Journal of Plant Physiology 234–235 (2019) 138–144

Contents lists available at ScienceDirect

Journal of Plant Physiology

journal homepage: www.elsevier.com/locate/jplph

A naturally optimized mass transfer process: The stomatal transpiration of T

plant leaves
⁎
Kai Xu, Liang Guo, Hong Ye
Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, 230027, People’s Republic of China

A R T I C LE I N FO A B S T R A C T

Keywords: Stomatal transpiration of leaves is a dominant pathway of plant physiological water loss. The leaf transpiration
Transpiration rate when stomata are fully open is commonly at the same level as the evaporation rate of a wet surface of the
Stomata distribution same area as that of the leaf area, although the cumulative area of the stomatal pores is typically less than 3% of
Extremal diffusion flux the leaf area. To elucidate the highly efficient diffusion of the stomatal array from the perspective of mass
Natural optimization
transfer theory, stomatal distribution characteristics of various kinds of leaves were obtained with optical mi-
croscope, and steady diffusions of water vapor from isolated zero-depth circular stomata, elliptical stomata, and
distributed stomatal arrays without airflow parallel to the surface were simulated with the finite element
method. It was found that the long perimeter of the elliptical stomata and the specific distribution characteristics
of the stomatal array are the dominant reasons for the highly efficient diffusion of the stomatal array on the
leaves. Furthermore, the simulation results reveal that extremal transpiration rates exist for the stomatal arrays
with different distribution characteristics. It was found that the transpiration rates of the vegetation tend to
approach the extremal values for flourishing development in the process of natural optimization.

1. Introduction stomatal areal density is approximately in the range of 50 to 300 /mm2,

Transpiration, the process by which water is transported through
tracheids or vessels and eventually lost into the surrounding atmo-
sphere as water vapor, plays a key role in the physiology of vascular
plants (Pieruschka et al., 2010), including several species of ferns as
well as all species of gymnosperms and angiosperms (Brodribb and
McAdam Scott, 2011). For broad-leaved vascular plants, such as dicots
and monocots, more than 90% of the water escapes from leaves
(Hopkins, 1999). Therefore, the transpiration process is closely related
to the anatomical structure of the leaves. Fig. 1 shows the schematic of
the cross section of a typical dicot leaf. As can be seen, the top and
bottom surfaces of the leaf are epidermis cell layers covered with waxy
cuticles. Stomata surrounded by paired guard cells are distributed on
the bottom surface. The guard cells increase or decrease in volume in
response to external and internal stimuli, and the resulting changes in
guard cell shape adjust stomatal aperture and thereby affect the flux of
gases between the leaf internal environment and the bulk atmosphere
(Lawson and Blatt, 2014). Generally, stomata exist on both the top
(adaxial) and bottom (abaxial) leaf surfaces or only on one surface,
typically the abaxial one as shown in Fig. 1. The length, width, and
depth (namely the thickness of the guard cells) of the spindly stomatal
pores are commonly in the order of micrometers. Considering that the
the cumulative area of stomatal pores when stomata are fully open
accounts for merely 0.2 to 2% of the leaf area (Nobel, 1999). Between
the epidermis layers of the dicot leaf, there are loosely arranged me-
sophyll cells generally classified into palisade and spongy tissues, and
an interconnected system of air space is formed inside the spongy tis-
sues. When the stomata are open, the air space is connected to the
ambient environment. Liquid water flowing inside the veins evaporates
into the air space, and the water vapor in the air space diffuses into the
ambient environment through the stomata or cuticles (Cannon et al.,
1979; Prado et al., 2013; Sack and Holbrook, 2006). The water vapor
transpiration rates of leaves from several kinds of angiosperms mea-
sured by porometer or portable photosynthesis system are in the range
of 2 to 10 mmol/(m2·s) (Pariyar et al., 2013; Pieruschka et al., 2010), of
which more than 90% of the water diffuses through stomata, indicating
that the stomata are the dominant passages of water vapor diffusion of
most plants leaves. It is noteworthy that the transpiration rates of the
leaves when stomata are fully open are approximately at the same level
as the evaporation rate of a wet surface of the same area as that of the
leaf (Willmer and Fricker, 1996), although the total area of the stomatal
pores is merely approximately 1% of the leaf area. The water vapor
transpiration of the natural leaves is a typical process of mass transfer;
however, a reasonable explanation for the characteristics of highly

⁎
Corresponding author.
E-mail address: hye@ustc.edu.cn (H. Ye).

https://doi.org/10.1016/j.jplph.2019.02.004
Received 11 November 2018; Received in revised form 9 February 2019; Accepted 9 February 2019
Available online 15 February 2019
0176-1617/ © 2019 Elsevier GmbH. All rights reserved.
K. Xu, et al.
Fig. 1. The schematic of the cross section of a typical dicot leaf.
efficient diffusion of the stomatal array is still unavailable from the
perspective of mass transfer theory.
The principal process of transpiration is the transport of water vapor
driven by the concentration gradient and airflow; thus, the gaseous
diffusion theory and the boundary layer effects should be the main
mechanisms. Stefan investigated a steady diffusion process of an iso-
lated circular water surface on an infinite impervious plane without
airflow based on the gaseous diffusion theory (Stefan, 1882), and a so-
called diameter law was summarized, stating that the vapor diffusion
rate is proportional to the diameter of the water surface rather than its
area. Troyer analyzed the case of gas diffusion from isolated circular
stomata through a boundary layer of a finite thickness into the atmo-
sphere, with the gas assumed to be at constant concentration in the
cross section of the stomata (Troyer, 1980). The results indicated that
treating the stomatal diffusion process as steady diffusion without air-
flow would involve no significant error if the boundary layer is 40 times
thicker than the stomatal radius. It is noteworthy that a boundary layer
thickness of 400 μm is smaller than most of the values calculated by
Nobel for a number of natural wind velocities and leaf dimensions
(Nobel, 1970), while the size of stomata is generally around 10 μm.
Therefore, employing the diameter law based on the gaseous diffusion
theory to investigate the diffusion phenomenon of the isolated circular
stomata is appropriate in the cases of realistic boundary layer and
stomatal dimensions. However, on the surfaces of the natural leaves,
the stomata are commonly spindly, which can be approximated as el-
lipses. Sayre found that the diffusion rate from an elliptical surface is
proportional to the semi-axes of the surface (Sayre, 1926). In order to
correct the diffusion rate, Cooke defined an eccentricity factor, which
needs to be determined by experiments (Cooke, 1967).
Different from the hypothetical isolated circular or elliptical sto-
mata, the stomata on the surface of natural leaves are commonly dis-
tributed discretely as a stomatal array. Based on the investigations of
the diffusion process in the vicinity of the surface of soil pore arrays, E.
Haghighi (Haghighi et al., 2013) suggested that the transport of the
water vapor in the vicinity region of the surface can be assumed to be
only governed by gaseous diffusion theory. From the perspective of
dynamic response, the regulatory mechanisms of stomata and soil pores
are quite different, while from the perspective of mass transfer theory
the structures of the discrete water vapor sources are similar and the
diffusion processes are both driven by the concentration gradient of
water vapor. Thus, the suggestion by E. Hagaighi is also appropriate for
the transport of the water vapor in the vicinity region of the leaf sur-
face. Moreover, interferences among the distributed stomata on the leaf
surface limit the space for the vapor diffusion from a single stoma
(Pickard, 1981), leading to a lower diffusion rate of the stomatal array
than the cumulative diffusion rate of the respective isolated stomata.
The diffusion experiments of the distributed stomata indicated that the
degree of interferences are associated with the stomatal sizes and in-
tervals between adjacent stomata (Ting and Loomis, 1963, 1965). A
distributed factor was employed to correct the diffusion rates (Bange,
1953; Cooke, 1967; Ting and Loomis, 1965), but the expression and
meaning of the factor are still unclear.
139
Journal of Plant Physiology 234–235 (2019) 138–144
In recent years, the finite element method was employed to in-
vestigate the stomatal transpiration from the perspective of mass
transfer theory. Defraeye simulated the process of stomatal transpira-
tion from a series of leaf surfaces with zero-depth stomata in a two-
dimensional domain (Defraeye et al., 2013). The results indicated that
the transpiration rate from the leaf surface was strongly dependent on
surface coverage and source size, providing insights into the dominant
geometrical parameters effecting the transpiration process of the sto-
matal array. In addition, considering that the actual stomata on the
natural leaves commonly have interior structures, Roth-Nebelsick in-
vestigated the influence of stomatal architecture on stomatal con-
ductance and on gas concentration gradient by simulating the diffusion
rates of water vapor and CO2 occurring in a set of three-dimensional
isolated stomata models with the finite element method (Roth-
Nebelsick, 2007). The results showed that the structures of the isolated
stomata profoundly influence the gas concentration in the vicinity of
the stomatal pores. Nevertheless, the impact of stomatal arrays on the
gas diffusion process of natural leaves was not considered in this study.
Roth-Nebelsick also explored the effects of stomata crypts on the
transpiration process with the three-dimensional finite element models
of encrypted stomata (Roth-Nebelsick et al., 2009). The results in-
dicated that in the leaves with open or partially closed stomata, crypts
reduce transpiration by less than 15% compared with the none-
ncrypted, superficially positioned stomata. Lehmann investigated the
effects of stomata clustering on the process of leaf gas exchange with
the numerical models involved with the stomata conductance of the
stomatal pore and substomatal cavity (Lehmann and Or, 2015). The
results indicated that stomata clustering increases the interactions
among concentration shells of the adjacent stomata, resulting in larger
diffusive resistance that may reduce fluxes by 5–15%.
From the perspective of mass transfer theory, the stomatal dis-
tribution characteristics should be the dominant factor affecting the
transpiration process of the natural leaves because the stomata are the
diffusion sources. However, the relationship between the highly effi-
cient diffusion characteristic of the natural leaves and the stomata
distribution pattern is still unclear. In this study, the stomatal dis-
tribution characteristics of various kinds of plant leaves were obtained
with optical microscope, and the steady diffusions from isolated zero-
depth sources and distributed stomatal arrays without airflow parallel
to the surface were simulated with finite element method in a three-
dimensional domain based on the stomatal distribution characteristics
of the natural leaves.
2. Methods and models
2.1. Measurements of stomatal distribution characteristics
To investigate the differences in stomatal distribution character-
istics among various kinds of plants, typical plants grown in the Yangtze
Plain of China were selected to provide healthy leaves as test samples,
including four kinds of deciduous plants (Sapium sebiferum, Bischofia,
Lotus, and Crape myrtle), five kinds of evergreen plants
(Chlorophytum, Mimosa, Camphor, Hedera, and Osmanthus), two
kinds of annual crops (Maize and Soybean), and one kind of gymnos-
perm (Ginkgo). The surface topographies at the central zones of the test
leaves were copied with cyanoacrylates on one surface of a microslide
with rubbing technique. It is noteworthy that all the test leaves were
picked at noon when the stomata were open entirely, and the rubbing
procedure was conducted immediately after the leaves were picked to
avoid the closing of the stomata. The stomatal imprint on the micro-
slides were recorded as images with an optical microscope. The para-
meters of shape and distribution of hundreds of stomata in the images
of each kinds of leaves were obtained with the Image-Pro software, and
the mean values were calculated as the stomatal distribution char-
acteristics.
K. Xu, et al. Journal of Plant Physiology 234–235 (2019) 138–144

2.2. Theoretical model of an isolated circular stoma

When water vapor diffuses steadily from an isolated zero-depth

circular source into a half space without airflow parallel to the surface,
the bulk behavior of water vapor follows Fick diffusion law:

⎧ J = − D ∇C
dC
⎨ dt = −∇J = 0 (1)
⎩
where J is the diffusion flux, C is the concentration of water vapor, D is
the binary diffusion coefficient of water vapor, and t is the time. Inside a
leaf, the mesophyll cells next to the stomata are the evaporation sites of
water vapor, and the water vapor concentration in the air space of the
mesophyll tissue is approximately saturated (Farquhar and Raschke,
1978; Roth-Nebelsick, 2007). The diffusion through the stomatal tube
can be considered one-dimensional (Parkhurst, 1994; Waggoner and
Zelitch, 1965), i.e., the concentration at every cross section of the sto-
matal tube is approximately constant. Thus, the stomatal transpiration
could be assumed as water vapor diffusion from a zero-depth stomata,
which is consistent with the assumption of the diameter law (Lee and
Gates, 1964). Assuming that the concentrations of water vapor are
equal to C0 in the circular source region and 0 at an infinite point, the
boundary conditions can be given as

⎧C = C0, 0 ≤ r < h, z = 0
⎪ ∂C
= 0, r > h, z = 0
⎨ ∂z
⎪C (r , z ) → 0, when(r 2 + z 2) → ∞ (2)
⎩
where h is the radius of the circular region, and r and z are the radial
and axial coordinates, respectively. According to equation (2), the
concentration gradient of water vapor at the circular region is
∂C 2C
lim (r , z ) = − 0 (h2 − r 2)−1/2 , 0 < r < h
z→0+ ∂z π (3)

The diffusion rate can be achieved by integrating the flux J in the

entire circular region:

Q= ∫ J ⋅dA = ∫ − D ∂∂Cz (r , 0)⋅dA = 4C0 Dh

s s (4)

where S represents the area of the circular region. Therefore, the dif-
fusion rate Q is proportional to the concentration difference between
the source region and the infinite point. When the concentrations at the
source and the infinite point are C0 and C∞ respectively, equation (4)
can be converted into
Q = 4(C0 − C∞ ) Dh (5)

which indicates that Q is proportional to the radius of the circular

source.

2.3. Computational models for isolated stomata and stomatal array

The theoretical analysis in section 2.2 is limited to the special case

of isolated circular source, and further investigations should be con-
ducted on the isolated elliptic source and distributed elliptic stomatal
array. Fig. 2a, b, and c are the computational models of an isolated
circular stoma, an isolated elliptical stoma, and a spatial repeat unit of a
stomatal array, respectively. In the isolated stomata models (Fig. 2a and
b), the diffusion source was circular or elliptic, located at the center of
the bottom of a hemispherical computational domain. The radius (h) of
the circular source or semi-axes (a and b) of the elliptic source were set
within tens of micrometers based on the dimensions of the stomata on
the leaves obtained from section 2.1, and the radius (R) of the hemi-
spherical computational domain was set as 1000 μm, which was de-
monstrated to be large enough with little effect on the diffusion flux
from the stomatal source after a series of preliminary simulations. The
maximum element size of free meshes in the diffusion space was set as
Fig. 2. The computational model of (a) isolated circular stomata, (b) isolated
elliptical stomata, and (c) stomatal array. In the stomatal array model, a spatial
periodic unit is employed and symmetric boundary conditions are imposed on
the symmetric surfaces.
10 μm, and the elements at the diffusion source were approximately
several microns or less.
In the stomatal array model (Fig. 2c), the diffusion source was el-
liptic and located at the center of the bottom square of a rectangular
cuboid diffusion space. The side faces of the space were imposed with
symmetric boundary condition based on the spatial repetition feature of
the stomatal array. The height (H) of the cuboid space was set as

140
K. Xu, et al.
1000 μm for the same reason as that for the isolated stomata model. The
interval (i) of the array, namely the side length of the bottom square of
the array unit or the distance between the adjacent stomata, was tens to
hundreds of micrometers based on the results obtained from section
2.1. It should be noted that the stomatal array in the computational
model was designed as a regular matric-like distribution pattern, which
is a simplification of the stomatal distribution patterns of the natural
leaves as shown in section 3.1. Furthermore, the zero-depth sources
were adopted in these models not because of the negligence of the in-
ternal structure of the stomata and the gradient in water vapor con-
centration that develops between the sites of evaporation and the sur-
face of the leaf, but because the effect of the stomatal distribution
characteristics on the diffusion process was the focus of this work. The
stomata provide a boundary condition of constant water vapor con-
centration as shown in Fig. 2.
In both of the models, the temperatures in the computational do-
mains were set as 27°C, which is a common temperature level in the
natural environments. The water vapor at the diffusion sources were
assumed to be saturated (C0 = 1.42 mol/m3), and the relative air hu-
midity of the atmosphere was set as 70% (C∞ = 0.994 mol/m3), which
is also a common level during daytime in the natural environments. The
binary diffusion coefficient was set under one atmospheric pressure
(D = 2.6 × 10−5 m2/s). The finite element method was adopted to
simulate the diffusion processes of the models.
3. Results and discussion
3.1. Stomatal distribution characteristics of plant leaves
Fig. 3 exhibits the surface topographies of seven kinds of leaves
obtained in section 2.1 as examples. As can be seen, the stomata are
commonly spindly and discretely distributed on the leaf surfaces. The
size of the stomata and the distance between adjacent stomata represent
the distribution characteristics of the stomatal array on the leaves.
Table 1 shows the mean values of the measured stomatal length (l),
stomatal width (w), and distance between adjacent stomata (d) of the
stomatal array images obtained from section 2.1. It can be seen that l
distributes in the range from 9.3 to 39. 3 μm, w in the range from 2.2 to
9.1 μm, and d in the range from 38.9 to 121.1 μm. It should be noted
that the direction distribution of the major axes of the spindly stomata
is not considered in this study, and major axes of the vapor sources are
all simplified to be in the same direction in the computational model,
whereas the direction of the stomata is randomly distributed except for
the Chlorophytum and Maize as shown in Fig. 3. Moreover, adjacent
stomatal rows can be observed on the adaxial and abaxial surfaces of
Maize, exhibiting a unique structure of stomatal distribution. Table 1
also shows the characteristic ratios of the stomatal arrays, i.e., the ratio
of width to length (w/l) and the ratio of distance to length (d/l). It can
be seen that w/l distributes in the range from 0.1 to 0.6, and d/l in the
range from 2.6 to 9.4. It is noteworthy that the d/l of Ginkgo is 9.40,
indicating that the stomatal density of the Ginkgo leaves is much
smaller than that of the other kinds of leaves. It is also noteworthy that
the stomata of the leaves were measured on the abaxial side except for
in Lotus because it is an aquatic plant and the stomata are merely
distributed on the adaxial side. The stomata on the adaxial sides of
Maize and Mimosa leaves were also measured to compare with the
stomata on the opposite sides. For monocots such as Maize, the mean
size and distance between the stomata on both sides of the leaves were
similar, but for dicots such as Mimosa, the distance between stomata on
the adaxial side was much larger than that on the abaxial side.
3.2. Diffusion from an isolated stoma
The diffusion fluxes of the isolated stomatal model were simulated
to investigate the relationship between the diffusion rate and the size of
the vapor sources. The value range of the source size was determined
141
Journal of Plant Physiology 234–235 (2019) 138–144
based on the results exhibited in Table 1, and the ratio of area (A) to
perimeter (P) of the vapor sources is considered as a geometrical
parameter establishing the relationship between A and P. The diffusion
rate (mol/s) is the integral of the diffusion flux (mol/m2⋅s) in the entire
vapor source. Fig. 4 represents the relationship between the diffusion
rate and A/P of the vapor sources. In this figure, eccentricity (e) re-
presents the ratio of the distance between the two focuses and the major
axis length of the elliptical sources. As can be seen, in the special case of
circular stomata (e = 0, A/ P = r /2) , the diffusion rate is proportional to
the radius of the diffusion sources, which is consistent with the theo-
retical result of equation (5), indicating that the radius can be regarded
as the characteristic length of the diffusion process of an isolated cir-
cular source.
When e is larger than zero, the diffusion sources are ellipse-shaped
and similar to the shape of the stomata on the natural leaves. As can be
seen from Fig. 4, for a series of elliptical sources with the same e, the
diffusion rates are proportional to A/P, indicating that A/P is appro-
priate to be taken as the characteristic length of the ellipse-shaped
diffusion sources. It also can be seen that the diffusion rates of the el-
liptical sources rise with the increasing e and are larger than that of
circular sources (e = 0) at the same A/P. It should be noted that the
area of the biological internal environment exposed to the surrounding
air for nutrient acquisition is commonly extremely limited for self-
protection, and the sum of all the opened stomatal area in daytime
accounts for merely 0.2 to 2% of the total area of the leaf surface; thus,
the area of the stomata should be utilized efficiently for water vapor
diffusion or CO2 uptake. Assuming a series of isolated vapor sources
with constant areas of 200, 400, and 600μm2 , respectively, Fig. 4 also
shows the diffusion rate of the vapor sources with the constant area and
different perimeters. As can be seen, the diffusion rate of the sources
increases with the decrease of A/P, indicating that the elliptical sources
with longer perimeter are more appropriate than the circular source for
transpiration.
Compared with the circular source, the remarkable geometrical
characteristic of the elliptic source is the longer perimeter; thus, the
diffusion characteristics of the marginal region of the vapor source
should be further investigated. Assuming a circular diffusion source
with a radius of 5 μm as an example, Fig. 5 shows the component of
diffusion flux in the diffusion region and the surrounding impermeable
region. As can be seen, in the diffusion region, there is no radial com-
ponent of diffusion flux but only axial component. The axial component
of diffusion flux increases along the radial direction and sharply in-
creases near the edge of the diffusion region. It also can be seen that in
the surrounding impermeable region, there is no axial component but
only a radial component sharply increasing while approaching to the
edge of the diffusion region. This phenomenon is the so-called edge
diffusion effect and attributed to the assumptions of the constant con-
centration in the diffusion region and the impermeable surface in the
remaining plane region. These assumptions are appropriate because the
stomata are commonly micron-scale with approximately constant sa-
turated water vapor in the stomatal tube and because the cuticle on the
leaf surface is impermeable. Thus, the diffusion rates of the spindly
stomata on the natural leaves are mainly dominated by the long peri-
meter of the stomata because of the edge diffusion effect.
3.3. Diffusion from distributed elliptical stomata
As the spindly stomata distribute on the leaf surfaces discretely
(Fig. 3), the relationship between the diffusion rates and the interaction
of the adjacent stomata should be further investigated by simulating the
diffusion process of the elliptical diffusion source array introduced in
section 2.3. The length of the major axis (2a) and minor axis (2b) of the
elliptical sources correspond to the average length (l) and width (w) of
the stomata, respectively, and the interval (i) of the sources in the
models correspond to the average distance (d) between the adjacent
stomata. The ratios of semi-minor axis to semi-major axis (b/a) of the
K. Xu, et al.
Fig. 3. Epidermis optical micrographs of seven kinds of the leaves as examples: (a) Hedera, (b) Camphor, (c) Mimosa, (d) Lotus, (e) Chlorophytum, (f) Gingko, (g)
Maize (adaxial surface), (h) Maize (abaxial surface).
ellipse array models were determined based on the ratios of width to
length (w/l) of the stomata shown in Table 1 (in the range from 0.1 to
0.6). It should be noted that a series of preliminary simulations in-
dicated that changes in b have a slight effect on the average diffusion
flux through the distributed elliptical sources because b was determined
to have a lower effect on the perimeter of the elliptical sources than a
when b/a was in the range from 0.1 to 0.6. Therefore, the value of b/a
was set as a constant of 0.5, and the influence of a on the average
diffusion flux of the distributed elliptical sources was investigated in the
following study.
142
Journal of Plant Physiology 234–235 (2019) 138–144
Different from the discussion on the diffusion from the isolated
sources, we focused on the average diffusion flux through the plane
distributed with elliptical stomata with different major axes and in-
terval. Here, the definition of average diffusion flux is similar to that of
the transpiration rate of the natural leaves. It is noteworthy that the
ratio of d/l was employed to describe the distribution characteristics of
the stomatal array instead of stomatal density, because d/l could es-
tablish the relationship between the size of the stomatal array and the
size of the isolated stomata. The parameter of i/2a was employed for
the same reason. Fig. 6a shows the average diffusion fluxes of the
K. Xu, et al. Journal of Plant Physiology 234–235 (2019) 138–144

Table 1
Measured stomatal parameters of the plant leaves.
leaf phenology species side stomatal parameters (μm) w/l d/l
l w d

angiosperms annual Maize adaxial* 39.24 3.97 121.10 0.10 3.09

abaxial 37.39 3.85 115.13 0.10 3.08
Soybean abaxial 18.48 3.64 82.32 0.20 4.45
evergreen Chlorophytum abaxial 21.36 9.05 109.20 0.42 5.11
Mimosa adaxial* 17.83 7.06 70.67 0.40 3.96
abaxial 18.49 6.83 49.43 0.37 2.67
Camphor abaxial 16.39 6.23 67.37 0.38 4.11
Osmanthus abaxial 9.32 5.32 43.86 0.57 4.71
Hedera abaxial 14.52 8.35 80.56 0.58 5.55
deciduous Crape myrtle abaxial 11.59 4.05 41.21 0.35 3.56
Bischofia abaxial 16.29 2.24 54.40 0.14 3.34
Sapium sebiferum abaxial 20.81 3.76 62.69 0.18 3.01
Lotus adaxial* 9.46 3.28 38.94 0.35 4.12
gymnosperm deciduous Ginkgo abaxial 12.49 4.91 117.39 0.39 9.40

Fig. 4. The relationships between the diffusion rates and the ratios of area to
perimeter of the elliptical sources. Eccentricity (e) represents the ratio of focal
length and major axis length of the elliptical sources. A and P represent the area
and perimeter of the elliptical sources, respectively.

Fig. 6. (a) The relationships between the average diffusion fluxes and ratios of
interval to major axis (i/2a) for distributed stomata with different major axis
(2a). (b) The extremal flux line connects the five extremal points corresponding
to (a), and the scattered points represent the stomatal distribution character-
istics of the various kinds of plant leaves. In this figure, i and 2a are equal to d
and l, respectively. The position of the point labeled with Arabidopsis was
determined with the microscopic images of stomata on the hypocotyls obtained
from the results of Wang (Wang et al., 2015).

Fig. 5. The components of the diffusion flux in and near the diffusion source
with a radius of 5 μm. The blue zone represents the axial component of the
diffusion flux, and the red zone represents the radial component (For inter-
pretation of the references to colour in this figure legend, the reader is referred
to the web version of this article).
elliptical array models with different 2a. As can be seen, the average
diffusion flux approaches an extremum when the elliptical array has a
specific i/2a, and the specific i/2a decreases with increasing 2a. Dif-
ferent from isolated stomata, the diffusion process of a stomata in the
stomatal array would affect the water vapor concentration field near
the adjacent stomata, resulting in a decreased diffusion flux of the ad-
jacent stomata. Thus, if i/2a is too small, which means that the stomata
are too crowded, the average diffusion flux would weaken with

143
K. Xu, et al.
decreasing i/2a as shown in Fig. 6a. Furthermore, if i/2a is too large,
which means that the stomata are too sparse, the average diffusion flux
would also weaken with increasing i/2a as shown in Fig. 6a. Thus, there
is an extremum when the elliptical array has a specific i/2a. It should be
noted that all the extrema of the average diffusion fluxes are close to
7.4mmol/m2⋅s , which is of the same level as the measured transpiration
rates of the natural leaves (in the range from 2 to 10 mmol/(m2·s)
(Pariyar, et al., 2013; Pieruschka, et al., 2010)), indicating the relia-
bility of the calculated results of the stomatal array models. Moreover,
it is noteworthy that the specific values of the extremum would change
with the constant temperature and relative air humidity set in the
computational models introduced in section 2.3.
The coordinates of the five extremum points shown in Fig. 6a were
mapped in Fig. 6b by the transition of i/2a to d/l, and the corresponding
five points in Fig. 6b could be connected to form an extremal flux line
which exhibits the optimal stomatal distribution characteristics for the
maximal average diffusion fluxes. The stomatal distribution parameters
(l and d/l) of the natural leaves shown in Table 1 are also drawn in
Fig. 6b. As can be seen, almost all of the points labeled with the plant
names are near the extremal flux line, indicating that the distribution
characteristics of the stomata on the leaf surfaces may tend to the ex-
tremal flux line rather than irregular distribution. Nevertheless, the
point labeled with Ginkgo is farthest away from the extremal flux line,
as illustrated in Fig. 6b. The d/l of the ginkgo is very large, which means
that the stomata on the leaf surface are too sparse or too small, resulting
in the lower average diffusion flux than the extremum. It is noteworthy
that Ginkgo is a relict gymnosperm plant thriving in the Mesozoic era,
and a high transpiration rate was not adaptive to the cold and dry
climate in the late Mesozoic era. Therefore, Ginkgo may have survived
due to the sparse distribution pattern of its stomata and low tran-
spiration rate.
Various angiosperms have flourished for more than 100 million
years after gymnosperms. It can be supposed that the stomatal dis-
tribution patterns tend to maximize the transpiration rate to match the
more vigorous physiological activities of the plants and expand the
ecophysiological niche space during the natural optimization of the
vegetation (McElwain et al., 2016). It is noteworthy that for the annual
crops such as Maize and Soybean, the corresponding stomatal para-
meter points are very close to the extremal flux line. The crops have
been domesticated by humans for a long period of time to produce the
economic benefit; thus, the crop seeds with the highest assimilation
rates as well as highest transpiration rates tend to be filtered out for
larger-scale planting. Therefore, the optimized stomatal distribution
patterns and the almost maximal transpiration rates of the crops may be
the results of natural optimization and artificial selection.
Besides the plant leaves, stomata also distribute on the hypocotyls.
Wang investigated the influence of Brassinosteroid (BR) signals on the
stomatal production in the hypocotyls of Arabidopsis (Wang et al.,
2015). Based on the microscopic images of stomata obtained by Wang,
stomatal length and distance between adjacent stomata are measured
with the Image-Pro software, and the point labeled with Arabidopsis is
shown in Fig. 6b as a preliminary exhibition of the relationship between
stomata on the hypocotyls and the extremal flux line of leaves. As can
be seen, this point is far away from the extremal flux line of leaves,
indicating the difference of distribution characteristic between stomata
on leaves and hypocotyls. Nevertheless, more species of plants and
microscopic images of stomata should be investigated to clarify the
reasons for this difference.
4. Conclusions
The edge diffusion effect of the spindly micron-scale stomata and
the stomatal distribution patterns approaching the extremal flux line
are the reasons for the high-efficiency diffusion characteristics of the
stomatal array of the natural leaves. The vegetation benefits from the
144
Journal of Plant Physiology 234–235 (2019) 138–144
stomatal distribution characteristics by facilitating the higher tran-
spiration rate for flourishing development from the perspective of mass
transfer theory. The optimization of the stomatal distribution patterns
of the natural leaves may be the result of natural optimization and
artificial selection.
Acknowledgement
This work was funded by the China National Natural Science
Foundation (contract grant number 51576188)
References
Bange, G.G.J., 1953. On the quantitative explanation of stomatal transpiration. Acta Bot.
Neerl. 2, 255–297. https://doi.org/10.1111/j.1438-8677.1953.tb00275.x.
Brodribb, T.J., McAdam Scott, A.M., 2011. Passive origins of stomatal control in vascular
plants. Science 331, 582–585. https://doi.org/10.1126/science.1197985.
Cannon, J.N., Krantz, W.B., Kreith, F., Naot, D., 1979. A study of transpiration from
porous flat plates simulating plant leaves. Int. J. Heat Mass Tran. 22, 469–483.
https://doi.org/10.1016/0017-9310(79)90013-9.
Cooke, J.R., 1967. Some theoretical considerations in stomatal diffusion: a field theory
approach. Acta Biotheor. 17, 95–124. https://doi.org/10.1007/BF01625092.
Defraeye, T., Verboven, P., Derome, D., Carmeliet, J., Nicolai, B., 2013. Stomatal tran-
spiration and droplet evaporation on leaf surfaces by a microscale modelling ap-
proach. Int. J. Heat Mass Tran. 65, 180–191. https://doi.org/10.1016/j.
ijheatmasstransfer.2013.05.075.
Farquhar, G.D., Raschke, K., 1978. On the resistance to transpiration of the sites of
evaporation within the leaf. Plant Physiol. 61, 1000–1005. https://doi.org/10.1104/
pp.61.6.1000.
Haghighi, E., Shahraeeni, E., Lehmann, P., Or, D., 2013. Evaporation rates across a
convective air boundary layer are dominated by diffusion. Water Resour. Res. 49,
1602–1610. https://doi.org/10.1002/wrcr.20166.
Hopkins, W.G., 1999. Introduction to Plant Physiology. John Wiley and Sons.
Lawson, T., Blatt, M.R., 2014. Stomatal size, speed, and responsiveness impact on pho-
tosynthesis and water use efficiency. Plant Physiol. 164, 1556–1570. https://doi.org/
10.1104/pp.114.237107.
Lee, R., Gates, D.M., 1964. Diffusion resistance in leaves as related to their stomatal
anatomy and micro-structure. Am. J. Bot. 51, 963–975. https://doi.org/10.1002/j.
1537-2197.1964.tb06725.x.
Lehmann, P., Or, D., 2015. Effects of stomata clustering on leaf gas exchange. New Phytol.
207, 1015–1025. https://doi.org/10.1111/nph.13442.
McElwain, J.C., Yiotis, C., Lawson, T., 2016. Using modern plant trait relationships be-
tween observed and theoretical maximum stomatal conductance and vein density to
examine patterns of plant macroevolution. New Phytol. 209, 94–103. https://doi.
org/10.1111/nph.13579.
Nobel, P.S., 1970. Introduction to Biophysical Plant Physiology.
Nobel, P.S., 1999. Physicochemical & Environmental Plant Physiology. Academic press.
Pariyar, S., Eichert, T., Goldbach, H.E., Hunsche, M., Burkhardt, J., 2013. The exclusion of
ambient aerosols changes the water relations of sunflower (Helianthus annuus) and
bean (Vicia faba) plants. Environ. Exp. Bot. 88, 43–52. https://doi.org/10.1016/j.
envexpbot.2011.12.031.
Parkhurst, D.F., 1994. Diffusion of CO2 and other gases inside leaves. New Phytol. 126,
449–479. https://doi.org/10.1111/j.1469-8137.1994.tb04244.x.
Pickard, W.F., 1981. Lumped circuit approximations for flux in systems governed by the
Laplace equation: the stoma of zero thickness. Chem. Eng. Sci. 36, 191–197. https://
doi.org/10.1016/0009-2509(81)80063-2.
Pieruschka, R., Huber, G., Berry, J.A., 2010. Control of transpiration by radiation. Proc.
Nati. Acad. Sci. 107, 13372–13377. https://doi.org/10.1073/pnas.0913177107.
Prado, K., Boursiac, Y., Tournaire-Roux, C., Monneuse, J.-M., Postaire, O., Da Ines, O.,
Schäffner, A.R., Hem, S., Santoni, V., Maurel, C., 2013. Regulation of Arabidopsis leaf
hydraulics involves light-dependent phosphorylation of aquaporins in veins. Plant
Cell 25, 1029–1039. https://doi.org/10.1105/tpc.112.108456.
Roth-Nebelsick, A., 2007. Computer-based studies of diffusion through stomata of dif-
ferent architecture. Ann. Bot. 100, 23–32. https://doi.org/10.1093/aob/mcm075.
Roth-Nebelsick, A., Hassiotou, F., Veneklaas, E.J., 2009. Stomatal crypts have small ef-
fects on transpiration: a numerical model analysis. Plant Physiol. 151, 2018–2027.
https://doi.org/10.1104/pp.109.146969.
Sack, L., Holbrook, N.M., 2006. Leaf hydraulics. Annu. Rev. Plant Biol. 57, 361–381.
https://doi.org/10.1146/annurev.arplant.56.032604.144141.
Sayre, J.D., 1926. Physiology of stomata of Rumex patientia. Ohio J. Sci. 57, 233–266.
Stefan, J., 1882. Über die Verdampfung aus einem kreisförmig oder elliptisch begrenzten
Becken. Ann. Phys. 253, 550–560.
Ting, I.P., Loomis, W.E., 1963. Diffusion through stomates. Am. J. Bot. 50, 866–872.
Ting, I.P., Loomis, W.E., 1965. Further studies concerning stomatal diffusion. Plant
Physiol. 40, 220–228.
Troyer, J.R., 1980. Diffusion from a circular stoma through a boundary layer: a field-
theoretical analysis. Plant Physiol. 66, 250–253. https://doi.org/10.1104/pp.66.2.
250.
Waggoner, P.E., Zelitch, I., 1965. Transpiration and the stomata of leaves. Science 150,
1413–1420. https://www.jstor.org/stable/1716983.
Wang, M., Yang, K., Le, J., 2015. Organ‐specific effects of brassinosteroids on stomatal
production coordinate with the action of TOO MANY MOUTHS. J. integ. plant biol
57, 247–255. https://doi.org/10.1111/jipb.12285.
Willmer, C., Fricker, M., 1996. Stomata Vol. 2 Springer Science & Business Media.