Infrared thermography in plant phenotyping for salinity tolerance in cereal

Xavier R.R. Sirault and Richard A. James

Author Affliations

CSIRO Plant Industry Canberra – High Resolution Plant Phenomics Centre

Overview

The recent advances made in the use of infrared thermal imaging (thermography) as a non invasive, high throughput technique for the screening of salinity tolerance in plants is reviewed. Taking wheat seedlings as an example, the methods and protocols used to impose a homogeneous salt stress to a large number of genotypes, as well as capturing infrared images of these genotypes and automatically processing the images are described in detail in this chapter. We also present the source code of the Matlab program applied to automatically identify plants and batch process IR images

Background

The growth of crop plants in response to salinity is initially reduced by the decrease in soil water potential (1). This is essentially a water stress that results from the osmotic effect of salts in the soil solution. This makes it harder for the roots to extract water from the soil, so is similar to that imposed by soil drying (2). Growth can also decrease as a result of the salts taken up by the plant to toxic concentrations. This “salt-specific effect” usually occurs later – over a period of weeks to months – as salts accumulate in the older leaves. This build-up of salt results in leaf injury and death and a subsequent reduction in the supply of assimilate to the growing regions (1, 3). Salt tolerance mechanisms in crop plants are generally described under three categories (4): 1) tolerance of high external salt concentrations in the soil, called “osmotic stress tolerance”, 2) control of salt uptake, and 3) tolerance of high internal salt concentrations in leaves, called “tissue tolerance”. Osmotic stress is the major cause of the reduction in growth and yield of cereals in saline soils. Genotypic variation for osmotic stress tolerance can be assessed by measuring stomatal conductance because it is likely that the factors controlling growth of plants under osmotic stress also regulate stomatal conductance. Stomatal conductance is reduced immediately with the onset of salinity (5), and follows a virtually identical response profile to that of growth (6). Additionally, over a timescale of days, the reduction in growth rate is often matched by a similar reduction in stomatal conductance (7-9). However, phenotyping for osmotic stress tolerance by measuring stomatal conductance is problematic. This is due to poor repeatability that largely results from stomatal sensitivity to small changes in environmental conditions. The precision and speed of porometry also affect the accuracy and reproducibility of measurements. An alternative approach for measuring stomatal conductance uses infrared (IR) thermography. Leaf temperature varies with transpiration rate (10), which is largely a function of stomatal conductance. As infrared measurements of leaf temperature correlate well with estimates of stomatal conductance (11-12), this technology can be used to assess osmotic stress tolerance in plants (11). Tolerance to osmotic stress is the only salt tolerance mechanism that is amenable for accurate and specific phenotyping using IR thermography. Infrared thermography can graphically display the surface temperature distribution of a plant by focusing the long-wave radiation emitted and reflected by that plant onto a temperature-sensitive detector called a micro-bolometer. The plant’s temperature determines how much radiation it will emit. Therefore, the measurement of thermal radiation emitted by the plant can be used as an indicator of its temperature. Screening for phenotypic variation in osmotic stress tolerance is contingent on the ability to measure the specific effects of osmotic stress on seedling temperature without the complication of other salinity-induced changes that may also impact on seedling temperature. This could include changes in leaf morphology, and the technique requires that measurements are taken before salt levels in the leaves build up to potentially toxic concentrations and before the age-associated decline in stomatal conductance (13). Taking these factors into consideration, IR thermography measurements should ideally be completed on young seedlings (leaf 2 – 3 stage for cereals), shortly after the final desired concentration of salinity is attained (3 – 5 d). Measurements on young seedlings have additional benefits such as more accurate IR measurements and greater high-throughput screening potential. Tolerance to osmotic stress is best assessed by simultaneously measuring the temperature of seedlings of the same variety treated with salt relative to non-salt controls. This overcomes complications due to any potential intrinsic variation in leaf temperature determinants such as stomatal conductance, waxiness or colour. It also overcomes the need for a temperature reference to determine the absolute temperature of the seedlings. Determining absolute leaf temperatures can be problematic due to a number of confounding factors such as changes in ambient temperature and humidity or reflected background radiation that could introduce errors into this measurement. In this chapter, we outline the process of growing plants in salt-stressed conditions suitable for early IR thermography measurements. The actual IR thermography measurement process is described for wheat and barley plants grown hydroponically along with details of the subsequent automated image capture and analysis protocols.

Materials/Equipment

Seed preparation Petri dishes (90 mm diameter). Filter paper (Whatman Grade 1 circles, 90 mm). 1% hypochlorite solution. Thiram 800 (or other suitable anti-fungal agent). Growing media 2 M CaCl2 stock solution. 5 M NaCl stock solution. Hoagland’s nutrient solution. 50:50 coarse river sand:perlite mix. Quartz gravel (granulation 8 – 12 mm). Aquarium pumps (>400 L / hr). Hydroponic containers. Acquiring thermal images Bronze tinted Polymethyl methacrylate plate (e.g. Perspex™ ca. 40 cm x 60 cm, PlastiX, Sydney). Infrared camera (ThermaCAM SC660 IR Camera (FLIR Systems Inc., Boston, MA, USA, or equivalent, i.e. thermal resolution better than 0.050 K). Firewire cable to connect camera to computer. Tripod on which to mount the infrared camera. Computer with a Firewire port and ThermaCam Researcher Professional software version 2.9 or above (FLIR Systems Inc.). Controlled-environment chamber with temperature and relative humidity control (e.g. Conviron chamber model PGC20 with CMP6050 Control System, Controlled Environments Ltd., Winnipeg, Canada). Processing thermal images Matlab software version 2009a or above (The Mathworks, Natick, MA, USA). The Imaging Processing Toolbox version 6.3 or above (The Mathworks, Natick,. MA, USA). Microsoft Excel version 2007 or above (Microsoft. Microsoft Excel. Redmond, Washington: Microsoft, 2007). Methods Seed preparation Select seeds for uniform shape and weight (for wheat and barley, select uniform seeds within a 5 mg weight range). Surface sterilise seeds with 1% hypochlorite or a fungicide such as Thiram (1.4g/L). Seed fungal infections on young seedlings can impair plant performance. For example, Na+ uptake to the shoot can increase 2-3 fold with a fungal infection on the seed and crown of a young salt-treated wheat seedling. Imbibe (1-2 h de-ionised H2O) and germinate on moist filter paper (infused with antifungal agent such as Thiram 800) in Petri dishes for two days at 4C. Plant only uniformly germinated seeds. Growing media There are a number of different ways to grow and treat plants with salinity. The key considerations to determine which method to use relate to whether the plants are nutritionally well supplied, are well drained, aerated and not waterlogged, and importantly, the constancy and homogeneity of the salinity treatment to which they are exposed. Below are described two different growing methods that when used carefully, account for these key issues. Supported hydroponics Supported hydroponics is preferable over hydroponics because it provides a supportive matrix around the roots. This prevents root breakage when solutions are changed or with bubbling for aeration. This matrix can vary. We use quartz gravel (14-15), whereas other groups have used polycarbonate pellets (16). The supported hydroponics system we use consists of a number of 40-L containers, each holding 40 square pots of 6.5 cm width and 15.8 cm depth. These contain quartz gravel and sit on a perforated stage (2 cm high) within the container. This setup sits upon a 60-L container of the identical width and length. This contains a 50-L reservoir of a nutrient and salt solution that is periodically pumped into the upper chamber and drained back into the reservoir for continual reuse. The frequency and duration of sub-irrigation events is controlled by a timer (Figure 1). Plant germinated seeds (one per pot) to a depth of about 1 cm. Carefully cover with fine quartz gravel, avoiding damage to either the roots or the small coleoptile. Water gently and cover pots with wet towels and insulation covers to maintain moist conditions during seedling establishment. Remove covers when coleoptiles emerge (24 – 48 h) and commence sub-irrigation with ¼ strength Hoagland’s nutrient solution at a frequency of 30 min. Increase to 1/2 modified Hoagland’s nutrient solution (P reduced from 1 mM to 100 μM) 7 days after emergence (DAE). Solutions should be changed weekly and pH monitored and adjusted regularly because the nutrient solution is weakly buffered. For the salt treatment, at 8 days after emergence, add 25 mM NaCl to the irrigation solution once or twice daily until the required salt concentration is reached. We typically use 100 – 150 mM NaCl for screening wheat and barley seedlings. The slow incremental addition of salt concentrations is critical to avoid plasmolysis of root cells and to allow the plant time to osmotically adjust so as not to lose turgor and wilt (Munns 2002). Supplemental Ca2+ (CaCl2) should be added with the commencement of the salt treatment to maintain a Na+:Ca2+ ratio of 15:1, based on the final concentration of the salinity. If supplemental Ca2+ is not added, other cations such as Na+ out-compete for uptake sites, resulting in Ca2+ deficiency. For the control plants, maintain a duplicate hydroponics system containing only 1/2 modified Hoagland’s nutrient solution (P concentration reduced from 1 mM to 100 μM). Sand/perlite media Pack pots with a 50:50 mix of coarse river-sand:perlite, leaving a 2 cm space at the top of the pots for irrigation events. The pots should be much taller than they are wide to assist with drainage and help prevent waterlogging (Passioura 2006). Fine mesh is required at the base of the pot to prevent loss of sand. It may be necessary to provide a larger matrix layer (e.g. large quartz gravel) in the base of the pot to assist with drainage. Flush pots several times with tap water to wash out any fine silt or dirt. Drainage to approximate field capacity is important after every flush or irrigation event to prevent waterlogging. After the initial free drainage, extra water can be removed by tilting pots at 45o for 5 – 10 min. Additionally, wicks can be inserted in the base of the pots or pots can be placed on layers of towels, depending on how open the base of the pot is for good contact of soil to towel layer. Plant germinated seeds 2 cm deep, one per pot. Sprinkle with water and cover pots to reduce evaporation and help maintain a humid environment. Remove covers when the coleoptiles begin to emerge. Keep the top layer of soil moist over the next few days to ensure root growth and seedling establishment. Commence gentle watering (~5 mL) with ¼ Hoagland’s nutrient solution at 3 (DAE) and increase to 1/2 Hoagland’s nutrient solution at 7 DAE. At 8 days after emergence, flush the pots with 25 mM NaCl, followed by adequate drainage to approximate field capacity. Once or twice daily, incrementally increase the concentration of the salt solution used for flushing by no more than 25 mM NaCl, until the desired NaCl concentration is achieved. Supplemental Ca2+ (CaCl2) to maintain a Na+:Ca2+ ratio of 15:1 based on the final concentration of the salinity is also advised when using the sand/perlite media. Pots should be watered daily to excess and then adequately drained (as above) with the final salt solution (also containing supplementary Ca2+). This will minimise the build-up of salts in the rhizosphere as the plants continue to take up water. A duplicate set of pots should be set up and watered daily with a 1/2 Hoagland’s nutrient solution for the control.

Units, terms, definitions

Procedure

The conditions in the controlled environment chamber should be set to maximise the IR signal. Set light intensity between 500 – 1000 μmol m-2 s-1; optimal for cereals such as wheat and barley. Similarly, by acquiring thermographs of seedlings in an atmosphere with a VPD ranging from 1.2 to 1.6 KPa, one can maximise transpiration rate while maintaining lower stomatal conductance. This keeps the seedling slightly cooler than the background, while maintaining higher effective sensitivity even at high salinity levels. Care is needed to prevent a build-up of CO2 when taking IR measurements in a confined space such as a controlled environment chamber. We have found that the CO2 concentration can increase by at least 50% over a period of 30 min, due to expired air from an operator (James et al. 2008). This led to a decrease in stomatal conductance and transpiration rate by about 30%, resulting in -hotter’ plants. To overcome this issue, the IR camera should be controlled remotely and the operator needs to minimize the time spent in the controlled environment chamber between readings. Age and growth stage of seedlings is important for the accuracy of IR measurements. Wheat and barley seedlings with 21/2 – 3 leaves is an ideal growth stage so that the leaves fall in the same orientation, perpendicular to the IR camera. On more advanced plants, some of the leaves go out of focus as they fall away from the perpendicular plane with the IR camera. Consequently, they give a different IR signal to those leaves in focus. Once the desired salt concentration is achieved, allow 2 – 3 d for the plants to equilibrate with the new conditions before commencing IR readings. Acquiring thermal images Thermal images of seedlings should be acquired between 11h00 and 14h00 in the controlled-environment chamber described above; both stomatal conductance and transpiration tend to be relatively constant over this time period. The background is particularly important when taking IR thermographs of plants. We used a bronze tinted acrylic (Perspex™) plate as a background. The benefit of Perspex™ is that it is readily available, it provides a homogenous background as it does not transmit IR radiation (it appears opaque in the IR) and it has a lower emissivity that results in an apparent temperature ca. 2 oC -hotter’ than the air temperature. Consequently, it provides a good contrast against the cooler transpiring seedlings. Note that at a high salt concentration, from and above 150 mM NaCl, the temperature difference between the plant and the background starts to shrink, eventually to less than 0.5C, so the image of the salt-treated plant will start blending into the image of the background. This means that the image-processing algorithm will have to be modified. Remember that the ability to automatically separate the pixels representing the plant from the background depends on a consistent difference in apparent temperature between the two. A highly reflecting surface, i.e. with very low emissivity, (e.g. a polished metal) is unsuitable as a background for thermography study. For acquisition of IR thermographs, a ThermaCAM SC660 infrared camera (FLIR systems Inc, MA, USA) was used. The SC660 infrared camera uses a focal plane array, un-cooled micro-bolometer with a 640 480-detector array that gives a very good spatial resolution of the plant – good spatial resolution is critical when one looks at small seedlings. This model has also a very good thermal resolution (0.045 K) and accuracy (∓1%). The camera is placed at a distance of ca. 90 cm from the plant and is left in the growth chamber for about two hours before starting the IR measurements series. This allows the optics of the IR camera to reach thermal equilibrium with the air temperature. A default emissivity value of 0.95 is entered in the camera settings. Emissivity of most vegetation is usually deemed to be ca. 0.95. The main problem in radiation thermometry is that the emissivity of the target is usually not exactly known. For plant material, values ranging from 0.94 to 0.97 have been measured. More importantly, the emissivity of a surface will vary with the angle of emission and the temperature of this surface; this is why care is taken to account for variability of leaf angles of the various plants. By screening seedlings rather than more advanced plants, leaf angles are no longer randomly distributed, providing a comparable surface to the IR camera. Similarly, the distance of the camera to the subject needs to be measured and entered into the camera settings together with air temperature and relative humidity. Anything between the surface being measured and the detector will remove some of the thermal radiation in some region of the spectrum. This error is called “Absorption error”. It can be corrected by the software if RH is measured before image acquisition. Place pots side by side in front of the Perspex™ plate (Figure 2) and avoid overlapping leaves. Pots should be ~10 cm apart. Allow a set amount of time of about 1 – 2 min before taking an IR image. During that time, focus on the image using the cameras controls for a clear IR thermograph (Figure 3), and make small adjustments to the orientation and location of pots as necessary. In this set-up the relative difference in leaf temperature due to salinity is measured. We do not try to determine absolute leaf temperature. As indicated previously, several factors (e.g. surface emittance, ambient temperature variation, humidity, background thermal radiation) introduce errors in the measurement of absolute leaf temperature. By measuring relative temperature differences between salt stressed and control seedling simultaneously, most of these problems are overcome. IR Images are stored directly in a folder on the computer hard drive. Processing thermal imaging Convert IR images into Matlab format files (x.mat extension) using ThermaCAM Researcher Pro or FLIR Quickreport software. Load files into Matlab software release 2011a (The MathWorks, Natick, MA, USA) and check that the matrices are read as matrices of raw temperature in unit of Kelvin. Transform matrices into 8-bit resolution, grey-level images (Figure 4) using the custom-built function -tmp2img.m’. – The function scales and normalizes the temperature data in a grey level intensity image; the Matlab code for the function -temp2img’ is indicated below: function I = temp2img(T) % function I = temp2img(T) NaV Function to convert an input matrix of temperatures (in Kelvin) to an 8 bit grey level image suitable for viewing. %% Inputs: T = matrix of temperatures % Outputs: I = 8 bit grey level image representing temperatures Tmax = max(T));  % Calculate the maximum temp in matrix Tmin = min(T(); % Calculate the minimum temp in matrix I_double = (T – Tmin)/(Tmax-Tmin); % Scale and normalise temp matrix I = uint8(floor(255*I_double)); % convert from double to uint8 Use the Matlab script (which labels the custom-built function, -tmep2img.m’) to automatically identify the seedlings in the 8-bit grey image. The code was written to batch process hundreds of IR images at a time. The detection algorithm relies on Otsu’s automatic thresholding method (11). This assumes only two populations of pixels in an image, i.e. background and foreground. If more than two populations of pixels are present, other methods are available to create a binary mask from the 8-bit grey level image. However, this will require the involvement of an image analyst. It is also possible to manually capture the average temperature of a leaf by “drawing” the contour of the leaf by hand using the FLIR Quickreport software (delivered with the IR camera at no cost). Obviously, this is a very tedious task, non-amenable to automation and subjective. Use the Matlab script (which labels the custom-built function, -tmep2img.m’) to automatically identify the seedlings in the 8-bit grey image. The code was written to batch process hundreds of IR images at a time. The detection algorithm relies on Otsu’s automatic thresholding method (11). This assumes only two populations of pixels in an image, i.e. background and foreground. If more than two populations of pixels are present, other methods are available to create a binary mask from the 8-bit grey level image. However, this will require the involvement of an image analyst. It is also possible to manually capture the average temperature of a leaf by “drawing” the contour of the leaf by hand using the FLIR Quickreport software (delivered with the IR camera at no cost). Obviously, this is a very tedious task, non-amenable to automation and subjective. A binary matrix is then computed to create a mask or silhouette of the plant (Figure 5) in the image. The mask is used to derive the temperature of the two seedlings in the thermograph by multiplying the original image by the mask values according to arithmetic array rules, i.e. element by element multiplication (Figure 6). The Matlab script then computes the average temperature for each seedling and reports the difference in temperature between the control and salt-treated seedling within a MS Excel spreadsheet

Other resources

Notes and troubleshooting tips

– If more than two populations of pixels are present, other methods are available to create a binary mask from the 8-bit grey level image. However, this will require the involvement of an image analyst. It is also possible to manually capture the average temperature of a leaf by “drawing” the contour of the leaf by hand using the FLIR Quickreport software (delivered with the IR camera at no cost). Obviously, this is a very tedious task, non-amenable to automation and subjective.

Links to resources and suppliers

Literature references

1. Munns R (1993) Physiological processes limiting plant growth in saline soil: some dogmas and hypotheses. Plant Cell Environ 16:15 -24 2. Epstein E (1980) Responses of plant to saline environments. In: Rains DW, Valentine RC, Hollaender A, (eds) Genetic engineering of osmoregulation. Plenum Press, New York 3. Munns R, Schachtman DP, Condon AG (1995) The significance of a two-phase growth response to salinity in wheat and barley. Aust J Plant Physiol 22:561-569 4. Munns R, James RA, Lauchli A (2006) Approaches to increasing the salt tolerance of wheat and other cereals. J Exp Bot 57:1025-1043 5. Rahnama A et al (2010) Stomatal conductance as a screen for osmotic stress tolerance in durum wheat growing in saline soil. Funct Plant Biol 37:255-263 6. Passioura JB, Munns R (2000) Rapid environmental changes that affect leaf water status induce transient surges or pauses in leaf expansion rate. Aust J Plant Physiol 27:941-948 7. Fricke W et al (2004) Rapid and tissue specific changes in ABA and in growth rate in response to salinity in barley leaves. J Exp Bot 55:1115-1123 8. Yeo AR, Caporn SJM, Flowers TJ (1985) The effect of salinity upon photosynthesis in rice: Gas exchange by individual leaves in relation to their salt content. J Exp Bot 36:124-148 9. Sibole JV et al (1998) Role of sodium in the ABA-mediated long-term growth response of bean to salt stress. Physiol Plant 104:299-305 10. Fuchs M (1990) Infrared measurement of canopy temperature and detection of plant water stress. Theor Appl Climatol 42:253 -261 11. Sirault XRR, James RA, Furbank RT (2009) A new screening method for osmotic component of salinity tolerance in cereals using infrared thermography. Funct Plant Biol 36:970-977 12. Jones HG (1999) Use of thermography for quantitative studies of spatial and temporal variation of stomatal conductance over leaf surfaces. Plant Cell Environ 22:1043-1055 13. James RA et al (2002) Factors affecting CO2 assimilation, leaf injury and growth in salt-stressed durum wheat. Funct Plant Biol 29:1393-1403 14. Munns R, James RA (2003) Screening methods for salinity tolerance: a case study with tetraploid wheat. Plant Soil 253:201-218 15. James RA et al (2008) Genetic variation in tolerance to the osmotic stress component of salinity stress in durum wheat. Funct Plant Biol 35:111-123 16. Genc Y, McDonald GK, Tester M (2007) Reassessment of tissue Na+ concentration as a criterion for salinity tolerance in bread wheat. Plant Cell Environ 30:1486-1498 17. Passioura JB (2006) The perils of pot experiments. Funct Plant Biol 33:1075-1079 18. Munns R (2002) Comparative physiology of salt and water stress. Plant Cell Environ 25:239 -250

Health, safety & hazardous waste disposal considerations

!

Leave a Reply