I am not sure what you are trying to do with such a few data using either of two multivariate statistics. First remember; multivariate statistical ordination analysis such as (PCA or CCA) aims at ordinating the data in such a way that you can extract two major conclusions about: (1) the number of clusters if any; and (2) the tendency or patterns of the data. CCA also correlates the clusters and patterns of the data with some exogenous variables. For example, in your data you may want to understand which plant species are well correlated using the three chemicals with two replicates. If so, you may see in your analysis (PCA or CCA) one cluster of two or may be all three species indicating they have the same concentrations for all three chemicals (Phenolic, Flavonoids, IC50). In this case, all three species are usually graphed on the same quadrant of a two dimensional analysis (e.g., PCA1 vs PCA2). If there is any cluster; that is each plant species (three dots in the graph) is well separated from each other then the concentration of chemicals (Phenolics, Flavonoids, and IC50) ARE COMPLETELY DIFFERENT BETWEEN PLANT SPECIES. The tendency, if any, usually explains which plant species have the maximum or minimum concentrations of chemicals. The major problem with PCA is the horseshoe effect that hides somehow this tendency and this is a simple justification why to use CCA. CCA also describes as I describe before, the correlation of the concentrations versus some other variables of the plant species, such as biomass or ash content, density, LAI, etc. Both PCA and CCA can be easily conducted in SPSS, although I am not familiar with this computer statistical software as I usually use SAS, it is a simple procedure that the same program may help you using the help icon in the program. Be sure which matrix are you using because sometimes it gives to other results; that is graphing for examples plant species x plant species and gives the concentrations of chemicals in the figure (Phenolics, Flavonoids, IC50) which are also interpretable for understanding better your data and analysis.

Should I be of further help, please do not hesitate to contact me at your earliest convenience.

Kind Regards

Sincerely

Jose Navar, Ph.D OSU

I think PCA is more preferred in such a scenario 

I am not sure what you are trying to do with such a few data using either of two multivariate statistics. First remember; multivariate statistical ordination analysis such as (PCA or CCA) aims at ordinating the data in such a way that you can extract two major conclusions about: (1) the number of clusters if any; and (2) the tendency or patterns of the data. CCA also correlates the clusters and patterns of the data with some exogenous variables. For example, in your data you may want to understand which plant species are well correlated using the three chemicals with two replicates. If so, you may see in your analysis (PCA or CCA) one cluster of two or may be all three species indicating they have the same concentrations for all three chemicals (Phenolic, Flavonoids, IC50). In this case, all three species are usually graphed on the same quadrant of a two dimensional analysis (e.g., PCA1 vs PCA2). If there is any cluster; that is each plant species (three dots in the graph) is well separated from each other then the concentration of chemicals (Phenolics, Flavonoids, and IC50) ARE COMPLETELY DIFFERENT BETWEEN PLANT SPECIES. The tendency, if any, usually explains which plant species have the maximum or minimum concentrations of chemicals. The major problem with PCA is the horseshoe effect that hides somehow this tendency and this is a simple justification why to use CCA. CCA also describes as I describe before, the correlation of the concentrations versus some other variables of the plant species, such as biomass or ash content, density, LAI, etc. Both PCA and CCA can be easily conducted in SPSS, although I am not familiar with this computer statistical software as I usually use SAS, it is a simple procedure that the same program may help you using the help icon in the program. Be sure which matrix are you using because sometimes it gives to other results; that is graphing for examples plant species x plant species and gives the concentrations of chemicals in the figure (Phenolics, Flavonoids, IC50) which are also interpretable for understanding better your data and analysis.

Should I be of further help, please do not hesitate to contact me at your earliest convenience.

Kind Regards

Sincerely

Jose Navar, Ph.D OSU

You have very few data . I suggest you to use standart anaova procedure for each flavonoid. BY this way you can test the variation among plants and between extraction methods for each phenolic compound. Or paired-t test may also performed for extraction methods. You can use the manual "http://www.academia.dk/BiologiskAntropologi/Epidemiologi/PDF/SPSS_Statistical_Analyses_using_SPSS.pdf" for the abovementioned statistics.