What I would recommend is that you treat a specimen that is identified to an order (Perciformes) with something like the following.
Species => Order Perciformes species undetermined.
The individual is still an instance of a species, however that species has yet to be determined.
What would work best is to have some standard way of writing the green string above.
This would allow the occurrences that are of individuals identified only to the Order Perciformes, to be interpreted as a species that falls somewhere within the Order Perciformes.
- Pete