APPENDIX S1. Supplementary methods.
SUPPLEMENTARY TABLES. TABLE S1. Primer details for microsatellite loci used to genotype black-backed jackals (Canis Mesomelas).
TABLE S2. Per-locus summary statistics as calculated in Cervus v3.0.7. The non-exclusion probabilities and combined non-exclusion probabilities (final row, italics) are relevant indicators of the power of the loci for parentage and sibship analyses.
TABLE S3. Summary statistics for 20 sampling localities (farms) with >1 sample and for all farms pooled. Produced using the basicStats command of the diveRsity package v1.9.90 in R v3.6.2 and RStudio v1.2.5033. Standard deviation was calculated across loci in Microsoft Excel (stdev.s). Sampling localities with only one sample are not shown.
TABLE S4. Summary statistics per year and for all years pooled. Produced using the basicStats command of the diveRsity package v1.9.90 in R v3.6.2 and RStudio v1.2.5033. Standard deviation was calculated across loci in Microsoft Excel (STDEV.S).
TABLE S5. Pairwise FST values between farms with the full dataset (below diagonal) and associated significance at a level of 0.05 (above diagonal), where significant values are indicated by a “+” and non-significant values by a “−”. Calculated in Arlequin 3.5.2.2.
TABLE S6. Pairwise FST values between farms with relatives removed (below diagonal) and associated significance at a level of 0.05 (above diagonal), where significant values are indicated by a “+” and non-significant values by a “−”. Calculated in Arlequin 3.5.2.2.
TABLE S7. Comparison of mean pairwise relatedness (r) between years and mean individual inbreeding coefficients (F) between years. P-values for the Wilcoxon tests for difference in means are shown on the inside of the table (bordered by grey), with P-values for inbreeding comparisons shown below the diagonal (bottom left) and P-values for relatedness comparisons shown above the diagonal (top right). The mean F for each year is shown in the left-most column “outside” the main table, with the mean r for each year shown in the top row “outside” the main table. The numbers in parentheses after each year are the number of observations/data points for that year (number of samples for F and number of pairwise relatedness comparisons for r).
SUPPLEMENTARY FIGURES. FIGURE S1. STRUCTURE HARVESTER results for (a) Delta K values and (b) probability (-LnPr) of K = 1–27 averaged over 20 runs and (c) genetic differentiation between the jackal sample locations (farms) based on STRUCTURE analysis (performed with K = 2–6) of 1 = GV, 2 = BB, 3 = BR, 4 = BD, 5 = DS, 6 = GG, 7 = HK, 8 = KD, 9 = KW, 10 = KK, 11 = KT, 12 = NG, 13 = ND, 14 = OG, 15 = RV, 16 = RE, 17 = RT, 18 = RD, 19 = SG, 20 = SK, 21 = VR, 22 = WK, 23 = CL, 24 = KR, 25 = WB and 26 = TD.
FIGURE S2. STRUCTURE HARVESTER results for (a) Delta K values and (b) probability (-LnPr) of K = 1–27 averaged over 20 runs and (c) genetic differentiation between the jackal sample locations (farms) based on STRUCTURE analysis (performed with K = 2–6 and K = 14) of 1 = GV, 2 = BB, 3 = BD, 4 = DS, 5 = GG, 6 = HK, 7 = KW, 8 = KT, 9 = NG, 10 = ND, 11 = OG, 12 = RV, 13 = RE, 14 = RD, 15 = SG, 16 = SK, 17 = VR, 18 = WK and 19 = CL. After removing relatives, some localities had no samples, hence fewer sampling localities as compared to the full dataset. Note: The Evanno method (DeltaK) does not evaluate K = 1.
FIGURE S3. Principal component analysis (PCA) of the different jackal sampling locations (farms) with related individuals removed.
FIGURE S4. Plot comparing the relatedness estimates using six estimators and simulated individuals of known relatedness. Di, Dyadic likelihood estimator “DyadML”; LL, Lynch-Li estimator; LR, Lynch and Ritland estimator; QG, Queller and Goodnight estimator; Tri, Triadic likelihood estimator “TrioML”; W, Wang estimator. Plot produced with ggplot2 3.3.0 (Wickham, 2016).
FIGURE S5. Results of the spatial autocorrelation analysis for A females and B males. The blue line indicates the autocorrelation coefficient of the data, with the 95% confidence interval at each distance class indicated by the black error bars, as determined by 1000 bootstrap resampling replicates. The red dashed lines indicate the 95% confidence interval around the null hypothesis (no spatial structure, i.e. rauto = 0), as determined by permutation (999 steps). Thus, if the error bars around the blue line do not overlap with the red dashed lines in a distance class, then genotypes were more (positive rauto) or less (negative rauto) similar than expected under the null hypothesis in that distance class. Such cases are indicated with an asterisk (*).