
## Legends for Supplementary Figures ##

Figure S1. Land-use composition across sampling locations. Stacked bars show the proportional cover of retained land-use classes for each sampling location. Colors indicate land-use classes. Locations are grouped by landscape category. Far_LoA, far low-agricultural landscapes; Near_LoA, near low-agricultural landscapes; SSF, small-scale farming landscapes.


Figure S2. PCA of land-use composition across sampling locations. Points represent sampling locations and colors indicate landscape category. Arrows show land-use class loadings.


Figure S3.Land-use composition differences between SSF and LoA landscapes. Violin plots show the proportional cover of each land-use class within 1-km-radius buffers. Points indicate sampling locations. P-values are from Wilcoxon rank-sum tests. SSF, small-scale farming landscapes; LoA, low-agricultural landscapes.


Figure S4. Land-use composition differences among SSF, Near_LoA, and Far_LoA landscapes. Violin and box plots show the proportional cover of each land-use class within 1-km-radius buffers. Points indicate sampling locations. P-values are from Kruskal–Wallis tests, with significant pairwise comparisons shown above panels.


Figure S5. Correlations among landscape-configuration metrics. Values indicate Pearson correlation coefficients among the 23 landscape metrics. Blue indicates positive correlations and red indicates negative correlations. Metric abbreviations are defined in Table S3.


Figure S6. Variable loading plots for the first four dimensions of the landscaoe metric ordination. Arrows represent landscape metrics projected onto the ordination space. Arrow length and direction indicate the correlation of each metric with the ordination axes, and color intensity represents the conttribution (%) of each metric to each dimension. Percentages in parentheses indicate the variance explained by each dimension.


Figure S7. Pairwise FST among sampling locations for Andrena flavipes. Darker and lighter colors represent lower and higher FST values, respectively.


Figure S8. Pairwise FST among sampling locations for Bombus terrestris. Darker and lighter colors represent lower and higher FST values, respectively.


Figure S9. Pairwise FST among sampling locations for Lasioglossum malachurum. Darker and lighter colors represent lower and higher FST values, respectively.


Figure S10.Principal component analysis (PCA) of filtered SNP datasets. PCA plots showing the first two principal components for a) Andrena flavipes, b) Bombus terrestris, and c) Lasioglossum malachurum. Individuals have different shapes according to landscape category. The percentage of total variation explained by each principal component is indicated on the corresponding axes.


Figure S11. fineRADstructure coancestry pattern in Andrena flavipes. Heatmap shows inferred coancestry among individuals based on the final filtered SNP dataset. Individuals are ordered according to fineRADstructure clustering.


Figure S12. fineRADstructure coancestry pattern in Bombus terrestris. Heatmap shows inferred coancestry among individuals based on the final filtered SNP dataset. Individuals are ordered according to fineRADstructure clustering.


Figure S13. fineRADstructure coancestry pattern in Lasioglossum malachurum. Heatmap shows inferred coancestry among individuals based on the final filtered SNP dataset. Individuals are ordered according to fineRADstructure clustering.


Figure S14. ALStructure results for Andrena flavipes. Bar plot shows individual ancestry coefficients inferred with ALStructure for (A) K = 2, and (B) K = 3 using the final filtered SNP dataset. Each vertical bar represents one individual and colors represent inferred ancestry.


Figure S15. ALStructure results for Bombus terrestris. Bar plot shows individual ancestry coefficients inferred with ALStructure for (A) K = 2, and (B) K = 3 using the final filtered SNP dataset. Each vertical bar represents one individual and colors represent inferred ancestry.


Figure S16. ALStructure results for Lasioglossum malachurum. Bar plot shows individual ancestry coefficients inferred with ALStructure for (A) K = 2, and (B) K = 3 using the final filtered SNP dataset. Each vertical bar represents one individual and colors represent inferred ancestry.


Figure S17. Isolation-by-distance analysis in Andrena flavipes. Points show pairwise relationships between log-transformed geographic distance and genetic distance among sampling locations. The blue line shows the linear trend and the gray band indicates the 95% confidence interval. Mantel test statistics are shown in the top-right corner of the panel.


Figure S18. Isolation-by-distance analysis in Bombus terrestris. Points show pairwise relationships between log-transformed geographic distance and genetic distance among sampling locations. The blue line shows the linear trend and the gray band indicates the 95% confidence interval. Mantel test statistics are shown in the top-right corner of the panel.


Figure S19. Isolation-by-distance analysis in Lasioglossum malachurum. Points show pairwise relationships between log-transformed geographic distance and genetic distance among sampling locations. The blue line shows the linear trend and the gray band indicates the 95% confidence interval. Mantel test statistics are shown in the top-right corner of the panel.


Figure S20. BayeScan outlier analysis for Andrena flavipes. Plot shows BayeScan locus-specific FST estimates and log(q-values) for the filtered SNP dataset. Candidate SNPs were identified using q < 0.05.


Figure S21. BayeScan outlier analysis for Bombus terrestris. Plot shows BayeScan locus-specific FST estimates and q-values for the filtered SNP dataset. Candidate SNPs were identified using q < 0.05.


Figure S22. BayPass C2 x BF statistics in Andrena flavipes. Points represent SNPs tested across SSF × LoA, SSF × Near_LoA, and SSF × Far_LoA contrasts using seed 7357. Red points indicate SNPs supported by both BF and C2 thresholds.


Figure S23. BayPass C2 x BF statistics in Bombus terrestris. Points represent SNPs tested across SSF × LoA, SSF × Near_LoA, and SSF × Far_LoA contrasts using seed 7357. Red points indicate SNPs supported by both BF and C2 thresholds.


Figure S24. BayPass C2 x BF statistics in Lasioglossum malachurum. Points represent SNPs tested across SSF × LoA, SSF × Near_LoA, and SSF × Far_LoA contrasts using seed 7357. Red points indicate SNPs supported by both BF and C2 thresholds.


Figure S25. BayPass local score analysis for the SSF x LoA contrast in Andrena flavipes. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S26. BayPass local score analysis for the SSF x Near_LoA contrast in Andrena flavipes. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S27. BayPass local score analysis for the SSF x Far_LoA contrast in Andrena flavipes. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S28. BayPass local score analysis for the SSF x LoA contrast in Bombus terrestris. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S29. BayPass local score analysis for the SSF x Near_LoA contrast in Bombus terrestris. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S30. BayPass local score analysis for the SSF x Far_LoA contrast in Bombus terrestris. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S31. BayPass local score analysis for the SSF x LoA contrast in Lasioglossum malachurum. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S32. BayPass local score analysis for the SSF x Near_LoA contrast in Lasioglossum malachurum. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S33. BayPass local score analysis for the SSF x Far_LoA contrast in Lasioglossum malachurum. The graph shows Lindley Scores along chromosomes, with dashed lines indicating significance thresholds.


Figure S34. COG functional categories represented among candidate genes in Andrena flavipes. Each bar represents the number of candidate genes assigned to each COG functional category based on eggNOG-mapper annotation.


Figure S35. COG functional categories represented among candidate genes in Bombus terrestris. Each bar represents the number of candidate genes assigned to each COG functional category based on eggNOG-mapper annotation.


Figure S36. COG functional categories represented among candidate genes in Lasioglossum malachurum. Each bar represents the number of candidate genes assigned to each COG functional category based on eggNOG-mapper annotation.
