Raw EPG data recorded by EPG Systems Stylet+d was manually annotated using EPG Systems Stylet+a software v01.30 (13-04-2016)/B27. Annotated waveforms were non-probing (np), pathway (C), phloem salivation (E1), phloem ingestion (E2), derailed stylet mechanics (F), xylem feeding (G) and intracellular puncture—potential drop (pd). Waveforms were identified based on the waveform pattern, amplitude, relative voltage level, R/emf origin, frequency, and the context of the waveform as described in the previous EPG studies of B. tabaci (Jiang et al., 1999 (link); Johnson and Walker, 1999 (link); Liu et al., 2012 (link); Civolani et al., 2014 (link); Zhou, 2014 ; Prado Maluta et al., 2017 (link)).
Annotation files were then directly passed to a modified version of the Ebert 3.0 program in SAS Enterprise Guide 7.1, SAS 9.4 statistical software (SAS Institute, Cary, NC, USA) for further analysis which produces the same parameters as the popular Sarria excel workbook (Sarria et al., 2009 (link); Ebert et al., 2015 (link)). The modified version is provided in theSupplementary Material . The modified version utilizes the series of BoxCox power transformation to determine the best possible transformation, as implemented in the PROC TRANSREG statement (Osborne, 2010 ). The results of this power transformation were inspected visually using histogram and Q-Q plots. For certain parameters, power transformations are unsuitable as they cannot approximate the necessary S-curve. Therefore, the Arcsine transformation was applied before the BoxCox transformation if this was necessary. The modification of the original Ebert 3.0 program also utilizes a macro script developed by Piepho (2012 ) that mitigates the case when varying standard error of a difference causes the traditional algorithm to fail to represent all significant differences of the means using the letter grouping. The Piepho algorithm solves the problem as it is able to generate a discontinous line display (Piepho, 2012 ). As a consequence, seeing discontinous assigned letters such as “ac” or “acd” is not uncommon (Piepho, 2014 (link); Poosapati et al., 2014 (link); Santos et al., 2015 (link); McCaghey et al., 2017 (link)).
Annotation files were then directly passed to a modified version of the Ebert 3.0 program in SAS Enterprise Guide 7.1, SAS 9.4 statistical software (SAS Institute, Cary, NC, USA) for further analysis which produces the same parameters as the popular Sarria excel workbook (Sarria et al., 2009 (link); Ebert et al., 2015 (link)). The modified version is provided in the
Full text: Click here