A table says “pond” in one row and “pond ” in another. To a person scanning the screen they look identical. A counting tool may treat them as different values. Before explaining a surprising category total, inspect the labels that produced it.
List the values before changing them
Make a frequency table of the original labels, including missing entries. Show text inside quotation marks or inspect its length so trailing spaces become easier to notice. In a tiny example, six “pond” records and four “pond ” records might represent ten observations of one habitat, but that interpretation needs confirmation.
Define the rule you intend to apply
Removing leading and trailing spaces is different from removing every space. The latter could turn meaningful multiword names into ambiguous strings. Changing case can also merge codes that were deliberately distinct. Check the data dictionary or ask the supplier before treating a visual similarity as permission to combine values.
Keep a reversible mapping
Create a second column for the cleaned label and keep a small mapping table: original value, replacement, reason. After grouping the cleaned column, check that the number of records is unchanged. A cleaner list of category names should not quietly remove observations.
Then review the unusual values rather than forcing all of them into the nearest familiar name. An unfamiliar category may contain real information. Add the decision to your transformation log so another reader can repeat it. This is a modest cleaning step, but documenting it separates a reproducible count from a total that merely looks tidier.