Genomes from historical Drosophila melanogaster specimens illuminate adaptive and demographic changes across more than 200 years of evolution.

The ability to perform genomic sequencing on long-dead organisms is opening new frontiers in evolutionary research. These opportunities are especially notable in the case of museum collections, from which countless documented specimens may now be suitable for genomic analysis-if data of sufficient quality can be obtained. Here, we report 25 newly sequenced genomes from museum specimens of the model organism Drosophila melanogaster, including the oldest extant specimens of this species. By comparing historical samples ranging from the early 1800s to 1933 against modern-day genomes, we document evolution across thousands of generations, including time periods that encompass the species' initial occupation of northern Europe and an era of rapidly increasing human activity. We also find that the Lund, Sweden population underwent local genetic differentiation during the early 1800s to 1933 interval (potentially due to drift in a small population) but then became more similar to other European populations thereafter (potentially due to increased migration). Within each century-scale time period, our temporal sampling allows us to document compelling candidates for recent natural selection. In some cases, we gain insights regarding previously implicated selection candidates, such as ChKov1, for which our inferred timing of selection favors the hypothesis of antiviral resistance over insecticide resistance. Other candidates are novel, such as the circadian-related gene Ahcy, which yields a selection signal that rivals that of the DDT resistance gene Cyp6g1. These insights deepen our understanding of recent evolution in a model system, and highlight the potential of future museomic studies.

Driving stroke quality improvement at scale in EDs across a nationwide network of hospitals: strategies and interventions.

OBJECTIVES: Reducing the treatment time while increasing the proportion of eligible stroke patients who receive intravenous tissue plasminogen activator (tPA) has been a priority for many quality improvement efforts. Recent studies have primarily focused on identifying interventions that reduce door-to-needle (DTN) time, while comparatively little has been done to determine whether these interventions also improve tPA rates. METHODS: In order to investigate interventions related to process improvements, an electronic dashboard serving as a stroke performance tool was implemented to store and retrieve patient outcome data. These data were used to study the efficacy of interventions designed to facilitate triage of stroke patients in the ED, and determine the individual interventions associated with the most significant improvements in the fraction of patients receiving tPA and in reducing the DTN time. Stroke performance data from the dashboard collected over a 2-year period (2015-2017) from 89 US hospitals were analysed with respect to interventions implemented by individual facilities, as verified by a hospital survey. RESULTS: A statistically significant association was found between increases in the fraction of patients receiving tPA and reductions in DTN time over the study period. These improvements in outcomes were most strongly associated with process interventions that allocate stroke-specific physical and human resources in the ED, most notably a designated emergency room space for stroke, and with workflows that decrease the time to key checkpoints for determining a patient's eligibility for tPA. CONCLUSIONS: Data from the stroke performance tool was leveraged to identify the programmes and process interventions that lead to improved patient outcomes and allow EDs to better prioritise process interventions and resources.

Higher Incidence of Ischemic Stroke in Patients Taking Novel Oral Anticoagulants.

Background and Purpose- The increased use of novel oral anticoagulants (NOACs) to control atrial fibrillation is largely driven by the assumption that they are equally effective as warfarin at preventing ischemic stroke while putting patients at lower risk of hemorrhages. To test this hypothesis, a retrospective study of the relative incidence of strokes among patients taking NOACs versus those taking warfarin is performed. Methods- Relative stroke incidence in the 2 groups of patients was compared using odds ratios and Fisher exact tests for significance using a data set of 71 365 on NOACs and 59 546 patients on warfarin. In addition, the 7033 patients with a record of both warfarin and NOAC use were analyzed as a separate cohort. Results- There is a significantly higher (odds ratio=1.29, <0.001) frequency of ischemic strokes among patients prescribed NOACs compared with those on warfarin. The relative frequency of ischemic strokes was also higher for every individual NOAC compared with warfarin (these higher frequencies are statistically significant for dabigatran and apixaban, though not for edoxaban and rivaroxaban). There is a lower incidence of intracranial hemorrhages and nontraumatic hemorrhages in general among patients taking NOACs, consistent with the published literature. Comparisons of the demographic and clinical profiles of the patients taking NOACs to those on warfarin do not show significantly higher background stroke risk in NOAC patients; in fact, patients on NOACs tend to be at lower background risk overall for ischemic strokes. Conclusions- Because NOAC use is associated with higher ischemic stroke risk together with a lower risk of hemorrhages than warfarin use, it can be concluded that patients on warfarin are more strongly anticoagulated. The observed effect could be a secondary consequence of dosage control or alternatively a result of different anticoagulant effects among the different medications.

Variance in estimated pairwise genetic distance under high versus low coverage sequencing: The contribution of linkage disequilibrium.

The mean pairwise genetic distance among haplotypes is an estimator of the population mutation rate θ and a standard measure of variation in a population. With the advent of next-generation sequencing (NGS) methods, this and other population parameters can be estimated under different modes of sampling. One approach is to sequence individual genomes with high coverage, and to calculate genetic distance over all sample pairs. The second approach, typically used for microbial samples or for tumor cells, is sequencing a large number of pooled genomes with very low individual coverage. With low coverage, pairwise genetic distances are calculated across independently sampled sites rather than across individual genomes. In this study, we show that the variance in genetic distance estimates is reduced with low coverage sampling if the mean pairwise linkage disequilibrium weighted by allele frequencies is positive. Practically, this means that if on average the most frequent alleles over pairs of loci are in positive linkage disequilibrium, low coverage sequencing results in improved estimates of θ, assuming similar per-site read depths. We show that this result holds under the expected distribution of allele frequencies and linkage disequilibria for an infinite sites model at mutation-drift equilibrium. From simulations, we find that the conditions for reduced variance only fail to hold in cases where variant alleles are few and at very low frequency. These results are applied to haplotype frequencies from a lung cancer tumor to compute the weighted linkage disequilibria and the expected error in estimated genetic distance using high versus low coverage.

Rapid and convergent evolution in the glioblastoma multiforme genome.

Determining which mutations drive tumor progression is a defining question in cancer genomics. We analyzed sequence evolution in Glioblastoma multiforme (GBM) by computing the number of parallel mutations and by estimating ω=dN/dS, a measure of the strength and direction of selection. The ω values of almost all 7617 mutated genes in GBM are much higher than in germline genes. We identified only 21 genes under significant positive selection in GBM, as well as 29 genes under significant purifying selection, including several zinc finger proteins. Therefore, most of the high ω values in the GBM genome are due to weaker purifying selection rather than positive selection. We also found multiple recurrent mutations in GBM, several of which are associated with patient survival time. Our results suggest that convergence and neutral evolution play a significant role in GBM, and that sites with recurrent mutations can serve as molecular diagnostics of the clinical course of GBM tumors.

An eQTL analysis of the human glioblastoma multiforme genome.

In this paper we use eQTL mapping to identify associations between gene dysregulation and single nucleotide polymorphism (SNP) genotypes in glioblastoma multiforme (GBM). A set of 532,954 SNPs was evaluated as predictors of the expression levels of 22,279 expression probes. We identified SNPs associated with fold change in expression level rather than raw expression levels in the tumor. Following adjustment for false discovery rate, the complete set of probes yielded 9257 significant associations (p<0.05). We found 18 eQTLs that were missense mutations. Many of the eQTLs in the non-coding regions of a gene, or linked to nearby genes, had large numbers of significant associations (e.g. 321 for RNASE3, 101 for BNC2). Functional enrichment analysis revealed that the expression probes in significant associations were involved in signal transduction, transcription regulation, membrane function, and cell cycle regulation. These results suggest several loci that may serve as hubs in gene regulatory pathways associated with GBM.

The origin and evolution of vertebrate glycine transporters.

In the vertebrate central nervous system, glycinergic neurotransmission is regulated by the action of the glycine transporters 1 and 2 (GlyT1 and GlyT2)--members of the solute carrier family 6 (SLC6). Several invertebrate deuterostomes have two paralogous glycine carrier genes, with one gene in the pair having greater sequence identity and higher alignment scores with respect to GlyT1 and the other paralog showing greater similarity to GlyT2. In phylogenetic trees, GlyT2-like sequences from invertebrate deuterostomes form a monophyletic subclade with vertebrate GlyT2, while invertebrate GlyT1-like proteins constitute an outgroup to both the GlyT2-like proteins and to vertebrate GlyT1 sequences. These results are consistent with the hypothesis that vertebrate GlyT1 and GlyT2 are, respectively, derived from GlyT1- and GlyT2-like genes in invertebrate deuterostomes. This implies that the gene duplication which gave rise to these paralogs occurred prior to the origin of vertebrates. GlyT2 subsequently diverged significantly from its invertebrate orthologs (i.e., through the acquisition of a unique N-terminus) as a consequence of functional specialization, being expressed principally in the lower CNS; while GlyT1 has activity in both the lower CNS and several regions of the forebrain.

The Precision and Power of Population Branch Statistics in Identifying the Genomic Signatures of Local Adaptation.

Population branch statistics, which estimate the branch lengths of focal populations with respect to two outgroups, have been used as an alternative to FST-based genome-wide scans for identifying loci associated with local selective sweeps. In addition to the original population branch statistic (PBS), there are subsequently proposed branch rescalings: normalized population branch statistic (PBSn1), which adjusts focal branch length with respect to outgroup branch lengths at the same locus, and population branch excess (PBE), which also incorporates median branch lengths at other loci. PBSn1 and PBE have been proposed to be less sensitive to allele frequency divergence generated by background selection or geographically ubiquitous positive selection rather than local selective sweeps. However, the accuracy and statistical power of branch statistics have not been systematically assessed. To do so, we simulate genomes in representative large and small populations with varying proportions of sites evolving under genetic drift or background selection (approximated using variable Ne), local selective sweeps, and geographically parallel selective sweeps. We then assess the probability that local selective sweep loci are correctly identified as outliers by FST and by each of the branch statistics. We find that branch statistics consistently outperform FST at identifying local sweeps. When background selection and/or parallel sweeps are introduced, PBSn1 and especially PBE correctly identify local sweeps among their top outliers at a higher frequency than PBS. These results validate the greater specificity of rescaled branch statistics such as PBE to detect population-specific positive selection, supporting their use in genomic studies focused on local adaptation.

The influence of demographic stochasticity on evolutionary dynamics and stability.

We derive the frequency-dependent selection coefficient caused by "demographic" stochasticity resulting from the random sampling of opponents an individual faces during behavioral "contests" with other individuals. The mean, variance, and higher moments of fitness all influence the direction and strength of selection. A frequency-dependent trait can be stable when an individual's fitness depends upon an infinite number of contests with other individuals and unstable when it depends upon a finite number of contests. Conversely, unstable equilibria for an infinite number of contests can be stable when there is a finite number of contests. At stable equilibria for a finite number of contests, higher moments of fitness can outweigh the joint influence of the first two moments so that natural selection favors "within-generation" or developmental-trait variation (also known as phenotypic plasticity) contrary to the claim that natural selection always acts against such variation. We use second-moment estimates of the fitness functions in a diffusion approximation to compute fixation probabilities of competing strategies. These estimates are shown to be qualitatively consistent with those derived from simulations when population sizes are sufficiently large to ignore the contribution of higher-moment terms. We also show that explicit solutions to the diffusion approximation only exist for pair-wise interactions that lead to positive frequency-dependent selection.

Lipidomic analysis reveals that phosphatidylglycerol and phosphatidylethanolamine are newly generated phospholipids in an early-divergent protozoan, Giardia lamblia.

The pathogenic protozoan Giardia lamblia is known to not synthesize membrane lipids de novo. Therefore, it is possible that lipids in the small intestine, where trophozoites colonize, play key roles in regulating the growth and differentiation of this important pathogen. The focus of the current study is to conduct a complete lipidomic analysis and to test the hypothesis that Giardia has some ability to generate new phospholipids (PLs). Using mass spectrometry, now we show that phosphatidylglycerols (PGs) are major PLs followed by phosphatidylcholines (PCs) and phosphatidylethanolamines (PEs) in non-encysting and encysting trophozoites, as well in cysts. The fatty acids attached to these PLs consist mostly of palmitate, palmitoleate, oleate, and linoleate. Results also indicate that PGs and PEs, unlike PCs, are not present in bovine bile and serum, the major sources of lipids of the culture medium, and that they could therefore be produced by fatty acid and headgroup remodeling reactions, circumventing the synthesis of entirely new PLs via de novo pathways. Genomic and transcriptional analyses show the presence of giardial phosphatidylglycerolphosphate synthase (gpgps) and phosphatidylserine decarboxylase (gpsd) genes, which are expressed throughout the life cycle. Bioinformatic and phylogenetic analyses further indicated that both genes are of prokaryotic origin and that they have undergone duplication in the course of evolution. Our studies suggest that the abundance of PG in Giardia is unique among eukaryotes and that its synthesis thus could serve as a potential target for developing new therapies against this waterborne parasite.

Discovery of stroke-related blood biomarkers from gene expression network models.

BACKGROUND: Identifying molecular biomarkers characteristic of ischemic stroke has the potential to aid in distinguishing stroke cases from stroke mimicking symptoms, as well as advancing the understanding of the physiological changes that underlie the body's response to stroke. This study uses machine learning-based analysis of gene co-expression to identify transcription patterns characteristic of patients with acute ischemic stroke. METHODS: Mutual information values for the expression levels among 13,243 quantified transcripts were computed for blood samples from 82 stroke patients and 68 controls to construct a co-expression network of genes (separately) for stroke and control samples. Page rank centrality scores were computed for every gene; a gene's significance in the network was assessed according to the differences in their network's pagerank centrality between stroke and control expression patterns. A hybrid genetic algorithm - support vector machine learning tool was used to classify samples based on gene centrality in order to identify an optimal set of predictor genes for stroke while minimizing the number of genes in the model. RESULTS: A predictive model with 89.6% accuracy was identified using 6 network-central and differentially expressed genes (ID3, MBTPS1, NOG, SFXN2, BMX, SLC22A1), characterized by large differences in association network connectivity between stroke and control samples. In contrast, classification models based solely on individual genes identified by significant fold-changes in expression level provided lower predictive accuracies: < 71% for any single gene, and even models with larger (10-25) numbers of gene transcript biomarkers gave lower predictive accuracies (≤ 82%) than the 6 network-based gene signature classification. miRNA:mRNA target prediction computational analysis revealed 8 differentially expressed micro-RNAs (miRNAs) that are significantly associated with at least 2 of the 6 network-central genes. CONCLUSIONS: Network-based models have the potential to identify a more statistically robust pattern of gene expression typical of acute ischemic stroke and to generate hypotheses about possible interactions among functionally relevant genes, leading to the identification of more informative biomarkers.

Regorafenib Prior to Selective Internal Radiation Therapy Using 90Y-Resin Microspheres for Refractory Metastatic Colorectal Cancer Liver Metastases: Analysis of Safety, Dosimetry, and Molecular Markers.

Background: This Phase II, open-label, study examined the safety of regorafenib followed by selective internal radiation therapy (SIRT) with regorafenib re-initiation in the treatment of metastatic colorectal cancer (mCRC) patients with liver metastases who are not surgical candidates. Methods: Patients received 160 mg regorafenib daily on a 21-day course followed by a 1 week washout prior to SIRT. Liver function was evaluated at 2 and 4 weeks after SIRT, and regorafenib re-initiated if liver function was normal. Patients were evaluated for safety, and restaged at weeks 6 and 12 following SIRT. In addition, protein and cytokine assays of blood were performed to identify candidate molecular biomarkers associated with outcomes. Individual patient voxel-based dosimetry assessment was performed post-SIRT. Results: Twenty-Five patients were enrolled and received a median 11 weeks regorafenib. Three patients received regorafenib, but not SIRT due to disease progression. The remaining 22 patients received SIRT with a median activity delivered to the liver of 38 mCi, mean normal liver dose of 14.98 Gy and tumor mean dose of 29.0 Gy with a tumor to normal ratio mean of 2.42. There were four treatment-related serious AEs and no treatment-related deaths. Median progression-free survival was 3.7 months and the median overall survival was 12.1 months. The relative densities of several biomolecules changed significantly during the course of treatment, most notably post-treatment increases in levels of sex-hormone binding globulin (SHBG) and decreased levels of the cytokine MIG (CXL9). Decreases in von Willebrand factor (VWF), the ankyrin repeat domain (ANKRD26), and MIG were associated with improved survival times. Post-treatment increases in alpha-2-macroglobulin (A2M) and the cytokine intercellular adhesion molecule (ICAM-1) were associated with reduced overall survival time, while increases in Eotaxin (CCL14) predicted longer overall survival times. Conclusions: The treatment of mCRC patients with liver metastases using regorafenib followed by SIRT was tolerable in this patient population. Further efficacy analysis of this treatment schema and analysis of potential molecular biomarkers using larger sample sizes is merited.

Novel role of sphingolipid synthesis genes in regulating giardial encystation.

Although encystation (cyst formation) is important for the survival of Giardia lamblia outside its human host, the molecular events that prompt encystation have not been fully elucidated. Here, we demonstrate that sphingolipids (SLs), which are important for the growth and differentiation of many eukaryotes, play key roles in giardial encystation. Transcriptional analyses showed that only three genes in the SL biosynthesis pathways are expressed and transcribed differentially in nonencysting and encysting Giardia trophozoites. While the putative homologues of giardial serine palmitoyltransferase (gSPT) subunit genes (gspt-1 and -2) are differentially expressed in nonencysting and encysting trophozoites, the giardial ceramide glucosyltransferase 1 gene (gglct-1) is transcribed only in encysting cells. l-Cycloserine, an inhibitor of gSPT, inhibited the endocytosis and endoplasmic reticulum/perinuclear targeting of bodipy-ceramide in trophozoites, and this could be reversed by 3-ketosphinganine. On the other hand, D-threo-1-phenyl-2-palmitoylamino-3-morpholino-1-propanol (PPMP), an inhibitor of glucosylceramide synthesis, blocked karyokinesis and reduced cyst production in culture. PPMP also altered the expression of cyst wall protein transcripts in encysting cells. Phylogenetic analyses revealed that the gspt genes are paralogs derived from an ancestral spt sequence that underwent gene duplication early in eukaryotic history. This ancestral sequence, in turn, was probably derived from prokaryotic aminoacyl transferases. In contrast, gglct-1 is found in both prokaryotes and eukaryotes without any evidence of gene duplication. These studies indicate that SL synthesis genes are involved in key events in giardial biology and could serve as potential targets for developing new therapies against giardiasis.

Selection against demographic stochasticity in age-structured populations.

It has been shown that differences in fecundity variance can influence the probability of invasion of a genotype in a population; i.e., a genotype with lower variance in offspring number can be favored in finite populations even if it has a somewhat lower mean fitness than a competitor. In this article, Gillespie's results are extended to population genetic systems with explicit age structure, where the demographic variance (variance in growth rate) calculated in the work of Engen and colleagues is used as a generalization of "variance in offspring number" to predict the interaction between deterministic and random forces driving change in allele frequency. By calculating the variance from the life-history parameters, it is shown that selection against variance in the growth rate will favor a genotypes with lower stochasticity in age-specific survival and fertility rates. A diffusion approximation for selection and drift in a population with two genotypes with different life-history matrices (and therefore different mean growth rates and demographic variances) is derived and shown to be consistent with individual-based simulations. It is also argued that for finite populations, perturbation analyses of both the mean and the variance in growth rate may be necessary to determine the sensitivity of fitness to changes in the life-history parameters.

Bivalent Chromatin Domains in Glioblastoma Reveal a Subtype-Specific Signature of Glioma Stem Cells.

Glioblastoma multiforme (GBM) can be clustered by gene expression into four main subtypes associated with prognosis and survival, but enhancers and other gene-regulatory elements have not yet been identified in primary tumors. Here, we profiled six histone modifications and CTCF binding as well as gene expression in primary gliomas and identified chromatin states that define distinct regulatory elements across the tumor genome. Enhancers in mesenchymal and classical tumor subtypes drove gene expression associated with cell migration and invasion, whereas enhancers in proneural tumors controlled genes associated with a less aggressive phenotype in GBM. We identified bivalent domains marked by activating and repressive chromatin modifications. Interestingly, the gene interaction network from common (subtype-independent) bivalent domains was highly enriched for homeobox genes and transcription factors and dominated by SHH and Wnt signaling pathways. This subtype-independent signature of early neural development may be indicative of poised dedifferentiation capacity in glioblastoma and could provide potential targets for therapy.Significance: Enhancers and bivalent domains in glioblastoma are regulated in a subtype-specific manner that resembles gene regulation in glioma stem cells. Cancer Res; 78(10); 2463-74. ©2018 AACR.

A Comparative Study of the Molecular Characteristics of Familial Gliomas and Other Cancers.

BACKGROUND: Familial cancers are those that co-occur among first-degree relatives without showing Mendelian patterns of inheritance. MATERIALS AND METHODS: In this analysis, we compare the genomic characteristics of familial and sporadic cancers, with a focus on low-grade gliomas (LGGs) using sequence and expression data from the Cancer Genome Atlas. RESULTS: Familial cancers show similar genomic and molecular biomarker profiles to sporadic cancers, consistent with the similarity in their clinical features. There are no statistically significant differences among somatic mutation, copy number variant, or gene expression patterns between familial and sporadic cancers; methylation profiles are the only class of molecular data to show significant differences. CONCLUSION: Familial cancers are likely driven by multiple, individually weak contributions to familiality (i.e. large numbers of alleles and/or shared environmental risks). Consequently, these risk factors tend to be obscured by stronger confounding variables such as clinical or molecular variation among cancer subtypes.

Molecular Predictors of Long-Term Survival in Glioblastoma Multiforme Patients.

Glioblastoma multiforme (GBM) is the most common and aggressive adult primary brain cancer, with <10% of patients surviving for more than 3 years. Demographic and clinical factors (e.g. age) and individual molecular biomarkers have been associated with prolonged survival in GBM patients. However, comprehensive systems-level analyses of molecular profiles associated with long-term survival (LTS) in GBM patients are still lacking. We present an integrative study of molecular data and clinical variables in these long-term survivors (LTSs, patients surviving >3 years) to identify biomarkers associated with prolonged survival, and to assess the possible similarity of molecular characteristics between LGG and LTS GBM. We analyzed the relationship between multivariable molecular data and LTS in GBM patients from the Cancer Genome Atlas (TCGA), including germline and somatic point mutation, gene expression, DNA methylation, copy number variation (CNV) and microRNA (miRNA) expression using logistic regression models. The molecular relationship between GBM LTS and LGG tumors was examined through cluster analysis. We identified 13, 94, 43, 29, and 1 significant predictors of LTS using Lasso logistic regression from the somatic point mutation, gene expression, DNA methylation, CNV, and miRNA expression data sets, respectively. Individually, DNA methylation provided the best prediction performance (AUC = 0.84). Combining multiple classes of molecular data into joint regression models did not improve prediction accuracy, but did identify additional genes that were not significantly predictive in individual models. PCA and clustering analyses showed that GBM LTS typically had gene expression profiles similar to non-LTS GBM. Furthermore, cluster analysis did not identify a close affinity between LTS GBM and LGG, nor did we find a significant association between LTS and secondary GBM. The absence of unique LTS profiles and the lack of similarity between LTS GBM and LGG, indicates that there are multiple genetic and epigenetic pathways to LTS in GBM patients.

The role of deleterious mutations in allopatric speciation.

Allopatric speciation is often assumed to occur as a consequence of adaptive divergence between two isolated populations. However, there are some scenarios in which reproductive isolation can be favored due to accumulated unconditionally deleterious mutations. If deleterious mutations have synergistic epistatic effects, it is shown here that the average fitness of recombinants between two parental lines with a given number of fixed mutations is lower than that of the parents in both the F1 and F2 generations. If individual mutations are only slightly deleterious, then they will tend to fixation at a high enough rate to cause lower hybrid fitness. If the fitness effects of mutation give rise to antagonistic epistasis, the hybrids tend to have a higher average fitness than the parental lines, suggesting a possible scenario for the origin of hybrid vigor. The other model of deleterious mutations investigated is the accumulation of knockout mutants in a duplicated gene family. While neutral in the parental lines, upon contact the F1 and later generations have a significant probability of carrying double knockouts. Under this scenario, selection may also favor reproductive isolation between the two lines. Even when the selection coefficients generated are too low to drive speciation, epistatic interactions between deleterious mutations offer a possible explanation for both outbreeding depression and hybrid vigor.

Evolution of variance in offspring number: the effects of population size and migration.

It was shown by Gillespie [1974. Am. Nat. 108, 145-151], that if two genotypes produce the same average number of offspring on but have a different variance associated within each generation, the genotype with a lower variance will have a higher effective fitness. Specifically, the effective fitness is w(e)=w-sigma(2)/N, where w is the mean fitness, sigma(2) is the variance in offspring number, and N is the total population size. The model also predicts that if a strategy has a higher arithmetic mean fitness and a higher variance than the competitor, the outcome of selection will depend on the population size (with larger population sizes favoring the high-variance, high-mean genotype). This suggests that for metapopulations with large numbers of (relatively) small demes, a strategy with lower variance and lower mean may be favored if the migration rate is low while higher migration rates (consistent with a larger effective population size) favor the opposite strategy. Individual-based simulation confirms that this is indeed the case for an island model of migration, though the effect of migration differs greatly depending on whether migration precedes or follows selection. It is noted in the appendix that while Gillespie [1974. Am. Nat. 108, 145-151] does seem to be heuristically accurate, it is not clear that the definition of effective fitness follows from his derivation.