Ecological Archives E093-189-A1

Jacob T. Gable, David W. Crowder, Tobin D. Northfield, Shawn A. Steffan, and William E. Snyder. 2012. Niche engineering reveals complementary resource use. Ecology 93:1994–2000. http://dx.doi.org/10.1890/12-0188.1

Appendix A. Methods related to the field and greenhouse experiments, eight figures showing data from the field and greenhouse experiments, and six tables showing results of the statistical analyses.

Caterpillar impacts on predator-aphid relationships in the open field

We surveyed arthropod communities (aphids, caterpillars, and predators) in open-field Brassica oleracea plantings at the Washington State University research station in Othello, Washington, USA. We raised B. oleracea plants for four weeks in a greenhouse with ambient lighting, and then field-planted seedlings by hand on 20 May 2009. Each plot included 100 B. oleracea plants organized in 10 rows of 10 plants with approximately 8 m of bare ground between plots (plots were 25 m²). Half of the plots were randomly assigned to receive a treatment to reduce caterpillar densities, while the other half were designated as controls. On treated plots, we applied a Bacillus thuringiensis (Bt) insecticide (kurstaki strain-trade name Dipel DF) at the recommended rate of 0.5 pounds/acre every other week. This chemical is applied in a water-based solution, and is highly toxic to Plutella xylostella caterpillars but not aphids or their predators (Talekar and Shelton 1993; Appendix Figs. A1, A2). Thus, these treatments were designed to reduce caterpillar presence and feeding in half the plots. Bt sprays were timed to not coincide with weeks when plots were sampled. Control plots were sprayed with water only, to mimic the moisture added to treatment plots without the Bt toxin.

Arthropods in these plots were censused on 7 July and 23 July 2009, on 10 randomly selected plants in the interior rows of each plot (no plants along plot perimeters were sampled to reduce edge effects). We first conducted timed (1 min) searches of the entire plant for any aphidophagous predators. All predators observed were identified to species. Every predator species observed was found in both Bt and non-Bt plots, with no significant difference in predator abundance across treatments (Appendix Fig. A2). Next, we divided the plant into thirds, consisting of the newest leaves, oldest leaves, and middle-aged leaves, and counted the number of Brevicoryne brassicae aphids and P. xylostella caterpillars on two randomly-selected leaves per section. We then counted the number of leaves per plant to obtain an estimate of the number of aphids and caterpillars per plant.

 

Impacts of caterpillar damage on predator niche overlap

We observed predator foraging behavior in open-field plots at Washington State University’s Tukey Orchard in Pullman, WA, USA. We established four, 4-m² plots of B. oleraceae plants, each including 16 evenly-spaced plants in 4 rows of 4 plants each, by transplanting 6-week-old starts from the greenhouse. Plots were separated from one another by 5 m of bare, plowed ground. From 5 July to 8 Aug 2011, two randomly-selected plots received a twice-weekly treatment of formulated Bt insecticide, at label rates, while two control plots were sprayed twice-weekly with water. We observed no feeding damage or caterpillars on plants treated with Bt; the other two plots experienced natural colonization and feeding damage by P. xylostella.

Behavioral observations were conducted on 29 July, 5 Aug, and 10 Aug 2011. Each day, we visually scanned plants in each plot for 15 minutes each hour, and repeated the observations from 1100 until 1500, for a total of 4 observation periods in each plot each day (12 observations for each plot total). When a predator was located on a plant, we recorded the predator’s species identity, whether the predator was at the leaf perimeter vs. in the leaf center (defined as being more than 5 mm from the leaf perimeter), and whether the predator was near any leaf edge (within 5 mm of an outer edge or an interior edge created by caterpillar feeding), or not near any edge (>5 mm from either a natural or caterpillar-caused edge) (Steffan 2010). Predator foraging along leaf edges and in leaf centers were not mutually exclusive, as caterpillar feeding created “interior” edges within leaves. Two predator species used in our experiments, Hippodamia convergens and Diaeretiella rapae, were abundant enough during the observation period for statistical analyses of predator behavior.

We randomly selected one leaf from 8 randomly-chosen plants in each plot, on 10 August 2011, for leaf-structure analysis. These leaves were removed and photographed individually on a 50 × 60 cm grid of paper. Leaves then were analyzed for total leaf area and the proportion of edge vs. non-edge area using the photo analysis software Image-J (Steffan 2010).

 

Caterpillar damage and predator diversity effects

We conducted a fully-factorial manipulation of two factors: caterpillar feeding damage (present or absent) and predator species richness (0, 1, 2, or 4 predator species). Caterpillar damage was established, where appropriate, by allowing P. xylostella larvae to feed and then removing them before predators or aphids were added (see below). Predators were manipulated within a substitutive design such that predator densities were constant (4 per mesocosm) across all richness levels, and within each richness level all possible combinations of our four predator species were established in different mesocosms in both the presence and absence of caterpillar feeding damage (see below). The four predator species included the parasitoid wasps Aphidius matricariae and Diaeretiella rapae, the ladybeetle Hippodamia convergens, and the damsel bug Nabis alternatus. Experimental units consisted of cylindrical mesocosms (0.25 m diameter × 0.3 m height), with 50 cm² mesh windows cut out on two sides and the lid for ventilation, and a 1 cm² hole in the bottom for drainage. Each bucket was filled to a depth of 0.1 m with potting soil, and was planted with a single, 4-week old B. oleraceae plant with two true leaves, and watered as needed. The greenhouse was maintained at 25° C, with ambient humidity and light.

To establish P. xylostella feeding damage, where called for in the experimental design, 10 third-instar caterpillars were released onto leaves to feed for 72 h. After 72 h, caterpillars were removed, and 20 B. brassicae aphids from a greenhouse colony were released into all replicates (caterpillar damaged and not), with 10 aphids placed on each of the two leaves. We allowed 72 h for aphids to establish feeding locations, after which we conducted initial aphid counts and released predators. There were four predator treatments: (1) No predator control (4 replicates × 2 caterpillar treatments = 8); (2) Predators in monoculture (each of the 4 different predator species in separate monocultures × 3 replicates per species × 2 caterpillar treatments = 24); (3) Two predator species (each of the 6 unique combinations of two predator species drawn from the pool of 4 candidate species × 2 replicates / unique combination × 2 caterpillar treatments = 24); (4) All four predator species (8 replicates × 2 caterpillar treatments = 16). Each mesocosm that included predators housed 4 individuals, distributed evenly between the predator species. Predators were allowed to feed in the mesocosms for 72 h, after which aphids were counted. Parasitoid larvae were allowed to develop in their aphid hosts for 10 days after predator removal such that parasitized aphids could be clearly identified, after which the number of parasitized aphids was counted (parasitized aphids were distinguished based on their gold coloration). The experiment was conducted in two temporal blocks of 72 cages each, distributed among treatments as described above, for a total of 144 replicate cages across the entire experiment.

Concurrent with this manipulative experiment, we observed predator behavior on a second set of plants either damaged or undamaged by caterpillar feeding. Our goal was to make certain that the impacts of caterpillar feeding on predator behavior that we had observed in the field, also occurred in our greenhouse mesocosms. Thus, we again focused on the predators D. rapae and H. convergens, which generated the most foraging observations in the field study (described above). Single B. oleraceae plants were transplanted into 40 cage-mesocosms, with each plant trimmed to a single leaf of similar size. In 20 cages, five P. xylostella third-instar caterpillars were released to feed for 72 h. After this period, caterpillars were removed and 20 B. brassicae aphids were released onto each plant in all 40 cages. After allowing 72 h for aphids to establish feeding locations, predators were released. Twenty experimental units were randomly designated for observations of D. rapae (10 with caterpillars and 10 without) and the remaining 20 were designated for observations of H. convergens (10 with caterpillars and 10 without). Four predators were released into each cage, with behavioral observations conducted the following day. Each plant was observed for 1 min every hour from 0900 until 1700 (a total of 9 observations per experimental unit, and 90 per treatment). During each period, we first recorded whether each of the four predators was on the plant or not. For all predators foraging on the plant, we also recorded whether the predator was at a natural leaf edge (defined as being within 5 mm of the leaf perimeter), near any leaf edge (within 5 mm of a natural outer edge or an interior edge created by caterpillar feeding), or not near any edge (>5 mm from either a natural or caterpillar-caused edge) (Steffan 2010).

Alongside the two experiments described above, we quantified the effects of P. xylostella damage on B. oleraceae leaf structure. A total of 12 mesocosm-cages, treated as described above with half damaged by caterpillars and the others not, had their leaves harvested and individually photographed on a 15 × 25 cm grid. Leaves were analyzed for total leaf area and the proportion of edge vs. non-edge area using Image-J, as described previously (Steffan 2010).

LITERATURE CITED

Steffan, S. A. 2010. Biodiversity and fear ecology: the cascading effects of species richness and nontrophic interactions. Dissertation. Washington State University, Pullman, Washington, USA.

Talekar, N. S., and A. M. Shelton. 1993. Management of the diamondback moth. Annual Review of Entomology 38:275–301.


 

 

   FIG. A1. Effects of Bt insecticidal applications on P. xylostella densities (caterpillars / leaf) in open-field B. oleracea plots that were surveyed in 2009 for caterpillars, aphids, and predators. In plots sprayed with Bt, caterpillar densities were notably lower than in untreated, control plots (two sample t test: t8 = -2.12, P = 0.067).


 

 

   FIG. A2. Effects of Bt insecticidal applications on (A) predator densities (predators / leaf) and (B) aphid densities (aphids [log10] / plant) in open-field B. oleracea plots that were surveyed in 2009 for caterpillars, aphids, and predators. Predator densities were not significantly different in Bt compared with control plots (two sample t test: t8 = -1.56, P = 0.16). Similarly, aphid densities were not significantly different in Bt compared with control plots (two sample t test: t8 = -0.43, P = 0.68).


 

 

   FIG. A3. (A) In the open-field behavioral observation experiment, the percentage of leaf surface area within 5 mm of any leaf edge was significantly higher for leaves fed-upon by caterpillars than undamaged leaves (two sample t test: t30 = -7.70, P < 0.0001). (B) Total leaf surface area was not significantly affected by caterpillar feeding (two sample t test: t30 = -1.30, P = 0.20).


 

 
   FIG. A4. The change in aphid abundance on no-predator control plants in greenhouse cages with caterpillars absent or present. Aphid abundance was not affected by caterpillar feeding (ANOVA: F1,12 = 0.49, P = 0.50), and these effects were consistent across blocks (ANOVA: block effect: F1,12 = 0.15, P = 0.71; block × caterpillar interaction: F1,12 = 0.35, P = 0.56).

 

 
   FIG. A5. (A) In the greenhouse-mesocosm predator-diversity experiments, the percentage of leaf surface area within 5 mm of any leaf edge was significantly higher for leaves fed-upon by caterpillars than undamaged leaves (ANOVA: F1,44 = 40.2, P < 0.0001). These differences were greater in block one compared with block two, although the trend was the same for both blocks (ANOVA: block effect: F1,44 = 0.55, P = 0.47; block × caterpillar effect: F1,44 = 5.57, P = 0.023). (B) In contrast, total leaf surface area was not significantly affected by caterpillar feeding (ANOVA: F1,44 = 0.010, P = 0.92), although leaves were larger in block one compared with block two (ANOVA: block effect: F1,44 = 53.0, P < 0.0001; block × caterpillar effect: F1,44 = 0.66, P = 0.42).

 

 
   FIG. A6. The change in aphid abundance (means ± SE) on plants in greenhouse cages with or without predators. Predator presence decreased aphid abundance, on leaves with or without caterpillars (see Table A5).

 

 
   FIG. A7. Observations of foraging behavior for two predator species, Diaeretiella rapae and Hippodamia convergens, in greenhouse cages (see Table A6 results of logistic regression models). Plants were either fed-upon by caterpillars or not prior to behavioral observations. Hippodamia convergens always foraged along leaf edges more frequently than D. rapae (n = 151, χ2 = 21.8, P < 0.0001). This trend was not affected by caterpillar presence (caterpillar effect: n = 151, χ2 = 0.00018, P = 0.99; species × caterpillar interaction: n = 151, χ2 = 0.00024, P = 0.99). Diaeretiella rapae always foraged more frequently in leaf centers than H. convergens (n = 151, χ2 = 10.7, P = 0.0011), and both species foraged more frequently in leaf centers when caterpillars were present (n = 151, χ2 = 17.1, P < 0.0001). Spatial complementarity was significantly reduced when caterpillars were present, as predators were more commonly found in the same location (i.e., the leaf center) when caterpillars were present compared to when they were absent (species × caterpillar interaction: n = 151, χ2 = 5.36, P = 0.021).

 

 
   FIG. A8. (A) In the greenhouse behavior experiment, the percentage of leaf surface area within 5 mm of any leaf edge was significantly higher for leaves fed-upon by caterpillars than undamaged leaves (two-sample t test: t22 = -11.2, P < 0.0001). (B) Total leaf surface area was not significantly affected by caterpillar feeding (two-sample t test: t22 = -0.050, P = 0.96).

 

TABLE A1. Results of generalized linear model examining effects of insecticide treatment (Bt vs. non-Bt, Treatment) and predator species richness on aphid densities in open-field B. oleracea plots. Bt sprays reduced caterpillar densities (Fig. A1), but did not directly harm aphids or their predators (Fig. A2).
 

 

TABLE A2. Results of logistic regression models examining effects of caterpillar density (Treatment: reduced with Bt or control) on the behavior of two predators, H. convergens and D. rapae, in open-field plots. The first model compared the proportion of predators foraging on leaf perimeters vs. leaf centers across treatments; the second model compared the proportion of predators foraging along any leaf edge vs. not along an edge across treatments.

 


TABLE A3. Results of ANOVA model examining effects of caterpillar feeding damage and predator species richness on change in aphid abundance in the greenhouse mesocosms. For simplicity, we show sum of squares rather than the effect sizes for each level of predator richness and associated interactions. Caterpillar feeding damage did not indirectly harm aphids (Fig. A2), but increasing predator species richness on plants with prior caterpillar feeding strengthened aphid suppression, while predator diversity did not significantly impact aphid suppression on plants without prior caterpillar feeding (Fig. 2B). Although aphid abundance differed across blocks, the interaction between caterpillar presence and predator richness was consistent across the two experimental blocks.

 

 

TABLE A4. Results of ANOVA model and LSD contrasts examining effects of individual predator species compared with a diverse, 4-species predator community, on change in aphid abundance in the greenhouse experiment. Separate models were run for mesocosms that had caterpillar feeding damage versus those were caterpillars were not present. Values followed by a different letter were significantly different (P < 0.05).

 

 

TABLE A5. Results of ANOVA model examining effects of caterpillar presence and predator presence on change in aphid abundance in the greenhouse experiment. Across both blocks of the experiment, predators significantly reduced aphid abundance with or without prior caterpillar feeding. These effects were consistent across the two experimental blocks.

 

 

TABLE A6. Results of logistic regression models examining effects of caterpillar density (present or absent) on the behavior of two predators, H. convergens and D. rapae, in greenhouse mesocosms. The first model compared the proportion of predators foraging on leaf perimeters vs. leaf centers across treatments; the second model compared the proportion of predators foraging along any leaf edge vs. not along an edge across treatments.

 

[Back to E093-189]