Here is an outline of an algorithm you can adapt:

my %h;
my $n=3;

sub add{
  my ($n,$v,@a)=@_; 
 (%h < $n and not exists $h{$_} and  $h{$_}=$v ) for @a
};

add($n,  # Allow $n unique elements 
     1,  # This is the First array to be added
         # Numbers below are the contens of the array to be added
    3,4,5,6,3,2,4,1); # First set of numbers

$n+=$n;                 # Now increase the number of elements allowed

add($n,2,4,3,2,7,8,2); # Second set of numbers

print  qq($_ => $h{$_} \n) for sort keys %h
--OUTPUT---
2 => 2
3 => 1
4 => 1
5 => 1
7 => 2
8 => 2
[download]

The VALUES of the hash indicate the source array it came from.

Yes - this can be optimized, and use of the global %h can be avoided - enhancements left as an exercise.