The zero-mean hypothesis in this test means that the mean value of time series data is zero, while the single-mean hypothesis means that time series data has only one mean value.

The significance of the zero-mean hypothesis of DF test in SAS is that if the mean value of data is not zero, the mean value of data may change after preprocessing steps such as difference, thus affecting the test result of unit root.

The significance of single mean hypothesis is that if there are multiple means in the data, the mean distribution of the data may change after preprocessing steps such as difference, thus affecting the test results of unit roots.