Allelic Dropout Rates
Here we recover high rates of detected allelic dropout in samples classified as low quality museum specimens from many rounds of PCR amplification of nuclear and mitochondrial DNA markers. The rates of allelic dropout averaged 12.6% across the five evaluated microsatellite loci for the pooled dataset. When we applied our ‘best practices’ method from this study the average rate of allelic dropout was reduced to 9.8% across all loci and including the LQMS. When data from the pooled run was evaluated, our rates of allelic dropout in the LQMS (average of 19.6%) conformed to rates reported for various studies of avian fecal allelic dropout (mean of 21%, Regnaut, Lucas, & Fumagalli, 2006) but was much higher than rates reported from chimpanzee fecal samples via GBS (7% rate of allelic dropout; Barbian et al., 2018). This may be due to the fact that Barbian et al., (2018) only genotyped samples at loci that had over 500 reads (counts.min = 500). After manually processing genotypes from all the a priori information ( the replicates, pooled run, and bioinformatically pooled data) the rates of dropout in the LQMS increased dramatically following processing to 79%. However, that was due to most genotypes lacking verification, and resulted in only four total genotypes called across all LQMS samples, with MVZ 2088 recovering three genotypes and MVZ 5211 recovering one. It is also worth mentioning that the LQMS samples had very high rates of individual allelic dropout due to unreliable genotypes which were ultimately removed in the manually processed (best practices) genotypes.
The HQMS samples performed better than expected, all recovering very low rates of allelic dropout (2.4% from the pooled run and <0.001% following our best practices). Two samples were collected from 1926 and one from 1975, and provide robust evidence for the utility of museum specimens for the recovery of microsatellite genotypes. These three samples performed on par with GBS studies derived from tissue samples (Darby et al., 2016).
The rate of allelic dropout in our tissue sample was higher upon raw results from CHIIMP, at 14.7% compared to 0.4% in Darby et al., (2016). However, this was partially due to one instance of primer region trimming (GS-4) for this sample. Following implementation of our best practices the rate was reduced to 8.5%, which is still higher than other studies of tissue samples (Darby et al., 2016). Additionally, the aberrant call of 251 from HSU 8180 (GSLA-52, pooled run) was removed during our best practices.
The pooled run CHIIMP results for dropout when separated by length were: 7.5% for short, 0.35% for medium and 48% for long loci. The higher rates for the short loci is likely attributed to non-specific amplification or higher amounts of stutter and PCR artifacts as identified from CHIIMP. Following our manually processed genotypes we recovered 0.0025%, 6.3%, and 36% occurrence of allelic dropout for short, medium and long loci.