Of puzzles and pavements: a quantitative exploration of leaf epidermal cell shape

Abstract

Epidermal cells of leaves are diverse: tabular pavement cells, trichomes, and stomatal complexes. Pavement cells from the monocot Zea mays (maize) and the eudicot Arabidopsis thaliana (Arabidopsis) have highly undulate anticlinal walls. The molecular basis for generating these undulating margins has been extensively investigated in these species. This has led to two assumptions: first, that particular plant lineages are characterized by particular pavement cell shapes; and second, that undulatory cell shapes are common enough to be model shapes. To test these assumptions, we quantified pavement cell shape in epidermides from the leaves of 278 vascular plant taxa. We found that monocot pavement cells tended to have weakly undulating margins, fern cells had strongly undulating margins, and eudicot cells showed no particular undulation degree. Cells with highly undulating margins, like those of Arabidopsis and maize, were in the minority. We also found a trend towards more undulating cell margins on abaxial leaf surfaces; and that highly elongated leaves in ferns, monocots and gymnosperms tended to have highly elongated cells. Our results reveal the diversity of pavement cell shapes, and lays the quantitative groundwork for testing hypotheses about pavement cell form and function within a phylogenetic context.

Keywords: cell shape; diversity; morphometrics; pavement cell; phylogeny.

Figure 1

Epidermal peel images, leaf or leaflet outlines, and segmented cells for a small subset of sampled taxa. (a) Microsorum pteropus (Polypodiaceae), abaxial cells (left); and Tectaria pseudosinuata (Tectariaceae), abaxial cells (right). (b) Araucaria sp. (Araucariaceae), adaxial cells (left), and Microcycas calocoma (Zamiaceae), adaxial cells (right). (c) Drimys winteri (Winteraceae), adaxial cells (left); and Chloranthus sp. (Chloranthaceae), adaxial cells (right). (d) Homalocladium platycladum (Polygonaceae), adaxial cells (left); and Catharanthus roseus (Apocyanaceae), adaxial cells (right). (e) Alstroemeria aurea (Alstromeriaceae), adaxial cells (left); and Hemerocallis fulva (Xanthorrhoeaceae), adaxial cells (right). Leaf and cell outlines coloured according to major taxonomic divisions.

Figure 2

Traditional shape descriptors describe variation in base cell shape and margin undulation. (a) Principal component (PC) analysis of all epidermal cells sampled from monocots (pink) and eudicots (green) using traditional shape descriptors of aspect ratio (AR), area (A), circularity (C), and solidity (S). In this analysis, 69.7% of shape variance in the dataset was explained by the first two PC. The vectors describing the morphospace (inset) demonstrate how each shape descriptor relates to the first two components. (b) An illustration of cell solidity (S) calculated as the ratio of cell area to the convex hull area and its results from four representative cells with constant AR; colouring of representative cells matches quartiles bellow in (d). (c) An illustration of AR calculation as the ratio of maximal width to maximal length and its results from four representative cells with constant solidity value; colouring of representative cells matches quartiles bellow in (e). (d) The distribution of solidity values for our entire dataset, coloured according to quartiles. Twenty‐four cells from the median of each quartile are displayed with the same colour coding for reference. (e) The distribution of AR values for our entire dataset coloured according to quartiles. Twenty‐four cells from the median of each quartile are displayed with the same colour coding for reference.

Figure 3

Solidity and aspect ratio distributions varied between clades. (a) Solidity (S) and (b) aspect ratio (AR) data are presented as distributions by clade for ferns, gymnosperms, early‐diverging angiosperm (EDA) lineages, monocots, and eudicots. Note that the mean S values for both Arabidopsis (S _At = 0.65) and maize (S _Zm = 0.64) fell within the tails of the eudicot and monocot distributions, respectively. By contrast, mean AR values for both Arabidopsis (AR_At = 0.71) and maize (AR_Zm = 0.33) lay close to the mean values for eudicots and monocots, respectively. Arabidopsis and maize mean AR and S values marked with blue dots. No sample size scaling has been applied.

Figure 4

Pavement cells in the ferns, gymnosperms, and monocots were characterized by particular shape metrics. A maximum‐likelihood phylogenetic reconstruction of taxa sampled in our dataset (centre) surrounded by data rings depicting cell aspect ratio (AR) and solidity (S), and leaf AR for both adaxial and abaxial leaf surfaces (see key for positional key). Branch lengths are not shown in this figure, although they were used in all analyses. Taxonomic groups are indicated by colour. One representative cell shape from each species is depicted on the outermost ring. Evidence for phylogenetic signal was especially prevalent for cell metrics in the core monocots (a), and ferns (solidity); and in the grasses (b), and gymnosperms (AR). There was evidence for some phylogenetic signal at low taxonomic levels (family and/or genus) in the eudicots, but not across the clade as a whole. Each grey dot indicates multiple species in the same genus. All data and species names can be found in Supporting Information Table S1. The datamatrix and original treefiles are in the data repository (Vofely et al., 2018).

Figure 5

Cells on abaxial leaf surfaces tended to have more undulating margins. When there was a difference in mean cell solidity between the adaxial and abaxial leaf surfaces, solidity was often lower (higher undulation) on the abaxial leaf face. All data can be found in Supporting Information Table S1.

Figure 6

Anisotropic leaves tended to have anisotropic cells, but not in the eudicots or early‐diverging angiosperms. In the full dataset (a) and, in the ferns, gymnosperms and monocots (b), mean cell aspect ratio and mean leaf aspect ratio were correlated (regardless of leaf side). This correlation was not evident in the eudicots and early‐diverging angiosperms (c). No correlations were detected between leaf aspect ratio and mean cell solidity (d); mean cell aspect ratio and mean cell solidity (e); or log (mean cell area) and mean cell solidity (f). Linear regression lines are shown, along with Spearman's rho (ρ) correlation coefficients. All data can be found in Supporting Information Table S1.