Practical implications of Markovian assumptions?

[DMills]

My experiments involve releasing captive-reared animals back to the wild, and and resighting them in a number of different areas (strata). The likely behaviour of these animals is to move relatively large distance when first released, then settle into a small home range. Common histories (in a 2 strata system) might include:

A0B0BB0B - has moved from A-B early in the trial then set up home range
AA0AA00A - has set up home range in strata A

All animals are released in strata A, and it is rare to find animals that move from strata B back to strata A.

While GOF tests for my current data set suggest reasonable fit, I will be conducting further trials, and am concerned that with higher resolution data and longer data sets, this may not be the case Am I correct in assuming that large numbers of the histories like those mentioned above would lead to violations of markovian assumptions with respect to transitions between strata? and if so, what is the least data intensive way to accomodate these problems?

Thanks
Dave Mills
Tasmanian Aquaculture and Fisheries Institute