In that case the hash of hash is exactly appropriate and it looks like my sample code should drop right into your application. Happy to help.

Update: there is enough information to generate the header too: ;)

...

my @species = sort keys %data;
my $nSpecies = @species;

# Print the header
print "begin data;\n";
printf "dimensions ntax=%d nchar=%d;\n", $nSpecies, $dnaLen + $morphLe
+n;
printf "format datatype=mixed (dna:1-%d, standard:%d-%d)  missing=? ga
+p=-;\n",
    $dnaLen, $dnaLen + 1, $dnaLen + $morphLen;
print "Matrix\n";

# Print the data
for my $species (sort keys %data) {
    $data{$species}{dna} ||= '?' x $dnaLen;
    $data{$species}{morph} ||= '?' x $morphLen;
    print "$species: $data{$species}{dna}$data{$species}{morph}\n";
}

print "end;\n";
[download]

Prints:

begin data;
dimensions ntax=5 nchar=32;
format datatype=mixed (dna:1-14, standard:15-32)  missing=? gap=-;
Matrix
species_1: ACCATGATACGATG001001010201001001
species_2: GGTTTCGACGCAGA002010200210120201
species_3: GGACTCAGCGACTA??????????????????
species_4: ??????????????001001110000000101
species_5: ??????????????111001001001000201
end;
[download]