I have a CCA triplot of taxa/ samples over 8 environmental gradients, and am trying to make sense of the result.

First, different theory sources all state that clustering of samples around the far end of an arrow representing an environmental gradient suggests that gradient is strong (has high values) in these samples (if I understand correctly). So I based my evaluation on that. However, by checking with the original (normalized) data, sometimes this is correct, but in other cases not.

Similarly, according to this statement, a line drawn from samples that came from high elevations orthogonally to the elevation vector in the triplot should touch that environmental vector near its far end, or even touch a prolongation of that far end. Equally, in some cases this is the case if I compare my triplot with the data sheet, in others absolutely not. I have checked my data sheet and CCA execution many times, but don't find mistakes there.

Is there anyone who can explain this? I am getting a little frustrated here, not being able to make good sense of my results after reading so much excellently explained CCA literature. Thank you in advance.

Serge