The experimental unit is the replicate (and not the birds in that replicate).

If you have one sex only, ten to twelve birds per replicate may be sufficient (see e.g. attached paper) but with mixed sexes the chance of uneven distribution of males and female over the replicates - thus increasing variability - , must be taken into account. I would choose in that case some 30 -50 animals per replicate.

What's the objective(s) of the proposed experiment?

I recommend increasing the number of birds per replicate, 3 may be insufficient.