F = 5.54, df = 3, 9, P = 0.02), with the sunflower/yarrow treatment (0.50 ± 0.10, n = 19) having significantly more larvae than broccoli (0.30 ± 0.00, n = 14), buckwheat (0.4 0 ± 0.10, n = 16), and the dill/fennel combination (0.40 ± 0.10, n = 19) (Table 3).
Broccoli head diameter differed significantly among treatments only in the fourth harvest in Tifton in 2011(F = 24.12, df = 3, 9, P = 0.01), with those in sunflower/yarrow (9.01 ± 0.43 cm) being significantly larger than those in buckwheat (8.33 ± 0.37 cm) and dill/fennel (7.44 ± 0.32 cm) treatments, but having comparable diameters to those in the broccoli (8.46 ± 0.61 cm) treatment (Table 4). Otherwise, broccoli diameter was unaffected by treatment.

Discussion
There were very few significant differences among treatments for the assessed variables. Given that the experiment was replicated five times in space and time, this finding suggests that, with this experimental design, the flowering plots had little or no impact on pest populations or on parasitoid activity.

The lepidopteran pest composition in the studied system varied considerably across years and locations, highlighting the considerable challenges of spatiotemporal variability in the pest complexes that are presented to pest managers, and which must be considered in developing effective farmscaping systems. For example, in 2010 in Athens and Tifton, and in 2011 in Tifton, P. xylostella was the most abundant lepidopteran pest, while in Athens in 2011 and Tifton in 2012, P. rapae was the most abundant lepidopteran pest (Fig. 1).This variation also underscores the need for spatiotemporal diversity in assessing farmscaping systems to ensure the technology is appropriate for local conditions.

Diadegma insulare was the dominant parasitoid of P. xylostella in both locations and across the years of the study (Fig. 2), and inflicted high levels of parasitism. Correspondingly, D. insulare is reported to be the main parasitoid of P. xylostella in the United States (Mitchell et al., 1997; Xu et al., 2001; Shelton et al., 2002). Access to floral resources has been shown to enhance longevity and fecundity of D. insulare (Idris and Grafius, 1995; Johanowicz and Mitchell, 2000; Lee and Heimpel, 2008) Microplitis plutellae was also relatively important at both locations, but occurred later in the season than D. insulare. Flower treatments exerted minimal effect on parasitoid efficacy.

Parasitism of P. rapae in Athens in 2011 and in Tifton in 2011 and 2012 was dominated by Tachinidae; however, P. puparum was also important in Athens in 2011 and in Tifton in 2012 (Fig. 3), when P. rapae was the most numerous herbivore in the system. Wold-Burkness et al. (2005) reported P. puparum as the most dominant parasitoid of P. rapae in cabbage, in Minnesota, but the generalist tachinid Compsilura concinnata (Meigen) was also important. Floral resources did not improve parasitism rates of by tachinids and P. puparum (Lee and Heimpel, 2005)

Significant differences occurred sporadically among treatments: for percent parasitism of all lepidopteran pupae in Athens in 2011; P. rapae larval density in Tifton in 2011; percent parasitism of P. rapae pupae (Table 2) in Athens 2011; P. xylostella larval density (Table 3) in Athens 2011; and broccoli diameter (Table 4) in the fourth harvest in Tifton 2011. Despite the occasional significant differences, the lack of consistency or pattern in the differences strongly suggests that these differences were random and did not reflect effects of the experimental treatments. Lee and Heimpel (2005) reported no significant differences in larval and pupal densities of P. rapae and P. xylostella in cabbage with buckwheat as a food source for natural enemies. They also found that overall parasitism of P. rapae by Tachinids and P. puparum, and of P. xylostella by D. insulare was not significantly influenced by buckwheat, similar to the present observations.

Various factors could have contributed to the general lack of significant differences in the variables and inconsistencies in those variables that showed significant differences, with respect to the treatments: the size of the experimental plots, the size of the floral subplots, and the choice of flower treatments.

The size of the plots (12X12 m) and distance between the plots (3 m) was possibly too small to yield differences given the likely mobility of the pests and natural enemies studied. Unfortunately, larger suitable parcels of land for the project were not available. Given the scale, it seems likely that the pests and at least some of the parasitoids were able to move readily among plots, thereby neutralizing any significant effects that the various treatments might have had. Tachinids are known to have long flight ranges, up to 200 m (Pfannenstiel et al., 2012; Romina et al., 2011). Many of the systems where access to non-host resources has significantly enhanced parasitism of pests have involved much larger plots. For example, Tylianakis et al. (2004) used a minimum of 4,800 m2, Zumoffen et al. (2012) used a minimum of 21,000 m2, Thomson and Hoffmann (2013) used 20,000 m2, and (Balmer et al., 2013) used 8085 m2, all of which awere considerably larger than ours (144 m2). Therefore, there might exist a plot size limit below which floral farmscaping might be ineffective in enhancing biological control for particular systems.

Another important factor may have been the size of the floral plot relative to the total area planted. In the present study the treatment plot occupied 2.8% of the total area and could have been too small to be effective. Pfiffner and Wyss (2004) recommended that at least 10% of intensively cultivated area be set aside as wildflower strips for natural enemy conservation, although the basis for this number was not explained. In the lettuce-alyssum system in California, ~4% of cultivated land is devoted to natural enemy conservation, and this system has been successful in managing the currant-lettuce aphid, Nasonovia ribisnigri (Mosley), with attracted hoverflies (Gillespie et al., 2011). The work by Gillespie et al. (2011) on the lettuce-alyssum system suggests that the area set aside for natural enemies can be reduced to 2% without any effect on hoverfly distribution in the crop section. However, Gillespie et al. (2011), studied a highly mobile predator on aphids, whereas less-mobile parasitoids might behave differently and might require a larger conservation area. Also the lettuce-alyssum system as studied by Gillespie et al. (2011) involved alternate rows of lettuce and sweet alyssum, maximizing edge:area ratio of treatment plots as opposed to having the flowers as border plants. This might be important in increasing the efficacy of flower plants as food sources for parasitoids. Increasing the size of the flower plots risks consuming area that could be devoted to the crop and can add difficulties to management practices (such as tillage), and reduces the acceptability of such an integrated flower system for growers.

The flowers used might have been inappropriate for the system and, therefore, not as effective at attracting or retaining relevant parasitoids of the pests in the broccoli system. Flowers are known to vary in their attraction to parasitoids (Zhu et al., 2013; Sivinski et al., 2011) . Bees, adult hoverflies, Lygus lineolaris (Palisot de Beauvois) (Hemiptera: Miridae), and Hippodamia convergens (Guérin-Méneville) (Coleoptera: Coccinellidae) were observed feeding on the floral resources and, therefore, it is also possible that these species attracted by the flowers could have competitively diminished the resources that were available to the parasitoids, supporting fewer parasitoids. Possible competition for floral resources among flower visitors has been documented in other systems (Hogg et al., 2011; Campbell et al., 2012; Ambrosino et al., 2006).

Considering these factors, we cannot conclude with certainty that these flower treatments were not effective in the system. But for future studies on this system, we would recommend using a larger area for the flower and the treatment plot as well as a greater distance between blocks to reduce insect spillover from adjacent plots. If increasing the plot size with this system becomes successful in enhancing biological control of the lepidopteran pests, as recorded in the studies with larger plot sizes mentioned above, that will suggest that there is a size threshold above which floral farmscaping might be effective at enhancing biological control, and, therefore, floral farmscaping may not be effective for managing lepidopteran pests in smallholder organic broccoli production. Further detailed analyses of the effects of the flowers used on the life histories of key parasitoids and pests in the system would be valuable. Understanding the key parasitoids in the system for respective pest species and the benefits of particular flower species for those parasitoids creates opportunities for developing appropriate flower blends to match the physical, nutritional, and phenological demands of the natural enemies.

Predation and Farmscaping in Broccoli and Cucumbers

Results
There were no significant differences among the treatments in number of chewing predators/plant for either cucumber or broccoli in any of the years and locations. Number of sucking predators/plant differed significantly among treatments (F = 6.53, df = 3, 9, P = 0.01) in cucumber in one year and site (Athens in 2011), with the cucumber control treatment (1.06 ± 0.14) having the highest sucking predator density and differing significantly from the dill/fennel treatment (0.56 ± 0.11), while having comparable sucking predator density to the buckwheat (0.99 ± 0.13) and yarrow/sunflower (0.84 ± 0.12) treatments. Total numbers of predators/plant differed significantly among treatments (F = 4.07, df = 3, 9, P = 0.04) only in cucumber in Athens 2010, with the cucumber control treatment having the highest predator density (0.44 ± 0.07) and differing significantly from the yarrow/sunflower (0.29 ± 0.09) and dill/fennel (0.24 ± 0.05) treatments, but not from buckwheat (0.40 ± 0.09). Buckwheat total predator density also differed significantly from both yarrow/sunflower and dill/fennel in Athens cucumbers in 2010, while yarrow/sunflower and dill/fennel did not differ significantly from one another (Table 5). Total predator density did not differ among treatments in cucumber for Athens 2011 (F = 3.30, df = 3, 9, P = 0.07) for Tifton 2011 (F = 0.59, df = 3, 9, P = 0.64), or for broccoli in Athens 2010 (F = 1.70, df = 3, 9, P = 0.24), Athens 2011 (F = 0.50, df = 3, 9, P = 0.69), Tifton 2010 (F = 1.41, df = 3, 9, P = 0.30), Tifton 2011 (F = 0.40, df = 3, 9, P = 0.76), and Tifton 2012 (F = 0.41, df = 3, 9, P = 0.75).

Predators obtained in the on-plant samples included the sucking predators Lygus lineolaris (Palisot de Beauvois) (Hemiptera: Miridae; phytozoophagous (Agustí and Cohen, 2000)), Orius insidiosus (Say) (Hemiptera: Anthocoridae), Geocoris spp. (Hemiptera: Geocoridae), and Nabis sp. (Hemiptera: Nabidae). Chewing predators consisted of Solenopsis invicta (Buren) (Hymenoptera: Formicidae), Coleomegilla maculata (DeGeer) (Coleoptera: Coccinellidae), Harmonia axyridis (Pallas) (Coleoptera: Coccinellidae), Coccinella septempunctata (L.)(Coleoptera: Coccinellidae), and Hippodamia convergens (Guérin-Méneville) (Coleoptera: Coccinellidae). For individual predator densities (Table 6), there were significant differences among treatments only in H. convergens densities in broccoli in Athens in 2010 (F = 1.96, df = 3, 9, P = 0.02) and in cucumber in Athens in 2011 (F = 4.98, df = 3, 9, P = 0.03), and Geocoris spp. density in Athens in 2011 (F = 3.14, df = 3, 9, P = 0.05). The broccoli treatment in broccoli in Athens in 2010 had the highest H. convergens density (0.31±0.07 per plant) and differed significantly from the buckwheat (0.0 8± 0.03) treatment but had comparable density to yarrow/sunflower (0.21 ± 0.06) and dill/fennel (0.19 ± 0.06) treatments. Dill/fennel treatment for cucumber in Athens in 2011 had the highest H. convergens density (0.15 ± 0.04), and differed significantly from yarrow/sunflower (0.05 ± 0.02) and cucumber (0.00 ± 0.00) treatments, but had comparable density to the buckwheat (0.09 ± 0.03) treatment. The control treatment had the highest Geocoris spp. density in the cucumber trial in Athens in 2011 (1.01 ± 0.14), differing significantly from the dill/fennel treatment (0.55 ± 0.11), but exhibiting similar density as in the buckwheat (0.85 ± 0.13) and yarrow/sunflower (0.76 ± 0.11) treatments.

The most abundant predators were H. convergens for broccoli in Tifton in 2011 (67.9% of all predators) and in 2012 (63.4%) and for cucumber in Athens in 2010 (30.7%); O. insidiosus for broccoli in Athens in 2010 (52.1%); L. lineolaris for broccoli in Athens in 2011 (39.8%); Geocoris spp. for cucumber in Athens 2011 (62.6%); and S. invicta for cucumber in Tifton (41.2%). Hippodamia convergens, despite not being the most abundant predator, had a high relative abundance in broccoli in Athens in 2010 and 2011 (Table 7).

Table 8 presents percent predation of beet armyworm eggs/plant by 72 h, separated by whether the eggs were chewed or sucked for both broccoli and cucumber studies in both locations, from 2010 to 2012. There were no significant differences among the treatments in any of the variables in either broccoli or cucumber trials for both locations, except for percent sucked beet armyworm eggs/plant by 72 h in Tifton in 2011, which had borderline significant differences among treatments (F = 1.73, df = 3, 9, P = 0.05). Dill/fennel treatment had the highest percentage of sucked beet armyworm eggs/plant by 72 h (34.70 ± 5.80) and significantly differed from buckwheat (20.40±3.90) and cucumber (18.40 ± 4.00) treatments, but not from sunflower/yarrow (24.50 ± 4.50) treatment.

Comparison of percentage predation of beet armyworm eggs/plant in broccoli and cucumber studies shows that predation was consistently higher in cucumber than in broccoli in each year and each location. More of the predated eggs were chewed than sucked in cucumber in each year of the study, while the results in broccoli were mixed (Table 8).

Most of the egg masses in both broccoli (approximately 98%) and cucumber (approximately 100%) studies were discovered (indicated by consumption of at least one egg in the mass) by predators by 72 h.

Percentage predation of beet armyworm eggs/plant increased with time in both crops. The increase was typically more rapid and achieved higher predation levels across the observation period in cucumber than in broccoli, with most of the predation in cucumber occurring within 48 h (Figs. 4 and 5).

Predators captured in the photographs in association with the beet armyworm eggs included the sucking predators/omnivores L. lineolaris [broccoli, Athens 2010 (n = 6) and 2011 (n = 3) and broccoli, Tifton 2010 (n = 2), 2011 (n=2) and 2012 (n = 6)] , O. insidiosus [broccoli, Athens 2011 (n = 1), Tifton 2011 (n=1)] , and Geocoris uliginosus (Say) [broccoli, Athens 2011 (n = 1)]. The chewing predators included S. invicta [broccoli, Tifton 2010 (n = 3)], C. maculata [broccoli, Athens 2010 (n = 1) and 2011 (n = 1)], and H. convergens [broccoli, Athens 2011 (n = 1) and broccoli, Tifton 2010 (n = 1), 2011 (n = 1), and 2012 (n = 1)].

Broccoli head diameter differed significantly among treatments only in the fourth harvest in Tifton in 2011(F = 24.12, df = 3, 9, P = 0.01), with those in sunflower/yarrow (9.01 ± 0.43 cm) being significantly larger than those in buckwheat (8.33 ± 0.37 cm) and dill/fennel (7.44±0.32 cm) treatments, but having comparable diameters to those in the broccoli (8.46 ± 0.61 cm) treatment (Table A1). Cucumber head diameter did not differ among treatments in any of the years, locations, and harvests (Table 9).

Discussion
Results from our experiments show very few significant differences among treatments in any of the variables 7evaluated in either the broccoli or cucumber studies for both locations. The exceptions were sucking predator density in cucumber in Athens 2011, total predator density in Athens 2010 (Table 5), Geocoris spp. density in cucumber in Athens in 2011, H. convergens density in broccoli in Athens in 2010 and in cucumber in Athens in 2011 (Table 6), and broccoli diameter at fourth harvest in Tifton in 2011. Considering that the experiment was replicated five times with broccoli and three times with cucumbers in space and time, this suggests that the flower treatments did not impact predator density or predation of beet armyworm eggs with this experimental design. The sporadic significant differences recorded in some of the variables were likely random and did not reflect effects of the experimental treatments.

The consistently higher predation of beet armyworm eggs in cucumber than broccoli, as can be seen in Figs. 4 and 5, may be attributable to higher predator density in cucumber than broccoli in Athens 2011. Predator species’ relative abundance might have also played a role as S. invicta, Geocoris sp., and H. axyridis, C. septempunctata were more abundant in cucumber than broccoli. However, differential predator density cannot explain the higher predation of beet armyworm eggs in cucumber than broccoli in Athens 2010 as predator densities for both crops were comparable. Nor do the predator numbers adequately explain the very low predation rates in Tifton 2012 broccoli relative to 2010 and 2011, since the predator numbers were highest in 2012.

Predation was expected to be higher in cucumber than broccoli for several reasons. First, cucumber was planted in summer when the temperature was higher and insects more active than in spring when broccoli was planted. Further, predator (and herbivore) species abundance is higher later in the season, which may contribute to higher predation rates. Finally, cucumber plant and growth architecture generates a very different micro-habitat than that presented by the more vertical and open structure of broccoli. Cucumbers expand and cover the ground, providing shade and a presumably more favorable micro-environment for insects relative to broccoli.

The plastic mulch may have significantly affected the microhabitat for foraging predators, and changed their activity against the sentinel eggs. Plastic mulch has been shown to affect insect communities in the field (Tuovinen et al., 2006; Žanic et al., 2009). The change from tillage and hand pulling of weeds in 2010 to plastic mulch and tillage for weed management in 2011 also may have affected the outcomes of the study. Predation rates in both broccoli and cucumbers declined from 2010 to 2011, and failed to rebound in 2012 in Tifton although predator numbers on plants were comparable with those observed in 2010.

Besides microhabitat variability, the predation results may also have been affected by the diverse feeding habits of the predators present. All of the predators observed are polyphagous, and many are omnivorous. The abundance and species composition of available herbivores in the broccoli crop varied across locations and years (Table A2), and these variations in herbivorous prey may have resulted in variation in predation across locations and years. Predators are known to switch prey based on preference (preferred vs. alternative prey) and, therefore, their responses to S. exigua eggs could have been influenced by temporal and spatial variations in the abundance of other prey species around them (Welch et al., 2012). Further, possibilities for intraguild predation may have varied within locations and across years, adding another element of variation into the predation outcomes with sentinel prey. In addition, the relative demands of carnivory and herbivory in the omnivores may also have added significant variation to the observed predation rates on the sentinel egg masses (Welch et al., 2012).

The lack of differences in predation observed here suggests no benefit obtained from the flower plantings. However, plot sizes may have significantly influenced the lack of treatment effect. Plot size and distance between plots (3 m) was possibly too small to generate treatment effects. Unfortunately, larger suitable parcels of land for the project were not available. Given the scale, it seems likely that many of the predators were able to move readily among plots, possibly negating any significant effects that the various treatments might have had. Predators, such as Geocoris sp., H. convergens, Lygus sp., and Nabis spp., are highly mobile and have been recorded to disperse more than 1 km (Sivakoff et al., 2012). If plot size was a constraint, this would suggest that there is a lower threshold in production plot size below which flower treatments may not be beneficial for predator action. Such a finding would be significant for smallholders and local producers who grow on limited land.

Another important factor may have been the size of the treatment plot relative to the total area planted. In the present study the flowering treatment area occupied ~3% of the total area and could have been too small to be effective. Pfiffner and Wyss (2004) recommended that at least 10% of intensively cultivated area be set aside as wildflower strips for natural enemy conservation, however, the basis for the number was not explained. In the lettuce-alyssum system in California, ~4% of cultivated land is devoted to natural enemy conservation, and this system has been successful in managing the currant-lettuce aphid, Nasonovia ribisnigri (Mosley) with attracted hoverflies (Gillespie et al., 2011). The work by Gillespie et al. (2011) on the lettuce-alyssum system suggests that the area set aside for natural enemies can be reduced to 2% without any effect on hoverfly abundance in the crop section. There are likely system-specific variations in the extent of area to be devoted to insectary plants relative to target crop. However, significantly increasing the size of the treatment plot relative to the crop area brings costs in terms of lost yield, and land and management expenses (e.g., labor, fuel, and time to manage weeds in the floral planting) that would need to be compensated for by pest suppression from natural enemies or other valued services to justify producer adoption. For example, increasing the area set aside for insectary plant in the lettuce-alyssum system from 2% to 8% reduced the yield of lettuce by ~7% (Brennan, 2013). Future studies with significantly larger plot sizes would be of value. But if increasing the plot size results in enhanced predation by predators as a result of access to floral resources, this may suggest that there is a plot size limit above which access to flower can improve pest management in broccoli and cucumber systems, but below which smallholders may expect little benefit.

Effects of Nectar on Parasitoid Fitness

Results
Longevity
At least one female A. rufotestaceus was observed feeding on each treatment used in this study. There was a significant food treatment effect on the longevity of female A. rufotestaceus (F = 11.10, df = 3, P < 0.001). The wasps that fed on Indian blanket, buckwheat, and 5% honey solution survived significantly longer than those that fed on water alone. Indian blanket yielded the numerically highest longevity (11 ± 1 d), followed by 5% honey solution, buckwheat, and water (10 ± 1, 9 ± 1, and 4 ± 0 d, respectively) (Fig. 6). Therefore, wasps that fed on Indian blanket, buckwheat, and 5% honey solution lived at least twice as long as those that fed on water alone.

Fecundity
After 5 d of feeding, the number of mature eggs (egg load) in the female wasps increased significantly from 80 ± 1 eggs at emergence to 109 ± 3 eggs (water), 123 ± 5 eggs (buckwheat), 134 ± 6 eggs (5% honey solution), and 138 ± 3 eggs (Indian blanket). Indian blanket and 5% honey solution significantly (F = 3.91, df = 3, P = 0.017) increased the female wasp fecundity in comparison to those fed on water alone. Buckwheat resulted in an intermediate egg load that did not differ significantly from any of the other treatments after five days of feeding (Fig. 7).

Correlation of traits
Wing length (F = 2.52, df = 3, P = 0.076), tibia length (F = 0.74, df = 3, P = 0.537), and head width (F = 0.51, df = 3, P = 0.677) did not differ significantly among treatments. However, there was a significant correlation between egg load and wing length for parasitoids in the buckwheat (? = 0.05) and 5% honey solution (? = 0.01) treatments. Hence, wing length explained 55% and 52% of the variability in the number of mature eggs of the wasps fed on buckwheat and 5% honey solution, respectively, as well as 51% of the variability in the number of mature eggs in newly emerged female wasps (Table 10).

Sugar Analyses
There were no significant differences among the treatments with respect to sugar content of the wasps after 24 h of exposure to the treatments (Table 11, fructose ?2 = 2.60, df = 3, P = 0.458; glucose ?2 = 0.68, df = 3, P = 0.877; sucrose ?2 = 3.39, df = 3, P = 0.336; and maltose ?2 = 1.05, df = 3, P = 0.789).

The five percent honey solution had significantly higher fructose (?2 = 7.53, df = 2, P = 0.020), glucose (?2 = 7.57, df = 2, P = 0.023), and maltose levels (?2 = 9.37, df = 2, P = 0.009) than Indian blanket and buckwheat nectar, but similar levels of sucrose as buckwheat (?2 = 0.13, df = 1, P = 0.724). Buckwheat nectar had significantly higher sucrose levels (?2 = 5.40, df = 1, P = 0.020) than Indian blanket nectar. The glucose/fructose ratio did not differ significantly (?2 = 3.50, df = 2, P = 0.174) among the treatments, whereas sucrose/hexose ratio differed significantly (?2 = 8.00, df = 2, P = 0.018) among the treatments. Buckwheat nectar had a significantly higher sucrose/hexose ratio (2.10±0.36) than the 5% honey solution (0.04±0.00) and Indian blanket nectar (0.34±0.21) and buckwheat nectar had comparable sucrose/hexose ratios.

Discussion
Female A. rufotestaceus provisioned with nectar from flowering plants (buckwheat and Indian blanket) or with 5% honey solution lived significantly longer than those with access to water only. This increased longevity with buckwheat and honey is consistent with findings in other wasps, such as Lysiphlebus testaceipes (Cresson) (Hymenoptera: Aphidiidae) by Hopkinson et al. (2013), and Microplitis croceipes (Cresson) (Hymenoptera: Braconidae) by Nafziger and Fadamiro (2011). The significant increase in longevity of the wasps with access to Indian blanket is interesting because this is the first time this plant has been shown to enhance survivorship of a natural enemy, although it has been used as part of commercial flower mixes to attract natural enemies (Braman et al., 2002). Despite production of considerable amounts of nectar by buckwheat (Sim and Choi, 1999), the Indian blanket treatment yielded somewhat higher wasp longevity than buckwheat, although the difference was not statistically significant. This result implies that the sugar contents of the two flower species were qualitatively comparable for the wasps, and that the two plant species produced enough nectar to sustain the wasps equally.

Buckwheat nectar is “sucrose-dominant” (Vattala et al., 2006), while Indian blanket nectar is composed primarily of glucose . Sugar consumption may increase osmotic pressure in insects, with physiological consequences such as destabilization of water balance. This increase is more rapid with consumption of nectars dominated by monosaccharides, such as glucose and fructose, than with those dominated by disaccharides, such as sucrose (Baker and Baker, 1983; Vattala et al., 2006). However, the differing nectar sugar compositions of buckwheat and Indian blanket did not affect the wasps’ longevity in our study, in agreement with the result obtained by Chen and Fadamiro (2006), in which the longevity of Pseudacteon tricuspis Borgmeier was similarly influenced by sucrose, fructose, and glucose intake. These results are the first time sugar contents of Indian blanket nectar have been reported and they appear to be similar to those of buckwheat, except in sucrose, where buckwheat was significantly higher.

Nectar quantity and accessibility did not matter for A. rufotestaceus with the flowers tested, as the wasps had comparable life spans despite buckwheat’s considerable and easily accessible nectar (Sim and Choi, 1999; Vattala et al., 2006) compared to Indian blanket’s flowers, which produce limited nectar and with more restricted access (Mani and Saravanan, 1999). However, A. rufotestaceus is a relatively large parasitoid, and may have experienced no difficulty in accessing nectar in Indian blanket flowers with its mouthparts.

The significant increase in the number of mature eggs by A. rufotestaceus from 80 ± 1 to at least 123 ± 5 with access to non-host food for five days indicates that the wasp is synovigenic, although females emerge with a large number of mature ova. Despite access to buckwheat nectar and pollen for five days by A. rufotestaceus, their egg load did not significantly differ from those that had access to water only. This result is inconsistent with results obtained for other parasitic wasps by Witting-Bissinger et al. (2008) in which buckwheat significantly enhanced wasp fecundity relative to the water, although they evaluated realized fecundity over the lifetime of the wasps. Indian blanket, on the other hand, significantly increased A. rufotestaceus egg load in all of the wasps in comparison to the water, again highlighting its potential for use as a farmscaping plant.

Although wasp size metrics, such as wing length, tibia length, and head width, often correlate with longevity and fecundity, as was observed in the significant positive correlations between number of mature eggs and wing length in buckwheat and 5% honey treatments, the lack of significant differences in these metrics among the treatments indicate that the differences observed in longevity and fecundity are independent of parasitoid size.

Even though access to sugar significantly enhanced longevity and fecundity (Indian blanket and 5% honey solution only) of the wasps relative to water, the results did not correspond with observed differences in the sugar contents of the wasps, in which there were no significant differences in the treatments in any of the sugars (Table 11). The lack of significant differences in sugar contents of the wasps might be a result of a short feeding time (24 h) and/or the timing of the feeding assessment (very shortly after adult emergence). Wasps emerge with sugar reserves, as can be seen from the sugar contents of the wasps fed with water alone; therefore, there may not have been sufficient time post-emergence for the parasitoids to expend their pre-adult reserves, and to switch to reliance on adult foods. Therefore, we anticipate that allowing the wasps longer time to feed would yield significant differences in their body sugar.
The increase in longevity and fecundity of A. rufotestaceus when provisioned with carbohydrate-rich food sources can have important biological control implications in agroecosystems. The longer lifespan and higher number of eggs recorded with these food sources indicates that providing the wasps access to these resources can afford them longer time to access pests and more eggs with which to parasitize them, possibly resulting in greater pest suppression. Further, ready availability of carbohydrate resources may retain the parasitoids more effectively in the area of targeted pest populations.
Indian blanket may be a potentially effective farmscaping plant for continental US agricultural systems for several reasons. First, it is native to the central US and south-central Canada, and exhibits a broad geographic range for growth. Second, it exhibits prolonged flowering periods (we have observed flowering for 7-8 months in the field in southern Georgia, and 3-4 month flowering periods are common in the Great Lakes region). Third, in warmer climates, it can persist for two or more growing seasons. Fourth, its nectar quality was comparable to buckwheat for survival and fecundity of the parasitoids in the present study. However, the relatively deep corollae may present problems for smaller parasitoids to access the nectar, and additional studies of its relative effects on pest and other beneficial species are needed.

Although positive results were obtained with these plants in the enclosed system used in our study, where the wasps had no choice but to feed on what was provided to them, it is important to extend this study to the field where the effects of the plant species on the wasps can be evaluated under natural conditions before they are deployed as farmscaping plants for management of N. viridula or other pests. Under field conditions the net benefit of these plants species for pest species can also be evaluated to ensure that the plants do not enhance pest risk.

Participation Summary

Educational & Outreach Activities

Participation Summary

Education/outreach description:

PUBLICATIONS:
Aduba, OL, DM Olson, JR Ruberson, PG Hartel, TL Potter. 2013. Flowering plant effects on adults of the stink bug parasitoid Aridelus rufotestaceus (Hymenoptera: Braconidae). Biological Control 67:344-349.

Aduba, OL, JR Ruberson, PG Hartel. Submitted. Does Floral Farmscaping Enhance Parasitoid Efficacy in Smallholder Broccoli? Annals of Applied Biology (in review)

Aduba, OA. 2013. Does floral farmscaping enhance biological control? PhD Dissertation, University of Georgia.

OUTREACH:
1. Field/Farm Tour: Georgia Organics Annual Meeting (Athens, GA, 19 Feb 2010); showed field trials at UGA Horticulture Farm in Athens, Georgia

2. Field Day Tour at Walker Farms: Georgia Organics Annual Meeting (Savannah, GA, 10 March 2011); showed flower plantings at farm of collaborator Relinda Walker (Walker Farms)

3. Class field training for ENTO 4500/6500 (Biological Control; UGA Tifton Campus): 2010, 2011. 10 student participants in 2010, 12 in 2011, 12 in 2012.

PRESENTATIONS OF WORK:
Aduba, OL, JR Ruberson, PG Hartel: Does floral farmscaping differentially affect a pest (Pieris rapae) and its parasitoids in broccoli? (Annual meeting of the Entomological Society of America, Knoxville, TN, 12 November 2012)

Aduba, OL, JR Ruberson, PG Hartel: Floral farmscaping effects on predator abundance and efficacy (Annual meeting of the Georgia Entomological Society, Statesboro, GA; 7 April 2012)

Ruberson, JR, AS Sparks: Bugging out: Recognizing and managing insect pests of vegetables (Annual Conference of Georgia Organics, Savannah, Georgia; 11 March 2011)

Ruberson, JR: Sustainable management of insect pests in the Southeastern United States (Invited mini-course of 8 lectures, Part of series “Special Lectures on International Development”. Kyushu University, Fukuoka, Japan; 22-23 May 2011)

TRAINING:
– Trained one PhD student (Obinna Aduba, PhD 2013, University of Georgia) in sustainable and organic vegetable production.

– Trained in-depth four undergraduate students in organic vegetable management and farmscaping these systems

– Exposed over 100 undergraduate and graduate students to farmscaping in organic vegetable systems as part of their curriculum using the field plots and project

Project Outcomes

Project outcomes:
  • It became apparent in the on-farm plantings that Indian blanket (Gaillardia pulchella, Family Asteraceae) was able to provide nearly continuous flowering over a period of 6-7 months over most of Georgia. Further, the plant was able to perennialize in the region, and regrowth was significant in years 2 and 3. The plant attracts a number of arthropod natural enemies, was found to enhance survival of a parasitic wasp in our studies. Therefore, this plant would be a very strong candidate as a farmscaping plant to enhance natural enemies.

    Despite the considerable replication of the studies in space (two locations) and time (two to three years), and the use of two crops, we were unable to demonstrate any costs or benefits to pest management of using floral farmscaping in an organic broccoli or cucumber system. The apparent obstacle to success was the limited size of the plots (12x12m), which may have been overwhelmed by the mobile pests and natural enemies in the study. This would suggest that in a diverse system, smaller growing areas may not benefit from use of floral farmscaping for insect management. Other factors may have been the limited size of the flower area (2x2m; ~3% of the total planted area) and the choice of flowers. However, visitation by arthropods to the flowers was common, so the flower selection was likely not a concern and they were readily discovered by natural enemies.

Economic Analysis

Not applicable

Farmer Adoption

The flower plantings at the farms of Relinda Walker (Walker Farms) and Rick Reed (Deep South Growers) were a part of the overall farm system. Relinda Walker had adopted floral farmscaping on her farm, and was using a variety of plantings. She had incorporated Indian blanket into floral strips beneath her solid-set irrigation system and plants were doing well and full of insects. Borage plantings were much less successful at all locations and were not adopted. Additional adoption was not documented.

Recommendations:

Areas needing additional study

The project failed to answer the initial question affirmatively. Additional work needs to be conducted on larger plantings to determine whether a plot size threshold exists above which an effect can be detected in pests and natural enemies. This would be very helpful in devising community-wide planning of plantings to enhance natural enemy function on an appropriate landscape scale. Additional work needs to be conducted on other flower species that may be appropriate for planting in the Southeast to selectively encourage natural enemies relative to pests. The results with Indian blanket as long-season flower resources are very encouraging for establishing log-term plantings to foster natural enemies in the region. Further, expansion of the work to consider a value-added component of cut flowers would be very useful in building local markets and farm incomes.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.