The physiological importance of developmental mechanisms that enforce proper stomatal spacing in Arabidopsis thaliana



  • Genetic and cell biological mechanisms that regulate stomatal development are necessary to generate an appropriate number of stomata and enforce a minimum spacing of one epidermal cell between stomata. The ability to manipulate these processes in a model plant system allows us to investigate the physiological importance of stomatal patterning and changes in density, therein testing underlying theories about stomatal biology.
  • Twelve Arabidopsis thaliana genotypes that have varied stomatal characteristics as a result of mutations or transgenes were analyzed in this study. Stomatal traits were used to categorize the genotypes and predict maximum stomatal conductance to water vapor (Anatomical gsmax) for individuals. Leaf-level gas-exchange measurements determined Diffusive gsmax, net carbon assimilation (A), water-use efficiency (WUE), and stomatal responses to increasing CO2 concentration.
  • Genotypes with proper spacing (< 5% of stomata in clusters) achieved Diffusive gsmax values comparable to Anatomical gsmax across a 10-fold increase in stomatal density, while lines with patterning defects (> 19% clustering) did not. Genotypes with clustering also had reduced A and impaired stomatal responses, while WUE was generally unaffected by patterning.
  • Consequently, optimal function per stoma was dependent on maintaining one epidermal cell spacing and the physiological parameters controlled by stomata were strongly correlated with Anatomical gsmax.


Recent research has uncovered an increasingly complex genetic framework that controls stomatal development (Lau & Bergmann, 2012). Stomata play a critical role in plant physiology because the flow of gases into and out of the plant – mainly water vapor, carbon dioxide, and oxygen – is controlled by the size of the stomatal pore and the number of stomata present. Not surprisingly, plants lacking stomata are limited in size and habitat range, while plants that have lost stomata as a result of mutation are inviable (MacAlister et al., 2007; Vaten & Bergmann, 2012). To ensure the production of such a critical feature, plants have evolved an elegant developmental program that derives stomata from undifferentiated epidermal cells through a series of intermediate cell types (Fig. 1a). In the model system Arabidopsis thaliana, this progression is controlled by families of peptide ligands and cell-surface receptors, a mitogen-activated protein kinase (MAPK) signaling cascade, a suite of positively acting transcription factors, and a combination of other cell biology actors (Lau & Bergmann, 2012).

Figure 1.

Arabidopsis thaliana stomatal development pathway and deviations that lead to altered rates of stomatal production and to patterning defects. (a) A cartoon depiction of normal stomatal development that can lead to different values of the stomatal index (SI; the ratio of stomata to total epidermal cells), which correlates with changes in stomatal density (SD; the number of stomata per leaf area). Below, in gray, are two defects that lead to stomatal mispatterning as manifested by the genotypes used in this study. (b) Differential interference contrast (DIC) images of the abaxial epidermis of six genotypes as arranged by the extent of stomatal clustering and density. Additional information about the stomatal traits of all genotypes can be found in Tables 1-3. Bar (in Col-0 panel), 50 μm; all images are at the same magnification.

This sophisticated network not only provides multiple levels of regulation that are dependent on tissue identity, internal signals, and local environmental conditions, but it allows for great plasticity in determining the final pattern and number of stomata (Lake et al., 2001; Casson et al., 2009; Abrash et al., 2011). The capacity of plants to modulate stomatal development has also been observed over historical time-scales and among many different plant species, thus highlighting the evolutionary importance of ‘optimizing’ stomatal numbers (Beerling & Chaloner, 1993; Franks & Beerling, 2009b). We now possess the ability to manipulate specific genes that control stomatal development, and this technology enables us to break some of the basic rules in stomatal patterning and density that have evolved over millennia.

A central paradigm in stomatal development is the one-cell epidermal spacing rule, whereby stomata are separated from any surrounding stomata by at least one epidermal cell (Geisler et al., 2000). This spacing buffer has long been presumed to provide the necessary positioning for proper guard cell function and is nearly ubiquitous across extant species of plants (Sachs, 1991). The dynamic ability of guard cells to alter the size of the stomatal pore in response to local conditions relies upon ionic exchange with neighboring cells, which may depend on this epidermal cell buffer (Outlaw, 1983; Bergmann & Sack, 2007; Kim et al., 2010). Additionally, turgor pressure in the surrounding epidermal cells can provide mechanical forces that promote guard cell function, which may be lost when stomata are in contact (Edwards et al., 1976; Franks & Farquhar, 2007). In Arabidopsis thaliana, the large suite of genes involved in maintaining proper stomatal patterning during development supports the importance of this spacing model. A significant proportion of these are involved in cell-to-cell communication, such as extracellular ligands (the EPIDERMAL PATTERNING FACTOR (EPF) family) and their receptors (TOO MANY MOUTHS (TMM) and the ERECTA (ER) family), which orient divisions in the stomatal lineage and altogether limit the final number of stomata produced (Shpak et al., 2005; Lee et al., 2012). Signaling converges on the regulation of a transcription factor, SPEECHLESS (SPCH), that is required to initiate the stomatal lineage via the establishment of an asymmetric division (MacAlister et al., 2007). Asymmetric division is also critical for positioning stomata relative to each other, and BREAKING OF AYSMMETRY IN THE STOMATAL LINEAGE (BASL) is necessary to execute this specialized division (Dong et al., 2009). Manipulation of the above genes, among others, can violate the one-cell spacing rule and enable us to determine the physiological relevance of this evolutionarily conserved trait (Fig. 1a).

In order to address these questions regarding stomatal development, gas-exchange dynamics, and plant physiology, 12 A. thaliana lines with varied stomatal density and patterning characteristics as a result of mutations or transgenes were analyzed in a series of experiments. Leaf-level gas-exchange measurements were used to determine the maximum rate of stomatal conductance to water vapor (Diffusive gsmax) for individuals of each line, in addition to parameters for net carbon assimilation (A), water-use efficiency (WUE), and stomatal responses to increasing CO2 concentration [CO2]. The stomatal traits of those leaves were then quantified in order to calculate Anatomical gsmax (this term has also previously been called gmax or gwmax, but we will use Anatomical gsmax throughout this paper for consistency), which is the estimated maximum rate of stomatal conductance to water vapor based upon stomatal density and size in a diffusion-based equation (Eqn 1 in the 'Materials and Methods' section; Brown & Escombe, 1900; Franks & Farquhar, 2001; Franks & Beerling, 2009b). We also quantified the extent of stomatal clustering – the percentage of stomata in contact with at least one other stoma – to categorize the various genotypes by stomatal patterning. Here we show that stomatal clustering does indeed impact gas-exchange dynamics, negatively affecting both stomatal opening and closing. In addition, physiological benefits from plasticity in stomatal density, such as altered rates of carbon assimilation, are lost in mutants with clustered stomata.

Materials and Methods

Plant material and growth conditions

All genotypes tested were in the Columbia (Col-0) ecotype of Arabidopsis thaliana (L.) Heynh, and Col-0 was used as the control in all experiments. The following previously described genotypes were used: tmm-1 (Nadeau & Sack, 2002), basl-2 and tmm-1;basl-2 (Dong et al., 2009), tmm-1;erl1;erl2 (Shpak et al., 2005), epf1-1 (Hara et al., 2007), epf1-1;epf2-1 (Hunt & Gray, 2009), SPCHpro::SPCH-YFP and SPCHpro::SPCH 2-4A-YFP (Lampard et al., 2008), and FAMApro::CA-MKK9-YFP (Lampard et al., 2009). SPCHpro::SPCH 1D 2-5A-YFP was generated by the same methodology outlined in Lampard et al. (2008) and SPCH SILENCE was a stably silenced (> four generations) line derived from SPCHpro::SPCH 2-4A-YFP. Additional information on all genotypes can be found in Tables 1-3. Seeds were surface-sterilized and stratified at 4°C for 3–5 d in 0.15% agarose solution and then sown directly into pots of size 3.25″ × 3.25″ × 3″ filled with Pro-Mix HP soil (Premier Horticulture, Quakerstown, PA, USA) and supplemented with Scott's Osmocote Classic 14-14-14 fertilizer (Scotts-Sierra, Marysville, OH, USA). At 10–14 d, seedlings were thinned so that only one seedling per pot was remaining. Plants were grown to maturity in growth chambers where the conditions were as follows: day : night cycle, 16 h : 8 h; day : night temperature, 22 : 20°C; c. 100 μmol photon m−2 s−1 unless denoted as low-light (50 μmol photon m−2 s−1) or high-light (200 μmol photon m−2 s−1) growth conditions.

Table 1. Summary of gas flux dynamics and stomatal patterning by genotype
Genotype (growth condition)Diffusive gsmax (mol m−2 s−1)Anatomical gsmax (mol m−2 s−1)Stomatal density (mm−2)Clustered stomata (mm−2)Per cent clusteringPlants tested (n)Reference for line
  1. Values are mean ± SE.

  2. Diffusive gsmax, the maximum stomatal conductance as measured by leaf-level gas-exchange experiments.

  3. Anatomical gsmax, the maximum stomatal conductance as measured by stomatal size and density (Eqn 1).

Low Clustering (LOC)
SPCH SILENCE0.299 ± 0.0290.452 ± 0.05084.8 ± 9.40.8 ± 0.81.24This study
Col-0 (low light)0.408 ± 0.0670.689 ± 0.105130.1 ± 19.40.0 ± 0.00.04Ecotype Control
Col-00.585 ± 0.0531.049 ± 0.072185.7 ± 12.70.0 ± 0.00.08Ecotype Control
SPCH-YFP0.788 ± 0.0481.106 ± 0.053182.6 ± 8.70.0 ± 0.00.08Lampard et al. (2008)
epf1 0.936 ± 0.0940.923 ± 0.059209.3 ± 12.68.9 ± 2.54.07Hara et al. (2007)
Col-0 (high light)1.707 ± 0.1762.127 ± 0.088308.0 ± 12.50.8 ± 0.50.34Ecotype control
epf1;epf2 1.715 ± 0.3241.928 ± 0.122441.6 ± 27.621.6 ± 5.84.98Hunt & Gray (2009)
SPCH 2-4A1.995 ± 0.2691.722 ± 0.204359.3 ± 42.52.9 ± 0.80.87Lampard et al. (2008)
High Clustering (HiC)
tmm;erl1;erl2 0.313 ± 0.0200.490 ± 0.049100.6 ± 10.119.5 ± 3.019.24Shpak et al. (2005)
tmm 0.481 ± 0.0870.761 ± 0.078174.0 ± 17.072.7 ± 8.441.96Nadeau & Sack (2002)
basl 0.630 ± 0.0351.016 ± 0.026223.2 ± 5.7100.5 ± 4.345.216Dong et al. (2009)
FAMA::KK90.824 ± 0.1341.164 ± 0.163263.3 ± 3.2146.7 ± 67.248.05Lampard et al. (2009)
SPCH 1D 2-5A0.872 ± 0.1621.656 ± 0.267376.8 ± 61.682.4 ± 28.419.16This study
tmm;basl 0.540 ± 0.0401.676 ± 0.146387.3 ± 29.8214.8 ± 28.555.54Dong et al. (2009)
Table 2. Measurements used to calculate the anatomical maximum rate of stomatal conductance to water vapor (Anatomical gsmax) for genotypes with Low Clustering (LoC)
 Pore length (μm)Pore depth (μm)amax (μm2)Size (mm2)Density (mm−2)Number of stomata measured (n)
  1. Values are mean ± SE.

  2. amax, mean maximum stomatal pore area.

Adaxial stomata
 SPCH SILENCE12.59 ± 0.174.95 ± 0.0563.27 ± 1.75234.54 ± 3.2684.8 ± 9.478
 Col-0 (low light)13.41 ± 0.305.10 ± 0.0771.88 ± 3.38268.76 ± 5.2352.5 ± 10.433
 Col-013.35 ± 0.315.02 ± 0.1071.17 ± 3.30249.22 ± 4.5180.5 ± 5.630
 SPCH-YFP13.53 ± 0.335.23 ± 0.1072.98 ± 3.65269.12 ± 7.1154.9 ± 2.323
 epf1 12.81 ± 0.345.54 ± 0.0865.71 ± 3.53280.05 ± 6.5975.9 ± 5.026
 Col-0 (high light)17.01 ± 0.285.39 ± 0.09114.59 ± 3.84345.15 ± 6.80129.9 ± 7.627
 epf1;epf2 12.87 ± 0.425.36 ± 0.1066.64 ± 4.09244.46 ± 5.83138.9 ± 11.023
 SPCH 2-4A12.31 ± 0.235.25 ± 0.0960.18 ± 2.17250.93 ± 5.46110.5 ± 7.628
Abaxial stomata
 SPCH SILENCE0.0 ± 0.00
 Col-0 (low light)13.36 ± 0.235.55 ± 0.0671.36 ± 2.35287.15 ± 4.8277.5 ± 9.760
 Col-013.47 ± 0.185.39 ± 0.0772.08 ± 1.89287.13 ± 4.61105.3 ± 8.260
 SPCH-YFP14.02 ± 0.305.63 ± 0.0779.21 ± 3.19296.64 ± 5.88127.7 ± 7.455
 epf1 11.10 ± 0.285.75 ± 0.0749.83 ± 2.49273.17 ± 6.52133.4 ± 9.145
 Col-0 (high light)14.99 ± 0.485.89 ± 0.0791.75 ± 5.54371.81 ± 11.78178.1 ± 6.938
 epf1;epf2 11.33 ± 0.335.36 ± 0.0753.63 ± 2.92257.51 ± 7.05302.7 ± 18.973
 SPCH 2-4A12.23 ± 0.295.59 ± 0.0760.89 ± 2.65277.02 ± 7.17248.8 ± 37.061
Table 3. Measurements used to calculate the anatomical maximum rate of stomatal conductance to water vapor (Anatomical gsmax) for genotypes with High Clustering (HiC)
 Pore length (μm)Pore depth (μm)amax (μm2)Size (mm2)Density (mm−2)Number of stomata measured (n)
  1. Values are mean ± SE.

  2. amax, mean maximum stomatal pore area.

Adaxial stomata (not in clusters)
 tmm;erl1;erl2 11.43 ± 0.225.19 ± 0.1051.83 ± 1.95244.23 ± 10.3530.9 ± 1.824
 tmm 11.91 ± 0.235.18 ± 0.1056.34 ± 2.16222.34 ± 9.1039.8 ± 5.026
 basl 11.63 ± 0.184.99 ± 0.0853.88 ± 1.75225.85 ± 4.11115.2 ± 3.852
 FAMA::KK912.76 ± 0.465.25 ± 0.1565.39 ± 4.70283.75 ± 9.78134.7 ± 45.917
 SPCH 1D 2-5A11.56 ± 0.285.03 ± 0.0953.33 ± 2.65240.59 ± 13.2795.2 ± 8.927
 tmm;basl 11.83 ± 0255.18 ± 0.0856.00 ± 2.34258.76 ± 6.5664.1 ± 3.241
Adaxial stomata in clusters
 tmm;erl1;erl2 9.49 ± 0.344.84 ± 0.1635.73 ± 2.55223.14 ± 14.162.0 ± 1.58
 tmm 11.14 ± 0.844.25 ± 0.2351.58 ± 7.62210.66 ± 16.225.7 ± 1.611
 basl 11.00 ± 0.204.73 ± 0.1148.36 ± 1.80213.83 ± 4.5560.5 ± 2.351
 FAMA::KK99.16 ± 0.464.55 ± 0.1834.61 ± 3.60184.37 ± 17.1776.7 ± 52.220
 SPCH 1D 2-5A11.23 ± 0.544.64 ± 0.1950.89 ± 5.11212.16 ± 13.2517.5 ± 4.512
 tmm;basl 11.65 ± 0.494.69 ± 0.1254.78 ± 4.46230.59 ± 15.8314.7 ± 3.616
Abaxial stomata (not in clusters)
 tmm;erl1;erl2 13.36 ± 0.356.01 ± 0.0871.87 ± 3.80341.29 ± 15.4152.2 ± 5.435
 tmm 12.85 ± 0.365.90 ± 0.1666.09 ± 3.47299.28 ± 10.4167.2 ± 9.124
 basl 11.96 ± 0.225.53 ± 0.0657.65 ± 1.99269.66 ± 4.6768.1 ± 3.075
 FAMA::KK914.00 ± 0.375.87 ± 0.1578.42 ± 3.87359.83 ± 13.5858.6 ± 15.526
 SPCH 1D 2-5A11.25 ± 0.305.41 ± 0.0751.04 ± 2.55248.32 ± 5.86216.7 ± 29.236
 tmm;basl 11.95 ± 0.395.34 ± 0.0758.31 ± 3.52245.01 ± 7.16123.1 ± 21.636
Abaxial stomata in clusters
 tmm;erl1;erl2 11.51 ± 0.975.94 ± 0.1956.13 ± 9.35331.99 ± 23.1317.6 ± 4.012
 tmm 9.74 ± 0.435.50 ± 0.1438.98 ± 3.18233.40 ± 15.6666.2 ± 8.024
 basl 11.23 ± 0.315.02 ± 0.1050.57 ± 1.09249.22 ± 4.5140.0 ± 2.932
 FAMA::KK99.41 ± 0.405.23 ± 0.1236.65 ± 3.35187.25 ± 8.5270.0 ± 17.330
 SPCH 1D 2-5A11.02 ± 0.415.18 ± 0.1248.32 ± 3.68235.07 ± 8.6164.5 ± 27.710
 tmm;basl 10.19 ± 0.395.10 ± 0.1143.72 ± 3.09233.62 ± 12.30200.2 ± 31.849

Leaf-level gas-exchange measurements

Measurements were taken on the largest and most accessible mature rosette leaves at 5–7 wk post germination using a LI-6400 Portable Photosynthesis System with the 6400-02B LED Light Source (Li-Cor Biosciences Inc., Lincoln, NE, USA). Only one leaf was sampled per plant and between four and 16 plants were sampled per genotype. To increase plant throughput, experiments were performed using a combination of the LI-6400 for gas-exchange measurements and a customized growth chamber to pre-incubate plants at different [CO2]. The chamber consisted of a glass aquarium of dimensions 20″ × 10″ × 12″ with a sealed plexiglass lid that had a small hatch through which to insert plants into and remove them from the chamber. [CO2] in the chamber was monitored and adjusted by an infrared gas analyzer (Infrared Industries Inc., Santa Barbara, CA, USA) attached to a solenoid valve and compressed CO2 gas tank. [CO2] below ambient were achieved by forced air circulation inside the chamber over a container of soda lime (Alfa Aesar, Ward Hill, MA, USA). The [CO2] in the chamber was raised stepwise from 100 to 350, 500, 750, and finally 1000 ppm (± 20 ppm) in advance of LI-6400 measurements at each of the same [CO2]. Other conditions in the chamber remained constant: relative humidity (RH) > 80%, and temperature c. 22°C (Taylor Indoor Hygrometer/Thermometer, Oak Brook, IL, USA); light, at 100–150 μmol photon m−2 s−1, was provided by blue and red LEDs mounted above the chamber (PAR38 Ultra Bright LED Light Bulbs; Plants were able to equilibrate at each [CO2] for at least 45 min before sampling with the LI-6400, with a longer incubation of 1–2 h taking place in the initial conditions at 100 ppm CO2. Conditions in the LI-6400 were as follows: flow rate, 500 μmol s−1 (± 1); temperature, 22°C (± 0.1); photosynthetically active radiation (PAR), 1000 μmol photons m−2 s−1 (± 1); incoming RH, 75–80%; [CO2], 100–1000 μmol mol−1 (ppm is used interchangeably). Incoming RH was controlled by a custom-made dew-point controller attached to the air input of the LI-6400 (Larry Giles; Carnegie Institute for Science, Stanford, CA, USA). Once the leaf was inside the LI-6400 chamber, gas-exchange was allowed to come to a steady state for 5–15 min and then measurements were recorded by the LI-6400 for 5 min. Final parameter values were determined as the average value over those final 5 min. Leaves were excised from the plant after completing LI-6400 measurements and prepared for stomatal phenotype analysis (see Stomatal trait analysis). Leaf size was determined after cutting the leaf at the inside boundary of the LI-6400 sensor head gasket so that the LI-6400 file could be recomputed with proper leaf area and stomatal ratio (recomputed after stomatal phenotype analysis; no leaves filled the entire 2 × 3 cm chamber). The 6400-02B LED Light Source and 2 × 3 cm default sized chamber were used in favor of the 6400-15 Arabidopsis Chamber because of the improved gas-exchange measurements we could attain with better temperature control and increased leaf area. The average A. thaliana leaf area we achieved in the 6-cm2 chamber was 3.76 cm2 (min: 1.97 cm2; max: 5.71 cm2), compared with a maximum leaf area of 0.79 cm2 with the 6400-15 chamber. This increase in leaf area allowed for greatly improved sensitivity and stability in measuring differences in conductance between plants with distinct, but sometimes very similar, stomatal densities.

Stomatal trait analysis

Rosette leaves were prepared for stomatal phenotype analysis in the following steps: (1) cleared with 7 : 1 ethanol:acetic acid solution overnight or longer; (2) softened for 30 min in 1 M potassium hydroxide; (3) rinsed with water; (4) mounted on slides with Hoyer's solution; and (5) visualized by differential interference contrast (DIC) microscopy on a Leica DM2500 microscope at ×20 magnification (0.32 mm−2 field of view). In the middle of the leaf, between the midvein and the edge, four images on both the abaxial and adaxial sides of the leaf were taken. Stomatal density (D; mm−2), number of stomata with proper spacing, number of stomata in contact with one or more stomata (i.e. in clusters), and cluster size were manually counted for all pictures and all leaves using the Cell Counter in Image J (NIH, Per cent clustering was calculated by (1) counting the total number of stomata per unit area, (2) counting the number of stomata that were in contact with another stoma in that area (stomata in clusters), and (3) dividing the number of stomata in clusters by the total number of stomata. Empirical observation of the 12 genotypes revealed a natural break point in the distribution, with six genotypes having fewer than 5% of stomata in clusters (Low Clustering (LoC)) and six genotypes (High Clustering (HiC)) having extensive clustering (> 19%, i.e. well above 5%). Stomatal dimensions were counted on two to four leaves per genotype (75 ≤  210 stomata) to calculate average dimensions for each genotype. For this study, stoma size (S; μm2) was defined as an ellipse with major axis equal to guard cell length and minor axis equal to the width of the entire stoma; pore depth (l; μm) was equal to guard cell width at the center of the stoma; mean maximum stomatal pore area (amax; μm2) was an ellipse with major axis equal to pore length and minor axis equal to 1/2 pore length.

Calculating Anatomical gsmax

Maximum stomatal conductance to water vapor as defined by stomatal anatomy (Anatomical gsmax; mol m−2 s−1) was estimated for each leaf using a double end-correction version of the equation by Franks & Farquhar (2001):

display math(Eqn 1)

(d, the diffusivity of water in air (m2 s−1, at 22°C); v, the molar volume of air (m3 mol−1, at 22°C); π, the mathematical constant, approximated to 3.142.) Total leaf Anatomical gsmax for each individual was calculated as the sum of Anatomical gsmax abaxial (gab) and Anatomical gsmax adaxial (gad) stomata using empirical values of D, l, and amax as calculated for stomata on each side of the leaf (Anatomical gsmax = gab + gad). D was determined independently for each leaf, while values of l and amax were genotype averages. For HiC genotypes, Anatomical gsmax was calculated for each leaf as the sum of four components: Anatomical gsmax of abaxial stomata, Anatomical gsmax of clustered abaxial stomata, Anatomical gsmax of adaxial stomata, and Anatomical gsmax of clustered adaxial stomata (Anatomical gsmax = gab + gab cluster + gad + gad cluster). Genotype averages for all stomatal dimensions used to calculate Anatomical gsmax can be found in Tables 2 and 3.

Determination of water-use efficiency (WUE)

The WUE of each plant was determined using leaf-level gas-exchange measurements collected from the LI-6400 at 350 ppm CO2. Intrinsic WUE was calculated for each leaf by dividing the photosynthetic rate (μmol CO2 m−2 s−1) by the stomatal conductance (mol H2O m−2 s−1). Instantaneous WUE was calculated for each leaf by dividing the photosynthetic rate (μmol CO2 m−2 s−1) by the transpiration rate (mmol H2O m−2 s−1).

Statistical analysis

All statistical analysis was performed using R ( Regression models were used to determine correlation coefficients (adjusted R2) and the statistical significance of co-variation between parameters (< 0.05). For instances where a significant correlation between parameters was identified, the equation of the relationship is provided in Supporting Information Table S1. Comparison of regression models between stomatal patterning groups was performed using ANCOVA to determine if any significant differences existed between regression slopes (effect of patterning on dependent variable) or y-intercepts (quantitative differences between patterning groups). Pair-wise comparison of means was performed using a Wilcoxon signed rank test for nonparametric samples.


Mature A. thaliana rosette leaves were sampled in order to analyze their stomatal traits, including: size, pore length and depth, density, and stomatal clustering on both abaxial and adaxial sides of the leaf. In all, 12 different A. thaliana lines with diverse stomatal traits were tested, including seven mutant lines (epf1, epf1;epf2, tmm, tmm;erl1;erl2, tmm;basl, and basl), five transgenic lines (SPCH SILENCE, SPCH-YFP, SPCH 2-4A, SPCH 1D 2-5A, and FAMA::MKK9), and the ecotype control, Col-0 (Fig. 1b, Table 1). These lines were chosen because of their overall wild-type appearance and restriction of the manipulated genes to the stomatal lineage, which should limit the differences between lines only to changes in stomatal patterning. We avoided other stomatal mutants, such as null mutants of spch or erecta, which have significant growth defects or pleiotropic affects on plant physiology and morphology, respectively, in addition to altering stomatal development (Masle et al., 2005; MacAlister et al., 2007). Consequently, we regularly combined genotypes into collective groups for analysis and individual lines are not often referenced in the text.

In addition to ten previously published lines, we included two new lines: SPCH SILENCE, which completely lacks stomata on the abaxial side of mature true leaves, while stomatal patterning on the adaxial surface is largely unaffected; and SPCH 1D 2-5A, which has increased stomatal density and c. 20% stomata in clusters on both sides of the leaf (Tables 1-3). We did, however, observe quantitative differences in the stomatal patterning of some published lines when comparing cotyledons to mature rosette leaves. For example, clustering in epf1 was calculated at 37% in cotyledons and 16% on primary leaf abaxial surfaces (Hara et al., 2007). We found that clustering had reduced even further in mature rosette leaves of epf1 – only 3.4% on abaxial surfaces. This phenomenon was common to multiple genotypes (data not shown) and could be the consequence of different growth conditions (soil versus culture plate) or, more interestingly, differences in the regulation of stomatal development with respect to leaf number; increases in leaf size that alter lineage proliferation requirements; or changes in the number of signals perceived and integrated by rosette leaves versus cotyledons and primary true leaves.

Based upon the observed patterns in stomatal spacing of rosette leaves, two categorical groups were defined: genotypes with no or minimal clusters (< 5% of stomata in clusters; Tables 1 and 2) which we will refer to as Low Clustering (LoC), and genotypes with a significant proportion of stomata in clusters (> 19%; Tables 1 and 3), which will be termed High Clustering (HiC). Six lines fell into the LoC category (listed in order of increasing stomatal density): SPCH SILENCE, Col-0, SPCH-YFP, epf1, SPCH 2-4A, and epf1;epf2; while six lines were categorized as HiC: tmm;erl1;erl2, tmm, basl, FAMA::KK9, SPCH 1D 2-5A, and tmm;basl (Fig. 1b, Table 1). Considered together, the above panel represents a near 10-fold range in stomatal density for genotypes with both properly spaced and clustered patterns of stomata.

In natural populations, there is a negative relationship between stomatal size and density that is enforced because of constraints on stomatal packing and organization in the epidermis (Hetherington & Woodward, 2003; Franks & Beerling, 2009b; Franks et al., 2009). Our mutant lines were consistent with this precedent, regardless of clustering, and had a negative log correlation between average values of stomatal size and density (Fig. 2a, abaxial stomata: < 0.05; R2 = 0.39 for LoC; < 0.001; R2 = 0.58 for HiC; Fig. 2b, adaxial stomata: < 0.05; R2 = 0.24 for LoC; < 0.05; R2 = 0.19 for HiC). These results were also in line with a recent study that focused on EPF family mutants grown under different [CO2] regimes (Doheny-Adams et al., 2012). For abaxial stomata, the slope of the regression lines were different between patterning groups (< 0.05; ANCOVA), while for adaxial stomata, the slopes of the regression lines were not different, but the y-intercept was (= 0.876 and < 0.001, respectively; ANCOVA). This indicated that clustering did have some effect on the empirical relationship between stomatal size and density in A. thaliana, but not so severely as to disrupt the accepted negative correlation between those parameters. Finally, the size of stomata in clusters was smaller than the size of properly spaced stomata for all genotypes with HiC (Table 3).

Figure 2.

Stoma size on the abaxial and adaxial sides of the leaf were negatively correlated with stomatal density, irrespective of clustering phenotypes. Logarithmic models (< 0.05) describe the data for two groups as categorically defined by stomatal patterning in Arabidopsis thaliana: individuals with < 5% clustering (Low Clustering (LoC); open circles; = 14; R2 = 0.39 in (a); R2 = 0.24 in (b); error bars: ± SE) and individuals with > 19% clustering (High Clustering (HiC); closed circles; = 14 and R2 = 0.58 in (a); = 19 and R2 = 0.19 in (b)). In (a), the slopes of the categorical regressions were different (< 0.05), while in (b), the slopes were not different, but the y-intercepts were (= 0.876 and < 0.001, respectively). Specific regression equations for all figures, along with R2 and P-values, can be found in Supporting Information Fig. S1. In this figure, only the size and density of nonclustered stomata were included for both LoC and HiC groups. Additional information on the size and density of clustered stomata in the HiC lines can be found in Table 3.

To understand the effects of clustering on stomatal opening, we obtained experimental values of the maximum stomatal conductance to water vapor (Diffusive gsmax) for comparison with predicted values of Anatomical gsmax. Conditions at the leaf surface were established that supported maximum stomatal opening, including low [CO2], saturating light intensity, high RH, and ample soil water. Then, leaf-level gas-exchange measurements were performed. Diffusive and Anatomical gsmax were measured for each leaf and pooled by patterning group (Fig. 3; n = 50 for LoC and = 41 for HiC). Although values of gsmax were positively correlated for both patterning groups (< 0.001; R2 = 0.60 for LoC; < 0.01; R2 = 0.16 for HiC) only the regression model for LoC was not significantly different from a 1 : 1 reference line that represents a perfect match between Anatomical and Diffusive gsmax (95% confidence intervals). Additionally, the slopes of the regression models were significantly different (< 0.001; ANCOVA), indicating that stomatal clustering disrupted the ability of Anatomical gsmax to accurately predict Diffusive gsmax.

Figure 3.

Stomatal clustering prevented Anatomical gsmax from being a reliable predictor of Diffusive gsmax. Leaf-level gas-exchange measurements of maximum stomatal conductance to water vapor (Diffusive gsmax) were compared to a calculated estimate of maximum stomatal conductance based on stomatal size and density (Anatomical gsmax). The regression model of Arabidopsis thaliana plants with Low Clustering (LoC; open circles; = 50; < 0.001; R2 = 0.60; dashed line) was not significantly different from a 1 : 1 reference line (solid gray) that represents a perfect match between Diffusive gsmax and Anatomical gsmax (95% confidence intervals). Plants with High Clustering had a regression model (HiC; closed circles; = 41; < 0.01; R2 = 0.16; solid line) that was significantly different from the reference line (95% confidence intervals) and the slope of the LoC regression model (< 0.001). Two LoC values of Diffusive gsmax were above 3.0 mol m−2 s−1 and are not shown in the graph but were included in the analysis.

This comparison of gsmax values demonstrated that plants with significant stomatal clustering could not get as close to the theoretical maximum rate of gas exchange (Anatomical gsmax) as plants with properly spaced stomata (Fig. 3). For each patterning group, the average proportion of Anatomical gsmax that was reached by Diffusive gsmax was 81.82 ± 4.38% (mean ± SEM) for LoC and 60.51 ± 3.06% for HiC. This significant difference (< 0.001) between maximum rates of conductance underscored the importance of the developmental mechanisms that ensure proper stomatal patterning.

These results established that Anatomical gsmax was a poor estimate of Diffusive gsmax for plants with HiC and suggested that stomata in clusters function differently than stomata that are properly spaced. This model presumes that stomatal function is reduced by some factor as a result of stomatal clustering and we sought to quantify this factor by testing additional predictions of gsmax. We compared the original Anatomical gsmax prediction, which assumed that all stomata, regardless of clustering, fully contributed to gsmax, with two other scenarios. The first alternative was to assume that the contribution to gsmax of a single cluster (irrespective of the number of stomata in it) was equivalent to the contribution of one stoma (termed ‘Clusters As One Stoma’ gsmax in Fig. 4; COS gsmax in the text). The second scenario assumed that only properly spaced stomata were functional and capable of contributing to gsmax (‘Only Nonclustered’ gsmax in Fig. 4; NC gsmax in the text). All predictions were calculated using Eqn 1, but stomatal density was altered according to each assumption about the number of functional stomata in clusters. For consistency in analysis of the clustering effect, we selected a single HiC genotype (basl) because of the regular nature of its clusters (predominantly pairs) and the large number of samples we had measured (= 16). All three predictions were correlated with Diffusive gsmax (Fig. 4: < 0.001; R2 = 0.65 for Anatomical gsmax; < 0.01; R2 = 0.44 for COS gsmax; and < 0.05; R2 = 0.20 for NC gsmax). However, only the COS gsmax prediction was not significantly different from the 1 : 1 reference line (95% confidence intervals). Furthermore, the average proportion of COS gsmax that was reached by Diffusive gsmax was 81.26 ± 3.32%, nearly identical to the percentage achieved by lines with LoC (81.82 ± 4.38%). Meanwhile, NC gsmax overestimated Diffusive gsmax (115.25 ± 6.08%) and Anatomical gsmax was at the average for HiC plants (61.61 ± 2.35%). These post hoc quantitative predictions suggested that stomata in clusters did, in fact, perform gas exchange, but conductance was compromised as a result of contact with other stomata. Furthermore, the closest approximation, COS gsmax, suggested that stomata in clusters were equivalent in function to a single stoma, inferring that the production of additional stomata within a single lineage provides no advantage unless they are correctly spaced.

Figure 4.

Inferring quantitative differences about the functionality of stomata in clusters provided a more accurate prediction of the Diffusive gsmax. To quantify the functionality of stomata in clusters in Arabidopsis thaliana, the observed values of maximum stomatal conductance to water vapor (Diffusive gsmax) were compared with three quantitative predictions of gsmax for basl (breaking of asymmetry in the stomatal lineage) plants (= 16): Anatomical gsmax (closed circles), which assumes that all stomata contribute equally to gsmax (< 0.001; R2 = 0.65; solid line); ‘Clusters As One Stoma’ gsmax (open squares), which assumes that a single cluster only contributes the equivalent of One Stoma to gsmax, regardless of how many stomata it contains (< 0.01; R2 = 0.44; dashed line); and ‘Only Nonclustered’ gsmax (closed squares), which assumes that stomata in clusters are nonfunctional and contribute nothing to gsmax (< 0.05; R2 = 0.20; solid line). Of these three predictions, Diffusive gsmax was only consistent (slope and y-intercept within 95% confidence intervals of solid gray line) with the ‘clusters as one stoma’ model.

The presence of at least one epidermal cell to buffer the space between stomata has also been a presumed requirement for the dynamic activities of guard cells. We could test the contribution of stomatal patterning to guard cell responses independent of stomatal density because we had access to pairs of LoC and HiC lines with similar stomatal densities (high: epf1;epf2 versus tmm;basl; medium: Col versus basl; and low: SPCH-SIL versus tmm;erl1;erl2; Fig. 5; n = 4 per genotype). Notably, there was a large disparity in the dynamic range (difference between highest and lowest values) of stomatal conductance when comparing the responses of each patterning group to increasing [CO2]. At 100 ppm CO2, the range in stomatal conductance for genotypes with LoC was 1.223 mol m−2 s−1, while genotypes with HiC only had a range of 0.227 mol m−2 s−1, a five-fold difference. At 1000 ppm CO2, the difference in dynamic range was even higher, almost eight-fold. If each genotype was considered independently, the per cent decrease in stomatal conductance was significantly reduced for lines with HiC (Fig. 5c; P < 0.05). Overall, the average decrease in stomatal conductance from 100 to 1000 ppm CO2 was 72% for LoC lines and 40% for HiC lines. The initial slope of the response curve (between 100 and 500 ppm CO2) was also compelling: the average rate of decrease for each LoC line was greater than any rate for HiC lines, although some differences were not significant (< 0.05). These discrepancies in the dynamic range, per cent decrease, and rate of decrease indicated that stomatal clustering limited the operational range of stomatal conductance and impaired the ability of stomata to rapidly respond to changes in the local environment, both critical aspects of optimal stomatal activity and leaf productivity.

Figure 5.

Stomatal clustering impaired responses to increasing [CO2]. Stomatal conductance (gs) was determined by leaf-level gas-exchange measurements of individual rosette leaves of mature Arabidopsis thaliana plants at stepwise [CO2] from 100 to 1000 ppm (a; = 4 per genotype; error bars: ± SE). Low Clustering (LoC) genotypes are shown as open symbols with dashed lines, while High Clustering (HiC) genotypes are shown as closed symbols with solid lines. (b) Average stomatal density and patterning of the leaves that were used in gas-exchange measurements for (a). Genotypes with HiC are shown as a stacked bar: the black bar is the density of properly spaced stomata, and the upper gray bar is the density of stomata in clusters. Letters a, b, and c above the bars represent three significantly different groups of stomatal density (< 0.05). (c) The average per cent decrease in gs from the initial rate at 100 ppm CO2 to the final rate at 1000 ppm. Genotypes indicated by the letter a had a significantly larger per cent decrease in gs than those genotypes indicated by the letter b or c (< 0.05).

Recent work has demonstrated that operational stomatal conductance can be estimated as a function of Anatomical gsmax and the environmental conditions at the leaf (Dow et al., 2013). Essentially, this model demonstrated that physiological responses of stomata to the environment were independent of stomatal density. Stomatal clustering appeared to alter this relationship, as normalized responses for HiC lines produced a different profile from LoC lines (Fig. 6). The normalized stomatal conductance for LoC lines dropped 50% as CO2 increased from 100 to 1000 ppm, while HiC lines only experienced a 20% decrease. These relative differences from Anatomical gsmax were another example of how clustering disrupted the normal signatures of stomatal function and allowed us to quantify the importance of maintaining stomatal spacing. Of general interest, this comparative approach may be a useful tool for identifying other defects in stomatal physiology, arising from either genetic mutation or natural variation, without having to consider differences in stomatal density or size.

Figure 6.

Stomatal clustering altered the relationship between operational stomatal conductance (gs) and Anatomical gsmax in Arabidopsis thaliana. Response curves (from Fig. 5a) were normalized to individual values of maximum stomatal conductance (Anatomical gsmax) and combined by patterning group (= 12 per group; error bars ± SE). Low Clustering (LoC), open circles; High Clustering (HiC), closed circles.

In addition to measuring parameters of stomatal conductance, analyses of net carbon assimilation (A) and WUE were necessary to understand the physiological outcomes of altering stomatal density and patterning. At 350 ppm ambient CO2, both patterning groups had intercellular [CO2] (Ci) values that were modestly correlated with Anatomical gsmax, as represented by linear models (Fig. 7a: < 0.05; R2 = 0.072 for LoC; < 0.001; R2 = 0.25 for HiC). Slopes of the regression models were not significantly different, indicating that stomatal patterning did not affect the relationship between Anatomical gsmax and Ci, but the intercept for LoC was significantly lower than that of HiC plants (Fig. 7a: = 0.369 and < 0.001, respectively; ANCOVA). This reduction in Ci for LoC plants can probably be explained by the enhancement in carbon assimilation compared with plants with HiC (Fig. 7b: < 0.001; R2 = 0.46 for LoC; = 0.457 for HiC). Previous reports have also linked Anatomical gsmax with carbon assimilation rates, via leaf nitrogen content or maximum rates of Rubisco carboxylation (Vcmax; Franks & Beerling, 2009a; Franks et al., 2009), but the positive relationship we identified was lost when stomatal patterning was disrupted. This implies that the organization of stomata on the epidermis and coordination with the underlying mesophyll tissues are both important for delivering gains in carbon assimilation.

Figure 7.

Stomatal clustering prevented a positive correlation between net carbon assimilation (A) and the Anatomical gsmax. Experimental values for intercellular [CO2] (Ci; a) and A (b) were determined by leaf-level gas-exchange measurements of individual rosette leaves of mature Arabidopsis thaliana plants at 350 ppm CO2. Ci was correlated with maximum stomatal conductance (Anatomical gsmax) for both patterning groups (= 50; < 0.05; R2 = 0.074 for Low Clustering (LoC); open circles; = 41; < 0.001; R2 = 0.25 for High Clustering (HiC); closed circles). The slopes of the regression models in (a) were not significantly different but the y-intercepts were (= 0.369 and < 0.001, respectively). (b) A was correlated with Anatomical gsmax only for plants with LoC (< 0.001; R2 = 0.46 for LoC; = 0.457 for HiC).

Intrinsic WUE is a parameter that denotes the ratio of net carbon assimilation to stomatal conductance (A/gs) and is a useful metric for comparing the photosynthetic efficiency of plants independent of the evaporative demands created by a given environment, provided that the environment is similar across measurements (Osmond et al., 1980). At 350 ppm CO2, intrinsic WUE was negatively correlated with Anatomical gsmax for both patterning groups, as described by linear models (Fig. 8a: < 0.01; R2 = 0.14 for LoC; < 0.001; R2 = 0.24 for HiC). Once again, the slopes of the regressions were not significantly different, indicating that stomatal patterning did not affect the relationship between Anatomical gsmax and intrinsic WUE. However, the intercept for LoC plants was significantly larger than that of HiC, demonstrating that lines without clustering held a quantitative intrinsic WUE advantage across the entire range of Anatomical gsmax (Fig. 8a: = 0.952 and < 0.05, respectively; ANCOVA). Furthermore, it should be noted that the negative relationship between intrinsic WUE and Anatomical gsmax does provide empirical evidence that plants with fewer stomata should have improved photosynthetic efficiency, per unit of water transpired, than plants with high-density stomata.

Figure 8.

Water-use efficiency (WUE) was not strongly affected by stomatal patterning. Experimental values for intrinsic WUE (a) and instantaneous WUE (b) were determined by leaf-level gas-exchange measurements of individual rosette leaves of mature Arabidopsis thaliana plants at 350 ppm CO2. Intrinsic WUE had a negative correlation with maximum stomatal conductance (Anatomical gsmax) for both patterning groups (= 50; < 0.01; R2 = 0.14 for Low Clustering (LoC); open circles; = 41; < 0.001; R2 = 0.24 for High Clustering (HiC); closed circles), while instantaneous WUE was only correlated with Anatomical gsmax for HiC plants (= 0.987 for LoC; < 0.05; R2 = 0.085 for HiC). (a) The slopes of the regression models were not significantly different but the y-intercepts were (= 0.962 and < 0.05, respectively).

True advantages in WUE, however, are dependent on prevailing climatic conditions. Instantaneous WUE denotes the ratio of net carbon assimilation to actual transpiration rates (A/E) and differs from intrinsic WUE because it does consider the evaporative demand at the leaf surface. Instantaneous WUE is a valuable metric to quantify the carbon–water balance of a plant over short time-scales (Seibt et al., 2008). At 350 ppm CO2, transpiration rates were not significantly different between patterning groups, both of which had a positive logarithmic correlation with Anatomical gsmax (Fig. S1: < 0.001; R2 = 0.51 for LoC; < 0.01; R2 = 0.23 for HiC). Only HiC plants had a correlation between Anatomical gsmax and instantaneous WUE (Fig. 8b: = 0.889 for LoC; < 0.05; R2 = 0.085 for HiC), but this difference between patterning groups was driven by low carbon assimilation rates for HiC plants with high Anatomical gsmax (Fig. 7b). With that in perspective, it seems that Anatomical gsmax plays a minimal role in WUE under favorable external conditions, regardless of stomatal patterning. Essentially, increases in Anatomical gsmax produced correspondingly higher rates of transpiration and carbon assimilation, resulting in consistent instantaneous WUE across wide ranges in Anatomical gsmax; this relationship was especially true for plants with LoC. Nonetheless, the effects of Anatomical gsmax on instantaneous WUE might change under more challenging external conditions, such as drought or low RH, as evidence from intrinsic WUE suggests.


Our gas-exchange measurements highlight the physiological importance of the developmental programs that control stomatal patterning in A. thaliana. Optimal stomatal function was dependent on the separation of mature stomata by at least one epidermal cell and plasticity in stomatal development had physiological benefits for the plant. The significance of proper stomatal spacing was evident in at least three aspects of stomatal physiology, as observed in lines of HiC: maximum stomatal opening, as quantified by Diffusive gsmax, was reduced; stomatal responses to external signals were impaired; and carbon assimilation was reduced, especially at high Anatomical gsmax.

The reduction in maximum stomatal conductance was most apparent when comparing Anatomical gsmax to Diffusive gsmax for the two patterning groups (Fig. 3). This analysis demonstrated that plants with significant stomatal clustering were impaired from completely opening their stomatal pores, as predicted by Anatomical gsmax. This was probably the result of mechanical failure in the guard cells, which could have arisen from a variety of sources. First, increases in guard cell turgor are dependent on the efflux of protons and subsequent influx of rectifying K+ and Cl ions that produce an osmotic flow into the guard cell (Kim et al., 2010). This process is dependent on the exterior of the guard cell being in contact with epidermal cells that provide the necessary ion pools, specifically K+, and water resources to facilitate changes in guard cell osmolarity and turgor pressure (Outlaw, 1983). Secondly, adjacent guard cells would create opposing forces as they increase in turgor pressure and this resistance to guard cell movement could prevent maximum pore opening for one or both of the stomata (Franks & Farquhar, 2007). This scenario has similarities to the ‘mechanical advantage’ that epidermal cells possess over guard cells, which generally means that equal increases in guard cell and epidermal turgor pressure close the stomatal pore (Franks et al., 1998). Finally, development of adjacent guard cells might disrupt the signaling and cell biological mechanisms that determine the organization of the cytoskeleton or structure of the cell wall and cellulose microfibrils. These cellular components impart differential mechanical properties that are required for guard cells to bend away from each other during increases in turgor.

An alternative hypothesis to mechanical failure in the guard cells of clustered stomata is the hydraulic limitations of the leaf. Recent work has nicely demonstrated the correlation between leaf water supply, photosynthetic capacity, and stomatal conductance (Brodribb et al., 2005, 2007). If the hydraulic supply were incapable of providing enough water to multiple stomata in very close proximity, then guard cell turgor pressure would be insufficient to completely open the stomatal aperture.

Irrespective of the exact mechanism that causes guard cell malfunction, we wanted to quantify the extent to which clustering impaired stomatal conductance. Approximations of Anatomical gsmax, based on the functionality of stomata in clusters, provided evidence that one cluster had approximately the same gas-exchange potential as one properly spaced stoma (Fig. 4). This result is consistent with a hydraulic limitations model, which predicts that the stomatal conductance from a cluster of stomata would be equivalent to that from a single stoma, assuming that the mutations causing stomatal clustering did not affect venation density and water supply (Brodribb & Jordan, 2011). This result might also be explained by considering the consequences of diffusion through adjacent stomata. During transpiration, diffusional shells are created at the exterior of the stoma pore, and this shell represents a multidimensional concentration gradient of water vapor that enables diffusion to occur (Willmer & Fricker, 1996). Overlap between diffusion shells, principally the result of stomatal proximity, decreases the average concentration gradient at the leaf surface and reduces transpiration rates (Ting & Loomis, 1963). Therefore, the rate of gaseous diffusion through adjacent stomata should already be reduced. This phenomenon alone could account for some of the decreased rates in Diffusive gsmax for lines with stomatal clustering, regardless of any functional deficiencies in guard cells or water supply issues.

Both the hydraulic and diffusion shell models, however, present dilemmas. If stomatal activity is intended to reduce water loss, as emphasized by the diffusion shell explanation, why aren't all stomata in clusters? From our other findings, the apparent answer is that stomatal clustering also impairs the dynamic responses of stomata (Fig. 5) and carbon assimilation potential (Fig. 7). The hydraulic model makes an assumption that there is a break in the correlation between vein density and stomatal density in HiC plants, which highlights a need for developmental coordination between leaf tissues and the malfunction of an underlying mechanism that connects stomatal and leaf vein development.

A more derived consequence of altering stomatal patterning was the potential impact on photosynthesis, as the empirical correlation between stomatal conductance and carbon assimilation has been a long-standing observation (Wong et al., 1979). Indeed, Anatomical gsmax was correlated with carbon assimilation, but stomatal clustering prevented such a relationship, especially at high stomatal density (Fig. 7b). One explanation for the influence of patterning could be the positioning of stomata relative to the interior tissues. Intuitively, stomata should be located in areas overlying the open spaces of the mesophyll to ensure a suitable path for gaseous diffusion into and out of the leaf (Pesacreta & Hasenstein, 1999). The improper positioning of stomata, which leads to clustering in the first place, could compromise this CO2 pathway, thus preventing an even distribution of CO2 through the mesophyll and decreasing the photosynthetic potential of the leaf. Extracellular and cross-tissue communication via known peptide ligands, such as STOMAGEN, and received by TMM or ERECTA family receptors, could potentially provide the spatial information to position stomata away from underlying cells (Sugano et al., 2009). Extracellular communication factors could also play a role in coordinating the developmental proliferation of independent tissue layers. As mentioned earlier, the relationship between stomata, vein density or pattern, and photosynthetic capacity might justify these observed differences in carbon assimilation between LoC and HiC plants.

Environmental responses are also an integral part of stomatal function, whereby guard cells must perceive and adjust the stomatal pore in accordance with a multitude of external signals. Stomatal clustering impaired these activities, delaying the appropriate closure of the pore in response to increasing [CO2] (Fig. 5). As previously described, loss of the epidermal mechanical advantage might provide an explanation for this lack of responsiveness: when guard cells lose turgor, physical pressure from adjacent epidermal cells is largely responsible for closing the stomatal pore (Franks et al., 1998; Franks & Farquhar, 2007). Contact of adjacent stomata prevented that mechanical force from being applied and probably resulted in stomata closing at a slower rate (initial slope of the response curves in Fig. 5a) and to a quantitatively lesser extent (Fig. 5c).

The recent expansion of the gene regulatory network controlling A. thaliana stomatal development revealed a preponderance of factors that negatively regulate stomatal production. Low-density stomatal patterning seems to be the default program for plants, even though we have demonstrated the benefits of high numbers of stomata, such as the correlation with increased carbon assimilation (Fig. 7), and other reports have shown that smaller stomata – another consequence of a high density of stomata (Fig. 2) – respond more quickly to environmental cues (Drake et al., 2013). Despite such tangible benefits of producing more stomata, there seems to be a biological cost associated with such proliferation, as indicated by the genetic investment in high numbers of stomatal repressors. That most plants do not possess high stomatal densities could be a result of the overriding evolutionary pressure of needing to prevent water loss, which does favor low stomatal density (Fig. 8). However, our results could also be explained in terms of developmental constraint; higher stomatal densities lead to a higher chance of contact between neighboring stomata, and this contact has considerably negative consequences. Here, the physiological cost of mispatterning stomata would drive the expansion of genes and genetic networks involved in the relative positioning of stomata. Considering these limitations together, there seems to exist a parallel need to optimize stomatal production while safeguarding against stomatal clustering, where developmental flexibility is influenced by local conditions but ultimately constrained by the patterning regimes that maintain guard cell function.


This work was supported by a Stanford University Bio-X Interdisciplinary Fellowship awarded to G.J.D. and funding from the Carnegie Institution for Science and the Gordon and Betty Moore Foundation. D.C.B. is an investigator of the Howard Hughes Medical Institute. We thank Larry Giles, Roland Pieruschka, and Jen Johnson for technical support with gas-exchange equipment and experiments, and Leila Moinpour for assistance in performing gas-exchange experiments and stomatal quantification.