In this lengthy post, I include discussion of our experience with
leaf area measurement, and a collation of the responses I received
to my query to the ecophys-list and forest-list last year. I
thank all of you who responded, and hope this summary will be of
help to many of the list-members.
My experience has been that the Li-Cor conveyor belt area meter is
unsuitable for spruce needles due to their small size. The parallax
problem gets worse as the needles get thinner, and changes with
orientation of the needles to the long axis of the light source.
When forced to use this system with spruce needles, we have sent the
needles through 6 to a dozen times, and hoped that the average was a
reasonable estimation (but it probably was an underestimate). This
prompted our original quest for a better solution. Our technical
wizard discusses our current status in the area of needle area
measurement below.
- Jim Stewart
Our solution to the conifer projected needle area problem has been
to use a flatbed scanner with Sigmascan Pro Image Analysis software.
The main problem, as others have mentioned in this discussion, is
the shadows cast behind the needles by the scanner lamp as it sweeps
beneath the scan bed. These shadows make it difficult to separate
needle from background on the basis of gray-scale intensity. We've
used a back-light, like Regent's system, except that ours is a
small, homemade, light table such as you use for sorting slides.
The light table has 3 - 18-inch fluorescent tubes and uses "opal
glass" available from your local glass shop as a diffuser (total
materials cost = $100). I built it with a hinge to sit neatly on
our HP scanner. It eliminates the problem of shadows, but the
scanner does pick up the fluorescent lamps' flicker (60 Hz?) as it
makes its scan. The flicker shows up as regular bands of bright and
slightly-less-bright light in the image. Contrast is much improved,
though, and we have had no trouble in seeing a sharp separation in
intensity between needle and non-needle pixels. Apparently there
are much higher-frequency fluorescent light ballasts available which
will eliminate the flicker problem, but I haven't got around to
trying them yet.
To analyze the scanned images we bought Sigmascan Image analysis
software from Jandel Scientific (www.jandel.com). It was
inexpensive ($700 CDN) and appeared quite user-friendly but we had
no end of trouble getting it to work on our Windows `95 Pentium
system (which it was designed for). Tech support was terrible -
they simply would not get back to us until we threatened to send
their software back. We received a demo version of their more
sophisticated Sigmascan Pro which works better, but still has a
number of frustrating glitches. To date, we've "made-do" with this
software, which actually would be really straight-forward and
versatile for doing all sorts of morphometric work - if Jandel ever
releases a version that works properly. Keep your receipt if you
buy this product (Sigmascan Pro = $1400 CDN). Optimas
(www.optimas.com) or Image-Pro (www.mediacy.com) look like better
bets but are considerably more expensive (>US$3000). We'd be happy
to hear about a cheap Windows or DOS-based analysis system if anyone
knows of one (come on, you developers - all it really has to do is
count dots....)!
- Ken Stadt
The following is a summary of the replies received from colleagues,
and members of the forest-list and env-phys-list. Since I didn't
ask if people wanted their names attached to their comments in this
broadcast posting, I erred on the side of discretion and left them
off. If anyone would like to be put in touch with a particular
commentator, I will arrange that if you send me a note.
-----------------------------
My tendency would be to go for Regent's McNeedle (or WinNeedle)
system if you have the money. Right now, the system gives you
projected surface area as well as number and length of needles.
These last two figures are great to have if you want to obtain
actual surface through volume displacement. If all goes well, next
summer, we will be working with Regent Guay to expand the capability
of the software in order to simplify the measurement of total needle
area. I give you this info with some reserve however. We have not
gone any further into this for the past few years. We used dry mass
as the standardization factor in all our past Pn measurements.
However, our new work will involve needle area determination and I
will therefore have to plunge once more into the prickly realm of
black spruce needle surface determinations! But again, time and
funding will determine how far we can go with all our wild ideas!
-------------------------------------------------
We have used a LAI-3100 area meter from LI-COR to measure the
surface area of basidiocarps of a fungus and to measure deciduous
leaves too. Accuracy of the digital machine seemed to be far
sufficient for our purposes. I would even say that measurements were
really accurate (+/- 0.03 cm2). Surface area of basidiocarps ranged
from 0.1 to 1.5 cm2.
-------------------------------------------------
NIH IMAGE: I have used the public domain NIH Image (see below) in
conjunction with a normal flatbed scanner. The problem with normal
scanners is that needles cast shadows and a thresholding value must
be determined and if possible, calibration with another leaf area
meter (e.g. Licor) is recommended. The advantage of using NIH is
that it's free and you can add your own macro programs to obtain
more detailed measurements. To my knowledge, NIH is only available
for Macintosh and PowerMacintosh computers.
METHOD: The one-sided projected leaf area was measured using a
flatbed scanner. The needles were placed on the scanner with their
upper surface facing down and properly aligned whenever possible for
acurate measurements. Needle analysis was performed on a Power
Macintosh computer using the public domain NIH Image program
(Rasband, 1995). The program was used to estimate for each sampled
seedling, the needle's average leaf area (ALA, mm2), average leaf
length (ALL, mm), and average leaf width (ALW, mm). ALL and ALW
were approximated assuming that the ellipsoidal shape of the needles
was highly eccentric, and resembled that of a long, narrow
rectangle. A special macro (program) was used with NIH to estimate
ALL and ALW and is available to anyone interested.
REFERENCE: Rasband, W. 1995. NIH Image documentation V. 1.59. U.S.
National Institutes of Health, (Program available from the Internet
by anonymous ftp from zippy.nimh.nih.gov or on floppy disk from
NTIS, 5285 Port Royal Rd., Springfield, VA 22161, part number
PB93-504868), 89.
Another alternative is to buy Win/MacNeedleue from Regent Instruments
(regent at riq.qc.ca). Their program performs several measurements and
you usually need a more expensive scanner (with overhead lighting).
-------------------------------------------------
One company offers a system which is used by collegues of mine at
dept. of Molecular Biology of Plants, Czech Academy of Sciences in
Ceske Budejovice is a software is produced by Delta -T Scan Devices
Ltd 128 Low Road Burwell, Cambridge CB5 oej, England tel: +44 638
742922 FAX: +44 638 743155
-------------------------------------------------
Dave Larsen and I (John Kershaw) developed a system based on a
hand-held scanner. There is an article in Tree Physiology in 1992
describing the system and comparing it with the LiCor system and a
complex image analysis system. For about 250.00 you can get a
system that is just as accurate as the 8000-12000 systems you have
probably been looking at. We have software to read the scanned
files which is free. Take a look at the artical and if interested
I can send you the software.
-------------------------------------------------
I've had excellent luck with NIH image, which is freeware for the
Macintosh and I think (soon?) for Windoze.
-------------------------------------------------
Optimas software would do it for you. We have developed fully
automated image analysis system based on this software. By the way,
I am pretty sure that we can help you in measurement of leaf area
should you prefer to hire us rather then developing expensive IA
system. Please do not hesitate to contact me if you need more
information. Sincerely, Anatoly Portnoy <portnoya at schuller.com>
-------------------------------------------------
A colleague of mine measures photosynthesis and transpiration in red
spruce. He uses an old LICor area meter -- the kind with a
transparent conveyor belt. Apparently the sensitivity can be
adjusted to measure the rather small areas of needles from single
spruce shoot. After years of removing needles fron the shoot axis
with razor blades, we found that you can remove needles by dipping
the shoot in liquid N2; they fracture off at the sterigmata. Kind of
fun for the first half hour or so. Needles are then sandwiched in
Saran wrap and run through the meter. Freezing doesn't affect needle
area.
Somewhere I have some papers co authored by Bill Smith at Wyoming
involving shoot architecture and needle area. The species would have
been Picea engelmanni, and maybe subalpine fir (?). I don't know how
he measured area, but I can dig the refs up if you really need them.
Also, NIH image is freeware that can be used at least to do the
areas (using a scanned image) if not the lengths and widths. I use
it to assess injury in samples of spruce needles. It is quite
pwerful, and has a nice macro language that really helps in
automating analysis. I think it runs on Macintosh only, although
recently i picked up a rumor that there is or will be a PC version.
I don't have the URL handy, but if you do a net search on NIH Image,
you'll find it.
-------------------------------------------------
I have used the Decagon or Delta-T video-cam equipment and found it
very finnicky and frustrating. Slight adjustments of contrast and
hue make large differences in area, and you can't just seem to fix
these settings once when calibrating. However, I know that everyone
in Forest Sciences at UBC is so disgusted with this machine that
they'd probably give it to you!
I have also used the LiCor leaf area meters (i forget the models,
both "portable", and counter-top models). They do not have the
ambiguities of calibration problem, but need the plastic conveyor
belts to be efficient. Those belts eventually get gummed up and
dirty, so need some maintenance. I would consider the big
counter-top model with variable resolution (unfortunately you have
to manually install a different lens) as the current standard for
leaf area measurement.
I have not tried the WindDendro/MacDendro technology out of Quebec,
but I like the idea of using a standard flatbed scanner that you can
use for other purposes too. I've played around with some spruce
and birch leaves on a scanner, but have not yet found software to
simply count the black pixels! (may have to buy the Dendro software
after all...)
-------------------------------------------------
I can recommend two leaf area machines that you might be interested
in: Mac/WinNeedle: Developed By Instruments Regent. the e-mail
address is regent at riq.qc.ca This is quite a good system that works
on Mac or IBM although I have never personnally used it. Others who
have used both Decagon and Mac/WinNeedle say that WinNeedle is
better (more automatic, nice and glossy etc). Pierre Bernier has a
system at Forestry Canada. I gave your name/address to Regent Guay
who developed the equipment and said that he would contact you. I
used his optical image analysis system for the tree ring analyses
that I did for my doctorate. It's a small company based here in
Quebec, that has alot of potential. They have sold other systems in
B.C. (Literature is available in English).
Decagon AgVision Monochrome System, Rot and leaf analysis. We have
had this since about 1990 and has been faithful but it could do alot
more. Simple but crude. 1-800-755-2751.
-------------------------------------------------
from REGENT Instruments: -clip- We're very responsive and we can
also help you to get good images with your scanner. Our product
support many models of scanners via the TWAIN standard. They also
support video cameras. They run on Mac or PCs with Windows (3.1 or
95).
One last aspect on Win/MacNEEDLE is that we are comitted to include
standard practices for needle measurements. We've heard of the way
surface area is measured in the Boreas project and we're discussing
with other people about the possibility to include it in NEEDLE as
well as new methods. Needle as it is now can measure projected area,
length and with of needles and we're looking at a new method for
surface area. There is room for experimentation here.
Regent Guay REGENT INSTRUMENTS INC. 165 Fatima Ave. Quebec, Qc. G1P
2C7 CANADA Tel: 418-683-3064 FAX: 418-683-8396 Internet:
regent at riq.qc.ca
-------------------------------------------------
Leaf area measurement in conifers is a rather complex task, and how
to accomplish that really depends on what the measurements are
needed for. There are many commercially available instruments to
measure shoot (or branch) projected area. I have up to now used DIAS
from Delta T (Cambridge, U.K.). However, these measurement systems
are rather expensive. What one really needs is a video camera and a
multimedia card for PC. As already pointed out, there exists some
shareware for analysing bitmaps or one may like to develop their
own software - one needs only to count the black pixels. With
moderate effort it is possible to reduce the cost at least by factor
10.
BUT, it is very IMPORTANT to calibrate the system properly: the
number of pixels you get considered 'black' strongly depends on
resolution, contrast etc. With commercially available devices there
is ususally some way to play around with 'threshold', and it is also
explained in manual, how calculated area should perform to consider
'threshold' "correct". However, it does not work!! With conifer
shoots you will always have some half-shade (i.e. penumbra)) and
semihalf-shade there, which is also likely to be different for
different shoots. Thus, I have always calibrated the measurement
system with objects of known size and corrected the estimates of
DIAS from these.
Another question is, if it is possible to calculate total needle
area of shoot from the measurements of shoot projected area.
Unfortunately, the answer is NO. The number of needles, and
accordingly their total area, per unit stem length depends on ligh
conditions the shoot was exposed to during its formation. You may
want to check:
Niinemets,U; Kull,O (1995): Effects of light availability and tree
size on the architecture of assimilative surface in the canopy of
Picea abies: variation in shoot structure. Tree Physiol. 15(12),
791-798.
Sprugel,DG; Brooks,JR; Hinckley,TM (1996): Effects of light on shoot
geometry and needle morphology in Abies amabilis. Tree Physiol.
16(1/2), 91-98.
Now, for a more physiologically oriented work it is sometimes
necessary to express the physiological quantities on either
projected or total needle surface area. IT IS NOT THE SAME! Again,
how much total needle area would respect to a certain amount of
projected needle area depends on long-time light conditions the
needles were exposed to:
Niinemets,U; Kull,O (1995): Effects of light availability and tree
size on the architecture of assimilative surface in the canopy of
Picea abies: variation in needle morphology. Tree Physiol. 15(5),
307-315.
Sprugel,DG; Brooks,JR; Hinckley,TM (1996): Effects of light on shoot
geometry and needle morphology in Abies amabilis. Tree Physiol.
16(1/2), 91-98.
Also, projected needle area as measured by DIAS is sensitive to
needle structure as affected by light conditions (refs. above). I
should therefore recommend to use direct morphological measurements
on needles to derive their total and projected surface areas rather
than to use some video or electronic-optiacal device.
-------------------------------------------------
The following items were extracted from Canopy-list by one of the
contributors
-------------------------------------------------
How would YOU measure the surface area of a Douglas-fir branch tip?
Assume the base diameter of about 1 cm, length of about .75m, and
approximately 50-60 primary and secondary twiglets with needles
extending from the main twig axis. Am aware of Pike et al methods,
i.e measuring area of many small cylinders (or cones) plus subset
of needles, any other suggestions? This is for the purpose of
determining available macrolichen substrate.
-------------------------------------------------
I have heard of a technique where the branch is dunked in a
container of small glass beads, which are subsequently washed off
and weighed. Assuming a uniform single layer coating, it should be
straight foward to determine the total area (calibrate the
technique using more easily measured surfaces). This might have
been in documentation from Li-Cor. Not sure if it could be used for
macrolichen studies.
-------------------------------------------------
For many poeple the term "leaf area density" was unfamiliar; they
were asking for the difference between LAI (index) and LAD
(density). Leaf area density may be seen as many measurements of
leaf area index taken at constant intervals along the tree (say
each meter H) minus the leaf area index measured at height below
(H-1 meters). Thus, this gives:
square meter (H) - square meter (H-1) = square meter per cubic meter
= Leaf area density between H and (H-1)
It is therefore possible to deduct LAD from many measurements of
LAI made along the canopy. My new request (a more specific one) is
to gather some informations about LAI recorded at different heights
and for North American trees (deciduous and coniferous). I thank
you...
-------------------------------------------------
At the recent Canopy conference in Sarasota, there was discussion
about how to measure leaf area lost to herbivory. NIH Image was
mentioned. I have used Image as a leaf area meter (and am working
on it as a particle counter: seeds, ovules, insects). It is very
easy to use and it is free! It works with color too, so you can
look at smut infections, leaf miners, etc. It is only available for
Macintosh computers, so DOS-jocks need not apply. Here is the scoop
from the user manual:
IMAGE: Updated Versions and Bug Reports
Updates to Image are available to Internet users via anonymous ftp
from zippy.nimh.nih.gov. Those without Internet access can get
updates from many Macintosh bulletin boards and user group
libraries. A reasonably current version, including Pascal source
code and example images, is available from any of the following
sources:
1) From a friend. The Image program, including source code and
documentation, is public domain and may be freely copied,
distributed and modified. However, if you modify Image, please
update the about box before distributing your version of the
program.
2) Via anonymous FTP from zippy.nimh.nih.gov[128.231.98.32]. Enter
"anonymous" as the user name and your e-mail address as the
password. The /pub/nih-image directory contains the latest version
of Image (nih image154_fpu.hqx or nih-image154_nonfpu.hqx),
documentation in Word format (nih-image154_docs.hqx), and complete
Think Pascal source code (nih image154_source.hqx). The directory
/pub/image/images contains sample TIFF and PICT images. The
directory /pub/image/image_spinoffs contains versions of Image
extended to do FFTs (ImageFFT), fractal analysis (ImageFractal), and
to support quantitative evaluation of cerebral blood flow, glucose
metabolism, and protein synthesis (Image/MG). There is a README file
(0README.txt) with information on the file formats used.
3) Library 9 (Graphics Tools) of the MACAPP forum on CompuServe.
Source code is in Library 6 of the MACDEV forum.
4) Twilight Clone BBS in Silver Spring, MD. The Clone has 16 lines
on sequential rollover, starting with 301-946-8677. To guarantee a
V.32 connection, call 946-5034. Image is cul1ently available at no
charge from the Twilight Clone.
5) Subscribe to the NIH Im~lge mailing list by sending a message
containing the line "subscribe nih-image <your name>" to
listserv at soils.umn.edu. Next obtain a list of the available NIH
Image archive files by sending an "index nih image" command to
listserv at soils.umn.edu. These files can then be retrieved by means
of a "get nih-image filename" command. The files are Binhexed and
broken into chunks less than 32K in size. The NIH mailing is
maintained by the Soil Science Department at the University of
Minnesota.
6) NTIS (National Technical Information Service), 5285 Port Royal
Road, Springfield, VA 22161, phone 703-487-4650, order number
PB93-504868 ($100 check, VISA, or Mastercard). Both the
zippy.nimh.nih.gov FTP site and the Twilight Clone BBS are likely to
have newer versions of Image than NTIS.
Bug reports and suggestions are welcome, as are connections or
additions to this manual. The author (Wayne Rasband) can be reached
at the following electronic mail addresses:
Internet: wayne at helix.nih.gov AppleLink: wayne @ helix.nih.gov
@ internet# CompuServe: >INTERNET: wayne at helix.nih.gov
-------------------------------------------------
I have used the technique of a surface bead coating on several
occasions to obtain surface area determinations of macrolichens and
bryophytes. I have found that chromatography beads (eg. Poropak R)
have a much higher surface adhesion and produce a much better
monolayer (which is critical to the success of this technique (eg.
Can. J. Bot. 67:167-176). This technique produces some very
interesting data on the surface area of non-vascular plants in
canopies, eg. in tropical montane forest (Can. J. Bot 69:2122-2129)
and is described by Larson and Kershaw (1976) Can. J. Bot.
65:182-191. Cheers, Darwyn Coxson, University of Northern British
Columbia
-------------------------------------------------
On the subject of measuring surface areas of branch tips ... I've
been thinking about a similar topic: the fractal surface of a
branch. It seems that if you use these beads, then the area you get
may depend on the size of your beads. This would be very interesting
(to me!). The protocol being plot the log of the surface area of the
branch tips as a function of the log of the bead size. Intuitively
this would match up with one of Mandlebrot's (sic?) methods of
estimating the fractal dimension of a surface. Wouldn't it? Any
comments?
As as aside, if you could asign a dimension (fractal or otherwise)
to open space what would you assign? and what of a flat planer,
simple leaf surface? And how about the airspace within a crown? Is
that simply 3-dimensional? Is it more or is it less?
Back to my first paragraph and direct measuring of fractal
dimension, this bead method would appea to have significant
advantages over the 1985 Nature paper by Morse, someone else, and
Lawton about arthropod body sizes and the fractal dimension of
vegetation.
-------------------------------------------------
Why not try it with glass beads for the smallest then using
something like "styrafoam" beads when you get up to a few mm? Seems
like the method should work, the only problem being a mechanical
one of getting beads made of a suitable material. Could be a pretty
straightforward technique.
-------------------------------------------------
Speaking of fractal measures of vegetation, there is a USDA/USFS
publication called "The Fractal Forest: Fractal Geometry and
Applications in Forest Science" that has recently been published.
You can call 612-649-5000 to order this free publication. The code
for the publication is GTR-NC-170
James D. Stewart Telephone: (403) 492-6827
Dept. of Renewable Resources Facsimile: (403) 492-1767
University of Alberta E-mail: james.stewart at ualberta.ca
