My suggestion is that you should double check the encoding.
Using UTF-8 as a character encoding does not ensure by itself things are right.
For example, sending UTF-8 which is then decoded as if it were ISO-8859-15 would explain such a problem.
Verify all the data sources, you might have one in an unexpected encoding. I agree with cms, do not use text replace to delete this, fix the problem at its source.
You can play with iconv to see how your data looks when converted from and to different encodings.
UPDATE: I see your page perfectly, so maybe it's your browser decoding the characters wrongly, check the character encoding settings on your browser