Hi,
I have data of abundance of some tardigrada species and some variables like substrate, altitude and month of sampling. I want to carry out a DCCA with Past for evaluate what relationship exist betwen the variation in species composition within samples and the environmental variables. How can I organize my data for this?
Canonical correspondence analysis is a multivariate technique that maximizes the relationship between a series of dependent variables and a series of independent variables. This relationship is made on the basis of multiple regression.
(more information).
It is very common that two matrices are used; one of dependent variables (for example, species abundance at X's sites) and another of independent or explanatory variables (for example, climatic variables of X's sites).
For our example we will use this type of data.
For our first matrix we have the data from 10 sites, with the abundance of 5 species for these sites, we will save it as a .CSV format file in the folder we are working on, with the name "species.csv", like this:
sitios,sp1,sp2,sp3,sp4,sp5
1,1,0,0,1,0
2,0,0,0,0,0
3,0,0,0,0,0
4,1,8,1,5,1
5,1,0,3,1,0
6,3,5,0,1,0
7,117,288,141,100,14
8,0,0,0,0,1
9,0,0,0,0,0
10,1,2,26,3,2
For the second matrix, we have the data from the same 10 sites with their respective values of 5 climatic variables, in a .CVS format file in the same folder above, with the name "clima.csv", like this:
sitios,v1,v2,v3,v4,v5
1,28.4,80.1,30.7,54.3,32.4
2,26.8,85.8,19.4,23.7,51.2
3,25.5,85.2,56.2,66.2,60.0
4,27.3,90.4,43.3,52.1,53.0
5,24.6,83.8,12.5,22.6,69.0
6,24.5,83.8,12.5,25.0,66.3
7,26.0,81.2,32.0,77.0,65.0
8,25.1,86.6,11.3,45.2,83.1
9,27.5,84.5,23.7,41.2,55.7
10,26.0,94.1,20.0,80.0,60.0
Having our data matrices ready, we will get to work on R:
The first thing we must do is download the packages we need to do our canonical correspondence analysis. These packages are "fields" and "CCA"; for this we execute the following commands:
> install.packages ("fields")
> install.packages ("CCA")
# We download our package from the desired mirror (I use "USA CA2"), to our computer.
> library (fields)
> library (CCA)
#Loaded to our work area in R, the packages just downloaded.
Once we have the packages loaded, we introduce in R, as objects, the two matrices that we believe are in our working folder:
> clima species species climate ñero gala gala
After this we only have to proceed to make the graphic results of the canonical correspondence analysis.
> img.matcor (number, type = 2)
#We obtain the result of the correlation function between and within the matrices, for example in the following graph we can observe a high correlation for the species matrix, but in the matrix of climatic variables we see a less marked correlation, likewise, we can see the cross-correlation diagram (that is, of both matrices).
barplot (gala $ cor, xlab = "Dimension", ylab = "Canonical correlations",
names.arg = 1: 5, ylim = c (0,1))
#We made the explanatory graph of each dimension to which the variables used were reduced, since it is on this that it will be decided with which dimensions or "components" to work.
plt.cc (gala)
#Finally, using the two dimensions that have the most "explanatory power", we make our correspondence diagram which will serve to group the sites, the variables and the correlated species (for example, in the following graph we can see the high similarity between the sites 9,3 and 2, and as the species are very close (very correlated))
Canonical correspondence analysis is a multivariate technique that maximizes the relationship between a series of dependent variables and a series of independent variables. This relationship is made on the basis of multiple regression.
(more information).
It is very common that two matrices are used; one of dependent variables (for example, species abundance at X's sites) and another of independent or explanatory variables (for example, climatic variables of X's sites).
For our example we will use this type of data.
For our first matrix we have the data from 10 sites, with the abundance of 5 species for these sites, we will save it as a .CSV format file in the folder we are working on, with the name "species.csv", like this:
sitios,sp1,sp2,sp3,sp4,sp5
1,1,0,0,1,0
2,0,0,0,0,0
3,0,0,0,0,0
4,1,8,1,5,1
5,1,0,3,1,0
6,3,5,0,1,0
7,117,288,141,100,14
8,0,0,0,0,1
9,0,0,0,0,0
10,1,2,26,3,2
For the second matrix, we have the data from the same 10 sites with their respective values of 5 climatic variables, in a .CVS format file in the same folder above, with the name "clima.csv", like this:
sitios,v1,v2,v3,v4,v5
1,28.4,80.1,30.7,54.3,32.4
2,26.8,85.8,19.4,23.7,51.2
3,25.5,85.2,56.2,66.2,60.0
4,27.3,90.4,43.3,52.1,53.0
5,24.6,83.8,12.5,22.6,69.0
6,24.5,83.8,12.5,25.0,66.3
7,26.0,81.2,32.0,77.0,65.0
8,25.1,86.6,11.3,45.2,83.1
9,27.5,84.5,23.7,41.2,55.7
10,26.0,94.1,20.0,80.0,60.0
Having our data matrices ready, we will get to work on R:
The first thing we must do is download the packages we need to do our canonical correspondence analysis. These packages are "fields" and "CCA"; for this we execute the following commands:
> install.packages ("fields")
> install.packages ("CCA")
# We download our package from the desired mirror (I use "USA CA2"), to our computer.
> library (fields)
> library (CCA)
#Loaded to our work area in R, the packages just downloaded.
Once we have the packages loaded, we introduce in R, as objects, the two matrices that we believe are in our working folder:
> clima species species climate ñero gala gala
After this we only have to proceed to make the graphic results of the canonical correspondence analysis.
> img.matcor (number, type = 2)
#We obtain the result of the correlation function between and within the matrices, for example in the following graph we can observe a high correlation for the species matrix, but in the matrix of climatic variables we see a less marked correlation, likewise, we can see the cross-correlation diagram (that is, of both matrices).
barplot (gala $ cor, xlab = "Dimension", ylab = "Canonical correlations",
names.arg = 1: 5, ylim = c (0,1))
#We made the explanatory graph of each dimension to which the variables used were reduced, since it is on this that it will be decided with which dimensions or "components" to work.
plt.cc (gala)
#Finally, using the two dimensions that have the most "explanatory power", we make our correspondence diagram which will serve to group the sites, the variables and the correlated species (for example, in the following graph we can see the high similarity between the sites 9,3 and 2, and as the species are very close (very correlated))
The above replies are completely agreeable and informative, along with them i suggest you to go through this article which details you about the basics and importance of CCA
Best wishes
Article Canonical correspondence analysis and relate multivariate me...
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology