Printer Friendly

Analysis of selection and random change in fossil species using reconstructed genetic parameters.

Patterns of phenotypic evolution based on multiple quantitative morphological characters range from gradual to highly punctuated in well-defined, geographically widespread lineages with detailed fossil records (Cheetham 1986; Wei and Kennett 1988; Geary 1991, 1992, 1993; Cheetham and Jackson 1994b). The close correspondence between living, quantitatively discriminated morphospecies and genetic differences validates these patterns as the morphological records of species-level evolution, suitable for evaluating natural selection and random genetic change as the agents of speciation (Michaux 1989; Jackson and Cheetham 1990, 1994; Gould 1991). However, the inference of process from pattern is complex and not always intuitive. On geologic time scales, gradual patterns may be too slow to imply prevalence of directional selection (Geary 1991), and punctuated patterns might result entirely from the interaction of stabilizing selection, mutation, and genetic drift, without necessary input from directional selection (Lynch 1990; Cheetham and Jackson 1994b).

Quantitative genetic methods provide a rigorous basis for retrospective evaluation of the effects of selection and random change (Lande 1977, 1979, 1980, 1987; Turelli et al. 1988; Lynch 1990). These methods are particularly appropriate for investigating phenotypic differentiation of species discriminated on multiple quantitative differences, although they require simplifying assumptions (e.g., selection by truncation; Lande 1979) that are not altogether reasonable as a model of natural selection (Manly 1985). In addition, application of quantitative genetics to analysis of fossil species is greatly limited by the need for estimates of trait heritabilities (the degree to which offspring resemble parents) and genetic covariances (the degree to which offspring-parent resemblances are similar between traits). In answer to the question, Can one predict the evolution of quantitative characters without genetics? Willis et al. (1991) concluded that responses to selection cannot be predicted and past forces of selection cannot be reconstructed without data on many individuals of known ancestry, normally obtained through breeding experiments. Empirical data indicate that phenotypic covariances (calculated without regard to ancestry) and genetic covariances can be similar (Lofsvold 1986; Cheverud 1988), but the relationship is at best one of proportionality, not equality (Willis et al. 1991).

For clonal organisms such as Bryozoa that grow by budding modules (zooids), which together form a colony, phenotypic variance components form a partial record of ancestry from which genetic parameters can be reconstructed. The among-colony component of zooidal morphologic variance thus can be obtained by partitioning the variance of zooids within colonies from total phenotypic variance. This parameter corresponds to the clonal repeatability measure of quantitative genetics, which sets an upper limit to estimates of heritability (Falconer 1981). For two living bryozoan species, genetic parameters based on repeatability were reconstructed with sufficient precision to reproduce estimates of forces of selection and random change based on breeding data (Cheetham et al. 1993). This precision was possible because non-additive genetic variance (attributable to dominance and epistasis) and among-colonies environmental variance are so small in these species, and their genetic and phenotypic covariances are so closely proportional.

In this paper, we reconstruct quantitative genetic parameters for eight Neogene species of the cheilostome bryozoan Metrarabdotos and use these estimates to analyze the possible roles of natural selection and random genetic change in the phenotypic divergence between species and in the phenotypic variation within species. Metrarabdotos is one of the taxa from which the most persuasive evidence for punctuated speciation has come (Gould 1991). Rates of phenotypic divergence across the genus as a whole and of major subsets of its species can be accounted for by prevalence of stabilizing selection, expressed as morphologic stasis within species (Cheetham and Jackson 1994b). Significant phenotypic change is thus concentrated in geologically brief episodes of speciation and could have resulted entirely from random genetic processes without necessary input from directional selection. Here we consider more specifically levels of selective mortality and rates of mutational input to phenotypic variance necessary to explain observed phenotypic differences between pairs of species inferred to represent ancestor and descendant. We use the same rate tests that we employed to explore the differentiation of living species of the cheilostome Stylopoma based on breeding data (Cheetham et al. 1993). In addition, we employ the rate tests to determine whether fluctuating changes in otherwise static phenotypes within species are either too fast or too slow to be explained by random genetic processes.
TABLE 1. Metrarabdotos species from Miocene-Pliocene interval of detailed
sampling in Dominican Republic. Species designations are as in Cheetham (1986,
1987) and Cheetham and Hayek (1988). First and last occurrences are in
millions of years ago (Ma) estimated from stratigraphic thickness between
biostratigraphic markers (Cheetham 1986). Apparent ancestors are
morphologically nearest-neighbor species (square root of Mahalanobis
[D.sup.2]; omitted for species with apparent origins outside interval)
determined by discriminant analysis of 19 species (Jackson and Cheetham

                                                   Apparent        Morphologic
Species               No.     First     Last       ancestor         distance

M. colligatum         19       8.0      5.0
M. auriculatum        18       8.0      1.8      coligatum            19.00
M. new species 9      11       7.3      3.9      auriculatum          16.84
M. new species 10     14       6.5      3.0      new species 9         8.49
M. new species 5       8       7.8      6.0      colligatum           11.18
M. lacrymosum         15       7.8      3.0
M. new species 3       6       7.8      4.0      lacrymosum           36.15
M. new species 4       7       7.1      5.5      new species 3        21.78


Eight fossil species of Metrarabdotos, comprising six pairs of inferred ancestor and descendant species, occur in sufficient abundance for this analysis. All are from an interval of detailed sampling [average spacing 0.26 million years (Ma) for the species considered] spanning approximately 4 Ma of Upper Miocene-Lower Pliocene deposits in the Dominican Republic (fig. 2a; Saunders et al. 1986; Cheetham 1986, 1987). This interval marks the major episode of radiation of Metrarabdotos in the Neogene of the tropical western Atlantic and southeastern North America, based on study of specimens from throughout those regions (Jackson and Cheetham 1994; Cheetham and Jackson 5994a,b). Each species analyzed here is represented at 6 to 19 (mean 12) stratigraphic levels by specimens sufficiently well preserved for morphometric analysis. Additional occurrences of less well preserved specimens were used to calculate confidence limits for first appearances of species in the detailed sampling interval (Marshall 1994; Cheetham and Jackson 1994b).

Phylogenetic relationships between Metrarabdotos species were inferred by morphologic and stratigraphic proximity (stratophenetics; Cheetham and Hayek 1988; Jackson and Cheetham 1994; Cheetham and Jackson 1994b). Although confidence limits on stratigraphic ranges make the order of first occurrences of species of some pairs uncertain, alternative (cladistic) methods of phylogenetic inference suggest relationships that are widely at odds with observed stratigraphic ranges (Jackson and Cheetham 1994; Cheetham and Jackson 1994b). Moreover, an important effect of inferring relationships between species by minimizing morphologic and stratigraphic distances is that tests of rates of morphologic change become significantly more conservative (Cheetham and Jackson 1994b). The observed first occurrences of species in five of the six pairs differ by more than 0.5 Ma. The sixth pair (colligatum, auriculatum) was included in the analysis, even though ages of first occurrence do not differ, because auriculatum initially occurs at much lower abundances than subsequently or than colligature in the same samples (Cheetham 1986).

In each species, 15 traits of skeletal morphology (Appendix) were measured on each of five zooids at the same stage of development in 122 colony fragments. Most of these specimens were collected from different levels or localities and thus clearly represent different colonies. The traits employed (seven measurements, one count, and seven coded characters) are a subset of those used to discriminate species (Cheetham 1986; Cheetham and Hayek 1988). The characters not included here are restricted to particular groups of zooids within colonies and therefore could not be used for partitioning variance and reconstructing genetic parameters. Omission of traits strongly correlated with species differences could bias the quantitative genetic tests on rates of species divergence (Lofsvold 1988). However, these species all differ at the same high level of significance (P [is less than] 0.001) in a discriminant analysis based only on the 15 characters used here as in the analysis based on all 46 characters (Cheetham 1986).

For each trait in each species, heritability was estimated as clonal repeatability, assuming that nonadditive genetic [Mathematical Expression Omitted] and general environmental [Mathematical Expression Omitted] variances are minor, as determined from breeding data in Stylopoma. Repeatabilities were obtained as among-colony components of phenotypic variance partitioned by single-classification ANOVA on the individual zooid data. Degrees of freedom averaged 14 (range 6-23) among colonies and 61 (range 28-96) within colonies.

For consistency with methods used in quantitative genetics, within-colony zooid data were treated as replicate measurements, and colony means were used to calculate phenotypic variance-covariance matrices, P (Cheetham et al. 1993). Linear rescaling was used on the seven metric traits (multiplied by 10) and the single count (divided by 2) to reduce differences in magnitude. Colony means of the ostensibly non-metric traits behave virtually as continuous variates, because of variability among zooids within colonies (Cheetham 1987), and thus were treated as metric (Cheetham et al. 1993).

Genetic variance-covariance matrices, G, were reconstructed for each species from P and the heritability estimates (repeatabilities). An important advantage of basing calculations on P is that it is estimated with greater precision than G for the same size sample (Willis et al. 1991). Reconstruction is based on the assumption that G and P are significantly proportional as shown with breeding data in both species of Stylopoma (Cheetham et al. 1993). As a result of proportionality, the matrix [P.sup.-1]G* (where G* is the genetic variance-covariance matrix "bent" to enhance its use in calculating index selection; Hill and Thompson 1978; Hayes and Hill 1981) is approximately equal to the identity matrix with heritabilities on the diagonal (i.e., with the off-diagonal elements or "coheritabilities" approximately zero). Thus, "bent" genetic variance-covariance matrices, G*, were calculated for each Metrarabdotos species by premultiplying by P a matrix of zeros with repeatabilities on the diagonal (Cheetham et al. 1993).

Tests for selection and random change were all calculated with the reconstructed genetic parameters as described in Cheetham et al. (1993), following procedures in Lofsvold (1986, 1988), Turelli et al. (1988), and Lynch (1990).

Tests on Species Divergence. -- As in the rate tests on living bryozoan species (Cheetham et al. 1993), we first calculated the range of possible divergence times over which observed phenotypic differences can be explained by mutation and genetic drift alone and then calculated the minimum selection intensities required to produce the same differences over these and shorter time scales consistent with the geologically abrupt appearances of these species in the record. In all calculations, time scales are in generations; for conversion to geologic time, we assumed generation times for Metrarabdotos to be on the order of 10 yr (Cheetham and Jackson 1994b).

As for living species, we used two tests of the null hypothesis of sufficiency of random change to explain species divergence, both appropriate for long-term change. Each test compares a calculated rate statistic with typical rates of mutational input to phenotypic variance for polygenicc traits. Rates of mutational input are unknown for bryozoans, but should fall within the range [10.sup.-4] to [10.sup.-2] per generation, characteristic of a wide variety of organisms, provided that environmental variance is not much less than 50% of phenotypic variance (Lynch 1988, 1990). Calculated values of the test statistics exceeding [10.sup.-2] thus imply directional selection, and those less than [10.sup.-4] imply stabilizing selection.

The rate statistic [Delta] of Lynch (1990) is based on partitioning the pooled variance in each log-transformed trait into within- and between-species components. We used Lynch's equations for two-species analysis listed in Cheetham et al. (1993). The statistic [([[Sigma]*.sub.m]/[Sigma]).sup.2] of Turelli et al. (1988) provides a more rigorous test, with attached probability levels (P [is less than] 0.05), but is more sensitive to precision in estimates of mutational input and divergence times. We used the equations of Turelli et al. (1988) for the mutation-drift-equilibrium model, as listed in Cheetham et al. (1993). In both procedures, we calculated t times the rate statistic and the range of t(in generations) over which divergence is consistent with the null hypothesis of random change. We followed Lynch (1990) in averaging rates across traits for each species and also calculated a multivariate rate based on canonical scores from discriminant analysis of each species pair to allow for covariance among traits.

We used the multivariate selection model of Lande (1979) to estimate the intensities of directional selection required to explain divergence on time scales equal to or less than the minimum numbers of generations consistent with the hypothesis of random change. We followed Lofsvold (1988) in calculating selection estimates as an index that weights traits by a gradient of selection coefficients based on the difference between the species mean vectors and the reconstructed "bent" genetic variance-covariance matrix, G*, with equations listed in Cheetham et al. (1993). The intensity of selection required to produce the observed difference in mean vectors is the truncation point b in standard-deviation units of the index, and the corresponding minimum selective mortality per generation is estimated from the cumulative normal distribution function. The value of the truncation model is that it estimates the minimum selection intensity required to produce the change, even if truncation is not a reasonable model of natural selection (Manly 1985).

Tests on Within-Species Fluctuation. -- For each species, the average rate of fluctuation in each trait was calculated as the absolute value of change in mean phenotype between samples divided by the inferred number of generations between samples, averaged for all samples. The maximum fluctuation rate is the largest of the between-sample rates (not necessarily corresponding to the greatest absolute phenotypic change). We tested the hypothesis that these rates of fluctuation are neither too fast nor too slow to be accounted for by random genetic change with the two-tailed version of the statistic [([[Sigma]*.sub.m]/[Sigma]).sup.2] (Turelli et al. 1988, eqs. 9 and 10) at P [is less than] 0.05. As for the tests on species divergence, the hypothesis is rejected for values greater than [10.sup.-2] or less than [10.sup.-4].


Trait Differences, Heritabilities, and Covariances. -- Mean values of each trait for the eight Metrarabdotos species are shown in table 2, and the significance of differences between pairs of species inferred to represent ancestor and descendant in table 3. Each trait is highly significantly different between species of at least one pair, and two traits (LAVS and LAVL) differ significantly (at least P [is less than] 0.05) in all species pairs. On average, species of a pair differ significantly in six traits, with at least two differences at P [is less than] 0.001. Thus, the subset of 15 traits is highly correlated with species differences, and omission of other traits is unlikely to bias the analysis of species divergence severely.

As estimated by clonal repeatability, individual traits in Metrarabdotos have virtually the same range of heritabilities (-0.1277 to 0.8647) as estimated from breeding data in Stylopoma (-0.1549 to 0.8178; Cheetham et al. 1993). However, average repeatability in Metrarabdotos (0.3142) is somewhat smaller than average heritability for Stylopoma (0.3400). Thus, nonadditive genetic and general environmental effects are even less likely to be confounded with heritability than in Stylopoma. On average, 54% of traits show significant heritability in Metrarabdotos species (compared with 65% in Stylopoma), and only one (O2AVL) is not significantly heritable in at least one species.

Tables 5-8 show the phenotypic variance-covariance matrices (P) from which the genetic variance-covariance matrices (G) were reconstructed for the eight species of Metrarabdotos. The assumption of constancy in G between inferred ancestor and descendant in a species pair, required by the model of selection and to a lesser TABULAR DATA OMITTED extent by the models of random change (Lande 1979; Turelli 1988), can be only indirectly tested, using P. However, standard statistical tests of the structure and homogeneity of variance-covariance matrices, not applicable to G, do apply to P, and their application is justified by the significant proportionality between G and P in Stylopoma (Cheetham et al. 1993).

Significant covariance between traits exists for all Metrarabdotos species with the possible exception of new species 3, as shown by Bartlett's sphericity test. Thus, a multivariate approach is clearly called for in the selection tests and possibly also in the tests for random change. TABULAR DATA OMITTED However, Box's M-test suggests that the assumption of homogeneity of variance-covariance matrices between species is violated for all pairs except one (lacrymosum, new species 3). Box's M is well known to be sensitive to departures from multivariate normality (Norusis 1992), and the P matrices of most of the Metrarabdotos species do show some similarity in structure, especially in the first eigenvector, accounting for 43% to 70% of the total phenotypic variance. For three related species (lacrymosum, new species 3, new species 4) similarity in structure includes both first and second eigenvectors, together accounting for more than 80% of the TABULAR DATA OMITTED TABULAR DATA OMITTED TABULAR DATA OMITTED TABULAR DATA OMITTED variance. However, because of the apparently significant heterogeneity of covariances, we calculated comparative values for selection intensities and rates of mutational input by using the estimates for each species of a pair independently (Cheetham et al. 1993).

Random Change versus Selection in Species Divergence. -- Table 11 shows the ranges of divergence times over which observed phenotypic differences among the species of Metrarabdotos can be explained by random change (mutation and drift) alone, based on the "typical" rate of mutational input to phenotypic variance of [10.sup.-4] to [10.sup.-2] per generation. Trait variances for both species of a pair enter calculations of Lynch's (1990) [Delta] yielding a single value, but only one (that of the "ancestor") is used for Turelli et al.'s (1988) [([[Sigma]*.sub.m]/[Sigma]).sup.2], yielding two sets of values for each species pair. Only slight heterogeneity in ranges of divergence times results from using the two sets of values. Six to 10 traits were used for different species pairs in calculating both univariate (means across traits) and multivariate (canonical discriminant scores) values for each rate statistic. We did not consider traits with nonsignificant differences between species of a pair, those with no variance within species, and those with no apparent heritability. Only one trait (O2AVL) was thus eliminated from all species comparisons.

As for Stylopoma (Cheetham et al. 1993), [Delta] and [([[Sigma]*.sub.m]/[Sigma]).sup.2] provide considerably different estimates of maximum divergence times consistent with random change for the Metrarabdotos species, but similar ones for the more critical minimum times. With [Delta], minimum divergence times for random change are also generally consistent between the multivariate (canonical discriminant scores) and univariate (mean across traits) methods of calculation. However, with [([[Sigma]*.sub.m]/[Sigma]).sup.2] many of the multivariate values are higher. The significant trait covariances in almost all species, indicated by Bartlett's test of sphericity, suggest that the multivariate values may be more appropriate. Based on the multivariate values for both rate statistics, minimum divergence times consistent with divergence by random change are [10.sup.3] to [10.sup.4] generations for four pairs of species (colligatum, auriculatum; new species 9, new species 10; colligatum, new species 5; new species 3, new species 4) and [10.sup.4] to [10.sup.5] generations for the other two (auriculatum, new species 9; lacrymosum, new species 3). On the assumption that generation TABULAR DATA OMITTED TABULAR DATA OMITTED times are about 10 yr for Metrarabdotos (Cheetham and Jackson 1993b), divergence times of [10.sup.4] generations or fewer are consistent with the abrupt appearances of species in samples spaced an average of 0.26 Ma apart.

Reconstructed forces of directional selection required for divergence of Metrarabdotos species pairs are shown in table 12 for three time scales equal to or shorter than the minimum numbers of generations over which phenotypic change could have been produced by random change alone ([10.sup.2] to [10.sup.4] generations). Calculated minimum per-generation selective mortalities are all greater than 6% for two of the four species pairs in which minimum divergence times consistent with random change are in the [10.sup.3] to [10.sup.4] range (colligatum, auriculatum, at [10.sup.3]; colligatum, new species 5, at [10.sup.3] to [10.sup.4]). These levels of selection are more than twice the 2.6% selective mortality measured in single episodes of severe intensity (Boag and Grant 1981; Price et al. 1984; Lofsvold 1988, and references therein). The results for these species of Metrarabdotos are consistent irrespective of which species' heritabilities and covariances are used in the calculations and clearly make directional selection unlikely as the sole basis for their phenotypic divergence. For the other two pairs of species in this group (new species 9, new species 10; new species 3, new species 4; both at [10.sup.3] to [10.sup.4] generations), minimum selective mortality falls below 1% for [10.sup.4] generations, and may also be under the most severe level of intensity for [10.sup.3] generations. However, results substituting heritabilities and covariances between species are conflicting.

For species pairs in which random change could account for divergence over a minimum of [10.sup.4] to [10.sup.5] generations, reconstruction of selection intensities gives clear results for one and conflicting results for the other. For the pair consisting of lacrymosum and new species 3, the rate of selective mortality required to produce the observed phenotypic change is less than 1% per generation for [10.sup.4] generations and at or below the 2.6% level for [10.sup.3] generations, using both sets of values in calculations. The greater feasibility of directional selection in this case may be related to the fact that the morphologic difference between these two species is almost twice that between any of the other species used in the analysis. The last species pair (auriculatum, new species 9) yielded the most conflicting and inconclusive results of all.

Random Change versus Selection within Species. -- Table 13 summarizes the two-tailed tests for random change averaged over 9 to 12 eligible traits within each Metrarabdotos species. Traits with no variation or no heritable variation were eliminated. Mean values across traits for both average and maximum fluctuation are all within the limits of acceptance of the null hypothesis of random change. The only individual trait not within the limits (ND in auriculatum with an upper limit of 5.0 x [10.sup.-5] for average fluctuation) is a trait that only rarely shows any variation and has a very low heritability. Thus, although stabilizing selection is required to explain the net rate of (virtually zero) change within species, the fluctuations are sufficiently large to be accounted for by mutation and drift and not so large as to require directional selection.


The dominant process underlying the overall pattern of phenotypic evolution in Metrarabdotos is almost certainly stabilizing selection, expressed in net morphologic stasis within species with durations of 1.6 to 6.2 Ma (Cheetham 1986; Cheetham and Jackson 1994b). This conclusion is not refuted by the results obtained here by focusing on the differentiation of individual pairs of species and on the details of phenotypic change within species. Although these results show that selection is unnecessary to explain the pace of temporal fluctuation of mean phenotypes within species, confinement of fluctuation to relatively narrow tracks of no net change is consistent with the action of stabilizing selection to "correct" random excursions from a phenotypic optimum. Such excursions are likely because of the flatness of the fitness function in the vicinity of the optimum (Lynch 1990). But, provided that they are not accompanied by attainment of reproductive isolation (Futuyma 1987), increasing steepness of the selection gradient away from the optimum makes reversal highly probable.

The corollary to stasis in Metrarabdotos species is concentration of phenotypic evolution in brief episodes of cladogenesis at or within the limits of stratigraphic resolution (Cheetham and Jackson 1994b). This is consistent with the basic tenet of punctuated equilibria theory associating evolutionary change with speciation (Eldredge and Gould 1972; Stanley 1975; Gould and Eldredge 1977). The hypothesis that mutation and genetic drift could account for the rate of species divergence in general (Cheetham and Jackson 1994b) is not rejected by the results obtained for differentiation of individual species pairs on time scales at or just within the average stratigraphic resolution of 0.26 Ma. However, contrary to the results based on the overall divergence pattern, feasibility of random change as the sole mechanism of differentiation for individual species pairs does not extend to divergence times shorter than [10.sup.3] to [10.sup.4] generations.

Despite the adequacy of random processes to account for abrupt divergence of all the Metrarabdotos species, directional selection remains a plausible alternative to explain phenotypic differentiation of at least some species. Minimum levels of selective mortality on time scales within the range of stratigraphic resolution are most reasonable for the pair of species (lacrymosum, new species 3) separated by the greatest morphologic distance. However, generalizing this result to all species is contradicted by the two species pairs (comprising three species; colligatum, auriculatum, new species 5) in which required levels of selective mortality are sufficiently high to make TABULAR DATA OMITTED TABULAR DATA OMITTED directional selection untenable as the sole mechanism of divergence. Moreover, the method of calculating minimum selective mortalities biases in favor of finding directional selection (Lofsvold 1988). Thus, directional selection may also be untenable as the sole basis for divergence in the three species pairs for which conflicting results were obtained by substituting estimates of heritabilities and trait covariances between species. However, phenotypic differentiation of all species pairs might be explained by reasonable levels TABULAR DATA OMITTED of directional selection acting in concert with founder effects or other random processes on even the most abbreviated time scales ([much less than] [10.sup.3] generations).

In general, these results support genetic models of speciation involving relatively sudden shifts between multiple adaptive peaks on which phenotypes remain more or less static through long-term stabilizing selection (Wright 1982; Lande 1985, 1987; Lynch 1990). Models based on shifting balance among local demes (Wright 1982) TABULAR DATA OMITTED and isolation on the peripheries of large distributions (Mayr 1954, 1963) are both plausible for Metrarabdotos and other cheilostomes, based on their Neogene to Holocene distribution patterns in tropical America (Cheetham and Jackson 1994a). Random processes may be sufficient to explain phenotypic differentiation at speciation in these models, or directional election may be required. In either case, the agents of speciation are different from the pervasive stabilizing selection required to explain phenotypic stasis within species. In this sense, the results presented here are consistent with the "stronger" version of punctuated equilibria theory decoupling speciation from forces acting within species (Maynard Smith 1988).


We thank J. Sanner for morphologic measurements and other help; Y. Ventocilla for processing samples; and P. Jung, J. Saunders, E. Vokes, and L. Collins for specimens and stratigraphic documentation. This work was supported by the Smithsonian Scholarly Studies Program, the Marie Bohrn Abbott Fund of the National Museum of Natural History, Smithsonian Tropical Research Institute, and grants to the Panama Paleontology Project from the National Science Foundation and National Geographic Society.


Boag, P. T., and P. R. Grant. 1981. Intense natural selection in a population of Darwin's ground finches (Geospizinae) in the Galapagos. Science 214:82-85.

Cheetham, A. H. 1986. Tempo of evolution in a Neogene bryozoan: rates of morphologic change within and across species boundaries. Paleobiology 12:190-202.

-----. 1987. Tempo of evolution in a Neogene bryozoan: Are trends in single morphologic characters misleading? Paleobiology 13:286-296.

Cheetham, A. H., and L. C. Hayek. 1988. Phylogeny reconstruction in the Neogene bryozoan Metrarabdotos: a paleontological evaluation of methodology. Historical Biology 1:65-83.

Cheetham, A. H., and J.B.C. Jackson. 1994a. Speciation, extinction, and the decline of arborescent growth in Neogene and Quaternary cheilostome Bryozoa of tropical America. In J.B.C. Jackson et al., eds. Evolution and environment in tropical America. University of Chicago Press, Chicago (in press).

-----. 1994b. Process from pattern: tests for selection versus random change in punctuated bryozoan speciation. In D. H. Erwin and R. L. Anstey, eds. New approaches to speciation in the fossil record. Columbia University Press, New York (in press).

Cheetham, A. H., J.B.C. Jackson, and L. C. Hayek. 1993. Quantitative genetics of bryozoan phenotypic evolution. I. Rate tests for random change versus selection in differentiation of living species. Evolution 47:1526-1538.

Cheverud, J. M. 1988. A comparison of genetic and phenotypic correlations. Evolution 42:958-968.

Eldredge, N., and S. J. Gould. 1972. Punctuated equilibria: an alternative to phyletic gradualism. Pp. 82-115 in T.J.M. Schopf, ed. Models in paleobiology. Freeman, Cooper, San Francisco.

Falconer, D.S. 1981. Introduction to quantitative genetics, 2d ed. Longman, New York.

Futuyma, D.J. 1987. On the role of species in anagenesis. American Naturalist 130:465-473.

Geary, D. H. 1991. Patterns of evolutionary tempo and mode in the radiation of Melanopsis (Gastropoda; Melanopsidae). Paleobiology 16:492-511.

-----. 1992. An unusual pattern of divergence between two fossil gastropods: ecophenotypy, dimorphism, or hybridization? Paleobiology 18:93-109.

-----. 1994. Tempo and mode in fossil molluscs: investigating organism-environment interactions, species and speciation. In D. H. Erwin and R. L. Anstey, eds. New approaches to speciation in the fossil record. Columbia University Press, New York (in press).

Gould, S.J. 1991. Opus 200. Natural history 8/91: 12-18.

Gould, S. J., and N. Eldredge. 1977. Punctuated equilibria: the tempo and mode of evolution reconsidered. Paleobiology 3:115-151.

Hayes, J. F., and W. G. Hill. 1981. Modification of estimates of parameters in the construction of genetic selection indices ("bending"). Biometrics 37: 483-493.

Hill, W. G., and R. Thompson. 1978. Probabilities of nonpositive definite between-group or genetic covariance matrices. Biometrics 34:429-439.

Jackson, J.B.C.,$and A. H. Cheetham. 1990. Evolutionary significance of morphospecies: a test with cheilostome Bryozoa. Science 248:579-583.

-----. 1994. Phylogeny reconstruction and the tempo of speciation in cheilostome Bryozoa. Paleobiology. In press.

Lande, R. 1977. Statistical tests for natural selection on quantitative characters. Evolution 31:442-444.

-----. 1979. Quantitative genetic analysis of multivariate evolution, applied to brain: body size allometry. Evolution 33:402-416.

-----. 1980, Genetic variation and phenotypic evolution during allopatric speciation. American Naturalist 116:463-479.

-----. 1985. Expected time for random genetic drift of a population between stable phenotypic states. Proceedings of the National Academy of Sciences, USA 82:7641-7645.

-----. 1987. The dynamics of peak shifts and the pattern of morphological evolution. Paleobiology 12:343-354.

Lofsvold, D. 1986. Quantitative genetics of morphological differentiation in Peromyscus. I. Tests of the homogeneity of genetic covariance structure among species and subspecies. Evolution 40:559-573.

-----. 1988. Quantitative genetics of morphological differentiation in Peromyscus. II. Analysis of selection and drift. Evolution 42:54-67.

Lynch, M. 1988. The rate of polygenic mutation. Genetical Research Cambridge 51:137-148.

-----. 1990. The rate of morphological evolution in mammals from the standpoint of the neutral expectation. American Naturalist 136:727-741.

Manly, B.F.J. 1985. The statistics of natural selection on animal populations. Chapman and Hall, London.

Marshall, C.R. 1994. Stratigraphy, the true order of species' originations and extinctions, and testing ancestor-descendant hypotheses among Caribbean Neogene bryozoans. In D. H. Erwin and R. L. Anstey, eds. New approaches to speciation in the fossil record. Columbia University Press, New York (in press).

Maynard Smith, J. 1988. Games, sex and evolution. Harvester-Wheatsheaf, Hemel-Hempstead, UK.

Mayr, E. 1954. Geographic speciation in tropical echinoids. Evolution 8:1-18.

-----. 1963. Animal species and evolution. Harvard University Press, Cambridge, Mass.

Michaux, B. 1989. Morphological variation of species through time. Biological Journal of the Linnean Society 38:239-255.

Norusis, M. J. 1992. SPSS for windows, advanced statistics, release 5. SPSS, Chicago.

Price, T. D., P. R. Grant, H. L. Gibbs, and P. T. Boag. 1984. Recurrent patterns of natural selection in a population of Darwin's finches. Nature 309:787-789.

Saunders, J. B., P. Jung, and B. Biju-Duval. 1986. Neogene paleontology in the northern Dominican Republic, 1. Field surveys, lithology, and age. Bulletin of American Paleontology 89(323):1-79.

Stanley, S.M. 1975. A theory of evolution above the species level. Proceedings of the National Academy of Sciences, USA 72:646-650.

Turelli, M. 1988. Phenotypic evolution, constant covariances, and the maintenance of additive variance. Evolution 42:1342-1347.

Turelli, M., J. H. Gillespie, and R. Lande. 1988. Rate tests for selection on quantitative characters during macroevolution and microevolution. Evolution 42: 1085-1089.

Wei, K., and J. P. Kennett. 1988. Phyletic gradualism and punctuated equilibrium in the late Neogene planktonic foraminiferal clade Globoconella. Paleobiology 14:345-363.

Willis, J. H., J. A. Coyne, and M. Kirkpatrick. 1991. Can one predict the evolution of quantitative characters without genetics? Evolution 45:441-444.

Wright, S. 1982. Character change, speciation, and the higher taxa. Evolution 36:427-443.


Abbreviations and Definitions of Traits

Traits are illustrated in Cheetham (1986, fig. 6) and are listed together with rank order in discriminant analysis in Cheetham and Hayek (1988, table 2). Metric traits (mm) were all multiplied by 10 and number of areolae was divided by two to reduce differences in magnitude.

LZ: Length of ordinary feeding zo-oid (mm).

WZ: Width of ordinary feeding zo-oid (mm)

LO: Length of orifice (mm).

WO: Width of orifice (mm).

LD: Distance between lateral denticles (mm).

LAVS: Length of shorter avicularium (mm).

LAVL: Length of longer avicularium (mm).

PAVS, PAVL: Position of shorter (longer) avicularium: 1, proximal to orifice; 2, lateral; 3, distal.

O1AVS, O1AVL: Distal orientation of shorter (longer) avicularium: 1, directed distally; 2, transversely; 3, proximally.

O2VS, O2AVL: Lateral orientation of shorter (longer) avicularium: 1, inward; 2, longitudinally; 3, outward.

NA: Number of areolae.

ND: Number of oral denticles: 1, single proximal only; 2, proximal and paired lateral; 3, lateral only.
COPYRIGHT 1994 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1994 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Quantitative Genetics of Bryozoan Phenotypic Evolution, part 2
Author:Cheetham, Alan H.; Jackson, Jeremy B.C.; Hayek, Lee-Ann
Date:Apr 1, 1994
Previous Article:A population memetics approach to cultural evolution in chaffinch song: differentiation among populations.
Next Article:Evidence for a mosaic hybrid zone in the grass shrimp Palaemonetes kadiakensis (Palaemonidae) as revealed by multiple genetic markers.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |