Canonical Correspondence Analysis in PAST (v2)
HTML-код
- Опубликовано: 11 сен 2024
- Doing CCA with PAST with some discussion of other ordination methods. (Staying with v2 because there are a few bugs in v3 that affect graphing.) PAST CCA has a permutations test for significance but the results look a little odd, so I didn't discuss it. Use at your own risk.
Dear Keith, Can you please explain me how we can use Aspects (Directions) in CCA, as it is one of the important environmental variable.
Dear Keith,
Very useful videos on PAST. They are helping me a lot with me MSc dissertation project. I was wondering, is there a function to attribute different colour/shape points for species in different groups in a CCA plot, e.g to distinguish between rare and common species?
Nicholas Hill Thanks! In PAST, not that I am aware of. You can, however, save as a metafile and edit with a graphics package.
Dear Keith, thank you for shedding your light on this topic. As I am currently applying CCA on a vegetation dataset with 11 environmental variables, I do have some questions. One would be: together with the eigenvalues, the impact in % of each eigenvalue (= impact of the corresponding environmental gradient) on the variance in the species composition is calculated in %. In my case all 11 environmental gradients neatly add up to 100%. However, does this suggest that this 100 % relates to te EXPLAINED variance in species composition? Because it would be too much luck if the 11 environmental gradients I choose would explain ALL the variance in the species composition, wouldn't it?
A second question is caused by the correlations between the different environmental gradients in my case: for example, elevation and temperature correlate -0.99034: they are almost the same (although opposed) gradient. However, in the result of my analysis, elevation explains 39.1% of the (explained?) variance in species composition, while temperature explains 32.5%. Together with the other environmental gradients/ variables, they add up to the 100% mentioned before. Isn't that odd? Since elevation and temperature correlate -0.99034, they represent (almost) the same (explained) variance, so it seems their impact in % is counted double.
I would be very happy if you would be so kind to tell me what you think about this.
Sincerely, Serge Mooijman
Thank you for this explanation, it was really useful! I have two questions though.
1) What information does the permutation test provide?
2) I read somewhere else that "gradients that are too short may manifest linear responses and may be better handled by redundancy analysis" What is your opinion about it, since apparently is the case of my data?
Thanks in advance
In general, a permutation test involves randomising (permutating) the data, then recalculating the test statistic, usually 999 times (adding the value for the unrandomised data gives 1,000 values). The value for the real (unrandomised) data is compared to these and if the real value is greater than 95% of the randomised value, the test is significant and the null is rejected. The null is that the real value could have occurred by chance.
"Response variables show unimodal distributions across objects. If dealing with a sites × species (or OTUs) matrix, this suggests that a sampling gradient must be long enough to allow the increase and decrease of a given species or OTU across the sites sampled. Gradients that are too short may manifest linear responses and may be better handled by redundancy analysis (RDA), although CCA may also handle linear relationships."
Above is one of the assumptions of CCA. I have no further comment. You may need to try different methods and see which is most useful for your data.
Dear Keith, are the data in the permutation procedure randomized by swapping the sample labels with the values within each environmental gradient column?
how i do in past Discriminant analysis . . ? ?
Hello Keith. Can one use CCA when one wishes to look at the possible effect of vegetation variables on the abundance of animals? The vegetation is biotic in nature but also act as part of the environment (abiotic). Are there other ways to differentiate the types of data that can be used? Thank you.
That would be fine. The vegetation can be considered environmental data for other organisms.
Thank you Keith.
Dear Keith. Thank you for this tutorial, it helps me a lot! But can you tell me how to put a legend of symbols (or colors) in the plot? I need it to make the graphic complete.
Thank you very much.
+Andrea Alvito Hi Andrea New video on how to do this.
how i do in past Discriminant analysis . . ? ?