Figure 1.
Impact of sample size on allele detection.
Percentage of samples in which all alleles at a real frequency ≥0.05 at each locus were detected at each sample size (of the 100 random replicates per size), for A) the ant dataset, B) the squirrel dataset, C) the albatross dataset and D) the kakī dataset. The vertical line shows the sample size at which all alleles at a real frequency ≥0.05 at all loci were detected in ≥95% of replicates.
Figure 2.
Impact of allele frequency and sample size on the accuracy of mean sample allele frequency.
Mean difference from the real allele frequency for each sample size (of the 100 random replicates per size) for A) the ant dataset, B) the squirrel dataset, C) the albatross dataset and D) the kakī dataset. Black circles represent alleles with a real frequency ≥0.05 (data for the same allele at different sample sizes linked by a line), white circles represent alleles with a real frequency between 0.05 and 0.01, and grey circles represent alleles with a real frequency ≤0.01.
Figure 3.
Impact of sample size on the precision of sample allele frequencies.
The range (lines) and mean ± one standard deviation (solid boxes) of allele frequencies of the 100 random replicates at each sample size for one common and one relatively rare allele in A) the ant dataset, B) the squirrel dataset, C) the albatross dataset and D) the kakī dataset. The alleles are: A) allele 159 at locus FE16 (grey boxes, real frequency = 0.176) and allele 116 at locus FE17 (black boxes, real frequency = 0.833); B) allele 196 at locus Scv8 (grey boxes, real frequency = 0.126) and allele 162 at locus Scv23 (black boxes, real frequency = 0.766); C) allele 187 at locus De35 (grey boxes, real frequency = 0.147) and allele 165 at locus D5 (black boxes, real frequency = 0.852); and D) allele 241 at locus Kakī_21 (grey boxes, real frequency = 0.112) and allele 200 at locus Kakī_27 (black boxes, real frequency = 0.745).
Figure 4.
Impact of sample size on the accuracy and precision of sample heterozygosity.
The range (lines) and mean ± one standard deviation (solid boxes) of expected heterozygosity (HE) for the 100 random replicates at each sample size for a locus with high and a locus with low expected heterozygosity in the empirical dataset for A) the ant dataset, B) the squirrel dataset, C) the albatross dataset and D) the kakī dataset. Loci are: A) locus FE37 (real HE = 0.75, black boxes) and locus FE17 (real HE = 0.29, grey boxes); B) locus Scv3 (real HE = 0.78, black boxes) and locus Scv23 (real HE = 0.39, grey boxes); C) locus De11 (real HE = 0.807, black boxes) and locus D5 (real HE = 0.268, grey boxes); and D) locus Kakī_21 (real HE = 0.712, black boxes) and locus Kakī_40 (real HE = 0.291, grey boxes).
Figure 5.
Impact of sample size on the accuracy and precision of mean heterozygosity across loci.
The range (lines) and mean ± one standard deviation (solid boxes) of mean expected heterozygosity (HE) for the 100 random replicates at each sample size for A) the ant dataset (9 loci), B) the squirrel dataset (5 loci), C) the albatross dataset (7 loci) and D) the kakī dataset (8 loci).
Figure 6.
Impact of sample size on mean genetic distance between samples and the true population.
Mean pairwise FST between the 100 random replicates and the empirical dataset for A) the ant dataset, B) the squirrel dataset, C) the albatross dataset and D) the kakī dataset at each sample size. Error bars are standard deviation.