The existence of an enemy-free space can play an important role in aphid host race formation processes, but little is known about the mechanisms that create an area of low predation pressure on particular host plants. In this paper we identify a mechanism generating lower predation pressure that promotes the maintenance of the different host races of the pea aphid (Acyrthosiphon pisum) complex, a well-studied model for ecological speciation. The pea aphid consists of at least 15 genetically distinct host races which are native to specific host plants of the legume family, but can all develop on the universal host plant Vicia faba. Previous work showed that hoverfly (Episyrphus balteatus) oviposition preferences contribute to the enemy-free space that helps to maintain the different pea aphid host races, and that higher amounts of honeydew are more attractive to ovipositing hoverflies. Here we demonstrated that aphid honeydew is produced in large amounts when aphid reproduction rate was highest, and so is an important oviposition cue for hoverflies under field conditions. However, on less suitable host plants, where honeydew production is reduced, pea aphids enjoy lower predation rates. A reduction in enemy pressure can mitigate the performance disadvantages of aphids colonizing a novel host, and probably plays an important role in pea aphid host race formation.

data_Vosteen et al. 2016_Journal of Animal Ecology

Vosteen et al. 2016_R Skript_Fig. 2 quantitative honeydew collection

Fig. 2 – honeydew collection: Dry weight of honeydew collected from different pea aphid clone – host plant combinations during 24 h. Plants were infested with 50 adult pea aphids. Honeydew amount was analyzed with a one-way-ANOVA. In cases of non-homogenous variances, data were log-transformed. The Tukey Honest Significant Difference method was used for post-hoc comparison.

Vosteen et al. 2016_R Skript_Fig. 3 honeydew attractivity

Fig. 3 – hoverfly oviposition experiment: To test which cues elicit hoverfly oviposition under field conditions, eggs were counted on plants that were previously aphid infested, non-infested plants treated with aphid honeydew and non-treated control plants in the presence and absence of pea aphids. Plants were left for 30 h in the field and hoverfly eggs were counted afterwards. A randomized block design was used in this experiments. Each block consisted of all six treatments. Number of hoverfly eggs was analyzed with a generalized linear mixed effects model with a poisson error distribution, using blocks as random effects (random intercept) and treatments as fixed effects. P-values for explanatory variables were obtained by sequentially deleting explanatory variables and comparison of the more complex model with the simpler model with a likelihood ratio test (Zuur et al. 2009). Factor level reductions were used to reveal differences between levels of a treatment (Crawley 2013).

Vosteen et al. 2016_R Skript_Fig. 4a relative attractivity

Fig. 4a – honeydew quality: Honeydew collected from 6 different pea aphid clone – host plant combinations was applied to Vicia faba. Plants were left for 30 h in the field and hoverfly eggs were counted afterwards. A randomized block design was used in this experiments. Each block consisted of all six treatments. Number of hoverfly eggs was analyzed with genaralized linear mixed effect model with poisson distribution.

Vosteen et al. 2016_R Skript_Fig. 4a honeydew quality.r

Vosteen et al. 2016_R Skript_Fig. 4b honeydew quantity

Fig. 4b – honeydew quantity: Six, 12 or 18 droplets of this honeydew collected from Pisum race aphids feeding on Vicia faba were applied to uninfested Vicia faba. Plants were left for 30 h in the field and hoverfly eggs were counted afterwards. A randomized block design was used in this experiments. Each block consisted of all three treatments. Number of hoverfly eggs was analyzed with a generalized linear mixed effect model.

Vosteen et al. 2016_R Skript_Fig. 5 less suitable plant 2014

Fig. 5 – less suitable plants I: Trifolium and Pisum race aphids were placed both on Trifolium pratense and Pisum sativum, so that each race was reared on its native host and a less suitable plant. Plants were left for 14 days in the field and number of hoverfly eggs, hoverfly larvae and surviving aphids was recorded several times. A randomized block design was used in this experiments. Each block consisted of all four treatments. Data was analyzed with mixed effects models using blocks as random effects (random intercept) and treatments as fixed effects. P-values for explanatory variables were obtained by sequentially deleting explanatory variables and comparison of the more complex model with the simpler model with a likelihood ratio test (Zuur et al. 2009). Factor level reductions were used to reveal differences between levels of a treatment (Crawley 2013). Number of hoverfly eggs and larvae were analyzed with generalized linear mixed models with either a Poisson error distribution, or a negative binomial error distribution (glmm with the lmer function of the lme4 package (Bates et al. 2014) whichever fitted best the data of a certain experiment. In cases were only a few hoverfly eggs or larvae were present and several plants without hoverfly eggs or larvae occurred, presence / absence data were used and analyzed with Bernoulli glmms.

Vosteen et al. 2016_R Skript_Fig. 6 less suitable plant 2014_Aug

Fig. 2 – less suitable plants II: Trifolium and Pisum race aphids were placed both on Trifolium pratense and Pisum sativum, so that each race was reared on its native host and a less suitable plant. Plants were left for 6 days in the field and number of hoverfly eggs, hoverfly larvae and surviving aphids was recorded two times. A randomized block design was used in this experiments. Each block consisted of all four treatments. Survival of aphid colonies was analyzed with a generalized linear mixed effect model with a binomial distribution. Presence of hoverflz eggs and hoverflz larvae was analyzed with a generalized linear mixed effect model with poisson error distribution.