The report reads from a single datasource, which has been pre-joined if necessary from multiple inputs. This means there may be quite a lot of fields to choose from when constructing the report. To make managing large reports easier, you can group the fields back into their logical categories so that users of the report can navigate the fields more effectively.
For example, if I have joined together a Customer table (Name, Address etc.) along with a Sales table (Item, Quantity etc.), the output is one big table, (Name, Address, Item, Quantity etc.). I might choose to define two groups, Customer and Sales, so that I can partition the fields between the groups, making it easier for the user to find the relevant field for reporting. This grouping is only an aid to navigation, it doesn't alter how the fields are structured internally.
The next tab, Fields, will allow you to partition items between the groups you define here.