THE EVOLUTION OF GENETIC CORRELATIONS: AN ANALYSIS OF PATTERNS

The genetic correlation is a central parameter of quantitative genetics, providing a measure of the rate at which traits respond to indirect selection (i.e., selection that does not act upon the traits under study, but some other trait with which they have genes in common). In this paper, I review the pattern of variation among four combinations of traits: life history × life history (L × L), morphological × morphological (M × M), life history × morphological (L × M), and behavioral × behavioral (B × B). A few other combinations were investigated, but insufficient data were obtained for separate analysis. A total of 1798 correlations, distributed over 51 different animal and plant species, were analyzed. The analysis was conducted at two levels: first by dividing the data set solely by trait combination, and second by blocking the data by trait combination and species. Because selection will tend to fix alleles that show positive correlations with fitness traits faster than those that are negative and because the latter are expected to arise more frequently by mutation, correlations between life-history traits are predicted to be more often negative than those between morphological traits. This prediction was supported, with the ranking in decreasing proportion of negative correlations being: L × L > L × M > B × B > M × M. The mean magnitude of the genetic correlation shows little variation among morphological and life-history combinations, and the distribution of values is remarkably flat. However, the estimated standard errors and the coefficient of variation (SE/r_G) are large, making it difficult to separate biological factors influencing the pattern of dispersion from experimental error. Analysis of the phenotypic and genetic correlations suggest that for the combinations M × M and L × M, but not L × L or B × B, the phenotypic correlation is an adequate estimate of the genetic correlation.     

Path analysis of natural selection on plant chemistry: the xylem resin of ponderosa pine

We analyzed the pattern of correlations among fitness components, herbivory, and resin characteristics in a natural all-aged stand of ponderosa pine, to infer the strength and mechanism of natural selection on plant chemistry. Male and female cone production were monitored yearly for 15 years, and levels of herbivory for 9 years in 165 trees. Resin flow rate and monoterpene composition were determined for these same trees. Multiple regression of fitness components on resin characteristics showed significant associations consistent with directional selection for increased resin flow rates and increased proportions of α- and β-pinene, myrcene and terpinolene. However, negative correlations among monoterpene fractions of the resin constrained the overall selection. Selective herbivory by aphids approached statistical significance and monoterpenes showed some (non-significant) effect as deterrents against deer browse. Resin characteristics were not correlated with attack by cone insects or porcupines. However, the association between resin characteristics and fitness is significantly different from that predicted by the path coefficients involving herbivores. Therefore the hypothesis that these herbivores mediate selection on the resin is not supported by the observed pattern of correlations among resin characteristics, herbivory, growth and fecundity. In this population, most of the association between resin characteristics and fitness appears to be mediated by some other factor independent of attack by herbivore species present. Received: 18 March 1996/Accepted: 18 July 1996     

EXPERIMENTAL EVOLUTION OF RESISTANCE IN BRASSICA RAPA: CORRELATED RESPONSE OF TOLERANCE IN LINES SELECTED FOR GLUCOSINOLATE CONTENT

The evolutionary response of plant populations to selection for increased defense may be constrained by costs of defense. The purpose of this study was to investigate such constraints on the evolution of defense due to a cost of defense manifested as a trade-off between defense and tolerance. Variation in the response to artificial damage (tolerance) among lines of Brassica rapa that had been artificially selected for foliar glucosinolate content (defense) was examined. Leaf area was removed from replicates of three selection lines (high glucosinolates, control, and low glucosinolates) at three damage levels (0%, 20%, and 60% damage). An external cost of defense would result in a statistically significant selection line by damage treatment interaction, with those selected for high defense expressing less tolerance than those selected for low defense. Damage treatment had a significant overall effect on estimated total fitness, with fitness declining with increasing damage level. Further, selection line also had a significant overall effect on estimated total fitness, with low-defense selection lines having higher fitness compared to both control and high-defense selection lines. More importantly, a cost of defense in terms of tolerance was demonstrated by a significant selection line-by-damage treatment interaction. This interaction was in the direction to demonstrate a genetic trade-off between defense and tolerance, with low-defense selection lines decreasing estimated total fitness in response to damage less than both control and high-defense selection lines. Variation in tolerance among selection lines was due to the greater ability of low-defense lines to maintain fruit and seed production despite the presence of damage. In terms of tolerance, this cost of glucosinolate production in B. rapa could constrain the evolution of increased defense and, in so doing, maintain individuals within the population that are poorly defended yet tolerant.     

Settling-site selection and survival of two scale insects,Ceroplastes rubens andC. ceriferus,on citrus trees

We studied settling-site selection and the resulting survival of two sessile scale insects, Ceroplastes rubens and C. ceriferus, in the citrus tree, Citrus unshiu, in central Japan. C. rubens preferred 0-year-old twigs most as a settling-site; the density of nymphs settling on 0-year-old twigs was significantly higher than those on ≥1-year-old twigs, and few nymphs settled on ≥3-year-old twigs. The mean survival rates from settling until reproduction in the next year were significantly higher on more preferred twigs than on less preferred ones. In C. ceriferus, nymphs significantly preferred 1- and 2-year-old twigs to 0- and ≥3-year-old ones, and the mean survival rates on the more preferred 1- and 2-year-old twigs were significantly higher than those on less preferred ≥3-year-old twigs. However, the survival rate on less preferred 0-year-old twigs was slightly higher than those on 1- and 2-year-old ones. Thus, in both species of scale, it was the preferred twigs which were more profitable sites for survival after settling, except for less preferred 0-year-old twigs for C. ceriferus. In both scale species, most mortality was due to growth cessation, which is believed to be related to the twig quality as a food source. Predators and parasitoids were minor mortality factors. Both species showed constant survival rates until the density of settled nymphs exceeded double the “upper-limit” density, whereupon they decreased drastically. Nymphs of C. rubens settling on twigs of high scale density showed a spacing-out distribution, those of C. ceriferus did not. In C. rubens, an increase in preference for originally less profitable twigs at the later stage of the settling season was observed, but not in C. ceriferus. Accordingly, individuals of C. rubens showed a stronger tendency to avoid conspecifics than did C. ceriferus. Although nymphs of the two scales clearly preferred more profitable sites, their settling-site selection did not agree with the predictions from the ideal free distribution theory (Fretwell and Lucas, 1970). The discrepancies were (1) frequent settling on less profitable sites at the early stage of the settling season, (2) insufficient utilization of the most profitable twigs, and (3) virtually 100% mortality on overcrowded twigs under conditions where unoccupied profitable twigs still remained. These discrepancies are thought due to the limited dispersal time of nymphs.     

Selection and proliferation of rapid growing cell lines from embryo derived cell cultures of yew tree (Taxus cuspidata Sieb. et Zucc)

Calli were induced from 300,000 embryos isolated from immature to mature stage of seeds collected on late September from 14 elite trees. When the embryos were cultured onto plastic Petri-dish containing 20 mL of modified B5 basal medium supplemented with 3% (w/v) sucrose, 500 mg/L casein hydrolysate, 250 mg/L myo-inositol, 0.5% (w/v) polyvinyl polypyrrolidon (PVPP), 2×MS vitamins, 0.5 mg/L gibberellic acid, and 10 mg/L 2,4-D after 2 weeks of culture, yellowish-white calli were immediately formed on the surfaces of embryos, and subcultured for 4 weeks in same culture medium. Because most of calli maintained for more than 3 months were revealed differences in their colors, surface texture, and growth rate, visual selection was made for first round screening. When the size of visually selected calli larger than 19 mm in their diameter were inoculated, persistent proliferation was observed. Among the plating methods tested for the selection of rapid growing cell lines at single cell and/or small cell aggregate level, 2-layer spread plating revealed as the best for single cell cloning. To enhance cell growth and maintain high rate of viability for long-term culture of yew cells in bioreactor, final cell volume less than 50% in SCV seemed to be the best. Time course study revealed that 30% of inoculum density was suitable for fed batch culture. Among the tested conditional media, the rate of 1∶2 (old medium: fresh medium) was recorded at the best for cell growth.     

EVOLUTION AND EXTINCTION IN A CHANGING ENVIRONMENT: A QUANTITATIVE-GENETIC ANALYSIS

Because of the ubiquity of genetic variation for quantitative traits, virtually all populations have some capacity to respond evolutionarily to selective challenges. However, natural selection imposes demographic costs on a population, and if these costs are sufficiently large, the likelihood of extinction will be high. We consider how the mean time to extinction depends on selective pressures (rate and stochasticity of environmental change, and strength of selection), population parameters (carrying capacity, and reproductive capacity), and genetics (rate of polygenic mutation). We assume that in a randomly mating, finite population subject to density-dependent population growth, individual fitness is determined by a single quantitative-genetic character under Gaussian stabilizing selection with the optimum phenotype exhibiting directional change, or random fluctuations, or both. The quantitative trait is determined by a finite number of freely recombining, mutationally equivalent, additive loci. The dynamics of evolution and extinction are investigated, assuming that the population is initially under mutation-selection-drift balance. Under this model, in a directionally changing environment, the mean phenotype lags behind the optimum, but on the average evolves parallel to it. The magnitude of the lag determines the vulnerability to extinction. In finite populations, stochastic variation in the genetic variance can be quite pronounced, and bottlenecks in the genetic variance temporarily can impair the population's adaptive capacity enough to cause extinction when it would otherwise be unlikely in an effectively infinite population. We find that maximum sustainable rates of evolution or, equivalently, critical rates of environmental change, may be considerably less than 10% of a phenotypic standard deviation per generation.     

Assessment of Optimal Selected Prognostic Factors

The identification and assessment of prognostic factors is one of the major tasks in clinical research. The assessment of one single prognostic factor can be done by recently established methods for using optimal cutpoints. Here, we suggest a method to consider an optimal selected prognostic factor from a set of prognostic factors of interest. This can be viewed as a variable selection method and is the underlying decision problem at each node of various tree building algorithms. We propose to use maximally selected statistics where the selection is defined over the set of prognostic factors and over all cutpoints in each prognostic factor. We demonstrate that it is feasible to compute the approximate null distribution. We illustrate the new variable selection test with data of the German Breast Cancer Study Group and of a small study on patients with diffuse large B‐cell lymphoma. Using the null distribution for a p‐value adjusted regression trees algorithm, we adjust for the number of variables analysed at each node as well. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)