Performance on Plant Species

Adult Delphastus catalinae of undetermined age and sex were starved before they were individually confined for 24 h in a 150-mm plastic Petri dish lined with filter paper. As a source of water, a damp cotton ball was placed in a small dish in the test arena. Dishes used for the predation study contained a circular leaf section of each plant species that was previously infested with B. argentifolii. When necessary, additional cuttings were added to present ≈50 whitefly nymphs and eggs per plant species. Leaf cuttings were glued on to a filter paper lining a dish (150 × 25 mm) and placed within the same dish. Five plant species were used: cotton (Gossypium hirsutum L. var “DP 458 B/R”®), tomato (Lycopersicon esculentum Miller cv “Burbank”) (Pennington Seed, Inc., Madison, GA), hibiscus (Hibiscus rosa-sinensis L.) (Espositos Nurseries, Tallahassee, FL), cowpea (Vigna unguiculata [L.] Walpers, cv “Mississippi Silver”) (Cross Seed Company, Charleston, SC), and collard (Brassica oleracea ssp. acephala DC, cv “Georgian”) (Cross Seed Company, Charleston, SC). Each Petri dish comprised one replicate; each treatment was replicated 10 times (treatments × replicates = 5 × 10). One adult predator was released in the center of each Petri dish and allowed to feed for 24 h. After the 24-h exposure, the predator was removed from each dish and the number of remaining eggs and nymphs were recorded. This experiment was conducted in a ThermoForma growth chamber (ThermoForma, Marietta, OH) maintained at 26°C, 60% RH and 14L: 10D photoperiod.

Host Stage Preference

Collard greens (cv “Georgian”) were grown in pots containing MetroMix 200, (Scotts-Sierra Horticulture Products Co. Marysville, OH). A maximum of 4 leaves per potted plant were covered with an organza bag (16 cm × 20 cm) to prevent whitefly infestation. All other leaves were removed. Oviposition sites on the plant were restricted by securing only one leaf between two cardboard sheets (15 cm × 15 cm) and foam (1.27 cm). Ten circular holes were previously bored in the underside cardboard with a No. 10 cork borer. The cardboard was held in place with a 9-gauge wire frame and binder clips. The plant stem was covered with aluminum foil. The plants were placed in screened cages (61 × 61 × 61 cm). Approximately 200-300 adult whiteflies were released into each cage and allowed a 24-h oviposition period. Infestations were staggered to assure adequate numbers of each lifestage: eggs, small (1^st-3^rd instars) and large nymphs (4^th-pupae) at the time of the experiment. Leaf cuttings containing ≈200 prey items of each stage were made with the cork borer and they were placed in a feeding arena (150 × 25 mm dish lined with filter paper). Each leaf cutting was glued to the filter paper to reduce wilting time. A single starved (8-h starvation period) adult D. catalinae was placed in the center of each arena and allowed 24 h to feed. Each dish was secured with a rubber band. After the 24-h exposure, the predators were removed and the numbers of remaining prey were recorded by lifestage (eggs, small and large nymphs). The experiment was replicated 10 times. These tests were conducted in a ThermoForma growth chamber under conditions described as above.

Data Analysis

Predation on different plant species was analyzed as a One-Way ANOVA on percentage of whitefly hosts eaten. Data were transformed by the arcsine transformation prior to analysis but are presented as untransformed means (Sokal & Rohlf 2003). Predation on different host stages was analyzed as a One-Way ANOVA on numbers of prey eaten. When treatment effects were significant, means were separated by Tukey's HSD (P = 0.05). All analyses were performed with Systat 10.2 (Systat Software, Inc., Point Richmond, CA).

Performance on Plant Species

The proportions of whitefly hosts consumed were significantly different among plant species (F = 25.22; df = 4, 45; P < 0.01; R² = 0.69). Predation was significantly highest on cotton, followed in rank order by collard, cowpea, tomato, and lowest on hibiscus and ranged from 98.6% to 46.1% (Fig. 1). Previous studies have shown that different plant species can have profound effects on predation rates. In Serangium parcesetosum Sicard (Coccinellidae) feeding on B. argentifolii (Legaspi et al. 1996), predation was highest on cucumbers, followed by tomato and cantaloupe, and lowest on hibiscus, which are similar to the results presented here. In the predatory stinkbug Podisus nigrispinus (Dallas) (Heteroptera: Pentatomidae) feeding on 4^th instar Spodoptera exigua (Hübner) (Lepidoptera: Noctuidae), attack rates were higher and handling times shorter on sweet pepper and eggplant, compared with tomato (De Clercq et al. 2000). Lower predator performance on tomato was attributed to the presence of glandular trichomes and allelochemicals (De Clercq et al. 2000). The architecture of host plant leaves may affect predation rates, such as that in the lady beetle Propylea quatuordecimpunctata (L.) (Coccinellidae) feeding on the Russian wheat aphid, Diuraphis noxia (Mordvilko) (Aphididae) (Messina & Hanks 1998). Lower predation rates on broad-leaved crested wheatgrass, Agropyron desertorum (Fisher ex Link) Schultes was attributed to the presence of refuges such as rolled leaves, compared with the slender-leaved Indian ricegrass, Oryzopsis hymenoides (Roemer & Schultes) Ricker (Messina et al. 1997). Moreover, predation rates may be affected by interactions between different biotic and abiotic factors. In Macrolophus pygmaeus Rambur (Hemiptera: Miridae) feeding on green peach aphids, Myzus persicae (Sulzer) (Homoptera: Aphididae), predation rates were higher in darkness than in a light environment, and differences were more marked on pepper plants rather than eggplant (Perdikis et al. 2004).

Host Stage Preference

Total numbers of each host stage were not significantly different at the start of the experiment; about 200 prey items of eggs, small and large nymphs were presented to individual predators at the start of the feeding period (F = 0.029; df = 2, 27; P = 0.972; R² = 0.002), thereby permitting analysis on numbers of prey eaten (rather than percentages). Higher numbers of eggs were consumed in the 24-h predation period, compared with small or large nymphs (F = 84.933; df = 2, 27; P < 0.01; R² = 0.863) (Fig. 2). Predators are known to display different prey-preference responses when presented with various life stages of a prey. In some studies, adult prey was preferred over the egg stage. When eggs of Colorado potato beetle Leptinotarsa decemlineata (Say) (Coleoptera: Chrysomelidae) and aphids (M. persicae) were presented to C. maculata in equal numbers, females did not prefer either prey at low densities, but preferred aphids at high densities (Hazzard & Ferro 1991). Other studies have shown preferences for the egg stage of prey. For example, the phoretic mite, Macrocheles peregrinus Krantz (Acari: Macrochelidae) feeding on the horn fly, Haematobia irritans exigua De Meijere (Diptera: Muscidae), showed a strong preference for eggs of other dipterans (Roth et al. 1988). Coleomegilla maculata lengi neonates displayed strong preferences for conspecific eggs, even in the presence of essential prey and benefited from such cannibalistic behavior (Gagne et al. 2002). When presented with a choice between younger or older eggs, C. maculata lengi larvae displayed preferences for younger eggs of Trichoplusia ni (Hübner) (Noctuidae) (Roger et al. 2001). Finally, lifestage preferences may change with predator specificity. Blackwood et al. (2001) tested 13 species of phytoseiid mites for preferences between eggs and larvae of the two-spotted spider mite, Tetranychus urticae Koch (Arachnida: Acari: Tetranychidae), and found that oligophagous, specialized spider mite predators generally displayed a preference for eggs, whereas more polyphagous, generalist predators showed no prey-stage preferences.

In this study, we found that the proportions of whitefly eggs and nymphs consumed were highest on the cotton disks, followed by collards, cowpea, and tomato, and lowest on hibiscus. The collard and cowpea had glabrous leaves while the leaves from the other plant species in this study had trichomes. Because this study was not performed with whole plants, the effects of leaf structure (e.g., trichomes, refuges) were not likely to be strong factors affecting predation rates. More likely is the possibility that D. catalinae responded differentially to volatile secondary compounds released by the plant leaf cuttings. Delphastus catalinae adults displayed a strong preference for B. argentifolii eggs followed by small, then large nymphs, despite their presence in equal numbers in the feeding arena. The data suggest that in an applied biological control context, D. catalinae adults are most likely to be effective against B. argentifolii in cotton and early in the season for other crops, or when the egg stage is most abundant.