Table 1.
Synopsis of immune gene regulation effects found by previous studies in comparison with our results.
Figure 1.
Gene expression changes upon infection.
Presented values are calculated with the 2-ddCt method. This method yields fold-change values for gene expression between defined sample groups (exposed compared to non-exposed). Error bars are standard errors calculated upon averaging dCt values within sample groups and transformed to fold change errors with error propagation. The solid line marks the value 1 and corresponds to no change between groups. Dashed lines mark the values 2 and 0.5, corresponding to doubled and halved gene expression upon treatment, respectively. Asterisks mark significance of effects as detectable in the univariate outputs of the overall MANOVA (Table S6a in File S1). Visualization of fold changes within the two collection sites can be found in Figure S1.
Figure 2.
Boxplots of gene expression-fold values for catsup (F1,77 = 4.253, P = 0.043), pelle (F1,77 = 10.54, P = 0.002), PGRP-LC (F1,77 = 5.898, P = 0.017), peroxiredoxin5 (F1,77 = 11.64, P = 0.001), relish (F1,77 = 5.381, P = 0.023) and serpin27a (F1,77 = 4.075, P = 0.047).
Neunforn results are presented in the left boxplot of each pair (in grey). All depicted genes are significantly different in expression between sites (Table S6a in File S1). Fold-expression values were calculated with dCt values (see main text) and are therefore on a scale defined by reference gene expression.
Table 2.
MANOVA results.
Table 3.
Linear discriminant analyses for the factors site, infection and colony.
Figure 3.
Discriminant analysis of gene expression by colony origin.
Shown are individuals from colonies from site Aesch (A1 to A4, represented by filled diamonds, triangles, circles and squares, respectively) and Neunforn (N1 to N4, represented by open diamonds, triangles, circles and squares). LD1 and LD2 represent the first two linear discriminant functions. The main contributions to LD1 come from the genes basket, peroxiredoxin5, jafrac and hopscotch. Within LD2, PGRP-S3, jafrac, basket and hopscotch have the highest coefficients. Despite considerable variation within colonies, clusters are already visible and become fairly distinguishable when taking into account all 7 discriminant functions (55.2% of cases classified correctly as opposed to 12.5% expected by chance, Table 3).