This study is concerned with statistical methods used for the analysis of comparative data (in which observations are not expected to be independent because they are sampled across phylogenetically related species). The phylogenetically independent contrasts (PIC), phylogenetic generalized least-squares (PGLS), and phylogenetic autocorrelation (PA) methods are compared. Although the independent contrasts are not orthogonal, they are independent if the data conform to the Brownian motion model of evolution on which they are based. It is shown that uncentered correlations and regressions through the origin using the PIC method are identical to those obtained using PGLS with an intercept included in the model. The PIC method is a special case of PGLS. Corrected standard errors are given for estimates of the ancestral states based on the PGLS approach. The treatment of trees with hard polytomies is discussed and is shown to be an algorithmic rather than a statistical problem. Some of the relationships among the methods are shown graphically using the multivariate space in which variables are represented as vectors with respect to OTUs used as coordinate axes. The maximum-likelihood estimate of the autoregressive parameter, ρ, has not been computed correctly in previous studies (an appendix with MATLAB code provides a corrected algorithm). The importance of the eigenvalues and eigenvectors of the connection matrix, W, for the distribution of ρ is discussed. The PA method is shown to have several problems that limit its usefulness in comparative studies. Although the PA method is a generalized least-squares procedure, it cannot be made equivalent to the PGLS method using a phylogenetic model.

Corresponding Editor: T. Garland Jr.

This paper examines aspects of genetic draft, the stochastic force induced by substitutions at one locus on the dynamics of a closely linked locus. Of particular interest is the role of population size on genetic draft. Remarkably, the rate of substitution of weakly selected advantageous mutations decreases with increasing population size, whereas that for deleterious mutations increases with population size. This dependency on population size is the opposite of that for genetic drift. Moreover, these rates are only weakly dependent on population size, again contrary to the strong dependency of drift-based dynamics. Four models of the strongly selected loci responsible for genetic draft are examined. Three of these exhibit a very weak dependency on population size, which implies that their induced effects will also be weakly dependent on population size. Together, these results suggest that population size and binomial sampling may not be relevant to a species' evolution. If this is the case, then a number of evolutionary conundrums are resolved.

Corresponding Editor: A. Caballero

The existence of areas of lower endemism and disjunction of New Zealand biota is typified by Nothofagus beech trees (hence “beech-gap”) and have been attributed to a variety of causes ranging from ancient fault-mediated displacement (20–25 million years ago) to Pleistocene glacial extirpation (<1.8 million years ago). We used cytochrome oxidase I and 12S mtDNA sequence data from a suite of endemic invertebrates to explore phylogeographic depth and patterns in South Island, New Zealand, where the “beech-gap” occurs. Phylogeographic structure and genetic distance data are not consistent with ancient vicariant processes as a source of observed pattern. However, we also find that phylogeographic patterns are not entirely congruent and appear to reflect disparate responses to fragmentation, which we term “gap,” “colonization,” and “regional.” Radiations among congenerics, and in at least one instance within a species, probably took place in the Pliocene (2–7 million years ago), possibly under the influence of the onset of mountain building. This orogenic phase may have had a considerable impact on the development of the biota generally. Some of the taxa that we studied do not appear to have suffered range reduction during Pleistocene glaciation, consistent with their survival throughout that epoch in alpine habitats to which they are adapted. Other taxa have colonized the beech-gap recently (i.e., after glaciation), whereas few among our sample retain evidence of extirpation in the most heavily glaciated zone.

Corresponding Editor: S. Karl

Stable polymorphisms are commonly observed in experimental bacterial populations grown in homogeneous media. Evidence is accumulating that metabolic interactions might be the main mechanism underlying the emergence and maintenance of such polymorphisms. To date, however, attempts to model the evolution of bacterial polymorphism have not considered metabolism as a possible component of polymorphism maintenance. Here, we propose a simulation approach to model the evolution of selected polymorphisms in a bacterial population. Using recent knowledge of the relationship between bacterial fitness and metabolism, we build a simple metabolic model and test the effect of resource competition on polymorphism. Without making an a priori hypothesis on fitness functions, we show that stable polymorphic situations could be observed under high nutrient competition, and we propose a functional, metabolism-based explanation to the debated issue of polymorphism maintenance.

Corresponding Editor: H. Ochman

Stressful environments may be considered as those that reduce fitness, sometimes due in part to the increased metabolic expenditure required to sustain life. Direct adaptation to a stressor is expected to increase fitness and reduce maintenance metabolism, with the latter leading to increased biomass production. In this study, we test the general hypothesis that such adaptation to one stressor can preadapt organisms to novel stressful environments. Six lines of Escherichia coli propagated for 2000 generations at 41–42°C (42 group), a stressful temperature, were compared to six control lines propagated for 2000 generations at 37°C (37 group) and to the common ancestor of both groups. We assayed biovolume yield (a measure of growth efficiency) and competitive fitness in the 42 group's selective high temperature environment as well as five novel stressful environments—acid, alkali, ethanol, high osmolarity and peroxide. As previously reported, at high temperature the 42 group had both higher yield and fitness than the 37 group and ancestor. In the novel environments, the 42 group generally produced yields higher than the 37 group (and marginally higher than the ancestor), but we found no differences in competitive fitness among the 37 and 42 groups and the ancestor. We also found that the performance of lines within groups was not correlated across stressful environments for either yield or relative fitness. Because previous adaptation to one stressor did not improve our measure of Darwinian fitness in novel stressful environments, we conclude that the 42 group shows no useful preadaptation, or cross-tolerance, to these types of environments.

Corresponding Editor: J. Kingsolver

We studied the selection response of the freshwater grazing zooplankter, Daphnia galeata, to increased abundance of cyanobacteria in its environment. Cyanobacteria are a poor-quality and often toxic food. Distinct genotypes of D. galeata were hatched from diapausing eggs extracted from three time horizons in the sediments of Lake Constance, Europe, covering the period 1962 to 1997, a time of change in both the prevalence of planktonic cyanobacteria and levels of phosphorus pollution. We assessed whether the grazers evolved to become more resistant to dietary cyanobacteria by exposing genetically distinct clones to two diets, one composed only of the nutritious green alga, Scenedesmus obliquus (good food), and the other a mixture of S. obliquus and the toxic cyanobacterium Microcystis aeruginosa (poor food). Genotype performance was measured as the specific rate of weight gain from neonate to maturity (g_j).

We evaluated evolutionary change in the Daphnia population using an analysis of reaction norms based on relative (log-transformed) changes in g_j. Log(g_j) is a measure of the proportional effect of dietary cyanobacteria on other fitness components of the Daphnia phenotype. For comparison, we also analyze absolute (i.e., nontransformed) changes in g_j and discuss the interpretations of the two approaches. Statistical results using a general linear model demonstrate a significant effect of genotype (showing differences in g_j among genotypes), a significant genotype × food-type interaction (showing differences in phenotypic plasticity among genotypes), and, in the case of log-transformed data, a significant sediment-genotype-age × food-type interaction. The latter shows that phenotypic plasticity evolved over the period studied.

Two constraints act on response to selection in the D. galeata–Lake Constance system. First, g_j on a diet containing poor food is highly correlated with g_j on a diet of good food, thus evolving resistance also meant evolving an increase in g_j on both diets. Second, because genotypes with a high g_j also grow to a large adult body size, which in turn increases Daphnia vulnerability to fish predation, we suggest that selection only acted to favor genotypes possessing a high potential g_j after cyanobacteria became prevalent. The presence of cyanobacteria depressed realized g_j and led to animals of small adult body size even if their genotypes had the potential for high g_j and large size. With realized g_j reduced, genotypes with an inherently high value could be selected even in the presence of predatory fish. The joint action of selection by dietary cyanobacteria and vulnerability to fish predation provides an explanation for the observed evolution of resistance to poor food through reduced phenotypic plasticity.

Corresponding Editor: A. Caballero

Trends in the evolution of the euglenid pellicle were described using phylogenetic methods on 18S rDNA, morphological, and combined data from 25 mostly phototrophic taxa. The tree topology from a total-evidence analysis formed a template for a synthetic tree that took into account conflicting results derived from the partitioned datasets. Pellicle character states that can only be observed with the assistance of transmission and scanning electron microscopy were phylogenetically mapped onto the synthetic tree to test a set of previously established homology statements (inferences made independently from a cladogram). The results permitted us to more confidently infer the ancestral-derived polarities of character state transformations and provided a framework for understanding the key cytoskeletal innovations associated with the evolution of phototrophic euglenids. We specifically addressed the character evolution of (1) the maximum number of pellicle strips around the cell periphery; (2) the patterns of terminating strips near the cell posterior end; (3) the substructural morphology of pellicle strips; (4) the morphology of the cell posterior tip; and (5) patterns of pellicle pores on the cell surface.

Corresponding Editor: M. Riley

The timing of life-history events in insects can have important consequences for both survival and reproduction. For insect herbivores with complex life histories, selection is predicted to favor those combinations of traits that increase the size at metamorphosis while minimizing the risk of mortality from natural enemies. Studies quantifying selection on life-history traits in natural insect herbivore populations, however, have been rare. The purpose of this study was to measure phenotypic selection imposed by elements of the first and third trophic levels on variation in two life-history traits, the timing of egg hatch and pupal mass, in a population of oak-feeding caterpillars, Psilocorsis quercicella (Lepidoptera: Oecophoridae). Larvae were collected from the field throughout each of two generations per year for three years and reared to determine the effects of the date of egg hatch on both the risk of attack from parasitoids and the pupal mass of the survivors. The direction and strength of phenotypic selection attributed to aspects of the first and third trophic levels, as well as their combined effects, on the date of egg hatch was measured for each of the six generations. Heritabilities of and genetic correlations between pupal mass and the date of adult emergence from diapause (the life-history trait expected to have the largest influence on the timing of egg hatch, and thus larval development) were estimated from laboratory matings. In four of the six generations examined, significant directional selection attributed to the first trophic level was detected, always favoring early-hatching cohorts predicted to experience higher leaf quality than late-hatching cohorts. Directional phenotypic selection by the third trophic level was detected in only one of three years, and in that year the direction of selection was in opposite directions during the two successive generations. The combined effect of selection by both trophic levels indicated that the third trophic level acted to either reduce or enhance the more predictable pattern of selection attributed to the first trophic level. In addition, I found evidence of truncation selection acting to increase the mean and decrease the variance of pupal mass during the pupa-adult transition in the laboratory. Pupal mass and diapause duration were found to vary significantly among full-sibling families; upper bounds for heritability estimates were 0.57 and 0.30, respectively. Furthermore, these two traits were found to be positively genetically correlated (families with larger pupae had longer diapause durations), resulting in a fitness trade-off, because larger pupae enjoy higher survival through metamorphosis and female fecundity but emerge later, when average leaf quality for offspring is generally poorer.

Corresponding Editor: J. Conner

To date, there is still no consensus on the real significance of fluctuating asymmetry (FA) in evolutionary biology. Some studies have established links between FA and Darwinian fitness, and in a number of cases intermediate heritabilities for FA have been reported. However, many claims have been raised against the generality of these findings. I therefore tested if FA of a sexually selected trait (wing length) is indeed related to male mating success in Drosophila buzzatii from field and laboratory samples and whether FA has detectable heritability. Single, unsuccessful males had greater asymmetry for wing length than their mating counterparts both in nature and under nonoptimal rearing environments, but the higher FA in single males is most likely due to a poorer average phenotypic condition because there was no evidence of a genetic basis for this trait. Further evidence of an increase in FA under larval food stress is suggested when comparing the magnitude of the FA levels between stressful and optimal environments. On methodological grounds, a linear model is suggested that allows directional asymmetry (DA) and any genetic variation of DA that may be present to be statistically eliminated from estimates of FA.

Corresponding Editor: J. Cheverud

Recent studies indicate that postcopulatory sexual selection may represent an important component of the speciation process by initiating reproductive isolation via the evolutionary divergence of fertilization systems. Using two geographically isolated populations of the polyandrous beetle Callosobruchus maculatus, we investigated divergence in fertilization systems by determining the extent of postcopulatory functional incompatibility. Through reciprocal, cross-population matings we were able to separately estimate the effects of male and female population origin and their interaction on the extent of last-male sperm precedence, female receptivity to further copulation and female oviposition. Our results indicate partial incompatibility between the fertilization systems of the two populations at all three functional levels. Males derived from the same population as females outcompete rival, allopatric males with respect to sperm preemption, sperm protection, and ability to stimulate female oviposition. This pattern is reciprocated in both populations indicating that postcopulatory, prezygotic events represent important mechanisms by which between-population gene flow is reduced. We suggest the partial gametic isolation observed is a by-product of the coevolution of male and female fertilization systems by a process of cryptic female choice. Our results are consistent with a mechanism akin to conventional mate choice models although they do not allow us to reject antagonistic sexual coevolution as the mechanism of cryptic female choice.

Corresponding Editor: C. López-Fanjul

Acanthochromis polyacanthus is an unusual tropical marine damselfish that uniquely lacks pelagic larvae and has lost the capacity for broad-scale dispersal among coral reefs. On the modern Great Barrier Reef (GBR), three color morphs meet and hydridize at two zones of secondary contact. Allozyme electrophoreses revealed strong differences between morphs from the southern zone but few differences between morphs from the northern counterpart, thus suggesting different contact histories. We explore the phylogeography of Acanthochromis polyacanthus with mitochondrial cytochrome b region sequences (alignment of 565 positions) obtained from 126 individuals representing seven to 12 fish from 13 sites distributed over 12 reefs of the GBR and the Coral Sea. The samples revealed three major clades: (1) black fish collected from the southern GBR; (2) bicolored fish collected from the GBR and one reef (Osprey) from the northern Coral Sea; (3) black and white monomorphs collected from six reefs in the Coral Sea. All three clades were well supported (72–100%) by bootstrap analyses. Sequence divergences were very high between the major clades (mean = 7.6%) as well as within them (2.0–3.6%). Within clades, most reefs segregated as monophyletic assemblages. This was revealed both by phylogenetic analyses and AMOVAs that showed that 72–90% of the variance originated from differences among groups, whereas only 5–13% originated within populations. These patterns are discussed in relation to the known geological history of coral reefs of the GBR and the Coral Sea. Finally, we ask whether the monospecific status of Acanthochromis should be revisited because the sequence divergences found among our samples is substantially greater than those recorded among well-recognized species in other reef fishes.

Corresponding Editor: L. Bernatchez

Classical models of the spatial structure of population genetics rely on the assumption of migration-drift equilibrium, which is seldom met in natural populations having only recently colonized their current range (e.g., postglacial). Population structure then depicts historical events, and counfounding effects due to recent secondary contact between recently differentiated lineages can further counfound analyses of association between geographic and genetic distances. Mitochondrial polymorphisms have revealed the existence of two closely related lineages of the lake cisco, Coregonus artedi, whose significantly different but overlaping geographical distributions provided a weak signal of past range fragmentation blurred by putative subsequent extensive secondary contacts. In this study, we analyzed geographical patterns of genetic variation at seven microsatellite loci among 22 populations of lake cisco located along the axis of an area covered by proglacial lakes 12,000–8000 years ago in North America. The results clearly confirmed the existence of two genetically distinct races characterized by different sets of microsatellite alleles whose frequencies varied clinally across some 3000 km. Equilibrium and nonequilibrium analyses of isolation by distance revealed historical signal of gene flow resulting from the nearly complete admixture of these races following neutral secondary contacts in their historical habitat and indicated that the colonization process occurred by a stepwise expansion of an eastern (Atlantic) race into a previously established Mississippian race. This historical signal of equilibrium contrasted with the current migration-drift disequilibrium within major extant watersheds and was apparently maintained by high effective population sizes and low migration regimes.

Corresponding Editor: T. Smith

The plethodontid salamander Desmognathus orestes, a member of the D. ochrophaeus species complex, is distributed in southwestern Virginia, eastern Tennessee, and western North Carolina. Previous allozyme analyses indicate that D. orestes consists of two distinct groups of populations (D. orestes ‘B’ and D. orestes ‘C’) with extensive intergradation and probable gene flow between these two groups. Spatially varying allele frequencies can reflect historical associations, current gene flow, or a combination of population-level processes. To differentiate among these processes, we use multiple markers to further characterize divergence among populations of D. orestes and assess the degree of intergradation between D. orestes ‘B’ and D. orestes ‘C’, specifically investigating variation in allozymes, mitochondrial DNA (mtDNA), and reproductive behavior among populations. On a broad scale, the mtDNA genealogies reconstruct haplotype clades that correspond to the species identified from previous allozyme analyses. However, at a finer geographic scale, the distributions of the allozyme and mtDNA markers for D. orestes ‘B’ and D. orestes ‘C’ are discordant. MtDNA haplotypes corresponding to D. orestes ‘B’ are more broadly distributed across western North Carolina than predicted by allozyme data, and the region of intergradation with D. orestes ‘C’ indicates asymmetric gene flow of these markers. Asymmetric mating may contribute to observed discordance in nuclear versus cytoplasmic markers. Results support describing D. orestes as a single species and emphasize the importance of using multiple markers to examine fine-scale patterns and elucidate evolutionary processes affecting gene flow when making species-level taxonomic decisions.

Corresponding Editor: B. Bowen

One of the most striking morphological transformations in vertebrate evolution is the transition from a lizardlike body form to an elongate, limbless (snakelike) body form. Despite its dramatic nature, this transition has occurred repeatedly among closely related species (especially in squamate reptiles), making it an excellent system for studying macroevolutionary transformations in body plan. In this paper, we examine the evolution of body form in the lizard family Anguidae, a clade in which multiple independent losses of limbs have occurred. We combine a molecular phylogeny for 27 species, our morphometric data, and phylogenetic comparative methods to provide the first statistical phylogenetic tests of several long-standing hypotheses for the evolution of snakelike body form. Our results confirm the hypothesized relationships between body elongation and limb reduction and between limb reduction and digit reduction. However, we find no support for the hypothesized sequence going from body elongation to limb reduction to digit loss, and we show that a burrowing lifestyle is not a necessary correlate of limb loss. We also show that similar degrees of overall body elongation are achieved in two different ways in anguids, that these different modes of elongation are associated with different habitat preferences, and that this dichotomy in body plan and ecology is widespread in limb-reduced squamates. Finally, a recent developmental study has proposed that the transition from lizardlike to snakelike body form involves changes in the expression domains of midbody Hox genes, changes that would link elongation and limb loss and might cause sudden transformations in body form. Our results reject this developmental model and suggest that this transition involves gradual changes occurring over relatively long time scales.

Corresponding Editor: T. Garland Jr.

The hemiclonal waterfrog Rana esculenta (RL genotype), a bisexual hybrid between R. ridibunda (RR) and R. lessonae (LL), eliminates the L genome from its germline and clonally transmits the R genome (hybridogenesis). Matings between hybrids produce R. ridibunda offspring, but they generally die at an early larval stage. Mortality may be due to fixed recessive deleterious mutations in the clonally inherited R genomes that were either acquired through the advance of Muller's ratchet or else frozen in these genomes at hemiclone formation. From this hypothesis results a straightforward prediction: Matings between different hemiclones, that is, between R. esculenta possessing different R genomes of independent origin, should produce viable R. ridibunda offspring because it is unlikely that different clonal lineages have become fixed for the same mutations. I tested this prediction by comparing survival and larval performance of tadpoles from within- and between-population crossings using R. esculenta from Seseglio (Se) in southern, Alpnach (Al) in central, and Elliker Auen (El) in northern Switzerland, respectively. Se is isolated from the other populations by the Alps. Enzyme electrophoresis revealed that parents from Se belonged to a single hemiclone that was different from all hemiclones found north of the Alps. Parents from Al also belonged to one hemiclone, but parents from El belonged to three hemiclones, one of which was indistinguishable from the one in Al. Rana esculenta from Se produced inviable tadpoles when crossed with other hybrids of their own population, but when crossed with R. esculenta from Al and El, tadpoles successfully completed metamorphosis, supporting the hypothesis I tested. Within-population crosses from Al were also inviable, but some within-population crosses from El, where three hemiclones were present, produced viable offspring. Only part of the crosses between Al and El were viable, but there was no consistent relationship between hemiclone combination and tadpole survival. When backcrossed with the parental species R. ridibunda, hybrids from all source populations produced viable offspring. Performance of these tadpoles with a sexual and a clonal genome was comparable to that of normal, sexually produced R. ridibunda tadpoles. Thus, in the heterozygous state, the deleterious mutations on the clonal R genomes did not appear to reduce tadpole fitness.

Corresponding Editor: J. Merilä

Fluctuating asymmetry (FA), a ubiquitous type of asymmetry of bilateral characters, often has been used as a measure of developmental instability in populations. FA is expected to increase in populations subjected to genetic stressors such as inbreeding or environmental stressors such as toxins or parasites, although results have not always been consistent. We tested whether FA in four skeletal size characters and mandible shape was greater in a population of wild-derived mice reared in the laboratory and subjected to one generation of inbreeding (F = 0.25) versus that in an outbred group (F = 0.00). FA did not significantly differ between the inbred and outbred groups, despite the fact that these two groups differed dramatically in fitness under seminatural population conditions. As far as we know, this is the first study to evaluate the relationship between FA and inbreeding in wild house mice, and our general conclusion is opposite that of earlier work on laboratory inbred strains of mice and their hybrids. Size for two of the characters was significantly less in inbreds than in outbreds, however, and there was a significant difference between inbreds and outbreds in the signed differences of right and left sides in one character (humerus length). Some of the mice in both groups also were heterozygous or homozygous carriers of the t-complex. Because mice carrying this chromosome 17 variant are known to have reduced fitness, we also tested whether they had greater FA than mice carrying non-t-haplotypes. The overall level of a composite FA index calculated from all four characters was in fact significantly higher in the t-bearing mice. These combined results suggest that FA is not a generally sensitive proxy measure for fitness, but can be associated with fitness reductions for certain genetic stressors.

Corresponding Editor: J. Merilä

We combine the methods of geometric morphometrics and multivariate quantitative genetics to study the patterns of phenotypic and genetic variation of mandible shape in random-bred mice. The data are the positions of 11 landmarks on the mandibles of 1241 mice from a parent-offspring breeding design. We use Procrustes superimposition to extract shape variation and restricted maximum likelihood to estimate the additive genetic and environmental components of variance and covariance. Matrix permutation tests showed that the genetic and phenotypic as well as the genetic and environmental covariance matrices were similar, but not identical. Likewise, principal component analyses revealed correspondence in the patterns of phenotypic and genetic variation. Patterns revealed in these analyses also showed similarities to features previously found in the effects of quantitative trait loci and in the phenotypes generated in gene knockout experiments. We used the multivariate version of the breeders' equation to explore the potential for short-term response to selection on shape. In general, the correlated response is substantial and regularly exceeds the direct response: Selection applied locally to one landmark usually produces a response in other parts of the mandible as well. Moreover, even selection for shifts of the same landmark in different directions can yield dramatically different responses. These results demonstrate the role of the geometry and anatomical structure of the mandible, which are key determinants of the patterns of the genetic and phenotypic covariance matrices, in molding the potential for adaptive evolution.

Corresponding Editor: J. Merilä

In natural populations, organisms experience simultaneously biotic (e.g., competitors and parasites) and abiotic (e.g., temperature and humidity) stresses. Thus, species must have the capacity to respond to combinations of stressors. How does interaction between biotic and abiotic stress affect organismal performance? To address this question, I studied stress resistance of adult Drosophila melanogaster that survived parasitic attack (as larvae) by the parasitoid Asobara tabida. To determine the impact of genotype on stress resistance, I measured survival under desiccation and starvation of flies within isofemale (genetic) lines. Survivors of parasitism had slightly reduced survivorship compared to unparasitized relatives when both were unstressed, and this difference was exacerbated by desiccation and starvation. These results indicate multiple stressors can compound each other's individual negative effects on fitness. Moreover, isofemale lines differed in their sensitivity to environmental stress and to parasitism. Consequently, genotypic differences in sensitivity to stress may reflect differences in investment priorities between traits that promote survival over other life-history characters.

Corresponding Editor: S. Pitnick

Data are reported showing large directional changes in the frequencies of some gene arrangements and arrangement combinations in certain natural populations of Drosophila robusta in the eastern United States. The changes involve the same X-chromosomal inversion differences in two of the three localities studied and similar autosomal inversions in all three. These genetic changes provide a rare opportunity to observe evolutionary forces at work in nature. They are interpreted as being due to natural selection.

Corresponding Editor: M. Noor

We used flow cytometry to measure genome size in 15 species from seven families and subfamilies of tetraodontiform fishes. Previous studies have found that smooth pufferfishes (Tetraodontidae) have the smallest genome of any vertebrate measured to date (0.7–1.0 picograms diploid). We found that spiny pufferfishes (Diodontidae, sister group to the smooth puffers) possess a genome that is about two times larger (1.6–1.8 pg). Mola mola, a member of the sister group to Diodontidae and Tetraodontidae, also has a relatively large genome (1.7 pg). Parsimony analysis of this pattern indicates that the plesiomorphic condition for Molidae (Diodontidae, Tetraodontidae) is a genome size of 1.6–1.8 pg, and that tiny genome size is a derived character unique to smooth puffers. However, an alternative explanation is that the ancestor of Tetraodontidae acquired a heritable tendency toward decreasing genome size, such as a new or modified deletion mechanism, and genome size in all of the tetraodontid lineages has been decreasing in parallel since the split from Diodontidae. Small genome size (1.1–1.3 pg) also appears to have evolved independently in some members of Balistoidea (triggerfishes and filefishes) within Tetraodontiformes.

Corresponding Editor: L. Bernatchez

Corresponding Editor: D. Futuyma