Computational protocol: Heterodera schachtii Nematodes Interfere with Aphid-Plant Relations on Brassica oleracea

Similar protocols

Protocol publication

[…] The experiment consisted of 84 pots, based on four herbivore treatments (no herbivores, nematodes only, aphids only, nematodes and aphids), seven replicates and a soil microbial treatment with three levels. The replicates were distributed randomly over 7 blocks (fully randomized block design) to account for spatial effects in the greenhouse. The three different soil microbial inoculum treatments where different dilutions of soil suspensions were added have been described previously (Hol et al. ). For all parameters presented here, we tested for 3-way interactions between nematodes, aphids, and microbes. The absence of significant 3-way interactions showed that there were no effects of microbial soil treatment on aphid-nematode interactions. Hence, the data shown here were pooled for all three soil treatments, resulting in N = 21 per herbivore treatment. For the analysis of aphid and nematode effects on plant parameters, a linear mixed effects model (lme) was used with the soil microbial inoculum treatment nested within block as random factor and the presence of nematode and aphids as categorical factors. For all variables (except %water) presented in Table  that were significantly affected by either nematodes or aphids, we verified that there was no significant interaction with soil microbial treatment when this was included as fixed factor. Water percentages showed a significant interaction between nematodes and soil microbial treatment, but ignoring this interaction still gave overall significant main effects of nematodes on moisture. The aphid × nematode interactions were not significant for any parameters in Table , and hence were removed from the model. Most variables were log-transformed to obtain normality of errors, which was assessed by visual inspection of q-q plots. When transformation did not result in sufficiently normal error distributions, non-parametric tests were performed to verify the outcome of the linear mixed effects model. In all cases, results were congruent with the outcomes from non-parametric tests, and thus only results from the linear mixed effects models (lme) are shown. One phloem sample was lost during processing. Phloem data were analyzed with the same linear mixed effects model (lme) as above, but including harvesting time as covariate. For parameters significantly affected by aphids or nematodes the correlation between numbers per plant and response value was tested with Spearman rank correlation tests.Aphid numbers per plant over time were used to fit an exponential growth curve and estimate population doubling times. ANCOVAs were used to test whether relations between plant parameters (%nitrogen in shoot, %water, and the individual sugars, amino acids, and glucosinolate concentrations) and aphid doubling times interacted with nematode presence. Plants with less than 10 aphid individuals (3 plants with nematodes, 6 plants without nematodes) were omitted from this analysis. To adjust P-values to correct for type 1 errors due to multiple testing, sharpened Benjamini and Hochberg () false discovery rate control was performed (Verhoeven et al. ) for the lmes and for the ANCOVAs. Significance levels were determined based on the distribution of the P-values, and estimates of the true alternative cases and may, therefore, vary between datasets. In Table , individual P-values are shown, but only those that were significant after correcting for multiple tests are printed in bold (P < 0.01). The multivariate technique partial least squares regression (PLSR) was used to investigate the relation between aphid doubling times and all primary and secondary metabolites of the plant simultaneously. Data were log-transformed, scaled to unit variance, and were mean-centered before analyses. The analysis was done for all aphid doubling times and separately for those on plants with nematodes and those on plants without nematodes (3 PLSR in total). The optimal number of latent structures was determined on the basis of “leave one out” validation. Significance of the model was determined by comparing the explained variation in aphid doubling time with the explained variation of 1,000 models using permutations of the aphid doubling time data. A model explaining more variation than 95 % of 1,000 permutations was considered significant. All analyses were done in R 3.0.0 (R Development Core Team ) using the ‘nlme’ and ‘pls’ packages (Mevik and Wehrens ). […]

Pipeline specifications

Software tools lme4, nlme
Application Mathematical modeling
Organisms Brassica oleracea, Caenorhabditis elegans
Diseases Nematode Infections, Tick Infestations