Multiplexing approaches
Sequenced alleles ranged in size from 71 bp (D3s1768) to 211 bp (D12s372) and nine loci contained cryptic alleles in at least one species (Tables S7-S10). The level of amplification and obtained sequence reads varied across samples/species and loci in all three approaches. The amplification of all loci in one pool (1-pool approach) was least effective, resulting in the lowest number of amplified loci (mean 25.9) and alleles (mean 36.9; Table 2). In some cases, the reason for allelic dropouts could be attributed to wrong primer pairing/primer mismatches (primer-dimer or off-target amplification of short products). Most loci amplified less efficiently than in the other two approaches and some (N=11) failed to amplify at all. Only nine loci recovered the same number of alleles as in the 5-pool approach. Interestingly, even though the number of amplified alleles was reduced from a mean of 60.1 to 36.9 compared to the 5-pool approach, the level of heterozygosity was not affected to the same extent with a reduction from 49.8 to 43.5% (Table 2).
The best results, i.e. the highest amplification levels for loci (mean 40.2) and alleles (mean 60.1), were generated applying the 5-pool approach (Table 2). Nevertheless, we observed again primer-dimers and short off-target PCR products potentially as a result of interacting primers from different loci. Moreover, three loci (D11s1366, D12s67.2, and D15s1007) neither amplified in the 1-pool nor in the 5-pool approach and were excluded from further testing. To further improve amplification success and to reduce primer interactions among primers of different loci (based on the knowledge obtained from the 1-pool and 5-pool approaches), we distributed the 42 remaining loci into three amplification pools containing 18, 12 and 12 loci, respectively (Table S5). Using the 3-pool approach, we were able to largely minimize primer interactions, but amplification success for loci (mean 37.8) and alleles (mean 55.5) per species was slightly reduced compared to the 5-pool approach, but higher than in the 1-pool approach (Table 2).