Locally adapted temperate tree populations exhibit genetic trade-offs among climate-related traits that can be exacerbated by selective breeding and are challenging to manage under climate change. To inform climatically adaptive forest management, we investigated the genetic architecture and impacts of selective breeding on four climate-related traits in 105 natural and 20 selectively bred lodgepole pine populations from western Canada. Growth, cold injury, growth initiation, and growth cessation phenotypes were tested for associations with 18,600 single-nucleotide polymorphisms (SNPs) in natural populations to identify “positive effect alleles” (PEAs). The effects of artificial selection for faster growth on the frequency of PEAs associated with each trait were quantified in breeding populations from different climates. Substantial shifts in PEA proportions and frequencies were observed across many loci after two generations of selective breeding for height, and responses of phenology-associated PEAs differed strongly among climatic regions. Extensive genetic overlap was evident among traits. Alleles most strongly associated with greater height were often associated with greater cold injury and delayed phenology, although it is unclear whether potential trade-offs arose directly from pleiotropy or indirectly via genetic linkage. Modest variation in multilocus PEA frequencies among populations was associated with large phenotypic differences and strong climatic gradients, providing support for assisted gene flow polices. Relationships among genotypes, phenotypes, and climate in natural populations were maintained or strengthened by selective breeding. However, future adaptive phenotypes and assisted gene flow may be compromised if selective breeding further increases the PEA frequencies of SNPs involved in adaptive trade-offs among climate-related traits.

The paradox of high genetic variation observed in traits under stabilizing selection is a long-standing problem in evolutionary theory, as mutation rates appear too low to explain observed levels of standing genetic variation under classic models of mutation-selection balance. Spatially or temporally heterogeneous environments can maintain more standing genetic variation within populations than homogeneous environments, but it is unclear whether such conditions can resolve the above discrepancy between theory and observation. Here, we use individual-based simulations to explore the effect of various types of environmental heterogeneity on the maintenance of genetic variation (V A ) for a quantitative trait under stabilizing selection. We find that V A is maximized at intermediate migration rates in spatially heterogeneous environments and that the observed patterns are robust to changes in population size. Spatial environmental heterogeneity increased variation by as much as 10-fold over mutation-selection balance alone, whereas pure temporal environmental heterogeneity increased variance by only 45% at max. Our results show that some combinations of spatial heterogeneity and migration can maintain considerably more variation than mutation-selection balance, potentially reconciling the discrepancy between theoretical predictions and empirical observations. However, given the narrow regions of parameter space required for this effect, this is unlikely to provide a general explanation for the maintenance of variation. Nonetheless, our results suggest that habitat fragmentation may affect the maintenance of V A and thereby reduce the adaptive capacity of populations.

The use of next‐generation sequencing (NGS) data sets has increased dramatically over the last decade, but there have been few systematic analyses quantifying the accuracy of the commonly used variant caller programs. Here we used a familial design consisting of diploid tissue from a single lodgepole pine (Pinus contorta) parent and the maternally derived haploid tissue from 106 full‐sibling offspring, where mismatches could only arise due to mutation or bioinformatic error. Given the rarity of mutation, we used the rate of mismatches between parent and offspring genotype calls to infer the single nucleotide polymorphism (SNP) genotyping error rates of FreeBayes, HaplotypeCaller, SAMtools, UnifiedGenotyper, and VarScan. With baseline filtering HaplotypeCaller and UnifiedGenotyper yielded more SNPs and higher error rates by one to two orders of magnitude, whereas FreeBayes, SAMtools and VarScan yielded lower numbers of SNPs and more modest error rates. To facilitate comparison between variant callers we standardized each SNP set to the same number of SNPs using additional filtering, where UnifiedGenotyper consistently produced the smallest proportion of genotype errors, followed by HaplotypeCaller, VarScan, SAMtools, and FreeBayes. Additionally, we found that error rates were minimized for SNPs called by more than one variant caller. Finally, we evaluated the performance of various commonly used filtering metrics on SNP calling. Our analysis provides a quantitative assessment of the accuracy of five widely used variant calling programs and offers valuable insights into both the choice of variant caller program and the choice of filtering metrics, especially for researchers using non‐model study systems.

The use of NGS datasets has increased dramatically over the last decade, however, there have been few systematic analyses quantifying the accuracy of the commonly used variant caller programs. Here we used a familial design consisting of diploid tissue from a single Pinus contorta parent and the maternally derived haploid tissue from 106 full-sibling offspring, where mismatches could only arise due to mutation or bioinformatic error. Given the rarity of mutation, we used the rate of mismatches between parent and offspring genotype calls to infer the SNP genotyping error rates of FreeBayes, HaplotypeCaller, SAMtools, UnifiedGenotyper, and VarScan. With baseline filtering HaplotypeCaller and UnifiedGenotyper yielded one to two orders of magnitude larger numbers of SNPs and error rates, whereas FreeBayes, SAMtools and VarScan yielded lower numbers of SNPs and more modest error rates. To facilitate comparison between variant callers we standardized each SNP set to the same number of SNPs using additional filtering, where UnifiedGenotyper consistently produced the smallest proportion of genotype errors, followed by HaplotypeCaller, VarScan, SAMtools, and FreeBayes. Additionally, we found that error rates were minimized for SNPs called by more than one variant caller. Finally, we evaluated the performance of various commonly used filtering metrics on SNP calling. Our analysis provides a quantitative assessment of the accuracy of five widely used variant calling programs and offers valuable insights into both the choice of variant caller program and the choice of filtering metrics, especially for researchers using non-model study systems.

The use of NGS datasets has increased dramatically over the last decade, however, there have been few systematic analyses quantifying the accuracy of the commonly used variant caller programs. Here we used a familial design consisting of diploid tissue from a single Pinus contorta parent and the maternally derived haploid tissue from 106 full-sibling offspring, where mismatches could only arise due to mutation or bioinformatic error. Given the rarity of mutation, we used the rate of mismatches between parent and offspring genotype calls to infer the SNP genotyping error rates of FreeBayes, HaplotypeCaller, SAMtools, UnifiedGenotyper, and VarScan. With baseline filtering HaplotypeCaller and UnifiedGenotyper yielded one to two orders of magnitude larger numbers of SNPs and error rates, whereas FreeBayes, SAMtools and VarScan yielded lower numbers of SNPs and more modest error rates. To facilitate comparison between variant callers we standardized each SNP set to the same number of SNPs using additional filtering, where UnifiedGenotyper consistently produced the smallest proportion of genotype errors, followed by HaplotypeCaller, VarScan, SAMtools, and FreeBayes. Additionally, we found that error rates were minimized for SNPs called by more than one variant caller. Finally, we evaluated the performance of various commonly used filtering metrics on SNP calling. Our analysis provides a quantitative assessment of the accuracy of five widely used variant calling programs and offers valuable insights into both the choice of variant caller program and the choice of filtering metrics, especially for researchers using non-model study systems.