Food shortage is a stress factor that commonly affects organisms in nature. Resistance to food shortage or starvation resistance (SR) is a complex quantitative trait with direct implications on fitness. However, surveys of natural genetic variation in SR at different geographic scales are scarce. Here, we have measured variation in SR in sets of lines derived from nine natural populations of Drosophila melanogaster collected in western Argentina. Our study shows that within population variation explained a larger proportion of overall phenotypic variance (80%) than among populations (7·2%). We also noticed that an important fraction of variation was sex-specific. Overall females were more resistant to starvation than males; however, the magnitude of the sexual dimorphism (SD) in SR varied among lines and explained a significant fraction of phenotypic variance in all populations. Estimates of cross-sex genetic correlations suggest that the genetic architecture of SR is only partially shared between sexes in the populations examined, thus, facilitating further evolution of the SD.

TP5, a P element inserted in the telomere-associated sequences of the X chromosome, represses the excision of other P elements in the germ line through a combination of maternal and zygotic effects. The maternal component of this repression is impaired by heterozygous mutations in the aubergine and Suppressor of variegation 205 genes; one mutation in the piwi gene also appears to impair repression. In the female germ line, the level of TP5 mRNA is increased by these impairing mutations. The impairing aubergine and piwi mutations also increase the level of germ-line mRNA from CP, a transgene that encodes the P-element transposase; however, the Suppressor of variegation 205 mutation does not. These findings are discussed in terms of a model of P-element regulation that involves post-transcriptional and chromatin re-organizing events mediated by maternally transmitted small RNAs derived from the telomeric P element.

Historical information can be used, in addition to pedigree, traits and genotypes, to map quantitative trait locus (QTL) in general populations via maximum likelihood estimation of variance components. This analysis is known as linkage disequilibrium (LD) and linkage mapping, because it exploits both linkage in families and LD at the population level. The search for QTL in the wild population of Soay sheep on St. Kilda is a proof of principle. We analysed the data from a previous study and confirmed some of the QTLs reported. The most striking result was the confirmation of a QTL affecting birth weight that had been reported using association tests but not when using linkage-based analyses.

Hu & Xu (2008) developed a statistical method for computing the statistical power for detecting a quantitative trait locus (QTL) located in a marker interval. Their method is based on the regression interval mapping method and allows experimenters to effectively investigate the power for detecting a QTL in a population. This paper continues to work on the power analysis of separating multiple-linked QTLs. We propose simple formulae to calculate the power of separating closely linked QTLs located in marker intervals. The proposed formulae are simple functions of information numbers, variance inflation factors and genetic parameters of a statistical model in a population. Both regression and maximum likelihood interval mappings suitable for detecting QTL in the marker intervals are considered. In addition, the issue of separating linked QTLs in the progeny populations from an F2 subject to further self and/or random mating is also touched upon. One of the primary keys to our approach is to derive the genotypic distributions of three and four loci for evaluating the correlation structures between pairwise unobservable QTLs in the model across populations. The proposed formulae allow us to predict the power of separation when several factors, such as sample sizes, sizes and directions of QTL effects, distances between QTLs, interval sizes and relative QTL positions in the intervals, are considered together at a time in different experimental populations. Numerical justifications and Monte Carlo simulations were provided for confirmation and illustration.

Prediction of genetic values is a central problem in quantitative genetics. Over many decades, such predictions have been successfully accomplished using information on phenotypic records and family structure usually represented with a pedigree. Dense molecular markers are now available in the genome of humans, plants and animals, and this information can be used to enhance the prediction of genetic values. However, the incorporation of dense molecular marker data into models poses many statistical and computational challenges, such as how models can cope with the genetic complexity of multi-factorial traits and with the curse of dimensionality that arises when the number of markers exceeds the number of data points. Reproducing kernel Hilbert spaces regressions can be used to address some of these challenges. The methodology allows regressions on almost any type of prediction sets (covariates, graphs, strings, images, etc.) and has important computational advantages relative to many parametric approaches. Moreover, some parametric models appear as special cases. This article provides an overview of the methodology, a discussion of the problem of kernel choice with a focus on genetic applications, algorithms for kernel selection and an assessment of the proposed methods using a collection of 599 wheat lines evaluated for grain yield in four mega environments.

Analysis of the temporal variation in allele frequencies is useful for studying microevolutionary processes. However, many statistical methods routinely used to test temporal changes in allele frequencies fail to establish a proper hypothesis or have theoretical or practical limitations. Here, a Bayesian statistical test is proposed in which the distribution of the distances among sampling frequencies is approached with computer simulations, and hypergeometric sampling is considered instead of binomial sampling. To validate the test and compare its performance with other tests, agent-based model simulations were run for a variety of scenarios, and two real molecular databases were analysed. The results showed that the simulation test (ST) maintained the significance value used (α=0·05) for a vast combination of parameter values, whereas other tests were sensitive to the effect of genetic drift or binomial sampling. The differences between binomial and hypergeometric sampling were more complex than expected, and a novel effect was described. This study suggests that the ST is especially useful for studies with small populations and many alleles, as in microsatellite or sequencing molecular data.