Supplementary on-line material
Photo S1. Glyphosate-selected progeny 4a and unselected VLR1 sprayed with glyphosate at 350 g ha-1. Plants were grown in Controlled
Environment Room (CER).
200
(a)
Seeds plant (n)
10
(b)
150
-1
-1
Tillers plant (n)
15
5
0
100
50
0
0
200
400
Glyphosate (g ha-1)
600
0
200
400
600
-1
Glyphosate (g ha )
Fig. S1. Reproductive output in surviving plants from progeny 4a (white squares) compared to unselected VLR1 (solid circles) at several
glyphosate doses. (a): Number of tillers per plant. (b): Number of seeds per plant. Symbols are mean ± 1 SE n = 3. Two-way ANOVA
indicates significant interaction between the factors (population*glyphosate dose) (P < 0.05).
Biomass (% control)
(a)
100
(b)
80
60
40
20
0
0
200
400
Glyphosate (g ha-1)
600
0
200
400
600
Glyphosate (g ha-1)
Fig S2 Dose-response curves (biomass) of glyphosate-selected progenies in CER. Solid circles and solid line represent the unselected original
population (VLR1), open circles and dashed line the glyphosate-selected progeny 3a, open squares and dotted line the selected the progeny 4a.
Symbols are mean ± 1 SE n = 3, lines are predicted values for biomass as percentage of untreated control.
200
Biomass (% control)
(a)
(b)
(c)
150
100
50
0
0
100
200
300
400
Glyphosate (g ha-1)
500
0
100
200
300
400
Glyphosate (g ha-1)
500
0
100
200
300
400
500
Glyphosate (g ha-1)
Fig. S3 Glyphosate dose-response curves (biomass) of glyphosate-selected progenies in field conditions. Solid circles and solid line represent
the unselected original population (VLR1), open circles and dashed lines the first glyphosate-selected progenies (1b, 3b), open triangles and
dash-dotted lines the second glyphosate-selected progenies (1c, 2c, 3c), open squares and dotted lines the third glyphosate-selected progenies
(1d, 2d, 3d). Symbols are mean ± 1 SE, n = 3, lines are predicted values for biomass as percentage of untreated control.
Table S1. Parameters of the log–logistic model [Y = c+(d-c)/(1+(x/G)^b)] used to calculate the LD50 values of selected and unselected progeny
and their R:S ratios from dose–response bioassays for different developmental plants stages. ANOVA analysis conducted for each non-linear
regression is highly significant (P < 0.001).
†† Residual
†††
Progeny
d
c
G
b
RMS††
Adj-R2†††
LD50
(g ha-1)
R:S
1 tiller stage
VLR1
4a
100
100
1
7
109
177
6
4
5
49
0.99
0.97
109
177
-1.62
2-tiller stage
VLR1
4a
100
100
7
72
216
263
7
7
8
70
0.99
0.54
221
>350
->1.58
Mean Square.
Adjusted-R2: Approximate coefficient of determination for non-linear models with a defined intercept calculated as Adjusted-R2 -(sums of
squares of the regression/corrected total sums of squares).
Qu-gene modelling simulations
Set of parameters: complete additive model, heritability = 0.5; 4 generations of selection (displayed on the x axis of each graph), initial gene
frequency = 0.1 (same for all genes), 10 simultaneous runs (10 lines displayed in the graphs). The population size prior selection was 200 plants
(which approximately corresponded to the number of plants sprayed at each herbicide dose in our selection experiments).
The graphs display the components of variance, the heritability (narrow sense on individual basis), gene frequency of the favourable alleles and
the accumulated population mean.
By analysis of the population mean the hypothesis of one minor additive gene reflects well the response to the selection observed in the
progenies 1c, 2c, 3c and 1d, 2d ,3d where no significant shift towards glyphosate resistance was observed between the second and third cycle of
selection (Figure S4b, S4c). Genes have the chance to be fixed in the population and therefore absence of genetic variation cannot result in any
further shift towards resistance. Also figure S4f (two additive genes and proportion of selection 0.1) seems to well describe what was observed
in the progenies d selected at 350 g glyphosate ha-1. Assuming an initial frequency of 0.1 no simulation can really be matched or well describe
the response to selection we obtained at 150 g glyphosate ha-1 with the progenies “b”. However, assuming an initial gene frequency of 0.5
together with the hypothesis of 1 minor additive gene the modelled response to selection is similar to what was observed (Figure S4g). There is
chance of getting that minor gene fixed. Probably the negligible response to selection observed in progeny 2b and 3b was due to an insufficient
selection intensity applied at such low glyphosate rate (150 g ha-1) and subsequent low heritability.
Figure S4a. One gene and selection proportion 0.5 (roughly 50% survival in the unselected population at 150 g glyphosate ha-1, see Table 1).
Figure S4b. One gene and selection proportion 0.2 (roughly 20% survival in the unselected population at 250 g glyphosate ha-1, see Table 1).
Figure S4c. One gene and selection proportion 0.1 (roughly 10% survival in the unselected population at 350 g glyphosate ha-1, see Table 1).
Figure S4d. Two genes and selection proportion 0.5
Figure S4e. Two genes and selection proportion 0.2
Figure S4f. Two genes and selection proportion 0.1
In the following simulation a different initial gene frequency was hypothesized based on the plant survival obtained in the unselected population
(e.g. approximately 50% plant survival at 150 g glyphosate ha-1 suggested an initial gene frequency of 0.5
Figure S4g. One gene, initial gene fr. 0.5, selection proportion 0.5