2.3 Data analyses
2.3.1 Characterization of the electrical signalling potential of
Bt and non-Bt cotton plants Descriptive analyses were conducted with
boxplots aiming to characterize the quantiles, medians, maximum and
minimum values, and outliers of the variables and time for the emission
of VPs (variation potentials) after the infestations with aphids on Bt
and non-Bt cotton plants and amplitude of VPs.
Correlation analyses were conducted between the variables VP amplitude
and signal emission time after infestation plants within each cotton
cultivar. The degree of correlation between the variables in each
condition was studied using Spearman’s rank correlation coefficient (P
<0.05) using the cor.test function of the R program.
Data on the number of signals per time interval after infestation of Bt
and non-Bt cotton plants with aphids were subjected to deviance
analysis, with the purpose of studying the interaction involving cotton
cultivar, aphid / plant density and time interval. A generalized linear
model with a quasi-Poisson distribution was used. The goodness of fit of
the model was evaluated with a simulated normal envelope using the hnp
package in the R program (Moral et al., 2017).
Deviance analysis was applied to study the interaction involving cotton
cultivar, aphid/plant density and period (photophase / scotophase) in
the number of VPs. Data were divided into four sections, three of which
corresponded to the data recorded during three days of observation, and
the last section corresponded to the accumulated data recorded during
the three days of evaluation. Negative binomial generalized linear
models were used for approximately the 1st and 2nd evaluation days,
while quasi-Poisson models were adopted for data recorded on the 3rd day
and total accumulated over the three days of evaluations. We used a
half-normal plot with a simulated envelope with the hnp package (Moral
et al., 2017) to assess the goodness-of-fit of the models.
2.3.2 Dispersal pattern of A. gossypii in Bt and
non-Bt cotton plants
The parameter k in each cotton cultivar and density was compared
by confidence intervals. Confidence intervals were generated from the
values of k for each block. We used the nonparametric bootstrap
technique, with 10,000 pseudoreplications, and for the resampled
parameter in each treatment, we used the R program boot package (Angelo
et al., 2019).
The probability of aphids occurring within each region of Bt and non-Bt
cotton plants in each treatment (cultivar and aphid density) was
estimated and compared with a multinomial linear model. The analyses to
estimate the probabilities and their comparisons were conducted with
nnet (Venables et al., 2012) and emmeans (Lenth, 2020) packages from R.