We used a topoisomerase based reaction to ligate our PCR product into pCR2.1 vector to then introduce it in E. coli. Theoretically, there should be 3 outcomes. One where the PCR product for the expression of GFP is pasted in a sense way, another for the antisense way, and no pasting at all. Therefore, we assumed we were going to get a a 33% to 50% "yield" for colonies expressing GFP. While the culture shows barely a couple of non expressing GFP bacteria.
1. Adenine overhangs are more likely to be added to some sequences than others so maybe the forward orientation was more likely to have an A on both ends
2. Your strain expresses T7 polymerase which is on the reverse orientation of pCR2.1
3. When you copied GFP you grabbed the promoter from the source template (or part of it that made a chimeric promoter with the plasmid)
4. You only checked a couple plates but due to the randomness of the process it seems like it's skewed, however if you checked dozens of plates it would even out to ~50%.
Did you just use normal Taq or something like Easy A?
From your picture it looks like you have a fair number of non-fluorescing colonies though.