Using CDendro to analyze tree growth with climatic trends

[Sarah Brewster]

I am working on a project that involves comparing white spruce growth with climatic patterns such as temperature and precipiation trends. I'm looking for some advice on how to do this using the CDendro program. I collect three white spruce cores from 43 different sites. I want to create a mean growth curve for each site using the curves from the three cores collected. I want a curve that is a representative sample for the whole stand of trees. How do I use CDendro to create a mean curve and then save that curve as its own file? Once I have a representative curve I want to compare that to temperature and precipation trends. Is there a way to import climate data into CDendro to do the comparison, or could I export the the data from my curves into Microsoft EXCEL or another statistics program?

Thanks for any help you can offer,
Sarah

[Lars-Ake]

I am working on a project that involves comparing white spruce growth with climatic patterns such as temperature and precipiation trends. I'm looking for some advice on how to do this using the CDendro program. I collect three white spruce cores from 43 different sites. I want to create a mean growth curve for each site using the curves from the three cores collected. I want a curve that is a representative sample for the whole stand of trees. How do I use CDendro to create a mean curve and then save that curve as its own file? Once I have a representative curve I want to compare that to temperature and precipation trends. Is there a way to import climate data into CDendro to do the comparison, or could I export the the data from my curves into Microsoft EXCEL or another statistics program?

Thanks for any help you can offer,
Sarah

Hello Sarah!
CDendro is mainly intended for crossdating and chronology building - not for doing the mathematics of climatic analysis!
Accordingly the mechanisms for creating a mean value out of a collection of cores (in this case three cores in each collection) are very much ad hoc (just creating mean values from simply (!) detrended data) and not based on scientific investigation on how to best create such mean values to mirror the climatic response of those three trees.

Anyhow, to use CDendro to roughly do what you want, this is the way:
Put your three cores (synchronized) into a new collection and save that collection as a .fil file or as a .rwl file.
It will make things easier if you can date your collection before you save it.

Then see that the radio button NegExp for detrend is selected in your collection and press the Create mean value sample button.

This will open a new window with your mean value. Save that as a .wid file (Save ring width data as...)

Then close all windows.

Start with the next collection of your next site and do the same operation ending up with a .wid file.

Continue this process for all your sites.

At last, create a new collection and add all your .wid files to that collection. Use the menu-command: Collections/Add to this collection.
If you have all your .wid files properly dated, all the added members will be properly synchrononized when you add them to your new collection.
Check the synchronization by running the command "Test towards rest of collection" (a button in the collection window).
If a member of the collection is not in syncronization, right-click on it and adjust its offset. Do the checking again until it all seems to be correct.
Then save your collection.

See that the radio button No detrending is selected and press the Create mean value sample button again.
A mean value of all your 43 members is now available in a new window.
If you save it as a .wid file, that is quite a suitable format to add into e.g. EXCEL.

Be aware that these manipulations are done quite roughly - mean values created without much statistical analysis behind it.
Though this method works very well for creating mean value curves used for crossdating.

--------------------
An alternative to the above method, would be to put all the members of you 43 collections into a huge collection.
Then check that all members are in synchronization as described above.
Then use the menu command Collections/Write collection in Tabular format.
That will export the data to a .txt file which you may manipulate in .e.g. EXCEL.
That way you will then have total control over the mathematics done on your data.
Then you have only used CDendro for administering your data and checking its synchronization.
-----------------------
Another alternative to the first method, would be to put all your data into that huge collection with all members.
See that it is synchronized and then create a mean value directly out of that with NegExp detrend selected.
I doubt that you will se a difference between this third method and the first method.

Though for scientific work on climate matters I recommend the second method where you have total control on all the mathematics!

Best regards/Lars-Ake

[taxelson]

Hi Sarah,

Using the detrend function you need to be aware of its mechanism. It will just try to find a negative exponential shaped trend, and if finding one it will "straight" the actual curve according to that. If not it will instead divide each ring width just with the mean of them all. In both cases the result will be curves with the mean value 1 (mm) (100 if saved as 3-digit standard decadal-file).

I am not familiar with White spruce but in Norwegian spruce (PCAB) around here (60°N 15°E) it is not unusual that the very oldest rings are narrow according to depression, and some decade later it quickly becomes fast growing (according to neighbor trees are cut or blown down) and than, from that maximum it starts to decline according to a normal "neg exp shape". That kind of complicated trends will not be managed by CDendro but cause CDendro using just a plain mean value line for "detrending". That will of course be very bad if the output curve is to be used for climatological studies.

By using the "Dito + Neg Exp detrending curve" radio button in the curve display mode, you can check the shape of the detrending curve, which CDendro intends to use, for each sample.

One way to get rid of a detected, useless (useless for climatological studies but maybe not useless for basic cross dating purposes), part of a measurement series, is to cut it away using the "Create sample from block" function. It is available in the Curve display mode. (Click on the curve to get the offset position of the selected point, and put the received (start and) end points in suitable fields before pressing the "Create sample from block" button). Than you can save the kept block (automatically opened in a new window) as a sample file. If the "parent file" is a .pos-file it is a good idea to save also the block as a .pos-fil in order to keep the opportunity to check it again in CooRecoorder if you like.

(Lars-Åke, It would be nice if there was an explanation on how to do this cropping process in the manual, with pedagogical screen shots! (or is it somewhere, already?)

/Torbjörn