- Email: [email protected]
Water transport in leaf vein systems and the flow velocity measurement with a new method
Accepted Manuscript Title: Water transport in leaf vein systems and the flow velocity measurement with a new method Author: Wangyu Liu Yuanqiang Luo Li Wang Tao Luo Yi Peng Lei Wu PII: DOI: Reference:
S0176-1617(16)30148-1 http://dx.doi.org/doi:10.1016/j.jplph.2016.06.022 JPLPH 52419
To appear in: Received date: Revised date: Accepted date:
6-12-2015 17-6-2016 27-6-2016
Please cite this article as: Liu Wangyu, Luo Yuanqiang, Wang Li, Luo Tao, Peng Yi, Wu Lei.Water transport in leaf vein systems and the flow velocity measurement with a new method.Journal of Plant Physiology http://dx.doi.org/10.1016/j.jplph.2016.06.022 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Water transport in leaf vein systems and the flow velocity measurement with a new method Wangyu Liu1*, Yuanqiang Luo1, Li Wang1, Tao Luo1, Yi Peng1, Lei Wu2 1
School of Mechanical and Automotive Engineering, South China University of
Technology, 381 Wushan Road, Tianhe District, Guangzhou 510641, People’s Republic of China 2
Department of Equipment Manufacturing, Zhongshan Torch Polytechnic, 60
Zhongshan Port Avenue, Torch Development Zone, Zhongshan 528436, People’s Republic of China *Corresponding author. Tel.: +86 20 87511967 E-mail address: [email protected] (W. Liu)
1
Abstract As an exploration to the nature, research about plants' physiological properties have never been suspended. Water transport in leaf vein systems is an essential part of plant growth and development. In this paper, a simple but efficient method combined the fluorescence labeling technology frequently used in bioresearch and the image-processing technology in the computer realm was developed to measure the flow velocity, which was used as a quantitative description to reveal the regulation of water transport in leaf vein systems. Three ordinary species of plants were selected for the experiments and the influence of the experimental conditions, such as the concentration of fluorescein and illumination intensity of LEDs, was investigated. Differences among the flow velocities of different leaf veins of the same leaf as well as the flow velocities of different species were shown in bar charts. The mean measured flow velocities of the midrib and secondary vein of Ficus virens Ait. var. sublanceolata (Miq.) Corner were 4.549 m/h and 3.174 m/h. As for Plumeria rubra L. cv. Acutifolia and Hamelia patens, that were 0.339 m/h and 0.463 m/h, 2.609 m/h and 2.586 m/h, respectively. With the algorithm developed in this paper, the variation of the flow velocity in leaf veins was investigated by setting a constant time interval. Then a verification of the flow velocity measured by the algorithm was performed. Finally, according to the natural conditions of a plant leaf, a simulation about the water transport in leaf vein systems was carried out, which is especially different from the previous research. Keywords: Algorithm; Experiment; Flow velocity measurement; Fluorescein; Im2
age-processing; Leaf vein systems
3
Introduction As an exploration of nature, research about plants' physiological properties has never been suspended. Water transport in leaf vein systems is an essential part of plant growth and development. The leaf vein architecture can influence the transport properties a lot. As a review, Roth et al. (2001) summarized knowledge of interrelationships between the form and function of leaf vein and the evolution of leaf vein patterns to investigate the functional properties of the leaf vein systems. The leaf vein networks with a large number of closed loops have been proven to be the optimal transport networks which have resilience to damage and fluctuations in load (Bohn and Magnasco, 2007; Corson, 2010; Katifori et al., 2010). Yet, there still lacks quantitative description of the water transport in real leaf vein systems. As a quantitative parameter, flow velocity can be a description to reveal the regulation of water transport in leaf vein systems. However, the measurement of flow velocity in leaf veins is seldom found in the previous research. Particle image velocimetry (PIV) is a new flow measurement technology developed from the 1980s, which combines optical technology, computer technology, photographic technology and image-processing technology (He, 2009). Hao et al. (2011) investigated the dynamic wetting characteristics of water droplets on silicon wafers with microscale regular pillars structures and fresh lotus leaves experimentally. The internal velocity distribution of water droplets on both of these super hydrophobic surfaces were studied with a PIV system. However, the situation will be of great difference when the fluid region to be measured is too 4
small to be beyond the optical resolution or particle size, then the traditional PIV technology is no longer applicable (He, 2009). In order to solve this problem, a Micro-PIV technology was developed, whereas, no application on plant leaves could be found. Soares et al. (2013) presented an alternative protocol for use with the PIV technique in fluid and particle flow monitoring, without the use of external particles seeded as targets in the cross-correlation of the flow images. The method was based on the dynamic laser speckle patterns, or bio-speckle laser (BSL) patterns, with grains varying over time. A test used a simulated speckle pattern in a micro-flow in a torn leaf reacting to the broken internal pressure was taken. The results confirmed this hypothesis regarding the use of BSL associated to a PIV technique and illustrated a protocol to deal with the boiling effect that undermined the translational information in the speckle patterns. It was a significant breakthrough of flow measurement in plant leaves, but the method and experimental conditions were too difficult and complex to be followed. Furthermore, there was not any quantitative data reported about the flow velocity in the article. Garbe et al. (2002) used a technique of active thermography for estimating the water velocity in plant leaves. In his experiment, a part of the leaf was heated by a laser or heating wire and the movement of the heated water parcel was then visualized by an infrared camera. The water velocity, which must be calculated from a complex set of equations, was not given, too. Furthermore, the difference of temperature might have a significant impact on the water velocity and the heated water flowing from the petiole was opposite to the nature flow. Hüve et al. (2002) used 5
the dyes neutral red and acid fuchsin for staining of the water transport pathways in leaf veins. The acid fuchsin was observed under a binocular microscope with a magnification of 2.5 to 5 times, to measure the time interval up to the first visibility of the dye reaching tiny dots of ink placed along the two or three selected veins every 5 mm. Then the transport velocity of the dye acid fuchsin in veins of different size, within intact and damaged leaflets, was calculated. The experimental results indicated that the transport velocity in small veins of damaged leaves was increased several fold compared to the measurement of transport velocity in veins of the same size in intact leaves, which ensured water transport to leaf areas distal from the cut. So far, this is the only flow velocity data in leaf veins reported by the literature. But it should be noticed that the measurement under the binocular microscope could only obtain the data from two or three selected veins once, which meant that the whole data of the same sample could not be obtained at the same time, and that might bring in serious errors caused by the diversity of different plant leaves, even of the same species. In order to use the flow velocity as a quantitative description to reveal the regulation of water transport in leaf vein systems, a simple but efficient method must be developed for the flow velocity measurement. The method of using a fluorescence probe as a tracer to discover the mysteries inside plants is well developed. As being mostly used by researchers, fluorescent proteins play an indispensable role in unraveling the mechanism that governs the plant growth and development. For example, Chapman et al. (2005) reviewed the engineering of fluorescent proteins which continued to pro6
duce new tools for in vivo studies. The motivation of these improvements was to make the fluorescent proteins brighter and easier to be imaged. Zwiewka and Friml (2012) introduced the fluorescence imaging-based forward genetic screens as a developed approach which had progressed at a remarkable pace in the field of biological imaging in the last decades to identify trafficking regulators in plants. Compared with the conventional organic fluorescent dye, quantum dots (QDs) is a new fluorescence material and has the advantages of abundant color, stable photochemical properties, little fluorescence scattering and photo-bleaching, and low biological toxicity. Consequently, it is widely applied in the fields of biological tag, human pathology, material science, plant cell detachment and mark, genomics, proteomics, microorganism, biological imaging, biochip and so on (Chen et al., 2010; Pang et al., 2009). However, the application of QDs on plant research just began a few years ago and it is mainly used in roots and stems, not in leaves (Al-Salim et al., 2011; Hu et al., 2010). Besides fluorescent proteins, fluorescent dyes are also effective tracers for plant physiology studies, especially for the visualization of water transport through the plant leaves. Salleo et al. (2003) perfused Prunus laurocerasus L. and Juglans regia L. with a 2% Phloxine B solution under pressure. Infiltration of leaves with Phoxine B revealed that P. laurocerasus major veins were largely leaky in the radial direction whereas those of J. regia leaflets showed prevailing axial water transport. Katifori et al. (2010) injected fluorescein post injury at the stem of a lemon leaf, of which a circular cut was made on the main vein. It was observed that the fluorescein flowed through the vein net7
work around the injury, closing a number of loops and eventually reaching the tip of the leaf. The experimental results indicated that the existence of a high density of loops in transport networks, such as leaf veins, has resilience to damage and fluctuations in load. Both Salleo and Katifori have studied the characteristics of water transport through the plant leaves using fluorescent dyes as tracers. But they haven't quantitatively analyzed the flow velocity of water transport in leaf vein systems. With the development of computer technology, various efficient algorithms were applied for leaf veins recognition and extraction (Chen et al., 2011; Li and Zhao, 2012; Li et al., 2006; Radha and Jeyalakshmi, 2014; Zheng and Wang, 2010). Whereas, they were only fit for static leaf images. Inspired by the achievements in plants image-processing, the authors of this paper have tried an algorithm based on image-processing to recognize the position of the fluorescent dye flowing through the leaf veins and then automatically calculate the flow velocity in a selected leaf vein in a definite time interval. In this paper, a simple but efficient method is developed to measure the flow velocity, which is used as a quantitative description to reveal the regulation of water transport in leaf vein systems.
Materials and Methods Plant material and tracer For plant material preparation, three ordinary species of plants with heat-resistance 8
in the campus, Ficus virens Ait. var. sublanceolata (Miq.) Corner (Fig. 1a), Plumeria rubra L. cv. Acutifolia (Fig. 1b) and Hamelia patens (Fig. 1c) were selected. During the experiments, the leaves of the three selected plants were cut and fed with a tracer fluorescein (C20H12O5). The diameter of the fluorescein molecule is about 0.870 nm, which is much smaller than the pits pores (about 5 nm). Thus the movement of fluorescein could be representative of the velocity of the water. Legible and smooth flowing processes were observed. Fig. 1 shows the flowing images of the leaves of the three selected plants during the experiment, on which the tracer fluorescein can be easily detected.
9
Fig. 1. The flowing images of the leaves of the three selected plants during the experiment: (a) Ficus virens Ait. var. sublanceolata (Miq.) Corner; (b) Plumeria rubra L. cv. Acutifolia; (c) Hamelia patens. The tracer fluorescein (C20H12O5) which emitted orange light under the blue light source was flowing through the veins from the petiole (at the right side, out of the field of view). The images were extracted from the experimental videos recorded by a camera equipped with the orange and the yellow 10
light filters under the blue light source. In order to achieve good results, we selected the intact leaf samples from the three selected plants above, which were picked up with the stem remaining and placed in a plastic bag with wet towels to prevent damage to the veins. In the laboratory the leaves were stored in water and examined on the day of sample preparation.
Experimental setup What we are interested in is the fluid flow in leaf veins. Therefore, we need to make the fluid visible. Our preferred approach is the one that could make the flow information digitized. Then, we tracked the flow by supplying the leaf with water dissolved fluorescein and recorded the flow from the petiole to leaf tip. After a lot of trial-and-error experiments, we arrived at the setup with good effect shown in Fig. 2, which is described in the following.
11
Fig. 2. Pictures of the experimental setup for the flow observation: (a) room lighting; (b) partial enlarged detail; (c) blue light source. The number of the blue LEDs was 10 at each side. Every lamp holder was flexible and could be turned on and off independently. The camera was set on a piece of foam and connected to a computer, which was put on another table separated from the experimental platform. The leaf’s upside was fixed to a black piece of cardboard, which was just of the 12
same size and shape of the leaf, with double-faced adhesive tape, due to the veins running near the backside of the leaf in the investigated species and therefore being more easily perceivable from that side. Then the cardboard with the leaf on it was mounted onto a black wooden frame which was fixed on the experimental platform. A large piece of black cardboard was placed behind the leaf as a black background from a distance in order to avoid unwanted light reflections (and noise). On the same note, the container with the water-fluorescein solution had to be hidden within the frame, lest the light from the solution make the videos nearly unusable. Considering that our aim was to make the light noise on the pictures as minimal as possible, the lighting and camera were chosen accordingly. As light sources, we used 20 blue ultra-bright SHL 5-LED modules which were fixed symmetrically at two sides of the camera object lens. Every lamp holder was flexible and could be turned on and off independently. This way of positioning the lens in the middle of the light sources helped to reduce noise from shadows. What's more, the relative position of the lens and light sources formed an angle between the incident light and reflected light, which further reduced the unwanted light reflections from the leaf surface with a polariscope Haida C-POL. The light source was chosen for two reasons: The blue light should theoretically stimulate the photosynthesis of the leaves, which meant the stomata were open and thus there was a flow through the veins. Additionally, the tracer fluorescein (C20H12O5) emits orange light under the blue light source, which could be filtered from the light source by using two color blocking lens, the orange and the 13
yellow light filters, thus further reducing noise light and making the water-fluorescein solution brighter on the recording videos. It should be pointed out that, if the orange light filter is strong enough to filter all blue light on the recording screen, then only an orange light filter is needed. But it should also be noticed that it must not be too strong to make the recording screen all turn orange that the water-fluorescein solution becomes invisible. The combination of an orange and a yellow light filters is the best choice of this case. For recording, we used a Canon EOS 700D camera equipped with a Canon Macro Lens EF 100mm f/2.8 USM. To minimize the vibrations as well, the camera was set on a piece of foam and all operations were taken on a computer connected with it, which was put on another table separated from the experimental platform (Fig. 2). It is similar to a fluorescence and reflectance imaging setup (Lenk et al., 2007). Before the experiment, the leaf was cut-off from the stem under water, in order to avoid introducing embolisms into the vein network that would inevitably reduce rates of flow. Then the leaf was mounted on the tape. The position of the cardboard with the leaf on it could be adjusted upwards or downwards to fit the leaves of different sizes. Then the petiole was placed into the container and the camera was turned on and switched to the video recording mode. The position of the camera was adjusted forwards or backwards to make sure that the recording screen contained almost the whole leaf. Afterwards, the light was switched from room lighting (Fig. 2a) to our blue light source (Fig. 2c). Then the polariscope C-POL was spun to a suitable place 14
in order to furthest weaken the unwanted light reflections onto the recording screen. All of these operations must be done expertly within 2 minutes to avoid the formation of embolisms near the base of the petiole, which will affect the water transport velocity. As soon as the recording was begun, the water-fluorescein solution was injected into the container until the bottom of the petiole was covered, ensuring that the solution could flow into the petiole. The recording could be stopped when the leaf was almost filled with the solution. All experiments in this paper were conducted between 5 January and 9 January 2015 in an enclosing room with a temperature between 24 ℃ and 28 ℃. The final effect of the recording videos is shown as in Fig. 1.
Flow velocity measurement algorithm After the recording of the flow process from the petiole to leaf tip, the next step was to extract the flow velocity from the videos. In order to recognize the position of the fluorescein flowing through the leaf veins in the recording videos, which was the most difficult part, an algorithm was designed based on image-processing using a programming software MATLAB R2014a. Firstly, images were extracted from the recording video every few frames. Then the image of the last frame, on which the fluorescein almost flowed through the leaf, was chosen and turned into a grayscale image (Fig. 3a). The grayscale image was processed to a binary image (Fig. 3b) by choosing a threshold, and then a leaf veins image (Fig. 3c) could be obtained by skeletonizing the binary image. On the leaf veins 15
image, the leaf veins were numbered in order and the coordinate of each point could be read from the software. The most important part of our algorithm is the recognition of the position of the fluorescein. A judging criteria of the brightness (grey level) was used here. For a selected leaf vein to be measured, a suitable zone of frames [A, B] was selected. In order to cut noises on the images, the grayscale images of frame A and frame B were subtracted by the grayscale image of the first frame of the video, respectively. Then the brightness of each point on the selected leaf vein could be obtained (Fig. 3d and Fig. 3e). After review and comparison of the videos, a threshold of the brightness was set and the point with the setting brightness was taken as the farthest position of the flowing fluorescein. The distance of the two points on the grayscale images of frame A and frame B (marked in Fig. 3d and Fig. 3e respectively) could be calculated with their coordinates read from the software. Finally, the flow velocity between the two points could be obtained because the time interval between frame A and frame B could be easily known. For better understanding of the algorithm, Fig. 3 shows some images of H. patens in the process of the flow velocity measurement algorithm.
16
Fig. 3. Images of H. patens in the process of the flow velocity measurement algorithm: (a) grayscale image; (b) binary image; (c) leaf veins image; (d) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame A; (e) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame B. The abscissa in (d) and (e) is the number of each point on the leaf veins image (c), which is numbered in order starting from the base of the petiole.
Simulation Besides the experiment, a simulation using the finite element analysis software ANSYS Fluent 15.0 was carried out to simulate the water transport in leaf vein sys17
tems. In the previous research, the structure of leaf vein systems was constantly imitated in the design of microchannel. Therefore, simulations on the structure of leaf vein systems were almost based on microchannel. Xin (2013) established a 3D leaf vein system model consisting of the midrib and secondary veins, and then simulated the water transport in the model. His simulation was based on a natural plant leaf instead of microchannel. Whereas, the boundary conditions of setting the petiole as an inlet, the ends of the midrib and secondary veins as outlets, the surfaces of the leaf veins as no slip walls were actually quite different from the natural conditions of a plant leaf. Basing on the image of the last frame extracted from the recording video, on which the fluorescein almost flowed through the leaf, a 2D sketch of the structure of leaf vein systems could be extracted using a leaf extraction software LEAF GUI (Price et al., 2011). Then a 3D model of the leaf with mesophyll could be established by extruding the 2D sketch. Fig. 4 shows the 3D models of the leaves of the three selected plants. There are two parts: the leaf veins and mesophyll. The cross section of the leaf veins was assumed to be rectangle for simplification of the model, of which the thickness was 0.2 mm.
The total thicknesses of the whole models were all 0.3 mm.
18
Fig. 4. 3D models of the leaves of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. There are two parts: the leaf veins and mesophyll. The cross section of 19
the leaf veins was assumed to be rectangle, of which the thickness was 0.2 mm.
The
total thicknesses of the whole models were all 0.3 mm. According to the natural conditions of a plant leaf, the part of mesophyll was set as a porous zone with a porosity of 0.2.
The boundary conditions were set as below:
(1) Inlet: the petiole was set as a velocity-inlet and the velocity magnitude was set as the flow velocity measured from the experiment above; (2) Outlet: in a natural plant leaf, the liquid water turns into vapor and finally flows out of the leaf from the stomata, which mostly distribute on the lower surface of the leaf. So the lower surface of the leaf model was set as a pressure-outlet with the pressure of 0; (3) Porous-jump: from the viewpoint of botanist, the leaf vein systems (or vascular systems) are generally leaked tubes (Kizilova, 2008; Kizilova, 2003) through which the water can permeate into the mesophyll tissue, namely the destination of water transportation. Hence, the side faces and ends of the leaf veins were set as a porous-jump zone with the face permeability of 1e-10 m2 and the porous medium thickness of 0.001 mm; (4) Wall: other faces undefined of the leaf model were set as no slip walls.
Results Leaf vein anatomy Table 1 shows the overview of the vein anatomy in the basal part of the three se20
lected plants. Here, the following terms of leaf veins were used: (1) midrib (the main vein in the middle of the leaf, extending from the leaf base); (2) secondary veins (veins branching off the midrib, almost reaching the margin); (3) small secondary veins (also veins branching off the midrib, but smaller and not reaching the margin); (4) small veins (interconnections of secondary veins and small secondary veins). Midrib and secondary veins are often referred to as large veins.
Table 1 Overview of the vein anatomy in the basal part of the three selected plants. Diameter of the
Xylem
largest xylem
cross-sectional
vessel (μm)
area (μm2)
Number of xylem Vein vessels
F. virens Midrib
35-50
20-31
1560
Secondary
11-24
15-23
388
Small secondary
11-20
5.5-9
187
Small vein
3-7
5-7
90
25-51
6564
P. rubra Midrib
45-63
21
Secondary
17-31
14.5-29.5
1918
Small secondary
15-22
6.5-15
251
Small vein
5-7
6.5-9
120
H. patens Midrib
40-55
18-35
1950
Secondary
15-26
13.5-27.5
1219
Small secondary
11-21
5-8.5
189
Small vein
3-6
5.5-7.5
100
The data in Table 1 were extracted from the scanning electron micrographs of different leaf veins of the three selected plants observed under a scanning electron microscopy (SEM) (FEI Quanta 200) in low vacuum mode. From Table 1, among the three selected plants, the number of xylem vessels in the leaf veins of F. virens is the least and the xylem cross-sectional area is the smallest. H. patens ranks in the middle position while P. rubra has the largest number of xylem vessels and xylem cross-sectional area in the leaf veins.
Influence of the experimental conditions In order to investigate the influence of the experimental conditions on flow velocity
22
in leaf veins, the concentration of fluorescein and illumination intensity of LEDs were taken into consideration (Li et al., 2008). At the concentration of 0.1 g/L, the fluorescence of the water-fluorescein solution was too weak to be observed under the blue light source. So the concentrations of 0.3 g/L, 0.5 g/L, 0.7 g/L and 0.9 g/L were chosen for comparison. Fig. 5 shows the measured flow velocity in different leaf veins of the three selected plants at different concentrations of fluorescein. Each data was the mean flow velocity of at least three leaf veins, except for the midrib. The four leaves of the same plant at different concentrations were picked up on the same branch to reduce the influence of the diversity of different plant leaves. However, since the four leaves could not be the same entirely, it could be inferred that the influence of the concentration of fluorescein on flow velocity was inconspicuous. The difference among flow velocities at different concentrations may be caused by the diversity of different plant leaves even of the same species.
23
Fig. 5. The measured flow velocity in different leaf veins of the three selected plants at different concentrations of fluorescein: (a) F. virens; (b) P. rubra; (c) H. patens. The different concentrations of the water-fluorescein solution were 0.3 g/L, 0.5 g/L, 0.7 g/L and 0.9 g/L, respectively. To investigate the influence of the illumination intensity, the number of LEDs was controlled by turning on or off every LED independently. As the number decreasing 24
from 20 to 14, then to 8, symmetrically, the illumination intensity of LEDs decreased and the flow velocity decreased accordingly. Unlike that of the concentration of fluorescein, the influence of the illumination intensity of LEDs was so obvious that the recording time increased rapidly as the number of LEDs decreased. So, only a qualitative experiment was performed in this case. The results of the influence of the experimental conditions indicate that the adequate illumination intensity is important for the plant growth and development, especially for the water transport in leaf veins.
The concentration of the impurities in
water influences the water transport inconspicuously.
Differences among leaf veins and species Since the influence of the concentration of fluorescein on flow velocity was inconspicuous as shown in Fig. 5, Fig. 6 shows the mean measured flow velocity in different leaf veins of the three selected plants. As for sample F. virens, the flow velocity was the fastest and decreased with the decreasing of vein size, which was as expected by Hüve et al. (2002). The flow velocities both in the midrib and the secondary veins were almost the same for sample H. patens, and then it decreased with the decreasing of vein size. The slowest flow velocity was measured in sample P. rubra, in which the flow velocity in the secondary veins was faster than that in the midrib, and then it decreased with the decreasing of vein size. The results indicate that the flow velocities of different species can be quite different, which are in agreement with the anatomical 25
structures shown in Table 1. Comparing the three selected plant leaves, there may be some relations between the flow velocity and the structure of the leaf veins network as well as the natural characteristic of heat-resistance, which are to be investigated in the future work.
Fig. 6. The mean measured flow velocity in different leaf veins of the three selected plants.
Variation of the flow velocity in leaf veins With the algorithm mentioned above, the flow velocity in a selected leaf vein in a definite time interval can be measured and calculated quickly. By setting a constant time interval, the variation of the flow velocity in leaf veins can be investigated. Fig. 7a, Fig. 7c, and Fig. 7e show the variation of the flow velocity in midribs of the three selected plants for every 200, 1000, and 300 frames (25 fps), correspondingly. The figures show that, when the fluorescein flowed through, the flow velocity in the mid26
rib was the fastest at the beginning, then decreased with fluctuation. We guessed that on one hand the little fluctuation of the flow velocity might be caused by the wall roughness on the internal surface of leaf veins. On the other hand, the complicated network of the leaf veins consisted of a lot of branches and confluences, which might also have a great impact on the flow velocity. Fig. 7b, Fig. 7d, and Fig. 7f show the variation of the flow velocity in the secondary veins of the three selected plants. For each plant, three different secondary veins were selected for comparison, which were marked with the corresponding color on a little picture inserted in each figure. The zone and interval of frames measured and the serial number of the selected leaf vein were illustrated in each figure, too. Similar variation in the secondary veins was observed as that in the midribs. Furthermore, we noticed that a little difference existed among the three selected secondary veins of each plant, which might be caused by their different positions on the leaf veins network, as well as the different wall roughness on the internal surfaces, which is to be discussed in the future work. Additionally, it must be noted that the data in Fig. 7 was all measured under the same experimental condition, where the concentration of fluorescein was 0.5 g/L and the number of LEDs was 20.
27
28
Fig. 7. Variation of the flow velocity in: (a) the midrib of F. virens; (b) the secondary veins of F. virens; (c) the midrib of P. rubra; (d) the secondary veins of P. rubra; (e) the midrib of H. patens; (f) the secondary veins of H. patens. The abscissa "Frame" in 29
(a), (c) and (e) refers to the frame number of the corresponding experimental video. The frames intervals in (a), (c), and (e) are 200, 1000, and 300 frames (25 fps), respectively. In (b), (d), and (f), three different secondary veins of each plant are selected and marked with the corresponding color on a little picture inserted in each figure. The zone and interval of frames measured and the serial number of the selected leaf vein are illustrated in each figure, too. The data was all measured under the same experimental condition, where the concentration of fluorescein was 0.5 g/L and the number of LEDs was 20.
Verification of the flow velocity In Fig. 8, a verification of the flow velocity measured by the algorithm developed in this paper in a selected leaf vein is shown. Once a leaf vein was selected, the flow velocities on the given points of frame (e.g. 901, 1101, 1301, ..., 3101 in Fig. 8) were measured quickly by the algorithm within three different time intervals (zone of frames), respectively. For example, as in Fig. 8, the flow velocities of the green dash curve with triangle symbols were measured within 300 frames, respectively. The zones of frames were [751, 1051], [951, 1251], [1151, 1451], …, [2951, 3251]. The three dash curves match well with each other, which can verify the flow velocity measured by the algorithm developed in this paper.
30
Fig. 8. Verification of the flow velocity measured by the algorithm in a selected leaf vein. The given points of frame are 901, 1101, 1301, ..., 3101. The flow velocities of the blue dash curve with circle symbols were measured within 200 frames, respectively. The zones of frames were [801, 1001], [1001, 1201], [1201, 1401], …, [3001, 3201]. The flow velocities of the red dash curve with square symbols were measured within 250 frames, respectively. The zones of frames were [776, 1026], [976, 1226], [1176, 1426], …, [2976, 3226]. The flow velocities of the green dash curve with triangle symbols were measured within 300 frames, respectively. The zones of frames were [751, 1051], [951, 1251], [1151, 1451], …, [2951, 3251].
Comparison between the experimental and simulated flow velocity in the midrib Fig. 9 shows the comparison between the experimental and simulated flow velocity in the midribs of the three selected plants. The simulated data were extracted every 5 mm along the center axis of the midrib, which is similar to the experimental method performed by Hüve et al. (2002). The dash curves are polynomial fitting of the ex31
perimental data. Since the data of the flow velocity in the secondary veins were so much and so various, only the comparison of the flow velocity in the midrib was performed in this section. Obviously, the simulated flow velocity matches well with the polynomial fitting of the experimental flow velocity, which also verifies the new method of flow velocity measurement in leaf vein systems developed in this paper.
32
Fig. 9. Comparison between the experimental and simulated flow velocity in the midribs of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. The simulated data were extracted every 5 mm along the center axis of the midrib. The dash curves are polynomial fitting of the experimental data.
Discussion With the development of measuring technique, many new methods of flow velocity measurement have been come up with. Bio-speckle technique (Yokoi and Aizu, 2010), particle image velocimetry (PIV) (Hao et al., 2011; Soares et al., 2013), laser induce fluorescent (LIF) technique (Charalampous and Hardalupas, 2011) and tomograph technique (Maad et al., 2010) are representations of the flow visualization technique of a new generation. Whereas, methods of flow velocity measurement in leaf veins could hardly be found. The development tendency of the flow visualization technique is the combination of several technologies. In this paper, a simple but efficient method 33
was developed to measure the flow velocity, which was used as a quantitative description to reveal the regulation of water transport in leaf vein systems. This method combined the fluorescence labeling technology frequently used in bioresearch and the image-processing technology in computer realm. Comparing with the three methods mentioned above developed by other researchers (Garbet et al., 2002; Hüve et al., 2002; Soares et al., 2013), the new method developed in this paper is simpler in operation and the quantitative data obtained are intuitive and reliable. The experimental conditions are easy to get and closer to the natural conditions. Most importantly, the whole data of the same sample can be obtained at the same time, avoiding the errors caused by the diversity of different plant leaves even of the same species. The diversity of plants exists naturally, which influences the data of the experiments about plants unavoidably. In the experiments performed in this paper, Fig. 5 showed the diversity of different plant leaves of the same species while Fig. 6 showed the diversity of different species. From the data shown in Fig. 6, plants could be divided into three kinds according to the relation between the flow velocities of the midrib and the secondary veins. The sample F. virens might be more similar as the broad bean (Vicia faba L. cv. Scirocco) (Hüve et al., 2002) in some respects because their relations of the flow velocities among different leaf veins were the same. When concentrating on a selected leaf vein, Hüve et al. (2002) measured the flow velocity every 5 mm under a binocular microscope with a magnification of 2.5 to 5 times in his experiments. In a similar way, we measured the flow velocity of a selected leaf 34
vein by setting a constant time interval in our experiments. In the meanwhile, because of the invisible blocking or interruption in the inside of leaf veins, the fluorescein did not flow through every leaf veins in order from the petiole to leaf tip. In order to obtain the data more accurately, the leaf veins selected for data extraction were the first ones through which the fluorescein flowed in the experiments preformed in this paper. The results of the experiments preformed above indicate that the new method developed in this paper is very efficient for flow velocity measurement, which is of high practical value to be used as a quantitative description to reveal the regulation of water transport in leaf vein systems. The algorithm developed in this paper is verified to be efficient and has wider future applications, although there is still an imperfection that the calculation of the flow velocity is not completely automatic. The choosing of the threshold of the brightness needs an experienced judgement, which may bring in some subjective errors. Another highlight is that the simulation carried out in this paper, the boundary conditions of which are actually according to the natural conditions of a plant leaf, is especially different from the previous research. But there is a little deficiency that some parameters, such as the porosity of the part of mesophyll, the face permeability and the porous medium thickness of the porous-jump zone, were set depending on experience instead of the measured value, which will doubtlessly reduce the accuracy of the simulation. As for the future work, in order to understand more about plants' physiological 35
properties and promote the technique on bionics design, more flow and thermal characteristics such as the pressure drop (Cochard et al., 2004; Zwieniecki et al., 2002) and temperature distribution (Garbe et al., 2002) of the water transport in leaf vein systems will be investigated. The advantages of plants with heat-resistance are to be investigated in the following experiments. A growth chamber will be used for plant cultivation. The growing conditions can be changed artificially to investigate their influences on the characteristics mentioned above. Some relative anatomy experiments will also be performed to discover the internal mechanism of the macroscopic phenomena. Furthermore, the sharpness of the recording videos and the efficiency of the flow velocity measurement algorithm are to be promoted to achieve a higher accuracy of the experimental data in the future.
Acknowledgements The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (NO. 51375169 & 11572128). And we wish to thank Prof. Eleni Katifori for helpful discussions on the experiment.
References Al-Salim N, Barraclough E, Burgess E, Clothier B, Deurer M, Green S, et al. Quantum dot transport in soil, plants, and insects. Science of the Total Environment. 2011;409:3237-48. Bohn S, Magnasco MO. Structure, scaling, and phase transition in the optimal transport network. Physical Review Letters. 2007;98:1-4. Chapman S, Oparka KJ, Roberts AG. New tools for in vivo fluorescence tagging. 36
Current Opinion in Plant Biology. 2005;8:565-73. Charalampous G, Hardalupas Y. Method to reduce errors of droplet sizing based on the ratio of fluorescent and scattered light intensities (laser-induced fluorescence/Mie technique). Applied Optics. 2011;50:3622-37. Chen J, Wang Q, Lu M, Zhang H, Liu B. Application of quantum dots in plant molecule fluorescent labeling. Scientia Silvae Sinicae. 2010;46:153-61. Chen Y, Lin P, He Y. Velocity representation method for description of contour shape and the classification of weed leaf images. Biosystems Engineering. 2011;109:186-95. Cochard H, Nardini A, Coll L. Hydraulic architecture of leaf blades: where is the main resistance? Plant, Cell & Environment. 2004;27:1257-67. Corson F. Fluctuations and redundancy in optimal transport networks. Physical Review Letters. 2010;104:1-4. Garbe C, Schurr U, Jaehne B. Thermographic measurements on plant leaves. AeroSense 2002: International Society for Optics and Photonics; 2002. p. 407-16. Hüve K, Remus R, Lüttschwager D, Merbach W. Water transport in impaired leaf vein systems. Plant Biology. 2002;4:603-11. Hao P, Yao Z, Zhang X. Study of dynamic hydrophobicity of micro-structured hydrophobic surfaces and lotus leaves. Science China: Physics, Mechanics and Astronomy. 2011;54:675-82. He W. Water transport mechanism and simulation in plant. Xi'an: Xi'an University of Architecture and Technology; 2009. Hu Y, Li J, Ma L, Peng Q, Feng W, Zhang L, et al. High efficiency transport of quantum dots into plant roots with the aid of silwet L-77. Plant Physiology and Biochemistry. 2010;48:703-09. Katifori E, Szöllősi GJ, Magnasco MO. Damage and fluctuations induce loops in optimal transport networks. Physical Review Letters. 2010;104:1-4. Kizilova N. Long-distance liquid transport in plants. Proceedings of the Estonian Academy of Sciences. 2008;57:179-203. Kizilova NN. Construction principles and control over transport systems organization in biological tissues. Physics and Control, 2003 Proceedings 2003 International Conference: IEEE; 2003. p. 302-07. Lenk S, Chaerle L, Pfündel EE, Langsdorf G, Hagenbeek D, Lichtenthaler HK, et al. Multispectral fluorescence and reflectance imaging at the leaf level and its possible applications. Journal of Experimental Botany. 2007;58:807-14. Li B, Zhao Z. Research on leaf recognition and simulation. Information Science and Control Engineering 2012 (ICISCE 2012), IET International Conference on: IET; 2012. p. 1-4. Li H-j, He X-k, Zeng A-j, Liu Y-j, Cao S. Effect of environment on fluorescence analysis of fluorescein sodium. Journal of Jilin Agricultural University. 2008;30:49-51, 63. Li Y, Chi Z, Feng DD. Leaf vein extraction using independent component analysis. 2006 SMC'06 IEEE International Conference on Systems, Man and Cybernetics: 37
IEEE; 2006. p. 3890-94. Maad R, Hjertaker B, Johansen G, Olsen Ø. Dynamic characterization of a high speed gamma-ray tomograph. Flow Measurement and Instrumentation. 2010;21:538-45. Pang J-f, Yu Z, Liang Z, Yin C-c, Liu H, Pei D, et al. The application of QDs and its development prospect in plant. Journal of Agricultural Science and Technology. 2009;11:19-26. Price CA, Symonova O, Mileyko Y, Hilley T, Weitz JS. Leaf extraction and analysis framework graphical user interface: segmenting and analyzing the structure of leaf veins and areoles. Plant Physiology. 2011;155:236-45. Radha R, Jeyalakshmi S. An effective algorithm for edges and veins detection in leaf images. 2014 World Congress on Computing and Communication Technologies (WCCCT): IEEE; 2014. p. 128-31. Roth-Nebelsick A, Uhl D, Mosbrugger V, Kerp H. Evolution and function of leaf venation architecture: a review. Annals of Botany. 2001;87:553-66. Salleo S, Raimondo F, Trifilo P, Nardini A. Axial-to-radial water permeability of leaf major veins: a possible determinant of the impact of vein embolism on leaf hydraulics? Plant, Cell & Environment. 2003;26:1749-58. Soares R, Barbosa H, Braga R, Botega J, Horgan G. Biospeckle PIV (Particle Image Velocimetry) for analyzing fluid flow. Flow Measurement and Instrumentation. 2013;30:90-98. Xin X. Investigation on the mechanism and law for heat and mass transfer in hardy plant blades. Qingdao: Qingdao University of Science and Technology; 2013. Yokoi N, Aizu Y. Improvement of estimation parameters in bio-speckle blood flow imaging. Optical Review. 2010;17:230-38. Zheng X, Wang X. Leaf vein extraction using a combined operation of mathematical morphology. 2010 2nd International Conference on Information Engineering and Computer Science (ICIECS): IEEE; 2010. p. 1-4. Zwieniecki M, Melcher P, Boyce C, Sack L, Holbrook N. Hydraulic architecture of leaf venation in Laurus nobilis L. Plant, Cell & Environment. 2002;25:1445-50. Zwiewka M, Friml J. Fluorescence imaging-based forward genetic screens to identify trafficking regulators in plants. Frontiers in Plant Science. 2012;3:1-14.
38
Table list Table 1 Overview of the vein anatomy in the basal part of the three selected plants.
39
Figure legends Fig. 1. The flowing images of the leaves of the three selected plants during the experiment: (a) Ficus virens Ait. var. sublanceolata (Miq.) Corner; (b) Plumeria rubra L. cv. Acutifolia; (c) Hamelia patens. The tracer fluorescein (C20H12O5) which emitted orange light under the blue light source was flowing through the veins from the petiole (at the right side, out of the field of view). The images were extracted from the experimental videos recorded by a camera equipped with the orange and the yellow lightfilters under the blue light source. Fig. 2. Pictures of the experimental setup for the flow observation: (a) room lighting; (b) partial enlarged detail; (c) blue light source. The number of the blue LEDs was 10 at each side. Every lamp holder was flexible and could be turned on and off independently. The camera was set on a piece of foam and connected with a computer, which was put on another table separated from the experimental platform. Fig. 3. Images of H. patens in the process of the flow velocity measurement algorithm: (a) grayscale image; (b) binary image; (c) leaf veins image; (d) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame A; (e) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame B. The abscissa in (d) and (e) is the number of each point on the leaf veins image (c), which is numbered in order starting from the base of the petiole. Fig. 4. 3D models of the leaves of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. There are two parts: the leaf veins and mesophyll. The cross section of the leaf veins was assumed to be rectangle, of which the thickness was 0.2 mm. And 40
the total thicknesses of the whole models were all 0.3 mm. Fig. 5. The measured flow velocity in different leaf veins of the three selected plants at different concentrations of fluorescein: (a) F. virens; (b) P. rubra; (c) H. patens. The different concentrations of the water-fluorescein solution were 0.3 g/L, 0.5 g/L, 0.7 g/L and 0.9 g/L, respectively. Fig. 6. The mean measured flow velocity in different leaf veins of the three selected plants. Fig. 7. Variation of the flow velocity in: (a) the midrib of F. virens; (b) the secondary veins of F. virens; (c) the midrib of P. rubra; (d) the secondary veins of P. rubra; (e) the midrib of H. patens; (f) the secondary veins of H. patens. The abscissa "Frame" in (a), (c) and (e) refers to the frame number of the corresponding experimental video. The frames intervals in (a), (c) and (e) are 200, 1000 and 300 frames (25 fps), respectively. In (b), (d) and (f), three different secondary veins of each plant are selected and marked with the corresponding color on a little picture inserted in each figure. The zone and interval of frames measured and the serial number of the selected leaf vein are illustrated in each figure, too. The data was all measured under the same experimental condition, where the concentration of fluorescein was 0.5 g/L and the number of LEDs was 20. Fig. 8. Verification of the flow velocity measured by the algorithm in a selected leaf vein. The given points of frame are 901, 1101, 1301, ..., 3101. The flow velocities of the blue dash curve with circle symbols were measured within 200 frames, respectively. The zones of frames were [801, 1001], [1001, 1201], [1201, 1401], …, [3001, 41
3201]. The flow velocities of the red dash curve with square symbols were measured within 250 frames, respectively. The zones of frames were [776, 1026], [976, 1226], [1176, 1426], …, [2976, 3226]. The flow velocities of the green dash curve with triangle symbols were measured within 300 frames, respectively. The zones of frames were [751, 1051], [951, 1251], [1151, 1451], …, [2951, 3251]. Fig. 9. Comparison between the experimental and simulated flow velocity in the midribs of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. The simulated data were extracted every 5 mm along the center axis of the midrib. The dash curves are polynomial fitting of the experimental data.
42
S0176-1617(16)30148-1 http://dx.doi.org/doi:10.1016/j.jplph.2016.06.022 JPLPH 52419
To appear in: Received date: Revised date: Accepted date:
6-12-2015 17-6-2016 27-6-2016
Please cite this article as: Liu Wangyu, Luo Yuanqiang, Wang Li, Luo Tao, Peng Yi, Wu Lei.Water transport in leaf vein systems and the flow velocity measurement with a new method.Journal of Plant Physiology http://dx.doi.org/10.1016/j.jplph.2016.06.022 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Water transport in leaf vein systems and the flow velocity measurement with a new method Wangyu Liu1*, Yuanqiang Luo1, Li Wang1, Tao Luo1, Yi Peng1, Lei Wu2 1
School of Mechanical and Automotive Engineering, South China University of
Technology, 381 Wushan Road, Tianhe District, Guangzhou 510641, People’s Republic of China 2
Department of Equipment Manufacturing, Zhongshan Torch Polytechnic, 60
Zhongshan Port Avenue, Torch Development Zone, Zhongshan 528436, People’s Republic of China *Corresponding author. Tel.: +86 20 87511967 E-mail address: [email protected] (W. Liu)
1
Abstract As an exploration to the nature, research about plants' physiological properties have never been suspended. Water transport in leaf vein systems is an essential part of plant growth and development. In this paper, a simple but efficient method combined the fluorescence labeling technology frequently used in bioresearch and the image-processing technology in the computer realm was developed to measure the flow velocity, which was used as a quantitative description to reveal the regulation of water transport in leaf vein systems. Three ordinary species of plants were selected for the experiments and the influence of the experimental conditions, such as the concentration of fluorescein and illumination intensity of LEDs, was investigated. Differences among the flow velocities of different leaf veins of the same leaf as well as the flow velocities of different species were shown in bar charts. The mean measured flow velocities of the midrib and secondary vein of Ficus virens Ait. var. sublanceolata (Miq.) Corner were 4.549 m/h and 3.174 m/h. As for Plumeria rubra L. cv. Acutifolia and Hamelia patens, that were 0.339 m/h and 0.463 m/h, 2.609 m/h and 2.586 m/h, respectively. With the algorithm developed in this paper, the variation of the flow velocity in leaf veins was investigated by setting a constant time interval. Then a verification of the flow velocity measured by the algorithm was performed. Finally, according to the natural conditions of a plant leaf, a simulation about the water transport in leaf vein systems was carried out, which is especially different from the previous research. Keywords: Algorithm; Experiment; Flow velocity measurement; Fluorescein; Im2
age-processing; Leaf vein systems
3
Introduction As an exploration of nature, research about plants' physiological properties has never been suspended. Water transport in leaf vein systems is an essential part of plant growth and development. The leaf vein architecture can influence the transport properties a lot. As a review, Roth et al. (2001) summarized knowledge of interrelationships between the form and function of leaf vein and the evolution of leaf vein patterns to investigate the functional properties of the leaf vein systems. The leaf vein networks with a large number of closed loops have been proven to be the optimal transport networks which have resilience to damage and fluctuations in load (Bohn and Magnasco, 2007; Corson, 2010; Katifori et al., 2010). Yet, there still lacks quantitative description of the water transport in real leaf vein systems. As a quantitative parameter, flow velocity can be a description to reveal the regulation of water transport in leaf vein systems. However, the measurement of flow velocity in leaf veins is seldom found in the previous research. Particle image velocimetry (PIV) is a new flow measurement technology developed from the 1980s, which combines optical technology, computer technology, photographic technology and image-processing technology (He, 2009). Hao et al. (2011) investigated the dynamic wetting characteristics of water droplets on silicon wafers with microscale regular pillars structures and fresh lotus leaves experimentally. The internal velocity distribution of water droplets on both of these super hydrophobic surfaces were studied with a PIV system. However, the situation will be of great difference when the fluid region to be measured is too 4
small to be beyond the optical resolution or particle size, then the traditional PIV technology is no longer applicable (He, 2009). In order to solve this problem, a Micro-PIV technology was developed, whereas, no application on plant leaves could be found. Soares et al. (2013) presented an alternative protocol for use with the PIV technique in fluid and particle flow monitoring, without the use of external particles seeded as targets in the cross-correlation of the flow images. The method was based on the dynamic laser speckle patterns, or bio-speckle laser (BSL) patterns, with grains varying over time. A test used a simulated speckle pattern in a micro-flow in a torn leaf reacting to the broken internal pressure was taken. The results confirmed this hypothesis regarding the use of BSL associated to a PIV technique and illustrated a protocol to deal with the boiling effect that undermined the translational information in the speckle patterns. It was a significant breakthrough of flow measurement in plant leaves, but the method and experimental conditions were too difficult and complex to be followed. Furthermore, there was not any quantitative data reported about the flow velocity in the article. Garbe et al. (2002) used a technique of active thermography for estimating the water velocity in plant leaves. In his experiment, a part of the leaf was heated by a laser or heating wire and the movement of the heated water parcel was then visualized by an infrared camera. The water velocity, which must be calculated from a complex set of equations, was not given, too. Furthermore, the difference of temperature might have a significant impact on the water velocity and the heated water flowing from the petiole was opposite to the nature flow. Hüve et al. (2002) used 5
the dyes neutral red and acid fuchsin for staining of the water transport pathways in leaf veins. The acid fuchsin was observed under a binocular microscope with a magnification of 2.5 to 5 times, to measure the time interval up to the first visibility of the dye reaching tiny dots of ink placed along the two or three selected veins every 5 mm. Then the transport velocity of the dye acid fuchsin in veins of different size, within intact and damaged leaflets, was calculated. The experimental results indicated that the transport velocity in small veins of damaged leaves was increased several fold compared to the measurement of transport velocity in veins of the same size in intact leaves, which ensured water transport to leaf areas distal from the cut. So far, this is the only flow velocity data in leaf veins reported by the literature. But it should be noticed that the measurement under the binocular microscope could only obtain the data from two or three selected veins once, which meant that the whole data of the same sample could not be obtained at the same time, and that might bring in serious errors caused by the diversity of different plant leaves, even of the same species. In order to use the flow velocity as a quantitative description to reveal the regulation of water transport in leaf vein systems, a simple but efficient method must be developed for the flow velocity measurement. The method of using a fluorescence probe as a tracer to discover the mysteries inside plants is well developed. As being mostly used by researchers, fluorescent proteins play an indispensable role in unraveling the mechanism that governs the plant growth and development. For example, Chapman et al. (2005) reviewed the engineering of fluorescent proteins which continued to pro6
duce new tools for in vivo studies. The motivation of these improvements was to make the fluorescent proteins brighter and easier to be imaged. Zwiewka and Friml (2012) introduced the fluorescence imaging-based forward genetic screens as a developed approach which had progressed at a remarkable pace in the field of biological imaging in the last decades to identify trafficking regulators in plants. Compared with the conventional organic fluorescent dye, quantum dots (QDs) is a new fluorescence material and has the advantages of abundant color, stable photochemical properties, little fluorescence scattering and photo-bleaching, and low biological toxicity. Consequently, it is widely applied in the fields of biological tag, human pathology, material science, plant cell detachment and mark, genomics, proteomics, microorganism, biological imaging, biochip and so on (Chen et al., 2010; Pang et al., 2009). However, the application of QDs on plant research just began a few years ago and it is mainly used in roots and stems, not in leaves (Al-Salim et al., 2011; Hu et al., 2010). Besides fluorescent proteins, fluorescent dyes are also effective tracers for plant physiology studies, especially for the visualization of water transport through the plant leaves. Salleo et al. (2003) perfused Prunus laurocerasus L. and Juglans regia L. with a 2% Phloxine B solution under pressure. Infiltration of leaves with Phoxine B revealed that P. laurocerasus major veins were largely leaky in the radial direction whereas those of J. regia leaflets showed prevailing axial water transport. Katifori et al. (2010) injected fluorescein post injury at the stem of a lemon leaf, of which a circular cut was made on the main vein. It was observed that the fluorescein flowed through the vein net7
work around the injury, closing a number of loops and eventually reaching the tip of the leaf. The experimental results indicated that the existence of a high density of loops in transport networks, such as leaf veins, has resilience to damage and fluctuations in load. Both Salleo and Katifori have studied the characteristics of water transport through the plant leaves using fluorescent dyes as tracers. But they haven't quantitatively analyzed the flow velocity of water transport in leaf vein systems. With the development of computer technology, various efficient algorithms were applied for leaf veins recognition and extraction (Chen et al., 2011; Li and Zhao, 2012; Li et al., 2006; Radha and Jeyalakshmi, 2014; Zheng and Wang, 2010). Whereas, they were only fit for static leaf images. Inspired by the achievements in plants image-processing, the authors of this paper have tried an algorithm based on image-processing to recognize the position of the fluorescent dye flowing through the leaf veins and then automatically calculate the flow velocity in a selected leaf vein in a definite time interval. In this paper, a simple but efficient method is developed to measure the flow velocity, which is used as a quantitative description to reveal the regulation of water transport in leaf vein systems.
Materials and Methods Plant material and tracer For plant material preparation, three ordinary species of plants with heat-resistance 8
in the campus, Ficus virens Ait. var. sublanceolata (Miq.) Corner (Fig. 1a), Plumeria rubra L. cv. Acutifolia (Fig. 1b) and Hamelia patens (Fig. 1c) were selected. During the experiments, the leaves of the three selected plants were cut and fed with a tracer fluorescein (C20H12O5). The diameter of the fluorescein molecule is about 0.870 nm, which is much smaller than the pits pores (about 5 nm). Thus the movement of fluorescein could be representative of the velocity of the water. Legible and smooth flowing processes were observed. Fig. 1 shows the flowing images of the leaves of the three selected plants during the experiment, on which the tracer fluorescein can be easily detected.
9
Fig. 1. The flowing images of the leaves of the three selected plants during the experiment: (a) Ficus virens Ait. var. sublanceolata (Miq.) Corner; (b) Plumeria rubra L. cv. Acutifolia; (c) Hamelia patens. The tracer fluorescein (C20H12O5) which emitted orange light under the blue light source was flowing through the veins from the petiole (at the right side, out of the field of view). The images were extracted from the experimental videos recorded by a camera equipped with the orange and the yellow 10
light filters under the blue light source. In order to achieve good results, we selected the intact leaf samples from the three selected plants above, which were picked up with the stem remaining and placed in a plastic bag with wet towels to prevent damage to the veins. In the laboratory the leaves were stored in water and examined on the day of sample preparation.
Experimental setup What we are interested in is the fluid flow in leaf veins. Therefore, we need to make the fluid visible. Our preferred approach is the one that could make the flow information digitized. Then, we tracked the flow by supplying the leaf with water dissolved fluorescein and recorded the flow from the petiole to leaf tip. After a lot of trial-and-error experiments, we arrived at the setup with good effect shown in Fig. 2, which is described in the following.
11
Fig. 2. Pictures of the experimental setup for the flow observation: (a) room lighting; (b) partial enlarged detail; (c) blue light source. The number of the blue LEDs was 10 at each side. Every lamp holder was flexible and could be turned on and off independently. The camera was set on a piece of foam and connected to a computer, which was put on another table separated from the experimental platform. The leaf’s upside was fixed to a black piece of cardboard, which was just of the 12
same size and shape of the leaf, with double-faced adhesive tape, due to the veins running near the backside of the leaf in the investigated species and therefore being more easily perceivable from that side. Then the cardboard with the leaf on it was mounted onto a black wooden frame which was fixed on the experimental platform. A large piece of black cardboard was placed behind the leaf as a black background from a distance in order to avoid unwanted light reflections (and noise). On the same note, the container with the water-fluorescein solution had to be hidden within the frame, lest the light from the solution make the videos nearly unusable. Considering that our aim was to make the light noise on the pictures as minimal as possible, the lighting and camera were chosen accordingly. As light sources, we used 20 blue ultra-bright SHL 5-LED modules which were fixed symmetrically at two sides of the camera object lens. Every lamp holder was flexible and could be turned on and off independently. This way of positioning the lens in the middle of the light sources helped to reduce noise from shadows. What's more, the relative position of the lens and light sources formed an angle between the incident light and reflected light, which further reduced the unwanted light reflections from the leaf surface with a polariscope Haida C-POL. The light source was chosen for two reasons: The blue light should theoretically stimulate the photosynthesis of the leaves, which meant the stomata were open and thus there was a flow through the veins. Additionally, the tracer fluorescein (C20H12O5) emits orange light under the blue light source, which could be filtered from the light source by using two color blocking lens, the orange and the 13
yellow light filters, thus further reducing noise light and making the water-fluorescein solution brighter on the recording videos. It should be pointed out that, if the orange light filter is strong enough to filter all blue light on the recording screen, then only an orange light filter is needed. But it should also be noticed that it must not be too strong to make the recording screen all turn orange that the water-fluorescein solution becomes invisible. The combination of an orange and a yellow light filters is the best choice of this case. For recording, we used a Canon EOS 700D camera equipped with a Canon Macro Lens EF 100mm f/2.8 USM. To minimize the vibrations as well, the camera was set on a piece of foam and all operations were taken on a computer connected with it, which was put on another table separated from the experimental platform (Fig. 2). It is similar to a fluorescence and reflectance imaging setup (Lenk et al., 2007). Before the experiment, the leaf was cut-off from the stem under water, in order to avoid introducing embolisms into the vein network that would inevitably reduce rates of flow. Then the leaf was mounted on the tape. The position of the cardboard with the leaf on it could be adjusted upwards or downwards to fit the leaves of different sizes. Then the petiole was placed into the container and the camera was turned on and switched to the video recording mode. The position of the camera was adjusted forwards or backwards to make sure that the recording screen contained almost the whole leaf. Afterwards, the light was switched from room lighting (Fig. 2a) to our blue light source (Fig. 2c). Then the polariscope C-POL was spun to a suitable place 14
in order to furthest weaken the unwanted light reflections onto the recording screen. All of these operations must be done expertly within 2 minutes to avoid the formation of embolisms near the base of the petiole, which will affect the water transport velocity. As soon as the recording was begun, the water-fluorescein solution was injected into the container until the bottom of the petiole was covered, ensuring that the solution could flow into the petiole. The recording could be stopped when the leaf was almost filled with the solution. All experiments in this paper were conducted between 5 January and 9 January 2015 in an enclosing room with a temperature between 24 ℃ and 28 ℃. The final effect of the recording videos is shown as in Fig. 1.
Flow velocity measurement algorithm After the recording of the flow process from the petiole to leaf tip, the next step was to extract the flow velocity from the videos. In order to recognize the position of the fluorescein flowing through the leaf veins in the recording videos, which was the most difficult part, an algorithm was designed based on image-processing using a programming software MATLAB R2014a. Firstly, images were extracted from the recording video every few frames. Then the image of the last frame, on which the fluorescein almost flowed through the leaf, was chosen and turned into a grayscale image (Fig. 3a). The grayscale image was processed to a binary image (Fig. 3b) by choosing a threshold, and then a leaf veins image (Fig. 3c) could be obtained by skeletonizing the binary image. On the leaf veins 15
image, the leaf veins were numbered in order and the coordinate of each point could be read from the software. The most important part of our algorithm is the recognition of the position of the fluorescein. A judging criteria of the brightness (grey level) was used here. For a selected leaf vein to be measured, a suitable zone of frames [A, B] was selected. In order to cut noises on the images, the grayscale images of frame A and frame B were subtracted by the grayscale image of the first frame of the video, respectively. Then the brightness of each point on the selected leaf vein could be obtained (Fig. 3d and Fig. 3e). After review and comparison of the videos, a threshold of the brightness was set and the point with the setting brightness was taken as the farthest position of the flowing fluorescein. The distance of the two points on the grayscale images of frame A and frame B (marked in Fig. 3d and Fig. 3e respectively) could be calculated with their coordinates read from the software. Finally, the flow velocity between the two points could be obtained because the time interval between frame A and frame B could be easily known. For better understanding of the algorithm, Fig. 3 shows some images of H. patens in the process of the flow velocity measurement algorithm.
16
Fig. 3. Images of H. patens in the process of the flow velocity measurement algorithm: (a) grayscale image; (b) binary image; (c) leaf veins image; (d) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame A; (e) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame B. The abscissa in (d) and (e) is the number of each point on the leaf veins image (c), which is numbered in order starting from the base of the petiole.
Simulation Besides the experiment, a simulation using the finite element analysis software ANSYS Fluent 15.0 was carried out to simulate the water transport in leaf vein sys17
tems. In the previous research, the structure of leaf vein systems was constantly imitated in the design of microchannel. Therefore, simulations on the structure of leaf vein systems were almost based on microchannel. Xin (2013) established a 3D leaf vein system model consisting of the midrib and secondary veins, and then simulated the water transport in the model. His simulation was based on a natural plant leaf instead of microchannel. Whereas, the boundary conditions of setting the petiole as an inlet, the ends of the midrib and secondary veins as outlets, the surfaces of the leaf veins as no slip walls were actually quite different from the natural conditions of a plant leaf. Basing on the image of the last frame extracted from the recording video, on which the fluorescein almost flowed through the leaf, a 2D sketch of the structure of leaf vein systems could be extracted using a leaf extraction software LEAF GUI (Price et al., 2011). Then a 3D model of the leaf with mesophyll could be established by extruding the 2D sketch. Fig. 4 shows the 3D models of the leaves of the three selected plants. There are two parts: the leaf veins and mesophyll. The cross section of the leaf veins was assumed to be rectangle for simplification of the model, of which the thickness was 0.2 mm.
The total thicknesses of the whole models were all 0.3 mm.
18
Fig. 4. 3D models of the leaves of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. There are two parts: the leaf veins and mesophyll. The cross section of 19
the leaf veins was assumed to be rectangle, of which the thickness was 0.2 mm.
The
total thicknesses of the whole models were all 0.3 mm. According to the natural conditions of a plant leaf, the part of mesophyll was set as a porous zone with a porosity of 0.2.
The boundary conditions were set as below:
(1) Inlet: the petiole was set as a velocity-inlet and the velocity magnitude was set as the flow velocity measured from the experiment above; (2) Outlet: in a natural plant leaf, the liquid water turns into vapor and finally flows out of the leaf from the stomata, which mostly distribute on the lower surface of the leaf. So the lower surface of the leaf model was set as a pressure-outlet with the pressure of 0; (3) Porous-jump: from the viewpoint of botanist, the leaf vein systems (or vascular systems) are generally leaked tubes (Kizilova, 2008; Kizilova, 2003) through which the water can permeate into the mesophyll tissue, namely the destination of water transportation. Hence, the side faces and ends of the leaf veins were set as a porous-jump zone with the face permeability of 1e-10 m2 and the porous medium thickness of 0.001 mm; (4) Wall: other faces undefined of the leaf model were set as no slip walls.
Results Leaf vein anatomy Table 1 shows the overview of the vein anatomy in the basal part of the three se20
lected plants. Here, the following terms of leaf veins were used: (1) midrib (the main vein in the middle of the leaf, extending from the leaf base); (2) secondary veins (veins branching off the midrib, almost reaching the margin); (3) small secondary veins (also veins branching off the midrib, but smaller and not reaching the margin); (4) small veins (interconnections of secondary veins and small secondary veins). Midrib and secondary veins are often referred to as large veins.
Table 1 Overview of the vein anatomy in the basal part of the three selected plants. Diameter of the
Xylem
largest xylem
cross-sectional
vessel (μm)
area (μm2)
Number of xylem Vein vessels
F. virens Midrib
35-50
20-31
1560
Secondary
11-24
15-23
388
Small secondary
11-20
5.5-9
187
Small vein
3-7
5-7
90
25-51
6564
P. rubra Midrib
45-63
21
Secondary
17-31
14.5-29.5
1918
Small secondary
15-22
6.5-15
251
Small vein
5-7
6.5-9
120
H. patens Midrib
40-55
18-35
1950
Secondary
15-26
13.5-27.5
1219
Small secondary
11-21
5-8.5
189
Small vein
3-6
5.5-7.5
100
The data in Table 1 were extracted from the scanning electron micrographs of different leaf veins of the three selected plants observed under a scanning electron microscopy (SEM) (FEI Quanta 200) in low vacuum mode. From Table 1, among the three selected plants, the number of xylem vessels in the leaf veins of F. virens is the least and the xylem cross-sectional area is the smallest. H. patens ranks in the middle position while P. rubra has the largest number of xylem vessels and xylem cross-sectional area in the leaf veins.
Influence of the experimental conditions In order to investigate the influence of the experimental conditions on flow velocity
22
in leaf veins, the concentration of fluorescein and illumination intensity of LEDs were taken into consideration (Li et al., 2008). At the concentration of 0.1 g/L, the fluorescence of the water-fluorescein solution was too weak to be observed under the blue light source. So the concentrations of 0.3 g/L, 0.5 g/L, 0.7 g/L and 0.9 g/L were chosen for comparison. Fig. 5 shows the measured flow velocity in different leaf veins of the three selected plants at different concentrations of fluorescein. Each data was the mean flow velocity of at least three leaf veins, except for the midrib. The four leaves of the same plant at different concentrations were picked up on the same branch to reduce the influence of the diversity of different plant leaves. However, since the four leaves could not be the same entirely, it could be inferred that the influence of the concentration of fluorescein on flow velocity was inconspicuous. The difference among flow velocities at different concentrations may be caused by the diversity of different plant leaves even of the same species.
23
Fig. 5. The measured flow velocity in different leaf veins of the three selected plants at different concentrations of fluorescein: (a) F. virens; (b) P. rubra; (c) H. patens. The different concentrations of the water-fluorescein solution were 0.3 g/L, 0.5 g/L, 0.7 g/L and 0.9 g/L, respectively. To investigate the influence of the illumination intensity, the number of LEDs was controlled by turning on or off every LED independently. As the number decreasing 24
from 20 to 14, then to 8, symmetrically, the illumination intensity of LEDs decreased and the flow velocity decreased accordingly. Unlike that of the concentration of fluorescein, the influence of the illumination intensity of LEDs was so obvious that the recording time increased rapidly as the number of LEDs decreased. So, only a qualitative experiment was performed in this case. The results of the influence of the experimental conditions indicate that the adequate illumination intensity is important for the plant growth and development, especially for the water transport in leaf veins.
The concentration of the impurities in
water influences the water transport inconspicuously.
Differences among leaf veins and species Since the influence of the concentration of fluorescein on flow velocity was inconspicuous as shown in Fig. 5, Fig. 6 shows the mean measured flow velocity in different leaf veins of the three selected plants. As for sample F. virens, the flow velocity was the fastest and decreased with the decreasing of vein size, which was as expected by Hüve et al. (2002). The flow velocities both in the midrib and the secondary veins were almost the same for sample H. patens, and then it decreased with the decreasing of vein size. The slowest flow velocity was measured in sample P. rubra, in which the flow velocity in the secondary veins was faster than that in the midrib, and then it decreased with the decreasing of vein size. The results indicate that the flow velocities of different species can be quite different, which are in agreement with the anatomical 25
structures shown in Table 1. Comparing the three selected plant leaves, there may be some relations between the flow velocity and the structure of the leaf veins network as well as the natural characteristic of heat-resistance, which are to be investigated in the future work.
Fig. 6. The mean measured flow velocity in different leaf veins of the three selected plants.
Variation of the flow velocity in leaf veins With the algorithm mentioned above, the flow velocity in a selected leaf vein in a definite time interval can be measured and calculated quickly. By setting a constant time interval, the variation of the flow velocity in leaf veins can be investigated. Fig. 7a, Fig. 7c, and Fig. 7e show the variation of the flow velocity in midribs of the three selected plants for every 200, 1000, and 300 frames (25 fps), correspondingly. The figures show that, when the fluorescein flowed through, the flow velocity in the mid26
rib was the fastest at the beginning, then decreased with fluctuation. We guessed that on one hand the little fluctuation of the flow velocity might be caused by the wall roughness on the internal surface of leaf veins. On the other hand, the complicated network of the leaf veins consisted of a lot of branches and confluences, which might also have a great impact on the flow velocity. Fig. 7b, Fig. 7d, and Fig. 7f show the variation of the flow velocity in the secondary veins of the three selected plants. For each plant, three different secondary veins were selected for comparison, which were marked with the corresponding color on a little picture inserted in each figure. The zone and interval of frames measured and the serial number of the selected leaf vein were illustrated in each figure, too. Similar variation in the secondary veins was observed as that in the midribs. Furthermore, we noticed that a little difference existed among the three selected secondary veins of each plant, which might be caused by their different positions on the leaf veins network, as well as the different wall roughness on the internal surfaces, which is to be discussed in the future work. Additionally, it must be noted that the data in Fig. 7 was all measured under the same experimental condition, where the concentration of fluorescein was 0.5 g/L and the number of LEDs was 20.
27
28
Fig. 7. Variation of the flow velocity in: (a) the midrib of F. virens; (b) the secondary veins of F. virens; (c) the midrib of P. rubra; (d) the secondary veins of P. rubra; (e) the midrib of H. patens; (f) the secondary veins of H. patens. The abscissa "Frame" in 29
(a), (c) and (e) refers to the frame number of the corresponding experimental video. The frames intervals in (a), (c), and (e) are 200, 1000, and 300 frames (25 fps), respectively. In (b), (d), and (f), three different secondary veins of each plant are selected and marked with the corresponding color on a little picture inserted in each figure. The zone and interval of frames measured and the serial number of the selected leaf vein are illustrated in each figure, too. The data was all measured under the same experimental condition, where the concentration of fluorescein was 0.5 g/L and the number of LEDs was 20.
Verification of the flow velocity In Fig. 8, a verification of the flow velocity measured by the algorithm developed in this paper in a selected leaf vein is shown. Once a leaf vein was selected, the flow velocities on the given points of frame (e.g. 901, 1101, 1301, ..., 3101 in Fig. 8) were measured quickly by the algorithm within three different time intervals (zone of frames), respectively. For example, as in Fig. 8, the flow velocities of the green dash curve with triangle symbols were measured within 300 frames, respectively. The zones of frames were [751, 1051], [951, 1251], [1151, 1451], …, [2951, 3251]. The three dash curves match well with each other, which can verify the flow velocity measured by the algorithm developed in this paper.
30
Fig. 8. Verification of the flow velocity measured by the algorithm in a selected leaf vein. The given points of frame are 901, 1101, 1301, ..., 3101. The flow velocities of the blue dash curve with circle symbols were measured within 200 frames, respectively. The zones of frames were [801, 1001], [1001, 1201], [1201, 1401], …, [3001, 3201]. The flow velocities of the red dash curve with square symbols were measured within 250 frames, respectively. The zones of frames were [776, 1026], [976, 1226], [1176, 1426], …, [2976, 3226]. The flow velocities of the green dash curve with triangle symbols were measured within 300 frames, respectively. The zones of frames were [751, 1051], [951, 1251], [1151, 1451], …, [2951, 3251].
Comparison between the experimental and simulated flow velocity in the midrib Fig. 9 shows the comparison between the experimental and simulated flow velocity in the midribs of the three selected plants. The simulated data were extracted every 5 mm along the center axis of the midrib, which is similar to the experimental method performed by Hüve et al. (2002). The dash curves are polynomial fitting of the ex31
perimental data. Since the data of the flow velocity in the secondary veins were so much and so various, only the comparison of the flow velocity in the midrib was performed in this section. Obviously, the simulated flow velocity matches well with the polynomial fitting of the experimental flow velocity, which also verifies the new method of flow velocity measurement in leaf vein systems developed in this paper.
32
Fig. 9. Comparison between the experimental and simulated flow velocity in the midribs of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. The simulated data were extracted every 5 mm along the center axis of the midrib. The dash curves are polynomial fitting of the experimental data.
Discussion With the development of measuring technique, many new methods of flow velocity measurement have been come up with. Bio-speckle technique (Yokoi and Aizu, 2010), particle image velocimetry (PIV) (Hao et al., 2011; Soares et al., 2013), laser induce fluorescent (LIF) technique (Charalampous and Hardalupas, 2011) and tomograph technique (Maad et al., 2010) are representations of the flow visualization technique of a new generation. Whereas, methods of flow velocity measurement in leaf veins could hardly be found. The development tendency of the flow visualization technique is the combination of several technologies. In this paper, a simple but efficient method 33
was developed to measure the flow velocity, which was used as a quantitative description to reveal the regulation of water transport in leaf vein systems. This method combined the fluorescence labeling technology frequently used in bioresearch and the image-processing technology in computer realm. Comparing with the three methods mentioned above developed by other researchers (Garbet et al., 2002; Hüve et al., 2002; Soares et al., 2013), the new method developed in this paper is simpler in operation and the quantitative data obtained are intuitive and reliable. The experimental conditions are easy to get and closer to the natural conditions. Most importantly, the whole data of the same sample can be obtained at the same time, avoiding the errors caused by the diversity of different plant leaves even of the same species. The diversity of plants exists naturally, which influences the data of the experiments about plants unavoidably. In the experiments performed in this paper, Fig. 5 showed the diversity of different plant leaves of the same species while Fig. 6 showed the diversity of different species. From the data shown in Fig. 6, plants could be divided into three kinds according to the relation between the flow velocities of the midrib and the secondary veins. The sample F. virens might be more similar as the broad bean (Vicia faba L. cv. Scirocco) (Hüve et al., 2002) in some respects because their relations of the flow velocities among different leaf veins were the same. When concentrating on a selected leaf vein, Hüve et al. (2002) measured the flow velocity every 5 mm under a binocular microscope with a magnification of 2.5 to 5 times in his experiments. In a similar way, we measured the flow velocity of a selected leaf 34
vein by setting a constant time interval in our experiments. In the meanwhile, because of the invisible blocking or interruption in the inside of leaf veins, the fluorescein did not flow through every leaf veins in order from the petiole to leaf tip. In order to obtain the data more accurately, the leaf veins selected for data extraction were the first ones through which the fluorescein flowed in the experiments preformed in this paper. The results of the experiments preformed above indicate that the new method developed in this paper is very efficient for flow velocity measurement, which is of high practical value to be used as a quantitative description to reveal the regulation of water transport in leaf vein systems. The algorithm developed in this paper is verified to be efficient and has wider future applications, although there is still an imperfection that the calculation of the flow velocity is not completely automatic. The choosing of the threshold of the brightness needs an experienced judgement, which may bring in some subjective errors. Another highlight is that the simulation carried out in this paper, the boundary conditions of which are actually according to the natural conditions of a plant leaf, is especially different from the previous research. But there is a little deficiency that some parameters, such as the porosity of the part of mesophyll, the face permeability and the porous medium thickness of the porous-jump zone, were set depending on experience instead of the measured value, which will doubtlessly reduce the accuracy of the simulation. As for the future work, in order to understand more about plants' physiological 35
properties and promote the technique on bionics design, more flow and thermal characteristics such as the pressure drop (Cochard et al., 2004; Zwieniecki et al., 2002) and temperature distribution (Garbe et al., 2002) of the water transport in leaf vein systems will be investigated. The advantages of plants with heat-resistance are to be investigated in the following experiments. A growth chamber will be used for plant cultivation. The growing conditions can be changed artificially to investigate their influences on the characteristics mentioned above. Some relative anatomy experiments will also be performed to discover the internal mechanism of the macroscopic phenomena. Furthermore, the sharpness of the recording videos and the efficiency of the flow velocity measurement algorithm are to be promoted to achieve a higher accuracy of the experimental data in the future.
Acknowledgements The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (NO. 51375169 & 11572128). And we wish to thank Prof. Eleni Katifori for helpful discussions on the experiment.
References Al-Salim N, Barraclough E, Burgess E, Clothier B, Deurer M, Green S, et al. Quantum dot transport in soil, plants, and insects. Science of the Total Environment. 2011;409:3237-48. Bohn S, Magnasco MO. Structure, scaling, and phase transition in the optimal transport network. Physical Review Letters. 2007;98:1-4. Chapman S, Oparka KJ, Roberts AG. New tools for in vivo fluorescence tagging. 36
Current Opinion in Plant Biology. 2005;8:565-73. Charalampous G, Hardalupas Y. Method to reduce errors of droplet sizing based on the ratio of fluorescent and scattered light intensities (laser-induced fluorescence/Mie technique). Applied Optics. 2011;50:3622-37. Chen J, Wang Q, Lu M, Zhang H, Liu B. Application of quantum dots in plant molecule fluorescent labeling. Scientia Silvae Sinicae. 2010;46:153-61. Chen Y, Lin P, He Y. Velocity representation method for description of contour shape and the classification of weed leaf images. Biosystems Engineering. 2011;109:186-95. Cochard H, Nardini A, Coll L. Hydraulic architecture of leaf blades: where is the main resistance? Plant, Cell & Environment. 2004;27:1257-67. Corson F. Fluctuations and redundancy in optimal transport networks. Physical Review Letters. 2010;104:1-4. Garbe C, Schurr U, Jaehne B. Thermographic measurements on plant leaves. AeroSense 2002: International Society for Optics and Photonics; 2002. p. 407-16. Hüve K, Remus R, Lüttschwager D, Merbach W. Water transport in impaired leaf vein systems. Plant Biology. 2002;4:603-11. Hao P, Yao Z, Zhang X. Study of dynamic hydrophobicity of micro-structured hydrophobic surfaces and lotus leaves. Science China: Physics, Mechanics and Astronomy. 2011;54:675-82. He W. Water transport mechanism and simulation in plant. Xi'an: Xi'an University of Architecture and Technology; 2009. Hu Y, Li J, Ma L, Peng Q, Feng W, Zhang L, et al. High efficiency transport of quantum dots into plant roots with the aid of silwet L-77. Plant Physiology and Biochemistry. 2010;48:703-09. Katifori E, Szöllősi GJ, Magnasco MO. Damage and fluctuations induce loops in optimal transport networks. Physical Review Letters. 2010;104:1-4. Kizilova N. Long-distance liquid transport in plants. Proceedings of the Estonian Academy of Sciences. 2008;57:179-203. Kizilova NN. Construction principles and control over transport systems organization in biological tissues. Physics and Control, 2003 Proceedings 2003 International Conference: IEEE; 2003. p. 302-07. Lenk S, Chaerle L, Pfündel EE, Langsdorf G, Hagenbeek D, Lichtenthaler HK, et al. Multispectral fluorescence and reflectance imaging at the leaf level and its possible applications. Journal of Experimental Botany. 2007;58:807-14. Li B, Zhao Z. Research on leaf recognition and simulation. Information Science and Control Engineering 2012 (ICISCE 2012), IET International Conference on: IET; 2012. p. 1-4. Li H-j, He X-k, Zeng A-j, Liu Y-j, Cao S. Effect of environment on fluorescence analysis of fluorescein sodium. Journal of Jilin Agricultural University. 2008;30:49-51, 63. Li Y, Chi Z, Feng DD. Leaf vein extraction using independent component analysis. 2006 SMC'06 IEEE International Conference on Systems, Man and Cybernetics: 37
IEEE; 2006. p. 3890-94. Maad R, Hjertaker B, Johansen G, Olsen Ø. Dynamic characterization of a high speed gamma-ray tomograph. Flow Measurement and Instrumentation. 2010;21:538-45. Pang J-f, Yu Z, Liang Z, Yin C-c, Liu H, Pei D, et al. The application of QDs and its development prospect in plant. Journal of Agricultural Science and Technology. 2009;11:19-26. Price CA, Symonova O, Mileyko Y, Hilley T, Weitz JS. Leaf extraction and analysis framework graphical user interface: segmenting and analyzing the structure of leaf veins and areoles. Plant Physiology. 2011;155:236-45. Radha R, Jeyalakshmi S. An effective algorithm for edges and veins detection in leaf images. 2014 World Congress on Computing and Communication Technologies (WCCCT): IEEE; 2014. p. 128-31. Roth-Nebelsick A, Uhl D, Mosbrugger V, Kerp H. Evolution and function of leaf venation architecture: a review. Annals of Botany. 2001;87:553-66. Salleo S, Raimondo F, Trifilo P, Nardini A. Axial-to-radial water permeability of leaf major veins: a possible determinant of the impact of vein embolism on leaf hydraulics? Plant, Cell & Environment. 2003;26:1749-58. Soares R, Barbosa H, Braga R, Botega J, Horgan G. Biospeckle PIV (Particle Image Velocimetry) for analyzing fluid flow. Flow Measurement and Instrumentation. 2013;30:90-98. Xin X. Investigation on the mechanism and law for heat and mass transfer in hardy plant blades. Qingdao: Qingdao University of Science and Technology; 2013. Yokoi N, Aizu Y. Improvement of estimation parameters in bio-speckle blood flow imaging. Optical Review. 2010;17:230-38. Zheng X, Wang X. Leaf vein extraction using a combined operation of mathematical morphology. 2010 2nd International Conference on Information Engineering and Computer Science (ICIECS): IEEE; 2010. p. 1-4. Zwieniecki M, Melcher P, Boyce C, Sack L, Holbrook N. Hydraulic architecture of leaf venation in Laurus nobilis L. Plant, Cell & Environment. 2002;25:1445-50. Zwiewka M, Friml J. Fluorescence imaging-based forward genetic screens to identify trafficking regulators in plants. Frontiers in Plant Science. 2012;3:1-14.
38
Table list Table 1 Overview of the vein anatomy in the basal part of the three selected plants.
39
Figure legends Fig. 1. The flowing images of the leaves of the three selected plants during the experiment: (a) Ficus virens Ait. var. sublanceolata (Miq.) Corner; (b) Plumeria rubra L. cv. Acutifolia; (c) Hamelia patens. The tracer fluorescein (C20H12O5) which emitted orange light under the blue light source was flowing through the veins from the petiole (at the right side, out of the field of view). The images were extracted from the experimental videos recorded by a camera equipped with the orange and the yellow lightfilters under the blue light source. Fig. 2. Pictures of the experimental setup for the flow observation: (a) room lighting; (b) partial enlarged detail; (c) blue light source. The number of the blue LEDs was 10 at each side. Every lamp holder was flexible and could be turned on and off independently. The camera was set on a piece of foam and connected with a computer, which was put on another table separated from the experimental platform. Fig. 3. Images of H. patens in the process of the flow velocity measurement algorithm: (a) grayscale image; (b) binary image; (c) leaf veins image; (d) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame A; (e) the brightness (grey level) of each point on the selected leaf vein on the grayscale image of frame B. The abscissa in (d) and (e) is the number of each point on the leaf veins image (c), which is numbered in order starting from the base of the petiole. Fig. 4. 3D models of the leaves of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. There are two parts: the leaf veins and mesophyll. The cross section of the leaf veins was assumed to be rectangle, of which the thickness was 0.2 mm. And 40
the total thicknesses of the whole models were all 0.3 mm. Fig. 5. The measured flow velocity in different leaf veins of the three selected plants at different concentrations of fluorescein: (a) F. virens; (b) P. rubra; (c) H. patens. The different concentrations of the water-fluorescein solution were 0.3 g/L, 0.5 g/L, 0.7 g/L and 0.9 g/L, respectively. Fig. 6. The mean measured flow velocity in different leaf veins of the three selected plants. Fig. 7. Variation of the flow velocity in: (a) the midrib of F. virens; (b) the secondary veins of F. virens; (c) the midrib of P. rubra; (d) the secondary veins of P. rubra; (e) the midrib of H. patens; (f) the secondary veins of H. patens. The abscissa "Frame" in (a), (c) and (e) refers to the frame number of the corresponding experimental video. The frames intervals in (a), (c) and (e) are 200, 1000 and 300 frames (25 fps), respectively. In (b), (d) and (f), three different secondary veins of each plant are selected and marked with the corresponding color on a little picture inserted in each figure. The zone and interval of frames measured and the serial number of the selected leaf vein are illustrated in each figure, too. The data was all measured under the same experimental condition, where the concentration of fluorescein was 0.5 g/L and the number of LEDs was 20. Fig. 8. Verification of the flow velocity measured by the algorithm in a selected leaf vein. The given points of frame are 901, 1101, 1301, ..., 3101. The flow velocities of the blue dash curve with circle symbols were measured within 200 frames, respectively. The zones of frames were [801, 1001], [1001, 1201], [1201, 1401], …, [3001, 41
3201]. The flow velocities of the red dash curve with square symbols were measured within 250 frames, respectively. The zones of frames were [776, 1026], [976, 1226], [1176, 1426], …, [2976, 3226]. The flow velocities of the green dash curve with triangle symbols were measured within 300 frames, respectively. The zones of frames were [751, 1051], [951, 1251], [1151, 1451], …, [2951, 3251]. Fig. 9. Comparison between the experimental and simulated flow velocity in the midribs of the three selected plants: (a) F. virens; (b) P. rubra; (c) H. patens. The simulated data were extracted every 5 mm along the center axis of the midrib. The dash curves are polynomial fitting of the experimental data.
42