Choosing factors for principle components analysis


I am running a PCA on a dataset that includes 43 water wells, each with up to five descriptive factors associated with it (temp, distance to significant location, depth, number of taxa, volume of flow). I am using XLSTAT to perform the analysis.

Running the PCA with the whole dataset, the first two factors accounts for 71% of the variation in the data (if I am interpreting the result "axes F1 and F2: 71%" correctly). I understand that the PCA by definition uses uncorrelated variables, but I'm not sure if using something like number of taxa (a result of the study rather than a way to describe the well) and volume of water (set by the well owner) should be included. Additionally, when I run the PCA with only three descriptive factors (temp, distance to significant location and depth) then the first two factors account for 91% of the variation.

My question is whether it is more valid to run the PCA with all five of the descriptive factors included or just with the three. Note that with the three factors, every result in the correlation table is significant at 0.05 (which makes me question its validity). I'm not a student, and do not have a professor to take this question to.

Thank you in advance for any assistance you can provide.