Figure S1. Growth profiles of plants used in this study, depicting growth rates (% change / hr) relative to plant position (segment #) for specific segments over time. Surface plots (left): Growth rates (% change per hour, vertical axis) are plotted against the number of segments (defined by optical marker tags) from the apex downwards, over the duration of the imaging period in 10 minutes intervals. The red-shaded, nearest profile denotes the last 10 minute interval before harvest, depicted in greater detail in the right-hand scatterplot. Scatter plots (right): Growth rates (% change in length per hour) are plotted against distance from the stem base for specific segments. Segments are numbered from the top of the plant downwards in the right-hand margin, where sampled segments are indicated as white on black. The LOWESS regression curve follows the best fit through the growth rate data for this plant over a given 10' interval. Green dotted lines represented 65% confidence intervals for the LOWESS regression curve. Asterix indicates the stem position approximating the maximum growth rate of the regression curve, while the closed triangle indicates the first position below the top of the stem where the growth rate falls to zero (segment 10). See Hall and Ellis (2012) for further explanation (i.e. GKP for study of cell expansion in Arabidopsis inflorescences). Figure S2. Absolute fluorescence intensity (In.abs) of selected pectin-specific antibodies. Bar plots are given for each antibody:tissue-type combination, where values represent the mean of the normalized absolute fluorescence intensity (In.abs) for 3 biological replicates for each stage (YNG, MGR, and CSS). Plots are arranged by tissue type (column) and antibody (row). The final column presents the mean of sum of all normalized absolute fluorescence intensities for each biological replicate (denoted ‘Sum’). Antibodies are arranged row-wise according to their specificity to the degree of methyl-esterified pectins (‘High’, ‘Mod’/moderate, and ‘Low’; top to bottom, respectively). Values (Table S3) were derived from Ibn values (Table S2) via the method described in Figure S3. Error bars represent 95% confidence intervals computed from standard deviations (alpha=0.05). Figure S3. Relative representation of glycan-specificities in various assemblages of available antibodies. Values (ordinate) are expressed as a percentage of the total number of antibodies in those respective assemblages. The abscissa depicts the cell wall glycan categories appearing as ‘superclades’ in Pattathil et al., 2010, or otherwise categorized on the basis of published descriptions. Collections of antibodies are listed in the figure key with their total number in brackets. From left to right (top to bottom in key); those listed in the CCRC database as of 2012-10-26 (‘CCRC.listed’), those studied in Pattathil et al., 2010 (‘Pattat..study’), those provided in the CCRC-SKCMA-AR1 kit (‘SKCMA.AR1’), those provided in the CCRC-SKCMA-AM1 kit (‘SKCMA.AM1’), and those examined in this study (‘Hall.study’). Figure S4. Methodology for computation of absolute and relative intensity scores Flow chart depicts matrices or vectors (single columns) of various values within the R workspace, arrow connectors indicate computational processes used to generate matrices/vectors. Equations within gray boxes indicate the corresponding functions for computing those matrices/vectors. Definitions of entities and variables are presented in the following description of the method of calculation of Īn and Īrel.adj from Ībc. A relative fluorescence intensity value for the mean of three biological replicates was computed for each antibody according to: Īrel=Ībc (an,tn) / ∑Ībc(n) (Eq. 1) where: Ībc=absolute, background normalized (bc) intensity of a specific antibody::tissuetype combination (mean of biological replicates) Īrel=relative fluorescence intensity for each antibody::tissue-type combination an= the antibody (ranging from 0 to 55) tn= the tissue number (ranging from 1 to 7) However, a subset of antibodies exhibited low fluorescence intensity in all tissue types. Though signals were close to background, some values were many times greater than others leading to artificially elevated Īrel scores among tissue types. Such noise disrupts clustering similarities among the remainder of antibodies at a given stage. Rather than eliminate these dim antibodies from analysis, an approach was taken to scale the Īrel scores where sums of Ībc scores fell below a certain threshold. In order to do so, a proportional sum of intensities (PSI) was first derived to determine the relative, overall brightness of the tissue with respect to the brightest overall antibody at that stage using the relation: PSIn=∑Ībc(n) / max({∑Ībc,max}) (Eq. 2) where: PSIn = the proportional sum of intensities for a given antibody (n) Ībc = mean fluorescent intensity for each tissue (Ibc) for a given antibody at a specific stage max({∑Ībc})= maximum of all sums of mean fluorescence intensities (i.e. 55 sums for 55 antibodies). The vector 'PSI' was then used to calculate a correction factor (CF) according to the logistic function: CFn=1 / [1+e(-Wt*(PSIn-Th)] (Eq. 3) where: CFn= correction factor computed for each antibody, outputted to a vector containing values for all antibodies. Wt=weighting factor controlling the steepness of the logistic transformation of PSIn score Th=thresholding factor regulating the CF score below which the thresholding correction is applied e= natural logarithm (~2.718281828) The CF values were are also used to compute scores (Īrel.adj) according to the formula: Īrel.adj= Īrel + (Īrel - Tn-1)*CFn (Eq. 4) where: Tn = number of tissues (seven in this experiment) Īrel.adj=the relative intensity score for each antibody::tissue-type combination, adjusted by thresholding. For examination of absolute intensities across multiple antibodies (see Figure S4 on pectin-related antibodies), ‘normalized absolute intensity’ (Īn.abs) wer computed according through the formula: Īn.abs = Ībc * CFn (Eq.5) In this case, CF parameters (‘Wt’ and ‘Th’) were adopted from hierarchical clustering optimizations.