Is the epitope accessible to the antibody? It maybe blocked or sequestered within the cells. What antigen retrieval are you using? Is there a cuticle present?

What is the expression level in each species? Maybe it's too low for detection. Could you check mRNA levels to ensure it's there?

Hi Liz, thanks for your answer.

The antibodies I'm using are for targets in various cell locations: Membrane, nucleus etc. No cuticle. in-situ stainings (for those of the relevant genes) show good signal.

Im not using any antigen retrieval methods, what do you recommend?

One of the first things I check when antibodies don't work is a control that I am sure is positive.  Make sure your antibodies have not degraded.  If that is not the case then I would trouble shoot other aspects of the system to make sure something is not failing.  If it really is a highly conserved antigen then you should see some reaction.

Thanks Will,

I Checked them on control species, and they worked.

Could there be any species specific factor that Im not aware of?

Yeah it is always possible that the antigens are actually not on the species you are examining.  If they are not close enough related to the ones that the antibodies were made on they might not match.  If for example you were looking at Symphylla and had antibodies for fly tissues perhaps they will not cross react.

In this specific experiment, the antibodies are for alpha-tubulin, and for engrailed (among others). Both are know to be very conserved, and the antibodies are known to be very cross reactive among all arthropods. Fixation methods also vary considerably, so im not sure the fixing method is to blame. 

(alpha-tubulin+DAPI)

If your positive controls work, then either something in the protocol is not working well in your species of interest, or the epitopes are not conserved in that particular species. The former is more likely for the antibodies you mention, but since they are widely used, there might be some literature specifying the specific peptide sequence they recognize, so if you have the protein sequence for your species, you can double check that. If you fail to get signal from both the tubulin and engrailed mABs, then I think the protocol is to blame.

Are these the alpha-tubulin and engrailed mABs from DSHB?

There is another possibility: while the mRNA is being expressed, it might not be getting turned into protein. It's a biologically interesting explanation, but you'd have to really prove that it's not your protocol before making that claim.