Evaluation of Beneficial Insect Habitat for Organic Farms

For Objective 1:
Results.
In 2003, all ten advertised seed species were present in Company 1 samples, while in 2004, three of the ten advertised species were absent (Table 1). In 2003, the non-advertised species Mexican Hat and Siberian Wallflower were both present in one of three canisters from Company 1 (Table 1). In 2004, 9 different weed species were identified in subsamples from Company 1 (Table 1).

Seed abundance by percent weight varied considerably in 2003 between each of the three canisters from Company 1 was observed. Bishop’s Flower had the largest amount of variation with values ranging from 1.6-20.5% per canister (Table 1). Yarrow and Evening Primrose values were also variable, ranging from 0.6-5.6% and 3.5-13.6% per canister, respectively. Company 1 was the only seed mixture that contained an inert organic material to “help evenly distribute seed and create a mulch”, according to the manufacturers label. The amount of the material in each canister was advertised as 50.7%, although values ranged from 29.9-54.8%. In 2004, Company 1 had much less variation in subsamples (Table 1). The greatest variation was seen in Nasturtium, with values ranging from 5.9-11.9%.

Large differences were seen in the percentage of seeds by relative numerical abundance in the seed mixture from Company 1. In 2003, four seed species (Yarrow, Evening Primrose, Black-Eyed Susan, and Strawflower) made up more than 77.0% of the total percentage by relative numerical abundance (Table 1). The greatest variation was seen in Baby Blue Eyes and Strawflower with values ranging from >0.1-1.8% and 0.4-22.4%, respectively. Candytuft, Mexican Hat and Siberian Wallflower all had large variations because each seed species was found in only one container. In 2004, much less variation in relative numerical abundance was observed between subsamples from the three containers. Four seed species (Yarrow, Bishop’s Flower, Strawflower and Black Eyed Susan) made nearly 80% of the total percentage by relative numerical abundance from Company 1 (Table 1). Nasturtiums had the largest differences between containers with values ranging from 0.1-0.3%.

Germination of seeds from Company 1 was highly variable in 2003, with six of the ten species averaging germination rates below 80% (Table 1). All Strawflower and Angelica seeds tested failed to germinate, and were found to be non-viable after being subjected to a Tetrazolium Test. Candytuft, which was present in one of three containers, had only 0.7% germination. Yarrow had the greatest range in percent germination with values from 6.0-94.0%. Nasturtium and Buckwheat also showed large ranges in germination, with values ranging from 38.3-88.0% and 42.0-64.0%, respectively. In 2004, less variation was seen between subsamples of Company 1 seeds and only two of the seven species present averaged less than 80.0% germination (Table 1). Strawflower was the poorest performer, with values of 1.0-4.0% germination.

In both years of this study, all 12 advertised seed species were present in Company 2 samples (Table 2). However, in 2003 and 2004, respectively, there were 14 and seven non-advertised seed species in samples. Company 2 offered a consistent product with regards to relative numeric abundance, with little variation noted between years (Table 2). The Clover/ Alfalfa group accounted for greater than 70% of all seeds, when measured either by weight or numerical abundance, in both years of this study. The remainder of the advertised seed species were present in relatively low numbers. In 2004, weed species made up just over 1% of the average percent relative numeric abundance. Company 2 offered a consistent product with regards to percent germination. In 2003, only the Fennel and Caraway group demonstrated less than 80.0% germination (Table 2). In 2004, all species had greater than 80.0% germination. The seed distributor for Company 2 included information about % composition, purity and germination. Results from both 2003 and 2004 tests were consistent with manufacturers information.

In 2003, Company 3 had 15 of 16 advertised species present and one non-advertised species (Table 3). The composition of the Company 3 seed mix was relatively uniform. The seed species with the largest variation by percent weight was Evening Primrose, which was found in two of three packages, with values ranging from 0-1.0%. Of the advertised species, Sweet Alyssum and Bishop’s Flower, had values ranging from 0.9-2.4% and 1.3-2.4%, respectively. Of the 16 seed species present in the mixture from Company 3 in 2003, five species made up more than 50.0% of the seeds based on the average relative numeric abundance. The most prevalent seeds were Candytuft, Siberian Wallflower and California Poppy. In 2003, Company 3 had two of 15 species with less than 80.0% germination (Table 3). Gayfeather and Blanket Flower germination ranged from 18.0-56.0% and 68.0-82.0%, respectively.

In 2004, Company 4 had all 18 advertised species and three non-advertised species present (Table 4). In 2004, the composition of advertised species in the Company 4 seed mix was very consistent (Table 4). Blanket Flower had the largest variation in percent weight between subsamples, with values ranging from 3.5-7.7%. Considerable variation was seen in non-advertised species because seeds were found in only one of three subsamples. In 2004, the five most prevalent seed species from Company 4 were New England Aster, Yarrow, Sweet Alyssum, Candytuft and California Poppy. These species combined made up greater than 70% of the seeds based on the average percent relative numeric abundance. In 2004, Company 4 had two of 17 seed species with average germination rates below 80.0% (Table 4). New England Aster and Globe Gilia had germination ranges from 42.2-62.6% and 48.0-71.0%, respectively.

For Objective 1:
Discussion.
Company 1 offered an inconsistent seed mixture regarding the presence/ absence of seeds advertised in each mixture. Only 82.0% of the species are present in any given container, and there is only a 17.0% chance that a randomly selected container will have all ten advertised species present. The documented seed lot numbers indicated all seeds were from the same lot for both years of this study, however, considerable differences were seen between the two years. While all seeds advertised were present in one of the three canisters analyzed in 2003, 30% of the advertised seeds were missing from all canisters in 2004. The inconsistent results observed in Company 1 suggest poor quality analysis performed by Clyde Robin’s.

Two of the advertised seed components in the Company 1 seed mix, Yarrow and Evening Primrose, are considered weedy species. (Uva et. al 1997, Royer and Dickinson 1999 and Haragan 1991). The seed mix package states that 00.0% of contents are weeds and that no noxious weeds are present in any canister. However, in both years of this study, the claims made by this manufacturer were proven false. As a result of the weeds, growers could face increased weed pressure by these invasive weeds, which can be very difficult to control once established.

Company 2 offered a very consistent product with regards to the presence or absence of advertised seeds in the seed mixture, with nearly all seeds advertised being present. However, Fennel, one of the species advertised by this California based company, is ranked among some of the most invasive plant species by the California Exotic Plant Council (1999). In addition, Yarrow is considered an invasive weedy pest and can be difficult to control once established (Uva et al. 1997). Thus, growers planting this beneficial insect habitat might inadvertently make their weed problems significantly worse, and leave themselves open for serious problems if California adds Fennel to its list of Federal Noxious Weeds.

Company 3 offered a fairly consistent product in 2003, with only one advertised species missing. One of the advertised plants, Siberian Wallflower is currently on the Kansas Noxious Weed Seeds List (USDA-AMS 2002) and Yarrow, which is considered an invasive weed (Uva et al. 1997), could pose future weed problems for growers. Company 4 offered a consistent product in 2004 with all advertised species present and no seeds advertised in the mixture considered invasive or noxious weeds.

Presence or absence of species not advertised as being part of the seed mixtures. Two weed species were found in seeds from Company 1 in 2003, one of which, (Siberian Wallflower), is on the Kansas Noxious Weed Seeds List (USDA-AMS 2002). In 2004 ten weed species were found, one of which (Barnyard Grass) is a noxious weed in Arkansas (USDA-AMS 2002). Thus, by planting this seed mixture growers could inadvertently increase weed pressure.

In addition to unadvertised seeds being present in the seed mixture from Company 1 in 2003, all life stages of live lesser grain borer beetles (Rhyzopertha dominica (Fabricius)) were observed in all canisters upon arrival. These insects were actively feeding on seeds within packages, and as a result, greatly affect the weight and germination of these seeds. If a grower had ordered the seed mixtures and left them unopened for a period of time before planting, it is likely that few if any seeds would germinate once planted. In 2004, all canisters from Company 1 arrived insect pest free.

In 2003, Company 2 had relatively few weeds. The seed distributor indicated that 4.4% of their total mixture were weeds and other crop seeds, although the ten weed species recorded had a combined weight of less than 1.0%. Three weed species, Orchard Grass, Tall Fescue and Chess are listed as noxious weeds in multiple states, although the seed distributor for Company 2 indicated in paperwork accompanying seeds that no noxious weeds were present. In 2004, few weed seeds were observed and no species found was listed as a noxious or invasive weed.

Company 3 offered a consistent product, with few weeds or debris. Siberian Wallflower, an advertised seed in this mixture, is on the Kansas list of noxious weeds (USDA-AMS 2002). One unadvertised species, Evening Primrose, was present but because of the invasive nature of this plant, could end up becoming a serious problem for growers (Haragan 1991).

The seed mixture from Company 4 had little debris and few weed species. Siberian Wallflower, advertised as being in the seed mixture, is listed on the Kansas list of invasive species (USDA-AMS 2002) and one of the weed species, Yarrow, can be invasive and could pose problems for growers (Uva et al. 1997).

Consistency of composition between containers of the same seed mixtures. Percent by weight. In 2003, the average percent weight of each of the seed species from Company 1 was inconsistent between canisters. The seed distributor printed a breakdown of the contents on the side of each canister, which state that 48.9% of the contents were seeds and that none of the 10 species are present in quantities greater than 5.0% of the total mixture (CRSC 2003). However, in 2003 and 2004, 50.0% and 57.0% of the species were present in greater than the advertised 5%, respectively. Company 1 also indicated that 50.7% of the total weight was “inert organic material to help evenly distribute seed and create a mulch” (CRSC 2003). Growers who ordered this product expecting to receive 170g of seeds, may be disappointed.

It is possible that some discrepancies in the average percentage by weight in 2003 were due to the presence of live lesser grain borer beetles, which seemed to feed on the larger seeds (Buckwheat, Nasturtium and Angelica). In 2004, less variation was seen between canisters and no insects were observed.

Little variation between years was observed in advertised species in Good Bug Blend. By far, the greatest number of seeds based on percent weight was the Clover/ Alfalfa. Because the majority of seeds in this Clover/ Alfalfa group were coated, a tremendous amount of weight was added. This could pose problems if growers ordered seed based on a seeding rate that did not take this into account.

Due to the competitive nature of some of the plants in this seed mixture, future management problems could arise for growers. A study by Forehand (2004) found that in experimental plots after one year of growth, Clover, Yarrow and Fennel were the only plants remaining in all plots, of the fourteen original species transplanted.

Company 3 does not offer the consumer any information other than the species advertised in the seed mixture. They offered a consistent product, with little variation in percent by weight noted between the three analyzed packets.

Company 4 did not offer any information other than the names and species of seeds present. The average percent by weight varied greatly and could pose a problem if a grower was expecting a roughly equal distribution of seed species.

Percentage by relative numerical abundance. In both years of this study, a considerable amount of variation in percent relative numerical abundance was observed in seeds obtained from company 1. In 2003, five of the 10 advertised species averaged greater than 90% of the total percentage of seeds and large variations between canisters was recorded. In 2004, four species made up nearly 80% of the total percentage of seeds by relative numeric abundance. This could be a disappointment for growers if they were expecting a roughly equal distribution of seeds.

In both years of this study, the Clover/ Alfalfa group clearly made up most of the seed mixture from Company 2 based on the percent relative numerical abundance.

Based on the three seed packets analyzed in 2003, Company 3 offered a consistent product with regards to the average percent by numerical abundance and little variation was noted between the packets. Of the 16 advertised seeds, five species (Sweet Alyssum, Baby Blue Eyes, Candtuft, California Poppy and Siberian Wallflower) made up greater then 50% of the total percentage of seeds. Because the seed distributor does not give the consumer any information about the percent composition, a grower might assume a relatively equal distribution of seeds between the 16 species. Clearly this is not the case regarding percent weight or relative numeric abundance.

The seed distributor for Company 4 also failed to provide consumers with expectations of their seed mix. Again, if a grower was expecting a roughly equal distribution of seeds, they might be disappointed to learn that five species make over 70% of the total mix by percent relative numerical abundance.

Percent Germination. Poor germination of Company 1 seeds was observed in 2003, most likely due to the presence of live seed-feeding beetles in all three canisters. In addition to direct seed damage by beetle feeding, high levels of fungal infestation were observed during germination tests with these seeds. These fungi are likely the result of the nutrient-rich insect frass left behind from foraging within each canister or physical injury to the seeds themselves by the beetles. Probably because of a lack of pest insects, germination was higher in 2004, likely attributed to. However, in seven of 10 advertised species present, two had average percent germinations below 80% germination.

In both 2003 and 2004, Company 2 demonstrated good germination, with one plant group in each year having lower than 80% germination. In 2003, the company also included a copy of the percent composition, germination and hard seed, date tested and country or state of origin of seven of the fourteen plant groupings. All but one species met or exceeded the results provided by Peaceful Valley.

Seed mixtures obtained from Companies 3 and 4 demonstrated good germination with only two species demonstrating less than 80% germination.

For Objective 2:
Study 1.
Results.
Habitat type had a significant impact on the total abundance and diversity of insects found in sample plots for each of the calculated indices (see Table 3 for statistics). Of all the potential habitats studied, Border Patrol™ generally had the highest overall diversity for the index values calculated (Table 3). Of the cut flower/herb plantings, Celosia had the highest overall diversity and abundance for Simpson’s Index, Shannon’s Index and Hill’s N1 and N2 diversity numbers (Table 3). With the exception of species evenness, fennel had significantly lower index values compared to all other plant communities studied (Table 3).

Beneficial parasitoid diversity was significantly affected by habitat type for all of the index values (Table 4). Good Bug Blend and Border Patrol™ had the highest diversity and richness index values for beneficial parasitoids, but the lowest species evenness values. In general, fennel had the lowest diversity and richness values for beneficial parasitoids but the highest species evenness.

Four of the six abundance and diversity index values for the beneficial predator feeding group were significantly influenced by habitat type (Table 4). Celosia and Good Bug Blend had the highest beneficial predator index values for the cut-flower/herb and commercial mixtures, respectively, and fennel had the lowest index values.

Herbivore crop pest diversity indices were all significantly influenced by habitat type (Table 4). For four of the six index values calculated for herbivore crop pests, Border Patrol™ and Good Bug Blend were significantly higher than all other habitat types, while fennel had the lowest index values.

None of the diversity index values were significantly affected by habitat type for the mixed parasitoid feeding group (Table 4). Only species richness was significantly affected by habitat in the non-crop parasitoid feeding group. Only Hill’s N2 and species richness index values were significantly affected by habitat for inconsequential predators. Overall, fennel had the lowest Hill’s N2 and species richness index values for inconsequential predators. Habitat significantly affected species evenness and richness for non-crop pests. The three commercial mixes had the highest non-crop pests index values, while fennel had the lowest index values overall.

The only diversity value for pollinators that was significantly affected by habitat was species richness, in which Border Patrol™ and Beneficial Insect Mix had the highest index values, and fennel the lowest (Table 4). Three of the abundance and diversity indices for the decomposer/fungal feeder group were significantly altered by habitat type (Table 4). There was no significant difference in the decomposer index values between the three commercially available seed mixes, although Border Patrol™ generally had the highest values. Fennel had the lowest overall index values of all habitat types for decomposers.

Moth feeding activity varied significantly among the various beneficial insect habitats (Table 5). The highest mean number of noctuid moth flower visits per minute was recorded in Border Patrol™, while the lowest values were in Good Bug Blend, Zinnia and Beneficial Insect Mix. Border Patrol™ had significantly higher mean hawk moth visits.

Habitat type significantly altered the mean number of carabid beetles collected in pitfall traps (Table 6). The numerical trend indicated Good Bug Blend and fennel had the highest values, while Celosia and Beneficial Insect Mix had the lowest. No significant difference was seen in the mean number of spiders collected in pitfall traps placed in the habitats.

For Objective 2:
Study 1.
Discussion.
Border Patrol™ was chosen for this study because it offered the greatest variety of flower types compared to other commercial seed mixtures. The Border Patrol™ seed mixture had high diversity and evenness of beneficial parasitoids, but also had the greatest abundance and diversity of crop feeding herbivores, mixed parasitoids, decomposers/ fungal feeders, and it also attracted the highest number of pest moths of the six habitats tested. Evening primrose, the largest plant in this mixture, has large cup-shaped flowers with long, tubular corollae that open at dusk and are accessible to adult Lepidoptera (Brickell and Zuk 1997). Because Border Patrol™ harbors comparatively high crop pest populations and comparatively high levels of pest moth feeding were observed in this habitat, planting it near crops may actually increase pest insect populations.

Good Bug Blend was chosen for this study because of the high proportion of plant species with small, easily accessible nectaries within flowers. This type of floral structure is purported to benefit small parasitoids (Colley and Luna 2000, Leius 1960, Luna and Jepson 2002, Patt et al. 1997, Wäckers 2004). In this study Good Bug Blend harbored high abundance and diversity of beneficial predators, parasitoids, and ground beetles. Since this seed mixture included plants with relatively small, shallow flowers, large pollinators and lepidopteran pests were apparently less able to feed. Along with Border Patrol™, Good Bug Blend also harbored the highest abundance and diversity of crop-feeding herbivores.

Beneficial Insect Mix was chosen for this study because the plant species present in this seed mixture represented “showy” types of flowers typically associated with cut-flower production or gardening. Lepidopteran pests were not highly attracted to this habitat. Because of the large number of plant species found in Beneficial Insect Mix, it was expected that a high diversity of insects would also be observed. High abundance and diversity values were only found for non-crop herbivores and non-crop parasitoids, and this mix ranked the lowest of the three commercial seed mixtures for numbers of beneficial parasitoids and predators. It is possible that the relatively large flowers that benefited pollinators were unable to feed microscopic (1-2 mm) Hymenoptera parasitoids. This idea is supported by the work of Patt et al. (1997), who evaluated the influence of floral architecture on two parasitic Hymenoptera.

Celosia was chosen for this study because it is commonly grown in North Carolina as a cut flower crop. Overall, these plants ranked among the highest abundance and diversity values for predators, both beneficial and those of no agronomic consequence, as well as parasitoids that demonstrate varied life histories. While Celosia was the most effective of the three cut flower/ herb plantings at attracting several different feeding groups of predators and parasitoids, the groups found were for the most part not considered useful in biological control of crop pests. The floral structure of Celosia has very tightly clustered flower heads, containing up to thousands of individual flowers (Brickell and Zuk 1997) with relatively shallow, easily accessible pollen (Moore et al. 1998). Celosia attracted intermediate numbers of noctuid moths and no hawk moths, probably a reflection of this floral structure.

Zinnia is a commonly grown cut flower in the southeastern United States (Greer 2000). The large, daisy-like flower heads are borne on solitary long stems and bloom throughout the summer months (Brickell and Zuk 1997). Zinnias, which are in the same family as sunflowers, reportedly attract various kinds of beneficial insects from many different feeding groups (DuFour 2000 and Jones and Gillett 2005). This study found these plants had some of the lowest index values of insect abundance and diversity. While well suited for a cut flower cash crop, Zinnia does not appear to be effective at attracting beneficial insect populations.

Fennel is often recommended for attracting beneficial organisms in agricultural landscapes (Dufour 2000 and Al-Doghairi and Crenshaw 1999), but recommendations for using this plant have not been based on scientific evidence. Several studies have documented feeding by parasitic Hymenoptera on fennel and other umbelliferous plants (Al-Doghairi and Crenshaw 1999, Baggen and Gurr 1998, Baggen et al. 2000, DuFour 2000, Hodgson and Lovei 1993, Maingay et al. 1991, Patt et al. 1997, Poncavage 1991). However, this study found fennel had the lowest species diversity and abundance for all indices and for all feeding groups. One explanation may be that 120-d transplants were used in this study, which did not begin flowering until late summer. Fennel had an intermediate number of noctuid moth visits, and no hawk moth visits, probably reflecting the small umbelliferous structure of the flowers. A high mean number of ground beetles were collected from fennel, possibly in response to numerous immature Lepidoptera feeding on foliage.

This study demonstrates that a wide variety of arthropods are attracted to commercially available beneficial insect habitats, not only the intended beneficials. Although beneficial insects were collected from all the plantings in this study, it is unclear whether they were feeding within or benefiting from the particular plant communities. More work is necessary to determine whether these habitat plants provide pollen, nectar, alternate hosts, or other resources to specific natural enemies that attack crop pests, and if they benefit field populations of these enemies and assist in pest management. For example, Good Bug Blend had the highest diversity of beneficial parasitoids and predators of the habitat plants tested in this study. However, Forehand (2005) found that parasitism of pest moth eggs and caterpillars was not changed when small organic tomato fields were surrounded by Good Bug Blend. This suggests that high abundance and diversity of beneficial insects in a habitat may not be a predictor of how or whether the habitat functions as a pest management tool under field conditions.

For Objective 2:
Study 2.
Results.
Experiment 1. In 2004, average numbers of crop parasitoids, crop predators, and pollinators observed were significantly affected by flower species (F = 6.60, df = 3, 5, P = 0.0344; F = 10.45, df = 9, 16, P < 0.0001; F = 12.43, df = 9, 16, P < 0.0001). Deleterious and non-crop parasitoids, and deleterious predators were not significantly affected by flower species (F = 1.57, df = 9, 16, P = 0.2064; F = 0.12, df = 5, 9, P = 0.9849; F = 2.99, df = 5, 9, P = 0.0731; F = 2.56, df = 1, 2, P = 0.2506). In 2005, flower species significantly affected the mean numbers of parasitoids, non-crop parasitoids, predators and pollinators (F = 41.79, df = 2, 4, P = 0.0021; F = 27.45, df = 4, 8, P < 0.0001; F = 9.08, df = 4, 8, P = 0.0045) but not deleterious parasitoids, or deleterious predators (F = 2.86, df = 3, 6, P = 0.1267; F = 9.74, df = 1, 2, P = 0.0891). Pollinators and deleterious parasitoids were affected by time of day observations were made (F = 12.69, df = 1, 10, P = 0.0052; F = 9.86, df = 1, 10, P = 0.0105) while crop parasitoids, non-crop parasitoids, crop predators, deleterious predators were not (F = 1.16, df = 1, 6, P = 0.3235; F = 3.19, df = 1, 10, P = 0.1042; F = 0.16, df = 1, 10, P = 0.6966; F = 0.70, df = 1, 4, P = 0.4487). The interaction between time of day and flower species significantly affected pollinators, deleterious parasitoids, and crop predators (F = 16.58, df = 4, 10, P = 0.0002; F = 7.38, df = 3, 8, P = 0.0108; F = 8.85, df = 4, 10, P = 0.0025) but not crop parasitoids, non-crop parasitoids, deleterious predators (F = 1.00, df = 2, 6, P = 0.4207; F = 0.98, df = 4, 10, P = 0.4620; F = 0.18, df = 1, 4, P = 0.6907). In 2004, significantly more crop parasitoids were found feeding from celery than other plants, while in 2005 they fed from fennel in significantly higher numbers than from other plant species (Table 2). In 2005, non-crop parasitoids fed in significantly greater numbers from buckwheat flowers. In 2004 and 2005, crop predators fed in significantly higher numbers from fennel than the remainder of the flower species. In 2005, significantly more crop predators were present on buckwheat at 9:30 than at 12:00. In 2004, higher mean numbers of pollinators were found feeding from blanket flower, but did not significantly differ from mean number of pollinators feeding from tickseed (Table 2). Mean numbers of pollinators feeding from tickseed, fennel, yarrow, daisy, black-eyed Susan, and California poppy were approximately equal while celery, red clover, (Trifolium repens L.), and buckwheat were fed upon least. In 2005, more pollinators were observed feeding from black-eyed Susan and buckwheat than all other plant species. More pollinators were observed at both black-eyed Susan and buckwheat at 9:30 than at 12:00. The effects of replication and date on mean numbers of insects feeding from flowers in 2004 were significant for crop parasitoids (F = 3.98, df = 2, 11, P = 0.0383; F = 14.48, df = 6, 7, P < 0.0001). Replication did not effect deleterious or non-crop parasitoids, deleterious predators, crop predators, pollinators (F = 0.54, df = 2, 13, P = 0.5883; F = 0.59, df = 2, 13, P = 0.5568; F = 3.33, df = 2, 10, P = 0.0779; F = 0.23, df = 2, 17, P = 0.7948; F = 2.91, df = 2, 17, P = 0.0619). Date played a significant role in the number of pollinators and non-crop parasitoids found feeding from flowers (F = 9.16, df = 6, 13, P < 0.0001; F = 3.13, df = 6, 13, P = 0.0139) but not deleterious parasitoids, deleterious predators, or crop predators (F = 1.06, df = 6, 13, P = 0.4063; F = 2.40, df = 6, 13, P = 0.1056; F = 1.86, df = 6, 13, P = 0.1029). In 2005, no effect of replication was found for any of the feeding guilds. Week significantly affected the number of pollinators and non-crop parasitoids observed on flowers (F = 34.21, df = 5, 10, P < 0.0001; F = 7.89, df = 5, 10, P = 0.0030) but did not affect mean numbers of crop predators, deleterious predators, crop parasitoids, non-crop or deleterious parasitoids (F = 0.83, df = 5, 10, P = 0.7181; F = 2.68, df = 5, 10, P = 0.0865; F = 2.03, df = 5, 10, P = 0.1599; F = 1.31, df = 5, 10, P = 0.3356). In experiment 2 flower species significantly affected abundance of mymarids and trichogrammatids but not scelionids (F = 11.81, df = 5, 10, P = 0.0006; F = 13.45, df = 5, 10, P = 0.0004; F = 1.83, df = 5, 10, P = 0.1947). Height (F = 21.47, df = 2, 44, P < 0.0001; F = 25.51, df = 2, 44, P < 0.0001; F = 8.25, df = 2, 44, P = 0.0009) and the interaction between flower species and height played a significant role in abundance of mymarids, scelionids, and trichogrammatids (F = 7.24, df = 10, 44, P < 0.0001; F = 6.69, df = 10, 44, P < 0.0001; F = 4.17, df = 10, 44, P = 0.0004). The interaction between flower species and flower removal significantly affected trichogrammatids (F = 7.16, df = 5, 12, P = 0.0026) but not mymarids or scelionids (F = 0.56, df = 5, 12, P = 0.7280; F = 1.35, df = 5, 12, P = 0.3104). Flower removal and the interaction between flower removal and height significantly affected abundance of scelionids (F = 6.76, df = 1, 12, P = 0.0232; F = 6.20, df = 2, 44, P = 0.0042). Flower removal and the interaction between flower removal and height did not significantly affect abundance of mymarids (F = 1.62, df = 1, 12, P = 0.2266; F = 2.26, df = 2, 44; P = 0.1167) or trichogrammatids (F = 0.18, df = 1, 12, P = 0.6818; F = 0.41, df = 2, 44, P = 0.6672). There was a significant three way interaction between flower species, flower removal, and height for scelionids and trichogrammatids (F = 2.64, df = 10, 44, P = 0.0130; F = 2.28, df = 10, 44, P = 0.0298), but not for mymarids (F = 1.69, df = 10, 44, P = 0.1123). Among the different heights, a significant flower effect was found for mymarids, scelionids, and trichogrammatids at height 2 (flower height ) (F = 5.08, df = 5, 10, P = 0.0141; F = 4.70, df = 5, 10, P = 0.0182; F = 5.78, df = 5, 10, P = 0.0092) and height 1 (0.5 times flower height) (F =12.55, df = 5, 10, P = 0.0005; F = 3.24, df = 5,10, P = 0.0536; F = 22.38, df = 5, 10, P < 0.0001). At the height 3 (1.5 times flower height), there was a significant flower effect on abundance of trichogrammatids (F = 5.58, df = 5, 10, P = 0.0103) but not on abundance of mymarids (F = 2.56, df = 5, 10, P = 0.0965) or scelionids (F = 1.04, df = 5, 10, P = 0.4479). For Objective 2:
Study 2.
Discussion.
Experiment 1 was designed to determine which flowering plants provide floral food resources to beneficial insects. Two feeding guilds, crop parasitoids and crop predators, were considered beneficial insects because of their ability to affect agricultural pests. Other insects feeding from floral structures were recorded as farmers may assume they are beneficial because they look (to the untrained eye) similar to beneficial insects. The latter have species that may be deleterious because of their potential to reduce numbers of pollinators or spiders through predation or parasitization (e.g. Pompilidae and Chrysididae) (Triplehorn and Johnson 2005). We also recorded numbers of pollinators and non-crop parasitoids which may benefit the farm through pollination of crops or reduction of turf grass pests but play no role in crop pest management.

Results from this study show that insects belonging to different feeding guilds preferentially feed from different flower species. Direct comparison of the two study years is impossible as planting design and plant species observed differed between years. However, in both years, the same feeding guilds were affected by flower species with the exception of non-crop parasitoids. Overall trends in the frequency of feeding visits made by beneficial insects can be seen in both years. Fennel received the greatest number of feeding visits from crop predators both years and in 2005 fennel was frequented most often by crop parasitoids. In 2004, celery was visited most often by crop parasitoids. This study reinforces the observation that umbelliferous flowers which have easily accessible nectaries are often frequented by beneficial insects (Patt et al. 1997). Celery however only bloomed for three weeks in only two of the three replicates. In addition, celery is a biennial and blooms after its second year of planting. If celery blooms for only a few weeks after its second year of planting it would unlikely be a desirable component of a beneficial insectary habitat.

In 2004, few insects were found feeding from buckwheat when all observations were conducted at noon. Buckwheat tends to wilt in hot weather and does not produce nectar in the afternoon (Lee and Heimpel 2003; Olson et al. 2005). By adding a morning observation we were able to see that buckwheat was attractive to pollinators and crop predators.

This study provides preliminary recommendations about beneficial insect habitat for North Carolina growers. Siberian wallflower (Erysimum hieracifolium L.) and dame’s rocket (Hesperis matronalis L.) were eliminated from data analysis because they received so few feeding visits and therefore are likely poor choices as habitat plantings to attract natural enemies in North Carolina. Other plants, such as celery and cilantro (Coriandrum sativum L.) exhibited a short blooming period, making them unsuitable insectary habitat plants as well. In 2004, flowering was inconsistent across replications and dates causing many gaps in the data. Fennel showed promising characteristics both in terms of a long bloom period and in its ability to attract beneficial insects. Fennel, however, causes contact and photodermatitis in humans (Simon et al. 1984), can be invasive, and is listed on the California Exotic Plant Pest List (1999). Additional studies are needed on the biology and phenology of plants used in this study to determine how (if) they are useful for insectary habitat. The current findings can be a starting point for future observational studies of beneficial insect flower-feeding in North Carolina.

In experiment 2 abundance of microhymenoptera caught on sticky traps was used as an indirect indicator of relative attractiveness of each plant species to the three parasitoid families studied. The assumption was made that if flowers were attractive to microhymenoptera, a greater number would be caught at height 2 (the height of flower heads) in the treatments where flowers had not been removed. Crabgrass was chosen as the control for this study because it offered a vegetative habitat without flowers. It was assumed that if flowers were attractive, more microhymenoptera would be caught in plots containing flowering habitat than in non-flowering controls.

Each microhymenopteran family responded differently to the plants in this study (Table 3). Mymarids were found in greatest abundance at height 1 in black-eyed Susan plots. Scelionids were most abundant in cock’s comb plots at height 2. The greatest number trichogrammatids were trapped in crabgrass control plots both at height 1 and height 3. None of the flowers assumed to attract microhymenoptera belong to the family Apiaceae or Polygonaceae. These findings are significant because both fennel and buckwheat have been listed as suitable beneficial insect habitat (Maingay et al. 1991; Stephens et al. 1998; Irvin. et al. 2000; English-Loeb et al. 2003). Similar to the present findings, past work on the Small Farm Unit found abundance and diversity of natural enemies sampled from various cut flower and herb species to be lowest in plots containing pure stands of fennel and highest in cock’s comb (Forehand 2004).

Little evidence was found in this study that flower removal affected the number of wasps caught on traps. For the majority of the plant species tested, numbers of trapped microhymenoptera were the same in subplots where flowers were present compared to subplots where flowers had been removed. Only scelionids were found in greater abundance at flower height in cock’s comb plots where flowers remained intact (Table 3). This finding was similar to that of Rebek et al. (2005) who found that the removal of inflorescences from four species of flowering plants in an ornamental landscape had no effect on abundance of natural enemies collected on sticky cards. Our results differ from results of Irvin et al. (2000) who found greater abundance of the leafroller parasitoid Dolichogenidea tasmanica in buckwheat plantings with flowers present than in plantings where flowers had been removed indicating an attraction to floral structures.

Overall, the abundance of sampled microhymenoptera in this study was not different in flower plots compared to control plots. Scelionids and mymarids were found in greater numbers in a few plots containing flowering plants than in the control plots. Of these plots, mymarids were solely found in higher numbers at height 1 in black-eyed Susan and scelionids in greater abundance at height 2 in cock’s comb (Table 3). These findings suggest the flowers themselves were not attractive to mymarids. English-Loeb et al. (2003) found parasitism by mymarids to increase in the presence of buckwheat flowers. However, mymarids were caged on buckwheat putting them in close proximity to flowers. Scelionids showed preferential attraction to cock’s comb plantings at flower height indicating a possible attraction to floral structures. At height 2, trichogrammatids were most abundant in yarrow plots where flowers had been removed. Trichogrammatids were most abundant at height 1 in un-mowed control plots but were also highly abundant in mowed crabgrass control plots and buckwheat plots where flowers had been removed. This shows that while trichogrammatids appeared to be attracted to some habitats, flowers were clearly not responsible for this attraction.

Future field studies could be conducted to investigate which vegetative qualities of plants, rather than flowers, determine relative attraction to microhymenoptera. If vegetative habitat is attractive to different microhymenoptera, it would be useful to determine which habitats are preferred. In the current study, mean numbers of trichogrammatids were significantly greater within the canopy (height 1) of un-mowed crabgrass plots than in the canopy of any other plant species studied (Table 3). Using paper models of plant foliage Lukianchuk and Smith (1997) determined that female T. minutum Riley had a greater foraging success on simple rather than complex surfaces. It may be that the vegetative qualities of grass in this study exhibited a less complex structure than the foliage of the flowering plants. Trichome-density on plant surfaces could have played a role in preference of some plants over others. Keller (1987) determined that walking speed of T. exiguum was influenced by leaf-trichome form and density, with less-densely pubescent leaves permitting the fastest walking speeds. Measures of trichome-density and type are generally used to evaluate host-finding ability of parasitoids but could be important if trichomes impede location of food sources. Quantification of foliar trichomes could also be valuable since trichomes can provide shelter to microhymenoptera (Cortesero et al. 2000). In the present study, mymarids were found in greatest abundance in black-eyed Susan plots at height 1 regardless of flower presence or absence. Black-eyed Susan and cock’s comb in our plots were similar with regard to height, leaf size and shape, amount of foliage, and canopy closure. Black-eyed Susan foliage was densely covered with trichomes while cock’s comb foliage was glabrous.

In a study by Thorpe (1985) vegetation type (soybeans vs. weedy margins) was not determined to be an important factor in parasitism rates by Trichogramma minutum Riley or T. pretiosum Riley. However, height was determined to be important, with higher levels of egg parasitism by T. minutum at greater heights and higher parasitism by T. pretiosum at lower heights. Microhymenoptera in the current study were trapped at different heights relative to the height of flowers. Because different plant species bloomed at different heights, conclusions about flight-level preferences of different microhymenoptera could not be drawn. Future research could be conducted with traps placed at constant heights relative to ground-level in plots containing flowering and non-flowering plants. This would allow one to analyze flight behavior of different microhymenoptera in varied habitats relative to a constant height. In the present study, cock’s comb and black-eyed Susan bloomed at approximately equal heights. Fennel flowers were well-above and yarrow flowers well below cock’s comb and black-eyed Susan inflorescences. The effects of these height differences could have played a role in the results obtained in the present study if microhymenoptera were present in fennel plots at the approximate height of cock’s comb flowers but traps were not placed there to monitor activity.

While this study did not directly quantify attraction of microhymenoptera to habitat, some insight to habitat preference was obtained using relative measures of abundance. Microhymenoptera were not found in greater abundance in plantings containing members of the families Apiaceae or Polygonaceae. Higher numbers of only one microhymenopteran family were found at flower height in only one plant species, cock’s comb, a member of the pigweed family, Amaranthaceae. Abundance of trapped microhymenoptera varied with plant species at different trap heights for each hymenopteran family. This indicates that while habitat appears to play an important role in abundance of microhymenoptera for the most part floral food resources do not appear to be the causative agent

For Objective 3:
Results.
Egg Fate. In both years of this study the density of moth eggs was not significantly different (F = 1.12; df = 1,74; P = 0.2925) between treatment and control plots (Tables 2, 3). There was significantly lower (F = 8.57; df = 1,4; P = 0.0429) survival of moth eggs in 2004 than in 2003, primarily as a result of increased parasitism (F = 85.21; df = 1,4; P = 0.0008). Identifiable predation was a minor component of egg mortality, but unidentified predation may be part of the approximately 35-52% of the eggs that met an unknown fate.

Neither parasitism nor predation of moth eggs in tomatoes was significantly altered by the presence of beneficial insect habitat around plots (parasitism: F = 0.26, df = 1, 5, P = 0.6325: chewing predation: F = 0.62, df = 1, 5, P = 0.4665: sucking predation: F = 0.52, df = 1,5, P = 0.5036). The percentage of eggs that met an unknown fate also was not significantly affected by habitat (F = 1.55; df = 1,5; P = 0.2685).

Larval Parasitism. Larval populations of Manduca spp. were significantly higher (F = 17.77; df = 1,4; P = 0.0135) in 2004 than in 2003 (Fig. 1). This provided an opportunity to test habitat effects in larval population densities that differed by approximately six-fold. However, densities of larvae in treatment and control plots were not significantly different (F = 0.05; df = 1,5; P = 0.8292). Presence of beneficial insect habitat did not significantly (F = 2.15; df = 1,5; P = 0.2021) influence parasitism levels in either year of the study.

Percent Habitat Cover. Only six of the original fourteen species in the beneficial insect habitat were present at the conclusion of the 2004 field season. Yarrow, fennel and clover were the only species present in every plot with average percent cover estimates of 14.8%, 10.4% and 6.4%, respectively (Forehand 2005). Celery was found in two plots (0.7%), while only one alfalfa and one dill plant occurred in a single subsample of a single plot (both 0.1%).

For Objective 3:
Discussion.
For beneficial insect habitat to be useful in pest management there must be a net gain in beneficial insects and a net reduction in insect pests (Landis et al. 2000). Research conducted at the Center for Environmental Farming Systems in 2003 compared insects collected from six different flowering plant communities, and found Peaceful Valley’s “Good Bug Blend” to harbor the highest populations of beneficial parasitoids and predators (Forehand 2005). Of the three commercial habitat blends examined by Forehand (2005) this blend should have been the best to test for pest management in a cropping system. However, after two years of evaluating its effectiveness in reducing pest moth eggs and larval numbers, there was no significant pest reduction when the habitat was compared with millet. The beneficial insect habitat used in this study was relatively diverse and should have been resource rich when compared with the monoculture of millet, which had few apparent resources for beneficial insects. Insect pests are generally not a problem in North Carolina millet. The millet in this study was kept mowed to approximately 15cm tall, and prevented from seeding in order to simulate a grassy field border, which is the most common border on organic farms in this area.

In a separate field study in fall 2003, ‘Good Bug Blend’ was planted according to supplier’s instructions and intense weed competition resulted in effectively no habitat plants in plots by July 2004 (unpublished data, M. Kroner, North Carolina State University, Raleigh, N.C). This current study occurred under what could be considered ideal conditions for the habitat to ‘work’. Organically produced plants of each species were started in a greenhouse, then transplanted to the field in the proportion that each occurred in the original seed mixture. Habitat areas were hand weeded the first year of the study, dead transplants replaced, and habitat was irrigated both years as needed. However, weeding was not done in habitat borders in 2004 and by the end of the field season the only habitat plants present in all plots were yarrow, fennel and clover. Despite the differences in plant composition of habitat borders, results of this study were the same in both years.

Peaceful Valley recommends planting 1-5% of farmland for “good results” (Anonymous 2004). The habitat borders in this study comprised approximately 36% of the total area of treatment plots and should have been more than sufficient to see a treatment effect.

Several factors were considered when determining plot size for this experiment. The effect of scale on experimental results has been a point of contention in entomology and ecology with no prevailing conclusions coming from the mixed results the literature (eg. Bommarco and Banks 2003 and Bengtsson et al. 2005). It seems the scale of the experiment should reflect the scale of the questions being asked. Large scale questions should be tested with large scale experiments. Likewise, field level questions should also be conducted at an appropriate scale. Since habitat manipulation at the level advocated by the suppliers of beneficial insect habitat is clearly at the field scale, plots were sized accordingly in this study. Plot size fairly reflected the scale of tomato production by North Carolina organic growers. Most growers use three or more staggered plantings, each of which are approximately the size of plots in this study.

One consideration in assessing beneficial insect habitat is whether it acts as a “source” or a “sink” for natural enemies (Carmona and Landis 1999). In theory, beneficial insects are intended to utilize resources within a habitat area as needed, but then move into a nearby crop to manage pest populations. However, if the habitat were too attractive and resource-rich, there would be no need for beneficial insects to enter the crop. In this study it is unlikely the habitat was acting as either a source or sink. There was no treatment effect, and observed levels of parasitism in tomato were very similar to what Schmidt (1998) found in tomatoes on four organic farms in central North Carolina.

Predators clearly play an important role in agroecosystems (Barbosa and Wratten 1998, Helenius 1998, DuFour 2000). While identifiable predation appeared to be of little consequence in regulating pest moth egg populations in this tomato cropping system, it is possible that some of the roughly 35-52% of eggs that were unrecovered may have resulted from unidentifiable predators. However, it is also likely these eggs were removed in part by heavy rains or leaf contact.

The egg parasitoids (Families: Trichogrammatidae, Scelionidae) and larval parasitoids (Family: Braconidae) we evaluated in this study represent a wide range of both sizes and life histories. Based on their size, it is likely that their floral foraging behaviors differ (Patt et al. 1997). The habitat in this study was diverse enough to provide floral structures to accommodate both. Another possible carbohydrate source besides nectar is from homopteran honeydew (Idoine and Ferro 1988, Baggen and Gurr 1998). Because aphid (Homoptera: Aphididae) colonies were present on tomatoes within all plots, it is possible that parasitoids and parasitism levels benefited regardless of the resources present in the beneficial insect habitat.

Because neither egg nor larval parasitoids appeared to be assisted by the habitat surrounding tomato plots, the results of this study can be generalized to a large segment of the agriculturally important parasitoids. Clearly, more work needs to be done to evaluate the effectiveness of beneficial insect habitats. It is also clear that the mere presence, or even increased abundance, of beneficial insects in specific plant communities does not mean that those plants will aid pest management in nearby crops. Commercial suppliers of beneficial insect habitat should bear this in mind when making claims about the value of their products.

The authors have worked very closely with organic growers for many years and most, if not all, feel very strongly that a strip of diverse flowering plants surrounding their production area or specific crops encourages beneficial insects and reduces pest insect pest problems. The intent of this study was to determine if this view is correct. Bengston et al. (2005) observed that the relatively large non-cropped areas (field margins, field borders, hedgerows, etc.) are important refuges for many organisms. We agree with this observation and conclude that the surrounding, highly plant-diverse, unmanaged habitat, surrounding organic farms is more important to pest and beneficial insect ecology than planting a strip of flowering plants. Planting flowering strips alone, without the support of other nearby refuges, is unlikely to provide the economic return needed to offset the cost of establishment and maintenance.