Results
Specimen Collection. For initial microsatellite PCR amplification and gel electrophoresis assessments, we sampled 48 adult specimens from populations of O. lignaria collected in California (El Dorado County [n = 16], Fresno County [n = 11], Placer County [n = 13]) and Idaho (Franklin County [n = 8]). Using other wild-caught specimens, we conducted a population genetic analysis of 35 O. lignaria female specimens from Birch Creek and 41 O. lignaria female specimens from Cub River.
Microsatellite mining and assessment. Based on the filtering criteria we defined for our microsatellite query with the O. lignaria genome assembly, we identified 48,816 imperfect microsatellites, 10,553 perfect microsatellites, and 263 compound microsatellites. Average imperfect microsatellite product size is 153 nts ± 0.14 SE (maximum = 500, minimum = 100); average perfect microsatellite size is 148 nts ± 0.27 SE (maximum = 241, minimum = 100); and average compound microsatellite size is 171 nts ± 2.35 SE (maximum = 353, minimum = 100). We provide the filtered imperfect, perfect, and compound microsatellite primer sets in Supplementary Tables 1, 2, and 3, respectively.
From the queried loci, we haphazardly selected 22 microsatellites for testing, 13 of which were perfect microsatellites and nine of which were imperfect microsatellites (Table 1). We did not include compound microsatellites in our study due to their small sample size of 263 and product size overlap with perfect microsatellites. In the 76 O. lignaria specimens tested from Cub River and Birch Creek, average amplification success of the 22 markers was 95% ± 1% SE (maximum = 100%, minimum = 80%) (Table 1). Oli61901 demonstrated the least amplification success with 20% of specimens not amplifying at this locus. Oli064 demonstrated 100% amplification success in all tested specimens, whereas Oli160, Oli84207, Oli119, Oli131, Oli216, Oli127, Oli076, Oli053, Oli76146, and Oli03483 demonstrated 99% amplification success in all tested specimens. Average genotyping error rate was 11% ± 3% SE (Median = 3%; maximum = 42%, minimum = 0%). Eight markers had a genotyping error rate of 0% (i.e., sampled specimens were genotyped consistently at these loci), and included Oli156, Oli105252, Oli104572, Oli963, Oli144609, Oli76146, Oli141, Oli103483. Markers with a genotyping error rate > 30% included Oli053, Oli076, Oli107, and Oli127 (Table 1). Furthermore, a Wilcoxon Rank-sum test found that perfect microsatellites (M = 0.11, n = 13) had significantly larger genotype error rates than imperfect microsatellites (M = 0, n = 9) (W = 20.5, P = 0.01).
The 22 novel microsatellite loci tested in our study were distributed across 16 scaffolds of the O. lignaria genome assembly (Supplementary Table 4). Oli127 and Oli84207 were placed in scaffold NW_023009259.1; Oli107, Oli119, Oli131, and Oli216 were placed in scaffold NW_023009260.1; and Oli104572, Oli105252, and Oli156 were placed in scaffold NW_023009264.1. Based on the start position of each locus, mean pairwise nucleotide distances between loci co-located on a scaffold exceed a nucleotide distance of 500,000 nts: NW_023009259.1DIST = 559,304 nts (n = 1), NW_023009260.1DIST = 2,369,751 nts ± 514,195 SE (n = 6), and NW_023009264.1DIST = 6,599,034 nts ± 3,062,327 SE (n = 3). Comparison of the loci with the annotatedO. lignaria genome assembly suggests that four of the tested microsatellite loci are in an intergenic space, whereas 18 of the loci are in an intragenic space of a predicted gene. The average estimated distance of the intergenic loci to the nearest gene was 8,803 nts. Additional information, including gene identity, of the 22 novel microsatellites examined in this study are available in Supplementary Table 4.
Sibship identity and population genetic analysis. Of the 35O. lignaria individuals surveyed in Birch Creek, 21 full-sibling families were identified. Family size of the surveyed specimens in Birch Creek range from one to three individuals (mean = 1.67 ± 0.16 SE). Of the 41 O. lignaria individuals surveyed in Cub River, 24 full-sibling families were identified. Family size of the surveyed specimens in Cub River ranged from one to five individuals (mean = 1.70 ± 0.19 SE) (Table 2).
Seven loci, Oli160, Oli053, Oli064, Oli076, Oli127, Oli61901, and Oli84207, were not in HWE after Bonferroni correction (Table 1). Of the markers tested for LD and not in HWE, Oli053 was in LD with Oli076. Conversely, the following LD marker comparisons had at least one marker that was in HWE: Oli216 (in HWE) was in LD with Oli053 (not in HWE); Oli131 (in HWE) was in LD with Oli076 (not in HWE); Oli101 (in HWE) was in LD with Oli156 (in HWE) and Oli119 (in HWE); and Oli119 (in HWE) was in LD with Oli127 (in HWE) (Table 1). For the final analysis, we elected to remove all seven loci that were not in HWE and removed Oli101 as it was in LD with two other markers (Oli156 and Oli119). The loci either in HWE or demonstrating LD also demonstrated high genotype error rates (Supplementary Tables 5, 6). These loci are unlikely to be reliable markers for O. lignaria population genetic analysis due to phenomena like null alleles or primers binding to multiple annealing sites. Removal of these eight loci resulted in 14 loci suitable for further population genetic analysis: Oli021, Oli107, Oli119, Oli131, Oli141, Oli144609, Oli156, Oli216, Oli963, Oli74400, Oli76146, Oli103483, Oli104572, Oli105252.
We found no significant difference between meanHO and Nei’s HE in the Birch Creek population (mean HE = 0.63 ± 0.05 SE, mean HO = 0.60 ± 0.07 SE; t = 0.32,df = 26, P = 0.75) or Cub River population (meanHE = 0.60 ± 0.06 SE, meanHO = 0.53 ± 0.06 SE; t = 0.71, df = 26, P = 0.48). Furthermore, we found no significant difference in average HE between the Birch Creek or Cub River populations (t = 0.38, df = 26, P = 0.71). In addition to heterozygosity comparisons across populations, we found no significant difference between the Birch Creek and Cub River populations in Simpson’s Index of Diversity (1 - D) (Birch Creek mean [1-D] = 0.61 ± 0.05 SE; Cub River mean [1-D] = 0.58 ± 0.06 SE; t = -0.35, df = 28, P = 0.73), allelic richness (Birch Creek mean AR = 0.40 ± 0.06 SE; Cub River mean AR = 0.33 ± 0.07 SE; t = 0.79, df = 26, P = 0.44), the number of alleles (NA) (Birch Creek mean NA = 6.86 ± 0.94 SE; Cub River mean NA = 7.36 ± 1.51 SE; t = -0.26, df = 26, P = 0.79), or private alleles (PA) (Birch Creek M PA = 2; Cub River M PA = 1.5; Wilcoxon rank-sum test, W = 112.5, P = 0.51). All population genetic diversity and richness statistics and indices are described in Table 3.
AMOVA found evidence for genetic differentiation between the two Intermountain North America populations (Table 4). Specifically, 1.87% of the genetic variance was partitioned between populations, 4.83% of the genetic variance was partitioned between individuals within a population, and 93.3% of the genetic variance was partitioned within individuals across populations (Table 4). While high variance across individuals in both populations was observed, we found that ΦST is highest in variance comparisons between populations (ΦST = 0.07) as opposed to between individuals within populations (ΦST = 0.04) (Table 4). Furthermore, based on 999 permutations of the data, we present statistical inference to support the conclusion that these populations are differentiated at all population stratifications (Table 4). While population structure estimates are low, estimates of ΦSTbetween the two populations is further supported by analogs of this index in FST (1.66 X 10-2,P = 0.01), GST (8.39 X 10-3, P = 0.01) and Jost’s D (1.18 X 10-2, P = 0.05). Finally, an assessment of the inbreeding with FIS implicates no evidence of inbreeding across populations (FIS = 4.20 x 10-3, P = 0.99).
Osmia subspecies, species, and clade marker amplification. In total, eight loci consistently amplified in all four clades: Oli101, Oli84207, Oli131, Oli216, Oli127, Oli076, Oli156, and Oli105252. Nine loci amplified in three species/subspecies ofapicata , 11 loci amplified in seven species/subspecies ofemarginata , and all 22 loci amplified 11 species/subspecies ofbicornis and two species/subspecies of ribifloris (Table 5, Supplementary Table 7). Average marker amplification rate across specimens was 21% ± 6% (n = 3), 51% ± 5% (n = 27), 22% ± 6% (n = 7), and 66% ± 7% (n = 11) in the cladesapicata , bicornis , emarginata , andribifloris, respectively.