3 RESULTS
3.1 Characterization of the electrical signalling potential of Bt and non-Bt cotton plants In descriptive analysis, it was possible to visualize that Bt cotton plants when infested with A. gossypiiemitted the first VPs (minimum value in boxplot) between time intervals of 0.31 h (60 aphids/plant) and 0.64 h (30 aphids/plant) (Figure 1). In the absence of aphids, only two Bt cotton plants emitted these electrophysiological signals. Non-Bt cotton plants emitted the first VPs after 0.80 h when kept at 30 aphids/plant and after 1.60 h at the density of 60 aphids/plant, as well as in the control (non-Bt and no aphids) (Fig. 1 ).
The Bt and non-Bt cotton plants infested with aphids emitted signals after 60 h of aphid infestation (Fig. 1 ), while in cotton plants used as a control, the maximum signal emission values were observed at 55 and 57 h in non-Bt and Bt cotton plants, respectively (Fig. 1 ). The boxplot with VP amplitude shows that the response variability (maximum values, not including outliers, and 3rd quantile) (Fig. 2 ) of non-Bt cotton plants at 30 aphids/plant density was lower than that of the other cotton plants infested with A. gossypii . The maximum VP found in the control cotton plants was near 28 mV. In general, the mean VP (points within the boxplots) was near all treatments, ranging from 11.22 mV (non-Bt cotton plants infested with 30 aphids/plant) to 17 mV (Bt – control cotton plants). Outliers (points out of boxplots) occurred for 30 aphid/plant (129 mV)-infested non-Bt cotton plants and 60 aphid/plant (116.60 mV)-infested Bt cotton plants (Fig. 2 ).
Spearman rank analysis revealed that there was no correlation between the amplitude (mV) of VP and time (h) to emission of signals by cotton plants after aphid infestation at all densities studied within each cultivar (Bt and not Bt), except at the densities of 30 aphids / non-Bt cotton plants (ρ = -0.2659; P = 0.0060) and 60 aphids/Bt cotton plants (ρ = - 0.3528; P = 0.00254).
Analyzing the amount of VPs emitted by the cotton plants, we observed that infestation-free plants emitted few signals, with an average accumulation of 0.75 (control – Bt cotton) and 2.50 signals (control – non-Bt cotton) over 72 h. Only two Bt cotton plants emitted electrical signals in the absence of aphid stress (Table 1 ).
In the accumulated emission of VPs over 72 h, it was verified that Bt cotton plants exposed to 60 aphids/plant density emitted fewer signals compared to the other conditions (P <0.05) under aphid stress. However, by assessing the emission within the intervals, the deviance analysis revealed that the signal emission pattern in each cultivar was influenced by the time interval and aphid density, as there was a significant interaction between these three factors (P = 0.0488) (Table 1 ).
The highest number of signals emitted by Bt cotton plants when exposed to aphids occurred in the time interval after infestation of 0-12 h (30 aphids/plant) and 0-12, 36-48 and 60-72 h (60 aphids/plant) (Table 2, Fig. 3 ).
When we compared the signal emission pattern between combined treatments involving aphid densities and cotton cultivars within each time interval, it was possible to verify a delay in terms of the production pattern of signalling on Bt cotton plants under stress with 60 insects/plant because until the time interval of 36 h after infestation, there was a lower signal emission by Bt cotton plants when exposed to 60 aphids/plant in relation to the other conditions of aphid density/Bt or non-Bt cotton cultivar (Table 2, Fig. 3 ).
In the time interval of 36‒48 h, the emission of signals by Bt cotton plants was lower only in relation to Bt cotton with 30 aphids/plant. Additionally, in the time interval of 60‒72 h, the signal production by cotton plants was higher when the Bt and non-Bt cotton plants were exposed to densities of 60 and 30 aphids, respectively, in relation to other conditions (Table 2, Fig. 3 ).
The deviance analysis on the interaction of the factors aphid density versus cotton cultivar versus light period within each studied day (1st, 2nd or 3rd day) and accumulated over these three days influencing the number of VPs emitted by plants shows that there was no interaction (P> 0.05) among the studied factors for the 1st and 2nd day and the accumulated days of exposure of Bt and non-Bt cotton plants to aphids. The factor density [F density = 2.29,P density = 0.1294 (1st day);F density = 0.0070,P density = 0.95 (2nd day);F density = 0.1734,P density = 0.6813 (cumulative total)], cultivar [F cultivate = 0.003,P cultivate = 0.95 (1st day);F cultivate = 3.0896,P cultivar = 0.07 (2nd day);F cultivate = 2.2726,P cultivate = 0.1466 (cumulative total)] and period [F period = 1.3882,P period = 0.2387 (1st day);F period = 1.0805, P period= 0.3679 (2nd day); F period = 0.2966,P period = 0.5918 (cumulative total)] did not affect the isolation of VPs emitted by cotton plants.
There was an interaction between the factors aphid density versuscotton cultivar versus light/dark phase (F = 7.7295,P = 0.04150) for the number of VPs observed during the 3rd day of exposure of Bt cotton plants and non-Bt to aphids. It was found that on the third day, there was a higher VP production by non-Bt cotton plants exposed to 30 aphids/plant density than Bt cotton plants exposed to the same density during the photophase (Table 2). In addition, VP production by Bt and non-Bt cotton plants exposed to 60 and 30 aphids/plant, respectively, was higher in the light phase than in the dark phase (Table 2 ).
3.2 Dispersal pattern of A. gossypii in Bt and non-Bt cotton plants The behaviour of A. gossypii , independent of the factors of exposure time of plants to aphids, aphid density and cotton cultivar, followed a highly within-plant aggregated distribution pattern (k < 2) (Table 3 ).
Comparisons of the k index, based on confidence interval values, revealed that the highest k index of aphid aggregation with 30 aphids/non-Bt cotton plants was found at 48 h and 72 h after infestation of cotton plants with A. gossypii (Table 3 ). However, non-Bt cotton plants exposed to that density had a lower aphid kaggregation index at 72 h of infestation in relation to Bt cotton with 60 aphids/plant (Table 3 ).
In other words, the dispersal rate of A. gossypii was higher on Bt cotton with 60 aphids/plant than on non-Bt cotton plants with 30 aphids/plant at 72 h (Table 3 ). In fact, according to the multinomial distribution in the within-plant distribution of A. gossypii (Fig. 4 a, b, c, d ), it was confirmed that with 30 aphids/non-Bt cotton plants at 72 h, the highest proportions of aphids were on the adaxial (0.18) and abaxial (0.49) regions of the leaf (leaf I); however, there was increased insect dispersal to other positions, such as leaf II, adjacent leaf I and main meristem (Fig. 4 a ). On the other hand, on Bt cotton plants infested with 60 aphids/plant, we observed the most dispersal pattern with 72 h of infestation, where there was clearly an increased insect dispersal, with 0.16 and 0.41 of aphids found in the adaxial and abaxial regions of leaf I, respectively, and 0.20 in the main meristem of the cotton plant.
No significant difference was observed among the treatments within the infestation times of 0 h, 24 h and 48 h in relation to the kindex and multinomial distribution, except for the treatment with 60 aphids/Bt cotton plants within 48 h, which showed the most dispersal behaviour because it reached more regions of the cotton plants (Fig. 4b ).
In the comparisons of aggregation level among the time intervals within non-Bt cotton exposed to 60 aphids/plant, we perceived that the highestk aggregation index was during the infestation time of 48 h (Table 3 ). With Bt cotton plants at a density of 30 aphids/cotton plants, it was found that there was no change in the aphid dispersal pattern at all time intervals (Table 3 ).