Blackwell Science, LtdOxford, UKBIJBiological Journal of the Linnean Society0024-4066The Linnean Society of London, 2003? 2003 80•• 99106 Original Article SPECIATIONAL CHANGE IN RATITE EVOLUTION J. CUBO Biological Journal of the Linnean Society, 2003, 80, 99–106. With 3 figures Evidence for speciational change in the evolution of ratites (Aves: Palaeognathae) JORGE CUBO* UMR CNRS 8570, Université Paris 6/7-2, Pl Jussieu, Case 7077, 75005 Paris, France Received 20 September 2002; accepted for publication 11 February 2003 To perform a comparative analysis of character associations framed in a phylogenetic context (e.g. independent contrasts), a model of character evolution must be assumed. According to phyletic gradualism, morphological change accumulates gradually over time within lineages, and speciation events do not have a major role. Under speciational models, morphological change is assumed to occur during or just after cladogenesis in both daughter species, and the resulting morphologies do not change over long periods of time (stasis), until the next cladogenetic event. A novel method is presented for comparing these models of character evolution that uses permutational multiple phylogenetic regressions. The addition of divergence times to well-corroborated phylogenetic trees and the utilization of the method developed in this paper allows the estimation of relative frequency of gradual change and speciational change from living organisms. This method is applied to a dataset from ratites with the conclusion that, for a range of morphological features, change tends to have been speciational rather than gradual. © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106. ADDITIONAL KEYWORDS: heterochrony – independent contrasts – phyletic gradualism – phylogenetic comparative method – punctuational change. INTRODUCTION To perform a comparative analysis of character associations framed in a phylogenetic context (e.g. independent contrasts), a model of character evolution must be assumed (Harvey & Pagel, 1991; Harvey & Purvis, 1991; Garland, Harvey & Ives, 1992; Pagel, 1999). In fact, 12 years ago, Harvey & Purvis (1991) and Martins & Garland (1991) showed that the result of a comparative analysis strongly depends on the model of evolution that has been assumed. However, most comparative studies do not determine an appropriate model of evolution for character change prior to choosing the comparative method to be used. The approach presented here allows a comparison of the relative efficacy of phyletic gradualism vs. speciational models prior to choosing the comparative method. Under phyletic gradualism, geological time is an appropriate predictor of the amount of morphological change that has occurred, and speciation does not *E-mail: [email protected] affect the rate of change (Gould & Eldredge, 1993; Mooers, Vamosi & Schluter, 1999). If we assume the gradual model of evolution, our comparative method should consider branch lengths between speciation events in terms of either geological time or genetic change (assuming a clock model). According to the speciational model, changes occur at the time of speciation in both daughter species and the resulting morphologies do not change over long periods of time (stasis), until the next cladogenetic event (Rohlf et al., 1990). (Note that under the punctuational model, changes occur at the time of speciation but only in a single daughter species, Rohlf et al., 1990.) Under the speciational model, morphological change between specified times is proportional to the number of speciation events that have occurred (Mooers et al., 1999). If the speciational model of character evolution is assumed, our comparative method should assume equal branch lengths between speciation events. Here I present a novel method for comparing these models of character evolution that uses permutational multiple phylogenetic regressions (Legendre, Lapointe & Casgrain, 1994; Böhning-Gaese & Oberrath, 1999). I © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 99 100 J. CUBO apply this method to a dataset from ratites and conclude that, for a range of morphological features, change tends to have been speciational rather than gradual. MATERIAL I measured total length and diaphyseal diameter of stylopodial bones (humerus and femur) and zeugopodial bones (ulna, radius and tibiotarsus) to the nearest 0.01 mm using a digital caliper in a sample of nine species of ratites. The number of specimens for each species is given in parentheses: Rhea americana (Linnaeus, 1758) (3), Rhea pennata (d’Orbigny, 1834) (3), Struthio camelus Linnaeus, 1758 (3), Dromaius novaehollandiae (Latham, 1790) (2), Casuarius casuarius (Linnaeus, 1758) (2), Casuarius bennetti Gould, 1857 (1), Casuarius unappendiculatus Blyth, 1860 (2), Apteryx australis Shaw, 1813 (3) and Apteryx owenii Gould, 1847 (1). Following Cubo et al. (2002), dimensionless shape variables were computed to be compared between species: for each bone, the ratio diaphyseal diameter/total length was calculated. In addition, the ratio wing length/leg ratio was also calculated (limb length was measured as stylopodial length + zeugopodial length). Mean values of these ratios for each species were used, assuming no sampling error because of small sample sizes. DEVELOPMENT OF THE METHOD I used multiple regression and the Mantel permutation test to compare the gradual and the speciational models of character evolution for ratite birds. Distance matrices of size 9 ¥ 9 were constructed (nine species, 36 comparisons). The dependent matrices (Y 1, Y2,…Y6) contain the morphological dissimilarity (regarding the different ratios described above) for each pair of species. The independent matrices (X i) contain the phylogenetic distances for each pair of species. These distances were quantified under assumptions of the gradual model (XGr) and the speciational model (XSpeciat) of evolution. In all cases, I used the topology of the phylogenetic tree of extant ratites published by Cooper et al. (2001) and by Haddrath & Baker (2001) (Figs 1A,2A,3A). Under the gradual model of evolution, phylogenetic distances were measured as divergence times. For each comparison, the distance between the two species being compared was computed as the geological time since their last common ancestor. Both molecular clock data (van Tuinen & Hedges, 2001) and the fossil record (Cracraft, 2001) agree that the split between Palaeognathae (ratites and tinamous) and Neognathae (all other modern birds) occurred prior to the Cretaceous–Tertiary extinction event. Within ratites, estimations of divergence times between the different clades found by two recent molecular studies are slightly different (Cooper et al., 2001; Haddrath & Baker, 2001). I used divergence times found by Cooper et al. (2001) (Fig. 1A) to construct matrix X Gr-1 and divergence times found by Haddrath & Baker (2001) (Fig. 2A) to construct matrix X Gr-2. In general, no divergence times are available for within-genus comparisons (Casuarius, Apteryx, Rhea). In these cases, the first occurrence in the fossil record of each one of these genera (the geological age of the oldest fossil of each genus, Unwin, 1993) was tentatively used as the date of divergence between the different species of each genus (Figs 1A,2A). The following method to quantify the phylogenetic distances under the speciational model of evolution (matrix XSpeciat) was used. For each comparison, the phylogenetic distance was measured as the number of speciation events that separates each pair of extant species being compared since their last common ancestor (Fig. 3a). In the cases of unresolved trichotomies (three species of Casuarius and three species of Apteryx), according to Lemen & Freeman (1989), the number of nodes between a species and the last common ancestor for that trichotomy was calculated as the average of all possible arrangements; that is, 1.67. Ideally, under the speciational model of evolution, all speciation events of ratite evolution should be incorporated, including those leading to extinct species. However, the total number of extinct species of a clade cannot be quantified. In order to minimize this problem, I used a ‘complete tree’ (Mooers, 1995) that contains at least N - 1 of the total known extant species of the in-group (in the case of ratites, N = 10), a necessary requirement for a valid test of the speciational model (Mooers et al., 1999). The goal of the present analysis was to determine in ratites which model of character change (gradual or speciational) best describes the evolution of the morphological features in question. For this, I used permutational multiple phylogenetic regressions (Legendre et al., 1994; Böhning-Gaese & Oberrath, 1999). I computed multiple regression between the dependent morphological-dissimilarity matrices (Yi) and the independent phylogenetic-distance matrices XGr-1 and XSpeciat. Similarly, the dependent morphological-dissimilarity matrices (Y i) were regressed on the independent phylogenetic-distance matrices XGr-2 and XSpeciat. Figures 1B, 2B and 3B show scatter plots of the dependent matrix containing the morphological dissimilarity in humerus shape for each pair of species vs. the independent matrices containing phylogenetic distances for each pair of species calculated under assumptions of the gradual and the speciational models of evolution. I was interested in considering in multiple regression only the © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 SPECIATIONAL CHANGE IN RATITE EVOLUTION 101 A B Figure 1. (A) Phylogenetic relationships among extant ratites (modified from Cooper et al., 2001). Branch lengths have been scaled proportional to divergence times (taken from Cooper et al., 2001; with the exception of within-genus divergence times that were estimated as the first occurrence in the fossil record of each one of these genera, Unwin, 1993). Under the gradual model of evolution, for each comparison, the phylogenetic distance between the two species being compared was computed as the geological time since their last common ancestor. (B) Scatter-plot of the dependent matrix containing the morphological dissimilarity in humerus shape for each pair of species vs. the independent matrix containing phylogenetic distances for each pair of species calculated under assumptions of the gradual model of evolution by using divergence dates of Fig. 1A. independent-matrix variables (corresponding to the gradual and the speciational models of evolution) that contribute significantly to the explanation of each dependent-matrix variable (the morphological-dissimilarity matrices). For this, I used a forward-selection procedure (Legendre et al., 1994). At each step, the independent-matrix variable whose multiple regression equation provides the most significant R2 coefficient is selected, provided that both this probability and the probability of the corresponding partial standardized regression coefficient P(b) are smaller than or equal to the predetermined Bonferroni-corrected P-to- enter value (P = 0.05). The significance of statistics of the multiple regression equation ( R2 and partial standardized regression coefficients, bi) could not be tested in the parametric way because the values of the matrix variables corresponding to the morphological differences (simple distance matrices) are not independent from each other and the independence of the observations is a fundamental condition of parametric testing (Legendre et al., 1994). In these cases, significance of model statistics should be tested through permutational tests (Harvey & Pagel, 1991). I used the Mantel permutation test: the model and its statistics were © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 102 J. CUBO A B Figure 2. (A) Phylogenetic relationships among extant ratites (modified from Haddrath & Baker, 2001). Branch lengths have been scaled proportional to divergence times (taken from Haddrath & Baker, 2001; with the exception of some withingenus divergence times, Casuarius, Apteryx, that were estimated as the first occurrence in the fossil record of each one of these genera, Unwin, 1993). Under the gradual model of evolution, for each comparison, the phylogenetic distance between the two species being compared was computed as the geological time since their last common ancestor. (B) Scatter-plot of the dependent matrix containing the morphological dissimilarity in humerus shape for each pair of species vs. the independent matrix containing phylogenetic distances for each pair of species calculated under assumptions of the gradual model of evolution by using divergence dates of Fig. 2A. recomputed by repeatedly randomizing the values of the matrices corresponding to the morphological differences to obtain null distributions against which to test the significance of the statistics of the actual regression (Legendre et al., 1994). The matrix variables containing the morphological differences were randomly permuted 999 times, the independentmatrix variables with the phylogenetic distances (corresponding to the gradual and speciational models of evolution) were held constant, and the statistics of the regression model were repeatedly computed. The significance of the actual statistics was tested by compar- ing them with the distribution of values obtained from permutations (Legendre et al., 1994). RESULTS Table 1 shows results of the multiple regression through a forward-selection procedure between the dependent morphological-dissimilarity matrices (Y i) and (i) the independent phylogenetic-distance matrices XGr-1 and XSpeciat, and (ii) the independent phylogenetic-distance matrices X Gr-2 and XSpeciat. Results obtained by using divergence times taken © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 SPECIATIONAL CHANGE IN RATITE EVOLUTION 103 A B Figure 3. (A) Phylogenetic relationships among extant ratites (modified from Cooper et al., 2001 and Haddrath & Baker, 2001). Under the speciational model of evolution, for each comparison, the phylogenetic distance was measured as the number of speciation events that separates each pair of extant species being compared since their last common ancestor. In the cases of unresolved trichotomies (Casuarius, Apteryx), according to Lemen & Freeman (1989), the number of nodes between a species and the last common ancestor for that trichotomy was calculated as the average of all possible arrangements, that is, 1.67. (B) Scatter-plot of the dependent matrix containing the morphological dissimilarity in humerus shape for each pair of species vs. the independent matrix containing phylogenetic distances for each pair of species calculated under assumptions of the speciational model of evolution represented in Fig. 3A. from Cooper et al. (2001) (XGr-1, Fig. 1A) are very similar to those obtained by using divergence times taken from Haddrath & Baker (2001) (XGr-2, Fig. 2A). Regarding wing bones, the phylogenetic-distance matrix constructed assuming a speciational model of evolution (XSpeciat) was the only independent variable that significantly explained morphological differences in humerus shape, ulna shape, radius shape and the ratio wing length/leg length. In most cases (Table 1), XSpeciat (speciational model) was the only selected independent variable at the predetermined Bonferronicorrected P-to-enter value P = 0.05. In a few cases (see Table 1), both the speciational and the gradual model were selected at P = 0.05. In these cases, I repeated the multiple regression through a forward-selection procedure by using a predetermined Bonferroni-corrected P-to-enter value P = 0.01. Again, XSpeciat (speciational model) was the only independent variable that significantly explained morphological differences. Regarding leg bones, neither the gradual model nor the speciational model explain morphological differences in femur shape and tibiotarsus shape. DISCUSSION Different approaches have been used to test for gradual vs. speciational models of character evolution from living organisms. Avise (1977) performed a test by © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 104 J. CUBO Table 1. Selection through multiple regression of an optimal subset of independent phylogenetic-distance matrices (X i, constructed assuming the gradual and the speciational models of evolution) to explain the dependent morphologicaldissimilarity matrices (Yi) in ratites. Abbreviations: b, standard partial regression coefficients; DR2, increment in multiple regression coefficient produced when a new variable is added to the regression model; XGr-1, gradual model, divergence times were taken from Cooper et al. (2001); XGr-2, gradual model, divergence times were taken from Haddrath & Baker (2001); XSpeciat, speciational model; Y1 = wing length/leg length; Y2 = humerus shape; Y3 = ulna shape; Y4 = radius shape; Y5 = femur shape; Y6 = tibiotarsus shape; *, independent variable selected to be added to the model; #, predetermined Bonferroni-corrected P-to-enter value = 0.01 (P = 0.05 in all other cases). Note that P-to-enter value in step 2 is half of P-to-enter value in step 1 Step 1 bi Step 2 P(b) DR2 P(R2) bi P(b) DR2 P(R2) Y1 XGr-1 XSpeciat* XGr-2 XSpeciat* 0.747 0.738 0.640 0.738 0.002 0.001* 0.002 0.001* 0.558 0.545 0.409 0.545 0.002 0.001* 0.002 0.001* 0.433 0.047 0.063 0.001 0.142 0.302 0.007 0.001 Y2 #XGr-1 #XSpeciat* XGr-2 XSpeciat* 0.496 0.754 0.531 0.754 0.015 0.001* 0.003 0.001* 0.264 0.569 0.282 0.569 0.015 0.001* 0.003 0.001* -0.359 0.013 0.043 0.001 -0.191 0.126 0.013 0.001 Y3 XGr-1 XSpeciat* XGr-2 XSpeciat* 0.347 0.635 0.328 0.635 0.043 0.002* 0.048 0.002* 0.120 0.403 0.108 0.403 0.043 0.002* 0.048 0.002* -0.515 0.052 0.088 0.001 -0.486 0.041 0.086 0.001 Y4 XGr-1 XSpeciat* #XGr-2 #XSpeciat* 0.376 0.659 0.349 0.659 0.039 0.001* 0.043 0.001* 0.142 0.434 0.122 0.434 0.039 0.001* 0.043 0.001* -0.484 0.040 0.078 0.001 -0.480 0.022 0.084 0.001 Y5 XGr-1 XSpeciat #XGr-2 #XSpeciat 0.307 -0.062 0.371 -0.062 0.046 0.385 0.016 0.385 0.094 0.004 0.137 0.004 0.066 0.711 0.016 0.711 Y6 XGr-1 XSpeciat XGr-2 XSpeciat 0.063 0.016 0.154 0.016 0.339 0.413 0.194 0.413 0.004 0.0003 0.024 0.0003 0.723 0.932 0.397 0.932 comparing genetic differences in a speciose and a depauperate group of fishes, assuming these groups have similar geological ages. Mayden (1986) has criticized this method on the basis that the assumption of equal antiquity was not justified and this author suggested the use of cladistic methods to overcome this problem (as sister clades have, necessarily, the same geological age). Mindell, Sites & Graur (1989) and Lemen & Freeman (1989) have used cladistic methods which are based on character polarization and the comparison of primitive and derived character states. This approach is useful to analyse discrete characters, where ancestral states are estimated through outgroup comparison by using parsimony. For continuous characters, ancestral states are estimated in part or whole from daughter taxa by assuming a © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 SPECIATIONAL CHANGE IN RATITE EVOLUTION set of assumptions (Brownian motion, maximum parsimony, . . . ). In this last case, ancestral states are not independent from character states of daughter taxa and comparisons between ancestors and descendants are not valid (Harvey & Purvis, 1991). The characters analysed in this paper, bone shape and the ratio wing length/leg length, vary continuously across species. Therefore, the method developed here avoids the estimation of ancestral states to test the gradual and the speciational models of character evolution. Other methods have been proposed to test these models that also avoid the estimation of ancestral states for continuous traits (Pagel, 1997, 1999; Mooers et al., 1999). While these last methods perform estimation of evolutionary parameters by using a maximum likelihood approach (Pagel, 1997, 1999; Mooers et al., 1999), the method developed in this paper (based on permutational multiple phylogenetic regressions) is not constrained to the estimation of such parameters. Table 1 shows that, for a range of morphological features of ratites, change tends to have been speciational rather than gradual. The speciational model significantly explains morphological variation of humerus shape, ulna shape, radius shape, as well as the variation of the ratio wing length/leg length. Morphological change is assumed to occur during or just after cladogenesis in both daughter species, and the resulting morphologies do not change over long periods of time (stasis), until the next cladogenetic event (Rohlf et al., 1990). Consequently, comparative analyses of character associations framed in a phylogenetic context (e.g. independent contrasts) in ratites should assume equal branch lengths between speciation events. Regarding leg bones (femur and tibiotarsus), the rate of character change would have been rapid enough to erase phylogenetic effects. In fact, the speciational and the gradual models of change are not mutually exclusive. As quoted above, in a few cases (see Table 1), both the speciational and the gradual models were selected at a predetermined Bonferronicorrected P-to-enter value P = 0.05, although the speciational model was the only selected variable at a predetermined Bonferroni-corrected P-to-enter value P = 0.01. We can wonder about the proximal factors (Cubo et al., 2000; Cubo, 2000) underlying the similar patterns of character evolution in the different wing bones of ratites. Ratite wing bones are underdeveloped and this has been interpreted by Cubo & Arthur (2000) as a case of paedomorphosis (the retention of ancestral juvenile character states in adult stages of descendants, Gould, 1977). Considering that the development of forelimbs is delayed relative to the development of hindlimbs in birds (Carrier & Leon, 1990), either the truncation or the retardation of somatic development (heterochrony) is likely to pro- 105 duce reduction of the size of all wing bones by correlated development (Cubo & Arthur, 2000). Heterochronic changes can be instantaneous in terms of geological time (Gould, 1977) and they are likely to produce patterns of speciational character evolution. Ratite wing bones do not play a function in locomotion, they do not undergo adaptation linked to flight, and this would have contributed to preserve the pattern of speciational character evolution generated by heterochronic changes. ACKNOWLEDGEMENTS I thank Paul Harvey (University of Oxford) and Jacques Castanet and Emmanuel de Margerie (Pierre & Marie Curie University, Paris) for a critical reading of a preliminary version of this manuscript. I also thank Philippe Casgrain (University of Montreal) for useful comments on statistical methods and two anonymous reviewers for comments on the submitted version of the manuscript. REFERENCES Avise JC. 1977. Is evolution gradual or rectangular? Evidence from living fishes. Proceedings of the National Academy of Sciences, USA 74: 5083–5087. Böhning-Gaese K, Oberrath R. 1999. Phylogenetic effects on morphological, life-history, behavioural and ecological traits of birds. Evolutionary Ecology Research 1: 347–364. Carrier D, Leon LR. 1990. Skeletal growth and function in the California gull (Larus californicus). Journal of Zoology, London 222: 375–389. Cooper A, Lalueza-Fox C, Anderson S, Rambaut A, Austin J, Ward R. 2001. Complete mitochondrial genome sequences of two extinct moas clarify ratite evolution. Nature 409: 704–707. Cracraft J. 2001. Avian evolution, Gondwana biogeography and the Cretaceous-Tertiary mass extinction event. Proceedings of the Royal Society of London, B 268: 459–469. Cubo J. 2000. Process heterochronies in endochondral ossification. Journal of Theoretical Biology 205: 343–353. Cubo J, Arthur W. 2000. Patterns of correlated character evolution in flightless birds: a phylogenetic approach. Evolutionary Ecology 14: 693–702. Cubo J, Azagra D, Casinos A, Castanet J. 2002. Heterochronic detection through a function for the ontogenetic variation of bone shape. Journal of Theoretical Biology 215: 57– 66. Cubo J, Fouces V, Gonzalez-Martin M, Pedrocci V, Ruiz X. 2000. Non-heterochronic developmental changes underlie morphological heterochrony in the evolution of the Ardeidae. Journal of Evolutionary Biology 13: 269–276. Garland T Jr, Harvey PH, Ives AR. 1992. Procedures for the analysis of comparative data using phylogenetically independent contrasts. Systematic Biology 41: 18–32. © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106 106 J. CUBO Gould SJ. 1977. Ontogeny and phylogeny. Cambridge, MA: Harvard University Press. Gould SJ, Eldredge N. 1993. Punctated equilibrium comes of age. Nature 366: 223–227. Haddrath O, Baker AJ. 2001. Complete mitochondrial DNA genome sequences of extinct birds: ratite phylogenetics and the vicariance biogeography hypothesis. Proceedings of the Royal Society of London B 268: 939–945. Harvey PH, Pagel M. 1991. The comparative method in evolutionary biology. Oxford: Oxford University Press. Harvey PH, Purvis A. 1991. Comparative methods for explaining adaptations. Nature 351: 619–614. Legendre P, Lapointe FJ, Casgrain P. 1994. Modeling brain evolution from behavior: a permutational regression approach. Evolution 48: 1487–1499. Lemen CA, Freeman PW. 1989. Testing macroevolutionary hypotheses with cladistic analysis: Evidence against rectangular evolution. Evolution 43: 1538–1544. Martins EP, Garland T Jr. 1991. Phylogenetic analyses of the correlated evolution of continuous characters: a simulation study. Evolution 45: 534–557. Mayden RL. 1986. Speciose and depauperate phylads and tests of punctuated and gradual evolution: fact or artifact? Systematic Zoology 35: 591–602. Mindell DP, Sites JW, Graur D. 1989. Speciational evolution: a phylogenetic test with allozymes in Sceloporus (Reptilia). Cladistics 5: 49–61. Mooers AØ. 1995. Tree balance and tree completeness. Evolution 49: 379–384. Mooers AØ, Vamosi SM, Schluter D. 1999. Using phylogenies to test macroevolutionary hypotheses of trait evolution in Cranes (Gruinae). American Naturalist 154: 249–259. Pagel M. 1997. Inferring evolutionary processes from phylogenies. Zoologica Scripta 26: 331–348. Pagel M. 1999. Inferring the historical patterns of biological evolution. Nature 401: 877–884. Rohlf FJ, Chang WS, Sokal RR, Kim J. 1990. Accuracy of estimated phylogenies: effects of tree topology and evolutionary model. Evolution 44: 1671–1684. van Tuinen M, Hedges SB. 2001. Calibration of avian molecular clocks. Molecular Biology and Evolution 18: 206– 213. Unwin DM. 1993. Aves. In: Benton MJ, ed. The fossil record 2. London: Chapman & Hall, 717–737. © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 99–106
© Copyright 2026 Paperzz