OSCILLATOR: A system for analysis of diurnal leaf growth using infrared photography combined with wavelet transformation

Abstract

Background: Quantification of leaf movement is an important tool for characterising the effects of environmental signals and the circadian clock on plant development. Analysis of leaf movement is currently restricted by the attachment of sensors to the plant or dependent upon visible light for time-lapse photography. The study of leaf growth movement rhythms in mature plants under biological relevant conditions, e.g. diurnal light and dark conditions, is therefore problematic.

Results: Here we present OSCILLATOR, an affordable system for the analysis of rhythmic leaf growth movement in mature plants. The system contains three modules: (1) Infrared time-lapse imaging of growing mature plants (2) measurement of projected distances between leaf tip and plant apex (leaf tip tracking growth-curves) and (3) extraction of phase, period and amplitude of leaf growth oscillations using wavelet analysis. A proof-of-principle is provided by characterising parameters of rhythmic leaf growth movement of different Arabidopsis thaliana accessions as well as of Petunia hybrida and Solanum lycopersicum plants under diurnal conditions. The amplitude of leaf oscillations correlated to published data on leaf angles, while amplitude and leaf length did not correlate, suggesting a distinct leaf growth profile for each accession. Arabidopsis mutant accession Landsberg erecta displayed a late phase (timing of peak oscillation) compared to other accessions and this trait appears unrelated to the ERECTA locus.

Conclusions: OSCILLATOR is a low cost and easy to implement system that can accurately and reproducibly quantify rhythmic growth of mature plants for different species under diurnal light/dark cycling.

Figure 1

Experimental setups and the procedure of leaf growth and movement analysis. (a) SLR Cameras are mounted to an aluminium frame inside a growth cabinet, IR illumination is provided from LED lights (far left and far right). (b) The camera frame is suspended above a tray containing randomised plants. (c) Image J plugins (File S1) allow tracking of the leaf tip throughout a virtual image stack, save the coordinates and project the trajectory. (d) The distance in mm from the leaf tip to the rosette centre is calculated, averaged and plotted against time. A best fit 2° polynomial regression line (red) is fitted to individual leaf curves and subtracted from the data. (e) The result is the residual from the regression line, here depicted as raw projected oscillations. Note: Originally decreasing distance between tip and centre indicated upward leaf movement. For clarity the residual projected oscillations were inverted to allow maximum upright leaf position to correspond to maximum peak height. (f) Harmonic noise is removed from individual leaf growth movement plots using wavelet analysis resulting in smoothed projected leaf oscillation curves. All data represent averages of 10 leaves: For 5 plants, 2 leaves per plant were tracked and the analysis was performed with these 10 leaves ( n = 10 ). Because of the high density, the SE’s were plotted for each data point and depicted as shading.

Figure 2

Natural variation in projected leaf lengths and projected leaf oscillations for selected Arabidopsis accessions. Projected lengths of selected accessions are depicted in the left column (a,c,e,g,i,k) and the inverted and smoothed projected oscillations in the right column (b,d,f,h,j,l). For all accessions 2 leaves per plant were analysed and in total 8 leaves (4 plants) were used for analysis ( n = 8 ) except for Cvi-0 where for one plant only one suitable leaf was tracked, ( n = 7 ). Error bars represent SE.

Figure 3

Natural variation of diurnal leaf growth oscillations in Arabidopsis. (a) Average amplitudes (day 2–6) of each accession. (b) correlation between reported angle [22] and measured average amplitude, (c) averaged period (day 2–6) of each accession and (d) averaged phase of smoothed projected oscillations (day 2 – 6) for the accessions. n = 8 leaves, except for Cvi-0 n = 7 , from four plants per accession. Error bars represent SE. One-way ANOVA was used to identify significant differences between the accessions. Individual contrasts were then identified in a post-hoc Tukey test. *; P < 0.05.

Figure 4

The ERECTA locus does not determine phase of leaf oscillations. (a) Comparison of smoothed projected oscillations between Ler-1 and Lan-0. (b) Timing of peak oscillations (phase) depicted per period, grey area indicates night. (c) No significant differences were observed for the average phase (between day 2 and day 6) of Ler-1 and Lan-0. Error bars represent SE ( n =8).

Figure 5

OSCILLATOR can be used for different species. (a-b) thirty-two day old petunia (a) and tomato (b) plants at the start of imaging. (c) Projected lengths were measured using OSCILLATOR. (d) From the projected lengths the projected oscillations were extracted, inversed and smoothed. Error bars represent SE, ( n  = 8).