R2 is a measure of the proportion of variation in the dependent variable accounted for by the independent variable(s). It is called the coefficient of determination. The higher it is, the better the model. I don't know of any limit to be adopted for it. The rule of thump should be to use any R2 at least 80%.
You should take decision on adj-R square rather than just make threshold of R square. To select an optimal number of lags with the help of ARDL, the best option is using adj-R square.