$idx = $arr[0];
for(0..$#arr) {
        $idx = $_ if( ($arr[$idx]<0) || (($arr[$_] >= 0) && ($arr[$_] 
+< $arr[$idx])) )
}
[download]

This returns strange results for certain corner cases:

c:\@Work\Perl\monks>perl -wMstrict -le
"use Data::Dump qw(pp);
 ;;
 for my $ar (
   [ ], [ -1 ], [ -99 ], [ 0, -99 ], [ -99, 0 ],
   ) {
   my $idx = $ar->[0];
   for (0..$#$ar) {
     $idx = $_ if( ($ar->[$idx]<0) || (($ar->[$_] >= 0) && ($ar->[$_] 
+< $ar->[$idx])) );
     }
   print qq{(@$ar) -> lnn index of }, pp $idx;
   }
"
() -> lnn index of undef
(-1) -> lnn index of 0
Use of uninitialized value in numeric lt (<) at -e line 1.
(-99) -> lnn index of -99
(0 -99) -> lnn index of 0
Use of uninitialized value in numeric lt (<) at -e line 1.
Use of uninitialized value in numeric lt (<) at -e line 1.
Use of uninitialized value in numeric lt (<) at -e line 1.
(-99 0) -> lnn index of -99
[download]

I'd hate to have to extend this to dealing with fractional numbers. The O(n) solutions here and here both return an unambiguous undef for lists containing no LNN, including empty lists.