Soil Microbial Biovolume and C Determination

Epifluorescence Microscopy and Image Analysis: Protocol for Soil Microbial Biovolume and C Determination

Materials

1. Buffer: 0.05 M Na 2 HPO 4 (7.8 g L -1 ) in 0.15 M NaCl (8.8 g L -1 )adjusted to pH 9.0.
2. DTAF stain: 2 mg (5-[4,6-dichlorotriazin-2-yl] aminofluorescein) in 10 mL buffer. Prepare fresh daily.
3. Calcofluor M2R (Fluorescent Brightener) stain: 2 mg mL -1 in water. Stain is stable at room temperature for 1 month.
4. Filter stains, buffer and wash water through 0.2 µm filter.

Slide Preparation

1. Homogenize 10g soil in 190 mL filtered waterfor 1 min at full speed in a Waring blender.
2. Allow coarse particles to settle for 30 sec. With a wide-borepipette remove a bulk (5 mL) sample to a capped test tube.
3. Add 0.1 mL formalin (40% aqueous formaldehyde) as a preservative.
4. Vortex this bulk sample before subsampling to prevent further sedimentation.
5. Place 4 µL drops of soil suspension onto each 6 mm diam.?well? of printed slides (Bellco toxoplasmolysis slides). Spread the suspension uniformly across the well with the pipette tip without touching the tip to the slide surface. Prepare 5 replicate smears of each soil suspension.
6. Allow the smears to air dry completely, this fixes the organisms to the slide. Prepare two slides for each sample; one will be stained for bacteria, the other for fungi.

Staining

Bacteria

1. Flood each smear (well) with 8 µL DTAF stain.
2. Store slides in a container with wet tissue (100% RH) for 30 min.
3. Remove excess stain by placing slides in a series of staining jars containing buffer, 3 changes (30 min each) and finally water (30 min).
4. Air dry slides flat.
5. Add a small drop of low-fluorescence immersion oil (Cargille type FF recommended) to each smear and cover the slide with a 50 × 25 mm cover slip.

Fungi

1. Flood each smear with 10 µL Calcafluor M2R stain.
2. Stain in a covered container with wet tissue (100% RH) for 2 h.
3. Rinse by soaking slides in water in a staining jar for 30 min, 3 times.
4. Air dry slides flat.
5. Add small drops of low fluorescence immersion oil (Cargille type FF recommended) to each well and cover slide with a 50 × 25 mm cover slip.

Biovolume estimation

Bacteria

All bacteria are non-selectively stained green by this procedure. Observe the bacteria using a fluorescence microscope with a 63 x oil-immersion objective and 1.6 x zoom factor. Excitation filter 450-490 nm, dichroic 510 nm and suppression < 515 nm. Collect 3 images from each smear (3 × 5 = 15 images per sample) (Princeton Instruments cooled CCD camera with 512 × 768 pixels each representing 0.089 × 0.089 µm at this magnification). Measured soil bacteria average 0.9 × 0.4 µm, which corresponds to about 10 × 5 pixels. The software procedure is set to discard any objects represented by fewer than 5 pixels (smaller than about 0.2 × 0.2 µm). The bacteria are detected, counted and measured by image analysis (IPLab Spectrum Software, Apple Macintosh PPC computer).

The volume ( bv ) of each bacteria is calculated from measurements of the large, l , and small, d , diameters of each cell by assuming that the cells are prolate spheroids.

Equation for biovolume of bacteria

Fungi

Observe the calcofluor M2R (blue) stained fungi under epifluorescence with a 10x objective. Excitation 340-380 nm, dichroic 400 nm and suppression < 430 nm. Fluorescent dyes usually fail to stain heavily melanized hyphae which may be common or prevalent in some soils. The stain detects polysaccharides and does not discriminate living from dead hyphae.

Collect 3 images from each of 5 smears. Pixels represent 0.9 × 0.9 µm at this magnification. Hyphal fragments are detected and their length and diameter measured by image analysis. Objects with a plan area of less than 100 pixels (81 µm 2 ) are discarded, as are objects with a “radial standard deviation of < 35” – this is a measure of the variance in diameter of the object in different directions, which serves to exclude objects which are not long and thin, including most fungal spores. Hyphal volume for each fragment is calculated assuming the hyphae are simple cylinders of length l and diameter d .

Equation for hyphal volume of Fungi

Mass of soil per image (field of view)

The initial suspension (200 mL) contains 10 g moist soil, each smear contains 4 µL of the suspension. Thus each smear contains:

Equation for mass of soil per suspension g soil per smear

If the diameter of the smear is 6 mm, the area is (9)(pi) mm 2 .

The area of the image is xy mm 2 , where x and y are the dimensions of the counting area in millimeters.

The mass of soil per field ( s ) is given by:

Equation for mass of soil per image g soil per image

The sum of the volumes of bacteria ( bv )or fungal hyphae ( hv ), divided by the appropriate mass of soil ( s ) represents biovolume per gram soil (µm 3 g -1 ).

Calculation of biomass

The biovolume is converted to biomass C using conversion factors for C contents per unit volume. We use 200 fgC µm -3 for bacteria and 150 fgC µm -3 for fungi. These factors represent “best compromise” values from the available literature.

Image processing procedures

Image processing steps to measure bacterial and fungal hyphal biovolume in soil smears using IPLab Spectrum v 3.0 image analysis software (Macintosh PPC) are shown below in the form of ?scripts? which are executable lists of processing steps. The software has the capability to repeat the script for each of a list of image files and can in this way automatically analyze numerous images as a batch process. One externally programmed function, the filter?Tophat?(author: D. Harris) , was written in C and incorporated into the IPLab Spectrum software as an external function.

Bacteria

Action	Comment
Set variable	Width of ?Tophat brim? 1
Set variable	Width of ?Tophat? 15
Set variable	Height of ?Tophat? 100
Start	label to mark start of looping section
Open	Opens image file from supplied list
Show image
Duplicate window	Copies original image, we work on the copy
Show image	Duplicate
Linear filter	convolution with the 3 × 3 kernel. 0,-1,0,-1,4,-1,0,-1,0 (approx. Laplacian)
Linear filter	convolution with the 3 × 3 kernel. 3,5,3,5,8,5,3,5,3 (approx. Gaussian s = 1)
Tophat	?horizontal? pass detects peaks in intensity profile, subtracts background. Result in new window
Rename window	renames result of Tophat – peaks h
Change window	back to copy of original
Rotate and scale	Rotates copy of original 90 degrees
Tophat	?vertical? pass detects peaks in intensity profile, subtracts background. Result in new window
Rename window	renames result of Tophat peaks v
Rotate and scale	rotates peaks v -90 degrees
Image arithmetic	Maximum value of peaks h and peaks v, result to peaks h which now contains the net result of peaks detected by both ?horizontal? and ?vertical? passes of Tophat. Background is zero
Dispose window	peaks v
Dispose window	Copy of original
Change window	peaks h
Segmentation	segments image at threshold intensity value 1
Change widow	original image
Transfer attributes	Copies overlay of segmented image (detected features) onto original image
Measurement options	labels each measurement with image name, erases segments excluded from measurement, fills holes in segments
Set measurements	Collect area, perimeter, major and minor axes for each detected feature, excludes those not meeting set criteria – area > 5 < 200 pixels
Measure segments	Measure features matching set criteria
Dispose window

Dispose window|Clean up| |loop|go to start and get another image from list| |end|

Fungi

Action	Comment
Set variable	Width of ?Tophat brim? 1
Set variable	Width of ?Tophat? 10
Set variable	Height of ?Tophat? 100
Start	label to mark start of looping section
Open	Opens image file from supplied list
Show image
Duplicate window	Copies original image, we work on the copy
Show image	show copy
Median filter	Sets each pixel to the median value of the 9 pixels in its 3 × 3 neighborhood. Noise reduction
Tophat	?horizontal? pass detects peaks in intensity profile, subtracts background. Result in new window
Rename window	Tophat h
Change window	copy of original
Rotate and scale	rotate copy of original 90 degrees
Tophat	?vertical? pass detects peaks in intensity profile, subtracts background. Result in new window
Rename window	Tophat v
Rotate and Scale	rotate Tophat v -90 degrees
Image arithmetic	Maximum value of Tophat h and Tophat v, result to Tophat h which now contains the net result of peaks detected by both ?horizontal? and ?vertical? passes of Tophat. Background is zero
Dispose window	Tophat v
Dispose window	Rotated copy
Change window	Tophat h
Median filter	Sets each pixel to the median value of the 9 pixels in its 3 × 3 neighborhood. Noise reduction
Duplicate Window	Marr-Hildreth – will be mask after edge enhancement.
Change data type	signed 16 bit – allows negative values
Show image
Linear filter	convolution with the 3 × 3 kernel. 0,-1,0,-1,4,-1,0,-1,0 (approx. Laplacian)
Linear filter	convolution with the 3 × 3 kernel. 3,5,3,5,8,5,3,5,3 (approx. Gaussian s = 1)
Point function	<10 —> 0, =>10 —>1, set negative and small positive pixel values to zero, others to 1
Change data type	revert to unsigned 16 bit
Image arithmetic	Marr-Hildreth x Tophat h – retains Tophat h within Marr-Hildreth mask. sets others to zero
Segmentation	segments image at threshold intensity value =>1
Dispose window	Marr-Hildreth
Change window	Original image
Transfer attributes	Copies overlay of segmented image (detected features) onto original image
Measurement options	labels each measurement with image name, erases segments excluded from measurement, fills holes in segments
Set measurements	set criteria – area > 100 pixels, radial standard deviation > 35
Measure segments	Measures area of each detected feature matching criteria
Set measurements	no criteria
Modify segments	erode each segment to a single connected line of pixels (skeletonize)
Measure segments	area of skeleton estimates length
dispose window	original
loop	return to start
end

The Top Hat transform operates on each row of pixels in the image, first in the normal orientation of the image, then on the image rotated 90 degrees. The Top Hat is a hat-shaped mask which is moved along the intensity profile of each row of pixels. At each pixel position the Top Hat is raised or lowered to the level of the maximum intensity under the?brim?. Those pixels with values lower than the crown of the hat are set to zero in a new image. Pixels with values greater than the crown of the hat are given values in the newimage equal to the original intensity minus the level of the hatbrim. The effect of the filter is to copy only sharp bright peaks to the new image, eliminating small or low frequency fluctuations in the intensity profile. This background subtraction enables the new image to be segmented at a constant threshold, typically zero(Fig. A1). The parameters of the Top Hat, width and height, define the characteristics of the detected features. Whilst a three dimensional Top Hat could be used to detect bacteria-like features in one pass, the two dimensional Top Hat, two pass process, has the advantage of detecting filaments such as hyphae in addition to compact features like cocci and short rods.

Picture of Top Hat transform

Figure: Top Hat transform