What follows is an alternative which is probably faster than the OP's due to the lack of lists and arrays:

$au = 'PAUL A. HUBBARD,<SUP>1</SUP>'
    .' WENFENG YU,<SUP>2</SUP>'
    .' HORST SCHULZ,<SUP>2</SUP>'
    .' JUNG-JA P. KIM<SUP>1</SUP>';

$au =~ s{,((?:<SUP>.*?</SUP>)?)\s*}
        {$1</contributor>\n<contributor>}g;
$au = "<contributor>$au</contributor>\n";

print($au);

__END__
<contributor>PAUL A. HUBBARD<SUP>1</SUP></contributor>
<contributor>WENFENG YU<SUP>2</SUP></contributor>
<contributor>HORST SCHULZ<SUP>2</SUP></contributor>
<contributor>JUNG-JA P. KIM<SUP>1</SUP></contributor>
[download]

So far, I've always found that split with a capture in the pattern is at least as easy to implement as a straight regexp.

Updated.