When adding SmartMarks over the Add SmartMarks task on a native PDF, all layers of the original PDF are merged into one layer.
The output (native) PDF contains only two layers:
Layer 1 contains all objects from all layers of the input file (also, the layers that were set to invisible). So, besides losing the different layer information, the view can also be different.
This is a known limitation: for native PDF files, there is no other choice, we can only stream them as one entity (embedded in a form). In this operation, the layers get lost. One ends with one layer containing the SmartMarks and another layer containing a reference to this form (containing the original input file).
Normalize the file, in this case the file is opened by the Add SmartMark task and one layer is added containing the SmartMarks (all the original layers are preserved), export to PDF again if needed.